Merge remote-tracking branch 'origin/main' into copilot/fix-single-path-indexing-issue

Co-authored-by: magyargergo <11230420+magyargergo@users.noreply.github.com>
This commit is contained in:
copilot-swe-agent[bot] 2026-04-21 12:18:15 +00:00 • committed by GitHub
commit e969b638d1
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
55 changed files with 5485 additions and 360 deletions

View file

@ -135,6 +135,9 @@ jobs:
# the type=semver patterns would not match. In that case we add an explicit
# type=raw tag using the version already verified above, so the same
# image-naming rules apply regardless of how the workflow was triggered.
# NOTE: We check `inputs.tag` rather than `github.event_name` because in a
# reusable workflow the github context is inherited from the caller —
# `github.event_name` would still be "push", not "workflow_call".
- name: Extract Docker metadata
id: meta
uses: docker/metadata-action@030e881283bb7a6894de51c315a6bfe6a94e05cf # v6.0.0
@ -145,7 +148,7 @@ jobs:
type=semver,pattern={{version}}
type=semver,pattern={{major}}.{{minor}}
type=semver,pattern={{major}}
type=raw,value=${{ steps.version.outputs.version }},enable=${{ github.event_name == 'workflow_call' }}
type=raw,value=${{ steps.version.outputs.version }},enable=${{ inputs.tag != '' }}
- name: Build and push
id: build

3
.gitignore vendored
View file

@ -103,3 +103,6 @@ gitnexus/vendor/**/node_modules/
local_docs/
# Local agent scratch / review prompts (never commit)
.tmp/
.agents/

View file

@ -47,6 +47,7 @@ Commands and gotchas live under **Repo reference** below and in **[CONTRIBUTING.
| Date | Version | Change |
|------|---------|--------|
| 2026-04-19 | 1.5.0 | Cross-repo impact (#794): `impact`/`query`/`context` accept `repo: "@<group>"` + `service`. Removed `group_query`/`group_contracts`/`group_status` MCP tools; added `gitnexus://group/{name}/contracts` and `gitnexus://group/{name}/status` resources. |
| 2026-04-16 | 1.4.0 | Fixed: web UI description, pre-commit behavior, MCP tools (7->16), added gitnexus-shared, removed stale vite-plugin-wasm gotcha. |
| 2026-04-13 | 1.3.0 | Updated GitNexus index stats after DAG refactor. |
| 2026-03-24 | 1.2.0 | Fixed gitnexus:start block duplication. |
@ -107,10 +108,14 @@ Indexed as **GitNexus** (4325 symbols, 10556 relationships, 300 execution flows)
| `tool_map` | MCP/RPC tool definitions | `gitnexus_tool_map({})` |
| `shape_check` | Response shape vs consumer access | `gitnexus_shape_check({route: "/api/users"})` |
| `group_list` | List repo groups | `gitnexus_group_list({})` |
| `group_query` | Cross-repo search in a group | `gitnexus_group_query({name: "myGroup", query: "auth"})` |
| `group_sync` | Rebuild group Contract Registry | `gitnexus_group_sync({name: "myGroup"})` |
| `group_contracts` | Inspect group contracts | `gitnexus_group_contracts({name: "myGroup"})` |
| `group_status` | Group staleness report | `gitnexus_group_status({name: "myGroup"})` |
| `query` (group mode) | Cross-repo search in a group (RRF-merged) | `gitnexus_query({repo: "@myGroup", query: "auth"})` |
| `context` (group mode) | 360° view across all member repos | `gitnexus_context({repo: "@myGroup", name: "validateUser"})` |
| `impact` (group mode) | Cross-repo blast radius via Contract Bridge | `gitnexus_impact({repo: "@myGroup", target: "X", direction: "upstream"})` |
> Group mode: pass `repo: "@<groupName>"` to fan out across all member repos, or `repo: "@<groupName>/<memberPath>"` to target a single member (path keys from `group.yaml`). Optional `service: "<monorepo/path>"` filters by service root. Group-level state (contracts, staleness) lives in the resources table below — there are **no** `group_query` / `group_context` / `group_impact` / `group_contracts` / `group_status` MCP tools.
>
> For a full walkthrough of setting up a group across multiple repos that communicate over gRPC, see [docs/guides/microservices-grpc.md](docs/guides/microservices-grpc.md).
## Impact Risk Levels
@ -128,6 +133,8 @@ Indexed as **GitNexus** (4325 symbols, 10556 relationships, 300 execution flows)
| `gitnexus://repo/GitNexus/clusters` | All functional areas |
| `gitnexus://repo/GitNexus/processes` | All execution flows |
| `gitnexus://repo/GitNexus/process/{name}` | Step-by-step execution trace |
| `gitnexus://group/{name}/contracts` | Group Contract Registry (provider/consumer rows + cross-links) |
| `gitnexus://group/{name}/status` | Per-member index + Contract Registry staleness report |
## Self-Check Before Finishing

View file

@ -42,10 +42,14 @@ Monorepo: **CLI/MCP** (`gitnexus/`) + **browser UI** (`gitnexus-web/`).
| `tool_map` | MCP/RPC tool definitions and handlers |
| `shape_check` | Response shape vs consumer property access mismatches |
| `group_list` | List repo groups or details for one group |
| `group_query` | Cross-repo search in a group (reciprocal rank fusion) |
| `group_sync` | Rebuild group Contract Registry (`contracts.json`) |
| `group_contracts` | Inspect group contracts and cross-links |
| `group_status` | Index and Contract Registry staleness per repo in a group |
| `group_sync` | Rebuild group Contract Registry (`contracts.json`) and bridge graph |
`query`, `context`, and `impact` are group-aware: pass `repo: "@<groupName>"` (or `"@<groupName>/<memberPath>"` to scope to one member) plus optional `service: "<monorepo/path>"`. Group-mode `query` merges per-repo results via Reciprocal Rank Fusion; group-mode `impact` runs the local walk in the chosen member and fans out across boundaries via the Contract Bridge (`gitnexus/src/core/group/cross-impact.ts`). The previously-planned `group_query`, `group_context`, `group_impact`, `group_contracts`, `group_status` MCP tools are intentionally not introduced — group-level state is exposed via resources instead:
| Resource URI | Purpose |
|--------------|---------|
| `gitnexus://group/{name}/contracts` | Contract Registry (provider/consumer rows + cross-links) |
| `gitnexus://group/{name}/status` | Per-member index + Contract Registry staleness |
## Where to change what
@ -55,6 +59,7 @@ Monorepo: **CLI/MCP** (`gitnexus/`) + **browser UI** (`gitnexus-web/`).
| Parsing/graph construction | `src/core/ingestion/pipeline-phases/` + `pipeline.ts` |
| Graph schema/DB | `src/core/lbug/` (`schema.ts`, `lbug-adapter.ts`) |
| MCP tools/resources | `src/mcp/server.ts`, `tools.ts`, `resources.ts` |
| Cross-repo groups (sync, contracts, `@<group>` routing) | `src/core/group/` (`service.ts`, `cross-impact.ts`, `sync.ts`, `bridge-db.ts`) |
| Search ranking | `src/core/search/` (BM25, hybrid fusion) |
| Embeddings | `src/core/embeddings/` + `src/core/run-analyze.ts` |
| Wiki generation | `src/core/wiki/` |

View file

@ -11,6 +11,8 @@ RUN npm ci --prefix gitnexus-shared
COPY gitnexus-shared ./gitnexus-shared
RUN npm run build --prefix gitnexus-shared
COPY gitnexus/package.json ./gitnexus/
COPY gitnexus-web/package.json gitnexus-web/package-lock.json ./gitnexus-web/
RUN npm ci --prefix gitnexus-web

View file

@ -9,7 +9,7 @@
<h2>Join the official Discord to discuss ideas, issues etc!</h2>
<a href="https://discord.gg/AAsRVT6fGb">
<a href="https://discord.gg/MgJrmsqr62">
<img src="https://img.shields.io/discord/1477255801545429032?color=5865F2&logo=discord&logoColor=white" alt="Discord"/>
</a>
<a href="https://www.npmjs.com/package/gitnexus">
@ -194,6 +194,7 @@ gitnexus analyze --force # Force full re-index
gitnexus analyze --skills # Generate repo-specific skill files from detected communities
gitnexus analyze --skip-embeddings # Skip embedding generation (faster)
gitnexus analyze --skip-agents-md # Preserve custom AGENTS.md/CLAUDE.md gitnexus section edits
gitnexus analyze --skip-git # Index folders that are not Git repositories
gitnexus analyze --embeddings # Enable embedding generation (slower, better search)
gitnexus analyze --verbose # Log skipped files when parsers are unavailable
gitnexus mcp # Start MCP server (stdio) — serves all indexed repos

View file

@ -0,0 +1,295 @@
# Using GitNexus across gRPC microservices
## When to use this guide
This guide is for teams whose product lives in **several separate Git repositories** — one per service — and whose services talk to each other over **gRPC** (possibly alongside HTTP and message topics). GitNexus indexes each repo independently, then a _group_ stitches the per-repo indexes into a single cross-repo view that the `impact`, `query`, and `context` tools can traverse. If your services live in one monorepo, much of this still applies — set each service as a member of a group and use the `service` prefix to scope queries — but the walkthrough assumes the harder multi-repo case.
## Mental model
- Each repository has its own `.gitnexus/` index (a LadybugDB graph of symbols, relationships, processes). `gitnexus analyze` in each repo produces that index completely independently.
- A **group** is a higher-level construct stored at `~/.gitnexus/groups/<group>/` that references the per-repo indexes by their registry name.
- Sync-time extractors walk each member repo and emit **contracts** — provider or consumer records keyed by a canonical `contractId` (`grpc::auth.AuthService/Login`, `http::GET::/orders`, etc.).
- The sync step matches providers and consumers that share a `contractId` and writes **cross-links** to `<groupDir>/contracts.json`. Those cross-links are what lets `impact({repo: "@<group>", target: "X"})` hop from one repo into another.
- Contracts come from three places: automatic contract extractors (`grpc-extractor`, `http-route-extractor`, `topic-extractor`), a manifest escape hatch (`config.links` in `group.yaml`), and — for same-name symbol matches where no contract is declared — the exact-match matching cascade in [`matching.ts`](../../gitnexus/src/core/group/matching.ts).
- Each repo stays editable and re-indexable on its own. Re-run `gitnexus analyze` in a repo when it changes, then `gitnexus group sync <group>` to refresh `contracts.json`. `gitnexus group status` reports which members are stale.
## Prerequisites
- GitNexus installed and runnable as `gitnexus` or `npx gitnexus` (see the root [README.md](../../README.md)).
- Each service repository checked out locally. No requirement that they share a parent directory — the group references them by registry name.
- Write access to `~/.gitnexus/` (the default gitnexus home; see `getDefaultGitnexusDir` in [`storage.ts`](../../gitnexus/src/core/group/storage.ts)).
## Step-by-step walkthrough
The example uses three services — a TypeScript API gateway, a Go orders service, and a Python inventory service — with gRPC between them. The gateway is an `orders` consumer; the orders service is both an `orders` provider and an `inventory` consumer; the inventory service is an `inventory` provider.
### 1. Index each repository
Run `analyze` from inside each service repo (or pass the path). The CLI surface lives in [`gitnexus/src/cli/analyze.ts`](../../gitnexus/src/cli/analyze.ts) and is wired in [`gitnexus/src/cli/index.ts`](../../gitnexus/src/cli/index.ts).
```bash
cd ~/code/gateway && npx gitnexus analyze
cd ~/code/orders && npx gitnexus analyze
cd ~/code/inventory && npx gitnexus analyze
```
Useful flags:
- `--force` — reindex even if up to date.
- `--embeddings` — generate embedding vectors (needed only if you want semantic search; the exact-match cross-repo cascade does **not** need them).
- `--name <alias>` — register the repo under a specific alias when two repos share a basename (e.g. two `api/` folders).
- `--skip-git` — index a checkout that isn't a git repo.
Each run writes a `.gitnexus/` folder in the repo and registers the repo in `~/.gitnexus/registry.json`. Confirm with `npx gitnexus list`.
### 2. Author `group.yaml`
Create the group directory and edit the config. Either use the CLI scaffolder or write the file directly — both produce the same shape consumed by [`config-parser.ts`](../../gitnexus/src/core/group/config-parser.ts).
```bash
npx gitnexus group create payments-platform
# or manually:
mkdir -p ~/.gitnexus/groups/payments-platform
$EDITOR ~/.gitnexus/groups/payments-platform/group.yaml
```
Minimal working `group.yaml`:
```yaml
version: 1
name: payments-platform
description: Gateway + orders + inventory (gRPC)
repos:
gateway: gateway
orders: orders
inventory: inventory
# Only add explicit links when the automatic extractors miss something —
# see "When automatic extraction isn't enough" below.
links: []
packages: {}
detect:
http: true
grpc: true
topics: true
shared_libs: true
embedding_fallback: false
matching:
bm25_threshold: 0.7
embedding_threshold: 0.65
max_candidates_per_step: 3
```
Field notes (schema in [`types.ts`](../../gitnexus/src/core/group/types.ts)):
- `version` — must be `1`. The parser rejects anything else.
- `name` — required; used for the group directory name and all CLI / MCP calls.
- `repos` — a mapping from **group path** (a logical name you choose; can be a hierarchy like `backend/orders`) to **registry name** (the name shown by `npx gitnexus list`). Both sides appear throughout the tooling: contract rows use the group path; `@<group>/<groupPath>` routes tools to a single member.
- `links` — optional manifest escape hatch, one entry per explicit cross-repo contract. Validated by the parser: `from` and `to` must be known repo paths, `type` must be one of `http | grpc | topic | lib | custom`, and `role` must be `provider | consumer`.
- `detect` — toggles per extractor family. Defaults (set in `config-parser.ts`) turn `http`, `grpc`, `topics`, and `shared_libs` on; disable the ones you don't use to speed up sync.
- `matching` — thresholds for the matching cascade. The exact match is always run; other strategies depend on indexer state.
### 3. Sync the group
```bash
npx gitnexus group sync payments-platform --verbose
```
What this does (see [`sync.ts`](../../gitnexus/src/core/group/sync.ts)):
1. Opens each member's per-repo LadybugDB.
2. Runs the HTTP, gRPC, and topic extractors against the source files.
3. Applies manifest `links` through [`manifest-extractor.ts`](../../gitnexus/src/core/group/extractors/manifest-extractor.ts).
4. Runs the exact-match cascade, joining providers and consumers that share a normalized `contractId`.
5. Writes `contracts.json` in the group directory.
Flags:
- `--exact-only` — stop after the exact cascade; skip BM25 and embedding fallback.
- `--skip-embeddings` — run exact plus BM25 but not embedding-based matching.
- `--allow-stale` — don't warn if a member's index is stale.
- `--json` — machine-readable output.
The same operation is available over MCP as `group_sync({ name: "payments-platform" })` — see [`tools.ts`](../../gitnexus/src/mcp/tools.ts).
### 4. Inspect the registry
Use `gitnexus group contracts` for the CLI view or read the `gitnexus://group/<name>/contracts` MCP resource for the same data.
```bash
npx gitnexus group contracts payments-platform --type grpc --json
```
A shortened response:
```json
{
"contracts": [
{
"contractId": "grpc::orders.OrderService/PlaceOrder",
"type": "grpc",
"role": "provider",
"repo": "orders",
"symbolRef": { "filePath": "internal/grpc/order_server.go", "name": "RegisterOrderServiceServer" },
"confidence": 0.8,
"meta": { "service": "OrderService", "method": "PlaceOrder", "source": "go_register" }
},
{
"contractId": "grpc::orders.OrderService/PlaceOrder",
"type": "grpc",
"role": "consumer",
"repo": "gateway",
"symbolRef": { "filePath": "src/clients/orders.ts", "name": "OrderServiceClient" },
"confidence": 0.75,
"meta": { "service": "OrderService", "source": "ts_generated_client" }
}
],
"crossLinks": [
{
"from": { "repo": "gateway", "symbolUid": "…", "symbolRef": { "filePath": "src/clients/orders.ts", "name": "OrderServiceClient" } },
"to": { "repo": "orders", "symbolUid": "…", "symbolRef": { "filePath": "internal/grpc/order_server.go", "name": "RegisterOrderServiceServer" } },
"type": "grpc",
"contractId": "grpc::orders.OrderService/PlaceOrder",
"matchType": "exact",
"confidence": 1.0
}
]
}
```
Staleness of the underlying indexes shows up in `npx gitnexus group status payments-platform` or the `gitnexus://group/<name>/status` resource.
### 5. Run cross-repo impact with `@<group>` routing
From any shell (you do **not** have to `cd` into a member repo), the normal `impact` / `query` / `context` tools accept `repo: "@<group>"` to fan out across all members, or `repo: "@<group>/<memberPath>"` to target one member. Routing is implemented in [`resolve-at-member.ts`](../../gitnexus/src/core/group/resolve-at-member.ts) and described in [`tools.ts`](../../gitnexus/src/mcp/tools.ts).
Example MCP calls:
```json
{"tool": "impact", "arguments": {
"repo": "@payments-platform/orders",
"target": "PlaceOrder",
"direction": "upstream",
"crossDepth": 2
}}
```
```json
{"tool": "query", "arguments": {
"repo": "@payments-platform",
"query": "retry logic around PlaceOrder"
}}
```
The CLI equivalents still exist for scripting:
```bash
npx gitnexus group impact payments-platform \
--repo orders --target PlaceOrder --direction upstream --cross-depth 2
```
Phase 1 walks within the anchor member; Phase 2 hops across the Contract Bridge wherever a cross-link endpoint matches an impacted symbol. See [`cross-impact.ts`](../../gitnexus/src/core/group/cross-impact.ts) for the bridge query.
## How gRPC extraction works
`GrpcExtractor` ([`grpc-extractor.ts`](../../gitnexus/src/core/group/extractors/grpc-extractor.ts)) runs two passes per member repo:
1. **Proto map.** Every `**/*.proto` file is parsed to enumerate `service Foo { rpc Bar(...) }` blocks and (transitively) resolve the package name. Each RPC method becomes a provider contract with `contractId = grpc::<package>.<Service>/<Method>` and `confidence = 0.85`. Parsing uses the vendored `tree-sitter-proto` grammar when available and falls back to a length-preserving manual parser (`extractServiceBlocks`) otherwise, so `.proto` extraction works on platforms where the grammar fails to build.
2. **Source scan.** Every source file whose extension matches [`GRPC_SCAN_GLOB`](../../gitnexus/src/core/group/extractors/grpc-patterns/index.ts) is parsed by its language plugin:
| Language | Provider signal | Consumer signal |
|----------|-----------------|-----------------|
| Go ([`go.ts`](../../gitnexus/src/core/group/extractors/grpc-patterns/go.ts)) | `pb.RegisterXxxServer(...)`, `pb.UnimplementedXxxServer` embedded in struct | `pb.NewXxxClient(conn)` |
| Java ([`java.ts`](../../gitnexus/src/core/group/extractors/grpc-patterns/java.ts)) | `extends XxxServiceGrpc.XxxServiceImplBase` (with or without `@GrpcService`) | `XxxServiceGrpc.newBlockingStub(...)`, `newStub(...)` |
| Python ([`python.ts`](../../gitnexus/src/core/group/extractors/grpc-patterns/python.ts)) | `add_XxxServicer_to_server(...)` (bare or `_pb2_grpc.` attribute form) | `XxxStub(channel)` (ignores `Mock`/`Test`/`Fake`/`Stub`) |
| Node / TS ([`node.ts`](../../gitnexus/src/core/group/extractors/grpc-patterns/node.ts)) | NestJS `@GrpcMethod('Service','Method')` | `@GrpcClient` field typed `XxxServiceClient`, `client.getService<X>('Service')`, `new XxxServiceClient(...)`, `new foo.bar.XxxService(...)` in files that call `loadPackageDefinition` |
For each source-scan detection the extractor looks up the short service name in the proto map and picks:
- `grpc::<package>.<Service>/<Method>` when a method is named and the service resolves against the proto map,
- `grpc::<package>.<Service>/*` (wildcard) when only the service is known, or
- `grpc::<ServiceName>/*` when no `.proto` is available at all.
Provider detections land at confidence 0.8 (with proto) or 0.65 (without); consumers at 0.75 or 0.55. NestJS `@GrpcMethod` is fixed at 0.8 because the decorator is self-describing.
### Matching
`matching.ts` lowercases the package/service segment before comparing contract ids, so bindings that capitalize names differently (`auth.AuthService` vs `auth.authservice`) still match. Method names are compared case-sensitively because gRPC's wire path is case-sensitive. Service-only wildcards (`grpc::pkg.Svc/*`) match any method on the same service during cross-linking.
### Known limitations
- **Ambiguous proto resolution.** If a short service name exists in more than one `.proto` file and the source-scan hit can't be narrowed down by shared directory segments (`resolveProtoConflict` refuses to guess), the extractor skips contract emission and logs a warning.
- **Proto packages must be resolvable locally.** Transitive imports that point outside the repo produce an empty package segment, which means the contract id collapses to `grpc::<Service>/<Method>`. Cross-repo matches still work as long as both sides agree on the empty package.
- **Rewrite rules are not implemented.** If the provider repo writes `grpc::orders.OrderService/PlaceOrder` and the consumer repo writes `grpc::orderspb.OrderService/PlaceOrder`, they won't cross-link automatically. Use `config.links` to declare the correspondence (see below).
- **One sync = one snapshot.** Contracts are extracted against the indexed snapshot of each repo. Re-index first, then re-sync; the `status` command and resource surface staleness.
## When automatic extraction isn't enough
The escape hatch is the `links` list in `group.yaml`, handled by [`ManifestExtractor`](../../gitnexus/src/core/group/extractors/manifest-extractor.ts). Each entry is a **one-directional** provider/consumer declaration:
```yaml
version: 1
name: payments-platform
repos:
gateway: gateway
orders: orders
inventory: inventory
links:
# Explicit gRPC method: use when naming mismatches stop the
# automatic matcher from cross-linking.
- from: gateway
to: orders
type: grpc
contract: OrderService/PlaceOrder
role: consumer
# Service-level link when you don't want to enumerate methods.
- from: orders
to: inventory
type: grpc
contract: InventoryService
role: consumer
# Works for HTTP too — use `METHOD::/path` form for the exact
# handler, or just `/path` for a method-agnostic wildcard.
- from: gateway
to: orders
type: http
contract: POST::/orders
role: consumer
```
What the manifest extractor does (see [`manifest-extractor.ts`](../../gitnexus/src/core/group/extractors/manifest-extractor.ts)):
1. Builds a canonical `contractId` with `buildContractId` — the same canonicalization used by the automatic extractors, so manifest links cross-match automatic contracts on the other side.
2. Tries to resolve each side to a real graph symbol (the `Route` node for HTTP, a `Function|Method` / `Class|Interface` for gRPC, a `Package|Module` for `lib`).
3. If resolution fails, falls back to a deterministic synthetic uid (`manifest::<repo>::<contractId>`) so both sides still line up in cross-impact — name-only links still work when the symbol isn't in the graph.
4. Emits both a provider and a consumer `StoredContract` (confidence `1.0`, `source: "manifest"`) and a `CrossLink` with `matchType: "manifest"`.
Use `links` for exactly the cases the extractor can't infer: different package names across repos (see #701), hand-rolled transports, cases where the provider repo isn't checked out locally but you still want a record, or any contract whose provider and consumer simply don't share a surface the extractors know how to pattern-match.
History: the manifest extractor used to be silently skipped by the sync pipeline; that was fixed in [#827](https://github.com/abhigyanpatwari/GitNexus/pull/827) (tracking issue #826). If you ever see `config.links` with zero cross-links in `contracts.json`, make sure you're on a build that includes that fix, then re-run `group sync`.
## Troubleshooting
1. **`contracts.json` is empty after a sync.** Either no member repo contained a recognizable gRPC pattern, or the extractors are disabled in `detect`. Confirm `detect.grpc: true` and re-run with `--verbose`.
2. **A known provider/consumer pair doesn't cross-link.** Most common cause: the package segment differs. Check the raw contract ids with `gitnexus group contracts <name> --unmatched` — if you see two same-method contracts with different package prefixes, add a manifest `links:` entry to bridge them (no automatic rewrite rules yet).
3. **`matchType: "manifest"` is missing entirely.** The extractor needs `config.links` to be non-empty and the sync pipeline to actually call it — verify you're on a post-#827 build. Empty contract rows for manifest links usually mean `resolveSymbol` couldn't find a graph match; the synthetic uid still lets cross-impact work, it just won't carry a file path.
4. **Ambiguous proto warnings.** Look for `[grpc-extractor] Ambiguous proto resolution` in the sync logs; that means a service name exists in multiple `.proto` files under the same repo and the path-distance heuristic couldn't pick a winner. Resolve by renaming the service or declaring the intended pairing in `config.links`.
5. **Cross-impact says "stale".** Both sides need a fresh per-repo index _and_ a fresh group sync. Order matters: `gitnexus analyze` in each changed repo, then `gitnexus group sync <name>`. Use `gitnexus group status <name>` to see which side is behind.
## Related docs and references
- [AGENTS.md](../../AGENTS.md) — authoritative list of MCP tools and resources, including group-mode routing and the `gitnexus://group/…` resources.
- [ARCHITECTURE.md](../../ARCHITECTURE.md) — overall data flow and the call-resolution DAG that the per-repo indexer uses.
- [`gitnexus/src/core/group/`](../../gitnexus/src/core/group/) — `service.ts`, `sync.ts`, `config-parser.ts`, `matching.ts`.
- [`gitnexus/src/core/group/extractors/grpc-extractor.ts`](../../gitnexus/src/core/group/extractors/grpc-extractor.ts) and [`grpc-patterns/`](../../gitnexus/src/core/group/extractors/grpc-patterns/) — gRPC detection.
- [`gitnexus/src/core/group/extractors/manifest-extractor.ts`](../../gitnexus/src/core/group/extractors/manifest-extractor.ts) — the `config.links` escape hatch.
- [`gitnexus/src/mcp/tools.ts`](../../gitnexus/src/mcp/tools.ts) — MCP tool schemas (`group_list`, `group_sync`, plus `@<group>` routing on `impact` / `query` / `context`).
- [`gitnexus/src/cli/group.ts`](../../gitnexus/src/cli/group.ts) — CLI command definitions and flags.
- Upstream issues: [#701](https://github.com/abhigyanpatwari/GitNexus/issues/701), [#826](https://github.com/abhigyanpatwari/GitNexus/issues/826), [#906](https://github.com/abhigyanpatwari/GitNexus/issues/906).

View file

@ -2,6 +2,12 @@
All notable changes to GitNexus will be documented in this file.
## [Unreleased]
### Performance
- **`analyze` ~33% faster** — moved FTS index creation from the analyze pipeline to first-use lazy initialisation. The 5 `CREATE_FTS_INDEX` calls cost ~440 ms each in LadybugDB regardless of table size (≈2 s fixed overhead) and dominated runtime on small repos and slow CI runners. The cost now amortises across the first `query`/`context` call in a session via a new `ensureFTSIndex` helper. Mini-repo `analyze` measured locally on Windows: 6.4 s → 4.0 s warm; on CI Windows runners (≈3× slower) restores comfortable headroom against the 30 s e2e test budget.
## [1.6.2] - 2026-04-18
### Added

View file

@ -19,7 +19,7 @@
"cors": "^2.8.5",
"express": "^4.19.2",
"glob": "^13.0.6",
"graphology": "^0.25.4",
"graphology": "^0.26.0",
"graphology-indices": "^0.17.0",
"graphology-utils": "^2.3.0",
"ignore": "^7.0.5",
@ -40,7 +40,7 @@
"tree-sitter-ruby": "^0.23.1",
"tree-sitter-rust": "0.23.1",
"tree-sitter-typescript": "^0.23.2",
"uuid": "^13.0.0"
"uuid": "^14.0.0"
},
"bin": {
"gitnexus": "dist/cli/index.js"
@ -50,7 +50,7 @@
"@types/cors": "^2.8.17",
"@types/express": "^4.17.21",
"@types/js-yaml": "^4.0.9",
"@types/node": "^20.0.0",
"@types/node": "^25.6.0",
"@types/uuid": "^10.0.0",
"@vitest/coverage-v8": "^4.0.18",
"gitnexus-shared": "file:../gitnexus-shared",
@ -2041,12 +2041,12 @@
"license": "MIT"
},
"node_modules/@types/node": {
"version": "20.19.37",
"resolved": "https://registry.npmjs.org/@types/node/-/node-20.19.37.tgz",
"integrity": "sha512-8kzdPJ3FsNsVIurqBs7oodNnCEVbni9yUEkaHbgptDACOPW04jimGagZ51E6+lXUwJjgnBw+hyko/lkFWCldqw==",
"version": "25.6.0",
"resolved": "https://registry.npmjs.org/@types/node/-/node-25.6.0.tgz",
"integrity": "sha512-+qIYRKdNYJwY3vRCZMdJbPLJAtGjQBudzZzdzwQYkEPQd+PJGixUL5QfvCLDaULoLv+RhT3LDkwEfKaAkgSmNQ==",
"license": "MIT",
"dependencies": {
"undici-types": "~6.21.0"
"undici-types": "~7.19.0"
}
},
"node_modules/@types/qs": {
@ -3316,13 +3316,12 @@
"license": "ISC"
},
"node_modules/graphology": {
"version": "0.25.4",
"resolved": "https://registry.npmjs.org/graphology/-/graphology-0.25.4.tgz",
"integrity": "sha512-33g0Ol9nkWdD6ulw687viS8YJQBxqG5LWII6FI6nul0pq6iM2t5EKquOTFDbyTblRB3O9I+7KX4xI8u5ffekAQ==",
"version": "0.26.0",
"resolved": "https://registry.npmjs.org/graphology/-/graphology-0.26.0.tgz",
"integrity": "sha512-8SSImzgUUYC89Z042s+0r/vMibY7GX/Emz4LDO5e7jYXhuoWfHISPFJYjpRLUSJGq6UQ6xlenvX1p/hJdfXuXg==",
"license": "MIT",
"dependencies": {
"events": "^3.3.0",
"obliterator": "^2.0.2"
"events": "^3.3.0"
},
"peerDependencies": {
"graphology-types": ">=0.24.0"
@ -5412,9 +5411,9 @@
}
},
"node_modules/undici-types": {
"version": "6.21.0",
"resolved": "https://registry.npmjs.org/undici-types/-/undici-types-6.21.0.tgz",
"integrity": "sha512-iwDZqg0QAGrg9Rav5H4n0M64c3mkR59cJ6wQp+7C4nI0gsmExaedaYLNO44eT4AtBBwjbTiGPMlt2Md0T9H9JQ==",
"version": "7.19.2",
"resolved": "https://registry.npmjs.org/undici-types/-/undici-types-7.19.2.tgz",
"integrity": "sha512-qYVnV5OEm2AW8cJMCpdV20CDyaN3g0AjDlOGf1OW4iaDEx8MwdtChUp4zu4H0VP3nDRF/8RKWH+IPp9uW0YGZg==",
"license": "MIT"
},
"node_modules/universalify": {
@ -5451,9 +5450,9 @@
}
},
"node_modules/uuid": {
"version": "13.0.0",
"resolved": "https://registry.npmjs.org/uuid/-/uuid-13.0.0.tgz",
"integrity": "sha512-XQegIaBTVUjSHliKqcnFqYypAd4S+WCYt5NIeRs6w/UAry7z8Y9j5ZwRRL4kzq9U3sD6v+85er9FvkEaBpji2w==",
"version": "14.0.0",
"resolved": "https://registry.npmjs.org/uuid/-/uuid-14.0.0.tgz",
"integrity": "sha512-Qo+uWgilfSmAhXCMav1uYFynlQO7fMFiMVZsQqZRMIXp0O7rR7qjkj+cPvBHLgBqi960QCoo/PH2/6ZtVqKvrg==",
"funding": [
"https://github.com/sponsors/broofa",
"https://github.com/sponsors/ctavan"

View file

@ -60,7 +60,7 @@
"cors": "^2.8.5",
"express": "^4.19.2",
"glob": "^13.0.6",
"graphology": "^0.25.4",
"graphology": "^0.26.0",
"graphology-indices": "^0.17.0",
"graphology-utils": "^2.3.0",
"ignore": "^7.0.5",
@ -81,7 +81,7 @@
"tree-sitter-ruby": "^0.23.1",
"tree-sitter-rust": "0.23.1",
"tree-sitter-typescript": "^0.23.2",
"uuid": "^13.0.0"
"uuid": "^14.0.0"
},
"optionalDependencies": {
"node-addon-api": "^8.0.0",
@ -97,7 +97,7 @@
"@types/cors": "^2.8.17",
"@types/express": "^4.17.21",
"@types/js-yaml": "^4.0.9",
"@types/node": "^20.0.0",
"@types/node": "^25.6.0",
"@types/uuid": "^10.0.0",
"@vitest/coverage-v8": "^4.0.18",
"tsx": "^4.0.0",

View file

@ -121,7 +121,7 @@ ${
groupNames && groupNames.length > 0
? `## Cross-Repo Groups
This repository is listed under GitNexus **group(s): ${groupNames.join(', ')}** (see \`~/.gitnexus/groups/\`). For blast radius across repository boundaries, use MCP tools \`group_impact\`, \`group_sync\`, \`group_query\`, \`group_contracts\`, \`group_status\`, and \`group_list\`. From the terminal: \`npx gitnexus group list\`, \`npx gitnexus group sync <name>\`, \`npx gitnexus group impact <name> --target <symbol> --repo <group-path>\`.
This repository is listed under GitNexus **group(s): ${groupNames.join(', ')}** (see \`~/.gitnexus/groups/\`). For cross-repo analysis, use MCP tools \`impact\`, \`query\`, and \`context\` with \`repo\` set to \`@<groupName>\` or \`@<groupName>/<memberPath>\` (paths match keys in that group’s \`group.yaml\`). Use \`group_list\` / \`group_sync\` for membership and sync. From the terminal: \`npx gitnexus group list\`, \`npx gitnexus group sync <name>\`, \`npx gitnexus group impact <name> --target <symbol> --repo <group-path>\`.
`
: ''

View file

@ -6,7 +6,13 @@
*/
import fs from 'fs/promises';
import { findRepo, unregisterRepo, listRegisteredRepos } from '../storage/repo-manager.js';
import {
findRepo,
unregisterRepo,
listRegisteredRepos,
assertSafeStoragePath,
UnsafeStoragePathError,
} from '../storage/repo-manager.js';
export const cleanCommand = async (options?: { force?: boolean; all?: boolean }) => {
// --all flag: clean all indexed repos
@ -27,6 +33,24 @@ export const cleanCommand = async (options?: { force?: boolean; all?: boolean })
const entries = await listRegisteredRepos();
for (const entry of entries) {
// Safety guard (#1003 review — @magyargergo): same rationale as
// remove.ts. `~/.gitnexus/registry.json` is user-writable, so a
// corrupted or hand-edited entry could point storagePath at the
// repo root, an empty string, or anywhere else — and
// fs.rm(recursive: true) on any of those would be catastrophic.
// Skip poisoned entries without touching disk, but keep going
// through the rest of the registry (preserves the existing
// per-repo error-tolerance semantics of `clean --all`).
try {
assertSafeStoragePath(entry);
} catch (err) {
if (err instanceof UnsafeStoragePathError) {
console.error(`Refusing to clean ${entry.name}: ${err.message}`);
continue;
}
throw err;
}
try {
await fs.rm(entry.storagePath, { recursive: true, force: true });
await unregisterRepo(entry.path);

View file

@ -184,6 +184,83 @@ export function registerGroupCommands(program: Command): void {
}
});
group
.command('impact <name>')
.description('Cross-repo impact for a symbol in one member repo of a group')
.requiredOption('--target <symbol>', 'Symbol or file name to analyze')
.requiredOption(
'--repo <groupPath>',
'Member path from group.yaml (e.g. app/backend), not the indexed repo name',
)
.option('--direction <dir>', 'upstream or downstream', 'upstream')
.option('--service <path>', 'Optional monorepo service directory prefix (path filter)')
.option(
'--subgroup <path>',
'Optional prefix limiting which group repos participate in cross fan-out',
)
.option('--max-depth <n>', 'Max graph traversal depth')
.option('--cross-depth <n>', 'Cross-repository hop depth')
.option('--min-confidence <n>', 'Minimum relation confidence (0–1)')
.option('--include-tests', 'Include test files in traversal', false)
.option('--timeout-ms <n>', 'Phase-1 local impact wall time in milliseconds')
.option('--json', 'JSON output')
.action(async (name: string, opts: Record<string, string | boolean | undefined>) => {
const { LocalBackend } = await import('../mcp/local/local-backend.js');
const backend = new LocalBackend();
try {
await backend.init();
const payload: Record<string, unknown> = {
name,
repo: opts.repo,
target: opts.target,
direction: (opts.direction as string) || 'upstream',
};
if (opts.service) payload.service = opts.service;
if (opts.subgroup) payload.subgroup = opts.subgroup;
if (opts.maxDepth !== undefined && opts.maxDepth !== '') {
const n = parseInt(String(opts.maxDepth), 10);
if (!Number.isNaN(n)) payload.maxDepth = n;
}
if (opts.crossDepth !== undefined && opts.crossDepth !== '') {
const n = parseInt(String(opts.crossDepth), 10);
if (!Number.isNaN(n)) payload.crossDepth = n;
}
if (opts.minConfidence !== undefined && opts.minConfidence !== '') {
const n = parseFloat(String(opts.minConfidence));
if (!Number.isNaN(n)) payload.minConfidence = n;
}
if (opts.timeoutMs !== undefined && opts.timeoutMs !== '') {
const n = parseInt(String(opts.timeoutMs), 10);
if (!Number.isNaN(n)) payload.timeoutMs = n;
}
if (opts.includeTests) payload.includeTests = true;
const raw = await backend.getGroupService().groupImpact(payload);
if (raw && typeof raw === 'object' && 'error' in raw) {
console.error(String((raw as { error: string }).error));
process.exitCode = 1;
return;
}
if (opts.json) {
console.log(JSON.stringify(raw, null, 2));
} else {
const summary = (raw as { summary?: Record<string, number> })?.summary;
const risk = (raw as { risk?: string })?.risk;
console.log(`Group impact for "${name}" (${String(opts.repo)}): risk=${risk ?? '?'}`);
if (summary) {
console.log(
` direct=${summary.direct ?? 0} processes=${summary.processes_affected ?? 0} cross=${summary.cross_repo_hits ?? 0}`,
);
}
}
} finally {
await backend.dispose().catch(() => {});
}
});
group
.command('query <name> <query>')
.description('Search execution flows across all repos in a group')

View file

@ -83,6 +83,15 @@ program
.option('--all', 'Clean all indexed repos')
.action(createLazyAction(() => import('./clean.js'), 'cleanCommand'));
program
.command('remove <target>')
.description(
'Delete the GitNexus index for a registered repo (by alias, name, or absolute path). ' +
'Unlike `clean`, does not require being inside the repo. Idempotent on unknown targets.',
)
.option('-f, --force', 'Skip confirmation prompt')
.action(createLazyAction(() => import('./remove.js'), 'removeCommand'));
program
.command('wiki [path]')
.description('Generate repository wiki from knowledge graph')
@ -151,6 +160,15 @@ program
.option('-r, --repo <name>', 'Target repository')
.action(createLazyAction(() => import('./tool.js'), 'cypherCommand'));
program
.command('detect-changes')
.alias('detect_changes')
.description('Map git diff hunks to indexed symbols and affected execution flows')
.option('-s, --scope <scope>', 'What to analyze: unstaged, staged, all, or compare', 'unstaged')
.option('-b, --base-ref <ref>', 'Branch/commit for compare scope (e.g. main)')
.option('-r, --repo <name>', 'Target repository')
.action(createLazyAction(() => import('./tool.js'), 'detectChangesCommand'));
// ─── Eval Server (persistent daemon for SWE-bench) ─────────────────
program

110
gitnexus/src/cli/remove.ts Normal file
View file

@ -0,0 +1,110 @@
/**
* Remove Command (#664)
*
* Delete the `.gitnexus/` index for a registered repo and unregister it
* from the global registry (~/.gitnexus/registry.json). The target is
* identified by alias / basename-derived name / remote-inferred name /
* absolute path — no `--repo` flag, just a positional argument so the
* destructive-command ergonomics match `clean` (which is also
* destructive but scoped to `process.cwd()`).
*
* Compared to `clean`:
* - `clean` acts on the repo discovered by walking up from cwd.
* - `remove` acts on any registered repo identified by name or path.
*
* Behaviour notes:
* - Idempotent on unknown targets: exits 0 with a warning so that
* `remove X && analyze Y` keeps working in scripts. Per #664:
* "behave atomically and idempotently so retries are safe".
* - Atomic order mirrors `clean`: fs.rm FIRST, then unregister. A
* partial failure leaves the registry pointing at a missing dir
* (recoverable by `listRegisteredRepos({ validate: true })` on
* next read) rather than the opposite, which would orphan
* .gitnexus/ directories on disk.
* - `-f` / `--force` matches the confirmation-skip semantics of
* `clean -f`. (Distinct from `analyze --force`, which re-indexes;
* here there is no pipeline, so no conflation.)
*/
import fs from 'fs/promises';
import {
readRegistry,
resolveRegistryEntry,
assertSafeStoragePath,
unregisterRepo,
RegistryNotFoundError,
RegistryAmbiguousTargetError,
UnsafeStoragePathError,
} from '../storage/repo-manager.js';
export const removeCommand = async (target: string, options?: { force?: boolean }) => {
// Read the registry snapshot once and pass it to the resolver — this
// lets us render the "before" state in the dry-run path without a
// second disk read.
const entries = await readRegistry();
let entry;
try {
entry = resolveRegistryEntry(entries, target);
} catch (err) {
if (err instanceof RegistryNotFoundError) {
// Idempotent: missing target is a no-op warning, not an error.
// The `availableNames` hint comes from the error itself so users
// can see what they might have meant.
console.warn(`Nothing to remove: ${err.message}`);
return;
}
if (err instanceof RegistryAmbiguousTargetError) {
// Duplicate aliases are allowed via --allow-duplicate-name (#829);
// refuse to guess which one the user meant — surface the full list
// and exit non-zero so scripts don't silently pick the wrong repo.
console.error(`Error: ${err.message}`);
process.exit(1);
}
throw err;
}
// Confirmation gate — same shape as `clean`. Default is a dry-run
// that describes what would be deleted; `--force` actually deletes.
if (!options?.force) {
console.log(`This will delete the GitNexus index for: ${entry.name}`);
console.log(` Path: ${entry.path}`);
console.log(` Storage: ${entry.storagePath}`);
console.log('\nRun with --force to confirm deletion.');
return;
}
// Safety guard (#1003 review — @magyargergo): refuse to proceed if
// the registry entry's `storagePath` isn't the canonical
// `<entry.path>/.gitnexus` subfolder. `~/.gitnexus/registry.json` is
// user-writable, so a corrupted or hand-edited entry could point
// storagePath at the repo root, an empty string (→ cwd), a parent
// dir, or anywhere else; `fs.rm(recursive: true, force: true)` on
// any of those would be a runtime disaster. Bail before touching
// disk, with an actionable hint for recovering a broken registry.
try {
assertSafeStoragePath(entry);
} catch (err) {
if (err instanceof UnsafeStoragePathError) {
console.error(`Error: ${err.message}`);
process.exit(1);
}
throw err;
}
// Deletion order: fs.rm first, then unregister. If fs.rm fails mid-way,
// the registry entry stays so the user can retry. If fs.rm succeeds but
// unregister throws (e.g. ENOSPC on registry write), the entry becomes
// orphaned — `listRegisteredRepos({ validate: true })` prunes those on
// next read, so the failure is self-healing.
try {
await fs.rm(entry.storagePath, { recursive: true, force: true });
await unregisterRepo(entry.path);
console.log(`Removed: ${entry.name}`);
console.log(` Path: ${entry.path}`);
console.log(` Storage: ${entry.storagePath}`);
} catch (err) {
console.error(`Failed to remove ${entry.name}:`, err);
process.exit(1);
}
};

View file

@ -164,3 +164,55 @@ export async function cypherCommand(
});
output(result);
}
function formatDetectChangesResult(result: any): string {
if (result?.error) return `Error: ${result.error}`;
const summary = result?.summary || {};
if ((summary.changed_count || 0) === 0) {
return 'No changes detected.';
}
const lines: string[] = [];
lines.push(`Changes: ${summary.changed_files || 0} files, ${summary.changed_count || 0} symbols`);
lines.push(`Affected processes: ${summary.affected_count || 0}`);
lines.push(`Risk level: ${summary.risk_level || 'unknown'}`);
lines.push('');
const changed = result?.changed_symbols || [];
if (changed.length > 0) {
lines.push('Changed symbols:');
for (const symbol of changed.slice(0, 15)) {
lines.push(` ${symbol.type} ${symbol.name} → ${symbol.filePath}`);
}
if (changed.length > 15) {
lines.push(` ... and ${changed.length - 15} more`);
}
lines.push('');
}
const affected = result?.affected_processes || [];
if (affected.length > 0) {
lines.push('Affected execution flows:');
for (const processInfo of affected.slice(0, 10)) {
const steps = (processInfo.changed_steps || []).map((s: any) => s.symbol).join(', ');
lines.push(` • ${processInfo.name} (${processInfo.step_count} steps) — changed: ${steps}`);
}
}
return lines.join('\n').trim();
}
export async function detectChangesCommand(options?: {
scope?: string;
baseRef?: string;
repo?: string;
}): Promise<void> {
const backend = await getBackend();
const result = await backend.callTool('detect_changes', {
scope: options?.scope || 'unstaged',
base_ref: options?.baseRef,
repo: options?.repo,
});
output(formatDetectChangesResult(result));
}

View file

@ -64,16 +64,16 @@ const FUNCTION_LIKE_TYPES = new Set([
* numbers don't apply.
*/
export const findFunctionNode = (root: any): any | null => {
if (FUNCTION_LIKE_TYPES.has(root.type)) return root;
for (let i = 0; i < root.namedChildCount; i++) {
const child = root.namedChild(i);
if (!child) continue;
if (FUNCTION_LIKE_TYPES.has(child.type)) return child;
const found = findFunctionNode(child);
if (found) return found;
// Iterative DFS — avoids stack overflow on deeply nested ASTs.
const stack = [root];
while (stack.length > 0) {
const node = stack.pop()!;
if (FUNCTION_LIKE_TYPES.has(node.type)) return node;
for (let i = node.namedChildCount - 1; i >= 0; i--) {
const child = node.namedChild(i);
if (child) stack.push(child);
}
}
return null;
};
@ -98,15 +98,15 @@ export const findDeclarationNode = (root: any): any | null => {
'impl_item', // Rust: impl
]);
if (CLASS_LIKE_TYPES.has(root.type)) return root;
for (let i = 0; i < root.namedChildCount; i++) {
const child = root.namedChild(i);
if (!child) continue;
if (CLASS_LIKE_TYPES.has(child.type)) return child;
const found = findDeclarationNode(child);
if (found) return found;
// Iterative DFS — avoids stack overflow on deeply nested ASTs.
const stack = [root];
while (stack.length > 0) {
const node = stack.pop()!;
if (CLASS_LIKE_TYPES.has(node.type)) return node;
for (let i = node.namedChildCount - 1; i >= 0; i--) {
const child = node.namedChild(i);
if (child) stack.push(child);
}
}
return null;
};

View file

@ -13,6 +13,12 @@ import { characterChunk } from './character-chunk.js';
import type { Chunk } from './character-chunk.js';
import { ensureAndParse, findDeclarationNode, findFunctionNode } from './ast-utils.js';
import { buildLineIndex, resolveChunkLines } from './line-index.js';
import {
CHUNKING_RULES,
CHUNK_MODE_AST_DECLARATION,
CHUNK_MODE_AST_FUNCTION,
type ChunkingRule,
} from './types.js';
/**
* Main chunkNode function: dispatches by label
@ -40,31 +46,39 @@ export const chunkNode = async (
];
}
// Only function-like labels get AST chunking
if (label === 'Function' || label === 'Method' || label === 'Constructor') {
try {
const astChunks = await astChunk(content, filePath, startLine, endLine, chunkSize, overlap);
if (astChunks.length > 0) return astChunks;
} catch {
// AST parsing failed — fall through to character fallback
}
const rule = CHUNKING_RULES[label];
if (!rule) {
return characterChunk(content, startLine, endLine, chunkSize, overlap);
}
if (label === 'Class' || label === 'Interface') {
try {
const declarationChunks = await declarationChunk(
label,
try {
if (rule.mode === CHUNK_MODE_AST_FUNCTION) {
const astChunks = await astChunk(
content,
filePath,
startLine,
endLine,
chunkSize,
overlap,
rule,
);
if (astChunks.length > 0) return astChunks;
}
if (rule.mode === CHUNK_MODE_AST_DECLARATION) {
const declarationChunks = await declarationChunk(
content,
filePath,
startLine,
endLine,
chunkSize,
overlap,
rule,
);
if (declarationChunks.length > 0) return declarationChunks;
} catch {
// AST parsing failed — fall through to character fallback
}
} catch {
// AST parsing failed — fall through to character fallback
}
// Character-based fallback for everything else
@ -83,6 +97,7 @@ const astChunk = async (
endLine: number,
chunkSize: number,
overlap: number,
rule: ChunkingRule,
): Promise<Chunk[]> => {
const tree = await ensureAndParse(content, filePath);
if (!tree) return [];
@ -121,8 +136,8 @@ const astChunk = async (
statements,
targetNode.startIndex,
targetNode.endIndex,
true,
true,
rule.includePrefix,
rule.includeSuffix,
);
};
@ -145,13 +160,13 @@ const FIELD_LIKE_MEMBER_TYPES = new Set([
]);
const declarationChunk = async (
label: 'Class' | 'Interface',
content: string,
filePath: string,
startLine: number,
endLine: number,
chunkSize: number,
overlap: number,
rule: ChunkingRule,
): Promise<Chunk[]> => {
const tree = await ensureAndParse(content, filePath);
if (!tree) return [];
@ -162,7 +177,7 @@ const declarationChunk = async (
const bodyNode = getDeclarationBodyNode(targetNode);
if (!bodyNode) return [];
const members = collectDeclarationUnits(bodyNode, label);
const members = collectDeclarationUnits(bodyNode, rule.groupFields);
if (members.length === 0) return [];
return chunkByUnits(
@ -174,8 +189,8 @@ const declarationChunk = async (
members,
targetNode.startIndex,
targetNode.endIndex,
false,
false,
rule.includePrefix,
rule.includeSuffix,
);
};
@ -237,14 +252,22 @@ const chunkByUnits = (
if (candidateEndOffset - chunkStartOffset > chunkSize) {
const oversizedUnit = units[chunkStartUnitIdx];
const oversizedStartOffset =
chunkStartUnitIdx === 0 && includeContainerPrefixOnFirstChunk
? containerStartOffset
: oversizedUnit.startIndex;
const oversizedEndOffset =
chunkStartUnitIdx === units.length - 1 && includeContainerSuffixOnLastChunk
? containerEndOffset
: oversizedUnit.endIndex;
const oversizedLineRange = resolveChunkLines(
lineOffsets,
oversizedUnit.startIndex,
oversizedUnit.endIndex,
oversizedStartOffset,
oversizedEndOffset,
baseStartLine,
);
const oversizedChunks = characterChunk(
content.slice(oversizedUnit.startIndex, oversizedUnit.endIndex),
content.slice(oversizedStartOffset, oversizedEndOffset),
oversizedLineRange.startLine,
oversizedLineRange.endLine,
chunkSize,
@ -252,8 +275,8 @@ const chunkByUnits = (
).map((chunk, offsetIdx) => ({
...chunk,
chunkIndex: chunks.length + offsetIdx,
startOffset: chunk.startOffset + oversizedUnit.startIndex,
endOffset: chunk.endOffset + oversizedUnit.startIndex,
startOffset: chunk.startOffset + oversizedStartOffset,
endOffset: chunk.endOffset + oversizedStartOffset,
}));
chunks.push(...oversizedChunks);
chunkStartUnitIdx += 1;
@ -325,7 +348,7 @@ const getDeclarationBodyNode = (node: any): any | null => {
const collectDeclarationUnits = (
bodyNode: any,
label: 'Class' | 'Interface',
groupFields: boolean,
): Array<{ startIndex: number; endIndex: number }> => {
const members: Array<{ startIndex: number; endIndex: number; groupable: boolean }> = [];
@ -335,7 +358,7 @@ const collectDeclarationUnits = (
members.push({
startIndex: child.startIndex,
endIndex: child.endIndex,
groupable: label === 'Class' && FIELD_LIKE_MEMBER_TYPES.has(child.type),
groupable: groupFields && FIELD_LIKE_MEMBER_TYPES.has(child.type),
});
}

View file

@ -30,6 +30,7 @@ import {
DEFAULT_EMBEDDING_CONFIG,
EMBEDDABLE_LABELS,
isShortLabel,
LABEL_METHOD,
LABELS_WITH_EXPORTED,
STRUCTURAL_LABELS,
collectBestChunks,
@ -43,6 +44,12 @@ import {
import { loadVectorExtension } from '../lbug/lbug-adapter.js';
const isDev = process.env.NODE_ENV === 'development';
/**
* Bump this when the embedding text template changes in a way that should
* invalidate existing vectors, such as metadata/header shape changes,
* structural container context changes, or preceding-context formatting rules.
*/
export const EMBEDDING_TEXT_VERSION = 'v2';
/**
* Compute a stable content fingerprint for an embeddable node.
@ -57,12 +64,13 @@ export const contentHashForNode = (
// Hash must be deterministic across runs, so exclude methodNames/fieldNames
// which are populated during the batch loop via AST extraction.
// Using only node.content ensures the hash stays stable.
// NOTE: A change to extractStructuralNames behavior requires bumping EMBEDDING_TEXT_VERSION.
const text = generateEmbeddingText(
{ ...node, methodNames: undefined, fieldNames: undefined },
node.content,
config,
);
return createHash('sha1').update(text).digest('hex');
return createHash('sha1').update(EMBEDDING_TEXT_VERSION).update('\n').update(text).digest('hex');
};
/**
@ -83,7 +91,7 @@ const queryEmbeddableNodes = async (
try {
let query: string;
if (label === 'Method') {
if (label === LABEL_METHOD) {
// Method has parameterCount and returnType
query = `
MATCH (n:Method)
@ -115,7 +123,7 @@ const queryEmbeddableNodes = async (
const rows = await executeQuery(query);
for (const row of rows) {
const hasExportedColumn = label === 'Method' || LABELS_WITH_EXPORTED.has(label);
const hasExportedColumn = label === LABEL_METHOD || LABELS_WITH_EXPORTED.has(label);
allNodes.push({
id: row.id ?? row[0],
name: row.name ?? row[1],
@ -126,7 +134,7 @@ const queryEmbeddableNodes = async (
endLine: row.endLine ?? row[6],
isExported: hasExportedColumn ? (row.isExported ?? row[7]) : undefined,
description: row.description ?? (hasExportedColumn ? row[8] : row[7]),
...(label === 'Method'
...(label === LABEL_METHOD
? {
parameterCount: row.parameterCount ?? row[9],
returnType: row.returnType ?? row[10],
@ -415,8 +423,15 @@ export const runEmbeddingPipeline = async (
}
}
let prevTail = '';
for (const chunk of chunks) {
const text = generateEmbeddingText(node, chunk.text, finalConfig);
const text = generateEmbeddingText(
node,
chunk.text,
finalConfig,
chunk.chunkIndex,
prevTail,
);
allTexts.push(text);
allUpdates.push({
nodeId: node.id,
@ -425,6 +440,7 @@ export const runEmbeddingPipeline = async (
endLine: chunk.endLine,
contentHash: hash,
});
prevTail = overlap > 0 ? chunk.text.slice(-overlap) : '';
}
}

View file

@ -10,7 +10,12 @@
*/
import type { EmbeddableNode, EmbeddingConfig } from './types.js';
import { DEFAULT_EMBEDDING_CONFIG, isShortLabel } from './types.js';
import {
CHUNKING_RULES,
DEFAULT_EMBEDDING_CONFIG,
STRUCTURAL_TEXT_MODE_DECLARATION,
isShortLabel,
} from './types.js';
/**
* Truncate description to max length at sentence/word boundary
@ -95,47 +100,62 @@ const generateCodeBodyText = (
node: EmbeddableNode,
codeBody: string,
config: Partial<EmbeddingConfig>,
prevTail?: string,
): string => {
const header = buildMetadataHeader(node, config);
const cleaned = cleanContent(codeBody);
return `${header}\n\n${cleaned}`;
const parts = [header];
if (prevTail) {
parts.push(`[preceding context]: ...${cleanContent(prevTail)}`);
}
parts.push('', cleanContent(codeBody));
return parts.join('\n');
};
/**
* Generate embedding text for Class nodes
* Signature + properties + method name list only (no method bodies)
* Method/field names come from AST extractors via node.methodNames/node.fieldNames.
*/
const generateClassText = (
node: EmbeddableNode,
codeBody: string,
config: Partial<EmbeddingConfig>,
): string => {
return generateStructuralTypeText(node, codeBody, config);
const getCompactContainerContext = (
cleanedContent: string,
declarationOnly: string,
): string | undefined => {
const source = declarationOnly || cleanedContent;
const nlIdx = source.indexOf('\n');
const firstLine = (nlIdx === -1 ? source : source.substring(0, nlIdx)).trim();
return firstLine ? `Container: ${firstLine}` : undefined;
};
const generateStructuralTypeText = (
node: EmbeddableNode,
codeBody: string,
config: Partial<EmbeddingConfig>,
chunkIndex?: number,
prevTail?: string,
): string => {
const header = buildMetadataHeader(node, config);
const parts: string[] = [header];
const isFirstChunk = chunkIndex === undefined || chunkIndex === 0;
const cleanedContent = cleanContent(node.content);
const declarationOnly = extractDeclarationOnly(cleanedContent);
const compactContainerContext = getCompactContainerContext(cleanedContent, declarationOnly);
if (node.methodNames?.length) {
if (compactContainerContext) {
parts.push(compactContainerContext);
}
if (prevTail) {
parts.push(`[preceding context]: ...${cleanContent(prevTail)}`);
}
if (isFirstChunk && node.methodNames?.length) {
parts.push(`Methods: ${node.methodNames.join(', ')}`);
}
if (node.fieldNames?.length) {
if (isFirstChunk && node.fieldNames?.length) {
parts.push(`Properties: ${node.fieldNames.join(', ')}`);
}
const declarationOnly = extractDeclarationOnly(cleanContent(node.content));
if (declarationOnly) {
if (isFirstChunk && declarationOnly) {
parts.push('', declarationOnly);
}
const cleanedChunk = cleanContent(codeBody);
if (cleanedChunk && cleanedChunk !== cleanContent(node.content)) {
if (cleanedChunk && cleanedChunk !== cleanedContent) {
parts.push('', cleanedChunk);
}
@ -229,6 +249,8 @@ export const generateEmbeddingText = (
node: EmbeddableNode,
codeBody: string,
config: Partial<EmbeddingConfig> = {},
chunkIndex?: number,
prevTail?: string,
): string => {
if (isShortLabel(node.label)) {
const header = buildMetadataHeader(node, config);
@ -236,15 +258,12 @@ export const generateEmbeddingText = (
return `${header}\n\n${cleaned}`;
}
if (node.label === 'Class') {
return generateClassText(node, codeBody, config);
const chunkingRule = CHUNKING_RULES[node.label];
if (chunkingRule?.structuralTextMode === STRUCTURAL_TEXT_MODE_DECLARATION) {
return generateStructuralTypeText(node, codeBody, config, chunkIndex, prevTail);
}
if (node.label === 'Interface') {
return generateStructuralTypeText(node, codeBody, config);
}
return generateCodeBodyText(node, codeBody, config);
return generateCodeBodyText(node, codeBody, config, prevTail);
};
/**

View file

@ -4,35 +4,76 @@
* Type definitions for the embedding generation and semantic search system.
*/
export const LABEL_FUNCTION = 'Function' as const;
export const LABEL_METHOD = 'Method' as const;
export const LABEL_CONSTRUCTOR = 'Constructor' as const;
export const LABEL_CLASS = 'Class' as const;
export const LABEL_INTERFACE = 'Interface' as const;
export const LABEL_STRUCT = 'Struct' as const;
export const LABEL_ENUM = 'Enum' as const;
export const LABEL_TRAIT = 'Trait' as const;
export const LABEL_IMPL = 'Impl' as const;
export const LABEL_MACRO = 'Macro' as const;
export const LABEL_NAMESPACE = 'Namespace' as const;
export const LABEL_TYPE_ALIAS = 'TypeAlias' as const;
export const LABEL_TYPEDEF = 'Typedef' as const;
export const LABEL_CONST = 'Const' as const;
export const LABEL_PROPERTY = 'Property' as const;
export const LABEL_RECORD = 'Record' as const;
export const LABEL_UNION = 'Union' as const;
export const LABEL_STATIC = 'Static' as const;
export const LABEL_VARIABLE = 'Variable' as const;
export const LABEL_CODE_ELEMENT = 'CodeElement' as const;
export const CHUNK_MODE_AST_FUNCTION = 'ast-function' as const;
export const CHUNK_MODE_AST_DECLARATION = 'ast-declaration' as const;
// CHUNK_MODE_CHARACTER exists for type completeness but is a no-op in CHUNKING_RULES —
// omit the entry entirely to get character fallback via chunker.ts dispatch.
export const CHUNK_MODE_CHARACTER = 'character' as const;
export const STRUCTURAL_TEXT_MODE_NONE = 'none' as const;
export const STRUCTURAL_TEXT_MODE_DECLARATION = 'declaration' as const;
export interface ChunkingRule {
mode:
| typeof CHUNK_MODE_AST_FUNCTION
| typeof CHUNK_MODE_AST_DECLARATION
| typeof CHUNK_MODE_CHARACTER;
includePrefix: boolean;
includeSuffix: boolean;
groupFields: boolean;
structuralTextMode: typeof STRUCTURAL_TEXT_MODE_NONE | typeof STRUCTURAL_TEXT_MODE_DECLARATION;
}
/**
* Node labels that need chunking (have code body, potentially long)
*/
export const CHUNKABLE_LABELS = [
'Function',
'Method',
'Constructor',
'Class',
'Interface',
'Struct',
'Enum',
'Trait',
'Impl',
'Macro',
'Namespace',
LABEL_FUNCTION,
LABEL_METHOD,
LABEL_CONSTRUCTOR,
LABEL_CLASS,
LABEL_INTERFACE,
LABEL_STRUCT,
LABEL_ENUM,
LABEL_TRAIT,
LABEL_IMPL,
LABEL_MACRO,
LABEL_NAMESPACE,
] as const;
/**
* Node labels that are short (no chunking needed, embed directly)
*/
export const SHORT_LABELS = [
'TypeAlias',
'Typedef',
'Const',
'Property',
'Record',
'Union',
'Static',
'Variable',
LABEL_TYPE_ALIAS,
LABEL_TYPEDEF,
LABEL_CONST,
LABEL_PROPERTY,
LABEL_RECORD,
LABEL_UNION,
LABEL_STATIC,
LABEL_VARIABLE,
] as const;
/**
@ -61,26 +102,78 @@ export const isShortLabel = (label: string): boolean =>
(SHORT_LABELS as readonly string[]).includes(label);
/**
* Node labels that have structural names (methods/fields) extractable via AST
* Node labels that have structural names (methods/fields) extractable via AST.
* Only labels that consume methodNames/fieldNames in their embedding text should
* be listed here — extra entries trigger wasted AST parses with no effect on output.
*/
export const STRUCTURAL_LABELS: ReadonlySet<string> = new Set([
'Class',
'Struct',
'Interface',
'Enum',
LABEL_CLASS,
LABEL_STRUCT,
LABEL_INTERFACE,
]);
/**
* Node labels that have isExported column in their schema
*/
export const LABELS_WITH_EXPORTED = new Set([
'Function',
'Class',
'Interface',
'Method',
'CodeElement',
LABEL_FUNCTION,
LABEL_CLASS,
LABEL_INTERFACE,
LABEL_METHOD,
LABEL_CODE_ELEMENT,
]) as ReadonlySet<string>;
/**
* Labels that need special chunking and/or structural text semantics.
* Any chunkable label omitted here intentionally falls back to characterChunk
* plus generateCodeBodyText (for example Enum/Trait/Impl/Macro/Namespace).
*/
type ChunkableLabel = (typeof CHUNKABLE_LABELS)[number];
export const CHUNKING_RULES: Readonly<Partial<Record<ChunkableLabel, ChunkingRule>>> = {
[LABEL_FUNCTION]: {
mode: CHUNK_MODE_AST_FUNCTION,
includePrefix: true,
includeSuffix: true,
groupFields: false,
structuralTextMode: STRUCTURAL_TEXT_MODE_NONE,
},
[LABEL_METHOD]: {
mode: CHUNK_MODE_AST_FUNCTION,
includePrefix: true,
includeSuffix: true,
groupFields: false,
structuralTextMode: STRUCTURAL_TEXT_MODE_NONE,
},
[LABEL_CONSTRUCTOR]: {
mode: CHUNK_MODE_AST_FUNCTION,
includePrefix: true,
includeSuffix: true,
groupFields: false,
structuralTextMode: STRUCTURAL_TEXT_MODE_NONE,
},
[LABEL_CLASS]: {
mode: CHUNK_MODE_AST_DECLARATION,
includePrefix: true,
includeSuffix: false,
groupFields: true,
structuralTextMode: STRUCTURAL_TEXT_MODE_DECLARATION,
},
[LABEL_INTERFACE]: {
mode: CHUNK_MODE_AST_DECLARATION,
includePrefix: true,
includeSuffix: false,
groupFields: false,
structuralTextMode: STRUCTURAL_TEXT_MODE_DECLARATION,
},
[LABEL_STRUCT]: {
mode: CHUNK_MODE_AST_DECLARATION,
includePrefix: true,
includeSuffix: false,
groupFields: true,
structuralTextMode: STRUCTURAL_TEXT_MODE_DECLARATION,
},
};
/**
* Embedding pipeline phases
*/

View file

@ -0,0 +1,551 @@
/**
* Cross-repo impact (Phase 1 local walk + Phase 2 bridge fan-out).
* All bridge Cypher for this feature lives in this module.
*/
import fsp from 'node:fs/promises';
import path from 'node:path';
import type {
BridgeHandle,
ContractType,
CrossRepoImpact,
GroupConfig,
GroupImpactResult,
MatchType,
OutOfScopeLink,
} from './types.js';
import type { GroupRepoHandle, GroupToolPort } from './service.js';
import { loadGroupConfig } from './config-parser.js';
import {
fileMatchesServicePrefix,
normalizeServicePrefix,
repoInSubgroup,
} from './group-path-utils.js';
import { getGroupDir } from './storage.js';
import { closeBridgeDb, openBridgeDbReadOnly, queryBridge, readBridgeMeta } from './bridge-db.js';
import { BRIDGE_SCHEMA_VERSION } from './bridge-schema.js';
/** Cross-boundary hops beyond this value are clamped (multi-hop reserved for future work). */
export const MAX_SUPPORTED_CROSS_DEPTH = 1;
/** Default wall-clock budget for the Phase 1 `impact` leg when callers omit `timeoutMs`. */
export const DEFAULT_LOCAL_IMPACT_TIMEOUT_MS = 30_000;
const CY_NEIGHBORS_UPSTREAM = `
MATCH (consumer:Contract)-[l:ContractLink]->(provider:Contract)
WHERE provider.repo = $localRepo
AND provider.symbolUid IN $uids
AND provider.role = 'provider'
RETURN consumer.repo AS neighborRepo,
consumer.symbolUid AS neighborUid,
consumer.filePath AS neighborFilePath,
l.matchType AS matchType,
l.confidence AS confidence,
l.contractId AS contractId,
consumer.type AS contractType
`;
const CY_NEIGHBORS_DOWNSTREAM = `
MATCH (consumer:Contract)-[l:ContractLink]->(provider:Contract)
WHERE consumer.repo = $localRepo
AND consumer.symbolUid IN $uids
AND consumer.role = 'consumer'
RETURN provider.repo AS neighborRepo,
provider.symbolUid AS neighborUid,
provider.filePath AS neighborFilePath,
l.matchType AS matchType,
l.confidence AS confidence,
l.contractId AS contractId,
provider.type AS contractType
`;
type BridgeNeighborRow = {
neighborRepo: string;
neighborUid: string;
neighborFilePath?: string;
matchType: string;
confidence: number;
contractId: string;
contractType: string;
};
export interface RunGroupImpactDeps {
port: GroupToolPort;
gitnexusDir: string;
}
function parseDirection(raw: unknown): 'upstream' | 'downstream' | null {
if (raw === 'upstream' || raw === 'downstream') return raw;
return null;
}
function clampCrossDepth(raw: unknown): { depth: number; warning?: string } {
const n = typeof raw === 'number' && Number.isFinite(raw) ? Math.floor(raw) : 1;
const d = n < 1 ? 1 : n;
if (d > MAX_SUPPORTED_CROSS_DEPTH) {
return {
depth: MAX_SUPPORTED_CROSS_DEPTH,
warning: `crossDepth was ${d}; multi-hop cross-boundary traversal beyond ${MAX_SUPPORTED_CROSS_DEPTH} is not implemented yet. Using crossDepth ${MAX_SUPPORTED_CROSS_DEPTH}.`,
};
}
return { depth: d };
}
export function validateGroupImpactParams(params: Record<string, unknown>):
| {
ok: true;
name: string;
repoPath: string;
target: string;
direction: 'upstream' | 'downstream';
maxDepth: number;
crossDepth: number;
crossDepthWarning?: string;
relationTypes?: string[];
includeTests: boolean;
minConfidence: number;
service?: string;
subgroup?: string;
timeoutMs: number;
}
| { ok: false; error: string } {
const name = String(params.name ?? '').trim();
const repoPath = String(params.repo ?? '').trim();
const target = String(params.target ?? '').trim();
if (!name) return { ok: false, error: 'name is required' };
if (!repoPath)
return { ok: false, error: 'repo is required (group repo path, e.g. app/backend)' };
if (!target) return { ok: false, error: 'target is required' };
if (
params.service !== undefined &&
params.service !== null &&
String(params.service).trim() === ''
) {
return { ok: false, error: 'service must not be an empty string' };
}
const direction = parseDirection(params.direction);
if (!direction) return { ok: false, error: 'direction must be upstream or downstream' };
let maxDepth = typeof params.maxDepth === 'number' && params.maxDepth > 0 ? params.maxDepth : 3;
if (maxDepth > 32) maxDepth = 32;
const { depth: crossDepth, warning: crossDepthWarning } = clampCrossDepth(params.crossDepth);
const relationTypes = Array.isArray(params.relationTypes)
? params.relationTypes.filter((t): t is string => typeof t === 'string')
: undefined;
const includeTests = Boolean(params.includeTests);
let minConfidence = typeof params.minConfidence === 'number' ? params.minConfidence : 0;
if (minConfidence < 0) minConfidence = 0;
if (minConfidence > 1) minConfidence = 1;
const service = normalizeServicePrefix(params.service);
const subgroup = typeof params.subgroup === 'string' ? params.subgroup : undefined;
let timeoutMs =
typeof params.timeoutMs === 'number' && params.timeoutMs > 0
? params.timeoutMs
: typeof params.timeout === 'number' && params.timeout > 0
? params.timeout
: DEFAULT_LOCAL_IMPACT_TIMEOUT_MS;
if (timeoutMs > 3_600_000) timeoutMs = 3_600_000;
return {
ok: true,
name,
repoPath,
target,
direction,
maxDepth,
crossDepth,
crossDepthWarning,
relationTypes,
includeTests,
minConfidence,
service,
subgroup,
timeoutMs,
};
}
async function resolveGroupRepo(
port: GroupToolPort,
config: GroupConfig,
repoPath: string,
): Promise<GroupRepoHandle | { error: string }> {
const registryName = config.repos[repoPath];
if (!registryName) {
return { error: `Unknown repo path "${repoPath}" in this group.` };
}
try {
return await port.resolveRepo(registryName);
} catch (e) {
return { error: e instanceof Error ? e.message : String(e) };
}
}
async function safeLocalImpact(
port: GroupToolPort,
repo: GroupRepoHandle,
impactParams: Parameters<GroupToolPort['impact']>[1],
timeoutMs: number,
): Promise<{ value: unknown; timedOut: boolean }> {
let timer: ReturnType<typeof setTimeout> | undefined;
const impactP = port.impact(repo, impactParams).catch((err) => ({
error: err instanceof Error ? err.message : String(err),
}));
const timeoutP = new Promise<'timeout'>((resolve) => {
timer = setTimeout(() => resolve('timeout'), timeoutMs);
});
const won = await Promise.race([
impactP.then((v) => ({ tag: 'impact' as const, v })),
timeoutP.then(() => ({ tag: 'timeout' as const })),
]);
if (timer !== undefined) clearTimeout(timer);
if (won.tag === 'timeout') {
return {
value: { error: 'Local impact timed out', partial: true },
timedOut: true,
};
}
return { value: won.v, timedOut: false };
}
export function collectImpactSymbolUids(
local: unknown,
servicePrefix: string | undefined,
): { uids: string[]; targetFilePath?: string } {
const uids = new Set<string>();
let targetFilePath: string | undefined;
const obj = local as Record<string, unknown> | null;
if (!obj || typeof obj !== 'object') return { uids: [], targetFilePath };
const target = obj.target as { id?: string; filePath?: string } | undefined;
if (target?.id) {
targetFilePath = typeof target.filePath === 'string' ? target.filePath : undefined;
if (fileMatchesServicePrefix(targetFilePath, servicePrefix)) {
uids.add(String(target.id));
}
}
const byDepth = obj.byDepth as Record<string | number, unknown> | undefined;
if (byDepth && typeof byDepth === 'object') {
for (const items of Object.values(byDepth)) {
if (!Array.isArray(items)) continue;
for (const it of items) {
const row = it as { id?: string; filePath?: string };
if (row?.id && fileMatchesServicePrefix(row.filePath, servicePrefix)) {
uids.add(String(row.id));
}
}
}
}
return { uids: [...uids], targetFilePath };
}
function extractProcessNames(impact: unknown): string[] {
const o = impact as { affected_processes?: Array<{ name?: string }> };
if (!o?.affected_processes) return [];
return o.affected_processes.map((p) => String(p.name ?? '')).filter(Boolean);
}
function mergeRisk(localRisk: string, cross: CrossRepoImpact[]): string {
const highConf = cross.some((c) => c.contract.confidence >= 0.85);
if (localRisk === 'CRITICAL') return 'CRITICAL';
if (cross.length >= 3) return 'CRITICAL';
if (highConf) return 'HIGH';
if (cross.length > 0 && (localRisk === 'LOW' || localRisk === 'UNKNOWN')) return 'MEDIUM';
return localRisk;
}
async function ensureBridgeReady(
groupDir: string,
): Promise<{ handle: BridgeHandle } | { error: string }> {
const meta = await readBridgeMeta(groupDir);
if (meta.version > 0 && meta.version !== BRIDGE_SCHEMA_VERSION) {
return {
error: `Bridge schema version mismatch (meta.json has ${meta.version}, expected ${BRIDGE_SCHEMA_VERSION}). Run gitnexus group sync for this group.`,
};
}
const dbPath = path.join(groupDir, 'bridge.lbug');
try {
await fsp.access(dbPath);
} catch {
return {
error: `No bridge.lbug in this group directory. Run gitnexus group sync (schema ${BRIDGE_SCHEMA_VERSION}).`,
};
}
const handle = await openBridgeDbReadOnly(groupDir);
if (!handle) {
return {
error: `Could not open bridge.lbug read-only (schema ${BRIDGE_SCHEMA_VERSION}). Run gitnexus group sync.`,
};
}
return { handle };
}
function rowToNeighbor(r: Record<string, unknown>): BridgeNeighborRow | null {
const neighborRepo = String(r.neighborRepo ?? r[0] ?? '');
const neighborUid = String(r.neighborUid ?? r[1] ?? '');
if (!neighborRepo || !neighborUid) return null;
return {
neighborRepo,
neighborUid,
neighborFilePath:
r.neighborFilePath !== undefined ? String(r.neighborFilePath) : String(r[2] ?? ''),
matchType: String(r.matchType ?? r[3] ?? 'exact'),
confidence: Number(r.confidence ?? r[4] ?? 0),
contractId: String(r.contractId ?? r[5] ?? ''),
contractType: String(r.contractType ?? r[6] ?? 'custom'),
};
}
export async function runGroupImpact(
deps: RunGroupImpactDeps,
params: Record<string, unknown>,
): Promise<GroupImpactResult | { error: string }> {
const parsed = validateGroupImpactParams(params);
if (parsed.ok === false) return { error: parsed.error };
const {
name,
repoPath,
target,
direction,
maxDepth,
crossDepth: _crossDepth,
crossDepthWarning,
relationTypes,
includeTests,
minConfidence,
service: servicePrefix,
subgroup,
timeoutMs,
} = parsed;
const groupDir = getGroupDir(deps.gitnexusDir, name);
let config: GroupConfig;
try {
config = await loadGroupConfig(groupDir);
} catch (e) {
return { error: e instanceof Error ? e.message : String(e) };
}
const resolved = await resolveGroupRepo(deps.port, config, repoPath);
if ('error' in resolved) return { error: resolved.error };
const impactParams: Parameters<GroupToolPort['impact']>[1] = {
target,
direction,
maxDepth,
relationTypes: relationTypes && relationTypes.length > 0 ? relationTypes : undefined,
includeTests,
minConfidence,
};
// Single shared deadline for Phase 1 (local walk) + Phase 2 (bridge fan-out).
// Phase 1 still gets the full budget; Phase 2 only uses whatever wall-clock
// time is left, so total work cannot exceed `timeoutMs`.
const deadline = Date.now() + Math.max(0, timeoutMs);
const { value: local, timedOut: localTimedOut } = await safeLocalImpact(
deps.port,
resolved,
impactParams,
timeoutMs,
);
if (localTimedOut) {
const base = local as Record<string, unknown>;
return {
local,
group: name,
cross: [],
outOfScope: [],
truncated: true,
truncatedRepos: [],
summary: {
direct: 0,
processes_affected: 0,
modules_affected: 0,
cross_repo_hits: 0,
},
risk: 'UNKNOWN',
timeoutMs,
truncationReason: 'timeout',
crossDepthWarning,
};
}
const localObj = local as Record<string, unknown> | null;
if (localObj?.error && typeof localObj.error === 'string') {
// Fail closed: the local-impact phase errored (missing symbol, graph-load
// failure, thrown exception wrapped by safeLocalImpact, or port-returned
// `{ error }`). Do NOT wrap it into a zero-hit success payload — callers
// branch on top-level `error`, and a blast-radius tool reporting "no
// impact" on the failure path is a false negative on a safety-critical
// signal. Bubble the error so consumers treat it as a failure.
return { error: `Local impact failed for ${repoPath}: ${localObj.error}` };
}
if (servicePrefix) {
const tf = (localObj?.target as { filePath?: string } | undefined)?.filePath;
if (!fileMatchesServicePrefix(tf, servicePrefix)) {
return {
local: {},
group: name,
cross: [],
outOfScope: [],
truncated: false,
truncatedRepos: [],
summary: {
direct: 0,
processes_affected: 0,
modules_affected: 0,
cross_repo_hits: 0,
},
risk: 'LOW',
timeoutMs,
crossDepthWarning,
};
}
}
const { uids } = collectImpactSymbolUids(local, servicePrefix);
if (uids.length === 0) {
const s = (local as { summary?: Record<string, number> })?.summary || {};
return {
local,
group: name,
cross: [],
outOfScope: [],
truncated: Boolean((local as { partial?: boolean }).partial),
truncatedRepos: [],
summary: {
direct: s.direct ?? 0,
processes_affected: s.processes_affected ?? 0,
modules_affected: s.modules_affected ?? 0,
cross_repo_hits: 0,
},
risk: String((local as { risk?: string }).risk ?? 'LOW'),
timeoutMs,
truncationReason: (local as { partial?: boolean }).partial ? 'partial' : undefined,
crossDepthWarning,
};
}
const bridgePrep = await ensureBridgeReady(groupDir);
if ('error' in bridgePrep) return { error: bridgePrep.error };
const handle = bridgePrep.handle;
const cross: CrossRepoImpact[] = [];
const outOfScope: OutOfScopeLink[] = [];
const truncatedRepos: string[] = [];
try {
const cypher = direction === 'upstream' ? CY_NEIGHBORS_UPSTREAM : CY_NEIGHBORS_DOWNSTREAM;
const rows = await queryBridge<Record<string, unknown>>(handle, cypher, {
localRepo: repoPath,
uids,
});
const neighbors: BridgeNeighborRow[] = [];
for (const raw of rows) {
const n = rowToNeighbor(raw);
if (n) neighbors.push(n);
}
neighbors.sort((a, b) => b.confidence - a.confidence);
const seen = new Set<string>();
for (const n of neighbors) {
if (servicePrefix && !fileMatchesServicePrefix(n.neighborFilePath, servicePrefix)) {
continue;
}
if (!repoInSubgroup(n.neighborRepo, subgroup)) {
// CrossLink convention: consumer -> provider
outOfScope.push({
from: direction === 'upstream' ? n.neighborRepo : repoPath,
to: direction === 'upstream' ? repoPath : n.neighborRepo,
contractId: n.contractId,
confidence: n.confidence,
});
continue;
}
const key = `${n.neighborRepo}\0${n.neighborUid}\0${n.contractId}`;
if (seen.has(key)) continue;
seen.add(key);
if (Date.now() > deadline) {
truncatedRepos.push(n.neighborRepo);
continue;
}
const regName = config.repos[n.neighborRepo];
if (!regName) continue;
let neighborHandle: GroupRepoHandle;
try {
neighborHandle = await deps.port.resolveRepo(regName);
} catch {
truncatedRepos.push(n.neighborRepo);
continue;
}
const fan = await deps.port.impactByUid(neighborHandle.id, n.neighborUid, direction, {
maxDepth,
relationTypes: relationTypes ?? [],
minConfidence,
includeTests,
});
if (fan == null) {
truncatedRepos.push(n.neighborRepo);
continue;
}
cross.push({
repo: regName,
repo_path: n.neighborRepo,
contract: {
id: n.contractId,
type: n.contractType as ContractType,
match_type: (n.matchType as MatchType) || 'exact',
confidence: n.confidence,
},
by_depth: ((fan as { byDepth?: unknown }).byDepth ?? {}) as Record<string, unknown[]>,
affected_processes: extractProcessNames(fan),
});
}
} finally {
await closeBridgeDb(handle);
}
const localSum = (local as { summary?: Record<string, number> })?.summary || {};
const localRisk = String((local as { risk?: string }).risk ?? 'LOW');
const localPartial = Boolean((local as { partial?: boolean }).partial);
const truncated = truncatedRepos.length > 0 || localPartial;
const result: GroupImpactResult = {
local,
group: name,
cross,
outOfScope,
truncated,
truncatedRepos: [...new Set(truncatedRepos)],
summary: {
direct: localSum.direct ?? 0,
processes_affected: localSum.processes_affected ?? 0,
modules_affected: localSum.modules_affected ?? 0,
cross_repo_hits: cross.length,
},
risk: mergeRisk(localRisk, cross),
timeoutMs,
truncationReason: truncated ? 'partial' : undefined,
crossDepthWarning,
};
return result;
}
export { normalizeServicePrefix, fileMatchesServicePrefix } from './group-path-utils.js';

View file

@ -3,33 +3,92 @@ import {
compilePatterns,
runCompiledPatterns,
unquoteLiteral,
type CompiledPatterns,
type LanguagePatterns,
type PatternSpec,
} from '../tree-sitter-scanner.js';
import type { HttpDetection, HttpLanguagePlugin } from './types.js';
/**
* PHP HTTP plugin — Laravel `Route::get/post/...` declarations.
* PHP HTTP plugin.
*
* Providers:
* - Laravel `Route::get/post/...`
*
* Consumers (string-literal URLs only):
* - Laravel HTTP client: `Http::get/post/put/delete/patch($url)`
* - Guzzle / generic object method: `$client->get/post/...($url)`
* - `file_get_contents($url)`
*
* The pipeline already uses `PHP.php_only` for ingesting plain `.php`
* files (see `core/tree-sitter/parser-loader.ts`), and we do the same
* here so Laravel route files are parsed with the right grammar dialect.
*
* Scope notes: consumer patterns match string literals only. URLs built
* via binary concatenation (`$base . '/path'`), `sprintf`, or config
* lookup (`config('services.foo.base').'/path'`) are intentionally left
* for a follow-up — they require constant-folding the surrounding
* scope to be meaningful.
*/
const LARAVEL_PATTERNS = compilePatterns({
name: 'php-laravel',
language: PHP.php_only,
patterns: [
{
meta: {},
query: `
(scoped_call_expression
scope: (name) @scope (#eq? @scope "Route")
name: (name) @method (#match? @method "^(get|post|put|delete|patch)$")
arguments: (arguments . (argument (string) @path)))
`,
},
],
} satisfies LanguagePatterns<Record<string, never>>);
const LARAVEL_ROUTE_SPEC: PatternSpec<Record<string, never>> = {
meta: {},
query: `
(scoped_call_expression
scope: (name) @scope (#eq? @scope "Route")
name: (name) @method (#match? @method "^(get|post|put|delete|patch)$")
arguments: (arguments . (argument (string) @path)))
`,
};
const HTTP_FACADE_SPEC: PatternSpec<Record<string, never>> = {
meta: {},
query: `
(scoped_call_expression
scope: (name) @scope (#eq? @scope "Http")
name: (name) @method (#match? @method "^(get|post|put|delete|patch)$")
arguments: (arguments . (argument (string) @path)))
`,
};
const GUZZLE_MEMBER_SPEC: PatternSpec<Record<string, never>> = {
meta: {},
query: `
(member_call_expression
name: (name) @method (#match? @method "^(get|post|put|delete|patch)$")
arguments: (arguments . (argument (string) @path)))
`,
};
const FILE_GET_CONTENTS_SPEC: PatternSpec<Record<string, never>> = {
meta: {},
query: `
(function_call_expression
function: (name) @fn (#eq? @fn "file_get_contents")
arguments: (arguments . (argument (string) @path)))
`,
};
interface PhpPatternBundle {
laravelRoute: CompiledPatterns<Record<string, never>>;
httpFacade: CompiledPatterns<Record<string, never>>;
guzzleMember: CompiledPatterns<Record<string, never>>;
fileGetContents: CompiledPatterns<Record<string, never>>;
}
const mk = (spec: PatternSpec<Record<string, never>>, suffix: string) =>
compilePatterns({
name: `php-${suffix}`,
language: PHP.php_only,
patterns: [spec],
} satisfies LanguagePatterns<Record<string, never>>);
const PHP_PATTERNS: PhpPatternBundle = {
laravelRoute: mk(LARAVEL_ROUTE_SPEC, 'laravel-route'),
httpFacade: mk(HTTP_FACADE_SPEC, 'http-facade'),
guzzleMember: mk(GUZZLE_MEMBER_SPEC, 'guzzle-member'),
fileGetContents: mk(FILE_GET_CONTENTS_SPEC, 'file-get-contents'),
};
/**
* Extract the inner text of a PHP `string` node. The tree-sitter-php
@ -39,11 +98,8 @@ const LARAVEL_PATTERNS = compilePatterns({
* child nodes.
*/
function phpStringText(node: import('tree-sitter').SyntaxNode): string | null {
// Most single-quoted strings expose their inner content through the
// full node text (including quotes), which unquoteLiteral strips.
const direct = unquoteLiteral(node.text);
if (direct !== null && direct !== node.text) return direct;
// Fall back to child string_content / string_value node if present.
for (const child of node.children) {
if (child.type === 'string_content' || child.type === 'string_value') {
return child.text;
@ -52,13 +108,32 @@ function phpStringText(node: import('tree-sitter').SyntaxNode): string | null {
return direct;
}
/**
* HTTP client helpers (`Http::`, Guzzle) are almost always called with
* a path relative to a configured base URL, or a full URL. File paths
* are rare. Accept both relative (`/api/...`) and absolute (`http(s)://`).
*/
function isHttpClientPath(path: string): boolean {
return path.startsWith('/') || path.startsWith('http://') || path.startsWith('https://');
}
/**
* `file_get_contents` is used for both HTTP and filesystem reads. Only
* emit a consumer contract when the URL is an absolute HTTP(S) URL to
* avoid false positives for local file paths and stream wrappers
* (`php://input`, `file://`, `data:`, ...).
*/
function isHttpUrlLiteral(path: string): boolean {
return path.startsWith('http://') || path.startsWith('https://');
}
export const PHP_HTTP_PLUGIN: HttpLanguagePlugin = {
name: 'php-http',
language: PHP.php_only,
scan(tree) {
const out: HttpDetection[] = [];
for (const match of runCompiledPatterns(LARAVEL_PATTERNS, tree)) {
for (const match of runCompiledPatterns(PHP_PATTERNS.laravelRoute, tree)) {
const methodNode = match.captures.method;
const pathNode = match.captures.path;
if (!methodNode || !pathNode) continue;
@ -74,6 +149,53 @@ export const PHP_HTTP_PLUGIN: HttpLanguagePlugin = {
});
}
for (const match of runCompiledPatterns(PHP_PATTERNS.httpFacade, tree)) {
const methodNode = match.captures.method;
const pathNode = match.captures.path;
if (!methodNode || !pathNode) continue;
const path = phpStringText(pathNode);
if (path === null || !isHttpClientPath(path)) continue;
out.push({
role: 'consumer',
framework: 'laravel-http',
method: methodNode.text.toUpperCase(),
path,
name: null,
confidence: 0.7,
});
}
for (const match of runCompiledPatterns(PHP_PATTERNS.guzzleMember, tree)) {
const methodNode = match.captures.method;
const pathNode = match.captures.path;
if (!methodNode || !pathNode) continue;
const path = phpStringText(pathNode);
if (path === null || !isHttpClientPath(path)) continue;
out.push({
role: 'consumer',
framework: 'guzzle',
method: methodNode.text.toUpperCase(),
path,
name: null,
confidence: 0.7,
});
}
for (const match of runCompiledPatterns(PHP_PATTERNS.fileGetContents, tree)) {
const pathNode = match.captures.path;
if (!pathNode) continue;
const path = phpStringText(pathNode);
if (path === null || !isHttpUrlLiteral(path)) continue;
out.push({
role: 'consumer',
framework: 'file-get-contents',
method: 'GET',
path,
name: null,
confidence: 0.7,
});
}
return out;
},
};

View file

@ -0,0 +1,42 @@
/**
* Shared service-path normalization for group tools (`service` monorepo filter)
* and subgroup membership checks.
*
* Inputs may originate from tree-sitter, the OS file API, or user-supplied
* MCP arguments, so both `\` and `/` separators are accepted. Internally we
* normalize to POSIX-style `/` for case-sensitive segment comparisons.
*/
function toPosix(p: string): string {
return p.replace(/\\/g, '/');
}
export function normalizeServicePrefix(service: unknown): string | undefined {
if (service === undefined || service === null) return undefined;
const s = toPosix(String(service)).trim().replace(/\/+$/, '');
return s.length > 0 ? s : undefined;
}
export function fileMatchesServicePrefix(
filePath: string | undefined,
prefix: string | undefined,
): boolean {
if (!prefix) return true;
if (!filePath) return false;
const normalized = toPosix(filePath);
return normalized === prefix || normalized.startsWith(`${prefix}/`);
}
/**
* True if `repoPath` is at or beneath `subgroup` (member-path prefix in
* `group.yaml`). Empty / missing `subgroup` matches every repo.
*
* @param exact When set, requires an exact equality match (no descendant repos).
*/
export function repoInSubgroup(repoPath: string, subgroup?: string, exact?: boolean): boolean {
if (!subgroup?.trim()) return true;
const s = toPosix(subgroup).replace(/\/+$/, '');
const r = toPosix(repoPath);
if (exact) return r === s;
return r === s || r.startsWith(`${s}/`);
}

View file

@ -0,0 +1,34 @@
/**
* Map MCP/CLI `@groupName` or `@groupName/memberPath` to a concrete member path in group.yaml.
*/
import { loadGroupConfig } from './config-parser.js';
import { getDefaultGitnexusDir, getGroupDir } from './storage.js';
export async function resolveAtGroupMemberRepoPath(
groupName: string,
explicitMemberPath: string | undefined,
): Promise<{ ok: true; repoPath: string } | { ok: false; error: string }> {
const trimmed = groupName.trim();
if (!trimmed) return { ok: false, error: 'Group name is empty.' };
try {
const groupDir = getGroupDir(getDefaultGitnexusDir(), trimmed);
const config = await loadGroupConfig(groupDir);
const keys = Object.keys(config.repos).sort((a, b) => a.localeCompare(b));
if (keys.length === 0) {
return { ok: false, error: `Group "${trimmed}" has no repos in group.yaml.` };
}
if (explicitMemberPath !== undefined && explicitMemberPath !== '') {
if (!(explicitMemberPath in config.repos)) {
return {
ok: false,
error: `Unknown member path "${explicitMemberPath}" in group "${trimmed}". Known paths: ${keys.join(', ')}`,
};
}
return { ok: true, repoPath: explicitMemberPath };
}
return { ok: true, repoPath: keys[0]! };
} catch (e) {
return { ok: false, error: e instanceof Error ? e.message : String(e) };
}
}

View file

@ -3,10 +3,24 @@
* DB access is injected via GroupToolPort so this module stays free of LocalBackend private API.
*/
import fsp from 'node:fs/promises';
import path from 'node:path';
import { checkStaleness } from '../git-staleness.js';
import { loadGroupConfig } from './config-parser.js';
import {
fileMatchesServicePrefix,
normalizeServicePrefix,
repoInSubgroup,
} from './group-path-utils.js';
import { getDefaultGitnexusDir, getGroupDir, listGroups, readContractRegistry } from './storage.js';
import { syncGroup } from './sync.js';
import type {
ContractRegistry,
CrossLink,
GroupConfig,
GroupContextResult,
StoredContract,
} from './types.js';
export interface GroupRepoHandle {
id: string;
@ -52,12 +66,149 @@ export interface GroupToolPort {
includeTests: boolean;
},
): Promise<unknown | null>;
context(
repo: GroupRepoHandle,
params: {
name?: string;
uid?: string;
file_path?: string;
include_content?: boolean;
},
): Promise<unknown>;
}
function repoInSubgroup(repoPath: string, subgroup?: string): boolean {
if (!subgroup?.trim()) return true;
const s = subgroup.replace(/\/+$/, '');
return repoPath === s || repoPath.startsWith(`${s}/`);
function isStoredContract(raw: unknown): raw is StoredContract {
if (!raw || typeof raw !== 'object') return false;
const o = raw as Record<string, unknown>;
return (
typeof o.contractId === 'string' &&
typeof o.type === 'string' &&
typeof o.repo === 'string' &&
typeof o.role === 'string' &&
(o.role === 'provider' || o.role === 'consumer') &&
typeof o.symbolUid === 'string' &&
typeof o.symbolName === 'string' &&
typeof o.confidence === 'number' &&
o.meta !== undefined &&
typeof o.meta === 'object' &&
o.meta !== null &&
o.symbolRef !== undefined &&
typeof o.symbolRef === 'object' &&
o.symbolRef !== null &&
typeof (o.symbolRef as Record<string, unknown>).filePath === 'string' &&
typeof (o.symbolRef as Record<string, unknown>).name === 'string'
);
}
function filterQueryByServicePrefix(
queryResult: {
processes?: Array<Record<string, unknown>>;
process_symbols?: Array<Record<string, unknown>>;
},
servicePrefix: string,
): { processes: Array<Record<string, unknown>>; process_symbols: Array<Record<string, unknown>> } {
const symbols = (queryResult.process_symbols || []).filter((s) =>
fileMatchesServicePrefix(
typeof s.filePath === 'string' ? s.filePath : undefined,
servicePrefix,
),
);
const allowed = new Set(
symbols.map((s) => String((s as { process_id?: string }).process_id ?? '')).filter(Boolean),
);
const processes = (queryResult.processes || []).filter((p) => allowed.has(String(p.id)));
return { processes, process_symbols: symbols };
}
function isCrossLink(raw: unknown): raw is CrossLink {
if (!raw || typeof raw !== 'object') return false;
const o = raw as Record<string, unknown>;
const from = o.from as Record<string, unknown> | undefined;
const to = o.to as Record<string, unknown> | undefined;
if (!from || !to) return false;
if (typeof from.repo !== 'string' || typeof to.repo !== 'string') return false;
return typeof o.contractId === 'string' && typeof o.type === 'string';
}
async function loadContractRegistryResilient(
groupDir: string,
): Promise<
{ ok: true; registry: ContractRegistry; skippedCorrupt: number } | { ok: false; error: string }
> {
const filePath = path.join(groupDir, 'contracts.json');
let raw: string;
try {
raw = await fsp.readFile(filePath, 'utf-8');
} catch (e) {
if ((e as NodeJS.ErrnoException).code === 'ENOENT') {
return { ok: false, error: `No contracts.json for this group. Run group_sync first.` };
}
return { ok: false, error: e instanceof Error ? e.message : String(e) };
}
let root: unknown;
try {
root = JSON.parse(raw);
} catch {
return { ok: false, error: 'contracts.json is not valid JSON' };
}
if (!root || typeof root !== 'object' || Array.isArray(root)) {
return { ok: false, error: 'contracts.json has an invalid root object' };
}
const base = root as Record<string, unknown>;
const contractsRaw = base.contracts;
const crossRaw = base.crossLinks;
let skippedCorrupt = 0;
const contracts: StoredContract[] = [];
if (Array.isArray(contractsRaw)) {
for (const row of contractsRaw) {
try {
if (isStoredContract(row)) {
contracts.push(row);
} else {
skippedCorrupt++;
console.warn('[group] skipping corrupt contract row in contracts.json');
}
} catch {
skippedCorrupt++;
console.warn('[group] skipping corrupt contract row in contracts.json');
}
}
}
const crossLinks: CrossLink[] = [];
if (Array.isArray(crossRaw)) {
for (const row of crossRaw) {
try {
if (isCrossLink(row)) {
crossLinks.push(row);
} else {
skippedCorrupt++;
console.warn('[group] skipping corrupt crossLinks row in contracts.json');
}
} catch {
skippedCorrupt++;
console.warn('[group] skipping corrupt crossLinks row in contracts.json');
}
}
}
const registry: ContractRegistry = {
version: typeof base.version === 'number' ? base.version : 0,
generatedAt: typeof base.generatedAt === 'string' ? base.generatedAt : '',
repoSnapshots:
base.repoSnapshots && typeof base.repoSnapshots === 'object' && base.repoSnapshots !== null
? (base.repoSnapshots as Record<string, { indexedAt: string; lastCommit: string }>)
: {},
missingRepos: Array.isArray(base.missingRepos) ? (base.missingRepos as string[]) : [],
contracts,
crossLinks,
};
return { ok: true, registry, skippedCorrupt };
}
export class GroupService {
@ -103,10 +254,14 @@ export class GroupService {
const name = String(params.name ?? '').trim();
if (!name) return { error: 'name is required' };
const groupDir = getGroupDir(getDefaultGitnexusDir(), name);
const registry = await readContractRegistry(groupDir);
if (!registry) {
return { error: `No contracts.json for group "${name}". Run group_sync first.` };
const loaded = await loadContractRegistryResilient(groupDir);
if (loaded.ok === false) {
if (loaded.error.includes('No contracts.json')) {
return { error: `No contracts.json for group "${name}". Run group_sync first.` };
}
return { error: loaded.error };
}
const { registry, skippedCorrupt } = loaded;
let contracts = registry.contracts;
if (params.type) contracts = contracts.filter((c) => c.type === params.type);
if (params.repo) contracts = contracts.filter((c) => c.repo === params.repo);
@ -119,41 +274,151 @@ export class GroupService {
);
contracts = contracts.filter((c) => !matchedIds.has(`${c.repo}::${c.contractId}`));
}
return { contracts, crossLinks: registry.crossLinks };
const out: Record<string, unknown> = { contracts, crossLinks: registry.crossLinks };
if (skippedCorrupt > 0) out.skippedCorrupt = skippedCorrupt;
return out;
}
async groupImpact(params: Record<string, unknown>): Promise<unknown> {
const { runGroupImpact } = await import('./cross-impact.js');
return runGroupImpact({ port: this.port, gitnexusDir: getDefaultGitnexusDir() }, params);
}
async groupContext(params: Record<string, unknown>): Promise<GroupContextResult> {
const name = String(params.name ?? '').trim();
const target = typeof params.target === 'string' ? params.target.trim() : '';
const uid = typeof params.uid === 'string' ? params.uid.trim() : undefined;
const file_path = typeof params.file_path === 'string' ? params.file_path : undefined;
const include_content = Boolean(params.include_content);
if (
params.service !== undefined &&
params.service !== null &&
String(params.service).trim() === ''
) {
return { group: name || '', error: 'service must not be an empty string', results: [] };
}
const servicePrefix = normalizeServicePrefix(params.service);
const subgroup = typeof params.subgroup === 'string' ? params.subgroup : undefined;
const subgroupExact = params.subgroupExact === true;
if (!name) {
return { group: '', error: 'name is required', results: [] };
}
if (!uid && !target) {
return { group: name, error: 'target or uid is required', results: [] };
}
const groupDir = getGroupDir(getDefaultGitnexusDir(), name);
let config: GroupConfig;
try {
config = await loadGroupConfig(groupDir);
} catch (e) {
return {
group: name,
target: target || uid,
service: servicePrefix,
error: e instanceof Error ? e.message : String(e),
results: [],
};
}
const memberEntries = Object.entries(config.repos).filter(([repoPath]) =>
repoInSubgroup(repoPath, subgroup, subgroupExact),
);
// Per-repo work is independent (each repo opens its own DB handle and the
// group-level result preserves repo iteration order via the indexed map).
// Errors are caught per repo so one slow/failed member does not block the rest.
const results: GroupContextResult['results'] = await Promise.all(
memberEntries.map(async ([repoPath, registryName]) => {
try {
const repoObj = await this.port.resolveRepo(registryName);
const payload = await this.port.context(repoObj, {
name: target || undefined,
uid,
file_path,
include_content,
});
if (servicePrefix) {
const st = (payload as { status?: string })?.status;
const sym = (payload as { symbol?: { filePath?: string } })?.symbol;
if (st === 'found' && !fileMatchesServicePrefix(sym?.filePath, servicePrefix)) {
return { repoPath, registryName, payload: {} };
}
}
return { repoPath, registryName, payload };
} catch (e) {
return {
repoPath,
registryName,
payload: { error: e instanceof Error ? e.message : String(e) },
};
}
}),
);
return {
group: name,
target: target || uid,
service: servicePrefix,
results,
};
}
async groupQuery(params: Record<string, unknown>): Promise<unknown> {
const name = String(params.name ?? '').trim();
const queryText = String(params.query ?? '').trim();
if (!name || !queryText) return { error: 'name and query are required' };
if (
params.service !== undefined &&
params.service !== null &&
String(params.service).trim() === ''
) {
return { error: 'service must not be an empty string' };
}
const servicePrefix = normalizeServicePrefix(params.service);
const limit = typeof params.limit === 'number' && params.limit > 0 ? params.limit : 5;
const subgroup = typeof params.subgroup === 'string' ? params.subgroup : undefined;
const subgroupExact = params.subgroupExact === true;
const groupDir = getGroupDir(getDefaultGitnexusDir(), name);
const config = await loadGroupConfig(groupDir);
const perRepo: Array<{ repo: string; score: number; processes: unknown[] }> = [];
for (const [repoPath, registryName] of Object.entries(config.repos)) {
if (!repoInSubgroup(repoPath, subgroup)) continue;
try {
const repoObj = await this.port.resolveRepo(registryName);
const queryResult = (await this.port.query(repoObj, {
query: queryText,
limit,
max_symbols: 10,
include_content: false,
})) as { processes?: Array<Record<string, unknown>> };
const processes = queryResult.processes || [];
const scored = processes.map((p, idx) => ({
...p,
_rrf_score: 1 / (idx + 1 + 60),
_repo: repoPath,
}));
perRepo.push({ repo: repoPath, score: 0, processes: scored });
} catch {
perRepo.push({ repo: repoPath, score: 0, processes: [] });
}
}
const memberEntries = Object.entries(config.repos).filter(([repoPath]) =>
repoInSubgroup(repoPath, subgroup, subgroupExact),
);
// Per-repo query is independent; run them concurrently and isolate
// failures so one slow/failed member does not block the rest.
const perRepo = await Promise.all(
memberEntries.map(async ([repoPath, registryName]) => {
try {
const repoObj = await this.port.resolveRepo(registryName);
const queryResult = (await this.port.query(repoObj, {
query: queryText,
limit,
max_symbols: 10,
include_content: false,
})) as {
processes?: Array<Record<string, unknown>>;
process_symbols?: Array<Record<string, unknown>>;
};
const processes = servicePrefix
? filterQueryByServicePrefix(queryResult, servicePrefix).processes
: queryResult.processes || [];
const scored = processes.map((p, idx) => ({
...p,
_rrf_score: 1 / (idx + 1 + 60),
_repo: repoPath,
}));
return { repo: repoPath, score: 0, processes: scored as unknown[] };
} catch {
return { repo: repoPath, score: 0, processes: [] as unknown[] };
}
}),
);
const allProcesses = perRepo.flatMap((r) => r.processes as Array<Record<string, unknown>>);
allProcesses.sort((a, b) => (b._rrf_score as number) - (a._rrf_score as number));
@ -184,13 +449,10 @@ export class GroupService {
}
> = {};
const fsp = await import('node:fs/promises');
const pathMod = await import('node:path');
for (const [repoPath, registryName] of Object.entries(config.repos)) {
try {
const repoObj = await this.port.resolveRepo(registryName);
const metaPath = pathMod.join(repoObj.storagePath, 'meta.json');
const metaPath = path.join(repoObj.storagePath, 'meta.json');
const metaRaw = await fsp.readFile(metaPath, 'utf-8').catch(() => '{}');
const meta = JSON.parse(metaRaw) as { lastCommit?: string; indexedAt?: string };

View file

@ -96,6 +96,9 @@ export interface RepoHandle {
storagePath: string;
}
/** Why local impact or fan-out stopped early (e.g. wall-clock budget exhausted). */
export type GroupImpactTruncationReason = 'timeout' | 'partial';
export interface GroupImpactResult {
local: unknown;
group: string;
@ -110,6 +113,36 @@ export interface GroupImpactResult {
cross_repo_hits: number;
};
risk: string;
/**
* Milliseconds budget applied to the **Phase 1 local impact** leg (`safeLocalImpact`).
* If the walk hits this wall first, expect `truncationReason: 'timeout'` and a partial `local` payload.
*/
timeoutMs?: number;
/** Present when local impact or fan-out stopped early (timeout, graph cap, etc.). */
truncationReason?: GroupImpactTruncationReason;
/**
* Human-readable note when `crossDepth` was clamped (e.g. multi-hop not implemented yet).
*/
crossDepthWarning?: string;
}
/** One repo’s `context` tool payload in a group-scoped context run. */
export interface GroupContextRepoEntry {
repoPath: string;
registryName: string;
payload: unknown;
}
/**
* Aggregated group `context`: explicit per-repo rows (no merged symbol payloads).
* Use top-level `error` only for unrecoverable failures, not for “no matches” or service scope misses.
*/
export interface GroupContextResult {
group: string;
target?: string;
service?: string;
error?: string;
results: GroupContextRepoEntry[];
}
export interface CrossRepoImpact {

View file

@ -144,6 +144,18 @@ let currentDbPath: string | null = null;
let ftsLoaded = false;
let vectorExtensionLoaded = false;
/**
* In-process cache of FTS indexes that have been ensured against the current
* connection. Prevents repeated `CALL CREATE_FTS_INDEX` round-trips inside a
* single CLI/MCP session — the first call to `ensureFTSIndex` for a given
* `(tableName, indexName)` pays the LadybugDB cost (~440 ms even when the
* index already exists on disk), subsequent calls are a Set lookup. Cleared
* by `closeLbug` so a re-init starts fresh.
*
* Key format: `${tableName}:${indexName}`.
*/
const ensuredFTSIndexes = new Set<string>();
/**
* Check if an error indicates a missing column or table (schema-level problem)
* rather than a transient/connection error. Used for legacy DB fallback logic.
@ -1037,6 +1049,7 @@ export const closeLbug = async (): Promise<void> => {
currentDbPath = null;
ftsLoaded = false;
vectorExtensionLoaded = false;
ensuredFTSIndexes.clear();
};
export const isLbugReady = (): boolean => conn !== null && db !== null;
@ -1219,6 +1232,29 @@ export const createFTSIndex = async (
}
};
/**
* Lazy-create an FTS index, caching the fact in-process.
*
* Used by `queryFTS` so that `analyze` doesn't pay the ~440 ms × 5 fixed
* LadybugDB cost up-front (it dominates analyze on small repos). Instead,
* the cost is moved to the first `query`/`context` call in a session,
* where it's amortised across many lookups.
*
* Safe to call repeatedly — the in-process Set guarantees only the first
* call hits LadybugDB. `closeLbug` clears the cache so re-init starts fresh.
*/
export const ensureFTSIndex = async (
tableName: string,
indexName: string,
properties: string[],
stemmer: string = 'porter',
): Promise<void> => {
const key = `${tableName}:${indexName}`;
if (ensuredFTSIndexes.has(key)) return;
await createFTSIndex(tableName, indexName, properties, stemmer);
ensuredFTSIndexes.add(key);
};
/**
* Query a full-text search index
* @param tableName - The node table name

View file

@ -35,6 +35,30 @@ interface PoolEntry {
const pool = new Map<string, PoolEntry>();
/**
* Listeners notified when a pool entry is torn down (LRU eviction, idle
* timeout, explicit close). Used by upper layers (e.g. the BM25 search
* module) to invalidate per-repo caches that must not outlive the pool
* entry that produced them.
*
* Listeners run synchronously inside `closeOne` after the pool entry has
* been removed; throwing listeners are isolated so one bad listener does
* not prevent others from firing or break teardown.
*/
type PoolCloseListener = (repoId: string) => void;
const poolCloseListeners = new Set<PoolCloseListener>();
/**
* Subscribe to pool-close events. Returns a disposer that removes the
* listener (handy for tests).
*/
export function addPoolCloseListener(listener: PoolCloseListener): () => void {
poolCloseListeners.add(listener);
return () => {
poolCloseListeners.delete(listener);
};
}
/**
* Shared Database cache keyed by resolved dbPath.
* Multiple repoIds pointing to the same path share one native Database
@ -159,6 +183,16 @@ function closeOne(repoId: string): void {
}
pool.delete(repoId);
// Notify listeners AFTER the pool entry is gone so any cache-invalidation
// they perform is consistent with `isLbugReady(repoId) === false`.
for (const listener of poolCloseListeners) {
try {
listener(repoId);
} catch {
// Isolate listener failures — teardown must complete.
}
}
}
/**

View file

@ -19,7 +19,6 @@ import {
executeQuery,
executeWithReusedStatement,
closeLbug,
createFTSIndex,
loadCachedEmbeddings,
} from './lbug/lbug-adapter.js';
import {
@ -220,17 +219,12 @@ export async function runFullAnalysis(
});
// ── Phase 3: FTS (85–90%) ─────────────────────────────────────────
progress('fts', 85, 'Creating search indexes...');
try {
await createFTSIndex('File', 'file_fts', ['name', 'content']);
await createFTSIndex('Function', 'function_fts', ['name', 'content']);
await createFTSIndex('Class', 'class_fts', ['name', 'content']);
await createFTSIndex('Method', 'method_fts', ['name', 'content']);
await createFTSIndex('Interface', 'interface_fts', ['name', 'content']);
} catch {
// Non-fatal — FTS is best-effort
}
// FTS indexes are created lazily on first `query`/`context` call instead
// of eagerly here. On small repos / CI runners the LadybugDB
// CREATE_FTS_INDEX cost is ~440 ms × 5 (≈2 s) regardless of table size,
// which dominated `analyze` runtime and pushed Windows CI past its
// 30 s test budget. Lazy creation is implemented in
// `core/search/bm25-index.ts` via `ensureFTSIndex`.
// ── Phase 3.5: Re-insert cached embeddings ────────────────────────
if (cachedEmbeddings.length > 0) {

View file

@ -3,14 +3,119 @@
*
* Uses LadybugDB's built-in full-text search indexes for keyword-based search.
* Always reads from the database (no cached state to drift).
*
* FTS indexes are created lazily on first query (via `ensureFTSIndex`) — see
* `lbug-adapter.ts` for the rationale. This keeps `analyze` fast (the
* ~440 ms × 5 LadybugDB CREATE_FTS_INDEX cost dominates pipeline time on
* small repos / CI runners) at the cost of paying that overhead on the
* first `query`/`context` call in a session.
*/
import { queryFTS } from '../lbug/lbug-adapter.js';
import { queryFTS, ensureFTSIndex } from '../lbug/lbug-adapter.js';
export interface BM25SearchResult {
filePath: string;
score: number;
rank: number;
nodeIds?: string[];
}
/**
* FTS schema served by `searchFTSFromLbug`. Centralised so that both the
* CLI/pipeline path and the MCP pool path use identical (table, index,
* properties) tuples and the lazy-create logic stays in one place.
*/
const FTS_INDEXES: ReadonlyArray<{
table: string;
indexName: string;
properties: readonly string[];
}> = [
{ table: 'File', indexName: 'file_fts', properties: ['name', 'content'] },
{ table: 'Function', indexName: 'function_fts', properties: ['name', 'content'] },
{ table: 'Class', indexName: 'class_fts', properties: ['name', 'content'] },
{ table: 'Method', indexName: 'method_fts', properties: ['name', 'content'] },
{ table: 'Interface', indexName: 'interface_fts', properties: ['name', 'content'] },
];
/**
* Per-process cache for the MCP pool path: tracks which `(repoId, table)`
* pairs have been ensured. The CLI/pipeline path gets its own cache inside
* `lbug-adapter.ts` keyed by table/index, scoped to the singleton connection.
*
* IMPORTANT: an entry is added ONLY when the index was confirmed to exist
* (CREATE_FTS_INDEX succeeded, or failed with `'already exists'`). Other
* failures (transient lock errors, missing extension, etc.) leave the key
* unset so the next query retries instead of silently caching the failure.
*
* Entries for a given repoId are invalidated when its pool is closed —
* see the `addPoolCloseListener` registration in `searchFTSFromLbug`.
*/
const ensuredPoolFTS = new Set<string>();
/**
* Drop all ensured-FTS cache entries for a given repoId.
*
* Called from the pool-close listener so that a pool teardown / recreation
* forces the next `searchFTSFromLbug` call to re-issue `CREATE_FTS_INDEX`
* against the fresh connection rather than trust stale ensure-state from a
* previous pool lifetime.
*
* Exported for tests; the listener wiring is internal.
*/
export function invalidateEnsuredFTSForRepo(repoId: string): void {
const prefix = `${repoId}:`;
for (const key of ensuredPoolFTS) {
if (key.startsWith(prefix)) ensuredPoolFTS.delete(key);
}
}
/**
* Tracks whether we've already wired the pool-close listener for this
* process. The pool adapter is dynamically imported, so registration
* happens lazily on the first MCP-pool-backed FTS query.
*/
let poolCloseListenerRegistered = false;
function registerPoolCloseListenerOnce(
addPoolCloseListener: (listener: (repoId: string) => void) => void,
): void {
if (poolCloseListenerRegistered) return;
poolCloseListenerRegistered = true;
addPoolCloseListener((repoId) => invalidateEnsuredFTSForRepo(repoId));
}
async function ensureFTSIndexViaExecutor(
executor: (cypher: string) => Promise<any[]>,
repoId: string,
table: string,
indexName: string,
properties: readonly string[],
): Promise<void> {
const key = `${repoId}:${table}:${indexName}`;
if (ensuredPoolFTS.has(key)) return;
const propList = properties.map((p) => `'${p}'`).join(', ');
try {
await executor(
`CALL CREATE_FTS_INDEX('${table}', '${indexName}', [${propList}], stemmer := 'porter')`,
);
// Index was created successfully — safe to cache.
ensuredPoolFTS.add(key);
} catch (e: any) {
// 'already exists' is the happy path (index persists on disk between
// process invocations) — cache it. Anything else is treated as a
// transient failure: surface a one-time warning and leave the key
// unset so the NEXT query retries rather than silently using a
// cached failure (which previously disabled BM25 for the whole
// process for that repo).
const msg = String(e?.message ?? '');
if (msg.includes('already exists')) {
ensuredPoolFTS.add(key);
} else {
console.warn(
`[gitnexus] FTS index ensure failed for repo "${repoId}" table "${table}" ` +
`(index "${indexName}"): ${msg || e}. Will retry on next query.`,
);
}
}
}
/**
@ -23,7 +128,7 @@ async function queryFTSViaExecutor(
indexName: string,
query: string,
limit: number,
): Promise<Array<{ filePath: string; score: number }>> {
): Promise<Array<{ filePath: string; score: number; nodeId: string }>> {
// Escape single quotes and backslashes to prevent Cypher injection
const escapedQuery = query.replace(/\\/g, '\\\\').replace(/'/g, "''");
const cypher = `
@ -40,6 +145,7 @@ async function queryFTSViaExecutor(
return {
filePath: node.filePath || '',
score: typeof score === 'number' ? score : parseFloat(score) || 0,
nodeId: node.nodeId || node.id || '',
};
});
} catch {
@ -73,8 +179,21 @@ export const searchFTSFromLbug = async (
// Use MCP connection pool via dynamic import
// IMPORTANT: FTS queries run sequentially to avoid connection contention.
// The MCP pool supports multiple connections, but FTS is best run serially.
const { executeQuery } = await import('../lbug/pool-adapter.js');
const poolMod = await import('../lbug/pool-adapter.js');
const { executeQuery, addPoolCloseListener } = poolMod;
// Register the pool-close listener lazily on first use so a teardown of
// the pool entry (LRU eviction, idle timeout, explicit close) drops the
// matching `ensuredPoolFTS` entries. Without this, stale ensure-state
// can outlive the pool that produced it.
registerPoolCloseListenerOnce(addPoolCloseListener);
const executor = (cypher: string) => executeQuery(repoId, cypher);
// Lazy-create FTS indexes on first query for this repo (analyze no longer
// creates them up-front, so we ensure them here). Cached per-process.
for (const { table, indexName, properties } of FTS_INDEXES) {
await ensureFTSIndexViaExecutor(executor, repoId, table, indexName, properties);
}
fileResults = await queryFTSViaExecutor(executor, 'File', 'file_fts', query, limit);
functionResults = await queryFTSViaExecutor(executor, 'Function', 'function_fts', query, limit);
classResults = await queryFTSViaExecutor(executor, 'Class', 'class_fts', query, limit);
@ -87,7 +206,12 @@ export const searchFTSFromLbug = async (
limit,
);
} else {
// Use core lbug adapter (CLI / pipeline context) — also sequential for safety
// Use core lbug adapter (CLI / pipeline context) — also sequential for safety.
// Lazy-create FTS indexes on first query (analyze no longer does it).
for (const { table, indexName, properties } of FTS_INDEXES) {
await ensureFTSIndex(table, indexName, [...properties]).catch(() => {});
}
fileResults = await queryFTS('File', 'file_fts', query, limit, false).catch(() => []);
functionResults = await queryFTS('Function', 'function_fts', query, limit, false).catch(
() => [],
@ -99,17 +223,13 @@ export const searchFTSFromLbug = async (
);
}
// Merge results by filePath, summing scores for same file
const merged = new Map<string, { filePath: string; score: number }>();
// Collect all node scores per filePath to track which nodes actually matched
const fileNodeScores = new Map<string, Array<{ score: number; nodeId: string }>>();
const addResults = (results: any[]) => {
for (const r of results) {
const existing = merged.get(r.filePath);
if (existing) {
existing.score += r.score;
} else {
merged.set(r.filePath, { filePath: r.filePath, score: r.score });
}
if (!fileNodeScores.has(r.filePath)) fileNodeScores.set(r.filePath, []);
fileNodeScores.get(r.filePath)!.push({ score: r.score, nodeId: r.nodeId });
}
};
@ -119,6 +239,19 @@ export const searchFTSFromLbug = async (
addResults(methodResults);
addResults(interfaceResults);
// Sum the top-3 highest-scoring nodes per file and collect their nodeIds.
// Summing all nodes naively inflates scores for files with many mediocre
// matches (e.g. test files) over files with a single highly-relevant symbol.
const merged = new Map<string, { filePath: string; score: number; nodeIds: string[] }>();
for (const [filePath, entries] of fileNodeScores) {
const top3 = [...entries].sort((a, b) => b.score - a.score).slice(0, 3);
merged.set(filePath, {
filePath,
score: top3.reduce((acc, e) => acc + e.score, 0),
nodeIds: top3.map((e) => e.nodeId).filter((id) => id),
});
}
// Sort by score descending and add rank
const sorted = Array.from(merged.values())
.sort((a, b) => b.score - a.score)
@ -128,5 +261,6 @@ export const searchFTSFromLbug = async (
filePath: r.filePath,
score: r.score,
rank: index + 1,
nodeIds: r.nodeIds,
}));
};

View file

@ -28,6 +28,7 @@ import {
type RegistryEntry,
} from '../../storage/repo-manager.js';
import { GroupService, type GroupToolPort } from '../../core/group/service.js';
import { resolveAtGroupMemberRepoPath } from '../../core/group/resolve-at-member.js';
import { collectBestChunks } from '../../core/embeddings/types.js';
import { EMBEDDING_TABLE_NAME, EMBEDDING_INDEX_NAME } from '../../core/lbug/schema.js';
import { PhaseTimer } from '../../core/search/phase-timer.js';
@ -227,6 +228,7 @@ export class LocalBackend {
impact: (r, p) => this.impact(r as RepoHandle, p),
query: (r, p) => this.query(r as RepoHandle, p),
impactByUid: (id, uid, d, o) => this.impactByUid(id, uid, d, o),
context: (r, p) => this.context(r as RepoHandle, p),
};
this.groupToolSvc = new GroupService(port);
}
@ -633,8 +635,17 @@ export class LocalBackend {
return this.handleGroupTool(method, params || {});
}
const p = params && typeof params === 'object' ? (params as Record<string, unknown>) : {};
if (
(method === 'impact' || method === 'query' || method === 'context') &&
typeof p.repo === 'string' &&
p.repo.startsWith('@')
) {
return this.callToolAtGroupRepo(method, p);
}
// Resolve repo from optional param (re-reads registry on miss)
const repo = await this.resolveRepo(params?.repo);
const repo = await this.resolveRepo((params as { repo?: string } | undefined)?.repo);
switch (method) {
case 'query':
@ -974,16 +985,30 @@ export class LocalBackend {
for (const bm25Result of bm25Results) {
const fullPath = bm25Result.filePath;
try {
const symbols = await executeParameterized(
repo.id,
`
MATCH (n)
WHERE n.filePath = $filePath
RETURN n.id AS id, n.name AS name, labels(n)[0] AS type, n.filePath AS filePath, n.startLine AS startLine, n.endLine AS endLine
LIMIT 3
`,
{ filePath: fullPath },
);
// Prefer direct nodeId lookup (exact FTS-matched nodes) over filePath fallback.
// Without this, LIMIT 3 on filePath returns arbitrary symbols rather than
// the nodes that actually scored highest in the BM25 index.
const nodeIds = bm25Result.nodeIds?.length ? bm25Result.nodeIds : null;
const symbols = nodeIds
? await executeParameterized(
repo.id,
`
MATCH (n)
WHERE n.id IN $nodeIds
RETURN n.id AS id, n.name AS name, labels(n)[0] AS type, n.filePath AS filePath, n.startLine AS startLine, n.endLine AS endLine
`,
{ nodeIds },
)
: await executeParameterized(
repo.id,
`
MATCH (n)
WHERE n.filePath = $filePath
RETURN n.id AS id, n.name AS name, labels(n)[0] AS type, n.filePath AS filePath, n.startLine AS startLine, n.endLine AS endLine
LIMIT 3
`,
{ filePath: fullPath },
);
if (symbols.length > 0) {
for (const sym of symbols) {
@ -2969,17 +2994,103 @@ export class LocalBackend {
return this.groupList(params);
case 'group_sync':
return this.groupSync(params);
case 'group_contracts':
return this.groupContracts(params);
case 'group_query':
return this.groupQuery(params);
case 'group_status':
return this.groupStatus(params);
default:
throw new Error(`Unknown group tool: ${method}`);
throw new Error(
`Unknown group tool: ${method}. Removed tools: use repo "@<groupName>" on impact, query, or context (optional "/<memberPath>"), or MCP resources.`,
);
}
}
/**
* Dispatch impact/query/context when `repo` is `@groupName` or `@groupName/memberPath`
* (group mode — not the global indexed-repo `repo` parameter).
*/
private async callToolAtGroupRepo(
method: string,
params: Record<string, unknown>,
): Promise<unknown> {
await this.refreshRepos();
if (
params.service !== undefined &&
params.service !== null &&
String(params.service).trim() === ''
) {
return { error: 'service must not be an empty string' };
}
const raw = String(params.repo).slice(1);
const slash = raw.indexOf('/');
const groupName = (slash === -1 ? raw : raw.slice(0, slash)).trim();
const memberRest = slash === -1 ? undefined : raw.slice(slash + 1).trim() || undefined;
const resolved = await resolveAtGroupMemberRepoPath(groupName, memberRest);
if (resolved.ok === false) return { error: resolved.error };
const svc = this.getGroupService();
if (method === 'impact') {
const impactArgs: Record<string, unknown> = {
name: groupName,
repo: resolved.repoPath,
target: params.target,
direction: params.direction,
};
if (params.maxDepth !== undefined) impactArgs.maxDepth = params.maxDepth;
if (params.crossDepth !== undefined) impactArgs.crossDepth = params.crossDepth;
if (params.relationTypes !== undefined) impactArgs.relationTypes = params.relationTypes;
if (params.includeTests !== undefined) impactArgs.includeTests = params.includeTests;
if (params.minConfidence !== undefined) impactArgs.minConfidence = params.minConfidence;
if (params.service !== undefined && params.service !== null)
impactArgs.service = params.service;
if (typeof params.subgroup === 'string') impactArgs.subgroup = params.subgroup;
if (params.timeoutMs !== undefined) impactArgs.timeoutMs = params.timeoutMs;
if (params.timeout !== undefined) impactArgs.timeout = params.timeout;
return svc.groupImpact(impactArgs);
}
if (method === 'query') {
const queryArgs: Record<string, unknown> = {
name: groupName,
query: params.query,
};
if (typeof params.task_context === 'string') queryArgs.task_context = params.task_context;
if (typeof params.goal === 'string') queryArgs.goal = params.goal;
if (typeof params.limit === 'number') queryArgs.limit = params.limit;
if (typeof params.max_symbols === 'number') queryArgs.max_symbols = params.max_symbols;
if (params.include_content !== undefined) queryArgs.include_content = params.include_content;
if (params.service !== undefined && params.service !== null)
queryArgs.service = params.service;
if (memberRest !== undefined) {
queryArgs.subgroup = memberRest;
queryArgs.subgroupExact = true;
}
return svc.groupQuery(queryArgs);
}
if (method === 'context') {
const targetSym =
typeof params.target === 'string' && params.target.trim() !== ''
? params.target.trim()
: typeof params.name === 'string' && params.name.trim() !== ''
? params.name.trim()
: undefined;
const contextArgs: Record<string, unknown> = {
name: groupName,
target: targetSym,
};
if (typeof params.uid === 'string') contextArgs.uid = params.uid;
if (typeof params.file_path === 'string') contextArgs.file_path = params.file_path;
if (params.include_content !== undefined)
contextArgs.include_content = params.include_content;
if (params.service !== undefined && params.service !== null)
contextArgs.service = params.service;
if (memberRest !== undefined) {
contextArgs.subgroup = memberRest;
contextArgs.subgroupExact = true;
}
return svc.groupContext(contextArgs);
}
throw new Error(`Internal: unsupported group-repo tool ${method}`);
}
private async groupList(params: Record<string, unknown>): Promise<unknown> {
return this.getGroupService().groupList(params);
}
@ -2988,18 +3099,45 @@ export class LocalBackend {
return this.getGroupService().groupSync(params);
}
private async groupContracts(params: Record<string, unknown>): Promise<unknown> {
return this.getGroupService().groupContracts(params);
/**
* MCP resource body for `gitnexus://group/{name}/contracts` (Issue #794).
*/
async readGroupContractsResource(
groupName: string,
filter: { type?: string; repo?: string; unmatchedOnly?: boolean },
): Promise<string> {
try {
const params: Record<string, unknown> = { name: groupName };
if (filter.type !== undefined) params.type = filter.type;
if (filter.repo !== undefined) params.repo = filter.repo;
if (filter.unmatchedOnly === true) params.unmatchedOnly = true;
const raw = await this.getGroupService().groupContracts(params);
return LocalBackend.formatGroupResourcePayload(raw);
} catch (e) {
return `error: ${e instanceof Error ? e.message : String(e)}`;
}
}
private async groupQuery(params: Record<string, unknown>): Promise<unknown> {
await this.refreshRepos();
return this.getGroupService().groupQuery(params);
/**
* MCP resource body for `gitnexus://group/{name}/status` (Issue #794).
*/
async readGroupStatusResource(groupName: string): Promise<string> {
try {
const raw = await this.getGroupService().groupStatus({ name: groupName });
return LocalBackend.formatGroupResourcePayload(raw);
} catch (e) {
return `error: ${e instanceof Error ? e.message : String(e)}`;
}
}
private async groupStatus(params: Record<string, unknown>): Promise<unknown> {
await this.refreshRepos();
return this.getGroupService().groupStatus(params);
private static formatGroupResourcePayload(raw: unknown): string {
if (raw && typeof raw === 'object' && 'error' in raw) {
const err = (raw as { error?: unknown }).error;
if (typeof err === 'string' && err.length > 0) {
return `error: ${err}`;
}
}
return JSON.stringify(raw, null, 2);
}
/**

View file

@ -84,38 +84,140 @@ export function getResourceTemplates(): ResourceTemplate[] {
description: 'Step-by-step execution trace',
mimeType: 'text/yaml',
},
{
uriTemplate: 'gitnexus://group/{name}/contracts',
name: 'Group Contract Registry',
description:
'Cross-repo contract registry for a repository group. Optional query: type, repo, unmatchedOnly (true|false).',
mimeType: 'text/yaml',
},
{
uriTemplate: 'gitnexus://group/{name}/status',
name: 'Group Index Status',
description: 'Per-repo index and contract-registry staleness for a repository group',
mimeType: 'text/yaml',
},
];
}
/**
* Parse a resource URI to extract the repo name and resource type.
*/
function parseUri(uri: string): { repoName?: string; resourceType: string; param?: string } {
if (uri === 'gitnexus://repos') return { resourceType: 'repos' };
if (uri === 'gitnexus://setup') return { resourceType: 'setup' };
/** Query parameters for `gitnexus://group/{name}/contracts` */
export type GroupContractsResourceFilter = {
type?: string;
repo?: string;
unmatchedOnly?: boolean;
};
// Repo-scoped: gitnexus://repo/{name}/context
const repoMatch = uri.match(/^gitnexus:\/\/repo\/([^/]+)\/(.+)$/);
if (repoMatch) {
const repoName = decodeURIComponent(repoMatch[1]);
const rest = repoMatch[2];
/** Normalized parse result for GitNexus MCP resource URIs */
export type ParsedGitnexusResource =
| { kind: 'repos' }
| { kind: 'setup' }
| {
kind: 'repo';
repoName: string;
resourceType: string;
param?: string;
}
| {
kind: 'group';
groupName: string;
resourceType: 'contracts';
contractsFilter: GroupContractsResourceFilter;
}
| { kind: 'group'; groupName: string; resourceType: 'status' };
function parseUnmatchedOnlyParam(raw: string | null): boolean | undefined {
if (raw === null) return undefined;
const v = raw.trim().toLowerCase();
if (v === 'true' || v === '1') return true;
if (v === 'false' || v === '0') return false;
return undefined;
}
/**
* Parse a GitNexus resource URI (repos, setup, per-repo, or per-group templates).
* Used by `readResource` and tests (round-trip / dispatch coverage).
*/
export function parseResourceUri(uri: string): ParsedGitnexusResource {
if (uri === 'gitnexus://repos') return { kind: 'repos' };
if (uri === 'gitnexus://setup') return { kind: 'setup' };
let u: URL;
try {
u = new URL(uri);
} catch {
throw new Error(`Unknown resource URI: ${uri}`);
}
if (u.protocol !== 'gitnexus:') {
throw new Error(`Unknown resource URI: ${uri}`);
}
if (u.hostname === 'group') {
const segments = u.pathname
.replace(/^\/+|\/+$/g, '')
.split('/')
.filter(Boolean);
if (segments.length < 2) {
throw new Error(
`Invalid group resource URI (expected gitnexus://group/{name}/contracts or .../status): ${uri}`,
);
}
const tail = segments[segments.length - 1]!;
if (tail !== 'contracts' && tail !== 'status') {
throw new Error(`Unknown group resource path in URI: ${uri}`);
}
const groupName = segments
.slice(0, -1)
.map((s) => decodeURIComponent(s))
.join('/');
if (!groupName) {
throw new Error(`Invalid group resource URI (empty group name): ${uri}`);
}
if (tail === 'status') {
return { kind: 'group', groupName, resourceType: 'status' };
}
const contractsFilter: GroupContractsResourceFilter = {};
const type = u.searchParams.get('type');
if (type && type.trim()) contractsFilter.type = type.trim();
const repo = u.searchParams.get('repo');
if (repo && repo.trim()) contractsFilter.repo = repo.trim();
if (u.searchParams.has('unmatchedOnly')) {
const coerced = parseUnmatchedOnlyParam(u.searchParams.get('unmatchedOnly'));
if (coerced !== undefined) contractsFilter.unmatchedOnly = coerced;
}
return { kind: 'group', groupName, resourceType: 'contracts', contractsFilter };
}
if (u.hostname === 'repo') {
const segments = u.pathname
.replace(/^\/+|\/+$/g, '')
.split('/')
.filter(Boolean);
if (segments.length < 2) {
throw new Error(`Unknown resource URI: ${uri}`);
}
const repoName = decodeURIComponent(segments[0]!);
const restEncoded = segments.slice(1);
const rest = restEncoded.map((s) => decodeURIComponent(s)).join('/');
if (rest.startsWith('cluster/')) {
return {
kind: 'repo',
repoName,
resourceType: 'cluster',
param: decodeURIComponent(rest.replace('cluster/', '')),
param: rest.replace(/^cluster\//, ''),
};
}
if (rest.startsWith('process/')) {
return {
kind: 'repo',
repoName,
resourceType: 'process',
param: decodeURIComponent(rest.replace('process/', '')),
param: rest.replace(/^process\//, ''),
};
}
return { repoName, resourceType: rest };
return { kind: 'repo', repoName, resourceType: rest };
}
throw new Error(`Unknown resource URI: ${uri}`);
@ -125,18 +227,23 @@ function parseUri(uri: string): { repoName?: string; resourceType: string; param
* Read a resource and return its content
*/
export async function readResource(uri: string, backend: LocalBackend): Promise<string> {
const parsed = parseUri(uri);
const parsed = parseResourceUri(uri);
// Global repos list — no repo context needed
if (parsed.resourceType === 'repos') {
if (parsed.kind === 'repos') {
return getReposResource(backend);
}
// Setup resource — returns AGENTS.md content for all repos
if (parsed.resourceType === 'setup') {
if (parsed.kind === 'setup') {
return getSetupResource(backend);
}
if (parsed.kind === 'group') {
if (parsed.resourceType === 'contracts') {
return backend.readGroupContractsResource(parsed.groupName, parsed.contractsFilter);
}
return backend.readGroupStatusResource(parsed.groupName);
}
const repoName = parsed.repoName;
switch (parsed.resourceType) {
@ -241,6 +348,10 @@ async function getContextResource(backend: LocalBackend, repoName?: string): Pro
lines.push(` - gitnexus://repo/${context.projectName}/processes: All execution flows`);
lines.push(` - gitnexus://repo/${context.projectName}/cluster/{name}: Module details`);
lines.push(` - gitnexus://repo/${context.projectName}/process/{name}: Process trace`);
lines.push(
' - gitnexus://group/{name}/contracts: Group contract registry (optional ?type=&repo=&unmatchedOnly=)',
);
lines.push(' - gitnexus://group/{name}/status: Group index / contract staleness');
return lines.join('\n');
}

View file

@ -15,9 +15,12 @@ export interface ToolDefinition {
{
type: string;
description?: string;
default?: any;
default?: unknown;
items?: { type: string };
enum?: string[];
minimum?: number;
maximum?: number;
minLength?: number;
}
>;
required: string[];
@ -55,7 +58,11 @@ Returns results grouped by process (execution flow):
- process_symbols: all symbols in those flows with file locations and module (functional area)
- definitions: standalone types/interfaces not in any process
Hybrid ranking: BM25 keyword + semantic vector search, ranked by Reciprocal Rank Fusion.`,
Hybrid ranking: BM25 keyword + semantic vector search, ranked by Reciprocal Rank Fusion.
GROUP MODE: set "repo" to "@<groupName>" to search all member repos in that group (merged via RRF), or "@<groupName>/<groupRepoPath>" to run against a single member (same path keys as in group.yaml). If you use "@<groupName>" only, the member repo defaults to the lexicographically first key in group.yaml "repos". Prefer resources for contracts/status (see migration from legacy group_* tools).
SERVICE: optional monorepo path prefix (POSIX-style, case-sensitive segments). When "repo" starts with "@", only processes whose symbols fall under that prefix are included. For a normal indexed repo name (no leading @), this field is currently ignored by the server.`,
inputSchema: {
type: 'object',
properties: {
@ -69,11 +76,19 @@ Hybrid ranking: BM25 keyword + semantic vector search, ranked by Reciprocal Rank
description:
'What you want to find (e.g., "existing auth validation logic"). Helps ranking.',
},
limit: { type: 'number', description: 'Max processes to return (default: 5)', default: 5 },
limit: {
type: 'number',
description: 'Max processes to return (default: 5)',
default: 5,
minimum: 1,
maximum: 100,
},
max_symbols: {
type: 'number',
description: 'Max symbols per process (default: 10)',
default: 10,
minimum: 1,
maximum: 200,
},
include_content: {
type: 'boolean',
@ -82,7 +97,14 @@ Hybrid ranking: BM25 keyword + semantic vector search, ranked by Reciprocal Rank
},
repo: {
type: 'string',
description: 'Repository name or path. Omit if only one repo is indexed.',
description:
'Indexed repository name or path, or group mode "@<groupName>" / "@<groupName>/<memberPath>" (member path keys from group.yaml). Omit when only one indexed repo exists.',
},
service: {
type: 'string',
minLength: 1,
description:
'Optional monorepo service root (relative path, "/" separators). In group mode (@repo), prefix-matches symbol file paths; ignored for a normal repo name. Empty string is rejected server-side.',
},
},
required: ['query'],
@ -156,7 +178,11 @@ AFTER THIS: Use impact() if planning changes, or READ gitnexus://repo/{name}/pro
Handles disambiguation: if multiple symbols share the same name, returns ranked candidates (each with a relevance score) for you to pick from. Use uid for zero-ambiguity lookup, or narrow the search with file_path and/or kind hints.
NOTE: ACCESSES edges (field read/write tracking) are included in context results with reason 'read' or 'write'. CALLS edges resolve through field access chains and method-call chains (e.g., user.address.getCity().save() produces CALLS edges at each step).`,
NOTE: ACCESSES edges (field read/write tracking) are included in context results with reason 'read' or 'write'. CALLS edges resolve through field access chains and method-call chains (e.g., user.address.getCity().save() produces CALLS edges at each step).
GROUP MODE: set "repo" to "@<groupName>" to run context in each member repo (aggregated list), or "@<groupName>/<groupRepoPath>" for one member. If you use "@<groupName>" only, the member defaults to the lexicographically first key in group.yaml "repos".
SERVICE: optional monorepo path prefix (case-sensitive path segments). When "repo" starts with "@", prefix-matches resolved symbol file paths; when a hit is outside the prefix, that member returns an empty payload for the symbol. Ignored for a normal indexed repo name.`,
inputSchema: {
type: 'object',
properties: {
@ -178,7 +204,14 @@ NOTE: ACCESSES edges (field read/write tracking) are included in context results
},
repo: {
type: 'string',
description: 'Repository name or path. Omit if only one repo is indexed.',
description:
'Indexed repository name or path, or group mode "@<groupName>" / "@<groupName>/<memberPath>". Omit if only one repo is indexed.',
},
service: {
type: 'string',
minLength: 1,
description:
'Optional monorepo service root (relative path). Applies in group mode (@repo) only; ignored for a normal repo name. Empty string is rejected server-side.',
},
},
required: [],
@ -273,7 +306,11 @@ TIP: Default traversal uses CALLS/IMPORTS/EXTENDS/IMPLEMENTS. For class members,
Handles disambiguation: when multiple symbols share the target name, returns ranked candidates (each with a relevance score) instead of silently picking one. Use target_uid for zero-ambiguity lookup, or narrow with file_path and/or kind hints.
EdgeType: CALLS, IMPORTS, EXTENDS, IMPLEMENTS, HAS_METHOD, HAS_PROPERTY, METHOD_OVERRIDES, METHOD_IMPLEMENTS, ACCESSES
Confidence: 1.0 = certain, <0.8 = fuzzy match`,
Confidence: 1.0 = certain, <0.8 = fuzzy match
GROUP MODE: set "repo" to "@<groupName>" for cross-repo impact anchored at the default member (lexicographically first key in group.yaml "repos"), or "@<groupName>/<groupRepoPath>" to choose the member (same path keys as in group.yaml). Phase-1 walk runs in that member; cross-boundary fan-out uses the group bridge.
SERVICE: optional monorepo path prefix (case-sensitive path segments). When "repo" starts with "@", scopes the local impact walk and cross-repo symbol paths to files under that prefix; ignored for a normal indexed repo name.`,
inputSchema: {
type: 'object',
properties: {
@ -298,8 +335,18 @@ Confidence: 1.0 = certain, <0.8 = fuzzy match`,
},
maxDepth: {
type: 'number',
description: 'Max relationship depth (default: 3)',
description: 'Max relationship depth (default: 3, server clamps to 1–32)',
default: 3,
minimum: 1,
maximum: 32,
},
crossDepth: {
type: 'number',
description:
'Cross-repository hop depth via contract bridge (default: 1; values above server maximum are clamped)',
default: 1,
minimum: 1,
maximum: 32,
},
relationTypes: {
type: 'array',
@ -308,10 +355,42 @@ Confidence: 1.0 = certain, <0.8 = fuzzy match`,
'Filter: CALLS, IMPORTS, EXTENDS, IMPLEMENTS, HAS_METHOD, HAS_PROPERTY, METHOD_OVERRIDES, METHOD_IMPLEMENTS, ACCESSES (default: usage-based, ACCESSES excluded by default)',
},
includeTests: { type: 'boolean', description: 'Include test files (default: false)' },
minConfidence: { type: 'number', description: 'Minimum confidence 0-1 (default: 0.7)' },
minConfidence: {
type: 'number',
description:
'Minimum edge confidence 0–1 (default: 0 when omitted; server clamps to 0–1)',
default: 0,
minimum: 0,
maximum: 1,
},
repo: {
type: 'string',
description: 'Repository name or path. Omit if only one repo is indexed.',
description:
'Indexed repository name or path, or group mode "@<groupName>" / "@<groupName>/<memberPath>". Omit if only one repo is indexed.',
},
service: {
type: 'string',
minLength: 1,
description:
'Optional monorepo service root (relative path). Applies when "repo" is group mode (@…); ignored for a normal repo name. Empty string is rejected server-side.',
},
subgroup: {
type: 'string',
description:
'Optional group subgroup prefix (member repo paths) limiting which repos participate in cross fan-out.',
},
timeoutMs: {
type: 'number',
description:
'Wall-clock budget in milliseconds for the Phase-1 local impact leg (default 30000)',
minimum: 1,
maximum: 3600000,
},
timeout: {
type: 'number',
description: 'Alias of timeoutMs (milliseconds) when timeoutMs is omitted',
minimum: 1,
maximum: 3600000,
},
},
required: ['target', 'direction'],
@ -429,49 +508,4 @@ WHEN TO USE: After changing group.yaml or re-indexing member repos.`,
required: ['name'],
},
},
{
name: 'group_contracts',
description: `Inspect contracts and cross-links from the group's contracts.json.
WHEN TO USE: Debug cross-repo links after group_sync.`,
inputSchema: {
type: 'object',
properties: {
name: { type: 'string', description: 'Group name' },
type: { type: 'string', description: 'Filter by contract type (http, topic, …)' },
repo: { type: 'string', description: 'Filter by group repo path (e.g. app/backend)' },
unmatchedOnly: { type: 'boolean', description: 'Only contracts with no cross-link' },
},
required: ['name'],
},
},
{
name: 'group_query',
description: `Run the query tool across all repos in a group and merge process results via reciprocal rank fusion.
WHEN TO USE: Semantic / hybrid search across a whole product group.`,
inputSchema: {
type: 'object',
properties: {
name: { type: 'string', description: 'Group name' },
query: { type: 'string', description: 'Search query' },
subgroup: { type: 'string', description: 'Limit to repo paths under this prefix' },
limit: { type: 'number', description: 'Max merged results (default 5)' },
},
required: ['name', 'query'],
},
},
{
name: 'group_status',
description: `Report index staleness (commit vs HEAD) and Contract Registry staleness (indexedAt) for each repo in a group.
WHEN TO USE: Before group_sync or when agents should refresh indexes.`,
inputSchema: {
type: 'object',
properties: {
name: { type: 'string', description: 'Group name' },
},
required: ['name'],
},
},
];

View file

@ -7,10 +7,50 @@
*/
import fs from 'fs/promises';
import { realpathSync } from 'fs';
import path from 'path';
import os from 'os';
import { getInferredRepoName } from './git.js';
/**
* Normalise a repo path for registry comparison across platforms
* (#664 review feedback from @evander-wang).
*
* Why this exists: `path.resolve` alone is NOT enough for
* cross-platform registry stability.
* - **macOS**: tmpdirs and `/var` are symlinks to `/private/var`.
* A child process that stored `/private/var/folders/.../repo` in
* the registry cannot later be matched by an outer caller that
* supplies the symlink form `/var/folders/.../repo`. `path.resolve`
* does not follow symlinks; `realpathSync.native` does.
* - **Windows**: GitHub runners surface tmpdirs in 8.3 short-name
* form (`RUNNERA~1\...`), but `process.cwd()` often returns the
* long form (`runneradmin\...`). `realpathSync.native` normalises
* both sides to the long-name canonical path.
*
* Fallback behaviour: if the path does not exist on disk (e.g. a user
* passed `gitnexus remove some-alias` and the alias misses every
* registry entry, or the caller is resolving a path that was deleted
* after registration), we return `path.resolve(p)` rather than
* throwing. This preserves the idempotent-on-missing semantics of
* `resolveRegistryEntry` / `remove`.
*
* Backwards compatibility: this function is applied to BOTH the
* caller-supplied input AND each stored `entry.path` at compare time
* inside `resolveRegistryEntry`, so registries written by older
* versions (where `registerRepo` only ran `path.resolve`) still match
* correctly. Newly-written entries are canonicalised at write time too
* so the registry stabilises over analyze/re-analyze cycles.
*/
export const canonicalizePath = (p: string): string => {
const resolved = path.resolve(p);
try {
return realpathSync.native(resolved);
} catch {
return resolved;
}
};
export interface RepoMeta {
repoPath: string;
lastCommit: string;
@ -359,13 +399,33 @@ export const registerRepo = async (
meta: RepoMeta,
opts?: RegisterRepoOptions,
): Promise<string> => {
// Preserve the caller's chosen path form in the registry — don't
// canonicalise at write time. This matters for two reasons:
// 1. `list` and error messages show the path the user actually
// knows (e.g. the 8.3 short form they typed), not a runtime-
// resolved long form they've never seen.
// 2. Keeps pre-existing #829 test assertions that compare
// `err.existingPath` against `path.resolve(tmpPath)` stable.
// Canonicalisation is applied at COMPARE points only (see below),
// which is where the cross-platform divergence actually matters.
const resolved = path.resolve(repoPath);
const { storagePath } = getStoragePaths(resolved);
// Canonical form used strictly for comparison — `realpathSync.native`
// expands macOS /var → /private/var and Windows 8.3 → long-name,
// falling back to `path.resolve` when the path doesn't exist.
const canonicalInput = canonicalizePath(repoPath);
const entries = await readRegistry();
const existingIdx = entries.findIndex((e) => {
const a = path.resolve(e.path);
const b = resolved;
// Canonicalise the STORED entry too so pre-canonicalisation
// registries (written by older versions, or paths passed in a
// different form) still match correctly. `canonicalizePath` falls
// back to `path.resolve` when the path no longer exists on disk,
// so stale entries that have been rm'd externally still resolve
// to a stable key instead of throwing.
const a = canonicalizePath(e.path);
const b = canonicalInput;
return process.platform === 'win32' ? a.toLowerCase() === b.toLowerCase() : a === b;
});
const existing = existingIdx >= 0 ? entries[existingIdx] : null;
@ -399,11 +459,14 @@ export const registerRepo = async (
// messages and list output #829 ships).
const explicitName = opts?.name !== undefined || isPreservedAlias;
if (explicitName && !opts?.allowDuplicateName) {
// Compare canonical-vs-canonical here too so `/var/foo` and
// `/private/var/foo` (same repo, different form) aren't treated as
// two colliding paths.
const collidingEntry = entries.find(
(e, i) =>
i !== existingIdx &&
e.name.toLowerCase() === name.toLowerCase() &&
path.resolve(e.path) !== resolved,
canonicalizePath(e.path) !== canonicalInput,
);
if (collidingEntry) {
throw new RegistryNameCollisionError(name, collidingEntry.path, resolved);
@ -435,12 +498,204 @@ export const registerRepo = async (
* Called after `gitnexus clean`.
*/
export const unregisterRepo = async (repoPath: string): Promise<void> => {
const resolved = path.resolve(repoPath);
// Canonicalise BOTH sides so an unregister call issued with the
// symlink form (`/var/folders/.../repo`) still matches an entry
// written with the realpath form (`/private/var/folders/.../repo`),
// and vice versa. Matches the semantics of `registerRepo` and
// `resolveRegistryEntry` post-#1003 review.
const resolved = canonicalizePath(repoPath);
const entries = await readRegistry();
const filtered = entries.filter((e) => path.resolve(e.path) !== resolved);
const matches = (a: string, b: string) =>
process.platform === 'win32' ? a.toLowerCase() === b.toLowerCase() : a === b;
const filtered = entries.filter((e) => !matches(canonicalizePath(e.path), resolved));
await writeRegistry(filtered);
};
/**
* Thrown by {@link resolveRegistryEntry} when no registered repo matches
* the caller's target string (by alias, basename, remote-inferred name,
* or resolved path). CLI callers that want idempotent "remove" semantics
* should catch this and exit 0 with a warning; non-idempotent callers
* (e.g. MCP tools) can surface the error directly.
*/
export class RegistryNotFoundError extends Error {
readonly kind = 'RegistryNotFoundError' as const;
constructor(
public readonly target: string,
public readonly availableNames: string[],
) {
const hint =
availableNames.length > 0
? ` Available: ${availableNames.join(', ')}.`
: ' No repositories are currently registered.';
super(`No registered repo matches "${target}".${hint}`);
this.name = 'RegistryNotFoundError';
}
}
/**
* Thrown by {@link resolveRegistryEntry} when the target string matches
* the `name` of two or more entries — only possible when the user
* previously registered duplicates via `analyze --name X
* --allow-duplicate-name` (#829). The error carries enough information
* for the caller to render an actionable disambiguation hint without
* string-matching on `.message`.
*
* `kind` is a string literal discriminant (same pattern as
* {@link RegistryNameCollisionError}) so callers can narrow via
* `err.kind === 'RegistryAmbiguousTargetError'` without importing the
* class.
*/
export class RegistryAmbiguousTargetError extends Error {
readonly kind = 'RegistryAmbiguousTargetError' as const;
constructor(
public readonly target: string,
public readonly matches: RegistryEntry[],
) {
const listing = matches.map((m) => ` - ${m.name} (${m.path})`).join('\n');
super(
`Multiple registered repos match "${target}":\n${listing}\n` +
`Pass the absolute path instead to disambiguate.`,
);
this.name = 'RegistryAmbiguousTargetError';
}
}
/**
* Thrown by {@link assertSafeStoragePath} when a registry entry's
* `storagePath` does NOT point at the expected `<entry.path>/.gitnexus`
* subfolder. CLI destructive commands (`remove`, `clean --all`) should
* catch this and exit non-zero without deleting anything — the usual
* cause is a corrupted or hand-edited `~/.gitnexus/registry.json`, and
* proceeding would mean `fs.rm(recursive: true)` on whatever odd path
* the entry is pointing at.
*/
export class UnsafeStoragePathError extends Error {
readonly kind = 'UnsafeStoragePathError' as const;
constructor(
public readonly entry: RegistryEntry,
public readonly expectedStoragePath: string,
public readonly actualStoragePath: string,
) {
super(
`Refusing to remove storage path for safety: expected ` +
`"${expectedStoragePath}" under the repo's .gitnexus subfolder, ` +
`but the registry entry has "${actualStoragePath}". ` +
`This usually means the registry entry is corrupted or was ` +
`hand-edited. Delete the entry manually from ~/.gitnexus/registry.json ` +
`and re-run analyze.`,
);
this.name = 'UnsafeStoragePathError';
}
}
/**
* Guard rail for destructive CLI paths (`remove` #664,
* `clean --all` #258, future MCP `remove` tool): verify that a
* registry entry's `storagePath` is the canonical `<repo>/.gitnexus`
* subfolder of its `path`. If not, throw {@link UnsafeStoragePathError}
* so the caller exits without touching disk.
*
* Why this exists (#1003 review — @magyargergo):
* - `~/.gitnexus/registry.json` is a plain-text user-writable file.
* A corrupted, hand-edited, or downgrade/upgrade-racing entry
* could plausibly end up with `storagePath === ""` (resolves to
* cwd), `storagePath === path` (the repo root!), `storagePath`
* equal to a parent/sibling of the repo, or simply any arbitrary
* filesystem path.
* - `fs.rm(recursive: true, force: true)` on ANY of those would be
* a runtime disaster — at best delete the user's working tree, at
* worst nuke an unrelated directory tree they happen to own.
* - `clean` (default, cwd-scoped) is safe by construction — it
* re-derives storagePath from `findRepo(cwd)` and never trusts
* the registry field. But `clean --all` DOES iterate the registry
* and trust each entry's stored storagePath (same shape as
* `remove`), so this helper must be wired into that loop too.
* - `server/api.ts` recomputes storagePath from `getStoragePath(entry.path)`
* and so is likewise safe-by-construction.
*
* Pure string check — does NOT require the paths to exist on disk.
* Windows: case-insensitive; POSIX: case-sensitive. Matches the
* comparison shape used elsewhere in this module.
*/
export const assertSafeStoragePath = (entry: RegistryEntry): void => {
const expected = path.join(path.resolve(entry.path), '.gitnexus');
const actual = path.resolve(entry.storagePath);
const matches =
process.platform === 'win32'
? expected.toLowerCase() === actual.toLowerCase()
: expected === actual;
if (!matches) {
throw new UnsafeStoragePathError(entry, expected, actual);
}
};
/**
* Resolve a user-supplied target string (from `gitnexus remove <target>`
* or equivalent MCP tool argument) to a single registry entry.
*
* Match precedence (first hit wins, subsequent tiers are only tried if
* the prior tier produces zero matches):
* 1. Exact resolved-path match (Windows: case-insensitive).
* Paths are unique by registry construction, so a path match can
* never be ambiguous.
* 2. Exact `name` match (case-insensitive). If ≥ 2 entries share the
* name — only possible via `--allow-duplicate-name` (#829) —
* throws {@link RegistryAmbiguousTargetError}.
*
* No fuzzy / partial matching — unambiguous, scriptable behaviour is
* more important than convenience for destructive commands.
*
* Throws {@link RegistryNotFoundError} if no entry matches.
*
* `entries` is passed in (rather than re-read) so callers that already
* hold the registry snapshot (e.g. to print a "before" state) can avoid
* a second disk read, and so tests can inject fixtures without touching
* `GITNEXUS_HOME`.
*/
export const resolveRegistryEntry = (entries: RegistryEntry[], target: string): RegistryEntry => {
// Tier 1: path match. Canonicalise BOTH sides so symlink and
// Windows-8.3 quirks don't cause a false miss — e.g. the caller
// passes `/var/folders/.../repo` while the registry has
// `/private/var/folders/.../repo` (both resolve to the same
// `realpath.native`). See `canonicalizePath` for the rationale.
//
// Canonicalising the STORED entry (not just the input) is what gives
// us backward-compat for registries written by versions that only
// ran `path.resolve` — both get canonicalised here at compare time.
const canonicalTarget = canonicalizePath(target);
const pathMatch = entries.find((e) => {
const a = canonicalizePath(e.path);
const b = canonicalTarget;
return process.platform === 'win32' ? a.toLowerCase() === b.toLowerCase() : a === b;
});
if (pathMatch) return pathMatch;
// Tier 2: name match. Case-insensitive on all platforms — registry
// name collisions are already filtered case-insensitively in
// `registerRepo`, so "APP" vs "app" are considered the same key.
const targetLower = target.toLowerCase();
const nameMatches = entries.filter((e) => e.name.toLowerCase() === targetLower);
if (nameMatches.length === 1) return nameMatches[0];
if (nameMatches.length > 1) {
throw new RegistryAmbiguousTargetError(target, nameMatches);
}
// Tier 3: miss. Build the available-names hint ONCE; resolveRepo-style
// disambiguated labels (`app (/path)`) are applied when the same name
// appears in multiple entries so the user sees the same hint shape as
// `-r <name>` errors.
const nameCounts = new Map<string, number>();
for (const e of entries) {
const key = e.name.toLowerCase();
nameCounts.set(key, (nameCounts.get(key) ?? 0) + 1);
}
const availableNames = entries.map((e) =>
(nameCounts.get(e.name.toLowerCase()) ?? 0) > 1 ? `${e.name} (${e.path})` : e.name,
);
throw new RegistryNotFoundError(target, availableNames);
};
/**
* List all registered repos from the global registry.
* Optionally validates that each entry's .gitnexus/ still exists.

View file

@ -345,6 +345,410 @@ describe('CLI end-to-end', () => {
}, 360000); // 6-min outer budget (4 × ~60s analyze calls + fixture setup)
});
// ─── gitnexus remove <target> (#664) ─────────────────────────────
//
// End-to-end regression guard for the remove command:
// 1. `remove <alias>` without --force is a dry-run (exit 0, preserves state)
// 2. `remove <alias> --force` deletes the .gitnexus/ directory
// AND unregisters from the global registry
// 3. `remove <unknown>` is idempotent (exit 0 with a warning)
// 4. `remove <ambiguous>` (two entries share the alias via
// --allow-duplicate-name) exits 1 with a disambiguation hint
// and leaves the registry unchanged.
//
// Every assertion reads the real registry.json on disk, so any
// regression in remove.ts → resolveRegistryEntry → unregisterRepo
// will surface here.
describe('remove <target> (#664)', () => {
it('dry-run lists, --force deletes, missing target is a no-op warning', () => {
const gnHome = fs.mkdtempSync(path.join(os.tmpdir(), 'gn-home-remove-'));
const repoA = makeMiniRepoCopy('remove-me', 'gn-rm-a-');
const parentA = path.dirname(repoA);
try {
// Index the repo under a custom alias so we can target it by
// name below. `--name` guarantees a stable alias regardless of
// how the host resolves the basename/remote-inferred name.
const r1 = runCliWithEnv(
['analyze', '--name', 'alias-a'],
repoA,
{ GITNEXUS_HOME: gnHome },
60000,
);
if (r1.status === null) return;
expect(
r1.status,
[`analyze exited with ${r1.status}`, `stdout: ${r1.stdout}`, `stderr: ${r1.stderr}`].join(
'\n',
),
).toBe(0);
const registryPath = path.join(gnHome, 'registry.json');
const afterIndex = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(afterIndex).toHaveLength(1);
expect(afterIndex[0].name).toBe('alias-a');
// Storage dir must exist before remove so we can assert its
// disappearance below.
const storagePath = afterIndex[0].storagePath;
expect(fs.existsSync(storagePath)).toBe(true);
// Dry-run: must NOT delete. Use parentA as cwd so the test
// never runs with the to-be-removed storage dir as its cwd.
//
// Assert the FULL dry-run output shape, not just the `--force`
// hint (#1003 senior-reviewer NIT): `remove.ts` prints the
// alias, the resolved path, AND the storage path. Verifying
// all three appear catches silent format regressions
// (e.g. a future refactor that accidentally drops one of the
// three `console.log` lines, or swaps `entry.path` for
// `entry.name` in the output).
const r2 = runCliWithEnv(['remove', 'alias-a'], parentA, { GITNEXUS_HOME: gnHome }, 15000);
if (r2.status === null) return;
expect(r2.status).toBe(0);
const r2Output = `${r2.stdout}${r2.stderr}`;
expect(r2Output).toMatch(/Run with --force/i);
expect(r2Output, 'dry-run must surface the alias').toContain('alias-a');
expect(r2Output, 'dry-run must surface the repo path').toContain(afterIndex[0].path);
expect(r2Output, 'dry-run must surface the storage path').toContain(storagePath);
expect(fs.existsSync(storagePath)).toBe(true);
// Registry still has the entry.
expect(JSON.parse(fs.readFileSync(registryPath, 'utf-8'))).toHaveLength(1);
// --force: must delete storage AND unregister.
const r3 = runCliWithEnv(
['remove', 'alias-a', '--force'],
parentA,
{ GITNEXUS_HOME: gnHome },
15000,
);
if (r3.status === null) return;
expect(
r3.status,
[
`remove --force exited with ${r3.status}`,
`stdout: ${r3.stdout}`,
`stderr: ${r3.stderr}`,
].join('\n'),
).toBe(0);
// Success-case output shape: `Removed: <alias>` header plus the
// same path-and-storagePath lines the dry-run prints (same NIT
// rationale — the success branch mirrors the dry-run's three
// console.log calls, so it has the same silent-regression risk).
const r3Output = `${r3.stdout}${r3.stderr}`;
expect(r3Output).toMatch(/Removed/i);
expect(r3Output, 'success output must surface the alias').toContain('alias-a');
expect(r3Output, 'success output must surface the repo path').toContain(afterIndex[0].path);
expect(r3Output, 'success output must surface the storage path').toContain(storagePath);
expect(fs.existsSync(storagePath)).toBe(false);
expect(JSON.parse(fs.readFileSync(registryPath, 'utf-8'))).toHaveLength(0);
// Idempotent: removing the same alias AGAIN must exit 0 with a
// warning (so `remove X && analyze Y` keeps working in scripts).
const r4 = runCliWithEnv(['remove', 'alias-a'], parentA, { GITNEXUS_HOME: gnHome }, 15000);
if (r4.status === null) return;
expect(r4.status).toBe(0);
expect(`${r4.stdout}${r4.stderr}`).toMatch(/Nothing to remove/i);
} finally {
fs.rmSync(gnHome, { recursive: true, force: true });
fs.rmSync(parentA, { recursive: true, force: true });
}
}, 180000); // 3-min outer budget (1 × ~60s analyze + 3 × fast remove calls)
it('ambiguous target (two entries share alias via --allow-duplicate-name) errors without mutating registry', () => {
const gnHome = fs.mkdtempSync(path.join(os.tmpdir(), 'gn-home-rm-amb-'));
const repoA = makeMiniRepoCopy('dup', 'gn-dup-a-');
const repoB = makeMiniRepoCopy('dup', 'gn-dup-b-');
const parentA = path.dirname(repoA);
const parentB = path.dirname(repoB);
try {
// Two repos registered under the same alias — only possible via
// --allow-duplicate-name (#829).
const r1 = runCliWithEnv(
['analyze', '--name', 'shared'],
repoA,
{ GITNEXUS_HOME: gnHome },
60000,
);
if (r1.status === null) return;
expect(r1.status).toBe(0);
const r2 = runCliWithEnv(
['analyze', '--name', 'shared', '--allow-duplicate-name'],
repoB,
{ GITNEXUS_HOME: gnHome },
60000,
);
if (r2.status === null) return;
expect(r2.status).toBe(0);
const registryPath = path.join(gnHome, 'registry.json');
const before = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(before).toHaveLength(2);
// `remove shared` must refuse to guess — exit 1, disambiguation hint.
const r3 = runCliWithEnv(
['remove', 'shared', '--force'],
parentA,
{ GITNEXUS_HOME: gnHome },
15000,
);
if (r3.status === null) return;
expect(r3.status).toBe(1);
const r3Output = `${r3.stdout}${r3.stderr}`;
expect(r3Output).toMatch(/Multiple registered repos match/i);
// Both paths must be surfaced in the hint so the user knows
// which ones to disambiguate between.
expect(r3Output).toMatch(/dup/);
// Registry unchanged — the failed resolution must NOT have
// mutated state.
const after = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(after).toHaveLength(2);
// And path-based remove still works: pass the absolute path of
// repoA and it resolves unambiguously.
//
// We pull the path from the registry snapshot rather than
// passing the outer `repoA` variable directly. This is the
// belt-and-suspenders for cross-platform path normalisation
// (#1003 review): the path the registry recorded has already
// gone through the analyze-side canonicalisation (which on
// macOS expands /var → /private/var and on Windows expands 8.3
// → long-name). Passing that exact string back to `remove`
// guarantees the comparison succeeds even on runners where the
// outer `repoA` is the symlink/short-name form. The code-side
// fix in `canonicalizePath` makes this redundant in practice,
// but the test shouldn't depend on the code fix being perfect
// on every platform — it should prove correctness against the
// registry contract.
const repoAEntry = before.find(
(e: { path: string }) =>
path.basename(e.path) === 'dup' && e.path.includes(path.basename(parentA)),
);
expect(
repoAEntry,
'repoA entry must exist in registry before path-remove step',
).toBeDefined();
const r4 = runCliWithEnv(
['remove', repoAEntry.path, '--force'],
parentA,
{ GITNEXUS_HOME: gnHome },
15000,
);
if (r4.status === null) return;
expect(
r4.status,
[
`remove-by-path exited with ${r4.status}`,
`stdout: ${r4.stdout}`,
`stderr: ${r4.stderr}`,
].join('\n'),
).toBe(0);
const finalEntries = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(finalEntries).toHaveLength(1);
// The survivor is repoB (its path stays in the registry).
expect(path.basename(finalEntries[0].path)).toBe('dup');
// And it's NOT the one we just removed.
expect(finalEntries[0].path).not.toBe(repoAEntry.path);
} finally {
fs.rmSync(gnHome, { recursive: true, force: true });
fs.rmSync(parentA, { recursive: true, force: true });
fs.rmSync(parentB, { recursive: true, force: true });
}
}, 240000); // 4-min outer budget (2 × ~60s analyze + 2 × fast remove)
it('refuses to proceed when a registry entry points storagePath outside <repo>/.gitnexus (#1003)', () => {
// Regression guard for the safety gap flagged by @magyargergo on
// PR #1003: `~/.gitnexus/registry.json` is a user-writable JSON
// file, so a corrupted or hand-edited entry could point
// storagePath at the repo root (catastrophic: rm the working
// tree) or at any other arbitrary path. `remove --force` must
// refuse to call fs.rm when storagePath isn't the canonical
// `<entry.path>/.gitnexus`. We verify:
// 1. Exit code 1 with the actionable "registry entry corrupted"
// hint.
// 2. The .gitnexus/ storage dir is UNTOUCHED.
// 3. The repo itself (entry.path) is UNTOUCHED.
// 4. The registry entry is NOT removed (no partial mutation).
const gnHome = fs.mkdtempSync(path.join(os.tmpdir(), 'gn-home-poison-'));
const repo = makeMiniRepoCopy('poisoned', 'gn-poison-');
const parent = path.dirname(repo);
try {
// Index the repo normally first so the registry has a valid
// entry we can then poison.
const r1 = runCliWithEnv(
['analyze', '--name', 'poisoned-alias'],
repo,
{ GITNEXUS_HOME: gnHome },
60000,
);
if (r1.status === null) return;
expect(r1.status).toBe(0);
const registryPath = path.join(gnHome, 'registry.json');
const original = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(original).toHaveLength(1);
// Poison the entry: set storagePath to the REPO ROOT itself.
// If the guard isn't in place, `remove --force` would call
// `fs.rm(repo, {recursive: true, force: true})` and wipe the
// entire working tree.
const poisoned = [{ ...original[0], storagePath: repo }];
fs.writeFileSync(registryPath, JSON.stringify(poisoned, null, 2));
// Sanity: storage dir and working tree both still exist.
expect(fs.existsSync(path.join(repo, '.gitnexus'))).toBe(true);
expect(fs.existsSync(repo)).toBe(true);
expect(fs.existsSync(path.join(repo, '.git'))).toBe(true);
// Attempt the remove — must FAIL without deleting anything.
const r2 = runCliWithEnv(
['remove', 'poisoned-alias', '--force'],
parent,
{ GITNEXUS_HOME: gnHome },
15000,
);
if (r2.status === null) return;
expect(
r2.status,
[`remove should have exited 1`, `stdout: ${r2.stdout}`, `stderr: ${r2.stderr}`].join(
'\n',
),
).toBe(1);
const r2Output = `${r2.stdout}${r2.stderr}`;
// Must surface the actionable "registry corrupted" hint, not
// just a raw fs.rm error.
expect(r2Output).toMatch(/Refusing to remove/i);
expect(r2Output).toMatch(/registry\.json/i);
// Repo + .gitnexus dir + .git dir must all still exist — the
// guard aborts BEFORE fs.rm. This is the whole point of the
// test: the working tree is not allowed to disappear.
expect(fs.existsSync(repo), 'repo working tree must survive').toBe(true);
expect(fs.existsSync(path.join(repo, '.gitnexus')), 'storage dir must survive').toBe(true);
expect(fs.existsSync(path.join(repo, '.git')), '.git must survive').toBe(true);
// Registry unchanged — no partial mutation.
const afterRegistry = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(afterRegistry).toHaveLength(1);
expect(afterRegistry[0].storagePath).toBe(repo); // still poisoned (we did that)
} finally {
fs.rmSync(gnHome, { recursive: true, force: true });
fs.rmSync(parent, { recursive: true, force: true });
}
}, 120000); // 2-min budget (1 × ~60s analyze + 1 × fast remove-refused)
});
// ─── clean --all: same safety guard applies (#1003 review) ───────
//
// The `clean --all` path iterates over the registry and calls
// `fs.rm(entry.storagePath)` — identical trust-the-registry pattern
// as `remove` had before the guard. A poisoned entry must be SKIPPED
// (not aborted), so clean --all preserves its existing per-repo
// error-tolerance semantics: one bad entry does not halt cleanup of
// the rest. We verify:
// 1. The poisoned entry is NOT deleted (working tree + .gitnexus
// survive), and the CLI prints a "Refusing to clean" message.
// 2. The poisoned entry is left in the registry (nothing was
// mutated for it).
// 3. A co-existing well-formed entry IS still cleaned (both its
// .gitnexus dir AND its registry entry are gone).
describe('clean --all with a poisoned registry entry (#1003)', () => {
it('skips poisoned entries, cleans valid ones, never deletes the working tree', () => {
const gnHome = fs.mkdtempSync(path.join(os.tmpdir(), 'gn-home-clean-poison-'));
const repoBad = makeMiniRepoCopy('bad-repo', 'gn-clean-bad-');
const repoGood = makeMiniRepoCopy('good-repo', 'gn-clean-good-');
const parentBad = path.dirname(repoBad);
const parentGood = path.dirname(repoGood);
try {
// Analyze both so the registry has two well-formed entries.
for (const [repo, alias] of [
[repoBad, 'bad-alias'],
[repoGood, 'good-alias'],
] as const) {
const r = runCliWithEnv(
['analyze', '--name', alias],
repo,
{ GITNEXUS_HOME: gnHome },
60000,
);
if (r.status === null) return;
expect(r.status, `analyze ${alias} exited ${r.status}: ${r.stdout}${r.stderr}`).toBe(0);
}
const registryPath = path.join(gnHome, 'registry.json');
const original = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(original).toHaveLength(2);
// Poison the 'bad-alias' entry by pointing its storagePath at
// the repo root itself. If the guard isn't wired into the
// clean --all loop, `clean --all --force` would fs.rm the
// working tree.
const poisoned = original.map((e: { name: string; storagePath: string; path: string }) =>
e.name === 'bad-alias' ? { ...e, storagePath: repoBad } : e,
);
fs.writeFileSync(registryPath, JSON.stringify(poisoned, null, 2));
// Sanity: both working trees and .gitnexus dirs still exist.
expect(fs.existsSync(repoBad)).toBe(true);
expect(fs.existsSync(path.join(repoBad, '.gitnexus'))).toBe(true);
expect(fs.existsSync(path.join(repoBad, '.git'))).toBe(true);
expect(fs.existsSync(path.join(repoGood, '.gitnexus'))).toBe(true);
// clean --all --force from a neutral cwd (parentBad), so the
// command isn't "inside" either repo.
const r = runCliWithEnv(
['clean', '--all', '--force'],
parentBad,
{ GITNEXUS_HOME: gnHome },
30000,
);
if (r.status === null) return;
// clean --all's per-entry error handling always exits 0 at
// the end (it only logs per-repo failures). The important
// assertions are on side effects, not the exit code.
const output = `${r.stdout}${r.stderr}`;
expect(output).toMatch(/Refusing to clean/i);
expect(output).toMatch(/bad-alias/);
// Poisoned repo: working tree + .gitnexus + .git all SURVIVE.
expect(fs.existsSync(repoBad), 'poisoned repo working tree must survive').toBe(true);
expect(
fs.existsSync(path.join(repoBad, '.gitnexus')),
'poisoned repo .gitnexus must survive (guard refused to rm repo root)',
).toBe(true);
expect(fs.existsSync(path.join(repoBad, '.git')), '.git must survive').toBe(true);
// Good repo: its .gitnexus IS gone (cleanup succeeded despite
// the poisoned sibling entry — per-entry error tolerance is
// preserved).
expect(
fs.existsSync(path.join(repoGood, '.gitnexus')),
'good repo .gitnexus should be cleaned',
).toBe(false);
// But the good repo's working tree stays (clean never touches
// anything outside .gitnexus).
expect(fs.existsSync(repoGood), 'good repo working tree must survive').toBe(true);
// Registry post-state: poisoned entry still present (skipped,
// not mutated); good entry unregistered.
const afterRegistry = JSON.parse(fs.readFileSync(registryPath, 'utf-8'));
expect(afterRegistry).toHaveLength(1);
expect(afterRegistry[0].name).toBe('bad-alias');
} finally {
fs.rmSync(gnHome, { recursive: true, force: true });
fs.rmSync(parentBad, { recursive: true, force: true });
fs.rmSync(parentGood, { recursive: true, force: true });
}
}, 240000); // 4-min budget (2 × ~60s analyze + 1 × fast clean --all)
});
describe('unhappy path', () => {
it('exits with error when no command is given', () => {
const result = runCliRaw([], MINI_REPO);

View file

@ -65,4 +65,51 @@ describe('group CLI', () => {
const blanketClosePattern = /closeLbug\s*\(\s*\)/;
expect(source).not.toMatch(blanketClosePattern);
});
it('group impact requires --target and --repo', () => {
const c = runGroup(['create', 'impcli']);
expect(c.status).toBe(0);
const r = runGroup(['impact', 'impcli']);
expect(r.status).not.toBe(0);
});
it('group impact runs with Issue #794 style flags (fixture-backed home)', () => {
const home = fs.mkdtempSync(path.join(os.tmpdir(), 'gitnexus-cli-impact-'));
try {
const gd = path.join(home, 'groups', 'test-group');
fs.mkdirSync(gd, { recursive: true });
fs.copyFileSync(
path.join(repoRoot, 'test', 'fixtures', 'group', 'group.yaml'),
path.join(gd, 'group.yaml'),
);
const r = spawnSync(
process.execPath,
[
'--import',
tsxImportUrl,
cliEntry,
'group',
'impact',
'test-group',
'--target',
'health',
'--repo',
'app/backend',
'--json',
],
{
cwd: repoRoot,
encoding: 'utf8',
timeout: 20000,
stdio: ['pipe', 'pipe', 'pipe'],
env: { ...process.env, GITNEXUS_HOME: home },
},
);
expect(r.status).not.toBe(0);
const msg = `${r.stderr}\n${r.stdout}`;
expect(msg).toMatch(/error|indexed|not found|repository/i);
} finally {
fs.rmSync(home, { recursive: true, force: true });
}
});
});

View file

@ -0,0 +1,74 @@
/**
* Group impact: exercise GroupService.groupImpact with fixture-backed group config
* and a stubbed port (no LadybugDB / bridge required when local impact yields no UIDs).
*/
import { describe, it, expect, vi, beforeAll, afterAll } from 'vitest';
import * as fs from 'node:fs';
import * as path from 'node:path';
import { fileURLToPath } from 'node:url';
import os from 'node:os';
import { GroupService, type GroupToolPort } from '../../../src/core/group/service.js';
const __dirname = path.dirname(fileURLToPath(import.meta.url));
const fixturesDir = path.resolve(__dirname, '../../fixtures/group');
let tmpHome: string;
beforeAll(() => {
tmpHome = fs.mkdtempSync(path.join(os.tmpdir(), 'gitnexus-grp-impact-int-'));
const groupDir = path.join(tmpHome, 'groups', 'test-group');
fs.mkdirSync(groupDir, { recursive: true });
fs.copyFileSync(path.join(fixturesDir, 'group.yaml'), path.join(groupDir, 'group.yaml'));
});
afterAll(() => {
if (tmpHome) fs.rmSync(tmpHome, { recursive: true, force: true });
});
function stubPort(): GroupToolPort {
return {
resolveRepo: vi.fn(async () => ({
id: 'stub',
name: 'stub',
repoPath: '/tmp/repo',
storagePath: '/tmp/.gitnexus',
})),
impact: vi.fn(async () => ({
target: {},
byDepth: {},
summary: { direct: 0, processes_affected: 0, modules_affected: 0 },
risk: 'LOW',
})),
query: vi.fn(),
impactByUid: vi.fn(),
context: vi.fn(),
};
}
describe('group impact integration', () => {
it('returns validation error when parameters are incomplete', async () => {
const svc = new GroupService(stubPort());
const r = (await svc.groupImpact({ name: 'x', direction: 'upstream' })) as { error: string };
expect(r.error).toMatch(/repo is required|target is required/);
});
it('runs happy-path stub against fixture group (stops before bridge when no symbol UIDs)', async () => {
const prev = process.env.GITNEXUS_HOME;
process.env.GITNEXUS_HOME = tmpHome;
try {
const svc = new GroupService(stubPort());
const r = (await svc.groupImpact({
name: 'test-group',
repo: 'app/backend',
target: 'health',
direction: 'upstream',
})) as { group?: string; error?: string; cross?: unknown[] };
expect(r.error).toBeUndefined();
expect(r.group).toBe('test-group');
expect(Array.isArray(r.cross)).toBe(true);
} finally {
if (prev === undefined) delete process.env.GITNEXUS_HOME;
else process.env.GITNEXUS_HOME = prev;
}
});
});

View file

@ -1,5 +1,33 @@
import { describe, it, expect } from 'vitest';
import { searchFTSFromLbug, type BM25SearchResult } from '../../src/core/search/bm25-index.js';
import { describe, it, expect, vi, beforeEach } from 'vitest';
import {
searchFTSFromLbug,
invalidateEnsuredFTSForRepo,
type BM25SearchResult,
} from '../../src/core/search/bm25-index.js';
vi.mock('../../src/core/lbug/lbug-adapter.js', async (importOriginal) => {
const actual = await importOriginal<typeof import('../../src/core/lbug/lbug-adapter.js')>();
return {
...actual,
queryFTS: vi.fn().mockResolvedValue([]),
};
});
// Pool adapter is dynamically imported by the MCP-pool path of
// `searchFTSFromLbug`. We mock it so we can drive the executor and the
// pool-close listener without spinning up a real LadybugDB pool.
const poolCloseListeners: Array<(repoId: string) => void> = [];
const mockExecuteQuery = vi.fn();
vi.mock('../../src/core/lbug/pool-adapter.js', () => ({
executeQuery: (repoId: string, cypher: string) => mockExecuteQuery(repoId, cypher),
addPoolCloseListener: (listener: (repoId: string) => void) => {
poolCloseListeners.push(listener);
return () => {
const idx = poolCloseListeners.indexOf(listener);
if (idx !== -1) poolCloseListeners.splice(idx, 1);
};
},
}));
describe('BM25 search', () => {
describe('searchFTSFromLbug', () => {
@ -32,5 +60,255 @@ describe('BM25 search', () => {
expect(result.score).toBe(1.5);
expect(result.rank).toBe(1);
});
it('accepts optional nodeIds field', () => {
const result: BM25SearchResult = {
filePath: 'src/index.ts',
score: 1.5,
rank: 1,
nodeIds: ['func:id1', 'func:id2'],
};
expect(result.nodeIds).toEqual(['func:id1', 'func:id2']);
});
});
describe('score aggregation', () => {
beforeEach(() => {
vi.clearAllMocks();
});
it('sums only top-3 scoring nodes per file when more than 3 match', async () => {
const { queryFTS } = await import('../../src/core/lbug/lbug-adapter.js');
// File table: empty; Function table: 5 hits for the same file; rest: empty
vi.mocked(queryFTS)
.mockResolvedValueOnce([]) // File
.mockResolvedValueOnce([
// Function — 5 hits, scores 10/9/8/7/6
{ filePath: 'src/views.py', score: 10, nodeId: 'func:node1', name: 'get_queryset' },
{ filePath: 'src/views.py', score: 9, nodeId: 'func:node2', name: 'post' },
{ filePath: 'src/views.py', score: 8, nodeId: 'func:node3', name: 'delete' },
{ filePath: 'src/views.py', score: 7, nodeId: 'func:node4', name: 'patch' },
{ filePath: 'src/views.py', score: 6, nodeId: 'func:node5', name: 'put' },
])
.mockResolvedValueOnce([]) // Class
.mockResolvedValueOnce([]) // Method
.mockResolvedValueOnce([]); // Interface
const results = await searchFTSFromLbug('queryset');
expect(results).toHaveLength(1);
expect(results[0].filePath).toBe('src/views.py');
// Only top-3 scores (10+9+8=27), not naive sum of all 5 (10+9+8+7+6=40)
expect(results[0].score).toBe(27);
expect(results[0].nodeIds).toEqual(['func:node1', 'func:node2', 'func:node3']);
});
it('propagates nodeIds for files with fewer than 3 matching nodes', async () => {
const { queryFTS } = await import('../../src/core/lbug/lbug-adapter.js');
vi.mocked(queryFTS)
.mockResolvedValueOnce([]) // File
.mockResolvedValueOnce([
// Function — 2 hits
{ filePath: 'src/models.py', score: 5, nodeId: 'func:m1', name: 'save' },
{ filePath: 'src/models.py', score: 3, nodeId: 'func:m2', name: 'delete' },
])
.mockResolvedValueOnce([]) // Class
.mockResolvedValueOnce([]) // Method
.mockResolvedValueOnce([]); // Interface
const results = await searchFTSFromLbug('model');
expect(results).toHaveLength(1);
expect(results[0].score).toBe(8); // 5+3
expect(results[0].nodeIds).toEqual(['func:m1', 'func:m2']);
});
it('filters out empty nodeIds', async () => {
const { queryFTS } = await import('../../src/core/lbug/lbug-adapter.js');
vi.mocked(queryFTS)
.mockResolvedValueOnce([]) // File
.mockResolvedValueOnce([
// Function — nodes with no id
{ filePath: 'src/utils.py', score: 5, nodeId: '', name: 'helper' },
{ filePath: 'src/utils.py', score: 3, nodeId: '', name: 'util' },
])
.mockResolvedValueOnce([]) // Class
.mockResolvedValueOnce([]) // Method
.mockResolvedValueOnce([]); // Interface
const results = await searchFTSFromLbug('util');
expect(results).toHaveLength(1);
expect(results[0].nodeIds).toEqual([]);
});
it('merges hits across multiple index tables for the same file', async () => {
const { queryFTS } = await import('../../src/core/lbug/lbug-adapter.js');
vi.mocked(queryFTS)
.mockResolvedValueOnce([
// File table
{ filePath: 'src/auth.py', score: 4, nodeId: 'file:auth', name: 'auth.py' },
])
.mockResolvedValueOnce([
// Function table
{ filePath: 'src/auth.py', score: 9, nodeId: 'func:login', name: 'login' },
])
.mockResolvedValueOnce([
// Class table
{ filePath: 'src/auth.py', score: 7, nodeId: 'cls:User', name: 'User' },
])
.mockResolvedValueOnce([]) // Method
.mockResolvedValueOnce([]); // Interface
const results = await searchFTSFromLbug('auth');
expect(results).toHaveLength(1);
// All 3 hits (scores 9+7+4=20) — each from a different table, all top-3
expect(results[0].score).toBe(20);
expect(results[0].nodeIds).toEqual(['func:login', 'cls:User', 'file:auth']);
});
it('ranks files by aggregated score descending', async () => {
const { queryFTS } = await import('../../src/core/lbug/lbug-adapter.js');
vi.mocked(queryFTS)
.mockResolvedValueOnce([]) // File
.mockResolvedValueOnce([
// Function — hits across two files
{ filePath: 'src/low.py', score: 2, nodeId: 'func:a', name: 'a' },
{ filePath: 'src/high.py', score: 9, nodeId: 'func:b', name: 'b' },
])
.mockResolvedValueOnce([]) // Class
.mockResolvedValueOnce([]) // Method
.mockResolvedValueOnce([]); // Interface
const results = await searchFTSFromLbug('fn');
expect(results[0].filePath).toBe('src/high.py');
expect(results[1].filePath).toBe('src/low.py');
expect(results[0].rank).toBe(1);
expect(results[1].rank).toBe(2);
});
});
describe('ensureFTS cache (MCP pool path)', () => {
const REPO = 'test-repo-fts-cache';
beforeEach(() => {
// Clean state so cases don't bleed into each other.
mockExecuteQuery.mockReset();
invalidateEnsuredFTSForRepo(REPO);
// Suppress the surfaced warn so test output stays readable.
vi.spyOn(console, 'warn').mockImplementation(() => {});
});
it('does NOT cache a transient CREATE_FTS_INDEX failure — second call retries', async () => {
// First call: every CREATE_FTS_INDEX fails transiently; QUERY_FTS_INDEX returns nothing.
mockExecuteQuery.mockImplementation(async (_repo: string, cypher: string) => {
if (cypher.includes('CREATE_FTS_INDEX')) {
throw new Error('transient lock error: Could not set lock');
}
return [];
});
const r1 = await searchFTSFromLbug('anything', 5, REPO);
expect(Array.isArray(r1)).toBe(true);
const createCallsAfterFirst = mockExecuteQuery.mock.calls.filter((c) =>
String(c[1]).includes('CREATE_FTS_INDEX'),
).length;
// 5 FTS index tables — all five attempted on first call.
expect(createCallsAfterFirst).toBe(5);
// Second call: CREATE succeeds this time. The bug being fixed: if the
// first failure was cached, we'd see ZERO additional CREATE calls.
mockExecuteQuery.mockReset();
mockExecuteQuery.mockResolvedValue([]);
await searchFTSFromLbug('anything', 5, REPO);
const createCallsOnRetry = mockExecuteQuery.mock.calls.filter((c) =>
String(c[1]).includes('CREATE_FTS_INDEX'),
).length;
expect(createCallsOnRetry).toBe(5);
});
it("treats 'already exists' as success and caches it (no retry on second call)", async () => {
mockExecuteQuery.mockImplementation(async (_repo: string, cypher: string) => {
if (cypher.includes('CREATE_FTS_INDEX')) {
throw new Error("Catalog exception: index 'file_fts' already exists");
}
return [];
});
await searchFTSFromLbug('anything', 5, REPO);
mockExecuteQuery.mockReset();
mockExecuteQuery.mockResolvedValue([]);
await searchFTSFromLbug('anything', 5, REPO);
const createCallsOnSecond = mockExecuteQuery.mock.calls.filter((c) =>
String(c[1]).includes('CREATE_FTS_INDEX'),
).length;
expect(createCallsOnSecond).toBe(0);
});
it('invalidateEnsuredFTSForRepo drops cached entries so next call re-issues CREATE', async () => {
// Prime the cache with successful creates.
mockExecuteQuery.mockResolvedValue([]);
await searchFTSFromLbug('anything', 5, REPO);
mockExecuteQuery.mockReset();
mockExecuteQuery.mockResolvedValue([]);
// Without invalidation: no re-CREATE.
await searchFTSFromLbug('anything', 5, REPO);
expect(
mockExecuteQuery.mock.calls.filter((c) => String(c[1]).includes('CREATE_FTS_INDEX')).length,
).toBe(0);
// After invalidation: next call re-issues CREATE for all 5 tables.
invalidateEnsuredFTSForRepo(REPO);
mockExecuteQuery.mockReset();
mockExecuteQuery.mockResolvedValue([]);
await searchFTSFromLbug('anything', 5, REPO);
expect(
mockExecuteQuery.mock.calls.filter((c) => String(c[1]).includes('CREATE_FTS_INDEX')).length,
).toBe(5);
});
it('a pool-close listener fired by the pool adapter invalidates this repo only', async () => {
const OTHER = 'other-repo';
mockExecuteQuery.mockResolvedValue([]);
// Prime both repos.
await searchFTSFromLbug('anything', 5, REPO);
await searchFTSFromLbug('anything', 5, OTHER);
// Confirm at least one listener was registered by the search module.
expect(poolCloseListeners.length).toBeGreaterThanOrEqual(1);
// Simulate the pool adapter closing REPO.
for (const l of poolCloseListeners) l(REPO);
mockExecuteQuery.mockReset();
mockExecuteQuery.mockResolvedValue([]);
await searchFTSFromLbug('anything', 5, REPO);
const createForRepo = mockExecuteQuery.mock.calls.filter(
(c) => c[0] === REPO && String(c[1]).includes('CREATE_FTS_INDEX'),
).length;
expect(createForRepo).toBe(5);
// OTHER repo's cache must remain intact — no re-CREATE for it.
mockExecuteQuery.mockReset();
mockExecuteQuery.mockResolvedValue([]);
await searchFTSFromLbug('anything', 5, OTHER);
const createForOther = mockExecuteQuery.mock.calls.filter(
(c) => c[0] === OTHER && String(c[1]).includes('CREATE_FTS_INDEX'),
).length;
expect(createForOther).toBe(0);
invalidateEnsuredFTSForRepo(OTHER);
});
});
});

View file

@ -209,11 +209,12 @@ describe('chunkNode', () => {
const result = await chunkNode('Class', content, 'test.ts', 1, 6, 90, 0);
expect(result).toHaveLength(2);
expect(result[0].text).toContain('class Parser {');
expect(result[0].text).toContain('options: ParserOptions;');
expect(result[0].text).toContain('cache: Map<string, any>;');
expect(result[1].text).toContain('parseJSON()');
expect(result[1].text).toContain('validate()');
expect(result[0].startLine).toBe(2);
expect(result[0].startLine).toBe(1);
expect(result[1].startLine).toBe(4);
});
@ -237,11 +238,44 @@ describe('chunkNode', () => {
const result = await chunkNode('Interface', content, 'test.ts', 10, 14, 500, 0);
expect(result).toHaveLength(1);
expect(result[0].text).toContain('interface Handler {');
expect(result[0].text).toContain('handle(event: Event): void;');
expect(result[0].text).toContain('validate(input: string): boolean;');
expect(result[0].text).toContain('readonly name: string;');
});
it('uses declaration-aware chunking for Struct labels', async () => {
const content = [
'struct User {',
' name: String,',
' email: String,',
' age: u32,',
' address: String,',
'}',
].join('\n');
const tree = makeDeclarationTree('struct_item', 'declaration_list', content, [
'name: String,',
'email: String,',
'age: u32,',
'address: String,',
]);
createParserForLanguage.mockResolvedValue({
parse: vi.fn().mockReturnValue(tree),
});
const result = await chunkNode('Struct', content, 'test.rs', 40, 45, 45, 0);
expect(result).toHaveLength(2);
expect(result[0].text).toContain('struct User {');
expect(result[0].text).toContain('name: String,');
expect(result[0].text).toContain('email: String');
const combinedText = result.map((chunk) => chunk.text).join('\n');
expect(combinedText).toContain('email: String');
expect(combinedText).toContain('age: u32');
expect(combinedText).toContain('address: String');
expect(result[0].startLine).toBe(40);
});
it('splits a function into multiple AST-aware chunks using snippet offsets', async () => {
const content = [
'function example() {',

View file

@ -43,6 +43,16 @@ describe('CLI help surface', () => {
expect(result.stdout).toContain('--repo <name>');
});
it('detect-changes help exposes compare scope and base-ref flags', () => {
const result = runHelp('detect-changes');
expect(result.status).toBe(0);
expect(result.stdout).toContain('gitnexus detect-changes|detect_changes [options]');
expect(result.stdout).toContain('--scope <scope>');
expect(result.stdout).toContain('--base-ref <ref>');
expect(result.stdout).toContain('--repo <name>');
});
it('wiki help shows provider, review, and verbose flags', () => {
const result = runHelp('wiki');

View file

@ -18,12 +18,19 @@ vi.mock('../../src/core/tree-sitter/parser-loader.js', () => ({
resolveLanguageKey: vi.fn((language: string) => language),
}));
vi.mock('gitnexus-shared', () => ({
const { getLanguageFromFilename } = vi.hoisted(() => ({
getLanguageFromFilename: vi.fn().mockReturnValue('typescript'),
}));
vi.mock('gitnexus-shared', () => ({
getLanguageFromFilename,
}));
import { chunkNode } from '../../src/core/embeddings/chunker.js';
const CLASS_PREV_TAIL_SAMPLE = 30;
const STRUCT_PREV_TAIL_SAMPLE = 20;
type FakeNode = {
type: string;
startIndex: number;
@ -183,10 +190,18 @@ describe('embedding-chunking integration', () => {
const chunks = await chunkNode(node.label, node.content, node.filePath, 20, 25, 90, 0);
expect(chunks).toHaveLength(2);
const secondText = generateEmbeddingText(node, chunks[1].text);
const secondText = generateEmbeddingText(
node,
chunks[1].text,
{},
chunks[1].chunkIndex,
chunks[0].text.slice(-CLASS_PREV_TAIL_SAMPLE),
);
expect(secondText).toContain('Class: Parser');
expect(secondText).toContain('Methods: parseJSON, validate');
expect(secondText).toContain('Properties: options, cache');
expect(secondText).toContain('Container: class Parser {');
expect(secondText).toContain('[preceding context]: ...');
expect(secondText).not.toContain('Methods: parseJSON, validate');
expect(secondText).not.toContain('Properties: options, cache');
expect(secondText).toContain('parseJSON(text: string)');
});
@ -220,9 +235,53 @@ describe('embedding-chunking integration', () => {
const text = generateEmbeddingText(node, chunks[0].text);
expect(text).toContain('Interface: Handler');
expect(text).toContain('Methods: handle, validate');
expect(text).toContain('Container: interface Handler {');
expect(text).toContain('readonly name: string;');
});
it('struct chunks retain structural container context', async () => {
getLanguageFromFilename.mockReturnValue('rust');
const node = makeNode({
label: 'Struct',
name: 'User',
fieldNames: ['name', 'email', 'age', 'address'],
content: `struct User {
name: String,
email: String,
age: u32,
address: String,
}`,
startLine: 40,
endLine: 45,
filePath: 'src/user.rs',
});
createParserForLanguage.mockResolvedValue({
parse: vi.fn().mockReturnValue(
makeDeclarationTree('struct_item', 'declaration_list', node.content, [
{ text: 'name: String,', type: 'field_definition' },
{ text: 'email: String,', type: 'field_definition' },
{ text: 'age: u32,', type: 'field_definition' },
{ text: 'address: String,', type: 'field_definition' },
]),
),
});
const chunks = await chunkNode(node.label, node.content, node.filePath, 40, 45, 45, 0);
expect(chunks).toHaveLength(2);
const secondText = generateEmbeddingText(
node,
chunks[1].text,
{},
chunks[1].chunkIndex,
chunks[0].text.slice(-STRUCT_PREV_TAIL_SAMPLE),
);
expect(secondText).toContain('Struct: User');
expect(secondText).toContain('Container: struct User {');
expect(secondText).not.toContain('Properties: name, email, age, address');
expect(secondText).toContain('age: u32,');
});
it('metadata is present in every chunk', () => {
const longContent = 'x'.repeat(3000);
const node = makeNode({

View file

@ -1,11 +1,17 @@
import { describe, it, expect, vi, beforeEach } from 'vitest';
import { createHash } from 'crypto';
import { contentHashForNode } from '../../src/core/embeddings/embedding-pipeline.js';
import {
contentHashForNode,
EMBEDDING_TEXT_VERSION,
} from '../../src/core/embeddings/embedding-pipeline.js';
import { generateEmbeddingText } from '../../src/core/embeddings/text-generator.js';
import type { EmbeddableNode, EmbeddingProgress } from '../../src/core/embeddings/types.js';
import { DEFAULT_EMBEDDING_CONFIG, EMBEDDABLE_LABELS } from '../../src/core/embeddings/types.js';
import { STALE_HASH_SENTINEL } from '../../src/core/lbug/schema.js';
const CLASS_CHUNK_SIZE = 90;
const CLASS_OVERLAP = 10;
// ────────────────────────────────────────────────────────────────────────────
// contentHashForNode
// ────────────────────────────────────────────────────────────────────────────
@ -32,6 +38,8 @@ describe('contentHashForNode', () => {
it('matches sha1(generateEmbeddingText(node, node.content))', () => {
const node = makeNode();
const expected = createHash('sha1')
.update(EMBEDDING_TEXT_VERSION)
.update('\n')
.update(generateEmbeddingText(node, node.content))
.digest('hex');
expect(contentHashForNode(node)).toBe(expected);
@ -56,6 +64,10 @@ describe('contentHashForNode', () => {
const hashWithFullDefaults = contentHashForNode(node, DEFAULT_EMBEDDING_CONFIG);
expect(hashWithEmptyConfig).toBe(hashWithFullDefaults);
});
it('exports a text template version marker', () => {
expect(EMBEDDING_TEXT_VERSION).toBe('v2');
});
});
// ────────────────────────────────────────────────────────────────────────────
@ -439,6 +451,118 @@ describe('runEmbeddingPipeline incremental filter', () => {
expect(vectorIndexCalls.length).toBeGreaterThanOrEqual(1);
});
it('does not inject preceding context when overlap is disabled', async () => {
const embedBatchSpy = vi
.fn()
.mockImplementation((texts: string[]) =>
Promise.resolve(texts.map(() => new Float32Array(384))),
);
vi.doMock('../../src/core/embeddings/embedder.js', () => ({
initEmbedder: vi.fn().mockResolvedValue(undefined),
embedBatch: embedBatchSpy,
embedText: vi.fn().mockResolvedValue(new Float32Array(384)),
embeddingToArray: vi.fn().mockImplementation((emb: Float32Array) => Array.from(emb)),
isEmbedderReady: vi.fn().mockReturnValue(true),
}));
vi.doMock('../../src/core/lbug/lbug-adapter.js', () => ({
loadVectorExtension: vi.fn().mockResolvedValue(undefined),
}));
const node = makeNode({
label: 'Class',
name: 'Parser',
content: `class Parser {
options: ParserOptions;
cache: Map<string, any>;
parseJSON() { return JSON.parse("{}"); }
validate() { return true; }
}`,
startLine: 1,
endLine: 6,
});
const executeQuery = mockExecuteQuery([node]);
const executeWithReusedStatement = mockExecuteWithReusedStatement();
const { runEmbeddingPipeline } =
await import('../../src/core/embeddings/embedding-pipeline.js');
await runEmbeddingPipeline(
executeQuery,
executeWithReusedStatement,
onProgress,
{ chunkSize: 90, overlap: 0 },
undefined,
undefined,
new Map(),
);
const embeddedTexts = embedBatchSpy.mock.calls.flatMap((call) => call[0] as string[]);
const laterChunks = embeddedTexts.slice(1);
expect(laterChunks.length).toBeGreaterThan(0);
for (const text of laterChunks) {
expect(text).not.toContain('[preceding context]:');
}
});
it('truncates preceding context to the configured overlap size', async () => {
const embedBatchSpy = vi
.fn()
.mockImplementation((texts: string[]) =>
Promise.resolve(texts.map(() => new Float32Array(384))),
);
vi.doMock('../../src/core/embeddings/embedder.js', () => ({
initEmbedder: vi.fn().mockResolvedValue(undefined),
embedBatch: embedBatchSpy,
embedText: vi.fn().mockResolvedValue(new Float32Array(384)),
embeddingToArray: vi.fn().mockImplementation((emb: Float32Array) => Array.from(emb)),
isEmbedderReady: vi.fn().mockReturnValue(true),
}));
vi.doMock('../../src/core/lbug/lbug-adapter.js', () => ({
loadVectorExtension: vi.fn().mockResolvedValue(undefined),
}));
const node = makeNode({
label: 'Class',
name: 'Parser',
content: `class Parser {
options: ParserOptions;
cache: Map<string, any>;
parseJSON() { return JSON.parse("{}"); }
validate() { return true; }
}`,
startLine: 1,
endLine: 6,
});
const executeQuery = mockExecuteQuery([node]);
const executeWithReusedStatement = mockExecuteWithReusedStatement();
const { runEmbeddingPipeline } =
await import('../../src/core/embeddings/embedding-pipeline.js');
await runEmbeddingPipeline(
executeQuery,
executeWithReusedStatement,
onProgress,
{ chunkSize: CLASS_CHUNK_SIZE, overlap: CLASS_OVERLAP },
undefined,
undefined,
new Map(),
);
const embeddedTexts = embedBatchSpy.mock.calls.flatMap((call) => call[0] as string[]);
const laterChunk = embeddedTexts.find((text) => text.includes('[preceding context]:'));
expect(laterChunk).toBeDefined();
expect(laterChunk).toContain('[preceding context]: ...');
const precedingContextLine = laterChunk
?.split('\n')
.find((line) => line.startsWith('[preceding context]: ...'));
expect(precedingContextLine).toBeDefined();
expect(precedingContextLine).toContain('ring, any>');
expect(precedingContextLine).not.toContain('parseJSON() {');
});
it('throws when DELETE for stale nodes fails with non-trivial error', async () => {
mockEmbedderSetup();

View file

@ -0,0 +1,282 @@
import { describe, it, expect, vi, beforeEach } from 'vitest';
import * as fs from 'node:fs';
import * as path from 'node:path';
import * as os from 'node:os';
import {
validateGroupImpactParams,
runGroupImpact,
MAX_SUPPORTED_CROSS_DEPTH,
DEFAULT_LOCAL_IMPACT_TIMEOUT_MS,
collectImpactSymbolUids,
fileMatchesServicePrefix,
} from '../../../src/core/group/cross-impact.js';
import type { GroupToolPort } from '../../../src/core/group/service.js';
import { writeBridgeMeta } from '../../../src/core/group/bridge-db.js';
import { BRIDGE_SCHEMA_VERSION } from '../../../src/core/group/bridge-schema.js';
function tmpGroup(): { tmpDir: string; groupDir: string; cleanup: () => void } {
const tmpDir = path.join(os.tmpdir(), `gitnexus-ci-${Date.now()}-${Math.random()}`);
const groupDir = path.join(tmpDir, 'groups', 'g1');
fs.mkdirSync(groupDir, { recursive: true });
fs.writeFileSync(
path.join(groupDir, 'group.yaml'),
`version: 1
name: g1
description: ""
repos:
app/backend: reg-be
app/frontend: reg-fe
links: []
packages: {}
detect:
http: true
grpc: true
topics: true
shared_libs: true
embedding_fallback: true
matching:
bm25_threshold: 0.7
embedding_threshold: 0.65
max_candidates_per_step: 3
`,
);
return {
tmpDir,
groupDir,
cleanup: () => fs.rmSync(tmpDir, { recursive: true, force: true }),
};
}
describe('cross-impact', () => {
beforeEach(() => {
vi.restoreAllMocks();
});
it('test_validateGroupImpactParams_rejects_bad_direction', () => {
const r = validateGroupImpactParams({
name: 'g',
repo: 'a',
target: 't',
direction: 'sideways',
});
expect(r.ok).toBe(false);
if (!r.ok) expect(r.error).toContain('direction');
});
it('test_validateGroupImpactParams_clamps_crossDepth_and_warns', () => {
const r = validateGroupImpactParams({
name: 'g',
repo: 'a',
target: 't',
direction: 'upstream',
crossDepth: 99,
});
expect(r.ok).toBe(true);
if (r.ok) {
expect(r.crossDepth).toBe(MAX_SUPPORTED_CROSS_DEPTH);
expect(r.crossDepthWarning).toBeDefined();
}
});
it('test_validateGroupImpactParams_default_timeout', () => {
const r = validateGroupImpactParams({
name: 'g',
repo: 'a',
target: 't',
direction: 'downstream',
});
expect(r.ok).toBe(true);
if (r.ok) expect(r.timeoutMs).toBe(DEFAULT_LOCAL_IMPACT_TIMEOUT_MS);
});
it('test_collectImpactSymbolUids_respects_service_prefix', () => {
const local = {
target: { id: 'a', filePath: 'services/auth/x.ts' },
byDepth: {
1: [{ id: 'b', filePath: 'other/y.ts' }],
},
};
const uids = collectImpactSymbolUids(local, 'services/auth').uids;
expect(uids).toContain('a');
expect(uids).not.toContain('b');
});
it('test_fileMatchesServicePrefix', () => {
expect(fileMatchesServicePrefix('services/auth/a.ts', 'services/auth')).toBe(true);
expect(fileMatchesServicePrefix('services/aut', 'services/auth')).toBe(false);
});
it('test_runGroupImpact_local_timeout_returns_truncation', async () => {
const { tmpDir, cleanup } = tmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
try {
let impactCalls = 0;
const port: GroupToolPort = {
resolveRepo: vi.fn(async () => ({
id: 'be',
name: 'reg-be',
repoPath: '/r',
storagePath: '/r/.gitnexus',
})),
impact: vi.fn(async () => {
impactCalls++;
await new Promise((r) => setTimeout(r, 200));
return { summary: { direct: 1 }, byDepth: { 1: [{ id: 'x' }] } };
}),
query: vi.fn(),
impactByUid: vi.fn(),
context: vi.fn(),
};
const r = await runGroupImpact(
{ port, gitnexusDir: tmpDir },
{
name: 'g1',
repo: 'app/backend',
target: 'Sym',
direction: 'upstream',
timeoutMs: 15,
},
);
expect(impactCalls).toBe(1);
expect('error' in r).toBe(false);
if (!('error' in r)) {
expect(r.truncationReason).toBe('timeout');
expect(r.truncated).toBe(true);
}
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
it('test_runGroupImpact_local_phase_error_bubbles_as_top_level_error', async () => {
// Regression for #1004: when the local-impact phase returns a structured
// `{ error: ... }` payload, groupImpact MUST surface it as a top-level
// `{ error }` instead of a zero-hit GroupImpactResult. Otherwise callers
// that branch on top-level `error` silently treat a failed analysis as
// "no impact across the group" — a false negative on the failure path
// of a blast-radius tool.
const { tmpDir, cleanup } = tmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
try {
const port: GroupToolPort = {
resolveRepo: vi.fn(async () => ({
id: 'be',
name: 'reg-be',
repoPath: '/r',
storagePath: '/r/.gitnexus',
})),
impact: vi.fn(async () => ({ error: 'symbol not found: Sym' })),
query: vi.fn(),
impactByUid: vi.fn(),
context: vi.fn(),
};
const r = await runGroupImpact(
{ port, gitnexusDir: tmpDir },
{
name: 'g1',
repo: 'app/backend',
target: 'Sym',
direction: 'upstream',
},
);
expect('error' in r).toBe(true);
if ('error' in r) {
expect(r.error).toContain('symbol not found: Sym');
expect(r.error).toContain('app/backend');
}
// And ensure we didn't silently fall back to a zero-hit success payload.
expect((r as { summary?: unknown }).summary).toBeUndefined();
expect((r as { cross?: unknown }).cross).toBeUndefined();
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
it('test_runGroupImpact_local_phase_thrown_exception_bubbles_as_top_level_error', async () => {
// Companion to the #1004 regression: safeLocalImpact wraps thrown
// exceptions from port.impact() as `{ error }` payloads. Those must
// bubble to the caller as top-level errors too, not be swallowed into
// an empty success payload.
const { tmpDir, cleanup } = tmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
try {
const port: GroupToolPort = {
resolveRepo: vi.fn(async () => ({
id: 'be',
name: 'reg-be',
repoPath: '/r',
storagePath: '/r/.gitnexus',
})),
impact: vi.fn(async () => {
throw new Error('graph-load failure: .gitnexus missing');
}),
query: vi.fn(),
impactByUid: vi.fn(),
context: vi.fn(),
};
const r = await runGroupImpact(
{ port, gitnexusDir: tmpDir },
{
name: 'g1',
repo: 'app/backend',
target: 'Sym',
direction: 'upstream',
},
);
expect('error' in r).toBe(true);
if ('error' in r) {
expect(r.error).toContain('graph-load failure');
}
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
it('test_runGroupImpact_bridge_schema_mismatch_returns_error', async () => {
const { tmpDir, groupDir, cleanup } = tmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
await writeBridgeMeta(groupDir, {
version: BRIDGE_SCHEMA_VERSION + 9,
generatedAt: new Date().toISOString(),
missingRepos: [],
});
try {
const port: GroupToolPort = {
resolveRepo: vi.fn(async () => ({
id: 'be',
name: 'reg-be',
repoPath: '/r',
storagePath: '/r/.gitnexus',
})),
impact: vi.fn(async () => ({
target: { id: 'u1', filePath: 'src/a.ts' },
summary: { direct: 1, processes_affected: 0, modules_affected: 0 },
byDepth: { 1: [{ id: 'u1', filePath: 'src/a.ts' }] },
risk: 'LOW',
})),
query: vi.fn(),
impactByUid: vi.fn(),
context: vi.fn(),
};
const r = await runGroupImpact(
{ port, gitnexusDir: tmpDir },
{
name: 'g1',
repo: 'app/backend',
target: 'Sym',
direction: 'upstream',
},
);
expect('error' in r).toBe(true);
if ('error' in r) {
expect(r.error).toContain('schema');
}
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
});

View file

@ -0,0 +1,87 @@
import { describe, it, expect } from 'vitest';
import {
fileMatchesServicePrefix,
normalizeServicePrefix,
repoInSubgroup,
} from '../../../src/core/group/group-path-utils.js';
describe('group-path-utils', () => {
describe('normalizeServicePrefix', () => {
it('returns undefined for null/undefined/empty', () => {
expect(normalizeServicePrefix(undefined)).toBeUndefined();
expect(normalizeServicePrefix(null)).toBeUndefined();
expect(normalizeServicePrefix('')).toBeUndefined();
expect(normalizeServicePrefix(' ')).toBeUndefined();
});
it('strips trailing slashes', () => {
expect(normalizeServicePrefix('services/auth/')).toBe('services/auth');
expect(normalizeServicePrefix('services/auth///')).toBe('services/auth');
});
it('normalizes Windows-style backslashes to POSIX', () => {
expect(normalizeServicePrefix('services\\auth')).toBe('services/auth');
expect(normalizeServicePrefix('app\\backend\\')).toBe('app/backend');
});
});
describe('fileMatchesServicePrefix', () => {
it('returns true when prefix is empty/undefined', () => {
expect(fileMatchesServicePrefix('any/file.ts', undefined)).toBe(true);
expect(fileMatchesServicePrefix('any/file.ts', '')).toBe(true);
});
it('returns false when filePath is missing but prefix is set', () => {
expect(fileMatchesServicePrefix(undefined, 'services/auth')).toBe(false);
});
it('matches exact prefix and descendants', () => {
expect(fileMatchesServicePrefix('services/auth', 'services/auth')).toBe(true);
expect(fileMatchesServicePrefix('services/auth/a.ts', 'services/auth')).toBe(true);
});
it('rejects partial-segment matches', () => {
expect(fileMatchesServicePrefix('services/aut', 'services/auth')).toBe(false);
expect(fileMatchesServicePrefix('services/authz/a.ts', 'services/auth')).toBe(false);
});
it('matches Windows-style file paths against POSIX prefix', () => {
expect(fileMatchesServicePrefix('services\\auth\\a.ts', 'services/auth')).toBe(true);
expect(fileMatchesServicePrefix('services\\authz\\a.ts', 'services/auth')).toBe(false);
});
});
describe('repoInSubgroup', () => {
it('matches every repo when subgroup is empty/undefined', () => {
expect(repoInSubgroup('any/repo', undefined)).toBe(true);
expect(repoInSubgroup('any/repo', '')).toBe(true);
expect(repoInSubgroup('any/repo', ' ')).toBe(true);
});
it('matches exact path and descendants by default', () => {
expect(repoInSubgroup('app/backend', 'app/backend')).toBe(true);
expect(repoInSubgroup('app/backend/sub', 'app/backend')).toBe(true);
expect(repoInSubgroup('app/frontend', 'app/backend')).toBe(false);
});
it('strips trailing slashes from subgroup', () => {
expect(repoInSubgroup('app/backend', 'app/backend/')).toBe(true);
expect(repoInSubgroup('app/backend/x', 'app/backend///')).toBe(true);
});
it('with exact=true matches only the exact repo', () => {
expect(repoInSubgroup('app/backend', 'app/backend', true)).toBe(true);
expect(repoInSubgroup('app/backend/sub', 'app/backend', true)).toBe(false);
});
it('rejects partial-segment matches', () => {
expect(repoInSubgroup('app/backendz', 'app/backend')).toBe(false);
});
it('normalizes Windows-style paths on both sides', () => {
expect(repoInSubgroup('app\\backend', 'app/backend')).toBe(true);
expect(repoInSubgroup('app/backend/x', 'app\\backend')).toBe(true);
expect(repoInSubgroup('app\\backend\\sub', 'app\\backend', true)).toBe(false);
});
});
});

View file

@ -0,0 +1,129 @@
/**
* Documents MCP → GroupService mapping: callers use `name` + concrete params;
* the "@group" string is interpreted only in LocalBackend.callTool (Issue #794).
*/
import { describe, it, expect, vi } from 'vitest';
import * as fs from 'node:fs';
import * as path from 'node:path';
import * as os from 'node:os';
import {
GroupService,
type GroupToolPort,
type GroupRepoHandle,
} from '../../../src/core/group/service.js';
function makeTmpGroup(): { tmpDir: string; cleanup: () => void } {
const tmpDir = path.join(os.tmpdir(), `gitnexus-gmode-${Date.now()}`);
const groupDir = path.join(tmpDir, 'groups', 'test-group');
fs.mkdirSync(groupDir, { recursive: true });
fs.writeFileSync(
path.join(groupDir, 'group.yaml'),
`version: 1
name: test-group
repos:
app/backend: test-backend
app/frontend: test-frontend
`,
);
return { tmpDir, cleanup: () => fs.rmSync(tmpDir, { recursive: true, force: true }) };
}
function makePort(overrides: Partial<GroupToolPort> = {}): GroupToolPort {
return {
resolveRepo: vi.fn(
async (name?: string): Promise<GroupRepoHandle> => ({
id: name || 'test',
name: name || 'test',
repoPath: '/tmp/repo',
storagePath: '/tmp/repo/.gitnexus',
}),
),
impact: vi.fn(async () => ({ target: {}, byDepth: {} })),
query: vi.fn(async () => ({
processes: [{ id: 'p1', heuristicLabel: 'Proc' }],
process_symbols: [
{ id: 's1', process_id: 'p1', filePath: 'services/auth/a.ts' },
{ id: 's2', process_id: 'p1', filePath: 'other/b.ts' },
],
})),
impactByUid: vi.fn(async () => null),
context: vi.fn(async () => ({
status: 'found',
symbol: { filePath: 'services/auth/x.ts', uid: 'u1', name: 'X' },
})),
...overrides,
};
}
describe('GroupService group-mode API surface', () => {
it('groupQuery uses name (never @-repo) and optional service filters processes', async () => {
const { tmpDir, cleanup } = makeTmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
try {
const query = vi.fn(async () => ({
processes: [{ id: 'p1' }],
process_symbols: [
{ id: 's1', process_id: 'p1', filePath: 'services/auth/a.ts' },
{ id: 's2', process_id: 'p1', filePath: 'other/b.ts' },
],
}));
const svc = new GroupService(makePort({ query }));
const r = (await svc.groupQuery({
name: 'test-group',
query: 'oauth',
service: 'services/auth',
})) as { results: Array<{ id?: string }> };
expect(query).toHaveBeenCalled();
expect(r.results.every((row) => row.id === 'p1')).toBe(true);
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
it('groupQuery rejects empty service string', async () => {
const { tmpDir, cleanup } = makeTmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
try {
const svc = new GroupService(makePort());
const r = await svc.groupQuery({ name: 'test-group', query: 'x', service: ' ' });
expect(r).toEqual({ error: 'service must not be an empty string' });
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
it('groupContext uses name + target (MCP maps @group to name)', async () => {
const { tmpDir, cleanup } = makeTmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
try {
const svc = new GroupService(makePort());
const r = await svc.groupContext({ name: 'test-group', target: 'MySym' });
expect(r.group).toBe('test-group');
expect(r.results).toHaveLength(2);
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
it('groupImpact with mock port returns structured result without @ in params', async () => {
const { tmpDir, cleanup } = makeTmpGroup();
vi.stubEnv('GITNEXUS_HOME', tmpDir);
try {
const svc = new GroupService(makePort());
const r = (await svc.groupImpact({
name: 'test-group',
repo: 'app/backend',
target: 't',
direction: 'upstream',
})) as { group?: string; error?: string };
expect(r.error).toBeUndefined();
expect(r.group).toBe('test-group');
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
});

View file

@ -2,16 +2,10 @@
import { describe, it, expect } from 'vitest';
import { GITNEXUS_TOOLS } from '../../../src/mcp/tools.js';
const GROUP_TOOL_NAMES = [
'group_list',
'group_sync',
'group_contracts',
'group_query',
'group_status',
];
const GROUP_TOOL_NAMES = ['group_list', 'group_sync'];
describe('Group MCP tools', () => {
it('all 5 group tools are registered', () => {
it('group_list and group_sync are registered', () => {
for (const name of GROUP_TOOL_NAMES) {
const tool = GITNEXUS_TOOLS.find((t) => t.name === name);
expect(tool, `tool ${name} should be registered`).toBeDefined();

View file

@ -517,6 +517,86 @@ Route::delete('/users/{id}', [UserController::class, 'destroy']);
});
});
describe('consumer extraction — PHP', () => {
it('extracts Laravel Http facade calls', async () => {
const dir = path.join(tmpDir, 'php-http-facade');
fs.mkdirSync(path.join(dir, 'app'), { recursive: true });
fs.writeFileSync(
path.join(dir, 'app/Client.php'),
`<?php
use Illuminate\\Support\\Facades\\Http;
class Client {
public function run() {
Http::get('/api/users');
Http::post('/api/orders/42');
Http::delete('/api/users/7');
}
}
`,
);
const contracts = await extractor.extract(null, dir, makeRepo(dir));
const consumers = contracts.filter((c) => c.role === 'consumer');
expect(consumers.find((c) => c.contractId === 'http::GET::/api/users')).toBeDefined();
expect(
consumers.find((c) => c.contractId === 'http::POST::/api/orders/{param}'),
).toBeDefined();
expect(
consumers.find((c) => c.contractId === 'http::DELETE::/api/users/{param}'),
).toBeDefined();
});
it('extracts Guzzle $client->method() calls', async () => {
const dir = path.join(tmpDir, 'php-guzzle');
fs.mkdirSync(path.join(dir, 'src'), { recursive: true });
fs.writeFileSync(
path.join(dir, 'src/ApiClient.php'),
`<?php
use GuzzleHttp\\Client;
class ApiClient {
public function run(Client $client) {
$client->get('/api/health');
$client->post('/api/orders/42');
}
}
`,
);
const contracts = await extractor.extract(null, dir, makeRepo(dir));
const consumers = contracts.filter((c) => c.role === 'consumer');
expect(consumers.find((c) => c.contractId === 'http::GET::/api/health')).toBeDefined();
expect(
consumers.find((c) => c.contractId === 'http::POST::/api/orders/{param}'),
).toBeDefined();
});
it('extracts file_get_contents HTTP calls', async () => {
const dir = path.join(tmpDir, 'php-fgc');
fs.mkdirSync(path.join(dir, 'src'), { recursive: true });
fs.writeFileSync(
path.join(dir, 'src/fetch.php'),
`<?php
function fetchRemote() {
$data = file_get_contents('https://example.test/api/items/1');
$local = file_get_contents('/tmp/local-file.txt');
return $data;
}
`,
);
const contracts = await extractor.extract(null, dir, makeRepo(dir));
const consumers = contracts.filter((c) => c.role === 'consumer');
expect(consumers.find((c) => c.contractId === 'http::GET::/api/items/{param}')).toBeDefined();
// file paths and stream wrappers must not emit consumer contracts
expect(consumers.find((c) => c.meta.path === '/tmp/local-file.txt')).toBeUndefined();
});
});
describe('provider extraction — FastAPI', () => {
it('extracts FastAPI @app.get decorator patterns', async () => {
const dir = path.join(tmpDir, 'fastapi');

View file

@ -44,6 +44,10 @@ function makePort(overrides: Partial<GroupToolPort> = {}): GroupToolPort {
impact: vi.fn(async () => ({ symbols: [] })),
query: vi.fn(async () => ({ processes: [] })),
impactByUid: vi.fn(async () => null),
context: vi.fn(async () => ({
status: 'found',
symbol: { filePath: 'services/auth/x.ts', uid: 'u1', name: 'X' },
})),
...overrides,
};
}
@ -233,6 +237,46 @@ describe('GroupService', () => {
cleanup();
}
});
it('test_groupContracts_skips_corrupt_contract_rows', async () => {
const { groupDir, cleanup, tmpDir } = makeTmpGroup();
try {
vi.stubEnv('GITNEXUS_HOME', tmpDir);
const badJson = `{
"version": 1,
"generatedAt": "2026-01-01T00:00:00.000Z",
"repoSnapshots": {},
"missingRepos": [],
"contracts": [
{ "not": "a-contract" },
{
"contractId": "http::GET::/ok",
"type": "http",
"repo": "app/backend",
"role": "provider",
"symbolUid": "u",
"symbolRef": { "filePath": "a.ts", "name": "f" },
"symbolName": "f",
"confidence": 1,
"meta": {}
}
],
"crossLinks": []
}`;
fs.writeFileSync(path.join(groupDir, 'contracts.json'), badJson, 'utf-8');
const svc = new GroupService(makePort());
const result = (await svc.groupContracts({ name: 'test-group' })) as {
contracts: unknown[];
skippedCorrupt?: number;
};
expect(result.contracts).toHaveLength(1);
expect(result.skippedCorrupt).toBe(1);
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
});
describe('groupSync', () => {
@ -328,6 +372,135 @@ describe('GroupService', () => {
cleanup();
}
});
it('test_groupQuery_subgroupExact_skips_descendant_member_paths', async () => {
const tmpDir = path.join(os.tmpdir(), `gitnexus-svc-nest-${Date.now()}`);
const groupDir = path.join(tmpDir, 'groups', 'nest-group');
fs.mkdirSync(groupDir, { recursive: true });
fs.writeFileSync(
path.join(groupDir, 'group.yaml'),
`version: 1
name: nest-group
repos:
app/frontend: fe-root
app/frontend/mobile: fe-nested
app/backend: be1
`,
);
try {
vi.stubEnv('GITNEXUS_HOME', tmpDir);
const query = vi.fn(async () => ({ processes: [{ name: 'p1' }] }));
const port = makePort({ query });
const svc = new GroupService(port);
const prefixOnly = (await svc.groupQuery({
name: 'nest-group',
query: 'x',
subgroup: 'app/frontend',
})) as { per_repo: Array<{ repo: string }> };
expect(prefixOnly.per_repo.map((r) => r.repo).sort()).toEqual([
'app/frontend',
'app/frontend/mobile',
]);
const exact = (await svc.groupQuery({
name: 'nest-group',
query: 'x',
subgroup: 'app/frontend',
subgroupExact: true,
})) as { per_repo: Array<{ repo: string }> };
expect(exact.per_repo).toHaveLength(1);
expect(exact.per_repo[0].repo).toBe('app/frontend');
} finally {
vi.unstubAllEnvs();
fs.rmSync(tmpDir, { recursive: true, force: true });
}
});
});
describe('groupImpact', () => {
it('test_groupImpact_returns_validation_error', async () => {
const svc = new GroupService(makePort());
const r = (await svc.groupImpact({})) as { error: string };
expect(r.error).toContain('name');
});
});
describe('groupContext', () => {
it('test_groupContext_requires_target_or_uid', async () => {
const svc = new GroupService(makePort());
const r = await svc.groupContext({ name: 'test-group' });
expect(r.error).toContain('target');
});
it('test_groupContext_iterates_repos', async () => {
const { cleanup, tmpDir } = makeTmpGroup();
try {
vi.stubEnv('GITNEXUS_HOME', tmpDir);
const port = makePort();
const svc = new GroupService(port);
const r = await svc.groupContext({ name: 'test-group', target: 'MySym' });
expect(r.group).toBe('test-group');
expect(r.results).toHaveLength(2);
expect(port.context).toHaveBeenCalledTimes(2);
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
it('test_groupContext_subgroupExact_skips_descendant_member_paths', async () => {
const tmpDir = path.join(os.tmpdir(), `gitnexus-ctx-nest-${Date.now()}`);
const groupDir = path.join(tmpDir, 'groups', 'nest-group');
fs.mkdirSync(groupDir, { recursive: true });
fs.writeFileSync(
path.join(groupDir, 'group.yaml'),
`version: 1
name: nest-group
repos:
app/frontend: fe-root
app/frontend/mobile: fe-nested
`,
);
try {
vi.stubEnv('GITNEXUS_HOME', tmpDir);
const port = makePort();
const svc = new GroupService(port);
await svc.groupContext({
name: 'nest-group',
target: 'X',
subgroup: 'app/frontend',
subgroupExact: true,
});
expect(port.context).toHaveBeenCalledTimes(1);
} finally {
vi.unstubAllEnvs();
fs.rmSync(tmpDir, { recursive: true, force: true });
}
});
it('test_groupContext_service_prefix_filters_payload', async () => {
const { cleanup, tmpDir } = makeTmpGroup();
try {
vi.stubEnv('GITNEXUS_HOME', tmpDir);
const port = makePort({
context: vi.fn(async () => ({
status: 'found',
symbol: { filePath: 'other/path/x.ts', uid: 'u1', name: 'X' },
})),
});
const svc = new GroupService(port);
const r = await svc.groupContext({
name: 'test-group',
target: 'MySym',
service: 'services/auth',
});
expect(r.results.every((x) => Object.keys(x.payload as object).length === 0)).toBe(true);
} finally {
vi.unstubAllEnvs();
cleanup();
}
});
});
describe('groupStatus', () => {

View file

@ -0,0 +1,230 @@
/**
* LocalBackend.callTool routes impact/query/context to GroupService when repo starts with "@".
*/
import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
import * as fs from 'node:fs';
import * as path from 'node:path';
import * as os from 'node:os';
const { lbugMocks } = vi.hoisted(() => ({
lbugMocks: {
initLbug: vi.fn().mockResolvedValue(undefined),
executeQuery: vi.fn().mockResolvedValue([]),
executeParameterized: vi.fn().mockResolvedValue([]),
closeLbug: vi.fn().mockResolvedValue(undefined),
isLbugReady: vi.fn().mockReturnValue(true),
},
}));
vi.mock('../../../src/core/lbug/pool-adapter.js', async (importOriginal) => {
const actual = await importOriginal<typeof import('../../../src/core/lbug/pool-adapter.js')>();
return { ...actual, ...lbugMocks };
});
vi.mock('../../../src/mcp/core/lbug-adapter.js', async (importOriginal) => {
const actual = await importOriginal<typeof import('../../../src/mcp/core/lbug-adapter.js')>();
return { ...actual, ...lbugMocks };
});
vi.mock('../../../src/storage/repo-manager.js', () => ({
listRegisteredRepos: vi.fn().mockResolvedValue([]),
cleanupOldKuzuFiles: vi.fn().mockResolvedValue({ found: false, needsReindex: false }),
}));
vi.mock('../../../src/core/search/bm25-index.js', () => ({
searchFTSFromLbug: vi.fn().mockResolvedValue([]),
}));
vi.mock('../../../src/mcp/core/embedder.js', () => ({
embedQuery: vi.fn().mockResolvedValue([]),
getEmbeddingDims: vi.fn().mockReturnValue(384),
}));
import { LocalBackend } from '../../../src/mcp/local/local-backend.js';
import { GroupService } from '../../../src/core/group/service.js';
describe('LocalBackend @group repo routing', () => {
let tmpDir: string;
let groupSpyQuery: ReturnType<typeof vi.spyOn>;
let groupSpyImpact: ReturnType<typeof vi.spyOn>;
let groupSpyContext: ReturnType<typeof vi.spyOn>;
beforeEach(() => {
tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), 'gitnexus-atgrp-'));
const groupDir = path.join(tmpDir, 'groups', 'g1');
fs.mkdirSync(groupDir, { recursive: true });
fs.writeFileSync(
path.join(groupDir, 'group.yaml'),
`version: 1
name: g1
repos:
app/backend: test-backend
app/frontend: test-frontend
`,
);
vi.stubEnv('GITNEXUS_HOME', tmpDir);
groupSpyQuery = vi
.spyOn(GroupService.prototype, 'groupQuery')
.mockResolvedValue({ via: 'query' });
groupSpyImpact = vi
.spyOn(GroupService.prototype, 'groupImpact')
.mockResolvedValue({ via: 'impact' });
groupSpyContext = vi.spyOn(GroupService.prototype, 'groupContext').mockResolvedValue({
group: 'g1',
results: [],
});
});
afterEach(() => {
vi.unstubAllEnvs();
vi.restoreAllMocks();
fs.rmSync(tmpDir, { recursive: true, force: true });
});
it('routes query to groupQuery with default member path (first sorted repos key)', async () => {
const backend = new LocalBackend();
const out = await backend.callTool('query', { repo: '@g1', query: 'login' });
expect(out).toEqual({ via: 'query' });
expect(groupSpyQuery).toHaveBeenCalledWith(
expect.objectContaining({ name: 'g1', query: 'login' }),
);
const arg = groupSpyQuery.mock.calls[0][0] as Record<string, unknown>;
expect(arg).not.toHaveProperty('repo');
});
it('routes query with explicit member path as exact subgroup (no descendant repo bleed)', async () => {
const backend = new LocalBackend();
await backend.callTool('query', { repo: '@g1/app/frontend', query: 'x' });
expect(groupSpyQuery).toHaveBeenCalledWith(
expect.objectContaining({
name: 'g1',
query: 'x',
subgroup: 'app/frontend',
subgroupExact: true,
}),
);
});
it('routes impact to groupImpact with resolved repo member path', async () => {
const backend = new LocalBackend();
const out = await backend.callTool('impact', {
repo: '@g1',
target: 'Sym',
direction: 'upstream',
});
expect(out).toEqual({ via: 'impact' });
expect(groupSpyImpact).toHaveBeenCalledWith(
expect.objectContaining({
name: 'g1',
repo: 'app/backend',
target: 'Sym',
direction: 'upstream',
}),
);
});
it('routes context to groupContext', async () => {
const backend = new LocalBackend();
await backend.callTool('context', { repo: '@g1', target: 'Sym' });
expect(groupSpyContext).toHaveBeenCalledWith(
expect.objectContaining({ name: 'g1', target: 'Sym' }),
);
});
it('maps MCP symbol name to groupContext target (does not overwrite group name)', async () => {
const backend = new LocalBackend();
await backend.callTool('context', { repo: '@g1', name: 'MyClass' });
expect(groupSpyContext).toHaveBeenCalledWith(
expect.objectContaining({ name: 'g1', target: 'MyClass' }),
);
});
it('returns error for unknown group name', async () => {
const backend = new LocalBackend();
const out = await backend.callTool('query', { repo: '@no-such-group', query: 'x' });
expect(out).toHaveProperty('error');
expect(String((out as { error: string }).error)).toMatch(
/not found|no such|unknown|exist|ENOENT/i,
);
});
it('returns error for unknown member path', async () => {
const backend = new LocalBackend();
const out = await backend.callTool('query', { repo: '@g1/not-a-member', query: 'x' });
expect(out).toHaveProperty('error');
expect(String((out as { error: string }).error)).toMatch(/Unknown member path/i);
});
it('rejects empty service without calling group tools', async () => {
const backend = new LocalBackend();
const out = await backend.callTool('query', { repo: '@g1', query: 'x', service: '' });
expect(out).toEqual({ error: 'service must not be an empty string' });
expect(groupSpyQuery).not.toHaveBeenCalled();
});
it('unknown group_* tools mention removal', async () => {
const backend = new LocalBackend();
await expect(backend.callTool('group_query', { name: 'g1', query: 'x' })).rejects.toThrow(
/Removed tools/,
);
});
it('removed group_contracts mentions migration', async () => {
const backend = new LocalBackend();
await expect(backend.callTool('group_contracts', { name: 'g1' })).rejects.toThrow(
/Removed tools/,
);
});
it('removed group_status mentions migration', async () => {
const backend = new LocalBackend();
await expect(backend.callTool('group_status', { name: 'g1' })).rejects.toThrow(/Removed tools/);
});
describe('Issue #794 manual smoke checklist (automated)', () => {
beforeEach(() => {
const groupDir = path.join(tmpDir, 'groups', 'myproduct');
fs.mkdirSync(groupDir, { recursive: true });
fs.writeFileSync(
path.join(groupDir, 'group.yaml'),
`version: 1
name: myproduct
repos:
app/backend: test-backend
app/frontend: test-frontend
`,
);
});
it.each([
{
method: 'impact',
params: { repo: '@myproduct', target: 'UserService.login', service: 'app/backend' },
spy: () => groupSpyImpact,
},
{
method: 'query',
params: { repo: '@myproduct', query: 'login', service: 'app/backend' },
spy: () => groupSpyQuery,
},
{
method: 'context',
params: { repo: '@myproduct', target: 'UserService.login', service: 'app/backend' },
spy: () => groupSpyContext,
},
])(
'$method with repo "@myproduct" routes to GroupService and forwards service',
async ({ method, params, spy }) => {
const backend = new LocalBackend();
await backend.callTool(method, params);
expect(spy()).toHaveBeenCalledWith(
expect.objectContaining({ name: 'myproduct', service: 'app/backend' }),
);
const callArg = spy().mock.calls[0][0] as Record<string, unknown>;
expect(
typeof callArg.repo === 'string' ? (callArg.repo as string).startsWith('@') : false,
).toBe(false);
},
);
});
});

View file

@ -15,7 +15,14 @@ import {
loadCLIConfig,
registerRepo,
listRegisteredRepos,
resolveRegistryEntry,
canonicalizePath,
assertSafeStoragePath,
RegistryNameCollisionError,
RegistryNotFoundError,
RegistryAmbiguousTargetError,
UnsafeStoragePathError,
type RegistryEntry,
type RepoMeta,
} from '../../src/storage/repo-manager.js';
import { parseRepoNameFromUrl, getInferredRepoName } from '../../src/storage/git.js';
@ -453,3 +460,334 @@ describe('getInferredRepoName + registerRepo (#979 — git remote inference)', (
}
});
});
// ─── resolveRegistryEntry (#664 — gitnexus remove <target>) ──────────
//
// The resolver is a pure function over a `RegistryEntry[]` snapshot, so
// these tests build synthetic entries inline and do NOT touch
// ~/.gitnexus. No GITNEXUS_HOME sandboxing needed. This also means the
// tests are platform-portable on Windows where realpath semantics on
// tmpdirs can diverge between runs (see the #955 CI pivot).
describe('resolveRegistryEntry (#664)', () => {
// A well-known synthetic registry with two same-name entries (which
// can only exist in reality after `--allow-duplicate-name` — #829) and
// one unique-name entry. Path prefixes differ across platforms so the
// tests stay meaningful regardless of `process.platform`.
const prefix = process.platform === 'win32' ? 'D:\\' : '/tmp/';
const pathA = `${prefix}projects${path.sep}gnx-a${path.sep}app`;
const pathB = `${prefix}projects${path.sep}gnx-b${path.sep}app`;
const pathW = `${prefix}work${path.sep}website`;
const entries: RegistryEntry[] = [
{
name: 'app',
path: pathA,
storagePath: `${pathA}${path.sep}.gitnexus`,
indexedAt: '2026-04-18T00:00:00.000Z',
lastCommit: 'aaaaaaa',
},
{
name: 'app',
path: pathB,
storagePath: `${pathB}${path.sep}.gitnexus`,
indexedAt: '2026-04-18T00:00:00.000Z',
lastCommit: 'bbbbbbb',
},
{
name: 'website',
path: pathW,
storagePath: `${pathW}${path.sep}.gitnexus`,
indexedAt: '2026-04-18T00:00:00.000Z',
lastCommit: 'ccccccc',
},
];
it('resolves by absolute path to the exact entry (path tier beats name tier)', () => {
const hit = resolveRegistryEntry(entries, pathA);
expect(hit).toBe(entries[0]);
expect(hit.path).toBe(pathA);
const hit2 = resolveRegistryEntry(entries, pathB);
expect(hit2).toBe(entries[1]);
expect(hit2.path).toBe(pathB);
});
it('resolves by unique name to the only matching entry', () => {
const hit = resolveRegistryEntry(entries, 'website');
expect(hit).toBe(entries[2]);
expect(hit.name).toBe('website');
});
it('name match is case-insensitive', () => {
expect(resolveRegistryEntry(entries, 'WEBSITE')).toBe(entries[2]);
expect(resolveRegistryEntry(entries, 'Website')).toBe(entries[2]);
});
it('path match is case-insensitive on Windows only', () => {
if (process.platform !== 'win32') {
// On POSIX, a differently-cased path must NOT match. Verify by
// lower-casing a mixed-case copy of pathW and expecting a miss.
const upper = pathW.toUpperCase();
expect(() => resolveRegistryEntry(entries, upper)).toThrow(RegistryNotFoundError);
return;
}
const upper = pathA.toUpperCase();
const hit = resolveRegistryEntry(entries, upper);
expect(hit).toBe(entries[0]);
});
it('throws RegistryAmbiguousTargetError when name matches multiple entries', () => {
// Two 'app' entries exist only because of --allow-duplicate-name
// (#829). The resolver MUST refuse to guess.
expect(() => resolveRegistryEntry(entries, 'app')).toThrow(RegistryAmbiguousTargetError);
try {
resolveRegistryEntry(entries, 'app');
} catch (e) {
expect(e).toBeInstanceOf(RegistryAmbiguousTargetError);
const err = e as RegistryAmbiguousTargetError;
expect(err.kind).toBe('RegistryAmbiguousTargetError');
expect(err.target).toBe('app');
expect(err.matches).toHaveLength(2);
// Error message must include both paths so the CLI can surface
// them without string-matching on `.message`.
expect(err.message).toContain(pathA);
expect(err.message).toContain(pathB);
}
});
it('throws RegistryNotFoundError when no entry matches', () => {
expect(() => resolveRegistryEntry(entries, 'nonexistent')).toThrow(RegistryNotFoundError);
try {
resolveRegistryEntry(entries, 'nonexistent');
} catch (e) {
expect(e).toBeInstanceOf(RegistryNotFoundError);
const err = e as RegistryNotFoundError;
expect(err.kind).toBe('RegistryNotFoundError');
expect(err.target).toBe('nonexistent');
// availableNames is disambiguated: 'app' appears twice, so both
// `app (path)` variants are included; 'website' is unique so it
// stays plain — matches the resolveRepo disambiguation shape.
expect(err.availableNames).toContain('website');
expect(err.availableNames.some((n) => n.startsWith('app ('))).toBe(true);
// Error message surfaces the hint.
expect(err.message).toContain('website');
}
});
it('throws RegistryNotFoundError with "no repositories registered" hint when registry is empty', () => {
try {
resolveRegistryEntry([], 'anything');
} catch (e) {
expect(e).toBeInstanceOf(RegistryNotFoundError);
const err = e as RegistryNotFoundError;
expect(err.availableNames).toEqual([]);
expect(err.message).toContain('No repositories are currently registered');
}
});
it('path match wins over name match (never ambiguous)', () => {
// Construct a pathological fixture where a registry entry's NAME
// happens to equal another entry's PATH. The path tier must win
// without triggering ambiguity.
const weird: RegistryEntry[] = [
{ ...entries[2] }, // 'website' at pathW
{
name: pathW, // degenerate: name equals another entry's path
path: `${prefix}elsewhere${path.sep}odd`,
storagePath: `${prefix}elsewhere${path.sep}odd${path.sep}.gitnexus`,
indexedAt: '2026-04-18T00:00:00.000Z',
lastCommit: 'ddddddd',
},
];
const hit = resolveRegistryEntry(weird, pathW);
// Must match the entry whose PATH is pathW, not the one whose NAME
// is pathW — because Tier 1 runs before Tier 2 and finds the path
// match first.
expect(hit.path).toBe(pathW);
expect(hit.name).toBe('website');
});
});
// ─── canonicalizePath (#1003 review — @evander-wang / @magyargergo) ──
//
// Shields `registerRepo`, `unregisterRepo`, and `resolveRegistryEntry`
// against cross-platform path-form divergence: macOS symlink expansion
// (/var → /private/var) and Windows 8.3 short-name expansion
// (RUNNERA~1 → runneradmin). The helper also underpins backwards
// compatibility with registries written by versions that only ran
// `path.resolve` — by canonicalising the stored entry at compare time,
// both pre- and post-fix entries converge to the same key.
//
// These tests avoid snapshotting a specific realpath value (that would
// be platform-fragile); instead they assert:
// - canonicalizePath is idempotent (f(f(x)) == f(x))
// - canonicalizePath falls back cleanly when the path doesn't exist
// - resolveRegistryEntry matches a stored entry even when the target
// and the stored value disagree on one-step normalisation (simulated
// via a fixture that stores the de-canonicalised form of a real
// existing path).
describe('canonicalizePath (#1003)', () => {
it('is idempotent — canonicalizePath(canonicalizePath(x)) === canonicalizePath(x)', async () => {
// Use the vitest project-root as a known-existing path. `os.tmpdir()`
// would work too but process.cwd() is guaranteed to exist for the
// test runner.
const p = process.cwd();
const once = canonicalizePath(p);
const twice = canonicalizePath(once);
expect(twice).toBe(once);
});
it('falls back to path.resolve when the target does not exist', () => {
// Construct a definitely-nonexistent path under tmpdir. Using
// random-ish segments so we don't collide with anything real.
const ghost = path.join(os.tmpdir(), 'gnx-never-exists-____', 'still-not-there');
const got = canonicalizePath(ghost);
// Must not throw, must not resolve to something weird — should be
// identical to `path.resolve(ghost)` since realpathSync.native will
// have thrown and we swallowed it.
expect(got).toBe(path.resolve(ghost));
});
it('returns an absolute path for relative input even when the path is missing', () => {
// Relative path that does not exist. Must still be absolute
// (fallback path: path.resolve normalises even non-existent inputs).
const rel = './does-not-exist-zzz-' + Date.now();
const got = canonicalizePath(rel);
expect(path.isAbsolute(got)).toBe(true);
});
});
describe('resolveRegistryEntry backward-compat with non-canonical stored paths (#1003)', () => {
it('matches a stored entry even when the target was passed in canonical form', async () => {
// Simulate the bug-producing scenario without depending on a real
// symlink/8.3 discrepancy (those are platform-specific and flaky to
// set up in CI). We take a REAL path that exists
// (canonicalizePath-stable), store a known-non-canonical copy of it
// in a fake RegistryEntry, then resolve with the canonical form and
// assert the match.
//
// Construct a non-canonical string that resolves to the same real
// path. `path.join` auto-normalises `.` and trailing separators, so
// we build the string by raw concat to keep it string-unequal to
// `realDir` until `canonicalizePath` runs.
const realDir = process.cwd();
const nonCanonical = realDir + path.sep + '.'; // e.g. /work/gitnexus/.
// Sanity: these are string-unequal before canonicalisation.
expect(nonCanonical).not.toBe(realDir);
const entries: RegistryEntry[] = [
{
name: 'stored-under-noncanonical-form',
path: nonCanonical,
storagePath: path.join(nonCanonical, '.gitnexus'),
indexedAt: '2026-04-20T00:00:00.000Z',
lastCommit: 'deadbee',
},
];
// Pass the canonical form as the target — resolver must still match.
const hit = resolveRegistryEntry(entries, realDir);
expect(hit).toBe(entries[0]);
});
});
// ─── assertSafeStoragePath (#1003 review — @magyargergo) ─────────────
//
// Guard rail against destroying more than the `.gitnexus/` subfolder.
// `~/.gitnexus/registry.json` is user-writable plain text, so a
// corrupted or hand-edited entry could put storagePath anywhere.
// These tests use synthetic `RegistryEntry` fixtures (no disk I/O)
// because the guard is a pure string check — it must not depend on
// the paths existing.
describe('assertSafeStoragePath (#1003)', () => {
const prefix = process.platform === 'win32' ? 'D:\\' : '/tmp/';
const repoPath = `${prefix}projects${path.sep}my-repo`;
const base: Omit<RegistryEntry, 'storagePath'> = {
name: 'my-repo',
path: repoPath,
indexedAt: '2026-04-21T00:00:00.000Z',
lastCommit: 'deadbee',
};
it('accepts the canonical <repo>/.gitnexus storage path', () => {
const entry: RegistryEntry = {
...base,
storagePath: path.join(repoPath, '.gitnexus'),
};
expect(() => assertSafeStoragePath(entry)).not.toThrow();
});
it('rejects when storagePath equals the repo path itself (would delete the code)', () => {
const entry: RegistryEntry = {
...base,
storagePath: repoPath, // catastrophic: rm the working tree
};
expect(() => assertSafeStoragePath(entry)).toThrow(UnsafeStoragePathError);
});
it('rejects when storagePath is a parent of the repo path', () => {
const entry: RegistryEntry = {
...base,
storagePath: path.dirname(repoPath), // also catastrophic
};
expect(() => assertSafeStoragePath(entry)).toThrow(UnsafeStoragePathError);
});
it('rejects when storagePath is empty (path.resolve falls back to cwd)', () => {
const entry: RegistryEntry = {
...base,
storagePath: '', // path.resolve('') === process.cwd() — would rm cwd
};
expect(() => assertSafeStoragePath(entry)).toThrow(UnsafeStoragePathError);
});
it('rejects when storagePath points somewhere totally unrelated', () => {
const entry: RegistryEntry = {
...base,
storagePath: `${prefix}some${path.sep}other${path.sep}place`,
};
expect(() => assertSafeStoragePath(entry)).toThrow(UnsafeStoragePathError);
});
it('rejects when storagePath is a sibling .gitnexus (right basename, wrong parent)', () => {
const entry: RegistryEntry = {
...base,
storagePath: path.join(`${prefix}different${path.sep}repo`, '.gitnexus'),
};
expect(() => assertSafeStoragePath(entry)).toThrow(UnsafeStoragePathError);
});
it('UnsafeStoragePathError carries the original entry + expected + actual paths', () => {
const entry: RegistryEntry = {
...base,
storagePath: `${prefix}evil${path.sep}path`,
};
try {
assertSafeStoragePath(entry);
} catch (e) {
expect(e).toBeInstanceOf(UnsafeStoragePathError);
const err = e as UnsafeStoragePathError;
expect(err.kind).toBe('UnsafeStoragePathError');
expect(err.entry).toBe(entry);
// Expected path is the canonical `<repo>/.gitnexus`.
expect(err.expectedStoragePath).toBe(path.join(path.resolve(repoPath), '.gitnexus'));
// Actual path is the corrupted value (resolved).
expect(err.actualStoragePath).toBe(path.resolve(entry.storagePath));
// Message must suggest the recovery action.
expect(err.message).toContain('registry.json');
}
});
it('Windows: storagePath match is case-insensitive to match register/unregister semantics', () => {
if (process.platform !== 'win32') return;
const entry: RegistryEntry = {
...base,
storagePath: path.join(repoPath.toUpperCase(), '.GITNEXUS'),
};
// Should accept because Windows paths are case-insensitive.
expect(() => assertSafeStoragePath(entry)).not.toThrow();
});
});

View file

@ -12,6 +12,7 @@ import { describe, it, expect, vi } from 'vitest';
import {
getResourceDefinitions,
getResourceTemplates,
parseResourceUri,
readResource,
} from '../../src/mcp/resources.js';
@ -36,6 +37,12 @@ function createMockBackend(overrides: Partial<Record<string, any>> = {}): any {
queryProcessDetail: vi
.fn()
.mockResolvedValue(overrides.processDetail ?? { error: 'Not found' }),
readGroupContractsResource: vi
.fn()
.mockResolvedValue(overrides.groupContractsBody ?? 'contracts: []\n'),
readGroupStatusResource: vi
.fn()
.mockResolvedValue(overrides.groupStatusBody ?? 'group: mock\n'),
...overrides,
};
}
@ -73,12 +80,12 @@ describe('getResourceDefinitions', () => {
});
describe('getResourceTemplates', () => {
it('returns 6 dynamic templates', () => {
it('returns 8 dynamic templates', () => {
const templates = getResourceTemplates();
expect(templates).toHaveLength(6);
expect(templates).toHaveLength(8);
});
it('includes context, clusters, processes, schema, cluster detail, process detail', () => {
it('includes context, clusters, processes, schema, cluster detail, process detail, group contracts/status', () => {
const templates = getResourceTemplates();
const uris = templates.map((t) => t.uriTemplate);
expect(uris).toContain('gitnexus://repo/{name}/context');
@ -87,6 +94,8 @@ describe('getResourceTemplates', () => {
expect(uris).toContain('gitnexus://repo/{name}/schema');
expect(uris).toContain('gitnexus://repo/{name}/cluster/{clusterName}');
expect(uris).toContain('gitnexus://repo/{name}/process/{processName}');
expect(uris).toContain('gitnexus://group/{name}/contracts');
expect(uris).toContain('gitnexus://group/{name}/status');
});
it('each template has uriTemplate, name, description, mimeType', () => {
@ -99,6 +108,61 @@ describe('getResourceTemplates', () => {
});
});
describe('parseResourceUri', () => {
it('parses group contracts without query', () => {
const p = parseResourceUri('gitnexus://group/acme/contracts');
expect(p).toEqual({
kind: 'group',
groupName: 'acme',
resourceType: 'contracts',
contractsFilter: {},
});
});
it('parses nested group name and contracts query params', () => {
const p = parseResourceUri(
'gitnexus://group/acme/billing/contracts?type=http&repo=app%2Fapi&unmatchedOnly=true',
);
expect(p.kind).toBe('group');
if (p.kind !== 'group' || p.resourceType !== 'contracts') throw new Error('unexpected');
expect(p.groupName).toBe('acme/billing');
expect(p.contractsFilter).toEqual({
type: 'http',
repo: 'app/api',
unmatchedOnly: true,
});
});
it('coerces unmatchedOnly false from string', () => {
const p = parseResourceUri('gitnexus://group/g1/contracts?unmatchedOnly=false');
expect(p.kind).toBe('group');
if (p.kind !== 'group' || p.resourceType !== 'contracts') throw new Error('unexpected');
expect(p.contractsFilter.unmatchedOnly).toBe(false);
});
it('parses group status', () => {
const p = parseResourceUri('gitnexus://group/my/product/status');
expect(p).toEqual({
kind: 'group',
groupName: 'my/product',
resourceType: 'status',
});
});
it('round-trips repo URI like legacy regex', () => {
const p = parseResourceUri('gitnexus://repo/my%20project/schema');
expect(p).toEqual({
kind: 'repo',
repoName: 'my project',
resourceType: 'schema',
});
});
it('rejects unknown group resource tail', () => {
expect(() => parseResourceUri('gitnexus://group/foo/bar')).toThrow('Unknown group resource');
});
});
// ─── readResource URI parsing ────────────────────────────────────────
describe('readResource', () => {
@ -149,6 +213,22 @@ describe('readResource', () => {
expect(result).toContain('No repositories indexed');
});
it('routes group contracts resource through backend', async () => {
const backend = createMockBackend();
const uri = 'gitnexus://group/g1/contracts?type=http&unmatchedOnly=true';
await readResource(uri, backend);
expect(backend.readGroupContractsResource).toHaveBeenCalledWith('g1', {
type: 'http',
unmatchedOnly: true,
});
});
it('routes group status resource through backend', async () => {
const backend = createMockBackend();
await readResource('gitnexus://group/acme/status', backend);
expect(backend.readGroupStatusResource).toHaveBeenCalledWith('acme');
});
it('routes gitnexus://repo/{name}/context correctly', async () => {
const backend = createMockBackend({
context: {

View file

@ -148,6 +148,65 @@ describe('text-generator', () => {
expect(text).toContain('class Parser {');
expect(text).toContain('parseJSON(text: string) { return JSON.parse(text); }');
});
it('generates Struct text with structural metadata', () => {
const node: EmbeddableNode = {
...baseNode,
label: 'Struct',
name: 'User',
fieldNames: ['name', 'age'],
content: `struct User {
name: String,
age: u32,
}`,
};
const text = generateEmbeddingText(node, node.content);
expect(text).toContain('Struct: User');
expect(text).toContain('Properties: name, age');
expect(text).toContain('Container: struct User {');
expect(text).toContain('struct User {');
});
it('keeps compact container context on later structural chunks', () => {
const node: EmbeddableNode = {
...baseNode,
label: 'Class',
name: 'Parser',
methodNames: ['parseJSON', 'validate'],
fieldNames: ['options', 'cache'],
content: `class Parser {
options: ParserOptions;
cache: Map<string, any>;
parseJSON(text: string) { return JSON.parse(text); }
validate() { return true; }
}`,
};
const text = generateEmbeddingText(
node,
'validate() { return true; }',
{},
1,
'parseJSON(text: string) { return JSON.parse(text); }',
);
expect(text).toContain('Class: Parser');
expect(text).toContain('Container: class Parser {');
expect(text).toContain('[preceding context]: ...parseJSON(text: string)');
expect(text).not.toContain('Methods: parseJSON, validate');
expect(text).not.toContain('Properties: options, cache');
});
it('adds preceding context to non-structural chunk text', () => {
const text = generateEmbeddingText(
baseNode,
'return JSON.parse(text);',
{},
1,
'function parseJSON(text: string): Result<any> {',
);
expect(text).toContain('Function: parseJSON');
expect(text).toContain('[preceding context]: ...function parseJSON');
expect(text).toContain('return JSON.parse(text);');
});
});
describe('Constructor label', () => {

View file

@ -0,0 +1,110 @@
import { beforeEach, describe, expect, it, vi } from 'vitest';
const initMock = vi.fn();
const callToolMock = vi.fn();
const writeSyncMock = vi.fn();
vi.mock('../../src/mcp/local/local-backend.js', () => ({
LocalBackend: class {
init = initMock;
callTool = callToolMock;
},
}));
vi.mock('node:fs', () => ({
writeSync: writeSyncMock,
}));
describe('direct CLI tool commands', () => {
beforeEach(() => {
vi.resetModules();
initMock.mockReset();
callToolMock.mockReset();
writeSyncMock.mockReset();
initMock.mockResolvedValue(true);
});
it('dispatches detect_changes with CLI-shaped arguments', async () => {
callToolMock.mockResolvedValue({
summary: {
changed_files: 1,
changed_count: 2,
affected_count: 1,
risk_level: 'low',
},
});
const { detectChangesCommand } = await import('../../src/cli/tool.js');
await detectChangesCommand({
scope: 'compare',
baseRef: 'main',
repo: 'gitnexus',
});
expect(callToolMock).toHaveBeenCalledWith('detect_changes', {
scope: 'compare',
base_ref: 'main',
repo: 'gitnexus',
});
expect(writeSyncMock).toHaveBeenCalledWith(1, expect.stringContaining('Risk level: low'));
});
it('prints "No changes detected." when changed_count is 0', async () => {
callToolMock.mockResolvedValue({
summary: { changed_files: 0, changed_count: 0, affected_count: 0, risk_level: 'low' },
});
const { detectChangesCommand } = await import('../../src/cli/tool.js');
await detectChangesCommand({});
expect(writeSyncMock).toHaveBeenCalledWith(1, expect.stringContaining('No changes detected.'));
});
it('prints error message when result contains an error', async () => {
callToolMock.mockResolvedValue({ error: 'index is stale' });
const { detectChangesCommand } = await import('../../src/cli/tool.js');
await detectChangesCommand({});
expect(writeSyncMock).toHaveBeenCalledWith(1, expect.stringContaining('Error: index is stale'));
});
it('truncates changed_symbols list beyond 15 and shows overflow count', async () => {
const symbols = Array.from({ length: 17 }, (_, i) => ({
type: 'function',
name: `fn${i}`,
filePath: `src/file${i}.ts`,
}));
callToolMock.mockResolvedValue({
summary: { changed_files: 17, changed_count: 17, affected_count: 0, risk_level: 'low' },
changed_symbols: symbols,
});
const { detectChangesCommand } = await import('../../src/cli/tool.js');
await detectChangesCommand({});
const output: string = writeSyncMock.mock.calls[0][1];
expect(output).toContain('function fn14 → src/file14.ts');
expect(output).not.toContain('fn15');
expect(output).toContain('... and 2 more');
});
it('truncates affected_processes list beyond 10', async () => {
const processes = Array.from({ length: 12 }, (_, i) => ({
name: `proc${i}`,
step_count: 3,
changed_steps: [{ symbol: `sym${i}` }],
}));
callToolMock.mockResolvedValue({
summary: { changed_files: 1, changed_count: 1, affected_count: 12, risk_level: 'low' },
affected_processes: processes,
});
const { detectChangesCommand } = await import('../../src/cli/tool.js');
await detectChangesCommand({});
const output: string = writeSyncMock.mock.calls[0][1];
expect(output).toContain('proc9');
expect(output).not.toContain('proc10');
});
});

View file

@ -2,7 +2,7 @@
* Unit Tests: MCP Tool Definitions
*
* Tests: GITNEXUS_TOOLS from tools.ts
* - All 16 tools are defined (per-repo + group_*)
* - All 13 tools are defined (per-repo + group_list/group_sync)
* - Each tool has valid name, description, inputSchema
* - Required fields are correct
* - Optional repo parameter is present on tools that need it
@ -10,17 +10,11 @@
import { describe, it, expect } from 'vitest';
import { GITNEXUS_TOOLS } from '../../src/mcp/tools.js';
const GROUP_TOOLS = new Set([
'group_list',
'group_sync',
'group_contracts',
'group_query',
'group_status',
]);
const GROUP_TOOLS = new Set(['group_list', 'group_sync']);
describe('GITNEXUS_TOOLS', () => {
it('exports all tools (7 base + 3 route/tool/shape + 1 api_impact + 5 group)', () => {
expect(GITNEXUS_TOOLS).toHaveLength(16);
it('exports all tools (7 base + 3 route/tool/shape + 1 api_impact + 2 group)', () => {
expect(GITNEXUS_TOOLS).toHaveLength(13);
});
it('contains all expected tool names', () => {
@ -101,23 +95,29 @@ describe('GITNEXUS_TOOLS', () => {
}
});
it('group_contracts has optional repo filter', () => {
const groupContracts = GITNEXUS_TOOLS.find((t) => t.name === 'group_contracts')!;
expect(groupContracts.inputSchema.properties).toHaveProperty('repo');
expect(groupContracts.inputSchema.required).not.toContain('repo');
});
it('group tools without backend repo param omit repo property', () => {
for (const name of ['group_list', 'group_status', 'group_sync', 'group_query'] as const) {
for (const name of ['group_list', 'group_sync'] as const) {
const tool = GITNEXUS_TOOLS.find((t) => t.name === name)!;
expect(tool.inputSchema.properties).not.toHaveProperty('repo');
}
});
it('group_query requires name and query', () => {
const groupQuery = GITNEXUS_TOOLS.find((t) => t.name === 'group_query')!;
expect(groupQuery.inputSchema.required).toContain('name');
expect(groupQuery.inputSchema.required).toContain('query');
it('impact, query, and context expose optional service with minLength', () => {
for (const n of ['impact', 'query', 'context'] as const) {
const tool = GITNEXUS_TOOLS.find((t) => t.name === n)!;
const svc = tool.inputSchema.properties.service;
expect(svc, n).toBeDefined();
expect(svc!.minLength).toBe(1);
}
});
it('impact schema bounds match cross-impact validation ranges', () => {
const impact = GITNEXUS_TOOLS.find((t) => t.name === 'impact')!;
expect(impact.inputSchema.properties.maxDepth.minimum).toBe(1);
expect(impact.inputSchema.properties.maxDepth.maximum).toBe(32);
expect(impact.inputSchema.properties.minConfidence.minimum).toBe(0);
expect(impact.inputSchema.properties.minConfidence.maximum).toBe(1);
expect(impact.inputSchema.properties.timeoutMs.maximum).toBe(3600000);
});
it('detect_changes scope has correct enum values', () => {