mirror of
https://github.com/agentscope-ai/ReMe.git
synced 2026-09-11 22:51:10 +00:00
### 1. Agent Wrapper(统一 Agent 后端抽象) - **`base_agent_wrapper.py`**:`reply()` 返回值从 `tuple[str, Any]` 改为 `dict`(含 `session_id` / `last_message` / `result` / 可选 `structured_output`);`reply_stream()` 改为产出统一的 `StreamChunk`。废弃 `add_tools()`,改为 `add_job_tools(names: list[str])`(按名解析 BaseJob)与 `add_skills()`;新增 `_resolve_job_tools()`、`_merged_kwargs()`、`_chunk()` 辅助方法及 `project_path` / `project_skills_root` 属性。 - **`as_agent_wrapper.py`(AgentScope 后端)**: - 会话持久化重写:`session_path` 落地到 `<vault>/<session_dir>/agentscope/`,`_load_state` 支持 `resume` / `session_id` / `fork_session`,并做 UUID 校验(`_validate_session_id`);`_cleanup_expired_sessions` 按天数清理过期会话。 - 新增内置工具集(`BypassAnalysisBash` + Edit/Glob/Grep/Read/Write),`BypassAnalysisBash` 绕过 AgentScope 自带 Bash 静态分析以让 permission_mode 生效;`_resolve_skills()` 把配置的 skill 暴露给后端,`_load_tool_env()` 注入项目 `.env`。 - `_event_to_chunk()` 把 20+ 种 AgentScope 事件(Reply/Text/Thinking/Data/ToolCall/ToolResult/ModelCall/ExceedMaxIters)归一化为 `StreamChunk`。 - **`cc_agent_wrapper.py`(Claude Code SDK 后端,+551 行)**: - 新增 `_CcFileSessionStore`:基于 vault 的文件型会话存储,实现 append(按 uuid 去重)/ load / list / delete / list_subkeys,并对路径做 `_safe_parts` + `resolve()` 防越界校验。 - `_build_options()`:统一构建 `ClaudeAgentOptions`,处理 skills、disallowed_tools(默认禁 `WebSearch`)、`.env` 注入、Claude Code 的 API 凭据解析(`_claude_code_api_env`,多级 base_url/api_key 回退)、`CLAUDE_CONFIG_DIR` 设置、skill 目录软链接(`_ensure_claude_skill_dir`)。 - `_raw_event_to_chunk()` / `_message_content_to_chunks()`:把 Anthropic 流式事件(message_start/delta/stop、content_block_*)与 SDK 消息块(AssistantMessage/UserMessage/ResultMessage/RateLimitEvent)转换为统一 `StreamChunk`;跟踪 block_id/block_type/tool_call_name 做关联;处理尾部 `"success"` 误报异常的吞掉逻辑。 ### 2. 统一流式协议(StreamChunk / ChunkEnum) - **`stream_chunk.py`**:`StreamChunk` 扩展为承载 AS + CC 双后端完整信息的统一结构,新增 `session_id` / `block_id` / `tool_call_id` / `tool_call_name` / `media_type` / `input_tokens` / `output_tokens` 等字段,纯文本流仍保持轻量。 - **`chunk_enum.py`**:补全生命周期标记 `REPLY_START` / `REPLY_END`,并文档化两套后端事件 → ChunkEnum 的映射。 ### 3. Index 模块重构(变化批次化 + dispatch) - 新增 `_change_batch.py`:`coalesce_changes()` 把同路径多次事件折叠为最终状态(结合 path 存在性判定),`bucket_changes()` 按 watchfiles.Change 分桶。 - 新增 `init_changes.py`(`InitChangesStep`):一次性扫描,对比 file_store / file_catalog 已索引节点计算 added/modified/deleted,写入 `context["changes"]` 后 dispatch。 - 新增 `update_changes.py`:抽象基类 `ChangeApplyStep` 统一 added/modified/deleted 处理与错误收集;`UpdateCatalogStep`(写 file_catalog)、`UpdateIndexStep`(写 file_store,含按后缀解析 chunker)。 - **`watch_changes.py`**:改用 `dispatch_step_specs`(基类提供的 `dispatch_steps()`),每批先 `coalesce_changes` 再 dispatch;默认参数调整(debounce 5000ms / step 1000ms / poll 5000ms)并暴露常量。 - 删除旧步骤:`clear_and_scan` / `foreach_dispatch` / `scan_changes` / `update_catalog`(旧) / `update_index`(旧);`clear_store.py` 取代 clear_and_scan。 ### 4. Evolve / Dream 模块(拆分为多步 pipeline) - 删除旧的单体 `auto_dream.py` / `dream.py` / `dream.yaml`,新增 `dream/` 子包,按 5 个步骤组织: - **`extract.py`**:扫描当日 day-index + daily 笔记,对比 file_catalog 找出 changed/deleted,调用 LLM 全局抽取 `units`(procedure/personal/wiki 三桶)与 `topics`,路径与桶做清洗/路由。 - **`integrate.py`**:逐个 unit 调用 LLM 写入 digest,结构化输出 `IntegrateOutcome`(CREATE/CORROBORATE/REFINE/CORRECT),失败 unit/路径收集回写。 - **`topics.py`**:写 `daily/<date>/interests.yaml`,结合当天已有 + 近 N 天做去重(`normalize_topic`),可走 LLM 或纯规则去重两条路径。 - **`proactive.py`**:读取当日 `interests.yaml`,作为主动推荐话题的入口。 - **`finish.py`**:把变更路径落盘到 dream file_catalog(checkpoint),渲染最终汇总摘要。 - 新增 `schema.py`(`DreamState` 等跨步骤共享状态与结构化输出模型)与 `utils.py`(状态存取、扫描打包、YAML 读写、结构化回复解析等公共函数)。 - `evolve/__init__.py` 导出全部新 step。 ### 5. auto_memory / auto_resource(适配新 Agent API) - **`auto_memory.py`**:会话路径迁移到 `<session_dir>/dialog/<session_id>.jsonl`;改用 `job_tools`;新增 `source_conversation` frontmatter 反向链接(`_session_link`);执行后刷新 day 索引(`refresh_day_index`),并对 session_id 做合法性校验。 - **`auto_resource.py`**:资源改用「同名 daily note」方案(`_compute_note_stem` 取文件 stem);批量处理 `changes: list[dict]`(`_handle_change` 逐项处理,返回逐项结果摘要);agent 会话 id 用稳定的 `uuid5`;同样刷新 day 索引。 ### 6. BaseStep 基类增强 - 新增 `dispatch_steps` / `dispatch_step_specs` 机制:`_resolve_dispatch_step()` 支持字符串或 dict 形式的 step spec,`dispatch_steps()` 复用当前 context 调用下游 step。 - 新增 `config_value()`:按 key 取 app config,缺失时回退 `ApplicationConfig` 默认值。 - 小幅清理:`language` 初始化、`copy()`、`Ref.__init__` 签名精简。 ### 7. Components 改动 - **`file_store/local_file_store.py`**:持久化改用 zstd 压缩(`.jsonl.zst`,通过新 `utils/jsonl_zst.py`);upsert 时先删除旧 chunk 的 keyword 文档;embedding 复用改为 `(text, embedding)` 键控,要求文本一致才复用;新增 `_matches_search_filter()` 对 vector/keyword 搜索做 path/path_prefix/metadata 的统一后过滤。 - **`keyword_index/bm25_index.py`**:索引文件名加入组件名 + tokenizer 指纹(sha256 前 12 位),快照/恢复时校验指纹防配置漂移;空索引 dump 时删除文件,加载失败抛错而非静默。 - **`file_chunker/markdown_file_chunker.py`**:弃用 `python-frontmatter`,改用内置 YAML 解析(非法 YAML 不阻断正文索引),并修正因 frontmatter 占用行号导致的 AST 行号偏移(`line_offset`)。 - **`cron_job.py`**:大幅简化(-187 行),由原来「dispatch 外部 job/step + 多种调度模式」改为「在自身 steps 上跑 cron 表达式」;`Application` 启动顺序随之调整为 base > stream > background > cron。 - 其余小调整:service(base/http/mcp)、file_graph、file_catalog、as_llm、as_embedding、tokenizer、prompt_handler、base_component 的签名/接口微调。 ### 8. Application 生命周期 - `_start()` 启动顺序明确为 components → base → stream → background → cron,启动失败会触发 `_close()` 回滚并 re-raise(不再吞异常)。 - 启动时创建 `session_dir` 目录;新增 `update_component()`(按类型/名就地更新已存在组件,不存在则报错)。 ### 9. File IO / 路径安全 - **`_path.py`**:`resolve_path` 增加 vault 越界防护(`is_relative_to` 校验),禁止 `.` / `..` 路径分量,支持 `allow_empty`。 - **`read.py`**:大文件(超过 `MAX_FILE_READ_BYTES`)走按行读取 `read_file_lines_safe`,避免一次性载入内存。 - **`_file_io.py` / `_daily_index.py` / `_path.py`** 等支持函数补齐(如 `refresh_day_index`、`read_file_lines_safe`)。 - **`env_utils.py`**:新增 `parse_env_file()`,`load_env()` 返回加载到的键值、支持 `override`、对无路径调用做幂等缓存。 ### 10. Config - `ApplicationConfig` 新增 `session_dir`(默认 `reme_session`)。 - `config_parser.py`:环境变量展开后做类型转换(`_convert_value`)、dot-notation 与 key=value 参数校验更严格、配置文件路径支持相对 `_CONFIG_DIR` 查找、根非 dict 报错。 - `default.yaml`:作业编排改用 `init_changes_step` + `dispatch_steps`(index/resource/digest 三个 watch loop 与 reindex);新增 `auto_dream`(4 步)、`proactive` 作业,移除旧 `dream`;file_catalog 增配 `resource` / `digest` / `dream` 实例;LLM 默认值与 Claude Code 凭据配置调整(tool_result_limit 50000、thinking_enable=false 等)。 ### 11. 其它 - 新增 `steps/common/add.py`(`AddStep` 算术 demo)、`channel/__init__.py` 与 common `__init__` 导出整理。 - 新增 4 篇文档:`docs4/auto_dream_logic_and_step_refactor.md`、`docs4/watch_loop_step_refactor_plan.md`、`docs4/todo.md`,以及 `reme_design.md` 更新。 **
315 lines
13 KiB
Python
315 lines
13 KiB
Python
"""Integration test for the auto_resource job.
|
|
|
|
Drives the ``auto_resource`` step against a real LLM. Three scenarios:
|
|
|
|
1. **CREATE (added)** / **UPDATE (modified)**: places a resource file in
|
|
``resource/{date}/``, calls ``auto_resource`` with a ``changes`` batch,
|
|
and expects the agent to write/update the same-name daily note.
|
|
|
|
2. **DELETE (deleted)**: seeds a resource note under
|
|
``daily/{date}/{resource_stem}.md``, calls ``auto_resource`` with a
|
|
deleted change, and expects the note file to be removed.
|
|
|
|
Requires LLM_API_KEY (and optionally LLM_BASE_URL / LLM_MODEL_NAME) in the
|
|
environment or a .env file at the repo root. Hits the real LLM API.
|
|
"""
|
|
|
|
import asyncio
|
|
import sys
|
|
from pathlib import Path
|
|
|
|
INTEGRATION_DIR = Path(__file__).resolve().parent
|
|
sys.path.insert(0, str(INTEGRATION_DIR))
|
|
|
|
# pylint: disable=wrong-import-position
|
|
from _vault_fixture import vault_env # noqa: E402
|
|
|
|
from reme4.steps.evolve.auto_resource import _compute_agent_session_id, _compute_note_stem # noqa: E402
|
|
|
|
RESOURCE_FILENAME = "project-roadmap.md"
|
|
RESOURCE_CONTENT_V1 = """\
|
|
# Project Roadmap 2026 Q3
|
|
|
|
## Goals
|
|
- Launch v2.0 API by July 15
|
|
- Migrate 80% of users to new auth system by August 1
|
|
- Reduce p99 latency to < 200ms
|
|
|
|
## Milestones
|
|
| Date | Milestone | Owner |
|
|
|------------|------------------------|---------|
|
|
| 2026-07-01 | API beta release | Alice |
|
|
| 2026-07-15 | API GA | Alice |
|
|
| 2026-08-01 | Auth migration done | Bob |
|
|
| 2026-08-15 | Performance target met | Charlie |
|
|
|
|
## Risks
|
|
- Auth migration blocked on legacy client deprecation (ETA: June 30)
|
|
- Performance target requires Redis cluster upgrade (budget approved)
|
|
"""
|
|
|
|
RESOURCE_CONTENT_V2 = """\
|
|
# Project Roadmap 2026 Q3 (Revised)
|
|
|
|
## Goals
|
|
- Launch v2.0 API by July 20 (delayed 5 days from original July 15)
|
|
- Migrate 80% of users to new auth system by August 1
|
|
- Reduce p99 latency to < 150ms (tightened from 200ms)
|
|
|
|
## Milestones
|
|
| Date | Milestone | Owner |
|
|
|------------|------------------------|---------|
|
|
| 2026-07-05 | API beta release | Alice |
|
|
| 2026-07-20 | API GA | Alice |
|
|
| 2026-08-01 | Auth migration done | Bob |
|
|
| 2026-08-15 | Performance target met | Charlie |
|
|
| 2026-08-20 | Post-launch review | Dave |
|
|
|
|
## Risks
|
|
- Auth migration blocked on legacy client deprecation (resolved June 28)
|
|
- Performance target requires Redis cluster upgrade (completed July 1)
|
|
- New risk: third-party OAuth provider rate limiting during migration
|
|
"""
|
|
|
|
|
|
def _read_text(p: Path) -> str:
|
|
return p.read_text(encoding="utf-8")
|
|
|
|
|
|
def _print_text_file(label: str, path: Path) -> str:
|
|
text = _read_text(path)
|
|
print("\n" + "=" * 70)
|
|
print(f"[{label}] {path} ({len(text)} bytes)")
|
|
print(f"[{label}] body:\n{text}")
|
|
print("=" * 70)
|
|
return text
|
|
|
|
|
|
def _print_message_files(label: str, paths: list[Path]) -> None:
|
|
for idx, path in enumerate(paths, 1):
|
|
if not path.is_file():
|
|
print(f"[{label}] message file missing: {path}")
|
|
continue
|
|
text = _read_text(path)
|
|
print("\n" + "=" * 70)
|
|
print(f"[{label}] message {idx}: {path} ({len(text)} bytes)")
|
|
print(text)
|
|
print("=" * 70)
|
|
|
|
|
|
def test_auto_resource_create():
|
|
"""CREATE branch: agent writes the same-name daily note and saves its AgentScope session."""
|
|
|
|
async def run():
|
|
with vault_env() as env:
|
|
app = await env.make_app()
|
|
try:
|
|
today = env.today
|
|
|
|
print("\n" + "=" * 70)
|
|
print("[setup] vault_root =", env.vault_dir)
|
|
print("[setup] today =", today)
|
|
print("=" * 70)
|
|
|
|
file_path = env.place_resource(RESOURCE_FILENAME, RESOURCE_CONTENT_V1)
|
|
note_stem = _compute_note_stem(RESOURCE_FILENAME)
|
|
agent_session_id = _compute_agent_session_id(file_path)
|
|
expected_session_jsonl = env.vault_dir / "reme_session" / "agentscope" / f"{agent_session_id}.jsonl"
|
|
|
|
print(f"[CREATE] file_path = {file_path}")
|
|
print(f"[CREATE] note_stem = {note_stem}")
|
|
print(f"[CREATE] agent_session_id = {agent_session_id}")
|
|
print(f"[CREATE] expected transcript = {expected_session_jsonl.relative_to(env.vault_dir)}")
|
|
|
|
with env.record_agents(prefix="agent_resource_create") as recorder:
|
|
response = await app.run_job(
|
|
"auto_resource",
|
|
changes=[{"path": file_path, "change": "added"}],
|
|
)
|
|
dumped = await recorder.dump()
|
|
for p in dumped:
|
|
print(f"[CREATE] agent memory dumped: {p}")
|
|
|
|
assert response.success is True, f"CREATE job failed: {response.answer!r}"
|
|
meta = response.metadata or {}
|
|
result_meta = (meta.get("results") or [{}])[0].get("metadata") or {}
|
|
assert result_meta.get("action") == "added", f"Unexpected action: {meta!r}"
|
|
assert result_meta.get("session_id") == note_stem, f"Unexpected session_id: {meta!r}"
|
|
assert result_meta.get("path") == f"daily/{today}/{note_stem}.md", f"Unexpected note path: {meta!r}"
|
|
note_path = env.vault_dir / "daily" / today / f"{note_stem}.md"
|
|
assert note_path.is_file()
|
|
|
|
assert expected_session_jsonl.is_file(), (
|
|
f"agent session not persisted at {expected_session_jsonl}; "
|
|
f"AgentScope files: "
|
|
f"{[p.name for p in (env.vault_dir / 'reme_session' / 'agentscope').glob('*.jsonl')]}"
|
|
)
|
|
|
|
note_text = _print_text_file("CREATE result.md", note_path)
|
|
_print_message_files("CREATE intermediate messages", [*dumped, expected_session_jsonl])
|
|
|
|
note_hits = [
|
|
needle
|
|
for needle in ("v2.0", "July 15", "Alice", "Bob", "p99", "200ms", "Redis")
|
|
if needle in note_text
|
|
]
|
|
print(f"[CREATE] landed note facts: {note_hits}")
|
|
assert (
|
|
len(note_hits) >= 3
|
|
), f"CREATE note missed expected facts {note_hits!r}\n--- NOTE ---\n{note_text}"
|
|
|
|
transcript = _read_text(expected_session_jsonl)
|
|
topic_hits = [
|
|
needle
|
|
for needle in ("v2.0", "July 15", "Alice", "Bob", "p99", "200ms", "Redis", file_path)
|
|
if needle in transcript
|
|
]
|
|
print(f"[CREATE] facts visible in transcript: {topic_hits}")
|
|
assert topic_hits, (
|
|
"agent transcript shows no signal it actually read the resource file; "
|
|
f"transcript head:\n{transcript[:500]}"
|
|
)
|
|
|
|
print("\n" + "=" * 70)
|
|
print("test_auto_resource_create passed")
|
|
print("=" * 70)
|
|
finally:
|
|
await env.close_all()
|
|
|
|
asyncio.run(run())
|
|
|
|
|
|
def test_auto_resource_update():
|
|
"""UPDATE branch: agent updates the same-name daily note and appends to its AgentScope session."""
|
|
|
|
async def run():
|
|
with vault_env() as env:
|
|
app = await env.make_app()
|
|
try:
|
|
today = env.today
|
|
|
|
print("\n" + "=" * 70)
|
|
print("[setup] vault_root =", env.vault_dir)
|
|
print("[setup] today =", today)
|
|
print("=" * 70)
|
|
|
|
# First run as "added" so the resource file exists and the
|
|
# initial transcript lands.
|
|
file_path = env.place_resource(RESOURCE_FILENAME, RESOURCE_CONTENT_V1)
|
|
note_stem = _compute_note_stem(RESOURCE_FILENAME)
|
|
agent_session_id = _compute_agent_session_id(file_path)
|
|
session_jsonl = env.vault_dir / "reme_session" / "agentscope" / f"{agent_session_id}.jsonl"
|
|
|
|
response = await app.run_job("auto_resource", changes=[{"path": file_path, "change": "added"}])
|
|
assert response.success is True, f"Initial create failed: {response.answer!r}"
|
|
assert session_jsonl.is_file(), "initial added run did not save the AgentScope session"
|
|
size_before = session_jsonl.stat().st_size
|
|
print(f"[UPDATE] transcript before modify ({size_before} bytes)")
|
|
|
|
# Now update the resource file and call with "modified".
|
|
env.place_resource(RESOURCE_FILENAME, RESOURCE_CONTENT_V2)
|
|
|
|
with env.record_agents(prefix="agent_resource_update") as recorder:
|
|
response = await app.run_job(
|
|
"auto_resource",
|
|
changes=[{"path": file_path, "change": "modified"}],
|
|
)
|
|
dumped = await recorder.dump()
|
|
for p in dumped:
|
|
print(f"[UPDATE] agent memory dumped: {p}")
|
|
|
|
assert response.success is True, f"UPDATE job failed: {response.answer!r}"
|
|
meta = response.metadata or {}
|
|
result_meta = (meta.get("results") or [{}])[0].get("metadata") or {}
|
|
assert result_meta.get("action") == "modified", f"Unexpected action: {meta!r}"
|
|
assert result_meta.get("session_id") == note_stem, f"Unexpected session_id: {meta!r}"
|
|
assert result_meta.get("path") == f"daily/{today}/{note_stem}.md", f"Unexpected note path: {meta!r}"
|
|
note_path = env.vault_dir / "daily" / today / f"{note_stem}.md"
|
|
assert note_path.is_file()
|
|
|
|
size_after = session_jsonl.stat().st_size
|
|
print(f"[UPDATE] transcript after modify ({size_after} bytes)")
|
|
assert size_after > size_before, (
|
|
f"transcript did not grow after modified run " f"({size_before} -> {size_after})"
|
|
)
|
|
|
|
note_text = _print_text_file("UPDATE result.md", note_path)
|
|
_print_message_files("UPDATE intermediate messages", [*dumped, session_jsonl])
|
|
|
|
note_hits = [
|
|
needle
|
|
for needle in ("July 20", "150ms", "Dave", "rate limiting", "resolved")
|
|
if needle in note_text
|
|
]
|
|
print(f"[UPDATE] landed note facts: {note_hits}")
|
|
assert (
|
|
len(note_hits) >= 2
|
|
), f"UPDATE note missed expected facts {note_hits!r}\n--- NOTE ---\n{note_text}"
|
|
|
|
transcript = _read_text(session_jsonl)
|
|
new_hits = [
|
|
needle
|
|
for needle in ("July 20", "150ms", "Dave", "rate limiting", "resolved")
|
|
if needle in transcript
|
|
]
|
|
print(f"[UPDATE] V2 facts visible in transcript: {new_hits}")
|
|
assert new_hits, "modified run added no V2 content to the transcript; " f"tail:\n{transcript[-800:]}"
|
|
|
|
print("\n" + "=" * 70)
|
|
print("test_auto_resource_update passed")
|
|
print("=" * 70)
|
|
finally:
|
|
await env.close_all()
|
|
|
|
asyncio.run(run())
|
|
|
|
|
|
def test_auto_resource_delete():
|
|
"""DELETE a resource note (change=deleted)."""
|
|
|
|
async def run():
|
|
with vault_env() as env:
|
|
app = await env.make_app()
|
|
try:
|
|
today = env.today
|
|
|
|
print("\n" + "=" * 70)
|
|
print("[setup] vault_root =", env.vault_dir)
|
|
print("[setup] today =", today)
|
|
print("=" * 70)
|
|
|
|
note_stem = _compute_note_stem(RESOURCE_FILENAME)
|
|
file_path = f"resource/{today}/{RESOURCE_FILENAME}"
|
|
|
|
seed_body = "---\nname: test\ndescription: test note\n---\n\nSome content.\n"
|
|
note_path = env.seed_daily_note(note_stem, seed_body)
|
|
assert note_path.is_file()
|
|
print(f"[DELETE] seeded note: {note_path}")
|
|
|
|
response = await app.run_job(
|
|
"auto_resource",
|
|
changes=[{"path": file_path, "change": "deleted"}],
|
|
)
|
|
|
|
assert response.success is True, f"DELETE job failed: {response.answer!r}"
|
|
meta = response.metadata or {}
|
|
result_meta = (meta.get("results") or [{}])[0].get("metadata") or {}
|
|
assert result_meta.get("action") == "deleted"
|
|
assert not note_path.is_file(), f"Note file still exists after delete: {note_path}"
|
|
|
|
print(f"[DELETE] note removed: {note_path}")
|
|
print("\n" + "=" * 70)
|
|
print("test_auto_resource_delete passed")
|
|
print("=" * 70)
|
|
finally:
|
|
await env.close_all()
|
|
|
|
asyncio.run(run())
|
|
|
|
|
|
if __name__ == "__main__":
|
|
print("=== auto_resource integration test ===")
|
|
test_auto_resource_create()
|
|
test_auto_resource_update()
|
|
test_auto_resource_delete()
|
|
print("\nAll integration tests passed!")
|