refactor(rust): extract inference-testing crate (#44873)

Move the shared test helpers out of litellm-inference's test-support
feature into a publish = false litellm-inference-testing crate used only
as a dev dependency by the format crates.

Also drop the dead src/constants.rs (OPENAI_DEFAULT_API_BASE had no
users) and declare the litellm-http/litellm-llms test-support features
on the crates that actually use them instead of relying on feature
unification through litellm-inference's dev-dependencies.

Co-authored-by: Yujong Lee <yujong@berri.ai>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

litellm-rust/Cargo.lock

@@ -4013,7 +4013,6 @@ dependencies = [
  "litellm-framer",
  "litellm-host",
  "litellm-http",
- "litellm-inference",
  "litellm-llms",
  "litellm-secrets",
  "litellm-tracing",
@@ -4042,6 +4041,7 @@ dependencies = [
  "litellm-host-native",
  "litellm-http",
  "litellm-inference",
+ "litellm-inference-testing",
  "litellm-llms",
  "litellm-llms-types",
  "litellm-secrets",
@@ -4068,6 +4068,7 @@ dependencies = [
  "litellm-host-native",
  "litellm-http",
  "litellm-inference",
+ "litellm-inference-testing",
  "litellm-llms",
  "litellm-llms-types",
  "litellm-secrets",
@@ -4096,6 +4097,7 @@ dependencies = [
  "litellm-host-native",
  "litellm-http",
  "litellm-inference",
+ "litellm-inference-testing",
  "litellm-llms",
  "litellm-llms-types",
  "litellm-secrets",
@@ -4125,6 +4127,7 @@ dependencies = [
  "litellm-host-native",
  "litellm-http",
  "litellm-inference",
+ "litellm-inference-testing",
  "litellm-llms",
  "litellm-llms-types",
  "litellm-secrets",
@@ -4138,6 +4141,16 @@ dependencies = [
  "wiremock",
 ]
 
+[[package]]
+name = "litellm-inference-testing"
+version = "0.1.0"
+dependencies = [
+ "futures-util",
+ "litellm-http",
+ "litellm-inference",
+ "litellm-secrets",
+]
+
 [[package]]
 name = "litellm-inference-transcription"
 version = "0.1.0"
@@ -4147,6 +4160,7 @@ dependencies = [
  "litellm-core-utils",
  "litellm-http",
  "litellm-inference",
+ "litellm-inference-testing",
  "litellm-llms",
  "litellm-secrets",
  "litellm-tracing",

litellm-rust/Cargo.toml

@@ -22,6 +22,7 @@ litellm-inference-responses = { path = "crates/inference-responses" }
 litellm-inference-messages = { path = "crates/inference-messages" }
 litellm-inference-chat = { path = "crates/inference-chat" }
 litellm-inference-ocr = { path = "crates/inference-ocr" }
+litellm-inference-testing = { path = "crates/inference-testing" }
 litellm-gateway-mcp = { path = "crates/gateway-mcp" }
 litellm-gateway = { path = "crates/gateway" }
 litellm-gateway-inference = { path = "crates/gateway-inference" }

litellm-rust/crates/inference-chat/Cargo.toml

@@ -22,7 +22,8 @@ tracing.workspace = true
 [dev-dependencies]
 litellm-cache-memory.workspace = true
 litellm-host-native.workspace = true
-litellm-inference = { workspace = true, features = ["test-support"] }
+litellm-http = { workspace = true, features = ["test-support"] }
+litellm-inference-testing.workspace = true
 litellm-tracing.workspace = true
 rstest.workspace = true
 tokio.workspace = true

litellm-rust/crates/inference-chat/tests/support/mod.rs

@@ -8,7 +8,7 @@ use std::{
     sync::{Arc, Mutex},
 };
 
-use litellm_inference::test_support::{http_config, no_secrets, provider_http, resources};
+use litellm_inference_testing::{http_config, no_secrets, provider_http, resources};
 use serde_json::Value;
 use wiremock::{Mock, MockServer, Request, ResponseTemplate, matchers::any};
 

litellm-rust/crates/inference-messages/Cargo.toml

@@ -27,7 +27,8 @@ tracing.workspace = true
 [dev-dependencies]
 litellm-cache-memory.workspace = true
 litellm-host-native.workspace = true
-litellm-inference = { workspace = true, features = ["test-support"] }
+litellm-http = { workspace = true, features = ["test-support"] }
+litellm-inference-testing.workspace = true
 rstest.workspace = true
 tokio.workspace = true
 wiremock.workspace = true

litellm-rust/crates/inference-messages/tests/messages/host.rs

@@ -5,8 +5,8 @@ use litellm_host::{
     interceptors::{ExecutionFacts, RequestContext, ResultSource, WireRequest},
     lifecycle::CallEvent,
 };
-use litellm_inference::test_support::{RecordingSecrets, no_secrets};
 use litellm_inference_messages::{MessagesCallResponse, route::Messages};
+use litellm_inference_testing::{RecordingSecrets, no_secrets};
 use litellm_llms::base_llm::messages::context::MessagesModelCapabilities as AnthropicModelCapabilities;
 use rstest::rstest;
 

litellm-rust/crates/inference-messages/tests/messages/main.rs

@@ -4,11 +4,11 @@ use std::{
 };
 
 use litellm_http::{HttpSettings, Resolution};
-use litellm_inference::test_support::RecordingSecrets;
 use litellm_inference_messages::{
     Error, MessagesCall, MessagesShaping,
     route::{Messages, MessagesMachine, MessagesOutput},
 };
+use litellm_inference_testing::RecordingSecrets;
 use litellm_llms_types::formats::messages::{MessagesRequest, MessagesResponse};
 use litellm_secrets::source::SecretSource;
 use rstest::fixture;

litellm-rust/crates/inference-messages/tests/messages/response.rs

@@ -3,10 +3,10 @@ use litellm_host::{
     lifecycle::ExecutionEvent,
 };
 use litellm_http::transport::Error as TransportError;
-use litellm_inference::test_support::{
+use litellm_inference_messages::{MessagesCallResponse, messages_body};
+use litellm_inference_testing::{
     RecordingSecrets, http_config, no_secrets, provider_http, resources,
 };
-use litellm_inference_messages::{MessagesCallResponse, messages_body};
 use rstest::rstest;
 
 use super::*;

litellm-rust/crates/inference-messages/tests/messages/secrets.rs

@@ -1,4 +1,4 @@
-use litellm_inference::test_support::RecordingSecrets;
+use litellm_inference_testing::RecordingSecrets;
 use rstest::rstest;
 
 use super::*;

litellm-rust/crates/inference-messages/tests/messages/stream.rs

@@ -5,11 +5,11 @@ use std::{
 
 use bytes::Bytes;
 use futures_util::{StreamExt, TryStreamExt};
-use litellm_inference::test_support::{RecordingSecrets, no_secrets};
 use litellm_inference_messages::{
     MessagesCallResponse,
     route::{Messages, MessagesStreamHead},
 };
+use litellm_inference_testing::{RecordingSecrets, no_secrets};
 use litellm_tracing::{Logger, Metadata, Record, Sink};
 use rstest::rstest;
 use tokio::{

litellm-rust/crates/inference-messages/tests/support/mod.rs

@@ -8,7 +8,7 @@ use std::{
     sync::{Arc, Mutex},
 };
 
-use litellm_inference::test_support::{http_config, provider_http, resources};
+use litellm_inference_testing::{http_config, provider_http, resources};
 use litellm_secrets::source::SecretSource;
 use serde_json::Value;
 use wiremock::{Mock, MockServer, Request, ResponseTemplate, matchers::any};

litellm-rust/crates/inference-ocr/Cargo.toml

@@ -29,7 +29,8 @@ litellm-auth-aws.workspace = true
 litellm-auth-gcp.workspace = true
 litellm-host-native.workspace = true
 litellm-http = { workspace = true, features = ["test-support"] }
-litellm-inference = { workspace = true, features = ["test-support"] }
+litellm-inference-testing.workspace = true
+litellm-llms = { workspace = true, features = ["test-support"] }
 rand.workspace = true
 rstest.workspace = true
 time.workspace = true

litellm-rust/crates/inference-ocr/tests/ocr/documents.rs

@@ -1,7 +1,7 @@
 use base64::Engine;
 use litellm_host::interceptors::WireRequest;
-use litellm_inference::test_support::{http_config, no_secrets, resources};
 use litellm_inference_ocr::types::OcrDocumentInput;
+use litellm_inference_testing::{http_config, no_secrets, resources};
 use rstest::rstest;
 use wiremock::{Mock, matchers::any};
 

litellm-rust/crates/inference-ocr/tests/ocr/main.rs

@@ -2,7 +2,6 @@ use litellm_host::{
     interceptors::{RequestContext, WireRequest},
     lifecycle::CallEvent,
 };
-use litellm_inference::test_support::{http_config, no_secrets, resources};
 use litellm_inference_ocr::{
     OcrRoute,
     document::prepare_document,
@@ -10,6 +9,7 @@ use litellm_inference_ocr::{
     types::{LiteLLMOcrRequest, OcrDocumentInput},
     wire::{OcrWireRequest, decode_request},
 };
+use litellm_inference_testing::{http_config, no_secrets, resources};
 use litellm_llms::base_llm::ocr::{error::Error, settings::OcrSettings};
 use litellm_llms_types::formats::ocr::{LiteLLMOcrResponse, OcrDocument};
 use serde_json::{Map, Value, json};

litellm-rust/crates/inference-ocr/tests/ocr/mistral.rs

@@ -1,7 +1,7 @@
 use std::sync::Arc;
 
 use litellm_http::{HttpSettings, Resolution};
-use litellm_inference::test_support::{RecordingSecrets, http_config, no_secrets, resources};
+use litellm_inference_testing::{RecordingSecrets, http_config, no_secrets, resources};
 use litellm_llms::{
     base_llm::ocr::transformation::{BaseOcrConfig, OCR_RESPONSE_MAX_BYTES},
     mistral::ocr::transformation::MistralOcrConfig,

litellm-rust/crates/inference-ocr/tests/resources.rs

@@ -11,8 +11,8 @@ use litellm_auth_gcp::{
 };
 use litellm_http::{HttpSettings, Resolution};
 use litellm_inference::resources::CoreResources;
-use litellm_inference::test_support::{RecordingSecrets, http_pool};
 use litellm_inference_ocr::wire::{OcrWireRequest, decode_request};
+use litellm_inference_testing::{RecordingSecrets, http_pool};
 use litellm_llms::base_llm::ocr::settings::OcrSettings;
 use rstest::{fixture, rstest};
 use serde_json::json;

litellm-rust/crates/inference-responses/Cargo.toml

@@ -26,7 +26,7 @@ tracing.workspace = true
 litellm-cache.workspace = true
 litellm-cache-memory.workspace = true
 litellm-host-native.workspace = true
-litellm-inference = { workspace = true, features = ["test-support"] }
+litellm-inference-testing.workspace = true
 litellm-tracing.workspace = true
 rstest.workspace = true
 wiremock.workspace = true

litellm-rust/crates/inference-responses/tests/responses.rs

@@ -6,11 +6,11 @@ use std::sync::Arc;
 
 use futures_util::TryStreamExt;
 use litellm_host::{call::HostedCompletion, lifecycle::CallEvent};
-use litellm_inference::test_support::{RecordingSecrets, no_secrets};
 use litellm_inference_responses::{
     route::Responses,
     types::{ResponsesCall, ResponsesOutput},
 };
+use litellm_inference_testing::{RecordingSecrets, no_secrets};
 use rstest::{fixture, rstest};
 use serde_json::json;
 use wiremock::ResponseTemplate;

litellm-rust/crates/inference-responses/tests/support/mod.rs

@@ -8,7 +8,7 @@ use std::{
     sync::{Arc, Mutex},
 };
 
-use litellm_inference::test_support::{http_config, provider_http, resources};
+use litellm_inference_testing::{http_config, provider_http, resources};
 use litellm_secrets::source::SecretSource;
 use serde_json::Value;
 use wiremock::{Mock, MockServer, Request, ResponseTemplate, matchers::any};

litellm-rust/crates/inference-testing/Cargo.toml

@@ -0,0 +1,13 @@
+[package]
+name = "litellm-inference-testing"
+version = "0.1.0"
+edition.workspace = true
+license.workspace = true
+repository.workspace = true
+publish = false
+
+[dependencies]
+futures-util.workspace = true
+litellm-http.workspace = true
+litellm-inference.workspace = true
+litellm-secrets.workspace = true

litellm-rust/crates/inference-testing/src/lib.rs

@@ -7,7 +7,7 @@ use litellm_http::{
 };
 use litellm_secrets::{SecretValue, source::SecretSource};
 
-use crate::resources::CoreResources;
+use litellm_inference::resources::CoreResources;
 
 pub fn http_pool() -> HttpClientPool {
     HttpClientPool::new(Arc::new(PublicDnsResolver))
@@ -32,7 +32,6 @@ pub fn http_config() -> HttpClientConfig {
     Resolution::from(&HttpSettings::default()).config
 }
 
-/// A secret source that answers from a fixed table and records every name it was asked for.
 pub struct RecordingSecrets {
     values: Vec<(String, String)>,
     fails: bool,

litellm-rust/crates/inference-transcription/Cargo.toml

@@ -17,7 +17,7 @@ serde_json.workspace = true
 tracing.workspace = true
 
 [dev-dependencies]
-litellm-inference = { workspace = true, features = ["test-support"] }
+litellm-inference-testing.workspace = true
 litellm-tracing.workspace = true
 rstest.workspace = true
 tokio.workspace = true

litellm-rust/crates/inference-transcription/tests/support/mod.rs

@@ -5,7 +5,7 @@
 
 use std::sync::{Arc, Mutex};
 
-use litellm_inference::test_support::{http_config, no_secrets, provider_http, resources};
+use litellm_inference_testing::{http_config, no_secrets, provider_http, resources};
 use serde_json::Value;
 use wiremock::{Mock, MockServer, Request, ResponseTemplate, matchers::any};
 

litellm-rust/crates/inference/AGENTS.md

@@ -5,7 +5,7 @@
   - `RouteError` (`src/error.rs`), `CallOptions` (`src/lib.rs`), `CallContext` (`src/context.rs`)
   - diagnostic spans (`src/diagnostic.rs`), outbound send and signing (`src/outbound.rs`), provider resolution (`src/provider.rs`), `CoreResources` (`src/resources.rs`)
   - response caching (`src/caching.rs`): `Cachable`, `StreamCachable`, `CacheRequest`, `CallCache`, `execute_unary`, `execute_streaming`, stream capture
-  - shared test helpers behind the `test-support` feature (`src/test_support.rs`)
+  - Shared test helpers belong in litellm-inference-testing, used only as a dev dependency
 - Nothing here names an API format; format-specific code, constants, tests and test builders live in their `inference-<fmt>` crate
 - Never depend on an `inference-*` crate from here
 

litellm-rust/crates/inference/Cargo.toml

@@ -5,9 +5,6 @@ edition.workspace = true
 license.workspace = true
 repository.workspace = true
 
-[features]
-test-support = []
-
 [dependencies]
 litellm-cache.workspace = true
 litellm-cache-response.workspace = true
@@ -32,7 +29,6 @@ time.workspace = true
 url.workspace = true
 
 [dev-dependencies]
-litellm-inference = { workspace = true, features = ["test-support"] }
 litellm-cache-memory.workspace = true
 litellm-http = { workspace = true, features = ["test-support"] }
 litellm-llms = { workspace = true, features = ["test-support"] }

litellm-rust/crates/inference/src/constants.rs

@@ -1 +0,0 @@
-pub const OPENAI_DEFAULT_API_BASE: &str = "https://api.openai.com";

litellm-rust/crates/inference/src/lib.rs

@@ -2,13 +2,10 @@ pub mod context;
 pub mod diagnostic;
 
 pub mod caching;
-pub mod constants;
 pub mod error;
 pub mod outbound;
 pub mod provider;
 pub mod resources;
-#[cfg(feature = "test-support")]
-pub mod test_support;
 
 pub use error::RouteError;