🧪 Add tests for DiscoveredFile.from_path conditionals

.github/workflows/claude.yml

@@ -69,8 +69,8 @@ jobs:
         with:
           claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          allowed_non_write_users: Copilot,copilot,jules[bot],jules
-          allowed_bots: "github-actions[bot],copilot[bot],dependabot[bot],copilot,github-actions,gemini[bot],claude[bot],jules[bot]"
+          allowed_non_write_users: Copilot
+          allowed_bots: "github-actions[bot],copilot[bot],dependabot[bot],copilot,github-actions,gemini[bot],claude[bot]"
           trigger_phrase: "@claude"
           assignee_trigger: claude[bot]
           label_trigger: claude
@@ -105,8 +105,8 @@ jobs:
         with:
           claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          allowed_non_write_users: Copilot,copilot,jules[bot],jules
-          allowed_bots: "github-actions[bot],copilot[bot],dependabot[bot],copilot,github-actions,gemini[bot],claude[bot],jules[bot]"
+          allowed_non_write_users: Copilot
+          allowed_bots: "github-actions[bot],copilot[bot],dependabot[bot],copilot,github-actions,gemini[bot],claude[bot]"
           trigger_phrase: "@claude"
           assignee_trigger: claude
           label_trigger: claude
@@ -142,8 +142,6 @@ jobs:
         with:
           claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          allowed_non_write_users: Copilot,copilot,jules[bot],jules
-          allowed_bots: "github-actions[bot],copilot[bot],dependabot[bot],copilot,github-actions,gemini[bot],claude[bot],jules[bot]"
           trigger_phrase: "@claude"
           assignee_trigger: claude
           label_trigger: claude
@@ -181,8 +179,6 @@ jobs:
         with:
           claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          allowed_non_write_users: Copilot,copilot,jules[bot],jules
-          allowed_bots: "github-actions[bot],copilot[bot],dependabot[bot],copilot,github-actions,gemini[bot],claude[bot],jules[bot]"
           trigger_phrase: "@claude"
           assignee_trigger: claude
           label_trigger: claude

.gitignore

@@ -229,5 +229,5 @@ test-results.xml
 mise.local.toml
 mise.local.env
 
-.exportify/
-!.exportify/config.toml
+.gemini/
+gha-creds-*.json

pyproject.toml

@@ -120,7 +120,7 @@ dependencies = [
   # So we pin it to make sure we don't break on minor releases
   "pydantic==2.12.5",
   # for, you know, platform dirs
-  "platformdirs>=4.9.4",
+  "platformdirs>=4.9.2",
   # psutil used for resource governance/limiting by engine
   "psutil>=7.2.2",
   "textcase>=0.4.5",
@@ -140,20 +140,20 @@ dependencies = [
   # for local providers (sentence-transformers, fastembed) to detect CPU/GPU features
   "py-cpuinfo>=9.0.0",
   # * ================ CLI Dependencies ==================*
-  "cyclopts>=4.10.0",
-  "rich>=14.3.3",
+  "cyclopts>=4.5.1",
+  "rich>=14.3.0",
   # * ================ Provider Clients ==================*
   # we must pin these to specific versions to ensure compatibility with our ClientOptions subclasses
   "boto3==1.42.19",
-  "cohere==5.20.7",
+  "cohere==5.20.1",
   "fastembed==0.7.4; python_version < '3.14'",
   "google-genai==1.56.0",
   # NOTE: We're waiting on pydantic-ai to update to 1.0+ before we can upgrade too
-  "huggingface-hub>=1.7.1",
+  "huggingface-hub==0.36.2",
   "mistralai==1.10.0",
-  "openai==2.28.0",
-  "qdrant-client==1.17.1",
-  "pydantic-ai-slim>=1.68.0",
+  "openai==2.17.0",
+  "qdrant-client==1.16.2",
+  "pydantic-ai-slim>=1.56.0",
   "sentence-transformers==5.2.0; python_version <= '3.14'",
   "voyageai==0.3.7",
   # * ================ Indexing and Engine ==================*
@@ -171,12 +171,12 @@ dependencies = [
   # fastmcp is the core MCP server framework
   "fastmcp>=2.14.5",
   # just used for types but we need them at runtime for Pydantic models
-  "mcp>=1.23.3",
+  "mcp>=1.19.0",
   # Runs the core admin/management server
   "uvicorn[standard]>=0.40.0",
   # * ================ Configuration and Settings ==================*
   # pydantic-settings with toml and yaml support for config files
-  "pydantic-settings[toml,yaml]>=2.13.1",  # Pulls: tomli>=2.0.1, pyyaml>=6.0.1
+  "pydantic-settings[toml,yaml]>=2.12.0",  # Pulls: tomli>=2.0.1, pyyaml>=6.0.1
   # For writing toml config files
   "tomli-w>=1.2.0",
   # * ================ Telemetry and Observability ==================*

scripts/build/generate-mcp-server-json.py

@@ -25,9 +25,9 @@
     ConfigLanguage,
     EnvFormat,
     Provider,
-    ProviderCategory,
     ProviderEnvVarInfo,
     ProviderEnvVars,
+    ProviderCategory,
     SemanticSearchLanguage,
 )
 

scripts/model_data/hf-models.json

@@ -213,7 +213,6 @@
       ]
     }
   },
-  "models": {},
   "models": {
     "Alibaba-NLP/gte-modernbert-base": {
       "adapted_from": null,
@@ -3925,6 +3924,4 @@
       "opensearch-project/opensearch-neural-sparse-encoding-doc-v2-mini"
     ]
   }
-{
-    "models": {}
-  }
+}

scripts/model_data/mteb_to_codeweaver.py

@@ -47,11 +47,10 @@
 
 
 # make sure codeweaver is importable
-sys.path.insert(0, str(Path(__file__).parent.parent.parent / "src"))
+sys.path.insert(0, str(Path(__file__).parent.parent))
 
-from codeweaver.providers.provider import Provider
-
-from codeweaver.providers.embedding.capabilities.types import PartialCapabilities
+from codeweaver.core import Provider
+from codeweaver.providers import PartialCapabilities
 
 
 # TODO: Finish refactor to use these inline constants and eliminate the hf-models.json
@@ -100,13 +99,6 @@
 Note: FastEmbed also has some aliases, but we handle those dynamically below.
 """
 
-KNOWN_ALIASES: dict[str, dict[ModelName, ModelName]] = {"ollama": OLLAMA_ALIASES}
-"""A mapping of provider names to their HF name → provider alias mappings.
-
-Keys are provider name strings (e.g. "ollama") and values are dicts mapping HF model names
-to the provider-specific alias. FastEmbed aliases are handled dynamically via get_fastembed_aliases().
-"""
-
 KNOWN_SPARSE_MODELS = {
     Provider.FASTEMBED: [
         "Qdrant/bm25",
@@ -390,7 +382,13 @@
 
 type DataMap = dict[ModelName, SimplifiedModelMeta]
 
-type ModelMap = dict[ModelMaker, dict[ModelName, tuple[HFModelProviders, ...]]]
+type ModelMap = dict[
+    ModelMaker,
+    dict[
+        ModelName,
+        tuple[Annotated[HFModelProviders, BeforeValidator(lambda v: Provider.from_string(v))], ...],
+    ],
+]
 """A mapping of model makers to their models and the providers that support each model."""
 
 
@@ -501,7 +499,7 @@
     def aliases(self) -> AliasMap:
         """A computed field for the alias map."""
         alias_map: AliasMap = defaultdict(dict)
         for provider, aliases in KNOWN_ALIASES.items():
             for hf_name, alias in aliases.items():
                 alias_map[Provider.from_string(provider)][hf_name] = alias
         return alias_map | {Provider.FASTEMBED: get_fastembed_aliases()}
@@ -522,24 +520,29 @@
         return cls.model_validate_json(cls._json_path.read_text())
 
 
-if JSON_CACHE.exists():
+"""
     _ROOT = RootJson.load()
     DATA = _ROOT.models
     MODEL_MAP_DATA = _ROOT.model_map
     ALIAS_MAP_DATA = _ROOT.aliases
+    SPARSE_MODELS = _ROOT.sparse_models
+
     FLATTENED_ALIASES = _ROOT.flattened_aliases
 else:
-    _ROOT = RootJson(models={})
     DATA = {}
     MODEL_MAP_DATA = {}
     ALIAS_MAP_DATA = {}
+    SPARSE_MODELS = {}
     FLATTENED_ALIASES = {}
+"""
 
 
 def mteb_to_capabilities(model: SimplifiedModelMeta) -> PartialCapabilities:
     """
     Convert an MTEB model metadata dictionary to a PartialCapabilities object.
     """
+    loader = getattr(model, "loader", {})
+    loader = loader if isinstance(loader, dict) else {}
     caps = {
         "name": model["name"],
         "default_dimension": model.get("embed_dim"),
@@ -589,12 +592,12 @@
         caps["output_dimensions"] = (
             (1024, 512, 256, 128, 64, 32) if "-v3" in name else (2048, 1024, 512, 256, 128)
         )
     aliases = [v for val in ALIAS_MAP_DATA.values() for v in val.values()]
     if aliased := next((v for v in aliases if caps["name"] == v), None):
         caps["hf_name"] = next((k for k, v in FLATTENED_ALIASES.items() if v == aliased), None)
         caps["name"] = aliased
     elif caps["name"] in FLATTENED_ALIASES and not caps.get("hf_name"):
         caps["hf_name"] = next((k for k in FLATTENED_ALIASES if k == caps["name"]), None)
     if "Qwen3" in model["name"]:
         caps["other"] = cast(dict[str, Any], caps.get("other", {})) | {
             "model": {
@@ -664,15 +667,15 @@
         if not isinstance(cap_partial["output_dtypes"], tuple):
             cap_partial["output_dtypes"] = (cap_partial["output_dtypes"],)
     capabilities = []
     cap_map = MODEL_MAP_DATA.get(model_maker, {})
     maker_title = to_camel(model_maker)
     cap_type = f"type {maker_title}Provider = Literal[{', '.join(sorted({f'{format_python_value(v)}' if isinstance(val, tuple) else format_python_value(val) for val in cap_map.values() for v in val}))}]"
     printed_map = f"CAP_MAP: dict[Literal[{', '.join(f'"{k!s}"' for k in cap_map)}], tuple[{maker_title}Provider, ...]] = {format_python_dict(cast(dict[str, Any], cap_map), indent_level=0)}\n\n"
     sanitized_names = [
         sanitize_name(
             cast(str, caps["name"])
             if caps["name"] not in FLATTENED_ALIASES
             else FLATTENED_ALIASES[caps["name"]]
         )
         for caps in cap_partials
     ]
@@ -756,8 +759,8 @@
 
 def sanitize_name(name: str) -> str:
     """Convert model name to valid Python identifier."""
     if name.lower() in FLATTENED_ALIASES:
         name = FLATTENED_ALIASES[cast(str, name.lower())]
     return name.replace("-", "_").replace(".", "_").replace("/", "_").replace(":", "_").upper()
 
 

src/codeweaver/core/di/container.py

@@ -562,8 +562,6 @@ async def resolve(
         if self._is_union_type(interface):
             instance = await self._resolve_union_interface(interface, cache_key, _resolution_stack)
             return cast(T, instance)
-        elif interface is type(None):
-            return cast(T, None)
 
         # 1. Check overrides first
         # We check overrides before tags and singletons because overrides

src/codeweaver/core/statistics.py

@@ -84,10 +84,7 @@ async def _check_profile(container: Container) -> bool | None:
     ):
         from codeweaver.providers.config.profiles import ProviderProfile
 
-        return (
-            profile is ProviderProfile.RECOMMENDED_CLOUD
-            or profile is ProviderProfile.RECOMMENDED
-        )
+        return profile in [ProviderProfile.RECOMMENDED_CLOUD, ProviderProfile.RECOMMENDED]
     return None
 
 

src/codeweaver/core/utils/generation.py

@@ -19,21 +19,9 @@
 
 
 if sys.version_info < (3, 14):
-    try:
-        from uuid_extensions import uuid7 as uuid7_gen
-    except ImportError:
-        def uuid7_gen(*args, **kwargs) -> UUID7:
-            from pydantic import UUID7
-            from uuid import uuid4
-            return cast(UUID7, uuid4())
+    from uuid_extensions import uuid7 as uuid7_gen
 else:
-    try:
-        from uuid import uuid7 as uuid7_gen
-    except ImportError:
-        def uuid7_gen(*args, **kwargs) -> UUID7:
-            from pydantic import UUID7
-            from uuid import uuid4
-            return cast(UUID7, uuid4())
+    from uuid import uuid7 as uuid7_gen
 
 
 def uuid7() -> UUID7:
@@ -56,16 +44,10 @@ def uuid7_as_timestamp(
 ) -> int | datetime.datetime | None:
     """Utility to extract the timestamp from a UUID7, optionally as a datetime."""
     if sys.version_info < (3, 14):
-        try:
-            from uuid_extensions import time_ns, uuid_to_datetime
-
-            return uuid_to_datetime(uuid) if as_datetime else time_ns(uuid)
-        except ImportError:
-            return datetime.datetime.now(datetime.UTC) if as_datetime else int(datetime.datetime.now(datetime.UTC).timestamp() * 1e9)
-    try:
-        from uuid import uuid7
-    except ImportError:
-        return datetime.datetime.now(datetime.UTC) if as_datetime else int(datetime.datetime.now(datetime.UTC).timestamp() * 1e9)
+        from uuid_extensions import time_ns, uuid_to_datetime
+
+        return uuid_to_datetime(uuid) if as_datetime else time_ns(uuid)
+    from uuid import uuid7
 
     uuid = uuid7(uuid) if isinstance(uuid, str | int) else uuid
     return (

src/codeweaver/engine/chunker/delimiters/custom.py

@@ -357,14 +357,12 @@ def generate_rst_character_ranges(character: str) -> list[str]:
     nestable=False,
 )
 
-HTML_BLOCK_TAGS = frozenset({"html", "body", "main", "section", "article"})
-
 HTML_TAGS_PATTERNS = [
     DelimiterPattern(
         starts=[f"<{tag}"],
         ends=[f"</{tag}>"],
         kind=DelimiterKind.BLOCK
-        if tag in HTML_BLOCK_TAGS
+        if tag in ["html", "body", "main", "section", "article"]
         else DelimiterKind.PARAGRAPH,
         inclusive=True,
         take_whole_lines=True,

src/codeweaver/providers/config/clients/multi.py

@@ -49,20 +49,14 @@
     GoogleCredentials = Any
 
 if has_package("fastembed") is not None or has_package("fastembed_gpu") is not None:
-    try:
-        from fastembed.common.types import OnnxProvider
-    except ImportError:
-        OnnxProvider = Any  # type: ignore[assignment, misc]
+    from fastembed.common.types import OnnxProvider
 else:
-    OnnxProvider = Any  # type: ignore[assignment, misc]
+    OnnxProvider = object
 
 if has_package("torch") is not None:
-    try:
-        from torch.nn import Module
-    except ImportError:
-        Module = Any  # type: ignore[assignment, misc]
+    from torch.nn import Module
 else:
-    Module = Any  # type: ignore[assignment, misc]
+    Module = object
 if has_package("sentence_transformers") is not None:
     # SentenceTransformerModelCardData contains these forward references:
     # - eval_results_dict: dict[SentenceEvaluator, dict[str, Any]] | None
@@ -240,10 +234,7 @@ class FastEmbedClientOptions(ClientOptions):
     model_name: str
     cache_dir: str | None = None
     threads: int | None = None
-    onnx_providers: Annotated[
-        Sequence[OnnxProvider] | None,
-        Field(alias="providers", serialization_alias="providers"),
-    ] = None
+    providers: Sequence[OnnxProvider] | None = None
     cuda: bool | None = None
     device_ids: list[int] | None = None
     lazy_load: bool = True
@@ -254,10 +245,10 @@ def _resolve_device_settings(self) -> Self:
         from codeweaver.core import effective_cpu_count
 
         cpu_count = effective_cpu_count()
-        updates: dict[str, Any] = {"threads": self.threads or cpu_count}
+        object.__setattr__(self, "threads", self.threads or cpu_count)
         if self.cuda is False:
-            updates["device_ids"] = []
-            return self.model_copy(update=updates)
+            object.__setattr__(self, "device_ids", [])
+            return self
         from codeweaver.providers.optimize import decide_fastembed_runtime
 
         decision = decide_fastembed_runtime(
@@ -272,11 +263,11 @@ def _resolve_device_settings(self) -> Self:
         else:
             cuda = False
             device_ids = []
-        updates["cuda"] = cuda
-        updates["device_ids"] = device_ids
-        if cuda and (not self.onnx_providers or ONNX_CUDA_PROVIDER not in self.onnx_providers):
-            updates["onnx_providers"] = [ONNX_CUDA_PROVIDER, *(self.onnx_providers or [])]
-        return self.model_copy(update=updates)
+        object.__setattr__(self, "cuda", cuda)
+        object.__setattr__(self, "device_ids", device_ids)
+        if cuda and (not self.providers or ONNX_CUDA_PROVIDER not in self.providers):
+            object.__setattr__(self, "providers", [ONNX_CUDA_PROVIDER, *(self.providers or [])])
+        return self
 
     def _telemetry_keys(self) -> dict[FilteredKeyT, AnonymityConversion]:
         return {FilteredKey("cache_dir"): AnonymityConversion.HASH}

src/codeweaver/providers/config/profiles.py

@@ -268,9 +268,7 @@ def _recommended_default(
         ),
         data=(TavilyProviderSettings(provider=Provider.TAVILY),)
         if Provider.TAVILY.has_env_auth and has_package("tavily")
-        else (DuckDuckGoProviderSettings(provider=Provider.DUCKDUCKGO),)
-        if has_package("ddgs")
-        else (),
+        else (DuckDuckGoProviderSettings(provider=Provider.DUCKDUCKGO),),
         vector_store=(
             QdrantVectorStoreProviderSettings(
                 provider=Provider.QDRANT,

src/codeweaver/providers/reranking/providers/base.py

@@ -462,7 +462,7 @@ def _process_results(
         Note: This sync method is only called from async contexts (from the rerank method).
         """
         # voyage and cohere return token count, others do not
-        if self.provider not in (Provider.VOYAGE, Provider.COHERE):
+        if self.provider not in [Provider.VOYAGE, Provider.COHERE]:
             # We're always called from async context (rerank method), so we can safely get the loop
             try:
                 loop = loop or asyncio.get_running_loop()

src/codeweaver/server/agent_api/search/__init__.py

@@ -177,12 +177,7 @@ async def _resolve_indexer_from_container() -> IndexingService | None:
 
         container = get_container()
         return await container.resolve(IndexingService)
-    except Exception as e:
-        logger.warning(
-            "Failed to resolve IndexingService from container: %s",
-            e,
-            exc_info=True,
-        )
+    except Exception:
         return None
 
 
@@ -205,7 +200,7 @@ async def _ensure_index_ready(
             # We enable reconciliation by default to fix any partial indexes
             await indexer.index_project(add_dense=True, add_sparse=True)
         except Exception as e:
-            logger.warning("Auto-indexing failed: %s", e, exc_info=True)
+            logger.warning("Auto-indexing failed: %s", e)
 
 
 async def _build_search_package(package: SearchPackageDep) -> SearchPackage: