cocoindex-code 0.1.12__tar.gz → 0.1.14__tar.gz

{cocoindex_code-0.1.12 → cocoindex_code-0.1.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cocoindex-code
-Version: 0.1.12
+Version: 0.1.14
 Summary: MCP server for indexing and querying codebases using CocoIndex
 Project-URL: Homepage, https://github.com/cocoindex-io/cocoindex-code
 Project-URL: Repository, https://github.com/cocoindex-io/cocoindex-code
@@ -17,7 +17,7 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Requires-Python: >=3.11
-Requires-Dist: cocoindex[litellm]==1.0.0a26
+Requires-Dist: cocoindex[litellm]==1.0.0a29
 Requires-Dist: einops>=0.8.2
 Requires-Dist: mcp>=1.0.0
 Requires-Dist: numpy>=1.24.0
@@ -165,6 +165,7 @@ Use the cocoindex-code MCP server for semantic code search when:
 | `COCOINDEX_CODE_EMBEDDING_MODEL` | Embedding model (see below) | `sbert/sentence-transformers/all-MiniLM-L6-v2` |
 | `COCOINDEX_CODE_BATCH_SIZE` | Max batch size for local embedding model | `16` |
 | `COCOINDEX_CODE_EXTRA_EXTENSIONS` | Additional file extensions to index (comma-separated, e.g. `"inc:php,yaml,toml"` — use `ext:lang` to override language detection) | _(none)_ |
+| `COCOINDEX_CODE_EXCLUDED_PATTERNS` | Additional glob patterns to exclude from indexing as a JSON array (e.g. `'["**/migration.sql", "{**/*.md,**/*.txt}"]'`) | _(none)_ |
 
 
 ### Root Path Discovery
@@ -297,9 +298,20 @@ claude mcp add cocoindex-code \
 
 Any model supported by LiteLLM works — see the [full list of embedding providers](https://docs.litellm.ai/docs/embedding/supported_embedding).
 
-### GPU-optimised local model
+### Local SentenceTransformers models
 
-If you have a GPU, [`nomic-ai/CodeRankEmbed`](https://huggingface.co/nomic-ai/CodeRankEmbed) delivers significantly better code retrieval than the default model. It is 137M parameters, requires ~1 GB VRAM, and has an 8192-token context window.
+Use the `sbert/` prefix to load any [SentenceTransformers](https://www.sbert.net/) model locally (no API key required).
+
+**Example — general purpose text model:**
+```bash
+claude mcp add cocoindex-code \
+  -e COCOINDEX_CODE_EMBEDDING_MODEL=sbert/nomic-ai/nomic-embed-text-v1 \
+  -- cocoindex-code
+```
+
+**GPU-optimised code retrieval:**
+
+[`nomic-ai/CodeRankEmbed`](https://huggingface.co/nomic-ai/CodeRankEmbed) delivers significantly better code retrieval than the default model. It is 137M parameters, requires ~1 GB VRAM, and has an 8192-token context window.
 
 ```bash
 claude mcp add cocoindex-code \
@@ -355,6 +367,7 @@ Returns matching code chunks with:
 | javascript | js | `.js` |
 | json | | `.json` |
 | kotlin | | `.kt`, `.kts` |
+| lua | | `.lua` |
 | markdown | md | `.md`, `.mdx` |
 | pascal | pas, dpr, delphi | `.pas`, `.dpr` |
 | php | | `.php` |

{cocoindex_code-0.1.12 → cocoindex_code-0.1.14}/README.md

@@ -130,6 +130,7 @@ Use the cocoindex-code MCP server for semantic code search when:
 | `COCOINDEX_CODE_EMBEDDING_MODEL` | Embedding model (see below) | `sbert/sentence-transformers/all-MiniLM-L6-v2` |
 | `COCOINDEX_CODE_BATCH_SIZE` | Max batch size for local embedding model | `16` |
 | `COCOINDEX_CODE_EXTRA_EXTENSIONS` | Additional file extensions to index (comma-separated, e.g. `"inc:php,yaml,toml"` — use `ext:lang` to override language detection) | _(none)_ |
+| `COCOINDEX_CODE_EXCLUDED_PATTERNS` | Additional glob patterns to exclude from indexing as a JSON array (e.g. `'["**/migration.sql", "{**/*.md,**/*.txt}"]'`) | _(none)_ |
 
 
 ### Root Path Discovery
@@ -262,9 +263,20 @@ claude mcp add cocoindex-code \
 
 Any model supported by LiteLLM works — see the [full list of embedding providers](https://docs.litellm.ai/docs/embedding/supported_embedding).
 
-### GPU-optimised local model
+### Local SentenceTransformers models
 
-If you have a GPU, [`nomic-ai/CodeRankEmbed`](https://huggingface.co/nomic-ai/CodeRankEmbed) delivers significantly better code retrieval than the default model. It is 137M parameters, requires ~1 GB VRAM, and has an 8192-token context window.
+Use the `sbert/` prefix to load any [SentenceTransformers](https://www.sbert.net/) model locally (no API key required).
+
+**Example — general purpose text model:**
+```bash
+claude mcp add cocoindex-code \
+  -e COCOINDEX_CODE_EMBEDDING_MODEL=sbert/nomic-ai/nomic-embed-text-v1 \
+  -- cocoindex-code
+```
+
+**GPU-optimised code retrieval:**
+
+[`nomic-ai/CodeRankEmbed`](https://huggingface.co/nomic-ai/CodeRankEmbed) delivers significantly better code retrieval than the default model. It is 137M parameters, requires ~1 GB VRAM, and has an 8192-token context window.
 
 ```bash
 claude mcp add cocoindex-code \
@@ -320,6 +332,7 @@ Returns matching code chunks with:
 | javascript | js | `.js` |
 | json | | `.json` |
 | kotlin | | `.kt`, `.kts` |
+| lua | | `.lua` |
 | markdown | md | `.md`, `.mdx` |
 | pascal | pas, dpr, delphi | `.pas`, `.dpr` |
 | php | | `.php` |

{cocoindex_code-0.1.12 → cocoindex_code-0.1.14}/pyproject.toml

@@ -23,7 +23,7 @@ classifiers = [
 
 dependencies = [
     "mcp>=1.0.0",
-    "cocoindex[litellm]==1.0.0a26",
+    "cocoindex[litellm]==1.0.0a29",
     "sentence-transformers>=2.2.0",
     "sqlite-vec>=0.1.0",
     "pydantic>=2.0.0",

{cocoindex_code-0.1.12 → cocoindex_code-0.1.14}/src/cocoindex_code/config.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import json
 import os
 from dataclasses import dataclass
 from pathlib import Path
@@ -42,6 +43,31 @@ def _discover_codebase_root() -> Path:
     return root if root is not None else cwd
 
 
+def _parse_json_string_list_env(var_name: str) -> list[str]:
+    """Parse an environment variable as a JSON array of strings."""
+    raw_value = os.environ.get(var_name, "")
+    if not raw_value.strip():
+        return []
+
+    try:
+        parsed = json.loads(raw_value)
+    except json.JSONDecodeError as exc:
+        raise ValueError(f"{var_name} must be a JSON array of strings, got invalid JSON") from exc
+
+    if not isinstance(parsed, list):
+        raise ValueError(f"{var_name} must be a JSON array of strings")
+
+    result: list[str] = []
+    for item in parsed:
+        if not isinstance(item, str):
+            raise ValueError(f"{var_name} must be a JSON array of strings")
+        item = item.strip()
+        if item:
+            result.append(item)
+
+    return result
+
+
 @dataclass
 class Config:
     """Configuration loaded from environment variables."""
@@ -50,8 +76,8 @@ class Config:
     embedding_model: str
     index_dir: Path
     device: str | None
-    trust_remote_code: bool
     extra_extensions: dict[str, str | None]
+    excluded_patterns: list[str]
 
     @classmethod
     def from_env(cls) -> Config:
@@ -76,16 +102,6 @@ class Config:
         # Device: auto-detect CUDA or use env override
         device = os.environ.get("COCOINDEX_CODE_DEVICE")
 
-        # trust_remote_code: opt-in via env var only.
-        # sentence-transformers 5.x+ supports Jina models natively, so
-        # auto-enabling this for jinaai/ models causes failures with
-        # transformers 5.x (removed find_pruneable_heads_and_indices).
-        trust_remote_code = os.environ.get("COCOINDEX_CODE_TRUST_REMOTE_CODE", "").lower() in (
-            "1",
-            "true",
-            "yes",
-        )
-
         # Extra file extensions (format: "inc:php,yaml,toml" — optional lang after colon)
         raw_extra = os.environ.get("COCOINDEX_CODE_EXTRA_EXTENSIONS", "")
         extra_extensions: dict[str, str | None] = {}
@@ -99,13 +115,16 @@ class Config:
             else:
                 extra_extensions[f".{token}"] = None
 
+        # Excluded file glob patterns
+        excluded_patterns = _parse_json_string_list_env("COCOINDEX_CODE_EXCLUDED_PATTERNS")
+
         return cls(
             codebase_root_path=root,
             embedding_model=embedding_model,
             index_dir=index_dir,
             device=device,
-            trust_remote_code=trust_remote_code,
             extra_extensions=extra_extensions,
+            excluded_patterns=excluded_patterns,
         )
 
     @property

{cocoindex_code-0.1.12 → cocoindex_code-0.1.14}/src/cocoindex_code/indexer.py

@@ -42,6 +42,7 @@ DEFAULT_INCLUDED_PATTERNS = [
     "**/*.txt",  # Plain text
     "**/*.rst",  # reStructuredText
     "**/*.php",  # PHP
+    "**/*.lua",  # Lua
 ]
 
 INCLUDED_PATTERNS = DEFAULT_INCLUDED_PATTERNS + [f"**/*{ext}" for ext in config.extra_extensions]
@@ -51,7 +52,7 @@ LANGUAGE_OVERRIDES: dict[str, str] = {
     ext: lang for ext, lang in config.extra_extensions.items() if lang is not None
 }
 
-EXCLUDED_PATTERNS = [
+DEFAULT_EXCLUDED_PATTERNS = [
     "**/.*",  # Hidden directories
     "**/__pycache__",  # Python cache
     "**/node_modules",  # Node.js dependencies
@@ -63,6 +64,8 @@ EXCLUDED_PATTERNS = [
     "**/.cocoindex_code",  # Our own index directory
 ]
 
+EXCLUDED_PATTERNS = DEFAULT_EXCLUDED_PATTERNS + config.excluded_patterns
+
 # Chunking configuration
 CHUNK_SIZE = 2000
 MIN_CHUNK_SIZE = 300

{cocoindex_code-0.1.12 → cocoindex_code-0.1.14}/src/cocoindex_code/query.py

@@ -106,7 +106,7 @@ async def query_codebase(
     db = coco_env.get_context(SQLITE_DB)
 
     # Generate query embedding.
-    query_embedding = await embedder.embed(query, True, query_prompt_name)
+    query_embedding = await embedder.embed(query, query_prompt_name)
 
     embedding_bytes = query_embedding.astype("float32").tobytes()
 

{cocoindex_code-0.1.12 → cocoindex_code-0.1.14}/src/cocoindex_code/shared.py

@@ -31,19 +31,15 @@ if config.embedding_model.startswith(SBERT_PREFIX):
     # Models that define a "query" prompt for asymmetric retrieval.
     _QUERY_PROMPT_MODELS = {"nomic-ai/nomic-embed-code", "nomic-ai/CodeRankEmbed"}
     query_prompt_name: str | None = "query" if _model_name in _QUERY_PROMPT_MODELS else None
-    # Models whose custom remote code is known-compatible with transformers 5.x.
-    _KNOWN_REMOTE_CODE_MODELS = {"nomic-ai/CodeRankEmbed"}
-    _trust = config.trust_remote_code or _model_name in _KNOWN_REMOTE_CODE_MODELS
     embedder = SentenceTransformerEmbedder(
         _model_name,
         device=config.device,
-        trust_remote_code=_trust,
+        trust_remote_code=True,
     )
     logger.info(
-        "Embedding model: %s | device: %s | trust_remote_code: %s",
+        "Embedding model: %s | device: %s",
         config.embedding_model,
         config.device,
-        _trust,
     )
 else:
     from cocoindex.ops.litellm import LiteLLMEmbedder