catlog22/Claude-Code-Workflow
1
0
Fork 0
mirror of https://github.com/catlog22/Claude-Code-Workflow.git synced 2026-02-10 02:24:35 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: CodexLens model detection, hybrid search stability, and JSON logging

Browse Source 
- Fix model installation detection using fastembed ONNX cache names
- Add embeddings_config table for model metadata tracking
- Fix hybrid search segfault by using single-threaded GPU mode
- Suppress INFO logs in JSON mode to prevent error display
- Add model dropdown filtering to show only installed models

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
catlog22
2025-12-22 21:49:10 +08:00
parent cf58dc0dd3
commit 8203d690cb
11 changed files with 302 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
codex-lens/src/codexlens/cli/commands.py
View File

@@ -35,8 +35,17 @@ from .output import (
app = typer.Typer(help="CodexLens CLI — local code indexing and search.")
def _configure_logging(verbose: bool) -> None:
level = logging.DEBUG if verbose else logging.INFO
def _configure_logging(verbose: bool, json_mode: bool = False) -> None:
"""Configure logging level.
In JSON mode, suppress INFO logs to keep stderr clean for error parsing.
Only WARNING and above are shown to avoid mixing logs with JSON output.
"""
if json_mode and not verbose:
# In JSON mode, suppress INFO logs to keep stderr clean
level = logging.WARNING
else:
level = logging.DEBUG if verbose else logging.INFO
logging.basicConfig(level=level, format="%(levelname)s %(message)s")
@@ -95,7 +104,7 @@ def init(
If semantic search dependencies are installed, automatically generates embeddings
after indexing completes. Use --no-embeddings to skip this step.
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
config = Config()
languages = _parse_languages(language)
base_path = path.expanduser().resolve()
@@ -314,7 +323,7 @@ def search(
# Force hybrid mode
codexlens search "authentication" --mode hybrid
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
search_path = path.expanduser().resolve()
# Validate mode
@@ -487,7 +496,7 @@ def symbol(
verbose: bool = typer.Option(False, "--verbose", "-v", help="Enable debug logging."),
) -> None:
"""Look up symbols by name and optional kind."""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
search_path = path.expanduser().resolve()
registry: RegistryStore | None = None
@@ -538,7 +547,7 @@ def inspect(
verbose: bool = typer.Option(False, "--verbose", "-v", help="Enable debug logging."),
) -> None:
"""Analyze a single file and display symbols."""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
config = Config()
factory = ParserFactory(config)
@@ -588,7 +597,7 @@ def status(
verbose: bool = typer.Option(False, "--verbose", "-v", help="Enable debug logging."),
) -> None:
"""Show index status and configuration."""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
registry: RegistryStore | None = None
try:
@@ -648,7 +657,7 @@ def status(
# Embedding manager not available
pass
except Exception as e:
logger.debug(f"Failed to get embeddings status: {e}")
logging.debug(f"Failed to get embeddings status: {e}")
stats = {
"index_root": str(index_root),
@@ -737,7 +746,7 @@ def projects(
- show <path>: Show details for a specific project
- remove <path>: Remove a project from the registry
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
registry: RegistryStore | None = None
try:
@@ -892,7 +901,7 @@ def config(
Config keys:
- index_dir: Directory to store indexes (default: ~/.codexlens/indexes)
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
config_file = Path.home() / ".codexlens" / "config.json"
@@ -1057,7 +1066,7 @@ def migrate(
This is a safe operation that preserves all existing data.
Progress is shown during migration.
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
base_path = path.expanduser().resolve()
registry: RegistryStore | None = None
@@ -1183,7 +1192,7 @@ def clean(
With path, removes that project's indexes.
With --all, removes all indexes (use with caution).
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
try:
mapper = PathMapper()
@@ -1329,7 +1338,7 @@ def semantic_list(
Shows files that have LLM-generated summaries and keywords.
Results are aggregated from all index databases in the project.
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
base_path = path.expanduser().resolve()
registry: Optional[RegistryStore] = None
@@ -1798,7 +1807,7 @@ def embeddings_generate(
codexlens embeddings-generate ~/.codexlens/indexes/project/_index.db # Specific index
codexlens embeddings-generate ~/projects/my-app --model fast --force # Regenerate with fast model
"""
_configure_logging(verbose)
_configure_logging(verbose, json_mode)
from codexlens.cli.embedding_manager import generate_embeddings, generate_embeddings_recursive

15
codex-lens/src/codexlens/cli/embedding_manager.py
View File

@@ -279,6 +279,21 @@ def generate_embeddings(
try:
with VectorStore(index_path) as vector_store:
# Check model compatibility with existing embeddings
if not force:
is_compatible, warning = vector_store.check_model_compatibility(
model_profile, embedder.model_name, embedder.embedding_dim
)
if not is_compatible:
return {
"success": False,
"error": warning,
}
# Set/update model configuration for this index
vector_store.set_model_config(
model_profile, embedder.model_name, embedder.embedding_dim
)
# Use bulk insert mode for efficient batch ANN index building
# This defers ANN updates until end_bulk_insert() is called
with vector_store.bulk_insert():

46
codex-lens/src/codexlens/cli/model_manager.py
View File

@@ -16,9 +16,11 @@ except ImportError:
# Model profiles with metadata
# Note: 768d is max recommended dimension for optimal performance/quality balance
# 1024d models are available but not recommended due to higher resource usage
# cache_name: The actual Hugging Face repo name used by fastembed for ONNX caching
MODEL_PROFILES = {
"fast": {
"model_name": "BAAI/bge-small-en-v1.5",
"cache_name": "qdrant/bge-small-en-v1.5-onnx-q", # fastembed uses ONNX version
"dimensions": 384,
"size_mb": 80,
"description": "Fast, lightweight, English-optimized",
@@ -27,6 +29,7 @@ MODEL_PROFILES = {
},
"base": {
"model_name": "BAAI/bge-base-en-v1.5",
"cache_name": "qdrant/bge-base-en-v1.5-onnx-q", # fastembed uses ONNX version
"dimensions": 768,
"size_mb": 220,
"description": "General purpose, good balance of speed and quality",
@@ -35,6 +38,7 @@ MODEL_PROFILES = {
},
"code": {
"model_name": "jinaai/jina-embeddings-v2-base-code",
"cache_name": "jinaai/jina-embeddings-v2-base-code", # Uses original name
"dimensions": 768,
"size_mb": 150,
"description": "Code-optimized, best for programming languages",
@@ -43,6 +47,7 @@ MODEL_PROFILES = {
},
"minilm": {
"model_name": "sentence-transformers/all-MiniLM-L6-v2",
"cache_name": "qdrant/all-MiniLM-L6-v2-onnx", # fastembed uses ONNX version
"dimensions": 384,
"size_mb": 90,
"description": "Popular lightweight model, good quality",
@@ -51,6 +56,7 @@ MODEL_PROFILES = {
},
"multilingual": {
"model_name": "intfloat/multilingual-e5-large",
"cache_name": "qdrant/multilingual-e5-large-onnx", # fastembed uses ONNX version
"dimensions": 1024,
"size_mb": 1000,
"description": "Multilingual + code support (high resource usage)",
@@ -59,6 +65,7 @@ MODEL_PROFILES = {
},
"balanced": {
"model_name": "mixedbread-ai/mxbai-embed-large-v1",
"cache_name": "mixedbread-ai/mxbai-embed-large-v1", # Uses original name
"dimensions": 1024,
"size_mb": 600,
"description": "High accuracy, general purpose (high resource usage)",
@@ -87,6 +94,23 @@ def get_cache_dir() -> Path:
return cache_dir
def _get_model_cache_path(cache_dir: Path, info: Dict) -> Path:
"""Get the actual cache path for a model.
fastembed uses ONNX versions of models with different names than the original.
This function returns the correct path based on the cache_name field.
Args:
cache_dir: The fastembed cache directory
info: Model profile info dictionary
Returns:
Path to the model cache directory
"""
cache_name = info.get("cache_name", info["model_name"])
return cache_dir / f"models--{cache_name.replace('/', '--')}"
def list_models() -> Dict[str, any]:
"""List available model profiles and their installation status.
@@ -106,13 +130,13 @@ def list_models() -> Dict[str, any]:
for profile, info in MODEL_PROFILES.items():
model_name = info["model_name"]
# Check if model is cached
# Check if model is cached using the actual cache name
installed = False
cache_size_mb = 0
if cache_exists:
# Check for model directory in cache
model_cache_path = cache_dir / f"models--{model_name.replace('/', '--')}"
# Check for model directory in cache using correct cache_name
model_cache_path = _get_model_cache_path(cache_dir, info)
if model_cache_path.exists():
installed = True
# Calculate cache size
@@ -166,7 +190,8 @@ def download_model(profile: str, progress_callback: Optional[callable] = None) -
"error": f"Unknown profile: {profile}. Available: {', '.join(MODEL_PROFILES.keys())}",
}
model_name = MODEL_PROFILES[profile]["model_name"]
info = MODEL_PROFILES[profile]
model_name = info["model_name"]
try:
# Download model by instantiating TextEmbedding
@@ -179,9 +204,9 @@ def download_model(profile: str, progress_callback: Optional[callable] = None) -
if progress_callback:
progress_callback(f"Model {model_name} downloaded successfully")
# Get cache info
# Get cache info using correct cache_name
cache_dir = get_cache_dir()
model_cache_path = cache_dir / f"models--{model_name.replace('/', '--')}"
model_cache_path = _get_model_cache_path(cache_dir, info)
cache_size = 0
if model_cache_path.exists():
@@ -224,9 +249,10 @@ def delete_model(profile: str) -> Dict[str, any]:
"error": f"Unknown profile: {profile}. Available: {', '.join(MODEL_PROFILES.keys())}",
}
model_name = MODEL_PROFILES[profile]["model_name"]
info = MODEL_PROFILES[profile]
model_name = info["model_name"]
cache_dir = get_cache_dir()
model_cache_path = cache_dir / f"models--{model_name.replace('/', '--')}"
model_cache_path = _get_model_cache_path(cache_dir, info)
if not model_cache_path.exists():
return {
@@ -281,9 +307,9 @@ def get_model_info(profile: str) -> Dict[str, any]:
info = MODEL_PROFILES[profile]
model_name = info["model_name"]
# Check installation status
# Check installation status using correct cache_name
cache_dir = get_cache_dir()
model_cache_path = cache_dir / f"models--{model_name.replace('/', '--')}"
model_cache_path = _get_model_cache_path(cache_dir, info)
installed = model_cache_path.exists()
cache_size_mb = None