codegraph-cli 2.0.0__tar.gz → 2.1.0__tar.gz

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/PKG-INFO

@@ -1,8 +1,8 @@
 Metadata-Version: 2.4
 Name: codegraph-cli
-Version: 2.0.0
+Version: 2.1.0
 Summary: AI-powered code intelligence CLI with multi-agent analysis, impact graphs, and conversational coding.
-Author-email: Ali Nasir <ali@codegraph.dev>
+Author-email: Ali Nasir <muhammadalinasir00786@gmail.com>
 License: MIT
 Project-URL: Homepage, https://github.com/al1-nasir/codegraph-cli
 Project-URL: Documentation, https://github.com/al1-nasir/codegraph-cli#readme
@@ -31,7 +31,6 @@ Requires-Dist: typer<1.0.0,>=0.12.0
 Requires-Dist: toml>=0.10.2
 Requires-Dist: lancedb>=0.4.0
 Requires-Dist: pyarrow>=14.0.0
-Requires-Dist: sentence-transformers>=2.2.0
 Requires-Dist: tree-sitter>=0.24.0
 Requires-Dist: tree-sitter-python>=0.23.0
 Requires-Dist: tree-sitter-javascript>=0.23.0
@@ -45,9 +44,13 @@ Requires-Dist: pytest-cov>=4.1.0; extra == "dev"
 Requires-Dist: pytest-mock>=3.11.0; extra == "dev"
 Requires-Dist: build>=1.0.0; extra == "dev"
 Requires-Dist: twine>=5.0.0; extra == "dev"
+Provides-Extra: embeddings
+Requires-Dist: torch>=2.0.0; extra == "embeddings"
+Requires-Dist: transformers<5.0.0,>=4.48.0; extra == "embeddings"
 Provides-Extra: all
 Requires-Dist: crewai>=0.80.0; extra == "all"
-Requires-Dist: google-generativeai>=0.5.0; extra == "all"
+Requires-Dist: torch>=2.0.0; extra == "all"
+Requires-Dist: transformers<5.0.0,>=4.48.0; extra == "all"
 Dynamic: license-file
 
 # CodeGraph CLI

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/__init__.py

@@ -1,4 +1,4 @@
 """CodeGraph CLI package."""
 
 __all__ = ["__version__"]
-__version__ = "2.0.0"
+__version__ = "2.0.1"

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/agents.py

@@ -6,7 +6,7 @@ from collections import deque
 from pathlib import Path
 from typing import Dict, List, Set
 
-from .embeddings import HashEmbeddingModel
+from .embeddings import HashEmbeddingModel, TransformerEmbedder
 from .llm import LocalLLM
 from .models import ImpactReport
 from .parser import PythonGraphParser

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/cli.py

@@ -10,6 +10,7 @@ import typer
 from . import __version__, config
 from .cli_chat import chat_app
 from .cli_setup import setup as setup_wizard, set_llm, unset_llm, show_llm
+from .cli_setup import set_embedding, unset_embedding, show_embedding
 from .cli_v2 import v2_app
 from .graph_export import export_dot, export_html
 from .orchestrator import MCPOrchestrator
@@ -35,6 +36,11 @@ app.command("set-llm")(set_llm)
 app.command("unset-llm")(unset_llm)
 app.command("show-llm")(show_llm)
 
+# Register embedding management commands
+app.command("set-embedding")(set_embedding)
+app.command("unset-embedding")(unset_embedding)
+app.command("show-embedding")(show_embedding)
+
 
 def version_callback(value: bool):
     """Print version and exit."""

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/cli_chat.py

@@ -281,7 +281,7 @@ def start_chat(
     new_session: bool = typer.Option(False, "--new", "-n", help="Force start a new session"),
 ):
     """Start interactive chat session."""
-    from .embeddings import HashEmbeddingModel
+    from .embeddings import get_embedder
     from .project_context import ProjectContext
     
     pm = ProjectManager()
@@ -294,7 +294,7 @@ def start_chat(
     
     # Initialize components
     context = ProjectContext(project, pm)
-    embedding_model = HashEmbeddingModel()
+    embedding_model = get_embedder()
     llm = LocalLLM(model=llm_model, provider=llm_provider, api_key=llm_api_key, endpoint=llm_endpoint)
     rag_retriever = RAGRetriever(context.store, embedding_model)
     

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/cli_setup.py

@@ -8,6 +8,7 @@ from typing import Optional
 import typer
 
 from . import config_manager
+from .embeddings import EMBEDDING_MODELS
 
 app = typer.Typer(help="Setup wizard for LLM provider configuration")
 
@@ -287,6 +288,12 @@ def setup():
         print_error("Failed to save configuration!")
         raise typer.Exit(code=1)
 
+    # Offer embedding setup
+    typer.echo("")
+    setup_emb = typer.confirm("Configure embedding model for semantic search?", default=True)
+    if setup_emb:
+        _interactive_embedding_setup()
+
 
 def set_llm(
     provider: str = typer.Argument(..., help="LLM provider: ollama, groq, openai, anthropic, gemini, openrouter"),
@@ -466,5 +473,156 @@ def show_llm():
     typer.echo("")
 
 
+# ===================================================================
+# Embedding model commands
+# ===================================================================
+
+def _interactive_embedding_setup():
+    """Interactive embedding model picker (called from setup wizard)."""
+    typer.echo("")
+    typer.echo(typer.style("╭──────────────────────────────────────────────╮", fg=typer.colors.CYAN))
+    typer.echo(typer.style("│", fg=typer.colors.CYAN) + typer.style("   Embedding Model Setup                       ", bold=True) + typer.style("│", fg=typer.colors.CYAN))
+    typer.echo(typer.style("╰──────────────────────────────────────────────╯", fg=typer.colors.CYAN))
+    typer.echo("")
+    typer.echo("Choose an embedding model for semantic code search:")
+    typer.echo("Larger models give better results but need more disk/RAM.\n")
+
+    # List models with numbers
+    model_keys = list(EMBEDDING_MODELS.keys())
+    for i, key in enumerate(model_keys, 1):
+        spec = EMBEDDING_MODELS[key]
+        name_col = f"{key}".ljust(12)
+        size_col = f"({spec['size']})".ljust(14)
+        desc = spec["description"]
+        typer.echo(f"  {i}) {name_col} {size_col} {desc}")
+
+    typer.echo("")
+
+    while True:
+        choice = typer.prompt(f"Enter choice [1-{len(model_keys)}]", type=str)
+        try:
+            idx = int(choice)
+            if 1 <= idx <= len(model_keys):
+                selected = model_keys[idx - 1]
+                break
+        except ValueError:
+            # Accept model key directly
+            if choice.strip() in model_keys:
+                selected = choice.strip()
+                break
+        print_error(f"Invalid choice. Enter 1-{len(model_keys)} or a model key.")
+
+    spec = EMBEDDING_MODELS[selected]
+
+    if selected != "hash":
+        typer.echo(f"\n  Model:    {typer.style(spec['name'], fg=typer.colors.CYAN)}")
+        typer.echo(f"  Download: {typer.style(spec['size'], fg=typer.colors.YELLOW)}")
+        typer.echo(f"  Dim:      {spec['dim']}")
+        print_info("Requires: pip install codegraph-cli[embeddings]")
+    else:
+        typer.echo(f"\n  Model: {typer.style('Hash Embedding (zero-dependency)', fg=typer.colors.CYAN)}")
+        print_info("No download needed, but no semantic understanding.")
+
+    success = config_manager.save_embedding_config(selected)
+    if success:
+        print_success(f"Embedding model set to: {selected}")
+        if selected != "hash":
+            print_info(f"Model will be downloaded on first use (~{spec['size']}).")
+            print_info("Re-index your project after changing embeddings: cg index <path>")
+    else:
+        print_error("Failed to save embedding config!")
+
+
+def set_embedding(
+    model: str = typer.Argument(
+        ...,
+        help="Embedding model key: qodo-1.5b, jina-code, bge-base, minilm, hash",
+    ),
+):
+    """Set the embedding model for semantic code search.
+
+    Available models (smallest to largest):
+
+        hash        0 bytes    No download, keyword-level only
+        minilm      ~80 MB     Tiny, fast, decent quality
+        bge-base    ~440 MB    Solid general-purpose
+        jina-code   ~550 MB    Code-aware, good quality
+        qodo-1.5b   ~6.2 GB   Best quality, code-optimized
+
+    Examples:
+        cg set-embedding minilm
+        cg set-embedding jina-code
+        cg set-embedding hash
+    """
+    model = model.lower().strip()
+
+    if model not in EMBEDDING_MODELS:
+        print_error(
+            f"Unknown model '{model}'. "
+            f"Choose from: {', '.join(EMBEDDING_MODELS.keys())}"
+        )
+        raise typer.Exit(code=1)
+
+    spec = EMBEDDING_MODELS[model]
+    success = config_manager.save_embedding_config(model)
+
+    if success:
+        print_success(f"Embedding model set to: {model}")
+        typer.echo(f"  Name: {typer.style(spec['name'], fg=typer.colors.CYAN)}")
+        typer.echo(f"  Dim:  {spec['dim']}")
+        if model != "hash":
+            typer.echo(f"  Size: {spec['size']} (downloaded on first use)")
+            print_info("Re-index your project after changing: cg index <path>")
+    else:
+        print_error("Failed to save configuration!")
+        raise typer.Exit(code=1)
+
+
+def unset_embedding():
+    """Reset embedding model to default (hash — no download)."""
+    success = config_manager.clear_embedding_config()
+    if success:
+        print_success("Embedding model reset to default (hash).")
+        print_info("No neural model will be used. Re-index to apply.")
+    else:
+        print_error("Failed to reset embedding config!")
+        raise typer.Exit(code=1)
+
+
+def show_embedding():
+    """Show current embedding model configuration."""
+    typer.echo("")
+    typer.echo(typer.style("╭──────────────────────────────────────────────╮", fg=typer.colors.CYAN))
+    typer.echo(typer.style("│", fg=typer.colors.CYAN) + typer.style("   Embedding Configuration                     ", bold=True) + typer.style("│", fg=typer.colors.CYAN))
+    typer.echo(typer.style("╰──────────────────────────────────────────────╯", fg=typer.colors.CYAN))
+
+    emb_cfg = config_manager.load_embedding_config()
+    current_key = emb_cfg.get("model", "hash")
+    spec = EMBEDDING_MODELS.get(current_key)
+
+    if spec is None:
+        typer.echo(f"  Model   {typer.style(current_key, fg=typer.colors.RED)} (unknown)")
+    else:
+        typer.echo(f"  Model   {typer.style(f' {current_key} ', bg=typer.colors.CYAN, fg=typer.colors.WHITE, bold=True)}")
+        typer.echo(f"  Name    {typer.style(spec['name'], bold=True)}")
+        typer.echo(f"  Dim     {spec['dim']}")
+        typer.echo(f"  Size    {spec['size']}")
+        typer.echo(f"  Desc    {spec['description']}")
+
+    typer.echo("")
+    typer.echo(typer.style("  Available Models", bold=True))
+    typer.echo(typer.style("  ─────────────────────────────────────────", dim=True))
+    for key, s in EMBEDDING_MODELS.items():
+        marker = typer.style(" *", fg=typer.colors.GREEN) if key == current_key else "  "
+        typer.echo(f"  {marker} {key.ljust(12)} {s['size'].ljust(12)} {s['description']}")
+
+    typer.echo("")
+    typer.echo(typer.style("  Quick Commands", bold=True))
+    typer.echo(typer.style("  ─────────────────────────────────────────", dim=True))
+    typer.echo(f"  {typer.style('cg set-embedding <model>', fg=typer.colors.YELLOW)}   Switch model")
+    typer.echo(f"  {typer.style('cg unset-embedding', fg=typer.colors.YELLOW)}         Reset to hash")
+    typer.echo("")
+
+
 if __name__ == "__main__":
     app()

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/config.py

@@ -13,10 +13,12 @@ SUPPORTED_EXTENSIONS = {".py"}
 
 # Load configuration from TOML file (if available)
 try:
-    from .config_manager import load_config
+    from .config_manager import load_config, load_embedding_config
     _toml_config = load_config()
+    _emb_config = load_embedding_config()
 except ImportError:
     _toml_config = {}
+    _emb_config = {}
 
 # LLM Provider Configuration — loaded from ~/.codegraph/config.toml (set via `cg setup` or `cg set-llm`)
 LLM_PROVIDER = _toml_config.get("provider", "ollama")
@@ -24,6 +26,9 @@ LLM_API_KEY = _toml_config.get("api_key", "")
 LLM_MODEL = _toml_config.get("model", "qwen2.5-coder:7b")
 LLM_ENDPOINT = _toml_config.get("endpoint", "http://127.0.0.1:11434/api/generate")
 
+# Embedding model — set via `cg set-embedding` (default: "hash" = no download)
+EMBEDDING_MODEL = _emb_config.get("model", "hash")
+
 
 def ensure_base_dirs() -> None:
     """Create base directories for local storage if needed."""

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/config_manager.py

@@ -78,11 +78,37 @@ def load_config() -> Dict[str, Any]:
         return DEFAULT_CONFIGS["ollama"].copy()
 
 
+def load_full_config() -> Dict[str, Any]:
+    """Load the entire TOML config (all sections)."""
+    if not CONFIG_FILE.exists() or toml is None:
+        return {}
+    try:
+        with open(CONFIG_FILE, "r") as f:
+            return toml.load(f)
+    except Exception:
+        return {}
+
+
+def _save_full_config(config: Dict[str, Any]) -> bool:
+    """Write entire config dict to TOML file, preserving all sections."""
+    if toml is None:
+        return False
+    BASE_DIR.mkdir(parents=True, exist_ok=True)
+    try:
+        with open(CONFIG_FILE, "w") as f:
+            toml.dump(config, f)
+        return True
+    except Exception:
+        return False
+
+
 def save_config(provider: str, model: str, api_key: str = "", endpoint: str = "") -> bool:
     """Save LLM configuration to TOML file.
     
+    Preserves other sections (e.g. ``[embeddings]``) in the file.
+    
     Args:
-        provider: Provider name (ollama, groq, openai, anthropic)
+        provider: Provider name (ollama, groq, openai, anthropic, gemini, openrouter)
         model: Model name
         api_key: API key for cloud providers
         endpoint: Custom endpoint (for Ollama)
@@ -90,32 +116,56 @@ def save_config(provider: str, model: str, api_key: str = "", endpoint: str = ""
     Returns:
         True if saved successfully, False otherwise
     """
-    if toml is None:
-        return False
-    
-    # Ensure directory exists
-    BASE_DIR.mkdir(parents=True, exist_ok=True)
+    config = load_full_config()
     
-    # Build config
-    config = {
-        "llm": {
-            "provider": provider,
-            "model": model,
-        }
+    config["llm"] = {
+        "provider": provider,
+        "model": model,
     }
-    
     if api_key:
         config["llm"]["api_key"] = api_key
-    
     if endpoint:
         config["llm"]["endpoint"] = endpoint
     
-    try:
-        with open(CONFIG_FILE, "w") as f:
-            toml.dump(config, f)
-        return True
-    except Exception:
-        return False
+    return _save_full_config(config)
+
+
+# ------------------------------------------------------------------
+# Embedding configuration
+# ------------------------------------------------------------------
+
+def load_embedding_config() -> Dict[str, Any]:
+    """Load embedding configuration from ``[embeddings]`` section.
+    
+    Returns:
+        Dict with at least ``model`` key, or empty dict.
+    """
+    full = load_full_config()
+    return full.get("embeddings", {})
+
+
+def save_embedding_config(model_key: str) -> bool:
+    """Save embedding model choice to config TOML.
+    
+    Preserves ``[llm]`` and other sections.
+    
+    Args:
+        model_key: One of the keys from ``EMBEDDING_MODELS``
+                   (e.g. ``"minilm"``, ``"jina-code"``, ``"hash"``).
+    
+    Returns:
+        True if saved successfully.
+    """
+    config = load_full_config()
+    config["embeddings"] = {"model": model_key}
+    return _save_full_config(config)
+
+
+def clear_embedding_config() -> bool:
+    """Remove ``[embeddings]`` section from config, resetting to default."""
+    config = load_full_config()
+    config.pop("embeddings", None)
+    return _save_full_config(config)
 
 
 def get_provider_config(provider: str) -> Dict[str, Any]:

{codegraph_cli-2.0.0 → codegraph_cli-2.1.0}/codegraph_cli/context_manager.py

@@ -56,7 +56,7 @@ class RepoMap:
           class: PythonGraphParser
           function: _resolve_call_edges
         codegraph_cli/embeddings.py
-          class: NeuralEmbedder
+          class: TransformerEmbedder
           class: HashEmbeddingModel
           function: get_embedder
           function: cosine_similarity