Add init_repo

mcp_codebase.py

@@ -42,7 +42,7 @@ OLLAMA_BASE_URL = "http://127.0.0.1:11434"
 os.environ["OLLAMA_HOST"] = OLLAMA_BASE_URL
 os.environ["OLLAMA_API_BASE"] = OLLAMA_BASE_URL
 
-CODEBASE_PATH = Path(os.environ.get("CODEBASE_PATH", "/home/popertots/Crussell"))  # change as needed
+CODEBASE_PATH = Path(os.environ.get("CODEBASE_PATH", "."))  # change as needed
 VECTOR_DB_PATH = Path(os.environ.get("VECTOR_DB_PATH", "./chroma_db"))
 BM25_INDEX_PATH = Path(os.environ.get("BM25_INDEX_PATH", "./bm25_index.json"))
 RAGIGNORE_PATH = CODEBASE_PATH / ".ragignore"
@@ -2244,6 +2244,92 @@ Example – read_file_lines_tool("src/auth/login.py", start=45, end=60) → full
 Note: This tool treats / as '/home/popertots/Crussell/' so adjust paths accordingly"""
     return EnhancedToon.file_content_results(read_file_lines(path, start, end), path)
 
+@mcp.tool()
+def init_repo(git_url: str) -> str:
+    """What it does – Initialise the RAG stack with a brand‑new Git repository or update an existing one.
+
+    The function will:
+    1. Clone the repo if it does not already exist locally,
+    2. Pull the latest changes if it does,
+    3. Reset the global `CODEBASE_PATH` to the local clone,
+    4. Remove any stale index artefacts,
+    5. Re‑build the vector/BM25 indexes with the same lock‑protected flow used by `rebuild_index`,
+    6. Return a human‑readable status message.
+
+    Next steps:
+        • Call `health_check` to confirm that the server is ready.
+        • Use `search_codebase` or `find_code_references` to explore the new code.
+        • Use `read_file_lines_tool` to inspect any hit in detail.
+    """
+    import shutil
+    from pathlib import Path
+    import subprocess
+
+    # ----- 1. Determine where to clone ---------------------------------
+    repo_name = git_url.split('/')[-1]
+    if repo_name.endswith('.git'):
+        repo_name = repo_name[:-4]
+    clone_dir = Path.cwd() / repo_name
+
+    # ----- 2. Clone / pull -----------------------------------------------
+    try:
+        if clone_dir.exists():
+            # Existing repo → pull
+            subprocess.run(
+                ["git", "-C", str(clone_dir), "pull"],
+                check=True,
+                stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE,
+                text=True,
+            )
+        else:
+            # New repo → clone
+            subprocess.run(
+                ["git", "clone", git_url],
+                cwd=str(Path.cwd()),
+                check=True,
+                stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE,
+                text=True,
+            )
+    except subprocess.CalledProcessError as e:
+        return f"❌ Failed to clone/pull {git_url}: {e.stderr.strip()}"
+
+    # ----- 3. Update the global CODEBASE_PATH --------------------------------
+    global CODEBASE_PATH
+    CODEBASE_PATH = clone_dir
+
+    # ----- 4. Remove stale indices ----------------------------------------
+    for artefact in ("chroma_db", "bm25_index.json", "embeddings"):
+        artefact_path = CODEBASE_PATH / artefact
+        if artefact_path.exists():
+            try:
+                if artefact_path.is_dir():
+                    shutil.rmtree(artefact_path)
+                else:
+                    artefact_path.unlink()
+            except Exception as e:
+                logger.warning(f"Could not delete {artefact_path}: {e}")
+
+    # ----- 5. Re‑build the index (protected by the startup lock) ------------
+    try:
+        rebuild_index()
+    except Exception as e:
+        return f"❌ Index rebuild failed: {e}"
+
+    # ----- 6. Report success -----------------------------------------------
+    branch_name = subprocess.check_output(
+        ["git", "-C", str(CODEBASE_PATH), "rev-parse", "--abbrev-ref", "HEAD"],
+        text=True,
+    ).strip()
+    git_hash = subprocess.check_output(
+        ["git", "-C", str(CODEBASE_PATH), "rev-parse", "HEAD"],
+        text=True,
+    ).strip()
+
+    return f"✅ Initialized repository at {CODEBASE_PATH} on branch {branch_name} with hash {git_hash}"
+
+
 @mcp.tool()
 def rebuild_index() -> str:
     """What it does – Re‑creates all embeddings, BM25, and metadata indexes after major code changes.
@@ -2275,17 +2361,6 @@ Example – rebuild_index() → '✅ Index rebuilt (13.2 s); 11,234 chunks'"""
 # -----------------------------
 # Startup
 # -----------------------------
-def startup():
-    with _startup_lock:
-        if VECTOR_DB_PATH.exists() and BM25_INDEX_PATH.exists():
-            try:
-                load_indexes()
-            except Exception as e:
-                logger.warning(f"Failed loading indexes: {e}; rebuilding")
-                build_indexes()
-        else:
-            logger.info("No existing indexes found, building from scratch...")
-            build_indexes()
 
 def signal_handler(sig, frame):
     """Handle graceful shutdown on Ctrl+C."""
@@ -2316,7 +2391,6 @@ if __name__ == "__main__":
     signal.signal(signal.SIGTERM, signal_handler)
 
     try:
-        startup()
         logger.info("="*60)
         logger.info("🚀 MCP RAG Server is ready!")
         logger.info("Use 'python serve_http.py' for HTTP server")