Files
george ff4da0cb9e Fix git sync and add repo browser with path selection
- Fix discover_files: rel_path always computed (was stuck at '.' at root),
  include_path_match now uses relative path, 'return' changed to 'continue'
- Fix ingest_git_source: files were cloned but ingested from wrong path
  (docs/repo-id instead of data/repos/repo-id). Now stages filtered files
  into DOCS_PATH/library_id before calling ingest_library.
- Add browse_repo_tree() for interactive repo exploration
- Add POST /api/v1/sources/browse endpoint to backend
- Add /sources/browse proxy route to webui
- Rewrite sources.html: browse repo, expand/collapse tree, check paths to
  include, then save source and sync

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-06 01:28:10 +01:00

407 lines
13 KiB
Python

# Git Source Operations for Repository Cloning and File Discovery
import os
import shutil
from subprocess import run
from pathlib import Path
from typing import List, Optional, Dict, Any
def get_repos_dir() -> Path:
"""Get the base directory for storing cloned repositories."""
# Default to ./data/repos in project root
return Path(__file__).parent.parent.parent / "data" / "repos"
def ensure_repos_dir():
"""Ensure the repos directory exists (idempotent)."""
repos_dir = get_repos_dir()
repos_dir.mkdir(parents=True, exist_ok=True)
return repos_dir
# Initialize repos directory at module load time (safe to run multiple times)
ensure_repos_dir()
class GitCloneError(Exception):
"""Exception for git clone/checkout failures."""
pass
def run_git(command: List[str]) -> None:
result = run(command, capture_output=True, text=True)
if result.returncode != 0:
error = (result.stderr or result.stdout or "unknown git error").strip()
raise GitCloneError(error)
def clone_or_update_repo(
repo_id: str,
repo_url: str,
branch: str,
repos_base: Optional[Path] = None
) -> Dict[str, Any]:
"""
Clone a git repository or update an existing clone.
Args:
repo_id: Unique identifier for this repository (used in paths)
repo_url: Git URL to clone from
branch: Branch name to checkout
repos_base: Base directory for repos (defaults to get_repos_dir())
Returns:
Dict with operation result including repo path and files found
Raises:
GitCloneError: If clone or checkout fails
"""
repos_base = repos_base or get_repos_dir()
repo_path = repos_base / repo_id
try:
if repo_path.exists():
# Update existing clone
print(f" [Git] Updating existing clone at {repo_path}")
# Fetch latest changes
run_git(["git", "-C", str(repo_path), "fetch", "origin"])
# Reset to branch
run_git(["git", "-C", str(repo_path), "reset", "--hard", "origin/" + branch])
else:
# Clone new repository
print(f" [Git] Cloning {repo_url} to {repo_path}")
run_git(
[
"git",
"clone",
"--branch",
branch,
"--single-branch",
repo_url,
str(repo_path),
]
)
print(f" [Git] Checked out branch: {branch}")
return {
"success": True,
"repo_path": str(repo_path),
"url": repo_url,
"branch": branch
}
except GitCloneError:
raise
except Exception as e:
raise GitCloneError(f"Failed to clone/update repo: {e}") from e
def discover_files(
repo_path: Path,
include_paths: Optional[List[str]] = None,
exclude_paths: Optional[List[str]] = None
) -> List[Dict[str, Any]]:
"""
Discover files in a git repository respecting include/exclude paths.
Args:
repo_path: Path to the cloned repository
include_paths: List of paths relative to repo root to include (if None, all dirs considered)
exclude_paths: List of paths relative to repo root to exclude
Returns:
List of dicts with format:
{
"path": "docs/hooks.md", # Relative to repo root
"full_path": "/full/path/to/repo/docs/hooks.md"
}
"""
include_patterns = None if include_paths is None else [
Path(p) for p in include_paths
]
exclude_patterns = set() if exclude_paths is None else {
Path(p) for p in exclude_paths
}
discovered = []
def should_exclude(rel_path: Path) -> bool:
rel_str = str(rel_path).replace("\\", "/")
for exc_pattern in exclude_patterns:
exc_str = str(exc_pattern).replace("\\", "/")
if rel_str == exc_str or rel_str.startswith(exc_str + "/"):
return True
return False
def file_under_include(rel_path: Path) -> bool:
"""True if the file is inside an include path."""
if not include_patterns:
return True
rel_str = str(rel_path).replace("\\", "/")
for inc_pattern in include_patterns:
inc_str = str(inc_pattern).replace("\\", "/")
if rel_str == inc_str or rel_str.startswith(inc_str + "/"):
return True
return False
def dir_may_contain_includes(rel_path: Path) -> bool:
"""True if this directory is inside, equal to, or a parent of any include path."""
if not include_patterns:
return True
rel_str = str(rel_path).replace("\\", "/")
for inc_pattern in include_patterns:
inc_str = str(inc_pattern).replace("\\", "/")
if rel_str == inc_str or rel_str.startswith(inc_str + "/") or inc_str.startswith(rel_str + "/"):
return True
return False
def walk_and_collect(current: Path, rel_prefix: Path):
try:
for entry in sorted(os.scandir(current), key=lambda e: e.name):
entry_path = current / entry.name
rel_path = rel_prefix / entry.name
if should_exclude(rel_path):
continue
if entry.is_file():
if file_under_include(rel_path):
discovered.append({
"path": str(rel_path).replace("\\", "/"),
"full_path": str(entry_path),
"is_binary": is_probably_binary(str(entry_path))
})
elif entry.is_dir():
if dir_may_contain_includes(rel_path):
walk_and_collect(entry_path, rel_path)
except PermissionError:
pass
def is_probably_binary(filepath: str) -> bool:
"""Simple binary detection based on file extension and first bytes."""
ext = Path(filepath).suffix.lower()
text_extensions = {'.md', '.txt', '.py', '.js', '.ts', '.json',
'.yaml', '.yml', '.html', '.css', '.sh', '.sql'}
if ext not in text_extensions:
# Check for null bytes in first 8KB
try:
with open(filepath, 'rb') as f:
chunk = f.read(8192)
return b'\x00' in chunk
except:
return False
return False
walk_and_collect(repo_path, Path())
return discovered
def browse_repo_tree(repo_path: Path, max_depth: int = 4) -> List[Dict[str, Any]]:
"""
Return a nested directory tree for browsing a cloned repository.
Returns a list of nodes:
{"path": "docs", "type": "dir", "children": [...]}
{"path": "README.md", "type": "file"}
Paths are relative to repo_path. Hidden entries (starting with '.') are skipped.
"""
SKIP_DIRS = {".git", "node_modules", "__pycache__", ".venv", "venv", ".tox"}
def build(current: Path, rel_prefix: Path, depth: int) -> List[Dict[str, Any]]:
if depth <= 0:
return []
nodes: List[Dict[str, Any]] = []
try:
entries = sorted(os.scandir(current), key=lambda e: (not e.is_dir(), e.name.lower()))
except PermissionError:
return []
for entry in entries:
if entry.name.startswith(".") or entry.name in SKIP_DIRS:
continue
rel_path = rel_prefix / entry.name
rel_str = str(rel_path).replace("\\", "/")
if entry.is_dir():
children = build(Path(entry.path), rel_path, depth - 1)
nodes.append({"path": rel_str, "type": "dir", "children": children})
else:
nodes.append({"path": rel_str, "type": "file"})
return nodes
return build(repo_path, Path(), max_depth)
async def ingest_git_source(
library_id: str,
name: str,
description: Optional[str] = None,
repo_url: str = None,
branch: str = "main",
include_paths: Optional[List[str]] = None,
exclude_paths: Optional[List[str]] = None,
repos_base: Optional[Path] = None
) -> Dict[str, Any]:
"""
Ingest a git repository as a new library.
Clones the repo (or updates if exists), discovers files in include paths,
and ingests them into the vector store via existing pipeline.
Args:
library_id: Unique identifier for this library
name: Library display name
description: Optional description
repo_url: Git repository URL to clone from
branch: Branch to checkout (default: main)
include_paths: Paths relative to repo root to include (if None, all dirs considered)
exclude_paths: Paths relative to repo root to exclude
Returns:
Dict with operation result
Raises:
GitCloneError: If git operations fail
"""
from .ingest import ingest_library
print(f"\n[Git Ingestion] Processing library: {library_id}")
print(f" Source: {repo_url or '(local)'}")
# Ensure repos directory exists
repos_base = repos_base or get_repos_dir()
repos_base.mkdir(parents=True, exist_ok=True)
repo_id = f"{library_id}-git"
# Clone or update the repo
clone_result = clone_or_update_repo(
repo_id=repo_id,
repo_url=repo_url,
branch=branch,
repos_base=repos_base
)
repo_path = Path(clone_result["repo_path"])
print(f" [Git] Found files in {repo_path}")
# Discover files respecting include/exclude paths
files = discover_files(
repo_path=repo_path,
include_paths=include_paths,
exclude_paths=exclude_paths
)
print(f" [Git] Discovered {len(files)} file(s)")
if not files:
return {
"success": True,
"library_id": library_id,
"message": "No files found matching include/exclude criteria",
"files_discovered": 0
}
# Stage only the filtered files into DOCS_PATH/library_id so that
# ingest_library reads exactly what discover_files selected.
from .config import settings
docs_dir = Path(settings.docs_path) / library_id
if docs_dir.exists():
shutil.rmtree(docs_dir)
docs_dir.mkdir(parents=True, exist_ok=True)
staged = 0
for file_info in files:
if file_info.get("is_binary"):
continue
src = Path(file_info["full_path"])
dst = docs_dir / file_info["path"]
dst.parent.mkdir(parents=True, exist_ok=True)
try:
shutil.copy2(src, dst)
staged += 1
except OSError as exc:
print(f" [Git] Warning: could not copy {src}: {exc}")
print(f" [Git] Staged {staged} file(s) to {docs_dir}")
result = await ingest_library(
library_id=library_id,
name=name,
description=description,
source_path=library_id,
)
return {
"success": result.get("success", False),
"library_id": library_id,
"name": name,
"files_discovered": len(files),
"chunks_created": result.get("chunks_created", 0),
"vectors_added": result.get("vectors_added", 0)
}
async def sync_sources(
sources_config: Dict[str, Any] = None,
repos_base: Optional[Path] = None
) -> List[Dict[str, Any]]:
"""
Sync all git sources defined in config.
Args:
sources_config: List of source configs (same format as docs_sources.yaml)
repos_base: Base directory for repos
Returns:
List of results for each source
"""
if sources_config is None:
# Load from default config file
import yaml
config_path = Path(__file__).parent.parent.parent / "docs_sources.yaml"
if not config_path.exists():
return [{"success": False, "error": f"Config not found: {config_path}"}]
with open(config_path) as f:
data = yaml.safe_load(f)
sources_config = data.get("sources", [])
results = []
for source in sources_config:
try:
result = await ingest_git_source(
library_id=source.get("library_id"),
name=source.get("name"),
description=source.get("description"),
repo_url=source.get("repo_url"),
branch=source.get("branch", "main"),
include_paths=source.get("include_paths"),
exclude_paths=source.get("exclude_paths"),
repos_base=repos_base
)
except GitCloneError as e:
result = {
"success": False,
"library_id": source.get("library_id", "unknown"),
"error": str(e)
}
except Exception as e:
result = {
"success": False,
"library_id": source.get("library_id", "unknown"),
"error": f"Unexpected error: {e}"
}
results.append(result)
return results