Files
Aperant/apps/backend/core/git_executable.py
T
Andy 7cb9e0a335 fix(worktree): prevent cross-worktree file leakage via environment variables (#1267)
* fix(worktree): prevent cross-worktree file leakage via environment variables

When pre-commit hook runs in a worktree, it sets GIT_DIR and GIT_WORK_TREE
environment variables. These variables were persisting and leaking into
subsequent git operations in other worktrees or the main repository, causing
files to appear as untracked in the wrong location.

Root cause confirmed by 5 independent investigation agents:
- GIT_DIR/GIT_WORK_TREE exports persist across shell sessions
- Version sync section runs git add without env isolation
- Tests already clear these vars but production code didn't

Fix:
- Clear GIT_DIR/GIT_WORK_TREE when NOT in a worktree context
- Add auto-detection and repair of corrupted core.worktree config
- Add comprehensive documentation explaining the bug and fix

* fix(worktree): add git env isolation to frontend subprocess calls

Extend worktree corruption fix to TypeScript frontend:

- Create git-isolation.ts utility with getIsolatedGitEnv()
- Fix task/worktree-handlers.ts: merge, preview, PR creation spawns
- Fix terminal/worktree-handlers.ts: all 10 execFileSync git calls
- Use getToolPath('git') consistently for cross-platform support

Clears GIT_DIR, GIT_WORK_TREE, GIT_INDEX_FILE, and author/committer
env vars to prevent cross-contamination between worktrees.

Part of fix for mysterious file leakage between worktrees.

* fix(pre-commit): improve robustness of git worktree handling

Address PR review findings:

1. Improve .git file parsing for worktree detection:
   - Use sed -n with /p to only print matching lines
   - Add head -1 to handle malformed files with multiple lines
   - Add directory existence check before setting GIT_DIR

2. Add error handling for git config --unset:
   - Wrap in conditional to detect failures
   - Print warning if unset fails (permissions, locked config, etc.)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* fix(worktree): align git env isolation between TypeScript and Python

Address PR review findings for consistency:

1. Add GIT_AUTHOR_DATE and GIT_COMMITTER_DATE to TypeScript
   GIT_ENV_VARS_TO_CLEAR array to match Python implementation

2. Add HUSKY=0 to Python get_isolated_git_env() to match
   TypeScript implementation and prevent double-hook execution

3. Add getIsolatedGitEnv() to listOtherWorktrees() for consistency
   with all other git operations in the file

Also fix ruff linting (UP045): Replace Optional[dict] with dict | None

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* fix(workspace): migrate remaining git calls to use run_git for env isolation

Replace all direct subprocess.run git calls in workspace.py with run_git()
to ensure consistent environment isolation across all backend git operations.

This prevents potential cross-worktree contamination from environment
variables (GIT_DIR, GIT_WORK_TREE, etc.) that could be set by pre-commit
hooks or other git configurations.

Converted 13 subprocess.run calls:
- git rev-parse --abbrev-ref HEAD
- git merge-base (2 locations)
- git add (10 locations)

Also:
- Remove now-unused subprocess import
- Fix cross-platform issue: use getToolPath('git') in listOtherWorktrees

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* fix(worktree): add unit tests and fix pythonEnv git isolation bypass

- Add comprehensive Python tests for git_executable module (20 tests)
- Add TypeScript tests for getIsolatedGitEnv utility (19 tests)
- Migrate GitLab handlers to use getIsolatedGitEnv for git commands
- Fix critical bug: getPythonEnv() now uses getIsolatedGitEnv() as base
  to prevent git env vars from being re-added when pythonEnv is spread

The pythonEnv bypass bug caused git isolation to be defeated when:
  env: { ...getIsolatedGitEnv(), ...pythonEnv, ... }
because pythonEnv contained a copy of process.env with git vars intact.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* fix(worktree): add git env isolation to execution-handlers

Add getIsolatedGitEnv() to 6 git commands in execution-handlers.ts:
- QA review rejection: reset, checkout, clean (lines 407-430)
- Task discard cleanup: rev-parse, worktree remove, branch -D (lines 573-599)

Without env isolation, these commands could operate on the wrong
repository if GIT_DIR/GIT_WORK_TREE vars were set from a previous
worktree operation.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* fix(pre-commit): clear git env vars when worktree detection fails

When .git is a file but WORKTREE_GIT_DIR parsing fails (empty or invalid
directory), any inherited GIT_DIR/GIT_WORK_TREE environment variables
were left in place, potentially causing cross-worktree contamination.

Now explicitly unsets these variables in the failure case, matching the
behavior of the main repo branch.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* chore(tests): remove unused pytest import

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-18 00:04:23 +01:00

199 lines
6.4 KiB
Python

#!/usr/bin/env python3
"""
Git Executable Finder and Isolation
====================================
Utility to find the git executable, with Windows-specific fallbacks.
Also provides environment isolation to prevent pre-commit hooks and
other git configurations from affecting worktree operations.
Separated into its own module to avoid circular imports.
"""
import os
import shutil
import subprocess
from pathlib import Path
# Git environment variables that can interfere with worktree operations
# when set by pre-commit hooks or other git configurations.
# These must be cleared to prevent cross-worktree contamination.
GIT_ENV_VARS_TO_CLEAR = [
"GIT_DIR",
"GIT_WORK_TREE",
"GIT_INDEX_FILE",
"GIT_OBJECT_DIRECTORY",
"GIT_ALTERNATE_OBJECT_DIRECTORIES",
# Identity variables that could be set by hooks
"GIT_AUTHOR_NAME",
"GIT_AUTHOR_EMAIL",
"GIT_AUTHOR_DATE",
"GIT_COMMITTER_NAME",
"GIT_COMMITTER_EMAIL",
"GIT_COMMITTER_DATE",
]
_cached_git_path: str | None = None
def get_isolated_git_env(base_env: dict | None = None) -> dict:
"""
Create an isolated environment for git operations.
Clears git environment variables that may be set by pre-commit hooks
or other git configurations, preventing cross-worktree contamination
and ensuring git operations target the intended repository.
Args:
base_env: Base environment dict to copy from. If None, uses os.environ.
Returns:
Environment dict safe for git subprocess operations.
"""
env = dict(base_env) if base_env is not None else os.environ.copy()
for key in GIT_ENV_VARS_TO_CLEAR:
env.pop(key, None)
# Disable user's pre-commit hooks during Auto-Claude managed git operations
# to prevent double-hook execution and potential conflicts
env["HUSKY"] = "0"
return env
def get_git_executable() -> str:
"""Find the git executable, with Windows-specific fallbacks.
Returns the path to git executable. On Windows, checks multiple sources:
1. CLAUDE_CODE_GIT_BASH_PATH env var (set by Electron frontend)
2. shutil.which (if git is in PATH)
3. Common installation locations
4. Windows 'where' command
Caches the result after first successful find.
"""
global _cached_git_path
# Return cached result if available
if _cached_git_path is not None:
return _cached_git_path
git_path = _find_git_executable()
_cached_git_path = git_path
return git_path
def _find_git_executable() -> str:
"""Internal function to find git executable."""
# 1. Check CLAUDE_CODE_GIT_BASH_PATH (set by Electron frontend)
# This env var points to bash.exe, we can derive git.exe from it
bash_path = os.environ.get("CLAUDE_CODE_GIT_BASH_PATH")
if bash_path:
try:
bash_path_obj = Path(bash_path)
if bash_path_obj.exists():
git_dir = bash_path_obj.parent.parent
# Try cmd/git.exe first (preferred), then bin/git.exe
for git_subpath in ["cmd/git.exe", "bin/git.exe"]:
git_path = git_dir / git_subpath
if git_path.is_file():
return str(git_path)
except (OSError, ValueError):
pass # Invalid path or permission error - try next method
# 2. Try shutil.which (works if git is in PATH)
git_path = shutil.which("git")
if git_path:
return git_path
# 3. Windows-specific: check common installation locations
if os.name == "nt":
common_paths = [
os.path.expandvars(r"%PROGRAMFILES%\Git\cmd\git.exe"),
os.path.expandvars(r"%PROGRAMFILES%\Git\bin\git.exe"),
os.path.expandvars(r"%PROGRAMFILES(X86)%\Git\cmd\git.exe"),
os.path.expandvars(r"%LOCALAPPDATA%\Programs\Git\cmd\git.exe"),
r"C:\Program Files\Git\cmd\git.exe",
r"C:\Program Files (x86)\Git\cmd\git.exe",
]
for path in common_paths:
try:
if os.path.isfile(path):
return path
except OSError:
continue
# 4. Try 'where' command with shell=True (more reliable on Windows)
try:
result = subprocess.run(
"where git",
capture_output=True,
text=True,
timeout=5,
shell=True,
)
if result.returncode == 0 and result.stdout.strip():
found_path = result.stdout.strip().split("\n")[0].strip()
if found_path and os.path.isfile(found_path):
return found_path
except (subprocess.TimeoutExpired, OSError):
pass # 'where' command failed - fall through to default
# Default fallback - let subprocess handle it (may fail)
return "git"
def run_git(
args: list[str],
cwd: Path | str | None = None,
timeout: int = 60,
input_data: str | None = None,
env: dict | None = None,
isolate_env: bool = True,
) -> subprocess.CompletedProcess:
"""Run a git command with proper executable finding and environment isolation.
Args:
args: Git command arguments (without 'git' prefix)
cwd: Working directory for the command
timeout: Command timeout in seconds (default: 60)
input_data: Optional string data to pass to stdin
env: Custom environment dict. If None and isolate_env=True, uses isolated env.
isolate_env: If True (default), clears git env vars to prevent hook interference.
Returns:
CompletedProcess with command results.
"""
git = get_git_executable()
if env is None and isolate_env:
env = get_isolated_git_env()
try:
return subprocess.run(
[git] + args,
cwd=cwd,
input=input_data,
capture_output=True,
text=True,
encoding="utf-8",
errors="replace",
timeout=timeout,
env=env,
)
except subprocess.TimeoutExpired:
return subprocess.CompletedProcess(
args=[git] + args,
returncode=-1,
stdout="",
stderr=f"Command timed out after {timeout} seconds",
)
except FileNotFoundError:
return subprocess.CompletedProcess(
args=[git] + args,
returncode=-1,
stdout="",
stderr="Git executable not found. Please ensure git is installed and in PATH.",
)