Add catch_dotenv hook and corresponding tests to manage .env files

pre_commit_hooks/catch_dotenv.py

@@ -0,0 +1,182 @@
+#!/usr/bin/env python
+from __future__ import annotations
+
+import argparse
+import os
+import re
+import tempfile
+from collections.abc import Sequence
+from typing import Iterable
+
+# --- Defaults / Constants ---
+DEFAULT_ENV_FILE = ".env"          # Canonical env file name
+DEFAULT_GITIGNORE_FILE = ".gitignore"
+DEFAULT_EXAMPLE_ENV_FILE = ".env.example"
+GITIGNORE_BANNER = "# Added by pre-commit hook to prevent committing secrets"
+
+_KEY_REGEX = re.compile(r"^\s*(?:export\s+)?([A-Za-z_][A-Za-z0-9_]*)\s*=")
+
+
+def _atomic_write(path: str, data: str) -> None:
+    """Write text to path atomically (best-effort)."""
+    # Using same directory for atomic os.replace semantics on POSIX.
+    fd, tmp_path = tempfile.mkstemp(dir=os.path.dirname(path) or ".")
+    try:
+        with os.fdopen(fd, "w", encoding="utf-8", newline="") as tmp_f:
+            tmp_f.write(data)
+        os.replace(tmp_path, path)
+    finally:  # Clean up if replace failed
+        if os.path.exists(tmp_path):  # pragma: no cover (rare failure case)
+            try:
+                os.remove(tmp_path)
+            except OSError:  # pragma: no cover
+                pass
+
+
+def ensure_env_in_gitignore(env_file: str, gitignore_file: str, banner: str) -> bool:
+    """Normalize `.gitignore` so it contains exactly one banner + env line at end.
+
+    Returns True if the file was created or its contents changed, False otherwise.
+    Strategy: read existing lines, strip trailing blanks, remove any prior occurrences of
+    the banner or env_file (even if duplicated), then append a single blank line,
+    banner, and env_file. Produces an idempotent final layout.
+    """
+    try:
+        if os.path.exists(gitignore_file):
+            with open(gitignore_file, "r", encoding="utf-8") as f:
+                lines = f.read().splitlines()
+        else:
+            lines = []
+    except OSError as exc:
+        print(f"ERROR: unable to read '{gitignore_file}': {exc}")
+        return False
+
+    original = list(lines)
+
+    # Trim trailing blank lines
+    while lines and not lines[-1].strip():
+        lines.pop()
+
+    # Remove existing occurrences (exact match after strip)
+    filtered: list[str] = [ln for ln in lines if ln.strip() not in {env_file, banner}]
+
+    if filtered and filtered[-1].strip():
+        filtered.append("")  # ensure single blank before banner
+    elif not filtered:  # empty file -> still separate section visually
+        filtered.append("")
+
+    filtered.append(banner)
+    filtered.append(env_file)
+
+    new_content = "\n".join(filtered) + "\n"
+    if original == filtered:
+        return False
+    try:
+        _atomic_write(gitignore_file, new_content)
+        return True
+    except OSError as exc:  # pragma: no cover
+        print(f"ERROR: unable to write '{gitignore_file}': {exc}")
+        return False
+
+
+def create_example_env(src_env: str, example_file: str) -> bool:
+    """Write example file containing only variable keys from real env file.
+
+    Returns True if file written (or updated), False on read/write error.
+    Lines accepted: optional 'export ' prefix then KEY=...; ignores comments & duplicates.
+    """
+    try:
+        with open(src_env, "r", encoding="utf-8") as f_env:
+            lines = f_env.readlines()
+    except OSError as exc:
+        print(f"ERROR: unable to read '{src_env}': {exc}")
+        return False
+
+    seen: set[str] = set()
+    keys: list[str] = []
+    for line in lines:
+        stripped = line.strip()
+        if not stripped or stripped.startswith('#'):
+            continue
+        m = _KEY_REGEX.match(stripped)
+        if not m:
+            continue
+        key = m.group(1)
+        if key not in seen:
+            seen.add(key)
+            keys.append(key)
+
+    header = [
+        '# Generated by catch-dotenv hook.',
+        '# Variable names only – fill in sample values as needed.',
+        '',
+    ]
+    body = [f"{k}=" for k in keys]
+    try:
+        _atomic_write(example_file, "\n".join(header + body) + "\n")
+        return True
+    except OSError as exc:  # pragma: no cover
+        print(f"ERROR: unable to write '{example_file}': {exc}")
+        return False
+
+
+def _has_env(filenames: Iterable[str], env_file: str) -> bool:
+    """Return True if any staged path refers to a target env file by basename."""
+    return any(os.path.basename(name) == env_file for name in filenames)
+
+
+def _find_repo_root(start: str = '.') -> str:
+    """Ascend from start until a directory containing '.git' is found.
+
+    Falls back to absolute path of start if no parent contains '.git'. This mirrors
+    typical pre-commit execution (already at repo root) but makes behavior stable
+    when hook is invoked from a subdirectory (e.g. for direct ad‑hoc testing).
+    """
+    cur = os.path.abspath(start)
+    prev = None
+    while cur != prev:
+        if os.path.isdir(os.path.join(cur, '.git')):
+            return cur
+        prev, cur = cur, os.path.abspath(os.path.join(cur, os.pardir))
+    return os.path.abspath(start)
+
+
+def _print_failure(env_file: str, gitignore_file: str, example_created: bool, gitignore_modified: bool) -> None:
+    parts: list[str] = [f"Blocked committing '{env_file}'."]
+    if gitignore_modified:
+        parts.append(f"Added to '{gitignore_file}'.")
+    if example_created:
+        parts.append("Example file generated.")
+    parts.append(f"Remove '{env_file}' from the commit and commit again.")
+    print(" ".join(parts))
+
+
+def main(argv: Sequence[str] | None = None) -> int:
+    """Main function for the pre-commit hook."""
+    parser = argparse.ArgumentParser(description="Block committing environment files (.env).")
+    parser.add_argument('filenames', nargs='*', help='Staged filenames (supplied by pre-commit).')
+    parser.add_argument('--create-example', action='store_true', help='Generate example env file (.env.example).')
+    args = parser.parse_args(argv)
+    env_file = DEFAULT_ENV_FILE
+    # Resolve repository root (directory containing .git) so writes happen there
+    repo_root = _find_repo_root('.')
+    gitignore_file = os.path.join(repo_root, DEFAULT_GITIGNORE_FILE)
+    example_file = os.path.join(repo_root, DEFAULT_EXAMPLE_ENV_FILE)
+    env_abspath = os.path.join(repo_root, env_file)
+
+    if not _has_env(args.filenames, env_file):
+        return 0
+
+    gitignore_modified = ensure_env_in_gitignore(env_file, gitignore_file, GITIGNORE_BANNER)
+    example_created = False
+    if args.create_example:
+        # Source env is always looked up relative to repo root
+        if os.path.exists(env_abspath):
+            example_created = create_example_env(env_abspath, example_file)
+
+    _print_failure(env_file, gitignore_file, example_created, gitignore_modified)
+    return 1  # Block commit
+
+
+if __name__ == '__main__':
+    raise SystemExit(main())

tests/catch_dotenv_test.py

@@ -0,0 +1,316 @@
+from __future__ import annotations
+
+import os
+import threading
+import time
+from pathlib import Path
+import shutil
+import re
+import pytest
+
+from pre_commit_hooks.catch_dotenv import main, ensure_env_in_gitignore, GITIGNORE_BANNER, DEFAULT_ENV_FILE, DEFAULT_EXAMPLE_ENV_FILE, DEFAULT_GITIGNORE_FILE
+
+# Tests cover hook behavior: detection gating, .gitignore normalization, example
+# file generation parsing edge cases, idempotency, and preservation of existing
+# content. Each test isolates a single behavioral contract.
+
+
+@pytest.fixture()
+def env_file(tmp_path: Path) -> Path:
+    """Copy shared resource .env into tmp workspace as the canonical .env.
+
+    All tests rely on this baseline content (optionally appending extra lines
+    for edge cases) to ensure consistent parsing behavior.
+    """
+    # __file__ => <repo_root>/tests/catch_dotenv_test.py
+    # parents[0] = <repo_root>/tests, parents[1] = <repo_root>
+    # Source file stored as test.env in repo (cannot commit a real .env in CI)
+    resource_env = Path(__file__).resolve().parents[1] / 'testing' / 'resources' / 'test.env'
+    dest = tmp_path / DEFAULT_ENV_FILE
+    shutil.copyfile(resource_env, dest)
+    return dest
+
+
+def run_hook(tmp_path: Path, staged: list[str], create_example: bool = False) -> int:
+    cwd = os.getcwd()
+    os.chdir(tmp_path)
+    try:
+        args = staged[:]
+        if create_example:
+            args.append('--create-example')
+        return main(args)
+    finally:
+        os.chdir(cwd)
+
+
+def test_no_env_file(tmp_path: Path, env_file: Path):
+    """Hook should no-op (return 0) if .env not staged even if it exists."""
+    (tmp_path / 'foo.txt').write_text('x')
+    assert run_hook(tmp_path, ['foo.txt']) == 0
+
+
+def test_blocks_env_and_updates_gitignore(tmp_path: Path, env_file: Path):
+    """Staging .env triggers block (exit 1) and appends banner + env entry."""
+    ret = run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    assert ret == 1
+    gi = (tmp_path / DEFAULT_GITIGNORE_FILE).read_text().splitlines()
+    assert gi[-2] == GITIGNORE_BANNER
+    assert gi[-1] == DEFAULT_ENV_FILE
+
+
+def test_env_present_but_not_staged(tmp_path: Path, env_file: Path):
+    """Existing .env on disk but not staged should not block commit."""
+    assert run_hook(tmp_path, ['unrelated.txt']) == 0
+
+
+def test_idempotent_gitignore(tmp_path: Path, env_file: Path):
+    """Re-running after initial normalization leaves .gitignore unchanged."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    g.write_text(f"{GITIGNORE_BANNER}\n{DEFAULT_ENV_FILE}\n")
+    first = run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    assert first == 1
+    content1 = g.read_text()
+    second = run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    assert second == 1
+    assert g.read_text() == content1  # unchanged
+
+
+def test_gitignore_with_existing_content_preserved(tmp_path: Path, env_file: Path):
+    """Existing entries stay intact; banner/env appended at end cleanly."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    g.write_text('node_modules/\n# comment line\n')  # no trailing newline section markers
+    run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    lines = g.read_text().splitlines()
+    # original content should still be at top
+    assert lines[0] == 'node_modules/'
+    assert '# comment line' in lines[1]
+    # Last two lines should be banner + env file
+    assert lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+
+
+def test_gitignore_duplicates_are_collapsed(tmp_path: Path, env_file: Path):
+    """Multiple prior duplicate banner/env lines collapse to single pair."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    g.write_text(f"other\n{GITIGNORE_BANNER}\n{DEFAULT_ENV_FILE}\n{GITIGNORE_BANNER}\n{DEFAULT_ENV_FILE}\n\n\n")
+    run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    lines = g.read_text().splitlines()
+    assert lines.count(GITIGNORE_BANNER) == 1
+    assert lines.count(DEFAULT_ENV_FILE) == 1
+    assert lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+
+
+def test_create_example(tmp_path: Path, env_file: Path):
+    """Example file includes discovered keys; values stripped to KEY=."""
+    ret = run_hook(tmp_path, [DEFAULT_ENV_FILE], create_example=True)
+    assert ret == 1
+    example = (tmp_path / DEFAULT_EXAMPLE_ENV_FILE).read_text().splitlines()
+    key_lines = [ln for ln in example if ln and not ln.startswith('#')]
+    # All key lines should be KEY=
+    assert all(re.match(r'^[A-Za-z_][A-Za-z0-9_]*=$', ln) for ln in key_lines)
+    # Spot check a few known keys from resource file
+    for k in ['BACKEND_CONTAINER_PORT=', 'ACCESS_TOKEN_SECRET=', 'SUPABASE_SERVICE_KEY=']:
+        assert k in key_lines
+
+
+def test_create_example_duplicate_key_variant_ignored(tmp_path: Path, env_file: Path):
+    """Appending whitespace duplicate of existing key should not duplicate in example."""
+    with open(env_file, 'a', encoding='utf-8') as f:
+        f.write('BACKEND_CONTAINER_PORT =999 # duplicate variant\n')
+    run_hook(tmp_path, [DEFAULT_ENV_FILE], create_example=True)
+    lines = (tmp_path / DEFAULT_EXAMPLE_ENV_FILE).read_text().splitlines()
+    key_lines = [ln for ln in lines if ln and not ln.startswith('#')]
+    assert key_lines.count('BACKEND_CONTAINER_PORT=') == 1
+
+
+def test_gitignore_without_trailing_newline(tmp_path: Path, env_file: Path):
+    """Normalization works when original .gitignore lacks trailing newline."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    g.write_text('existing_line')  # no newline at EOF
+    run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    lines = g.read_text().splitlines()
+    assert lines[0] == 'existing_line'
+    assert lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+
+
+def test_ensure_env_in_gitignore_normalizes(tmp_path: Path, env_file: Path):
+    """Direct API call collapses duplicates and produces canonical tail layout."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    g.write_text(f"{GITIGNORE_BANNER}\n{DEFAULT_ENV_FILE}\n{GITIGNORE_BANNER}\n{DEFAULT_ENV_FILE}\n\n")
+    modified = ensure_env_in_gitignore(DEFAULT_ENV_FILE, str(g), GITIGNORE_BANNER)
+    assert modified is True
+    lines = g.read_text().splitlines()
+    # final two lines should be banner + env
+    assert lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+    # only one occurrence each
+    assert lines.count(GITIGNORE_BANNER) == 1
+    assert lines.count(DEFAULT_ENV_FILE) == 1
+
+
+def test_source_env_file_not_modified(tmp_path: Path, env_file: Path):
+    """Hook must not alter original .env (comments and formatting stay)."""
+    original = env_file.read_text()
+    run_hook(tmp_path, [DEFAULT_ENV_FILE], create_example=True)
+    assert env_file.read_text() == original
+
+
+def test_large_resource_env_parsing(tmp_path: Path, env_file: Path):
+    """Generate example from resource env; assert broad key coverage & format."""
+    ret = run_hook(tmp_path, [DEFAULT_ENV_FILE], create_example=True)
+    assert ret == 1
+    example_lines = (tmp_path / DEFAULT_EXAMPLE_ENV_FILE).read_text().splitlines()
+    key_lines = [ln for ln in example_lines if ln and not ln.startswith('#')]
+    assert len(key_lines) > 20
+    assert all(re.match(r'^[A-Za-z_][A-Za-z0-9_]*=$', ln) for ln in key_lines)
+    for k in ['BACKEND_CONTAINER_PORT=', 'SUPABASE_SERVICE_KEY=', 'ACCESS_TOKEN_SECRET=']:
+        assert k in key_lines
+
+
+def test_failure_message_content(tmp_path: Path, env_file: Path, capsys):
+    """Hook stdout message should contain key phrases when blocking commit."""
+    ret = run_hook(tmp_path, [DEFAULT_ENV_FILE], create_example=True)
+    assert ret == 1
+    out = capsys.readouterr().out.strip()
+    assert "Blocked committing" in out
+    assert DEFAULT_GITIGNORE_FILE in out
+    assert "Example file generated" in out
+    assert "Remove '.env'" in out
+
+
+def test_create_example_when_env_missing(tmp_path: Path, env_file: Path):
+    """--create-example with no .env staged or present should no-op (exit 0).
+
+    Uses env_file fixture (requirement: all tests use fixture) then removes the
+    copied .env to simulate absence.
+    """
+    env_file.unlink()
+    ret = run_hook(tmp_path, ['unrelated.txt'], create_example=True)
+    assert ret == 0
+    assert not (tmp_path / DEFAULT_EXAMPLE_ENV_FILE).exists()
+
+
+def test_gitignore_is_directory_error(tmp_path: Path, env_file: Path, capsys):
+    """If .gitignore path is a directory, hook should print error and still block."""
+    gitignore_dir = tmp_path / DEFAULT_GITIGNORE_FILE
+    gitignore_dir.mkdir()
+    ret = run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    assert ret == 1  # still blocks commit
+    out = capsys.readouterr().out
+    assert "ERROR:" in out  # read failure logged
+
+
+def test_env_example_overwrites_existing(tmp_path: Path, env_file: Path):
+    """Pre-existing example file with junk should be overwritten with header & keys."""
+    example = tmp_path / DEFAULT_EXAMPLE_ENV_FILE
+    example.write_text('junk=1\nSHOULD_NOT_REMAIN=2\n')
+    run_hook(tmp_path, [DEFAULT_ENV_FILE], create_example=True)
+    content = example.read_text().splitlines()
+    assert content[0].startswith('# Generated by catch-dotenv')
+    assert any(ln.startswith('BACKEND_CONTAINER_PORT=') for ln in content)
+    assert 'junk=1' not in content
+    assert 'SHOULD_NOT_REMAIN=2' not in content
+
+
+def test_large_gitignore_normalization_performance(tmp_path: Path, env_file: Path):
+    """Very large .gitignore remains normalized quickly (functional smoke)."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    # Generate many lines with scattered duplicates of banner/env
+    lines = [f"file_{i}" for i in range(3000)] + [GITIGNORE_BANNER, DEFAULT_ENV_FILE] * 3
+    g.write_text("\n".join(lines) + "\n")
+    start = time.time()
+    run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    elapsed = time.time() - start
+    result_lines = g.read_text().splitlines()
+    assert result_lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+    assert result_lines.count(GITIGNORE_BANNER) == 1
+    assert result_lines.count(DEFAULT_ENV_FILE) == 1
+    # Soft performance expectation: should finish fast (< 0.5s on typical dev machine)
+    assert elapsed < 0.5
+
+
+def test_concurrent_gitignore_writes(tmp_path: Path, env_file: Path):
+    """Concurrent ensure_env_in_gitignore calls result in canonical final state."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    # Seed with messy duplicates
+    g.write_text(f"other\n{GITIGNORE_BANNER}\n{DEFAULT_ENV_FILE}\n\n")
+
+    def worker():
+        ensure_env_in_gitignore(DEFAULT_ENV_FILE, str(g), GITIGNORE_BANNER)
+
+    threads = [threading.Thread(target=worker) for _ in range(5)]
+    for t in threads:
+        t.start()
+    for t in threads:
+        t.join()
+    lines = g.read_text().splitlines()
+    assert lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+    assert lines.count(GITIGNORE_BANNER) == 1
+    assert lines.count(DEFAULT_ENV_FILE) == 1
+
+
+def test_mixed_staged_files(tmp_path: Path, env_file: Path):
+    """Staging .env with other files still blocks and only normalizes gitignore once."""
+    other = tmp_path / 'README.md'
+    other.write_text('hi')
+    ret = run_hook(tmp_path, [DEFAULT_ENV_FILE, 'README.md'])
+    assert ret == 1
+    lines = (tmp_path / DEFAULT_GITIGNORE_FILE).read_text().splitlines()
+    assert lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+
+
+def test_already_ignored_env_with_variations(tmp_path: Path, env_file: Path):
+    """Pre-existing ignore lines with spacing normalize to single canonical pair."""
+    g = tmp_path / DEFAULT_GITIGNORE_FILE
+    g.write_text(f"  {DEFAULT_ENV_FILE}  \n{GITIGNORE_BANNER}\n   {DEFAULT_ENV_FILE}\n")
+    run_hook(tmp_path, [DEFAULT_ENV_FILE])
+    lines = g.read_text().splitlines()
+    assert lines[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+    assert lines.count(DEFAULT_ENV_FILE) == 1
+
+
+def test_subdirectory_invocation(tmp_path: Path, env_file: Path):
+    """Running from a subdirectory still writes gitignore/example at repo root."""
+    sub = tmp_path / 'subdir'
+    sub.mkdir()
+    # simulate repository root marker
+    (tmp_path / '.git').mkdir()
+    # simulate running hook from subdir while staged path relative to repo root
+    cwd = os.getcwd()
+    os.chdir(sub)
+    try:
+        ret = main(['../' + DEFAULT_ENV_FILE])
+        gi = (tmp_path / DEFAULT_GITIGNORE_FILE).read_text().splitlines()
+    finally:
+        os.chdir(cwd)
+    assert ret == 1
+    assert gi[-2:] == [GITIGNORE_BANNER, DEFAULT_ENV_FILE]
+
+
+def test_atomic_write_failure_gitignore(monkeypatch, tmp_path: Path, env_file: Path, capsys):
+    """Simulate os.replace failure during gitignore write to exercise error path."""
+    def boom(*a, **k):
+        raise OSError('replace-fail')
+    monkeypatch.setattr('pre_commit_hooks.catch_dotenv.os.replace', boom)
+    modified = ensure_env_in_gitignore(DEFAULT_ENV_FILE, str(tmp_path / DEFAULT_GITIGNORE_FILE), GITIGNORE_BANNER)
+    assert modified is False
+    out = capsys.readouterr().out
+    assert 'ERROR: unable to write' in out
+
+
+def test_atomic_write_failure_example(monkeypatch, tmp_path: Path, env_file: Path, capsys):
+    """Simulate os.replace failure when writing example env file."""
+    def boom(*a, **k):
+        raise OSError('replace-fail')
+    monkeypatch.setattr('pre_commit_hooks.catch_dotenv.os.replace', boom)
+    ok = False
+    # create_example_env requires source .env to exist; env_file fixture provides it in tmp_path root
+    cwd = os.getcwd()
+    os.chdir(tmp_path)
+    try:
+        ok = main([DEFAULT_ENV_FILE, '--create-example']) == 1
+    finally:
+        os.chdir(cwd)
+    # hook still blocks; but example creation failed -> message should not claim Example file generated
+    assert ok is True
+    out = capsys.readouterr().out
+    assert 'Example file generated' not in out
+    assert 'ERROR: unable to write' in out