collect_files() in extract.py ignores .graphifyignore

[Leo86868]

Bug

collect_files() in extract.py does not filter files through .graphifyignore. This means _rebuild_code() in watch.py (which calls collect_files()) scans all files regardless of ignore patterns.

Meanwhile, detect() in detect.py correctly loads and applies .graphifyignore via _load_graphifyignore() + _is_ignored().

Impact

On a project with node_modules/ in .graphifyignore, collect_files() returns 40,272 files instead of 210. The resulting Louvain clustering takes ~17 minutes and produces a polluted graph.

Steps to Reproduce

1. Create a project with a node_modules/ directory and a .graphifyignore containing node_modules/
2. Run _rebuild_code(Path('.')) from graphify.watch
3. Observe that collect_files() returns files from node_modules/

Expected Behavior

collect_files() should respect .graphifyignore the same way detect() does.

Suggested Fix

Have collect_files() accept an optional root parameter and call _load_graphifyignore(root) + _is_ignored() to filter results, similar to how detect() does it in detect.py lines 306-333.

Workaround

Monkey-patching collect_files before calling _rebuild_code:

from graphify.detect import _load_graphifyignore, _is_ignored
from graphify.extract import collect_files as _original
from graphify import extract as _mod

def _filtered(target, *, follow_symlinks=False):
    files = _original(target, follow_symlinks=follow_symlinks)
    root = target if target.is_dir() else target.parent
    patterns = _load_graphifyignore(root)
    return [f for f in files if not _is_ignored(f, root, patterns)] if patterns else files

_mod.collect_files = _filtered

Environment

• graphify 0.4.6
• Python 3.13
• macOS

[safishamsi]

Fixed in v4 (commit 271ee0a). collect_files() now loads .graphifyignore via _load_graphifyignore and filters through _is_ignored the same way detect() does. An optional root parameter lets callers specify where to look for the ignore file. Thanks for the detailed report and the workaround.