fix(extract): default the AST cache location to CWD, not the analyzed source tree (#1774)#1802
Conversation
… source tree (Graphify-Labs#1774) extract() without cache_root derived the cache root from the inferred common parent of the input files, so analyzing a corpus outside CWD silently created graphify-out/cache/ inside the analyzed source tree (read-only corpora, other people's checkouts). The cache is an output: effective_root now defaults to Path('.'), matching cache_dir()'s own default. The inferred root still anchors node ids and symbol resolution unchanged, and an explicit cache_root still wins (CLI/watch behavior is untouched — they already pass it).
|
Thanks @SimiSips — you had the right diagnosis and the right default (the cache is an output; it belongs at CWD, not inside the analyzed tree). Landed in I extended it because Net: your behavior for the in-CWD case is preserved, and out-of-CWD corpora keep portable relative keys. I added a regression test asserting keys stay relative for a corpus outside CWD (the property the one-liner would have lost). Also mirrors the location/anchor split that #1747 introduced for the stat index. Appreciate the fix. |
…1774) With no explicit cache_root, extract() wrote graphify-out/cache/ under the inferred common parent of the inputs — the analyzed source tree — so scanning a read-only/foreign corpus silently polluted it. The naive fix (point the root at CWD) breaks two other things that shared the same parameter: file_hash keys become absolute/non-portable for an out-of-CWD corpus, and the XAML/C# project-scan boundary would scan CWD instead of the corpus. So split cache LOCATION from key/id ANCHOR: load_cached/save_cached gain a cache_root arg for where the dir lives, while `root` (inferred common parent) still anchors file_hash keys, source_file relativization, node ids, and the XAML boundary. extract() now locates the cache at CWD (or cache_root) but anchors on `root`; the parallel worker tuple carries both. Existing callers passing cache_root (CLI, watcher) are unchanged. Adopts @SimiSips's #1802 (the CWD default + the two location tests) and adds the decoupling plus a regression test that keys stay relative for a corpus outside CWD — the property the one-line version would have lost. Co-Authored-By: SimiSips <SimiSips@users.noreply.github.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2 participants
Fixes #1774.
Problem
extract()called withoutcache_rootderives the cache root from the inferred common parent of the input files — i.e. the analyzed source tree. Analyzing a corpus outside CWD (a read-only knowledge base, someone else's checkout) silently createsgraphify-out/cache/inside that corpus, defeating output isolation via CWD. Reproduced on currentv8(0.9.13). The CLI and watcher are unaffected — they already passcache_rootexplicitly (added in 01fb51b for #350) — so this only bites library callers andpython -m graphify.extract.Fix
One line:
effective_root(the cache location) now defaults toPath('.')whencache_rootisNone, matchingcache_dir()'s own default. An explicitcache_rootstill wins.Deliberately not the patch suggested in the issue (defaulting
rootitself to CWD):rootalso anchors node-id relativization (path.relative_to(root)) and_augment_symbol_resolution_edges(), and pointing it at CWD would degrade ids for corpora outside CWD. Only the cache side changes; id/symbol anchoring is untouched.Notes
file_hash()keys include the path relative to the cache root, so existing source-tree caches of library callers go cold once and re-extract (CLI/watch caches are unaffected). No correctness impact.source_filepaths — an already-handled pass-through in_relativize_source_files_in/_absolutize_source_files_in.Tests
tests/test_extract_cache_location.py: default cache lands in CWD and the corpus stays untouched; the CWD cache round-trips on a second run; explicitcache_rootstill wins.uv run pytest tests/ -q— 3162 passed, 3 skipped on macOS (Python 3.12).