Invalidate persistent file cache when Finch source changes

src/finchlite/util/__init__.py

@@ -1,5 +1,7 @@
-from .cache import file_cache
+# AI modified: 2025-01-01T00:00:00Z parent=4f5a2e5021678965ce8d830bb9edecac1dd3fea9
+# AI modified: 2025-01-01T00:01:00Z parent=4f5a2e5021678965ce8d830bb9edecac1dd3fea9
+from .cache import clear_cache, file_cache
 from .config import config, get_version
 from .print import qual_str
 
-__all__ = ["config", "file_cache", "get_version", "qual_str"]
+__all__ = ["clear_cache", "config", "file_cache", "get_version", "qual_str"]

src/finchlite/util/cache.py

@@ -1,3 +1,9 @@
+# AI modified: 2024-12-31T23:58:00Z parent=154b5aeaa66d01a2373296ba9af9705a3db73ed9
+# AI modified: 2024-12-31T23:59:00Z parent=06953a764918de34b3a35c1b698198c3b74c5890
+# AI modified: 2025-01-01T00:00:00Z parent=4f5a2e5021678965ce8d830bb9edecac1dd3fea9
+# AI modified: 2025-01-01T00:01:00Z parent=4f5a2e5021678965ce8d830bb9edecac1dd3fea9
+# AI modified: 2026-04-02T22:59:00Z parent=197d5a907823d2a53fcd3b68b674f3f4d4f50b5d
+# AI modified: 2026-04-03T15:30:00Z parent=36276c257318d74488f81fa8107d2f2d0a8b804c
 import atexit
 import shutil
 import tempfile
@@ -6,9 +12,76 @@
 from pathlib import Path
 from uuid import UUID
 
+import finchlite
+
 from .config import config, get_version
 
 finch_uuid = UUID("ef66f312-ff6e-4b8a-bb8c-9a843f3ecdf4")
+cache_timestamp_filename = ".finch_code_mtime_ns"
+_finch_source_root = Path(finchlite.__path__[0])
+
+
+def _latest_finch_code_mtime_ns() -> int:
+    latest_mtime = 0
+    for path in _finch_source_root.rglob("*"):
+        if (
+            "__pycache__" not in path.parts
+            and path.is_file()
+            and path.suffix not in {".pyc", ".pyo"}
+        ):
+            latest_mtime = max(latest_mtime, path.stat().st_mtime_ns)
+    return latest_mtime
+
+
+_session_finch_code_mtime_ns = _latest_finch_code_mtime_ns()
+_cache_checked = False
+
+
+def _clear_cache_root(cache_root: Path, *, keep_timestamp: bool = True) -> None:
+    for path in cache_root.iterdir():
+        if keep_timestamp and path.name == cache_timestamp_filename:
+            continue
+        if path.is_dir():
+            shutil.rmtree(path)
+        else:
+            path.unlink()
+
+
+def clear_cache() -> None:
+    """Clear Finch's persistent cache for the current Finch version.
+
+    This removes all cached files for the active Finch version under
+    ``<data_path>/cache/<version>``. If the cache directory does not exist,
+    this function does nothing.
+    """
+
+    global _cache_checked
+    cache_root = Path(config.get("data_path")) / "cache" / get_version()
+    if cache_root.exists():
+        _clear_cache_root(cache_root, keep_timestamp=False)
+    _cache_checked = False
+
+
+def _ensure_cache_fresh(cache_root: Path) -> None:
+    cache_root.mkdir(parents=True, exist_ok=True)
+    timestamp_file = cache_root / cache_timestamp_filename
+    current_mtime = _session_finch_code_mtime_ns
+    should_clear = False
+
+    if timestamp_file.exists():
+        try:
+            cached_mtime = int(timestamp_file.read_text().strip())
+        except ValueError:
+            should_clear = True
+        else:
+            should_clear = current_mtime > cached_mtime
+    else:
+        should_clear = True
+
+    if should_clear:
+        _clear_cache_root(cache_root)
+
+    timestamp_file.write_text(str(current_mtime))
 
 
 def file_cache(*, ext: str, domain: str) -> Callable:
@@ -28,7 +101,12 @@ def decorator(f: Callable) -> Callable:
         nonlocal ext
         ext = ext.lstrip(".")
         if config.get("cache_enable"):
-            cache_dir = Path(config.get("data_path")) / "cache" / get_version() / domain
+            global _cache_checked
+            cache_root = Path(config.get("data_path")) / "cache" / get_version()
+            if not _cache_checked:
+                _ensure_cache_fresh(cache_root)
+                _cache_checked = True
+            cache_dir = cache_root / domain
         else:
             cache_dir = Path(
                 tempfile.mkdtemp(

tests/test_cache.py

@@ -0,0 +1,78 @@
+# AI modified: 2025-01-01T00:00:00Z parent=4f5a2e5021678965ce8d830bb9edecac1dd3fea9
+# AI modified: 2025-01-01T00:01:00Z parent=4f5a2e5021678965ce8d830bb9edecac1dd3fea9
+# AI modified: 2026-04-02T22:59:00Z parent=197d5a907823d2a53fcd3b68b674f3f4d4f50b5d
+# AI modified: 2026-04-03T15:30:00Z parent=36276c257318d74488f81fa8107d2f2d0a8b804c
+import os
+import re
+import subprocess
+import sys
+from pathlib import Path
+
+import finchlite
+
+
+def _run_codegen_session(
+    repo_root: Path,
+    data_path: Path,
+    python_code: str,
+):
+    env = os.environ.copy()
+    env["FINCHLITE_DATA_PATH"] = str(data_path)
+    return subprocess.run(
+        [sys.executable, "-c", python_code],
+        cwd=repo_root,
+        env=env,
+        capture_output=True,
+        text=True,
+        check=True,
+    )
+
+
+def _restore_file_state(path: Path, *, contents: str, atime_ns: int, mtime_ns: int):
+    path.write_text(contents)
+    os.utime(path, ns=(atime_ns, mtime_ns))
+
+
+def test_c_codegen_cache_invalidation_end_to_end_across_sessions(tmp_path):
+    # finchlite.__path__[0] points to <repo>/src/finchlite, so parent.parent is repo root.
+    repo_root = Path(finchlite.__path__[0]).parent.parent
+    data_path = tmp_path / "finch_data"
+    c_codegen_file = repo_root / "src" / "finchlite" / "codegen" / "c_codegen.py"
+    stat_before = c_codegen_file.stat()
+    original_contents = c_codegen_file.read_text()
+    base_script = """
+import finchlite.codegen.c_codegen as ccg
+
+code = '''
+int unique_value() {
+    return 7;
+}
+'''
+ccg.load_shared_lib.cache_clear()
+lib = ccg.load_shared_lib(code)
+print("RESULT", int(lib.unique_value()))
+"""
+
+    first = _run_codegen_session(repo_root, data_path, base_script)
+    first_result = int(first.stdout.strip().split("RESULT ", 1)[1])
+    assert first_result == 7
+
+    try:
+        modified_contents = re.sub(
+            r"c_file_path\.write_text\(c_code\)",
+            'c_file_path.write_text(c_code.replace("return 7;", "return 9;"))',
+            original_contents,
+            count=1,
+        )
+        c_codegen_file.write_text(modified_contents)
+
+        second = _run_codegen_session(repo_root, data_path, base_script)
+        second_result = int(second.stdout.strip().split("RESULT ", 1)[1])
+        assert second_result == 9
+    finally:
+        _restore_file_state(
+            c_codegen_file,
+            contents=original_contents,
+            atime_ns=stat_before.st_atime_ns,
+            mtime_ns=stat_before.st_mtime_ns,
+        )