node: Use hash based cache path in FilesystemBasedCache

FilesystemBasedCache previously used an escaped version of the full
cache key as the filename. Since cache keys include full URLs, this
could exceed the 255-byte filename limit. So use SHA256 based cache
path. For backward compatibility, legacy cache entries are  migrated to
the new cache path.

Fixes #443

Author: bbhtt

node/flatpak_node_generator/cache.py

@@ -2,6 +2,7 @@
 import re
 import tempfile
 import types
+import hashlib
 from pathlib import Path
 from typing import IO, Iterator, Optional, Type
 
@@ -149,16 +150,34 @@ class FilesystemBucketRef(Cache.BucketRef):
         def __init__(self, key: str, cache_root: Path) -> None:
             super().__init__(key)
             self._cache_root = cache_root
+            self._cache_path = self._cache_root / self._hash_key(key)
+            self._legacy_cache_path = (
+                self._cache_root / FilesystemBasedCache._escape_key(key)
+            )
+
+        @staticmethod
+        def _hash_key(key: str) -> str:
+            return hashlib.sha256(key.encode('utf-8')).hexdigest()
 
-            self._cache_path = self._cache_root / FilesystemBasedCache._escape_key(key)
+        def _migrate_cache_path(self) -> None:
+            if not self._cache_path.exists() and self._legacy_cache_path.exists():
+                try:
+                    self._legacy_cache_path.rename(self._cache_path)
+                except OSError:
+                    pass
 
         def open_read(self) -> Optional[Cache.BucketReader]:
+            self._migrate_cache_path()
+
             try:
                 fp = self._cache_path.open('rb')
             except FileNotFoundError:
-                return None
-            else:
-                return FilesystemBasedCache.FilesystemBucketReader(fp)
+                try:
+                    fp = self._legacy_cache_path.open('rb')
+                except FileNotFoundError:
+                    return None
+
+            return FilesystemBasedCache.FilesystemBucketReader(fp)
 
         def open_write(self) -> Cache.BucketWriter:
             target = self._cache_path

node/tests/test_cache.py

@@ -0,0 +1,93 @@
+import hashlib
+from pathlib import Path
+
+from flatpak_node_generator.cache import Cache, FilesystemBasedCache
+
+
+def test_cache_uses_hashed_filename(tmp_path: Path) -> None:
+    cache = FilesystemBasedCache(tmp_path)
+    Cache.instance = cache
+
+    key = 'remote-url-metadata:size:https://example.com/very/long/url'
+    bucket = cache.get(key)
+
+    with bucket.open_write() as writer:
+        writer.write(b'123')
+
+    files = list(tmp_path.iterdir())
+    assert len(files) == 1
+
+    filename = files[0].name
+
+    assert len(filename) == 64
+    assert filename == hashlib.sha256(key.encode('utf-8')).hexdigest()
+
+
+def test_cache_migrates_legacy_file(tmp_path: Path) -> None:
+    cache = FilesystemBasedCache(tmp_path)
+    Cache.instance = cache
+
+    key = 'remote-url-metadata:size:https://example.com/legacy'
+    legacy_name = FilesystemBasedCache._escape_key(key)
+    legacy_path = tmp_path / legacy_name
+
+    legacy_path.write_bytes(b'legacy-data')
+
+    bucket = cache.get(key)
+
+    reader = bucket.open_read()
+    assert reader is not None
+    assert reader.read_all() == b'legacy-data'
+    reader.close()
+
+    files = list(tmp_path.iterdir())
+    assert len(files) == 1
+
+    expected_hash = hashlib.sha256(key.encode('utf-8')).hexdigest()
+    assert files[0].name == expected_hash
+
+
+def test_cache_fallback_if_migration_fails(tmp_path: Path, monkeypatch) -> None:
+    cache = FilesystemBasedCache(tmp_path)
+    Cache.instance = cache
+
+    key = 'remote-url-metadata:size:https://example.com/fallback'
+    legacy_name = FilesystemBasedCache._escape_key(key)
+    legacy_path = tmp_path / legacy_name
+
+    legacy_path.write_bytes(b'fallback-data')
+
+    def fail_rename(self, target):
+        raise OSError('rename failed')
+
+    monkeypatch.setattr(Path, 'rename', fail_rename)
+
+    bucket = cache.get(key)
+
+    reader = bucket.open_read()
+    assert reader is not None
+    assert reader.read_all() == b'fallback-data'
+    reader.close()
+
+    assert legacy_path.exists()
+
+    expected_hash = hashlib.sha256(key.encode('utf-8')).hexdigest()
+    assert not (tmp_path / expected_hash).exists()
+
+
+def test_cache_never_creates_escaped_filename(tmp_path: Path) -> None:
+    cache = FilesystemBasedCache(tmp_path)
+    Cache.instance = cache
+
+    key = 'remote-url-metadata:size:https://example.com/test'
+    escaped_name = FilesystemBasedCache._escape_key(key)
+
+    bucket = cache.get(key)
+
+    with bucket.open_write() as writer:
+        writer.write(b'data')
+
+    assert not (tmp_path / escaped_name).exists()
+
+    expected_hash = hashlib.sha256(key.encode('utf-8')).hexdigest()
+    assert (tmp_path / expected_hash).exists()