Fix cache eviction bug in ParallelStoreReader (#34)

Author: maxrjones

src/obspec_utils/obspec.py

@@ -607,15 +607,19 @@ def _get_chunks(self, chunk_indices: list[int]) -> dict[int, bytes]:
             # Store in cache
             for chunk_idx, data in zip(needed, results):
                 self._cache[chunk_idx] = bytes(data)
-                # Move to end (most recently used)
-                self._cache.move_to_end(chunk_idx)
 
-                # Evict oldest if over capacity
-                while len(self._cache) > self._max_cached_chunks:
-                    self._cache.popitem(last=False)
+        # Mark all requested chunks as recently used
+        for i in chunk_indices:
+            self._cache.move_to_end(i)
 
-        # Return requested chunks from cache
-        return {i: self._cache[i] for i in chunk_indices}
+        # Build return dict before eviction
+        result = {i: self._cache[i] for i in chunk_indices}
+
+        # Evict oldest if over capacity
+        while len(self._cache) > self._max_cached_chunks:
+            self._cache.popitem(last=False)
+
+        return result
 
     def read(self, size: int = -1, /) -> bytes:
         """

tests/test_parallel_reader.py

@@ -41,3 +41,16 @@ def test_parallel_reader_caching():
     # First chunk refetched
     reader.seek(0)
     assert reader.read(4) == b"0123"
+
+
+def test_parallel_reader_read_spanning_more_chunks_than_cache():
+    """Read spanning more chunks than max_cached_chunks should succeed."""
+    memstore = MemoryStore()
+    # 20 bytes = 5 chunks of 4 bytes each
+    memstore.put("test.txt", b"0123456789ABCDEFGHIJ")
+
+    reader = ParallelStoreReader(
+        memstore, "test.txt", chunk_size=4, max_cached_chunks=2
+    )
+
+    assert reader.read(20) == b"0123456789ABCDEFGHIJ"