fix: invalid compressed file for huge file

pdbstore/io/file.py

@@ -287,3 +287,18 @@ def _explore_dirs(rootdir: str, recursive: bool = False) -> List[Path]:
             files_list.append(Path(file))
 
     return files_list
+
+
+def get_file_size(path: PathLike) -> int:
+    """Get file size
+
+    :param path: The file path
+    :return: The file size
+    """
+    if not path:
+        return 0
+    file_path: Path = util.str_to_path(path)
+    if not file_path or not file_path.exists():
+        return 0
+
+    return file_path.stat().st_size

pdbstore/store/entry.py

@@ -12,6 +12,9 @@
 class TransactionEntry:
     """A SymbolStore transaction entry representation"""
 
+    # File size limit to disable compression
+    MAX_COMPRESSED_FILE_SIZE: int = 2147482624
+
     def __init__(
         self,
         store: "Store",  # type: ignore[name-defined] # noqa: F821
@@ -102,6 +105,15 @@ def commit(self, force: Optional[bool] = False) -> bool:
         if not dest_dir.is_dir():
             dest_dir.mkdir(parents=True)
 
+        if self.compressed:
+            # Sanity check to limit compression for file having size with less than 2GB
+            # 2GB is the limit of cab files as per Microsoft documentation
+            if self.MAX_COMPRESSED_FILE_SIZE < io.file.get_file_size(self.source_file):
+                self.compressed = False
+                PDBStoreOutput().warning(
+                    f"Disable compression for {self.source_file} since file size is more than 2GB"
+                )
+
         if self.compressed:
             PDBStoreOutput().debug(
                 f"Compressing {self.source_file} to {str(dest_dir / (self.file_name[:-1] + '_'))}"

tests/unit/test_file.py

@@ -5,6 +5,7 @@
 
 from pdbstore.exceptions import ReadFileError
 from pdbstore.io import file
+from pdbstore.typing import Generator
 
 NEWLINES_FILE_CONTENT = "first line\nsecond line"
 
@@ -13,7 +14,7 @@
 
 
 @pytest.fixture(name="file_access")
-def fixture_file_access(tmp_path, request) -> Path:
+def fixture_file_access(tmp_path, request) -> Generator[Path, None, None]:
     """Generate temporary history file"""
     dest = tmp_path / f"file-{time.time()}.bin"
     with open(dest, "wb") as hfp:
@@ -84,3 +85,18 @@ def test_text_file_with_split_windows(file_access):
 
     content = file.read_text_file(file_access, True)
     assert content == NEWLINES_FILE_CONTENT.split("\n")
+
+
+@pytest.mark.parametrize("file_access", [[TEXT_FILE_WITH_CRLF]], indirect=True)
+def test_valid_file_size(file_access):
+    """test valid file size behavior"""
+    assert file.get_file_size(file_access) > 0
+
+
+@pytest.mark.parametrize(
+    "file_path",
+    [None, "", "/invalid/path"],
+)
+def test_invalid_file_size(file_path):
+    """test invalid file size behavior"""
+    assert file.get_file_size(file_path) == 0

tests/unit/test_transaction_entry.py

@@ -185,3 +185,19 @@ def test_extract_failure(tmp_path, tmp_store, test_data_native_dir, fake_process
         entry.compressed = False
         with pytest.raises(exceptions.CopyFileError):
             entry.extract(tmp_path)
+
+
+def test_large_compressed_file(tmp_store, test_data_native_dir):
+    """test no compress for very large file"""
+    with mock.patch("pdbstore.io.file.get_file_size") as _get_file_size:
+        _get_file_size.return_value = TransactionEntry.MAX_COMPRESSED_FILE_SIZE + 10
+        entry = TransactionEntry(
+            tmp_store,
+            "dummylib.pdb",
+            "1972BE39B97341928816018A8ECD08D91",
+            test_data_native_dir / "dummylib.pdb",
+            True,
+        )
+        assert entry.commit() is True
+        assert entry.is_compressed() is False
+        assert entry.stored_path.exists()