Fix is_valid_gzip on empty file

tests/handlers/compression/test_gzip.py

@@ -0,0 +1,9 @@
+from pathlib import Path
+
+from unblob.handlers.compression.gzip import MultiVolumeGzipHandler
+
+
+def test_multivolume_is_valid_gzip_empty_file(tmp_path: Path):
+    empty = tmp_path / "empty"
+    empty.touch()
+    assert not MultiVolumeGzipHandler().is_valid_gzip(empty)

tests/test_file_utils.py

@@ -83,13 +83,13 @@ def fake_file() -> File:
 
 class TestFile:
     def test_file_from_empty_bytes(self):
-        with pytest.raises(InvalidInputFormat):
+        with pytest.raises(ValueError):  # noqa: PT011
             File.from_bytes(b"")
 
     def test_file_from_empty_file(self, tmp_path):
         file_path = tmp_path / "file"
         file_path.touch()
-        with pytest.raises(InvalidInputFormat):
+        with pytest.raises(ValueError, match="cannot mmap an empty file"):
             File.from_path(file_path)
 
 

unblob/file_utils.py

@@ -40,17 +40,13 @@ class SeekError(ValueError):
     """Specific ValueError for File.seek."""
 
 
-class InvalidInputFormat(Exception):
-    pass
-
-
 class File(mmap.mmap):
     access: int
 
     @classmethod
     def from_bytes(cls, content: bytes):
         if not content:
-            raise InvalidInputFormat("Can't create File from empty bytes.")
+            raise ValueError("Can't create File from empty bytes.")
         m = cls(-1, len(content))
         m.write(content)
         m.seek(0)
@@ -59,12 +55,14 @@ def from_bytes(cls, content: bytes):
 
     @classmethod
     def from_path(cls, path: Path, access=mmap.ACCESS_READ):
+        """Create File.
+
+        Needs a valid non-empty file,
+        raises ValueError on empty files.
+        """
         mode = "r+b" if access == mmap.ACCESS_WRITE else "rb"
         with path.open(mode) as base_file:
-            try:
-                m = cls(base_file.fileno(), 0, access=access)
-            except ValueError as exc:
-                raise InvalidInputFormat from exc
+            m = cls(base_file.fileno(), 0, access=access)
             m.access = access
             return m
 
@@ -124,6 +122,10 @@ def tell(self):
         return self._file.tell() - self._offset
 
 
+class InvalidInputFormat(Exception):
+    pass
+
+
 class Endian(enum.Enum):
     LITTLE = "<"
     BIG = ">"

unblob/handlers/archive/sevenzip.py

@@ -107,11 +107,7 @@ class MultiVolumeSevenZipHandler(DirectoryHandler):
 
     def calculate_multifile(self, file: Path) -> Optional[MultiFile]:
         paths = sorted(
-            [
-                p
-                for p in file.parent.glob(f"{file.stem}.*")
-                if p.resolve().exists() and p.stat().st_size > 0
-            ]
+            [p for p in file.parent.glob(f"{file.stem}.*") if p.resolve().exists()]
         )
         if not paths:
             return None

unblob/handlers/compression/gzip.py

@@ -156,7 +156,12 @@ class MultiVolumeGzipHandler(DirectoryHandler):
     PATTERN = Glob("*.gz.*")
 
     def is_valid_gzip(self, path: Path) -> bool:
-        with File.from_path(path) as f:
+        try:
+            file = File.from_path(path)
+        except ValueError:
+            return False
+
+        with file as f:
             try:
                 fp = SingleMemberGzipReader(f)
                 if not fp.read_header():
@@ -167,11 +172,7 @@ def is_valid_gzip(self, path: Path) -> bool:
 
     def calculate_multifile(self, file: Path) -> Optional[MultiFile]:
         paths = sorted(
-            [
-                p
-                for p in file.parent.glob(f"{file.stem}.*")
-                if p.resolve().exists() and p.stat().st_size > 0
-            ]
+            [p for p in file.parent.glob(f"{file.stem}.*") if p.resolve().exists()]
         )
 
         # we 'discard' paths that are not the first in the ordered list,