Merge pull request #787 from AndrewFasano/fix_786

fix(gzip): empty files are invalid and should not be parsed

tests/test_file_utils.py

@@ -81,6 +81,18 @@ def fake_file() -> File:
     return File.from_bytes(b"0123456789abcdefghijklmnopqrst")
 
 
+class TestFile:
+    def test_file_from_empty_bytes(self):
+        with pytest.raises(InvalidInputFormat):
+            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):
+            File.from_path(file_path)
+
+
 class TestStructParser:
     def test_parse_correct_endianness(self):
         test_content = b"\x01\x02\x03\x04"

unblob/file_utils.py

@@ -40,11 +40,17 @@ 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.")
         m = cls(-1, len(content))
         m.write(content)
         m.seek(0)
@@ -55,7 +61,10 @@ def from_bytes(cls, content: bytes):
     def from_path(cls, path: Path, access=mmap.ACCESS_READ):
         mode = "r+b" if access == mmap.ACCESS_WRITE else "rb"
         with path.open(mode) as base_file:
-            m = cls(base_file.fileno(), 0, access=access)
+            try:
+                m = cls(base_file.fileno(), 0, access=access)
+            except ValueError as exc:
+                raise InvalidInputFormat from exc
             m.access = access
             return m
 
@@ -115,10 +124,6 @@ def tell(self):
         return self._file.tell() - self._offset
 
 
-class InvalidInputFormat(Exception):
-    pass
-
-
 class Endian(enum.Enum):
     LITTLE = "<"
     BIG = ">"

unblob/handlers/archive/sevenzip.py

@@ -107,7 +107,11 @@ 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()]
+            [
+                p
+                for p in file.parent.glob(f"{file.stem}.*")
+                if p.resolve().exists() and p.stat().st_size > 0
+            ]
         )
         if not paths:
             return None

unblob/handlers/compression/gzip.py

@@ -167,7 +167,11 @@ 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()]
+            [
+                p
+                for p in file.parent.glob(f"{file.stem}.*")
+                if p.resolve().exists() and p.stat().st_size > 0
+            ]
         )
 
         # we 'discard' paths that are not the first in the ordered list,