fixup! fix(secret): skip non-seekable files during scanning

Author: salome-voltz

ggshield/core/scan/file.py

@@ -3,7 +3,7 @@
 
 from ggshield.utils.files import ListFilesMode, is_path_binary, list_files, url_for_path
 
-from .scannable import NonSeekableFileError, Scannable
+from .scannable import Scannable
 
 
 class File(Scannable):
@@ -30,16 +30,13 @@ def is_longer_than(self, max_utf8_encoded_size: int) -> bool:
             # We already have the encoded size, easy
             return self._utf8_encoded_size > max_utf8_encoded_size
 
-        try:
-            with self.path.open("rb") as fp:
-                (
-                    result,
-                    self._content,
-                    self._utf8_encoded_size,
-                ) = Scannable._is_file_longer_than(fp, max_utf8_encoded_size)
-            return result
-        except NonSeekableFileError:
-            raise
+        with self.path.open("rb") as fp:
+            (
+                result,
+                self._content,
+                self._utf8_encoded_size,
+            ) = Scannable._is_file_longer_than(fp, max_utf8_encoded_size)
+        return result
 
     def _read_content(self) -> None:
         if self._content is None:

ggshield/core/scan/scannable.py

@@ -149,11 +149,12 @@ def _is_file_longer_than(
         Raises DecodeError if the file cannot be decoded.
         """
         # Get the byte size
-        assert fp.seekable()
+        # Note: IOBase.seekable() returns True on some non-seekable files like /proc/self/mounts
         try:
             byte_size = fp.seek(0, SEEK_END)
-        except OSError:
-            raise NonSeekableFileError("File does not support seeking operations")
+            fp.seek(0, SEEK_SET)
+        except OSError as exc:
+            raise NonSeekableFileError() from exc
 
         if byte_size > max_utf8_encoded_size * UTF8_TO_WORSE_OTHER_ENCODING_RATIO:
             # Even if the file used the worst encoding (UTF-32), encoding the content of
@@ -162,10 +163,6 @@ def _is_file_longer_than(
             return True, None, None
 
         # Determine the encoding
-        try:
-            fp.seek(0, SEEK_SET)
-        except OSError:
-            raise NonSeekableFileError("File does not support seeking operations")
         charset_matches = charset_normalizer.from_fp(fp)
         charset_match = charset_matches.best()
         if charset_match is None:

tests/unit/core/scan/test_scannable.py

@@ -52,7 +52,5 @@ def test_file_non_seekable(mock_open, tmp_path):
     test_file.write_text("test content")
     file_obj = File(test_file)
 
-    with pytest.raises(
-        NonSeekableFileError, match="File does not support seeking operations"
-    ):
+    with pytest.raises(NonSeekableFileError):
         file_obj.is_longer_than(1000)