diffoscope/comparators/debian.py

@@ -217,7 +217,10 @@ class DebControlFile(File):
 class DotChangesFile(DebControlFile):
     DESCRIPTION = "Debian .changes files"
     FILE_EXTENSION_SUFFIX = {".changes"}
-    FILE_TYPE_RE = re.compile(r"^(ASCII text|UTF-8 Unicode text)")
+
+    # .changes files can be identified "data" if they contain non-printable
+    # characters (Re: reproducible-builds/diffoscope#286)
+    FILE_TYPE_RE = re.compile(r"^(ASCII text|UTF-8 Unicode text|data)")
 
     @classmethod
     def recognizes(cls, file):
@@ -226,13 +229,19 @@ class DotChangesFile(DebControlFile):
 
         try:
             file._deb822 = Changes(filename=file.path)
-        except ChangesFileException:
+        except ChangesFileException as exc:
+            logger.warning(
+                f"Rejecting {file.path} as a Debian .changes file: {exc}"
+            )
             return False
 
         try:
+            file.validation_msg = None
             file._deb822.validate("sha256", check_signature=False)
-        except FileNotFoundError:
-            return False
+        except FileNotFoundError as exc:
+            # Continue even though this .changes file may be invalid
+            file.validation_msg = f"{os.path.basename(file.path)} is missing referenced files: {exc}"
+            logger.warning(file.validation_msg)
 
         return True
 
@@ -242,6 +251,10 @@ class DotChangesFile(DebControlFile):
         if differences is None:
             return None
 
+        for x in (self, other):
+            if x.validation_msg:
+                differences.add_comment(x.validation_msg)
+
         other_deb822 = self._get_deb822(other)
 
         files = zip(self._deb822.get("Files"), other_deb822.get("Files"))

tests/comparators/test_debian.py

@@ -113,7 +113,8 @@ def test_dot_changes_invalid(tmpdir):
     shutil.copy(TEST_DOT_CHANGES_FILE1_PATH, dot_changes_path)
     # we don't copy the referenced .deb
     identified = specialize(FilesystemFile(dot_changes_path))
-    assert not isinstance(identified, DotChangesFile)
+    # ... but it is identified regardless
+    assert isinstance(identified, DotChangesFile)
 
 
 def test_dot_changes_no_differences(dot_changes1):