Expose the diff line number of each patched file (#122)

tests/samples/debdiff.diff

@@ -0,0 +1,7 @@
+diff -u -N original/added.txt new/added.txt
+--- original/added.txt	1970-01-01 01:00:00.000000000 +0100
++++ new/added.txt	2025-03-26 12:42:27.672906377 +0000
+@@ -0,0 +1 @@
++new file
+Binary files Binary files /t/p1/a.png and /t/p2/a.png differ
+Binary files Binary files /t/p1/b.png and /t/p2/b.png differ

tests/test_parser.py

@@ -284,30 +284,58 @@ def test_parse_diff_with_new_and_modified_binary_files(self):
         self.assertFalse(res[0].is_removed_file)
         self.assertTrue(res[0].is_added_file)
         self.assertFalse(res[0].is_binary_file)
+        self.assertEqual(res[0].diff_line_no, 1)
 
         # second file is added
         self.assertFalse(res[1].is_modified_file)
         self.assertFalse(res[1].is_removed_file)
         self.assertTrue(res[1].is_added_file)
         self.assertTrue(res[1].is_binary_file)
+        self.assertEqual(res[1].diff_line_no, 4)
 
         # third file is modified
         self.assertTrue(res[2].is_modified_file)
         self.assertFalse(res[2].is_removed_file)
         self.assertFalse(res[2].is_added_file)
         self.assertTrue(res[2].is_binary_file)
+        self.assertEqual(res[2].diff_line_no, 8)
 
         # fourth file is removed
         self.assertFalse(res[3].is_modified_file)
         self.assertTrue(res[3].is_removed_file)
         self.assertFalse(res[3].is_added_file)
         self.assertTrue(res[3].is_binary_file)
+        self.assertEqual(res[3].diff_line_no, 11)
 
         # fifth empty file is added
         self.assertFalse(res[4].is_modified_file)
         self.assertFalse(res[4].is_removed_file)
         self.assertTrue(res[4].is_added_file)
         self.assertFalse(res[4].is_binary_file)
+        self.assertEqual(res[4].diff_line_no, 15)
+
+    def test_parse_debdiff_binary_file_line_numbers(self):
+        # issue #122 / PR #123: a binary change without hunks should still
+        # expose the diff line number where its entry appears.
+        utf8_file = os.path.join(self.samples_dir, 'samples/debdiff.diff')
+        with open(utf8_file, 'r') as diff_file:
+            res = PatchSet(diff_file)
+
+        self.assertEqual(len(res), 3)
+
+        # first file has a hunk; entry starts at the +++ line (3)
+        self.assertEqual(res[0].path, 'new/added.txt')
+        self.assertEqual(res[0].diff_line_no, 3)
+        self.assertEqual(res[0][0][0].diff_line_no, 5)
+
+        # the binary entries carry the line number of their "Binary files" line
+        self.assertEqual(res[1].path, '/t/p2/a.png')
+        self.assertTrue(res[1].is_binary_file)
+        self.assertEqual(res[1].diff_line_no, 6)
+
+        self.assertEqual(res[2].path, '/t/p2/b.png')
+        self.assertTrue(res[2].is_binary_file)
+        self.assertEqual(res[2].diff_line_no, 7)
 
     def test_parse_round_trip_with_binary_files_in_diff(self):
         """Parse git diff with binary files though round trip"""

unidiff/patch.py

@@ -209,7 +209,8 @@ def __init__(self, patch_info: Optional[PatchInfo] = None,
                  target_timestamp: Optional[str] = None,
                  is_binary_file: bool = False,
                  source_mode: Optional[str] = None,
-                 target_mode: Optional[str] = None) -> None:
+                 target_mode: Optional[str] = None,
+                 diff_line_no: Optional[int] = None) -> None:
         super(PatchedFile, self).__init__()
         self.patch_info = patch_info
         self.source_file = source
@@ -220,6 +221,9 @@ def __init__(self, patch_info: Optional[PatchInfo] = None,
         # git file modes (e.g. '100644', '100755', '120000'); None if unknown
         self.source_mode = source_mode
         self.target_mode = target_mode
+        # 1-based line number in the diff where this file entry starts; useful
+        # to locate files that have no hunks (e.g. binary changes)
+        self.diff_line_no = diff_line_no
 
     def __repr__(self) -> str:
         return "<PatchedFile: %s>" % self.path
@@ -453,7 +457,7 @@ def _parse(self, diff: Iterable, encoding: Optional[str],
         patch_info = None
 
         diff_lines = enumerate(diff, 1)
-        for unused_diff_line_no, line in diff_lines:
+        for diff_line_no, line in diff_lines:
             if encoding is not None:
                 line = line.decode(encoding)
 
@@ -466,7 +470,8 @@ def _parse(self, diff: Iterable, encoding: Optional[str],
                 source_file = is_diff_git_header.group('source')
                 target_file = is_diff_git_header.group('target')
                 current_file = PatchedFile(
-                    patch_info, source_file, target_file, None, None)
+                    patch_info, source_file, target_file, None, None,
+                    diff_line_no=diff_line_no)
                 self.append(current_file)
                 patch_info.append(line)
                 continue
@@ -542,7 +547,8 @@ def _parse(self, diff: Iterable, encoding: Optional[str],
                     # add current file to PatchSet
                     current_file = PatchedFile(
                         patch_info, source_file, target_file,
-                        source_timestamp, target_timestamp)
+                        source_timestamp, target_timestamp,
+                        diff_line_no=diff_line_no)
                     self.append(current_file)
                     patch_info = None
                 else:
@@ -585,7 +591,8 @@ def _parse(self, diff: Iterable, encoding: Optional[str],
                     current_file.is_binary_file = True
                 else:
                     current_file = PatchedFile(
-                        patch_info, source_file, target_file, is_binary_file=True)
+                        patch_info, source_file, target_file, is_binary_file=True,
+                        diff_line_no=diff_line_no)
                     self.append(current_file)
                 patch_info = None
                 current_file = None