gh-156505: Speed up difflib.HtmlDiff for lopsided replacements (GH-156506)

Author: felix314159

Lib/difflib.py

@@ -31,7 +31,7 @@
            'unified_diff', 'diff_bytes', 'HtmlDiff', 'Match']
 
 from heapq import nlargest as _nlargest
-from collections import namedtuple as _namedtuple
+from collections import deque as _deque, namedtuple as _namedtuple
 from types import GenericAlias
 lazy from _colorize import can_colorize, get_theme
 
@@ -1571,7 +1571,7 @@ def _line_pair_iterator():
         is defined) does not need to be of module scope.
         """
         line_iterator = _line_iterator()
-        fromlines,tolines=[],[]
+        fromlines, tolines = _deque(), _deque()
         while True:
             # Collecting lines of text until we have a from/to pair
             while (len(fromlines)==0 or len(tolines)==0):
@@ -1584,8 +1584,8 @@ def _line_pair_iterator():
                 if to_line is not None:
                     tolines.append((to_line,found_diff))
             # Once we have a pair, remove them from the collection and yield it
-            from_line, fromDiff = fromlines.pop(0)
-            to_line, to_diff = tolines.pop(0)
+            from_line, fromDiff = fromlines.popleft()
+            to_line, to_diff = tolines.popleft()
             yield (from_line,to_line,fromDiff or to_diff)
 
     # Handle case where user does not want context differencing, just yield

Lib/test/test_difflib.py

@@ -200,6 +200,26 @@ def test_mdiff_catch_stop_iteration(self):
             [((1, '\x00-2\x01'), (1, '\x00+3\x01'), True)],
         )
 
+    def test_mdiff_lopsided_replace(self):
+        self.assertEqual(
+            list(difflib._mdiff(["a\n"] * 4, ["b\n"])),
+            [
+                ((1, '\x00-a\n\x01'), (1, '\x00+b\n\x01'), True),
+                ((2, '\x00-a\n\x01'), ('', '\n'), True),
+                ((3, '\x00-a\n\x01'), ('', '\n'), True),
+                ((4, '\x00-a\n\x01'), ('', '\n'), True),
+            ],
+        )
+        self.assertEqual(
+            list(difflib._mdiff(["a\n"], ["b\n"] * 4)),
+            [
+                ((1, '\x00-a\n\x01'), (1, '\x00+b\n\x01'), True),
+                (('', '\n'), (2, '\x00+b\n\x01'), True),
+                (('', '\n'), (3, '\x00+b\n\x01'), True),
+                (('', '\n'), (4, '\x00+b\n\x01'), True),
+            ],
+        )
+
 
 patch914575_from1 = """
    1. Beautiful is beTTer than ugly.