👌 Skip the substitutions in normalize when there is nothing to replace

markdown_it/rules_core/normalize.py

@@ -5,15 +5,18 @@
 from .state_core import StateCore
 
 # https://spec.commonmark.org/0.29/#line-ending
-NEWLINES_RE = re.compile(r"\r\n?|\n")
-NULL_RE = re.compile(r"\0")
+NEWLINES_RE = re.compile(r"\r\n?")
 
 
 def normalize(state: StateCore) -> None:
-    # Normalize newlines
-    string = NEWLINES_RE.sub("\n", state.src)
+    string = state.src
+
+    # Normalize newlines. Most input has no "\r", so skip the substitution then.
+    if "\r" in string:
+        string = NEWLINES_RE.sub("\n", string)
 
     # Replace NULL characters
-    string = NULL_RE.sub("\ufffd", string)
+    if "\0" in string:
+        string = string.replace("\0", "\ufffd")
 
     state.src = string

tests/test_api/test_main.py

@@ -633,3 +633,24 @@ def test_src_char_code_follows_src():
     state.src = "xyz"
     with pytest.warns(DeprecationWarning):
         assert list(state.srcCharCode) == [120, 121, 122]
+
+
+@pytest.mark.parametrize(
+    "src,expected",
+    [
+        ("a\r\nb", "a\nb"),
+        ("a\rb", "a\nb"),
+        ("a\r\r\nb", "a\n\nb"),
+        ("a\n\rb", "a\n\nb"),
+        ("a\0b\r\n\0", "a�b\n�"),
+        ("plain\ntext", "plain\ntext"),
+    ],
+)
+def test_normalize(src, expected):
+    """Line endings become ``\\n`` and NULL characters become U+FFFD."""
+    from markdown_it.rules_core.normalize import normalize
+    from markdown_it.rules_core.state_core import StateCore
+
+    state = StateCore(src, MarkdownIt(), {})
+    normalize(state)
+    assert state.src == expected