From 9e490251ada38c76517a9cb5e7ffa4f4bfbe962c Mon Sep 17 00:00:00 2001
From: Apoorva Verma <vermaapoorva0510@gmail.com>
Date: Wed, 1 Jul 2026 06:01:53 +0530
Subject: [PATCH 1/2] Fix JS extractor line numbers being off on CRLF files

The // and <!-- comment rules matched with .*, which eats the \r of a
CRLF, so the comment kept a trailing \r counted as a newline (and the \n
again). Restrict both to [^\r\n] so the pair counts once.
---
 babel/messages/jslexer.py         |  4 ++--
 tests/messages/test_js_extract.py | 17 +++++++++++++++++
 tests/messages/test_jslexer.py    | 13 +++++++++++++
 3 files changed, 32 insertions(+), 2 deletions(-)

diff --git a/babel/messages/jslexer.py b/babel/messages/jslexer.py
index 4d9e66fca..c8c697d56 100644
--- a/babel/messages/jslexer.py
+++ b/babel/messages/jslexer.py
@@ -43,8 +43,8 @@ class Token(NamedTuple):
 
 _rules: list[tuple[str | None, re.Pattern[str]]] = [
     (None, re.compile(r'\s+', re.UNICODE)),
-    (None, re.compile(r'<!--.*')),
-    ('linecomment', re.compile(r'//.*')),
+    (None, re.compile(r'<!--[^\r\n]*')),
+    ('linecomment', re.compile(r'//[^\r\n]*')),
     ('multilinecomment', re.compile(r'/\*.*?\*/', re.UNICODE | re.DOTALL)),
     ('dotted_name', dotted_name_re),
     ('name', name_re),
diff --git a/tests/messages/test_js_extract.py b/tests/messages/test_js_extract.py
index af6dfb144..6e53a6c3a 100644
--- a/tests/messages/test_js_extract.py
+++ b/tests/messages/test_js_extract.py
@@ -189,3 +189,20 @@ def test_inside_nested_template_string():
     )
 
     assert messages == [(1, 'Greetings!', [], None), (1, 'This is a lovely evening.', [], None), (1, 'The day is really nice!', [], None)]
+
+
+@pytest.mark.parametrize('comment', ['// comment', '<!-- comment'])
+def test_line_numbers_are_the_same_for_lf_and_crlf(comment):
+    source = f"{comment}\n{comment}\n{comment}\nmsg = _('Bonjour')\n"
+
+    def line_numbers(newline):
+        buf = BytesIO(source.replace('\n', newline).encode('utf-8'))
+        return [lineno for lineno, _, _, _ in extract.extract('javascript', buf)]
+
+    assert line_numbers('\n') == line_numbers('\r\n') == [4]
+
+
+def test_translator_comment_has_no_trailing_carriage_return():
+    buf = BytesIO("// NOTE: hello\r\nmsg = _('Bonjour à tous')\r\n".encode())
+    messages = list(extract.extract_javascript(buf, ('_',), ['NOTE:'], {}))
+    assert messages[0][3] == ['NOTE: hello']
diff --git a/tests/messages/test_jslexer.py b/tests/messages/test_jslexer.py
index 00afc78ae..7d9120343 100644
--- a/tests/messages/test_jslexer.py
+++ b/tests/messages/test_jslexer.py
@@ -123,3 +123,16 @@ def test_jsx():
         ('jsx_tag', '</comp2', 8),
         ('operator', '>', 8),
     ]
+
+
+def test_line_comment_does_not_eat_carriage_return():
+    # A `//` comment must not swallow the trailing `\r` of a CRLF line ending,
+    # otherwise the line count for the following code is off by one per comment.
+    source = "// comment\r\ngettext('foo')\r\n"
+    assert list(jslexer.tokenize(source)) == [
+        ('linecomment', '// comment', 1),
+        ('name', 'gettext', 2),
+        ('operator', '(', 2),
+        ('string', "'foo'", 2),
+        ('operator', ')', 2),
+    ]

From ada27e466aeb648dc3508309568944fc6de3210e Mon Sep 17 00:00:00 2001
From: Aarni Koskela <akx@iki.fi>
Date: Fri, 30 Oct 2026 12:48:08 +0200
Subject: [PATCH 2/2] Simplify tests

---
 tests/messages/test_js_extract.py | 33 +++++++++----------------------
 tests/messages/test_jslexer.py    | 13 ------------
 2 files changed, 9 insertions(+), 37 deletions(-)

diff --git a/tests/messages/test_js_extract.py b/tests/messages/test_js_extract.py
index 6e53a6c3a..cb3e8eca3 100644
--- a/tests/messages/test_js_extract.py
+++ b/tests/messages/test_js_extract.py
@@ -41,14 +41,16 @@ def test_various_calls():
     ]
 
 
-def test_message_with_line_comment():
-    buf = BytesIO("""\
-// NOTE: hello
-msg = _('Bonjour à tous')
-""".encode('utf-8'))
+@pytest.mark.parametrize('newline', ['\n', '\r\n'])
+def test_message_with_line_comment(newline):
+    buf = BytesIO(newline.join([
+        "<!-- ignored",
+        "// NOTE: hello",
+        "msg = _('Bonjour à tous')",
+        "",
+    ]).encode('utf-8'))
     messages = list(extract.extract_javascript(buf, ('_',), ['NOTE:'], {}))
-    assert messages[0][2] == 'Bonjour à tous'
-    assert messages[0][3] == ['NOTE: hello']
+    assert messages == [(3, '_', 'Bonjour à tous', ['NOTE: hello'])]
 
 
 def test_message_with_multiline_comment():
@@ -189,20 +191,3 @@ def test_inside_nested_template_string():
     )
 
     assert messages == [(1, 'Greetings!', [], None), (1, 'This is a lovely evening.', [], None), (1, 'The day is really nice!', [], None)]
-
-
-@pytest.mark.parametrize('comment', ['// comment', '<!-- comment'])
-def test_line_numbers_are_the_same_for_lf_and_crlf(comment):
-    source = f"{comment}\n{comment}\n{comment}\nmsg = _('Bonjour')\n"
-
-    def line_numbers(newline):
-        buf = BytesIO(source.replace('\n', newline).encode('utf-8'))
-        return [lineno for lineno, _, _, _ in extract.extract('javascript', buf)]
-
-    assert line_numbers('\n') == line_numbers('\r\n') == [4]
-
-
-def test_translator_comment_has_no_trailing_carriage_return():
-    buf = BytesIO("// NOTE: hello\r\nmsg = _('Bonjour à tous')\r\n".encode())
-    messages = list(extract.extract_javascript(buf, ('_',), ['NOTE:'], {}))
-    assert messages[0][3] == ['NOTE: hello']
diff --git a/tests/messages/test_jslexer.py b/tests/messages/test_jslexer.py
index 7d9120343..00afc78ae 100644
--- a/tests/messages/test_jslexer.py
+++ b/tests/messages/test_jslexer.py
@@ -123,16 +123,3 @@ def test_jsx():
         ('jsx_tag', '</comp2', 8),
         ('operator', '>', 8),
     ]
-
-
-def test_line_comment_does_not_eat_carriage_return():
-    # A `//` comment must not swallow the trailing `\r` of a CRLF line ending,
-    # otherwise the line count for the following code is off by one per comment.
-    source = "// comment\r\ngettext('foo')\r\n"
-    assert list(jslexer.tokenize(source)) == [
-        ('linecomment', '// comment', 1),
-        ('name', 'gettext', 2),
-        ('operator', '(', 2),
-        ('string', "'foo'", 2),
-        ('operator', ')', 2),
-    ]