Keep translator comments next to the translation function call (#1196)

* Add reproducer for issue 1195

* Keep translator comments next to the translation function call, even if the text is further away

Fixes #1195

diff --git a/babel/messages/extract.py b/babel/messages/extract.py
index dcf62f24fbda475721a777c8c10256f6f1242319..2c0ccb323a520a6507c930337a7d37a549f14039 100644 (file)
--- a/babel/messages/extract.py
+++ b/babel/messages/extract.py
@@ -508,7 +508,7 @@ def extract_python(
     :rtype: ``iterator``
     """
     funcname = lineno = message_lineno = None
-    call_stack = -1
+    call_stack = []  # line numbers of calls
     buf = []
     messages = []
     translator_comments = []
@@ -526,7 +526,7 @@ def extract_python(
     current_fstring_start = None
 
     for tok, value, (lineno, _), _, _ in tokens:
-        if call_stack == -1 and tok == NAME and value in ('def', 'class'):
+        if not call_stack and tok == NAME and value in ('def', 'class'):
             in_def = True
         elif tok == OP and value == '(':
             if in_def:
@@ -535,12 +535,12 @@ def extract_python(
                 in_def = False
                 continue
             if funcname:
-                call_stack += 1
+                call_stack.append(lineno)
         elif in_def and tok == OP and value == ':':
             # End of a class definition without parens
             in_def = False
             continue
-        elif call_stack == -1 and tok == COMMENT:
+        elif not call_stack and tok == COMMENT:
             # Strip the comment token from the line
             value = value[1:].strip()
             if in_translator_comments and \
@@ -555,7 +555,7 @@ def extract_python(
                     in_translator_comments = True
                     translator_comments.append((lineno, value))
                     break
-        elif funcname and call_stack == 0:
+        elif funcname and len(call_stack) == 1:
             nested = (tok == NAME and value in keywords)
             if (tok == OP and value == ')') or nested:
                 if buf:
@@ -565,17 +565,20 @@ def extract_python(
                     messages.append(None)
 
                 messages = tuple(messages) if len(messages) > 1 else messages[0]
-                # Comments don't apply unless they immediately
-                # precede the message
-                if translator_comments and \
-                        translator_comments[-1][0] < message_lineno - 1:
-                    translator_comments = []
+
+                if translator_comments:
+                    last_comment_lineno = translator_comments[-1][0]
+                    if last_comment_lineno < min(message_lineno, call_stack[-1]) - 1:
+                        # Comments don't apply unless they immediately
+                        # precede the message, or the line where the parenthesis token
+                        # to start this message's translation call is.
+                        translator_comments.clear()
 
                 yield (message_lineno, funcname, messages,
                        [comment[1] for comment in translator_comments])
 
                 funcname = lineno = message_lineno = None
-                call_stack = -1
+                call_stack.clear()
                 messages = []
                 translator_comments = []
                 in_translator_comments = False
@@ -619,9 +622,9 @@ def extract_python(
 
             elif tok != NL and not message_lineno:
                 message_lineno = lineno
-        elif call_stack > 0 and tok == OP and value == ')':
-            call_stack -= 1
-        elif funcname and call_stack == -1:
+        elif len(call_stack) > 1 and tok == OP and value == ')':
+            call_stack.pop()
+        elif funcname and not call_stack:
             funcname = None
         elif tok == NAME and value in keywords:
             funcname = value

diff --git a/tests/messages/test_extract.py b/tests/messages/test_extract.py
index d5ac3b2cae6830b86b3a59aa758f5e10c4098aed..492d233039ae46a41576a03548b384ea88fff343 100644 (file)
--- a/tests/messages/test_extract.py
+++ b/tests/messages/test_extract.py
@@ -561,3 +561,46 @@ t2 = _(f'\xe5\xe4\xf6' f'\xc5\xc4\xd6')
         messages = list(extract.extract('python', buf, extract.DEFAULT_KEYWORDS, [], {}))
         assert len(messages) == 1
         assert messages[0][1] == 'åäöÅÄÖ'
+
+
+def test_issue_1195():
+    buf = BytesIO(b"""
+foof = {
+    'test_string': StringWithMeta(
+        # NOTE: Text describing a test string
+        string=_(
+            'Text string that is on a new line'
+        ),
+    ),
+}
+""")
+    messages = list(extract.extract('python', buf, {'_': None}, ["NOTE"], {}))
+    message = messages[0]
+    assert message[0] in (5, 6)  # Depends on whether #1126 is in
+    assert message[1] == 'Text string that is on a new line'
+    assert message[2] == ['NOTE: Text describing a test string']
+
+
+def test_issue_1195_2():
+    buf = BytesIO(b"""
+# NOTE: This should still be considered, even if
+#       the text is far away
+foof = _(
+
+
+
+
+
+
+
+
+
+            'Hey! Down here!')
+""")
+    messages = list(extract.extract('python', buf, {'_': None}, ["NOTE"], {}))
+    message = messages[0]
+    assert message[1] == 'Hey! Down here!'
+    assert message[2] == [
+        'NOTE: This should still be considered, even if',
+        'the text is far away',
+    ]