python-babel · akx · Oct 1, 2023 · Jun 14, 2023 · Jul 1, 2023 · Aug 1, 2023
diff --git a/babel/messages/extract.py b/babel/messages/extract.py
@@ -55,7 +55,8 @@ class _FileObj(SupportsRead[bytes], SupportsReadline[bytes], Protocol):
         def seek(self, __offset: int, __whence: int = ...) -> int: ...
         def tell(self) -> int: ...
 
-    _Keyword: TypeAlias = tuple[int | tuple[int, int] | tuple[int, str], ...] | None
+    _SimpleKeyword: TypeAlias = tuple[int | tuple[int, int] | tuple[int, str], ...] | None
+    _Keyword: TypeAlias = dict[int | None, _SimpleKeyword] | _SimpleKeyword
 
     # 5-tuple of (filename, lineno, messages, comments, context)
     _FileExtractionResult: TypeAlias = tuple[str, int, str | tuple[str, ...], list[str], str | None]
@@ -400,56 +401,65 @@ def extract(
                    options=options or {})
 
     for lineno, funcname, messages, comments in results:
-        spec = keywords[funcname] or (1,) if funcname else (1,)
+        specs = keywords[funcname] or (1,) if funcname else (1,)
+        if not isinstance(specs, dict): # for backwards compatibility
+            specs = {None: specs}
         if not isinstance(messages, (list, tuple)):
             messages = [messages]
         if not messages:
             continue
 
         # Validate the messages against the keyword's specification
-        context = None
-        msgs = []
-        invalid = False
-        # last_index is 1 based like the keyword spec
-        last_index = len(messages)
-        for index in spec:
-            if isinstance(index, tuple):
-                context = messages[index[0] - 1]
+
+        # None matches all arities.
+        for arity in (None, len(messages)):
+            try:
+                spec = specs[arity]
+            except KeyError:
+                continue
+            context = None
+            msgs = []
+            invalid = False
+            # last_index is 1 based like the keyword spec
+            last_index = len(messages)
+            for index in spec:
+                if isinstance(index, tuple):
+                    context = messages[index[0] - 1]
+                    continue
+                if last_index < index:
+                    # Not enough arguments
+                    invalid = True
+                    break
+                message = messages[index - 1]
+                if message is None:
+                    invalid = True
+                    break
+                msgs.append(message)
+            if invalid:
                 continue
-            if last_index < index:
-                # Not enough arguments
-                invalid = True
-                break
-            message = messages[index - 1]
-            if message is None:
-                invalid = True
-                break
-            msgs.append(message)
-        if invalid:
-            continue
 
-        # keyword spec indexes are 1 based, therefore '-1'
-        if isinstance(spec[0], tuple):
-            # context-aware *gettext method
-            first_msg_index = spec[1] - 1
-        else:
-            first_msg_index = spec[0] - 1
-        if not messages[first_msg_index]:
-            # An empty string msgid isn't valid, emit a warning
-            filename = (getattr(fileobj, "name", None) or "(unknown)")
-            sys.stderr.write(
-                f"{filename}:{lineno}: warning: Empty msgid.  It is reserved by GNU gettext: gettext(\"\") "
-                f"returns the header entry with meta information, not the empty string.\n"
-            )
-            continue
+            # keyword spec indexes are 1 based, therefore '-1'
+            if isinstance(spec[0], tuple):
+                # context-aware *gettext method
+                first_msg_index = spec[1] - 1
+            else:
+                first_msg_index = spec[0] - 1
+            if not messages[first_msg_index]:
+                # An empty string msgid isn't valid, emit a warning
+                filename = (getattr(fileobj, "name", None) or "(unknown)")
+                sys.stderr.write(
+                    f"{filename}:{lineno}: warning: Empty msgid.  It is reserved by GNU gettext: gettext(\"\") "
+                    f"returns the header entry with meta information, not the empty string.\n"
+                )
+                continue
 
-        messages = tuple(msgs)
-        if len(messages) == 1:
-            messages = messages[0]
+            msgs = tuple(msgs)
+            if len(msgs) == 1:
+                msgs = msgs[0]
 
-        if strip_comment_tags:
-            _strip_comment_tags(comments, comment_tags)
-        yield lineno, messages, comments, context
+            if strip_comment_tags:
+                _strip_comment_tags(comments, comment_tags)
+            yield lineno, msgs, comments, context
 
 
 def extract_nothing(

diff --git a/babel/messages/frontend.py b/babel/messages/frontend.py
@@ -1111,30 +1111,46 @@ def parse_mapping(fileobj, filename=None):
 def parse_keywords(strings: Iterable[str] = ()):
     """Parse keywords specifications from the given list of strings.
 
-    >>> kw = sorted(parse_keywords(['_', 'dgettext:2', 'dngettext:2,3', 'pgettext:1c,2']).items())
+    >>> kw = sorted(parse_keywords(['_', 'dgettext:2', 'dngettext:2,3', 'pgettext:1c,2', 'polymorphic:1', 'polymorphic:2,2t', 'polymorphic:3c,3t']).items())
     >>> for keyword, indices in kw:
     ...     print((keyword, indices))
     ('_', None)
     ('dgettext', (2,))
     ('dngettext', (2, 3))
     ('pgettext', ((1, 'c'), 2))
+    ('polymorphic', {None: (1,), 2: (2,), 3: ((3, 'c'),)})
     """
     keywords = {}
     for string in strings:
         if ':' in string:
             funcname, indices = string.split(':')
         else:
             funcname, indices = string, None
+        number = None
+        if indices:
+            inds = []
+            for x in indices.split(','):
+                if x[-1] == 't':
+                    number = int(x[:-1])
+                elif x[-1] == 'c':
+                    inds.append((int(x[:-1]), 'c'))
+                else:
+                    inds.append(int(x))
+            inds = tuple(inds)
+        else:
+            inds = None
         if funcname not in keywords:
-            if indices:
-                inds = []
-                for x in indices.split(','):
-                    if x[-1] == 'c':
-                        inds.append((int(x[:-1]), 'c'))
-                    else:
-                        inds.append(int(x))
-                indices = tuple(inds)
-            keywords[funcname] = indices
+            if number is None:
+                # For best backwards compatibility, collapse {None: x} into x.
+                keywords[funcname] = inds
+            else:
+                keywords[funcname] = {number: inds}
+        else:
+            if isinstance(keywords[funcname], tuple) or keywords[funcname] is None:
+                keywords[funcname] = {None: keywords[funcname]}
+            else:
+                assert isinstance(keywords[funcname], dict)
+            keywords[funcname][number] = inds
     return keywords
 
 

diff --git a/tests/messages/test_frontend.py b/tests/messages/test_frontend.py
@@ -17,15 +17,15 @@
 import time
 import unittest
 from datetime import datetime
-from io import StringIO
+from io import BytesIO, StringIO
 
 import pytest
 from freezegun import freeze_time
 from setuptools import Distribution
 
 from babel import __version__ as VERSION
 from babel.dates import format_datetime
-from babel.messages import Catalog, frontend
+from babel.messages import Catalog, extract, frontend
 from babel.messages.frontend import (
     BaseError,
     CommandLineInterface,
@@ -1320,6 +1320,35 @@ def test_parse_keywords():
     }
 
 
+def test_parse_keywords_with_t():
+    kw = frontend.parse_keywords(['_:1', '_:2,2t', '_:2c,3,3t'])
+
+    assert kw == {
+        '_': {
+            None: (1,),
+            2: (2,),
+            3: ((2, 'c'), 3),
+        }
+    }
+
+def test_extract_messages_with_t():
+    content = rb"""
+_("1 arg, arg 1")
+_("2 args, arg 1", "2 args, arg 2")
+_("3 args, arg 1", "3 args, arg 2", "3 args, arg 3")
+_("4 args, arg 1", "4 args, arg 2", "4 args, arg 3", "4 args, arg 4")
+"""
+    kw = frontend.parse_keywords(['_:1', '_:2,2t', '_:2c,3,3t'])
+    result = list(extract.extract("python", BytesIO(content), kw))
+    expected = [(2, '1 arg, arg 1', [], None),
+                (3, '2 args, arg 1', [], None),
+                (3, '2 args, arg 2', [], None),
+                (4, '3 args, arg 1', [], None),
+                (4, '3 args, arg 3', [], '3 args, arg 2'),
+                (5, '4 args, arg 1', [], None)]
+    assert result == expected
+
+
 def configure_cli_command(cmdline):
     """
     Helper to configure a command class, but not run it just yet.