Do not linktrail if following text is not [a-z]?

See wiktectract issue #1604
tatuylonen/wiktextract#1604
https://en.wikipedia.org/wiki/Help:Wikitext#Blend_link

This should not be merged as is, because it will create problems in
other extractors that might rely on different behavior.

In the best-case scenario, there might be two different camps:
1) Languages that use spaces that want to do linktrailing
2) Languages without spaces that can't do linktrailing

If this is the case, we might be able to get away with a
kludge that checks whether the script of the last character
in the link matches the script of the first character after
the link.

Author: kristian-clausal

src/wikitextprocessor/parser.py

@@ -1029,7 +1029,7 @@ def text_fn(ctx: "Wtp", token: str) -> None:
         and not node.children[-1].children
         and not ctx.suppress_special
     ):
-        m = re.match(r"(?s)(\w+)(.*)", token)
+        m = re.match(r"(?s)([a-z]+)(.*)", token)
         if m:
             node.children[-1].children.append(m.group(1))
             token = m.group(2)

src/wikitextprocessor/parserfns.py

@@ -9,7 +9,7 @@
 from collections.abc import Callable, Sequence
 from datetime import datetime, timezone
 from pathlib import Path
-from typing import TYPE_CHECKING, Optional, Union
+from typing import TYPE_CHECKING, Any, Optional, Union
 
 import dateparser
 
@@ -1185,7 +1185,7 @@ def parse_timestamp(
     if not dt:
         dt = "now"
 
-    settings: dateparser._Settings = {"RETURN_AS_TIMEZONE_AWARE": True}
+    settings: dict[str, Any] = {"RETURN_AS_TIMEZONE_AWARE": True}
     if loc in ("", "0"):
         dt += " UTC"
 
@@ -1206,15 +1206,15 @@ def parse_timestamp(
         # php's strtotime() (which is the original function used)
         # but we can handle special cases here and hope
         # people on wiktionary don't go crazy with weird formatting
-        t = dateparser.parse(dt, settings=settings)
+        t = dateparser.parse(dt, settings=settings)  # type: ignore
         if t is None:
             m = re.match(
                 r"([^+]*)\s*(\+\s*\d+\s*(day|year|month)s?)\s*$", orig_dt
             )
             if m:
-                main_date = dateparser.parse(m.group(1), settings=settings)
-                add_time = dateparser.parse(m.group(2), settings=settings)
-                now = dateparser.parse("now", settings=settings)
+                main_date = dateparser.parse(m.group(1), settings=settings)  # type: ignore
+                add_time = dateparser.parse(m.group(2), settings=settings)  # type: ignore
+                now = dateparser.parse("now", settings=settings)  # type: ignore
                 if main_date and add_time is not None and now is not None:
                     # this is just a kludge: dateparser parses "+2 days" as
                     # "2 days AGO". The now-datetime object is used to check

tests/test_parser.py

@@ -1101,7 +1101,7 @@ def test_link12(self):
         self.assertEqual(link.kind, NodeKind.LINK)
         self.assertEqual(link.largs, [["foo"], ["\n[bar"]])
 
-    def test_link_trailing(self):
+    def test_link_trailing_1(self):
         tree = self.parse("test", "[[Help]]ing heal")
         self.assertEqual(len(tree.children), 2)
         a, b = tree.children
@@ -1110,6 +1110,15 @@ def test_link_trailing(self):
         self.assertEqual(a.children, ["ing"])
         self.assertEqual(b, " heal")
 
+    def test_link_trailing_not_latin(self):
+        tree = self.parse("test", "[[appellāre]]の直説法所相現在第 foo")
+        self.assertEqual(len(tree.children), 2)
+        a, b = tree.children
+        self.assertEqual(a.kind, NodeKind.LINK)
+        self.assertEqual(a.largs, [["appellāre"]])
+        self.assertEqual(a.children, [])
+        self.assertEqual(b, "の直説法所相現在第 foo")
+
     def test_url1(self):
         tree = self.parse("test", "this https://wikipedia.com link")
         self.assertEqual(len(tree.children), 3)