jenkins-bot merged this change.

View Change

Use the old FILE_LINK_REGEX regex for Python versions older than 2.7.4

This patch reverts c25f22164f88b1bf91dd329d0566afed685ce530 and limits
the usage of the new regex introduced in 8ede851d8c0655b7eb021e69003e9
to Python 2.7.4+.

Although the old regex may get stuck in catastrophic backtracking, but the
chances of that situation happening are much lower and we have been using
if for a long time (since 2016-10-31).

Bug: T191161
Change-Id: I1567886eeacc82b3a393d818f98cdfb0216b1b31
---
M pywikibot/textlib.py
1 file changed, 35 insertions(+), 23 deletions(-)

diff --git a/pywikibot/textlib.py b/pywikibot/textlib.py
index 694e781..8f3da8a 100644
--- a/pywikibot/textlib.py
+++ b/pywikibot/textlib.py
@@ -103,32 +103,44 @@
 # [[ or ]].
 # The namespace names must be substituted into this regex.
 # e.g. FILE_LINK_REGEX % 'File' or FILE_LINK_REGEX % '|'.join(site.namespaces)
-FILE_LINK_REGEX = r"""
-    \[\[\s*
-    (?:%s)  # namespace aliases
-    \s*:
-    ((?=(?P<filename>
-        [^]|]+  # * quantifier may crash on Python 2.7.2 (T191161)
-    ))(?P=filename))?
-    (
-        \|
+if sys.version_info[:3] >= (2, 7, 4):
+    FILE_LINK_REGEX = r"""
+        \[\[\s*
+        (?:%s)  # namespace aliases
+        \s*:
+        (?=(?P<filename>
+            [^]|]*
+        ))(?P=filename)
         (
+            \|
             (
-                (?=(?P<inner_link>
-                    \[\[.*?\]\]
-                ))(?P=inner_link)
-            )?
-            ((?=(?P<other_chars>
-                [^\[\]]+  # * quantifier may crash on Python 2.7.2 (T191161)
-            ))(?P=other_chars))?
-        |
-            (?=(?P<not_wikilink>
-                \[[^]]*\]
-            ))(?P=not_wikilink)
-        )*?
-    )??
+                (
+                    (?=(?P<inner_link>
+                        \[\[.*?\]\]
+                    ))(?P=inner_link)
+                )?
+                (?=(?P<other_chars>
+                    [^\[\]]*
+                ))(?P=other_chars)
+            |
+                (?=(?P<not_wikilink>
+                    \[[^]]*\]
+                ))(?P=not_wikilink)
+            )*?
+        )??
+        \]\]
+    """
+else:
+    # Python 2.7.2 and 2.7.3 re bug (T191161)
+    FILE_LINK_REGEX = r"""
+    \[\[\s*(?:%s)\s*:[^|]*?\s*
+      (\|
+        ( ( \[\[ .*? \]\] )? [^[]*?
+         | \[ [^]]*? \]
+        )*
+      )?
     \]\]
-"""
+    """
 
 NON_LATIN_DIGITS = {
     'ckb': u'٠١٢٣٤٥٦٧٨٩',

To view, visit change 423463. To unsubscribe, visit settings.