From e041a77cc9cf60fe7bce435cd19d7e2b1c662263 Mon Sep 17 00:00:00 2001
From: Branko Majic <branko@majic.rs>
Date: Wed, 21 Feb 2018 19:59:04 +0100
Subject: [PATCH] Enforce the internal urlize markdown extensions works only
 with links stand as "separate words":

- Introduced additional tests to ensure links are rendered correctly
  when in middle of text.
- Introduced additional tests to ensure links are not rendered in case
  of invalid patterns.
- Ensure the link pattern is matched only for standalone words (with
  "spacing" around the links).
---
 src/wiki/plugins/links/mdx/urlize.py |  3 ++-
 tests/plugins/links/test_urlize.py   | 26 ++++++++++++++++++++++++++
 2 files changed, 28 insertions(+), 1 deletion(-)

diff --git a/src/wiki/plugins/links/mdx/urlize.py b/src/wiki/plugins/links/mdx/urlize.py
index 3dc630c6..901ce1c2 100644
--- a/src/wiki/plugins/links/mdx/urlize.py
+++ b/src/wiki/plugins/links/mdx/urlize.py
@@ -91,7 +91,8 @@ class UrlizePattern(markdown.inlinepatterns.Pattern):
         to standard matching flags added by parent class.
         """
 
-        return re.compile(r'^(.*?)%s(.*?)$' % URLIZE_RE, re.DOTALL | re.UNICODE | re.IGNORECASE)
+        # Ensure links are matched only if they stand on their own to avoid bad matches etc.
+        return re.compile(r'^(|.*?\s)%s(\s.*?|)$' % URLIZE_RE, re.DOTALL | re.UNICODE | re.IGNORECASE)
 
     def handleMatch(self, m):
         """
diff --git a/tests/plugins/links/test_urlize.py b/tests/plugins/links/test_urlize.py
index 079c123a..37750eef 100644
--- a/tests/plugins/links/test_urlize.py
+++ b/tests/plugins/links/test_urlize.py
@@ -120,6 +120,24 @@ FIXTURE_POSITIVE_MATCHES = [
         'http://example.com/my/path?param1=value1&param2=value2',
         EXPECTED_PARAGRAPH_TEMPLATE % ('http://example.com/my/path?param1=value1&amp;param2=value2', 'http://example.com/my/path?param1=value1&amp;param2=value2')
     ),
+
+    # Link positioned somewhere within the text, but around whitespace boundary.
+    (
+        'This is link myhost.example.com',
+        "<p>This is link " + EXPECTED_LINK_TEMPLATE % ('http://myhost.example.com', 'myhost.example.com') + "</p>"
+    ),
+    (
+        'myhost.example.com is the link',
+        "<p>" + EXPECTED_LINK_TEMPLATE % ('http://myhost.example.com', 'myhost.example.com') + " is the link</p>"
+    ),
+    (
+        'I have best myhost.example.com link ever',
+        "<p>I have best " + EXPECTED_LINK_TEMPLATE % ('http://myhost.example.com', 'myhost.example.com') + " link ever</p>"
+    ),
+    (
+        'I have best\nmyhost.example.com link ever',
+        "<p>I have best\n" + EXPECTED_LINK_TEMPLATE % ('http://myhost.example.com', 'myhost.example.com') + " link ever</p>"
+    ),
 ]
 
 
@@ -139,6 +157,14 @@ FIXTURE_NEGATIVE_MATCHES = [
         'example-.com',
         '<p>example-.com</p>'
     ),
+    (
+        '-example.com',
+        '<p>-example.com</p>'
+    ),
+    (
+        'my.-example.com',
+        '<p>my.-example.com</p>'
+    ),
 ]
 
 
-- 
2.45.2