Optimize several regexes from quadratic time to linear time

Part of the discussion in #798.

Signed-off-by: Anders Kaseorg <[email protected]>

Author: andersk

Diff for: markdown/inlinepatterns.py

@@ -147,10 +147,10 @@ def build_inlinepatterns(md, **kwargs):
 NOT_STRONG_RE = r'((^|\s)(\*|_)(\s|$))'
 
 # <http://www.123.com>
-AUTOLINK_RE = r'<((?:[Ff]|[Hh][Tt])[Tt][Pp][Ss]?://[^>]*)>'
+AUTOLINK_RE = r'<((?:[Ff]|[Hh][Tt])[Tt][Pp][Ss]?://[^<>]*)>'
 
 # <[email protected]>
-AUTOMAIL_RE = r'<([^> \!]*@[^> ]*)>'
+AUTOMAIL_RE = r'<([^<> !]*@[^@<> ]*)>'
 
 # <...>
 HTML_RE = r'(\<([a-zA-Z/][^\>]*?|\!--.*?--)\>)'
@@ -433,7 +433,7 @@ def get_stash(m):
 
 class LinkInlineProcessor(InlineProcessor):
     """ Return a link element from the given match. """
-    RE_LINK = re.compile(r'''\(\s*(?:(<.*?>)\s*(?:(['"])(.*?)\2\s*)?\))?''', re.DOTALL | re.UNICODE)
+    RE_LINK = re.compile(r'''\(\s*(?:(<[^<>]*>)\s*(?:('[^']*'|"[^"]*")\s*)?\))?''', re.DOTALL | re.UNICODE)
     RE_TITLE_CLEAN = re.compile(r'\s')
 
     def handleMatch(self, m, data):
@@ -467,8 +467,8 @@ def getLink(self, data, index):
         if m and m.group(1):
             # Matches [Text](<link> "title")
             href = m.group(1)[1:-1].strip()
-            if m.group(3):
-                title = m.group(3)
+            if m.group(2):
+                title = m.group(2)[1:-1]
             index = m.end(0)
             handled = True
         elif m:

Diff for: tests/misc/html.html

@@ -20,7 +20,7 @@ <h1>Block level html</h1>
 </div>
 
 <p>And of course <script>blah</script>.</p>
-<p><a href="script&gt;stuff&lt;/script">this <script>link</a></p>
+<p><a href="&lt;script&gt;stuff&lt;/script&gt;">this <script>link</a></p>
 <p>Some funky <x\]> inline stuff with markdown escaping syntax.</p>
 <p><img scr="foo.png" title="Only one inline element on a line." /></p>
 <p>And now a line with only an opening bracket:</p>