diff options
| author | Anders Kaseorg <andersk@mit.edu> | 2019-03-06 12:36:27 -0800 |
|---|---|---|
| committer | Waylan Limberg <waylan.limberg@icloud.com> | 2019-03-06 20:47:05 -0500 |
| commit | cb47805e10333dfbebc094d9f07fdf18e200aa89 (patch) | |
| tree | 4df80a2245df40a27e1b151d49f7279855b70e39 | |
| parent | 4b11593b681966e5c9a7a70c84a19546fc2e185c (diff) | |
| download | python-markdown-cb47805e10333dfbebc094d9f07fdf18e200aa89.tar.gz | |
Optimize several regexes from quadratic time to linear time
Part of the discussion in #798.
Signed-off-by: Anders Kaseorg <andersk@mit.edu>
| -rw-r--r-- | markdown/inlinepatterns.py | 10 | ||||
| -rw-r--r-- | tests/misc/html.html | 2 |
2 files changed, 6 insertions, 6 deletions
diff --git a/markdown/inlinepatterns.py b/markdown/inlinepatterns.py index 88e8ead..74cda18 100644 --- a/markdown/inlinepatterns.py +++ b/markdown/inlinepatterns.py @@ -147,10 +147,10 @@ IMAGE_REFERENCE_RE = IMAGE_LINK_RE 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]?://[^<>]*)>' # <me@example.com> -AUTOMAIL_RE = r'<([^> \!]*@[^> ]*)>' +AUTOMAIL_RE = r'<([^<> !]*@[^@<> ]*)>' # <...> HTML_RE = r'(\<([a-zA-Z/][^\>]*?|\!--.*?--)\>)' @@ -433,7 +433,7 @@ class HtmlInlineProcessor(InlineProcessor): 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 @@ class LinkInlineProcessor(InlineProcessor): 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 --git a/tests/misc/html.html b/tests/misc/html.html index 5380bbd..293e6cc 100644 --- a/tests/misc/html.html +++ b/tests/misc/html.html @@ -20,7 +20,7 @@ Html with various attributes. </div> <p>And of course <script>blah</script>.</p> -<p><a href="script>stuff</script">this <script>link</a></p> +<p><a href="<script>stuff</script>">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> |