test_markdown.py

#! /usr/bin/env python

import re

from markdown2 import markdown


link_patterns = [
  # issueNNN --> /help/issueNNN
  (re.compile(r'\b(issue)\s*(\d+)\b'), r'/help/\g<1>\g<2>'),
  # web URL
  (re.compile(r'''(?xi)
  (?<!['"])                               (?# not preceded by a quotes, avoid recursive processing of href="..." )
\b
( (?# Capture group 1: entire matched URL)
  (?:ftp|https?)://                       (?# URL schema)
  (?:                                     (?# optional userinfo part: )
    [a-z0-9._~\-%!$&'()*+,;=:]+             (?# see RFC 3986, sec. 3.2.1 )
    @                                       (?# literal '@' mark )
  )?
  (?:                                     (?# host, either one of the following: )
    [a-z0-9][a-z0-9.-]+\.(?:[a-z][a-z]+)    (?# looks like an FQDN )
    |
    [0-9]+(?:\.[0-9]+){3}                   (?# IPv4 dotted-quad )
    |
    \[ [0-9a-f]+(?::[0-9a-f]+){7} \]        (?# IPv6 full address literal )
    |
    \[ ([0-9a-f]+(?: :[0-9a-f]+ )* )? :: ([0-9a-f]+(?: :[0-9a-f]+)* ) \]
                                            (?# looks like an IPv6 abbreviated address literal )
  )
  (?: :[0-9]+ )?                          (?# optional port number )
  (?:                                     (?# optional path: )
    /                                       (?# literal slash )
    [a-z0-9._~\-%!$&'()*+,;=:@]*            (?# see RFC 3986, sec. 2.2, 2.3, and 3.3 )
  )*                                        (?# repeat zero or more times )
  (?:                                     (?# optional query part: )
    [?]                                     (?# literal question mark )
    [a-z0-9._~\-%!$&'()*+,;=:@?/]*          (?# see RFC 3986, sec. 2.2, 2.3, and 3.4 )
  )?                                        (?# at most once )
  (?:                                     (?# optional fragment part: )
    [#]                                     (?# literal hash mark )
    [a-z0-9._~\-%!$&'()*+,;=:@?/]*          (?# see RFC 3986, sec. 2.2, 2.3, and 3.5 )
  )?                                        (?# at most once )
)
  '''), r'\g<1>'),
  # "naked" domain, assume `http://` URL
  (re.compile(r'''(?xi)
  (?<!/)                       (?# not preceded by a /, avoid matching http://example.com )
  (?<!\.)                      (?# not preceded by a ., avoid matching just `com` in example.com )
  (?<!@)                       (?# not preceded by a @, avoid matching foo@_gmail.com_ )
  \b
  ( (?# capture group 1: domain name)
    [a-z0-9][a-z0-9.-]+\.(?:[a-z][a-z]+) (?# looks like a domain name )
    /?                           (?# optional slash )
    (?!@)                        (?# not followed by a @, avoid matching "foo.na" in "foo.na@example.com" )
  )
  \b
  '''), r'http://\g<1>')
]


def _to_markdown(text):
  html = markdown(
    text,
    extras=['footnotes', 'fenced-code-blocks', 'link-patterns', 'footnotes'],
    link_patterns=link_patterns)
  # Ensure the code is converted to utf-8
  return html.encode('utf-8')

def test_link_patterns():
  for src, expected in (
      ("issue1",   '<p><a href="/help/issue1">issue1</a></p>'),
      ("issue 2",  '<p><a href="/help/issue2">issue 2</a></p>'),
      ("nonlink1", '<p>nonlink1</p>'),
      ("http://www.google.com",
                   '<p><a href="http://www.google.com">http://www.google.com</a></p>'),
      ("http://www.google.com/",
                   '<p><a href="http://www.google.com/">http://www.google.com/</a></p>'),
      ("http://www.google.com:80/",
                   '<p><a href="http://www.google.com:80/">http://www.google.com:80/</a></p>'),
      ("http://www.google.com:80",
                   '<p><a href="http://www.google.com:80">http://www.google.com:80</a></p>'),
      ("http://www.google.com/search",
                   '<p><a href="http://www.google.com/search">http://www.google.com/search</a></p>'),
      ("http://www.google.com/search?q=regexp",
                   '<p><a href="http://www.google.com/search?q=regexp">http://www.google.com/search?q=regexp</a></p>'),
      ("[google](https://google.com)",
                   '<p><a href="https://google.com">google</a></p>'),
      ("""
[Google][1]

[1]: http://www.google.com
""",
                   '<p><a href="http://www.google.com">Google</a></p>'),
      ("This is text with http://www.example.org/foo/bar/something link",
       '<p>This is text with <a href="http://www.example.org/foo/bar/something">http://www.example.org/foo/bar/something</a> link</p>'),
      ("Dotted-quad addresses are also supported, as in http://192.0.2.5/",
       '<p>Dotted-quad addresses are also supported, as in <a href="http://192.0.2.5/">http://192.0.2.5/</a></p>'),
      ("IPv6 full address literals, as in http://[2001:DB8:0:0:0:0:0:1]",
       '<p>IPv6 full address literals, as in <a href="http://[2001:DB8:0:0:0:0:0:1]">http://[2001:DB8:0:0:0:0:0:1]</a></p>'),
      ("IPv6 abbreviated address literals, as in http://[::1]",
       '<p>IPv6 abbreviated address literals, as in <a href="http://[::1]">http://[::1]</a></p>'),
      ("IPv6 abbreviated address literals w/ port number, as in http://[::1]:8080",
       '<p>IPv6 abbreviated address literals w/ port number, as in <a href="http://[::1]:8080">http://[::1]:8080</a></p>'),
    ("FTP URLs, as in ftp://ftp.example.org/",
       '<p>FTP URLs, as in <a href="ftp://ftp.example.org/">ftp://ftp.example.org/</a></p>'),
    ("FTP URLs with explicit username, as in ftp://anonymous@ftp.example.org",
       '<p>FTP URLs with explicit username, as in <a href="ftp://anonymous@ftp.example.org">ftp://anonymous@ftp.example.org</a></p>'),
    ("FTP URLs with explicit username, as in ftp://anonymous@ftp.example.org",
       '<p>FTP URLs with explicit username, as in <a href="ftp://anonymous@ftp.example.org">ftp://anonymous@ftp.example.org</a></p>'),
    ("FTP URLs with username and password (INSECURE!), as in ftp://anonymous:guesswho@ftp.example.org",
       '<p>FTP URLs with username and password (INSECURE!), as in <a href="ftp://anonymous:guesswho@ftp.example.org">ftp://anonymous:guesswho@ftp.example.org</a></p>'),
  ):
    actual = _to_markdown(src).strip()
    assert actual == expected, \
      ("Rendering of '%(src)s'"
       " expected to be:\n\t '%(expected)s'\n"
       " but gotten:\n\t '%(actual)s'\n instead."
       % locals())

if __name__ == "__main__":
    test_link_patterns()