Mercurial > p > roundup > code
annotate roundup/dehtml.py @ 5565:b3d5085dd04e REST-rebased
added custom parsing properties from arguments
committer: Ralf Schlatterbeck <rsc@runtux.com>
| author | Chau Nguyen <dangchau1991@yahoo.com> |
|---|---|
| date | Wed, 30 Jan 2019 10:26:34 +0100 |
| parents | c749d6795bc2 |
| children | b74f0b50bef1 |
| rev | line source |
|---|---|
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
1 |
|
5376
64b05e24dbd8
Python 3 preparation: convert print to a function.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5305
diff
changeset
|
2 from __future__ import print_function |
|
5417
c749d6795bc2
Python 3 preparation: unichr.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5416
diff
changeset
|
3 from roundup.anypy.strings import u2s, uchr |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
4 class dehtml: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
5 def __init__(self, converter): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
6 if converter == "none": |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
7 self.html2text = None |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
8 return |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
9 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
10 try: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
11 if converter == "beautifulsoup": |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
12 # Not as well tested as dehtml. |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
13 from bs4 import BeautifulSoup |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
14 def html2text(html): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
15 soup = BeautifulSoup(html) |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
16 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
17 # kill all script and style elements |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
18 for script in soup(["script", "style"]): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
19 script.extract() |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
20 |
|
5416
56c9bcdea47f
Python 3 preparation: unicode.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5411
diff
changeset
|
21 return u2s(soup.get_text('\n', strip=True)) |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
22 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
23 self.html2text = html2text |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
24 else: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
25 raise ImportError # use |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
26 except ImportError: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
27 # use the fallback below if beautiful soup is not installed. |
|
5411
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
28 try: |
|
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
29 # Python 3+. |
|
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
30 from html.parser import HTMLParser |
|
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
31 from html.entities import name2codepoint |
|
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
32 except ImportError: |
|
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
33 # Python 2. |
|
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
34 from HTMLParser import HTMLParser |
|
9c6d98bf79db
Python 3 preparation: update HTMLParser / htmlentitydefs imports.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5376
diff
changeset
|
35 from htmlentitydefs import name2codepoint |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
36 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
37 class DumbHTMLParser(HTMLParser): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
38 # class attribute |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
39 text="" |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
40 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
41 # internal state variable |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
42 _skip_data = False |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
43 _last_empty = False |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
44 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
45 def handle_data(self, data): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
46 if self._skip_data: # skip data if in script or style block |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
47 return |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
48 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
49 if ( data.strip() == ""): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
50 # reduce multiple blank lines to 1 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
51 if ( self._last_empty ): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
52 return |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
53 else: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
54 self._last_empty = True |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
55 else: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
56 self._last_empty = False |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
57 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
58 self.text=self.text + data |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
59 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
60 def handle_starttag(self, tag, attrs): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
61 if (tag == "p" ): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
62 self.text= self.text + "\n" |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
63 if (tag in ("style", "script")): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
64 self._skip_data = True |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
65 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
66 def handle_endtag(self, tag): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
67 if (tag in ("style", "script")): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
68 self._skip_data = False |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
69 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
70 def handle_entityref(self, name): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
71 if self._skip_data: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
72 return |
|
5417
c749d6795bc2
Python 3 preparation: unichr.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5416
diff
changeset
|
73 c = uchr(name2codepoint[name]) |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
74 try: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
75 self.text= self.text + c |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
76 except UnicodeEncodeError: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
77 # print a space as a placeholder |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
78 pass |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
79 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
80 def html2text(html): |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
81 parser = DumbHTMLParser() |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
82 parser.feed(html) |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
83 parser.close() |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
84 return parser.text |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
85 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
86 self.html2text = html2text |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
87 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
88 if "__main__" == __name__: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
89 html=''' |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
90 <body> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
91 <script> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
92 this must not be in output |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
93 </script> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
94 <style> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
95 p {display:block} |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
96 </style> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
97 <div class="header"><h1>Roundup</h1> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
98 <div id="searchbox" style="display: none"> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
99 <form class="search" action="../search.html" method="get"> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
100 <input type="text" name="q" size="18" /> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
101 <input type="submit" value="Search" /> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
102 <input type="hidden" name="check_keywords" value="yes" /> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
103 <input type="hidden" name="area" value="default" /> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
104 </form> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
105 </div> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
106 <script type="text/javascript">$('#searchbox').show(0);</script> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
107 </div> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
108 <ul class="current"> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
109 <li class="toctree-l1"><a class="reference internal" href="../index.html">Home</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
110 <li class="toctree-l1"><a class="reference external" href="http://pypi.python.org/pypi/roundup">Download</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
111 <li class="toctree-l1 current"><a class="reference internal" href="../docs.html">Docs</a><ul class="current"> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
112 <li class="toctree-l2"><a class="reference internal" href="features.html">Roundup Features</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
113 <li class="toctree-l2 current"><a class="current reference internal" href="">Installing Roundup</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
114 <li class="toctree-l2"><a class="reference internal" href="upgrading.html">Upgrading to newer versions of Roundup</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
115 <li class="toctree-l2"><a class="reference internal" href="FAQ.html">Roundup FAQ</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
116 <li class="toctree-l2"><a class="reference internal" href="user_guide.html">User Guide</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
117 <li class="toctree-l2"><a class="reference internal" href="customizing.html">Customising Roundup</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
118 <li class="toctree-l2"><a class="reference internal" href="admin_guide.html">Administration Guide</a></li> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
119 </ul> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
120 <div class="section" id="prerequisites"> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
121 <h2><a class="toc-backref" href="#id5">Prerequisites</a></h2> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
122 <p>Roundup requires Python 2.5 or newer (but not Python 3) with a functioning |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
123 anydbm module. Download the latest version from <a class="reference external" href="http://www.python.org/">http://www.python.org/</a>. |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
124 It is highly recommended that users install the latest patch version |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
125 of python as these contain many fixes to serious bugs.</p> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
126 <p>Some variants of Linux will need an additional “python dev” package |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
127 installed for Roundup installation to work. Debian and derivatives, are |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
128 known to require this.</p> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
129 <p>If you’re on windows, you will either need to be using the ActiveState python |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
130 distribution (at <a class="reference external" href="http://www.activestate.com/Products/ActivePython/">http://www.activestate.com/Products/ActivePython/</a>), or you’ll |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
131 have to install the win32all package separately (get it from |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
132 <a class="reference external" href="http://starship.python.net/crew/mhammond/win32/">http://starship.python.net/crew/mhammond/win32/</a>).</p> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
133 </div> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
134 </body> |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
135 ''' |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
136 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
137 html2text = dehtml("dehtml").html2text |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
138 if html2text: |
|
5376
64b05e24dbd8
Python 3 preparation: convert print to a function.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5305
diff
changeset
|
139 print(html2text(html)) |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
140 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
141 try: |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
142 # trap error seen if N_TOKENS not defined when run. |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
143 html2text = dehtml("beautifulsoup").html2text |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
144 if html2text: |
|
5376
64b05e24dbd8
Python 3 preparation: convert print to a function.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5305
diff
changeset
|
145 print(html2text(html)) |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
146 except NameError as e: |
|
5376
64b05e24dbd8
Python 3 preparation: convert print to a function.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5305
diff
changeset
|
147 print("captured error %s"%e) |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
148 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
149 html2text = dehtml("none").html2text |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
150 if html2text: |
|
5376
64b05e24dbd8
Python 3 preparation: convert print to a function.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5305
diff
changeset
|
151 print("FAIL: Error, dehtml(none) is returning a function") |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
152 else: |
|
5376
64b05e24dbd8
Python 3 preparation: convert print to a function.
Joseph Myers <jsm@polyomino.org.uk>
parents:
5305
diff
changeset
|
153 print("PASS: dehtml(none) is returning None") |
|
5305
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
154 |
|
e20f472fde7d
issue2550799: provide basic support for handling html only emails
John Rouillard <rouilj@ieee.org>
parents:
diff
changeset
|
155 |
