Mercurial > p > roundup > code
comparison roundup/dehtml.py @ 5838:b74f0b50bef1
Fix CI deprication warning for HTMLParser convert_charrefs under py3.
/home/travis/build/roundup-tracker/roundup/roundup/dehtml.py:81:
DeprecationWarning: The value of convert_charrefs will become True in
3.5. You are encouraged to set the value explicitly.
parser = DumbHTMLParser()
| author | John Rouillard <rouilj@ieee.org> |
|---|---|
| date | Sat, 06 Jul 2019 17:36:25 -0400 |
| parents | c749d6795bc2 |
| children | 1700542408f3 |
comparison
equal
deleted
inserted
replaced
| 5837:883c9e90b403 | 5838:b74f0b50bef1 |
|---|---|
| 27 # use the fallback below if beautiful soup is not installed. | 27 # use the fallback below if beautiful soup is not installed. |
| 28 try: | 28 try: |
| 29 # Python 3+. | 29 # Python 3+. |
| 30 from html.parser import HTMLParser | 30 from html.parser import HTMLParser |
| 31 from html.entities import name2codepoint | 31 from html.entities import name2codepoint |
| 32 pyver=3 | |
| 32 except ImportError: | 33 except ImportError: |
| 33 # Python 2. | 34 # Python 2. |
| 34 from HTMLParser import HTMLParser | 35 from HTMLParser import HTMLParser |
| 35 from htmlentitydefs import name2codepoint | 36 from htmlentitydefs import name2codepoint |
| 37 pyver=2 | |
| 36 | 38 |
| 37 class DumbHTMLParser(HTMLParser): | 39 class DumbHTMLParser(HTMLParser): |
| 38 # class attribute | 40 # class attribute |
| 39 text="" | 41 text="" |
| 40 | 42 |
| 73 c = uchr(name2codepoint[name]) | 75 c = uchr(name2codepoint[name]) |
| 74 try: | 76 try: |
| 75 self.text= self.text + c | 77 self.text= self.text + c |
| 76 except UnicodeEncodeError: | 78 except UnicodeEncodeError: |
| 77 # print a space as a placeholder | 79 # print a space as a placeholder |
| 78 pass | 80 self.text= self.text + ' ' |
| 79 | 81 |
| 80 def html2text(html): | 82 def html2text(html): |
| 81 parser = DumbHTMLParser() | 83 if pyver == 3: |
| 84 parser = DumbHTMLParser(convert_charrefs=True) | |
| 85 else: | |
| 86 parser = DumbHTMLParser() | |
| 82 parser.feed(html) | 87 parser.feed(html) |
| 83 parser.close() | 88 parser.close() |
| 84 return parser.text | 89 return parser.text |
| 85 | 90 |
| 86 self.html2text = html2text | 91 self.html2text = html2text |
| 128 known to require this.</p> | 133 known to require this.</p> |
| 129 <p>If you’re on windows, you will either need to be using the ActiveState python | 134 <p>If you’re on windows, you will either need to be using the ActiveState python |
| 130 distribution (at <a class="reference external" href="http://www.activestate.com/Products/ActivePython/">http://www.activestate.com/Products/ActivePython/</a>), or you’ll | 135 distribution (at <a class="reference external" href="http://www.activestate.com/Products/ActivePython/">http://www.activestate.com/Products/ActivePython/</a>), or you’ll |
| 131 have to install the win32all package separately (get it from | 136 have to install the win32all package separately (get it from |
| 132 <a class="reference external" href="http://starship.python.net/crew/mhammond/win32/">http://starship.python.net/crew/mhammond/win32/</a>).</p> | 137 <a class="reference external" href="http://starship.python.net/crew/mhammond/win32/">http://starship.python.net/crew/mhammond/win32/</a>).</p> |
| 138 <script> | |
| 139 < HELP > | |
| 140 </script> | |
| 133 </div> | 141 </div> |
| 134 </body> | 142 </body> |
| 135 ''' | 143 ''' |
| 136 | 144 |
| 137 html2text = dehtml("dehtml").html2text | 145 html2text = dehtml("dehtml").html2text |