BeautifulSoup vs. loose & chars

[John Nagle]

I've been parsing existing HTML with BeautifulSoup, and occasionally
hit content which has something like "Design & Advertising", that is,
an "&" instead of an "&". Is there some way I can get BeautifulSoup
to clean those up? There are various parsing options related to "&"
handling, but none of them seem to do quite the right thing.

If I write the BeautifulSoup parse tree back out with "prettify",
the loose "&" is still in there. So the output is
rejected by XML parsers. Which is why this is a problem.
I need valid XML out, even if what went in wasn't quite valid.

John Nagle

[placid]

John Nagle wrote:

I've been parsing existing HTML with BeautifulSoup, and occasionally
hit content which has something like "Design & Advertising", that is,
an "&" instead of an "&". Is there some way I can get BeautifulSoup
to clean those up? There are various parsing options related to "&"
handling, but none of them seem to do quite the right thing.

If I write the BeautifulSoup parse tree back out with "prettify",
the loose "&" is still in there. So the output is
rejected by XML parsers. Which is why this is a problem.
I need valid XML out, even if what went in wasn't quite valid.

John Nagle

So do you want to remove "&" or replace them with "&" ? If you want
to replace it try the following;

import urllib, sys

try:
location = urllib.urlopen(url)
except IOError, (errno, strerror):
sys.exit("I/O error(%s): %s" % (errno, strerror))

content = location.read()
content = content.replace("&", "&")
To do this with BeautifulSoup, i think you need to go through every
Tag, get its content, see if it contains an "&" and then replace the
Tag with the same Tag but the content contains "&"

Hope this helps.
Cheers

[Felipe Almeida Lessa]

On 26 Dec 2006 04:22:38 -0800, placid <Bu****@gmail.comwrote:

So do you want to remove "&" or replace them with "&amp;" ? If you want
to replace it try the following;

I think he wants to replace them, but just the invalid ones. I.e.,

This & this &amp; that

would become

This &amp; this &amp; that
No, i don't know how to do this efficiently. =/...
I think some kind of regex could do it.

--
Felipe.

[Duncan Booth]

"Felipe Almeida Lessa" <fe**********@gmail.comwrote:

On 26 Dec 2006 04:22:38 -0800, placid <Bu****@gmail.comwrote:

>So do you want to remove "&" or replace them with "&amp;" ? If you
want to replace it try the following;

I think he wants to replace them, but just the invalid ones. I.e.,

This & this &amp; that

would become

This &amp; this &amp; that
No, i don't know how to do this efficiently. =/...
I think some kind of regex could do it.

Since he's asking for valid xml as output, it isn't sufficient just to
ignore entity definitions: HTML has a lot of named entities such as
&nbsp; but xml only has a very limited set of predefined named entities.
The safest technique is to convert them all to numeric escapes except
for the very limited set also guaranteed to be available in xml.

Try this:

from cgi import escape
import re
from htmlentitydefs import name2codepoint
name2codepoint = name2codepoint.copy()
name2codepoint['apos']=ord("'")

EntityPattern =
re.compile('&(?:#(\d+)|(?:#x([\da-fA-F]+))|([a-zA-Z]+));')

def decodeEntities(s, encoding='utf-8'):
def unescape(match):
code = match.group(1)
if code:
return unichr(int(code, 10))
else:
code = match.group(2)
if code:
return unichr(int(code, 16))
else:
return unichr(name2codepoint[match.group(3)])
return EntityPattern.sub(unescape, s)

>>escape(

decodeEntities("This & this &amp; that&nbsp;&eacute;")).encode(
'ascii', 'xmlcharrefreplace')
'This &amp; this &amp; that é'
P.S. apos is handled specially as it isn't technically a
valid html entity (and Python doesn't include it in its entity
list), but it is an xml entity and recognised by many browsers so some
people might use it in html.

[John Nagle]

Felipe Almeida Lessa wrote:

On 26 Dec 2006 04:22:38 -0800, placid <Bu****@gmail.comwrote:

>So do you want to remove "&" or replace them with "&amp;" ? If you want
to replace it try the following;

I think he wants to replace them, but just the invalid ones. I.e.,

This & this &amp; that

would become

This &amp; this &amp; that
No, i don't know how to do this efficiently. =/...
I think some kind of regex could do it.

Yes, and the appropriate one is:

krefindamp = re.compile(r'&(?!(\w|#)+;)')
...
xmlsection = re.sub(krefindamp,'&amp;',xmlsection)

This will replace an '&' with '&amp' if the '&' isn't
immediately followed by some combination of letters, numbers,
and '#' ending with a ';' Admittedly this would let something
like '&xx#2;', which isn't a legal entity, through unmodified.

There's still a potential problem with unknown entities in the output XML, but
at least they're recognized as entities.

John Nagle

[Andreas Lysdal]

Duncan Booth skrev:

"Felipe Almeida Lessa" <fe**********@gmail.comwrote:

>On 26 Dec 2006 04:22:38 -0800, placid <Bu****@gmail.comwrote:

>>So do you want to remove "&" or replace them with "&amp;" ? If you
want to replace it try the following;

I think he wants to replace them, but just the invalid ones. I.e.,

This & this &amp; that

would become

This &amp; this &amp; that
No, i don't know how to do this efficiently. =/...
I think some kind of regex could do it.

Since he's asking for valid xml as output, it isn't sufficient just to
ignore entity definitions: HTML has a lot of named entities such as
&nbsp; but xml only has a very limited set of predefined named entities.
The safest technique is to convert them all to numeric escapes except
for the very limited set also guaranteed to be available in xml.

Try this:

from cgi import escape
import re
from htmlentitydefs import name2codepoint
name2codepoint = name2codepoint.copy()
name2codepoint['apos']=ord("'")

EntityPattern =
re.compile('&(?:#(\d+)|(?:#x([\da-fA-F]+))|([a-zA-Z]+));')

def decodeEntities(s, encoding='utf-8'):
def unescape(match):
code = match.group(1)
if code:
return unichr(int(code, 10))
else:
code = match.group(2)
if code:
return unichr(int(code, 16))
else:
return unichr(name2codepoint[match.group(3)])
return EntityPattern.sub(unescape, s)

>>>escape(

decodeEntities("This & this &amp; that&nbsp;&eacute;")).encode(
'ascii', 'xmlcharrefreplace')
'This &amp; this &amp; that é'
P.S. apos is handled specially as it isn't technically a
valid html entity (and Python doesn't include it in its entity
list), but it is an xml entity and recognised by many browsers so some
people might use it in html.

Hey i fund this site:
http://www.htmlhelp.com/reference/ht...s/symbols.html

I hope that its what you mean.

/Scripter47

[Duncan Booth]

Andreas Lysdal <an*****@riddergarn.dkwrote:

>P.S. apos is handled specially as it isn't technically a
valid html entity (and Python doesn't include it in its entity
list), but it is an xml entity and recognised by many browsers so some
people might use it in html.

Hey i fund this site:
http://www.htmlhelp.com/reference/ht...s/symbols.html

I hope that its what you mean.

Try
http://www.w3.org/TR/html4/sgml/entities.html#entities
for a more complete list.

[Frederic Rentsch]

John Nagle wrote:

Felipe Almeida Lessa wrote:

>On 26 Dec 2006 04:22:38 -0800, placid <Bu****@gmail.comwrote:

>>So do you want to remove "&" or replace them with "&amp;" ? If you want
to replace it try the following;

I think he wants to replace them, but just the invalid ones. I.e.,

This & this &amp; that

would become

This &amp; this &amp; that
No, i don't know how to do this efficiently. =/...
I think some kind of regex could do it.

Yes, and the appropriate one is:

krefindamp = re.compile(r'&(?!(\w|#)+;)')
...
xmlsection = re.sub(krefindamp,'&amp;',xmlsection)

This will replace an '&' with '&amp' if the '&' isn't
immediately followed by some combination of letters, numbers,
and '#' ending with a ';' Admittedly this would let something
like '&xx#2;', which isn't a legal entity, through unmodified.

There's still a potential problem with unknown entities in the output XML, but
at least they're recognized as entities.

John Nagle

Here's another idea:

>>s = '''<htmlhtm tag should not translate

& should be &amp;
&xx#2; isn't a legal entity and should translate
{ is a legal entity and should not translate

</html>

>>import SE # http://cheeseshop.python.org/pypi/SE/2.3
HTM_Escapes = SE.SE (definitions) # See definitions below the

dotted line

>>print HTM_Escapes (s)

<htmlhtm tag should not translate
&gt; &amp; should be &amp;
&gt; &amp;xx#2; isn&quot;t a legal entity and should translate
&gt; { is a legal entity and should not translate
</html>

Regards

Frederic
------------------------------------------------------------------------------
definitions = '''

# Do # Don't do
# " =&nbsp;" &nbsp;== # 32 20
(34)=&dquot; &dquot;== # 34 22
&=&amp; &amp;== # 38 26
'=&quot; &quot;== # 39 27
<=&lt; &lt;== # 60 3c

>=&gt; &gt;== # 62 3e

©=&copy; &copy;== # 169 a9
·=&middot; &middot;== # 183 b7
»=&raquo; &raquo;== # 187 bb
À=&Agrave; &Agrave;== # 192 c0
Á=&Aacute; &Aacute;== # 193 c1
Â=&Acirc; &Acirc;== # 194 c2
Ã=&Atilde; &Atilde;== # 195 c3
Ä=&Auml; &Auml;== # 196 c4
Å=&Aring; &Aring;== # 197 c5
Æ=&AElig; &AElig;== # 198 c6
Ç=&Ccedil; &Ccedil;== # 199 c7
È=&Egrave; &Egrave;== # 200 c8
É=&Eacute; &Eacute;== # 201 c9
Ê=&Ecirc; &Ecirc;== # 202 ca
Ë=&Euml; &Euml;== # 203 cb
Ì=&Igrave; &Igrave;== # 204 cc
Í=&Iacute; &Iacute;== # 205 cd
Î=&Icirc; &Icirc;== # 206 ce
Ï=&Iuml; &Iuml;== # 207 cf
Ð=&Eth; &Eth;== # 208 d0
Ñ=&Ntilde; &Ntilde;== # 209 d1
Ò=&Ograve; &Ograve;== # 210 d2
Ó=&Oacute; &Oacute;== # 211 d3
Ô=&Ocirc; &Ocirc;== # 212 d4
Õ=&Otilde; &Otilde;== # 213 d5
Ö=&Ouml; &Ouml;== # 214 d6
Ø=&Oslash; &Oslash;== # 216 d8
Ù=&Ugrve; &Ugrve;== # 217 d9
Ú=&Uacute; &Uacute;== # 218 da
Û=&Ucirc; &Ucirc;== # 219 db
Ü=&Uuml; &Uuml;== # 220 dc
Ý=&Yacute; &Yacute;== # 221 dd
Þ=&Thorn; &Thorn;== # 222 de
ß=&szlig; &szlig;== # 223 df
à=&agrave; &agrave;== # 224 e0
á=&aacute; &aacute;== # 225 e1
â=&acirc; &acirc;== # 226 e2
ã=&atilde; &atilde;== # 227 e3
ä=&auml; &auml;== # 228 e4
å=&aring; &aring;== # 229 e5
æ=&aelig; &aelig;== # 230 e6
ç=&ccedil; &ccedil;== # 231 e7
è=&egrave; &egrave;== # 232 e8
é=&eacute; &eacute;== # 233 e9
ê=&ecirc; &ecirc;== # 234 ea
ë=&euml; &euml;== # 235 eb
ì=&igrave; &igrave;== # 236 ec
í=&iacute; &iacute;== # 237 ed
î=&icirc; &icirc;== # 238 ee
ï=&iuml; &iuml;== # 239 ef
ð=&eth; &eth;== # 240 f0
ñ=&ntilde; &ntilde;== # 241 f1
ò=&ograve; &ograve;== # 242 f2
ó=&oacute; &oacute;== # 243 f3
ô=&ocric; &ocric;== # 244 f4
õ=&otilde; &otilde;== # 245 f5
ö=&ouml; &ouml;== # 246 f6
ø=&oslash; &oslash;== # 248 f8
ù=&ugrave; &ugrave;== # 249 f9
ú=&uacute; &uacute;== # 250 fa
û=&ucirc; &ucirc;== # 251 fb
ü=&uuml; &uuml;== # 252 fc
ý=&yacute; &yacute;== # 253 fd
þ=&thorn; &thorn;== # 254 fe
(xff)=ÿ # 255 ff
&#== # All numeric codes
"~<(.|\n)*?>~==" # All HTM tags '''

If the ampersand is all you need to handle you can erase the others
in the first column. You need to keep the second column though, except
the last entry, because the tags don't need protection if '<' and
'>' in the first column are gone.
Definitions are easily edited and can be kept in text files.
The SE constructor accepts a file name instead of a definitions string:

>>HTM_Escapes = SE.SE ('definition_file_name')

-------------------------------------------------------------------