Using regular expressions in Python, I am trying to remove all XML-type elements in a string, except those containing QUOTE, eg <QUOTE>, </QUOTE> or <QUOTE A="B"> should remain, but others such as <EXAMPLE> or <TEST A="B"> should be removed. I've created this, which replaces all elements but can't work out the not part:
re.sub(r'</?[\w= \-"]+>', '', s)
Any ideas anyone?
I believe a negative lookahead assertion will do what you want:
import re
regex = r'<(?!/?QUOTE\b)[^>]+>'
tests = [
'a plain old string',
'a string with <SOME> <XML TAGS="stuff">',
'a string with <QUOTE>, </QUOTE>, and <QUOTE with="data">',
'a string that has <QUOTEA> tags </QUOTEB>',
]
for i in tests:
result = re.sub(regex, '', i)
print('{}\n{}\n'.format(i, result))
EDIT: How it works
Lookahead assertions, as the name suggests, "look ahead" in the string being matched, but don't consume the characters they're matching. You can do positive ((?=...)) and negative ((?!...)) lookaheads. (There are also positive and negative lookbehind assertions.)
So, the regex shown matches < for the beginning of a tag, then does a negative lookahead for QUOTE with an optional / before it (/?) and a word boundary behind it (\b). If that's matched, the regex does not match, and that tag is ignored. If it's not matched, the regex goes on to eat one or more non-> characters, and the closing >. I guess you might want to have it eat any whitespace following the tag, too - I didn't do that.
/? must be out of the lookahead assertion
QUOTE and /QUOTE are ignored./ is matched by [^>] . I was wandering absent-mindedly among postsI'd first replace QUOTE with some weird symbol that doesn't appear in the text, like maybe ^:
s = re.sub(r'(</?)QUOTE','\1^',s)
Then get rid of the XML tags that don't contain your weird symbol:
s = re.sub(r'</?[\w= \-"]+>','',s)
Then put the QUOTEs back in:
s = re.sub(r'(</?)\^','\1QUOTE',s)
EDIT: You can always combine these into one line by composition:
s = re.sub(r'(</?)\^','\1QUOTE',re.sub(r'</?[\w= \-"]+>','',re.sub(r'(</?)QUOTE','\1^',s)))
rmalouf's approach should work.
Here is a potential one-liner.
re.sub(r'<[/]?[^Q][^U][^O][^T][^E][^>]*>', '', s)
[/]? should match the /, when it is present.
[^>]*> matches everything else inside the tag, and the tag closer.
If you are expecting no other tags that start with Q, you can shorten it further:
re.sub(r'<[/]?[^Q][^>]*>', '', s)
