Python non-greedy regex to clean xml

问题:

I have an 'xml file' file that has some unwanted characters in it

<data>
  <tag>blar </tag><tagTwo> bo </tagTwo>
  some extra 
  characters not enclosed that I want to remove
  <anothertag>bbb</anothertag>
</data>

I thought the following non-greedy substitution would remove the characters that were not properly encased in <sometag></sometag>

re.sub("</([a-zA-Z]+)>.*?<","</\\1><",text)
            ^          ^ ^     ^      text is the xml txt.  
         remember tag, | |     put tag back without and reopen next tag
               read everything until the next '<' (non-gready) 

This regex seems only to find the position indicated with the [[]] in </tag>[[]]<tagTwo> What am I doing wrong?

EDIT: The motivation for this question has been solved (see comments, I had a stray & in the xml file which was causing it not to parse - it had nothing to do with the characters that I want to delete). However, I am still curious as to whether the regex is possible (and what was wrong with my attempt) and so I don't delete the question.

回答1:

The dot does not match newlines unless you specify the re.DOTALL flag.

re.sub("</([a-zA-Z]+)>.*?<","</\\1><",text, flags=re.DOTALL)

should work fine. (If it does not, my python is at fault, not the regex. Please correct.)

I think it is good practise to be as precise as possible when defining character classes that are to be repeated. This helps to prevent catastrophic backtracking. Therefore, I'd use [^<]* instead of .*? with the added bonus that it now finds stray characters after the last tag. This would not need the re.DOTALL flag any longer, since [^<] does match newlines.

回答2:

 "</[^>]+?>[^<>]+?<" 

in ipython:

In [1]: a="<data>  <tag>blar </tag><tagTwo> bo </tagTwo>  some extra   characters not enclosed that I want to remove  <anothertag>bbb</anothertag></data>"

In [2]: import re

In [3]: re.sub( "(</[^>]+?>)[^<>]+?<" ,"\\1<",a)
Out[3]: '<data>  <tag>blar </tag><tagTwo> bo </tagTwo><anothertag>bbb</anothertag></data>'