3.4 KiB
| isHidden | menupriority | kind | created_at | title | subtitle | author_name | author_uri | tags | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| false | 1 | article | 2010-05-19T22:20:34+02:00 | How to repair a cutted XML? | and how to do it without any parsor? | Yann Esposito | yannesposito.com |
|
For my main page, you can see, a list of my latest blog entry. And you have the first part of each article. To accomplish that, I needed to include the begining of the entry and to cut it somewhere. But now, I had to repair this cutted HTML.
Here is an example:
Introduction
The first paragraph
Another long paragraph
Après avoir coupé, j'obtiens :
Introduction
The first paragraph
Argh! In the middle of an <img> tag.
In fact, it is not as difficult as it should sound first. The secret is, you don't need to keep the complete tree structure to repair it, but only the list of not closed parents.
Given with our example, when we are after the first paragraph. we only have to close the div for class corps and the XML is repaired. Of course, when you cut inside a tag, you sould go back, as if you where just before it. Delete this tag and all is ok.
Then, all you have to do, is not remember all the XML tree, but only the heap containing your parents. Suppose we treat the complete first example, the stack will pass through the following state, in order:
[]
[div]
[div, div]
[div, div, p]
Introduction
[div, div]
[div]
[div, p]
The first paragraph
[div]
[div]
[div, p]
Another long paragraph
[div]
[]
The algorihm, is then really simple:
let res be the XML as a string ;
read res and each time you encouter a tag:
if it is an opening one:
push it to the stack
else if it is a closing one:
pop the stack.
remove any malformed/cutted tag in the end of res
for each tag in the stack, pop it, and write:
res = res + closed tag
return res
And res contain the repaired XML.
Finally, this is the code in ruby I use. The xml variable contain the cutted XML.
# repair cutted XML code by closing the tags
# work even if the XML is cut into a tag.
# example:
# transform ' toto hello ]*$/m,'')
depth-=1
depth.downto(0).each { |x| res<<= %{} }
res
end
I don't know if the code can help you, but the raisonning should definitively be known.