Re: [libxml++] HTML Parser Subclass (based on domparser.cc)

[Christophe de VIENNE <cdevienne alphacent com>]

Hi Laurent,

This looks interesting, I'll have a closer look on it a bit later.
However the integration into libxml++ will not occur in 2.6, since we 
froze the API on monday.
We'll soon discuss the start of a new unstable branch, in which the 
HtmlParser should probably have its place.

Thanks,

Christophe


Laurent Hoss a écrit :

> Hi all,
>
> I discovered the cool libxml++ yesterday on my quest for the best C++ 
> XML Parser (bindings, coz libxml2 seems to be the best C parser anyway 
> ;). Libxml is not new to me though, I used it extensively in Perl 
> thanx to the very complete XML::LibXML CPAN Module.
> Now one of my main motivations is to parse HTML Files into a DOM tree 
> where I can extract nodes with XPATH.
> In perl that was easy , it has the html parser included.
> Therefore after a thorough search in the API I was a bit disappointed 
> that there was no HTML Parser support in libxml++...
> but thanks to the clean API's of  libxml(++)  and after a little 
> reading , I had no difficulties at all building my own subclass (based 
> on domparser.cc) except some little quirks (like extra encoding 
> parameter in some html parser functions) :)
>
> In fact libxml2 has a really tolerant html parser (I used it in perl 
> for mirroring/parsing whole dynamic websites :D ), it even returns a 
> good XML Document when it had parser Errors, but to get a Doc returned 
> in such a case one has to turn off the 'wellformedness' check, which I 
> did in my temporary htmlparser Implementation.
> ( Unfort. there's always a segfault at the end of a run of my edited 
> 'dom_xpath/main.cc' html parsing example app , when ignoring 
> '!context_->wellFormed'  ?!  experimenting done in 
> 'HtmlParser::parse_context' method )
>
> I hope HTML Parsing can be included in the main distr. ( maybe better 
> with wellFormed check on )...
> To compile the whole library with my htmlparser class, I added the 
> class in all the files (Makefile.am files, libxml++.h...) containing 
> 'domparser'.
>
> Included are the c++ and include files of htmlparser class (or should 
> I've taken diffs from the domparser.cc/h originals ?) plus my html 
> parsing example, which shows all the //a[ href] links with their 
> attribute contents.
>
> Hopefully the segfault can be easily solved with the knowledge of the 
> lead developpers ( I don't have yet ;).
> I guess its just something I'm missing, else I'll try to find the 
> mem.leak using a debugger (or is there a better way ??)
>
> Thanx,
> Laurent