Re: [xml] two Patches regarding HTMLtree.c

[paul-libxml sponagl de]

Hi Daniel,

> > +    if (cur->type == HTML_PRESERVE_NODE) {
> > +       if (cur->content != NULL) {
> > +#ifndef XML_USE_BUFFER_CONTENT
> > +        xmlBufferWriteCHAR(buf, (const char *)cur->content);
> > +#else
> > +        xmlBufferWriteCHAR(buf, (const char *) xmlBufferContent(cur->content));
> > +#endif
> > +       }
> > +       return;
> > +    }
>
>   I assume it's to avoid doing escaping in scripts 

Scripts won't be processed 'preserved' without the patch, so
e.g. Javascripts will be processed as text to escape. the thing i did
was a diff between nodeDump and nodeDumpOutput.  (There are a view
more inconsistencies between those functions, but i had not the time
to track further on)

> >      /*
> >       * Get specific HTmL info for taht node.
> > @@ -565,17 +575,6 @@
> >     xmlBufferWriteCHAR(buf, cur->name);
> >     xmlBufferWriteChar(buf, ">");
> >      }
> > -#if 0
> > -    if (!htmlIsAutoClosed(doc, cur)) {
> > -   xmlBufferWriteChar(buf, "</");
> > -   xmlBufferWriteCHAR(buf, cur->name);
> > -   xmlBufferWriteChar(buf, ">");
> > -    }
>
>   Hum, I decided on purpose to retain this code around, any reason why
> to want to remove it ?

dumping to memory delivered _two_ closing tags

> > -#else
> > -    xmlBufferWriteChar(buf, "</");
> > -    xmlBufferWriteCHAR(buf, cur->name);
> > -    xmlBufferWriteChar(buf, ">");
> > -#endif
>
>   Can you explain the reason why closing tag should be removed ?
>
> >      if (cur->next != NULL) {
> >          if ((cur->next->type != HTML_TEXT_NODE) &&
> >         (cur->next->type != HTML_ENTITY_REF_NODE))
> > @@ -927,17 +926,9 @@
> >         (cur->children != cur->last))
> >         xmlOutputBufferWriteString(buf, "\n");
> >      }
> > -#if 0
> > -    if (!htmlIsAutoClosed(doc, cur)) {
> > -   xmlOutputBufferWriteString(buf, "</");
> > -   xmlOutputBufferWriteString(buf, (const char *)cur->name);
> > -   xmlOutputBufferWriteString(buf, ">");
> > -    }
> > -#else
> >      xmlOutputBufferWriteString(buf, "</");
> >      xmlOutputBufferWriteString(buf, (const char *)cur->name);
> >      xmlOutputBufferWriteString(buf, ">");
> > -#endif
>
>
>   Again this removes uncompiled code left around on purpose, but this
> won't fix anything.
>
> >      if (cur->next != NULL) {
> >          if ((cur->next->type != HTML_TEXT_NODE) &&
> >         (cur->next->type != HTML_ENTITY_REF_NODE))
>
>   Please send some explanations, I will then look at integrating your
> patch.

I had the problem that the #if 0 was ignored by gcc at this time.
(Please don't ask why - i'm still wondering - the preparsing-output
showed that it's still in the code) - so i decided to remove all the
stuff - Maybe the bad weather that day ;)

bye
Paul