merge of three branches

Very likely unstable DO NO USE

create branch 'DRUPAL-5--2'.

Sprout from master 2008-06-27 19:46:27 UTC Dan Morrison <dan@coders.co.nz> 'Merging DRUPAL-5--1 back into HEAD'
Delete:
    drunkard2drupal.xslt
    install-htmltidy.inc

I hope I'm doing this right

Doxygen comments: don't use @mainpage (this is reserved for an index page for *all* contrib modules), but @defgroup/@ingroup grouping. and don't use @package: this is reserved for java packages [1].

[1] http://www.stack.nl/~dimitri/doxygen/commands.html#cmdpackage

eg/url?DEBUGLEVEL=3
- code was there before but wasn't actually working

http://drupal.org/node/111740
untested, and applying to a 4-month old release.
Testing and re-merging to do...

This has been (still is) mostly undocumented, but if your source page has
<meta name='tag' content='widget'/>
or <a rel='tag'>widget</a>
in it, before or after the XSL transform, the page will be tagged with the taxonomy term 'widget' ... if it exists in any vocab so far.

To support free-tagging was harder. I needed to define the taxonomy as well.
<a rel='tag'>Products:widget B</a>
will tag the page with the term 'widget B' in the vocab 'Products' ... on-the-fly if free-tagging is enabled.

Removed a bug re CCK imports

Will absorb top-level (unwrapped) text if found in the raw body import, previously only element nodes were being found.

renamed the classes on the tre-view to avoid conflict with some themes - class=page becomes class=page-item so global settings for 'pgae' do not get applied
added a MAX_TITLE_LENGTH config option in case the default (24 characters) wasn't long enough.

Import_html confused about file:/// paths when using realpath()
http://drupal.org/node/103221

import converts &lt; to < and &gt; to >
http://drupal.org/node/97532

Input can either be id="byline" or id="field_byline" to populate a CCK 'byline' field.

Also support for HTML/non-HTML input text areas

If your input contains ANY classes at all, the content of those classes
becomes available for absorbing into the node.
If the input contains any ID elements, those too are added to the node
object before saving.
most of these extra blocks do not get serialized, as there is no handler
for them
BUT

If you are importing to CCK, AND can manipulate your input so as to
define ids of the form field_* - this block will be put in that cck
field.

So...
if your input contains
...
[div id='footer'][b]blah[/b][/div]
...
and yo have a CCK node definition that includes a field called 'footer'
- you need to work on your XSL import massager to make a conversion to
[div id='field_footer'][b]blah[/b][/div]

Once that is, done, the import_html process can detect that this is a
ver special ID in the source document, and inject it into the CCK object
as
$node->field_footer[0][value] = "[b]blah[/b]";

and things become good

http://drupal.org/node/97835

http://drupal.org/node/97750

Fixed 'do NOT add to menu' checkbox
Repositioned settings summary
Moved some of the options from final import stage back to settings and
step 1. The final form was getting cluttered.

Nasty invalid code was giving nasty messages (log length limit didn't render the closing [pre])
4.7 doesn't allow me to clear form_set_errors any more. it was halting all further processing when one page
failed.

http://drupal.org/node/46775

Also reviewed the docs, and fixed a problem with the hidden files regexps.
Changed the auto-teaser to be safe HTML when importing, as some older-formatted sites (tables) broke things frequently.

Got htmltidy for it working again. Added error message capture and display
Reviewed the xml toString function.
Had to give up on getting PHP4 to include XSL sub-templates. Still dunno why it refuses.
Tuned the behaviours on first install a little