BBCodeParser (another CVS commit)

I've spent quite a bit more time on BBCodeParser. One thing kind of led to the next. Please take a look if you can, because I'd expect there to be new bugs in a rewrite of this size. (I'd call it version 2.0-beta) Link: http://pricepages.org/bbcode/BBCodeParser.zip * The unit tests have been updated with more bugs and closed bugs. I rearranged the tests so all of the tests that deal with broken code should be in separate functions. If the Text_Wiki_BBCode folks (hi bertrand!) are bored, they can run the unit tests with said function commented out to check for bugs in parsing/rendering. Note that there will be small differences that mess with the tests, such as attribute ordering. * More currently open bugs that have been fixed: 5609 4844 3447 1979 373 2580 3775 * I rewrote the _buildTagArray() function. It had been using half of the execution time in my tests. A TODO: in the comments said "rewrite this function". It is now almost 2x faster than the original function in my informal tests. * I rewrote _validateTagArray() in an effort to make it more useful. It would be easier to read, too, but there is quite a bit more going on there. But commenting lines probably equal code lines. * Output should _always_ be XHTML 1.1 compatible. * Per discussion on the Developer list, you can now pass a flag that determines the action on an error or warning during parsing. Actions include: correcting the problem, aborting parsing, ignoring the invalid tag, and deleting the invalid tag. Examples: If you only accept valid input, set both warn and error to 'abort' and parsing will be aborted as soon as a problem is found. If you accept invalid feedback, but want to give the user feedback which tags caused problems, then set both error and warn to 'ignore' and the bad tags will be displayed. If you want to make the output as pretty as possible, then you want to auto correct when you can and delete when you can't. Set the options to delete on error, and correct on warn. (There are twelve combinations to fit various situations.) * I've attempted to maximize BC, but here are the only ways in which BC is broken (to my knowledge): Custom list numbering is now ignored for XHTML 1.1 reasons. I reverted back to single quotes by default. HTML is now automatically escaped to fix a few "bugs" (I believe that Text_Wiki also automatically escapes HTML). Old filters are incompatible with the new filters (the 'allowed' tag variable has been replaced with a more powerful set of variables and they use a different format). ~Seth