[John Cowan]
>In particular: TagSoup, which parses arbitrary HTML, has a PYX output mode
>of just such a form. PYX differs trivially from ESIS in that in ESIS,
>attribute lines precede the start-tag line, whereas in PYX they follow.
>I am not sure why Sean McGrath made this change, but I have followed him.
I made the change in order to make it easier to process the data with a top-down, recursive descent parser approach.
I.e.
peek at next start_tag
call handler for the right start_tag
That way, start_tag handlers could handle all the attribute processing, rather than the main
parsing loop having to stack up attributes for some yet-to-be determined element type.
(Hey! This is my first post to xml-dev in.....years :-)
Regards,
Sean
