Any ideas on a nice BSP view pretty printer?

Question

Hi there,
I am wondering whether somebody already tried to extend the SAP standard pretty printer by an own functionality which fits better for pages with mixed content, containing not only BSP-elements but also HTML-tags.
The standard pretty printer considers everything as plain text which is not a BSP-element and leaves it unaltered. Only BSP-elements are arranged according to their tree structure.
This produces ugly results for pages with mixed content.
1.) Requirement
A Pretty Printer for HTML-based BSP views should cover the following:
(a) HTML-Tags and BSP-elements together form the chain. As in following example:
<z:table table_id="testTab" binding="//test/gt_test" rows="5">
  <z:tableBody position="top">
    <colgroup>
      <col width="4%">
      <col width="96%">
    </colgroup>
  </z:tableBody>
  <z:column name="type" header="<%=zcl_bsp_util=>dtel_text_s('EDI_TTYP')%>"
            listPos="10"/>
  <z:column name="text" listPos="20"></z:column>
</z:table>
(b) Certain HTML formatting tags like I, B, etc. should not be added to the stack but be treated inline.
(c) HTML-attribute names should be converted to lower case, the attribute values should be quoted.
(d) Like for BSP-elements, there should be a threshold after which the attributes should appear in separate lines. Instead of a certain maximum "number of attributes", this could also be a maximum number of characters per line.
(e) Scripting Tags (<% ... %>) should start at a new line, if the content contains new-lines.
(f) Evaluation Scripting (<%= ... %>) should be inline, if they are contained within text nodes (and not within HTML oder BSP element nodes).
(g) The Pretty Printer should be parametrizable (with parameters like indent-depth,... ).
2.) Exits?
Of course, one could always use an external HTML editor and write an add-on for that editor which produces the desired result. Or one could take the view content into the clipboard, paste it into a plain text editor like UltraEdit or TextPad which allows external commands, apply an external, self-written pretty printer to it and then paste the result back into the View. But it would be easier to use the "Pretty Printer" button directly to get the result. Therefore the question arises at which point the standard Pretty Printer Functionality could be modified or extended.
I didn't find an exit. But the pretty-print itself is doubly-decoupled in the standard which minimizes the number of points where an extension has to be made:
2.a) First decoupling point: The Fcode handler method pretty_printer_ext( ) in cl_o2_page is calling
dynamically a method in cl_o2_co2_processor:
* call pretty printer
  create object pp type ('CL_O2_CO2_PROCESSOR').
  appl_str = me->pagekey-applname.
  page_str = me->pagekey-pagekey.
  try.
      CALL METHOD pp->('PRETTY_PRINT')
        EXPORTING
          indent              = 2
          ATTRIBUTE_NORMALIZE = 'X'
          ATTRIBUTE_THRESHOLD = 2
          BSP_UNFORMATTED     = source
          appl                = appl_str
          page                = page_str
        IMPORTING
          test_ok             = test_ok
          BSP_FORMATTED       = pp_source.
    catch cx_o2_co2_exception into o2ex.
2.b) Second decoupling point:
The method pretty_print() of class cl_o2_co2_processor (fancy names, by the way. It seems that many former chemistry students had joined the BSP team ),
there is a second delegation, to a local class, as follows:
  data pretty type ref to LCL_BSP_PRETTY_PRINT.
  pretty = LCL_BSP_PRETTY_PRINT=>getInstance( processor = me ).
The real logic for doing the Pretty Print, finally, is performed in this local class.
3.) Implementation.
Since, up to my knowledge, there is no HTML parser in the ABAP world,
the implementation could be based on existing HTML Parsers, like the Perl module Parser::HTML, or the Java package http://htmlparser.sourceforge.net/ . The pretty printer could then be called as external OS command.
Are there any ideas out there? Or is there somebody who already worked in this direction?
Regards,
Rüdiger

Answer 1

For those who are interested. I found out that the Dave Ragget's classic "HTML Tidy" (see <a href="http://tidy.sourceforge.net/">http://tidy.sourceforge.net/</a> )  is available in ABAP! There is a class CL_HTMLTIDY (at least here in my SAP_BASIS 700 system) which wraps calls of HTML Tidy by kernel modules.
The class is undocumented, but there are 2 OSS notes of this year for it, so it might well be that it is alive.
HTML Tidy can surely not directly be used for Pretty Printing BSP views, but for designing test cases or for doing things like HTML parsing, it might be a helpful tool.