Form2WSDL Project Roadmap

Itinerary

  1. Run as a Perl script, output some sample data to test the setup. 11/10/2004
  2. Fetch the page inputted by the user using LWP::UserAgent. Put this HTML into a variable and output it. 27/10/2004
  3. Pipe this code through an external program (most likely HTML Tidy) to ensure no validation errors. 4/11/2004
  4. Split the page into discrete components, hopefully through a pre-existing Perl module. Dump everything outside form elements. Maybe store metadata about the fetched page for the report at the end. 15/11/2004
  5. Load the form elements into an array.Take each one and make the necessary arrangements to convert the text and hidden elements into “equivalent” WSDL descriptions. 20/02/2005
  6. Test. Then go through the same process for the more difficult form elements — select and radio. 26/02/2005
  7. Write some test cases, package these with the program.
  8. Develop regex algorithms for guessing at useful parameter types.
  9. Large scale testing on real world examples.

Current Outstanding Issues

  1. HTML::Form doesn’t extract the name or id of a form, which we could use in preference to referring to operations as ‘FormNameN’ We’ll have to do some full HTML parsing to get at these.
  2. Operations usually have names that are readable to humans, but HTML forms don’t have any text that would seem to fit.
  3. A form may have two submit buttons, with name attributes to differentiate them. These should be treated as two separate instances of the form.

Icing