Improvements to the debugging show method of PGML.

[drgrice1]

This was something that was useful for me while I was working on writing the CodeMirror 6 PGML parser. It just formats the things in the PGML stack that are hashes in a nicer way, actually showing the contents instead of something like HASH(0x5d3919403d78) as it previously did.

The script in the attached zip can be used to test this:
pgml-parse.zip

The script requires that the PG_ROOT environment variable be set to the PG root directory. Execute the script by running ./pgml-parse.pl -f path/to/problem.pg. The script works with any version of PG that has PGML, but you can see the difference in the output. The combine entries (which almost any problem will have) will be output as HASH(0x...) with the develop branch, and will show the hash in a "prettier" way with this pull request.

This does not affect normal PGML usage in problems at all.

Just cleaning up local branches.

[pstaabp]

A couple of things. First, I'm getting some uninitialized string errors when running this on problems. Doesn't see to affect the output.

Also, would it be useful to put the pgml-parse.pl script in the bin or make a bin/dev directory for this?

[drgrice1]

I am not seeing any warnings when using the script. Do you have the PG_ROOT environment variable set? If so, the warnings are probably actually coming from the problem you are testing. You are testing with a problem that has BEGIN_PGML/END_PGML in it right? You probably will get uninitialized string warnings from the script if not.

The script is not really that versatile and was rather quickly thrown together. I don't know if it is worth adding to the repository.

[pstaabp]

I do have PG_ROOT defined.

I'm seeing Use of uninitialized value in concatenation (.) or string at /opt/webwork/pg/lib/PGalias.pm line 34. for every problem. Maybe the pg_environment is not loading.

Other errors that I'm seeing are some of my problems like: /Contrib/Fitchburg/Algebra/intervals/interval_notation.pg and I'm getting the following:

Use of uninitialized value $string in substitution (s///) at (eval 233) line 1005.
Use of uninitialized value $string in substitution (s///) at (eval 233) line 1006.
Use of uninitialized value in concatenation (.) or string at (eval 233) line 996.
Use of uninitialized value $string in substitution (s///) at (eval 233) line 1005.
Use of uninitialized value $string in substitution (s///) at (eval 233) line 1006.
Use of uninitialized value in concatenation (.) or string at (eval 233) line 996.
Use of uninitialized value $string in substitution (s///) at (eval 233) line 1005.
Use of uninitialized value $string in substitution (s///) at (eval 233) line 1006.
Use of uninitialized value in concatenation (.) or string at (eval 233) line 996.

[drgrice1]

The PGalias.pm warnings are because of a change in the code with the security vulnerability fixes in #1219 that the script doesn't account for.

The other warnings are actually expected. What is happening is that the PG_restricted_eval is encountering variables that are not defined because the script doesn't do a full render of the problem. It only parses the PGML section. For my tests that I worked with for development of the PG CodeMirror 6 editor I actually implemented special handling for this to define the needed variables.

Here is a hacked up version of the script that will remove the warnings, although will be less accurate for the actual result.
pgml-parse.zip

Note that this script really is not the point of this pull request. The PGML show method has been there pretty much since PGML was created and this is just a minor tweak on its output. A script that utilizes it is something different entirely. @dpvc probably has scripts that do so as well.