Parse bridge flag [TG-4115]

[thk123]

Based on #2526 ✅

tidy up removed to separate PR

First some tidy up - there are four different tables in the java bytecode spec that have different values but were all using the same macros in the parser. This is confusing (and led to some nonsense lines, reading ACC_PROTECTED on a class for example, which isn't legal. Instead pulled each table into an enum, linking to the docs. To allow bitwise & and name scoping, namespaces were required.

Then the actual work - parse the ACC_BRIDGE which is a flag for methods that the compiler generates to deal with type erasure.

I didn't add a method to code_typet as this is very Java specific. Is it worth introducing a java_code_typet for java method types?

[thk123]

TG PR: diffblue/test-gen#2034

[allredj]

Passed Diffblue compatibility checks (cbmc commit: 10ac97e).
Build URL: https://travis-ci.com/diffblue/test-gen/builds/78217964

[peterschrammel]

Using enum class should remove this weird namespace.

[thk123]

Unfortunately using enum class means it cannot be implicitly converted to int (to be used in the logical and later). I could replace the word namespace with class, but that feels weirder. @thomasspriggs suggested writing a method called read_flag, but as they would all be unique types I'd either have to write one per type, or template it and lose type safety (not too bad, but I prefer this option) - let me know what you'd prefer me to do.

[thomasspriggs]

How about doing away with the defines / enums and using this approach instead?

  struct flagst {
    bool is_public : 1;
    bool is_private : 1;
    bool is_protected : 1;
    bool is_static : 1;
    bool is_final : 1;
    void : 4;
    bool is_interface : 1;
    bool is_abstract : 1;
    void : 1;
    bool is_synthetic : 1;
    bool is_annotation : 1;
    bool is_enum : 1;
    void : 1;
  } flags = *reinterpret_cast<flagst*>(&access_flags);

  parsed_class.is_abstract = flags.is_abstract;
  parsed_class.is_enum = flags.is_enum; 
  parsed_class.is_final = flags.is_final; 
  parsed_class.is_interface = flags.is_interface;
  parsed_class.is_synthetic = flags.is_synthetic;
  parsed_class.is_annotation = flags.is_annotation;

[thk123]

@peterschrammel I really like Tom's suggestion above, if you'd like it I'll go ahead and change it for that with some unit tests to check we get each method correct.

[thk123]

@peterschrammel Going to pull this refactor out into a different PR to unblock this - but could you express an opinion on whether you like Toms suggested approach (I think I'd probably put the struct directly into parse_treet::classt to avoid the is_abstract=is_abstract duplication.

[allredj]

This PR failed Diffblue compatibility checks (cbmc commit: ac9bc2d).
Build URL: https://travis-ci.com/diffblue/test-gen/builds/78728751
Status will be re-evaluated on next push.
Please contact @peterschrammel, @thk123, or @allredj for support.

Common spurious failures:

• the cbmc commit has disappeared in the mean time (e.g. in a force-push)
• the author is not in the list of contributors (e.g. first-time contributors).

[thomasspriggs]

LGTM

[peterschrammel]

@thomasspriggs, needs a rebase.

[peterschrammel]

Replaced by #2582.