FYI We have released version 1.12.4 this evening.

I triggered the unit tests in CI and some of them require an update. If you want any help, let me know. 🙂

FYI We have released version 1.12.4 this evening.

I triggered the unit tests in CI and some of them require an update. If you want any help, let me know. 🙂

I can test this out to see if it fixes some of the parse exceptions in our code base.

Thanks for this contribution. Instead of a boolean, we could support a list of regular expressions instead, for added granularity. What do you think?

I would be down to change this behavior. Are you thinking a single regex or multiple regexs? Would the idea be to match specific text inside of the ParseException error message?

ie.

Skipping function company::DEFINE_string(logs_directory, "", "Path to the directory."). Error reported from parser was: Expected ')', found ','  (at char 15), (line:1, col:16)

So we could add a regex that would be something like company::DEFINE to only match specific member_kind, member_symbol or arglist values?

FYI We have released version 1.12.4 this evening.
I triggered the unit tests in CI and some of them require an update. If you want any help, let me know. 🙂

I can test this out to see if it fixes some of the parse exceptions in our code base.

I still see parse exceptions in our code base. I think some are probably due to our usage of gflags as well as selectively including specific source for documentation. In either case, I still think having a mechanism to disable these prints would be helpful from our end.

@JasperCraeghs I updated with the regex approach. Here are my local repo test results:

Default for doxylink_parse_exception_ignore_regex_list

Skipping function company::DEFINE_bool(show, false, "Enable visualization"). Error reported from parser was: Expected ')', found ','  (at char 12), (line:1, col:13)
...
Skipping function company::anonymous_namespace{builder.cc}::kRange3D(8.f, 8.f, 3.f). Error reported from parser was: Expected ')', found '.'  (at char 2), (line:1, col:3)
...
Skipping function transform::getAngle(transform_pb2.Rotation2f rot). Error reported from parser was: Expected ')', found '.'  (at char 14), (line:1, col:15)
...

With the following doxylink_parse_exception_ignore_regex_list = [r".*DEFINE.*", r".*anonymous_namespace.*"]

Ignoring parsing exceptions using `re.compile('.*DEFINE.*|.*anonymous_namespace.*')`
...
Skipping function transform::getAngle(transform_pb2.Rotation2f rot). Error reported from parser was: Expected ')', found '.'  (at char 14), (line:1, col:15)
...

• I replaced print with report_warning so that it becomes a proper Sphinx warning that can be converted into an error with sphinx-build -W
• Combining the regex patterns with | without grouping (with parentheses) may lead to unexpected results, I think. I chose to evaluate each regex one by one. Python stores the compiled patterns in cache.
• I added a test case.
• I renamed the configuration variable doxylink_parse_exception_ignore_regex_list to doxylink_parse_error_ignore_regexes. I'll update the PR description to reflect our recent changes.

Please review my contribution.

• I replaced print with report_warning so that it becomes a proper Sphinx warning that can be converted into an error with sphinx-build -W
• Combining the regex patterns with | without grouping (with parentheses) may lead to unexpected results, I think. I chose to evaluate each regex one by one. Python stores the compiled patterns in cache.
• I added a test case.
• I renamed the configuration variable doxylink_parse_exception_ignore_regex_list to doxylink_parse_error_ignore_regexes. I'll update the PR description to reflect our recent changes.

Please review my contribution.

Thanks for updating this @JasperCraeghs, your changes look great 🎉!

When this merges do you have a plan for what the release timeline looks like?

Cheers!

Released as 1.13.0