bug: GitHub `query` and `involved_issues` don't work with fine-grained access tokens

[muni-corn]

The current documentation for bugwarrior says:

If you want to write your own github query, as described at https://help.github.com/articles/searching-issues/:

query = "assignee:octocat is:open"

Note that this search covers both issues and pull requests, which github treats as a special kind of issue.

...but it seems that GitHub has changed its mind on this.

If I set:

involved_issues = true

# or
query = "involves:muni-corn archived:false"

then I get this error on bugwarrior pull:

ERROR:bugwarrior.collect:Worker for [github] failed: Non-200 status code 422; 'https://api.github.com/search/issues?q=involves:muni-corn%20state:open&per_page=100'; '{"message":"Query must include \'is:issue\' or \'is:pull-request\'","documentation_url":"https://docs.github.com/rest/search/search#search-issues-and-pull-requests","status":"422"}'
Traceback (most recent call last):
  File "/nix/store/d7pmp2f48nx8q56mwfzqqzib4rfgaic8-python3-3.13.12-env/lib/python3.13/site-packages/bugwarrior/collect.py", line 47, in _aggregate_issues
    for issue in service.issues():
                 ~~~~~~~~~~~~~~^^
  File "/nix/store/d7pmp2f48nx8q56mwfzqqzib4rfgaic8-python3-3.13.12-env/lib/python3.13/site-packages/bugwarrior/services/github.py", line 435, in issues
    self.get_query(
    ~~~~~~~~~~~~~~^
        'involves:{user} state:open'.format(user=self.config.username)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    )
    ^
  File "/nix/store/d7pmp2f48nx8q56mwfzqqzib4rfgaic8-python3-3.13.12-env/lib/python3.13/site-packages/bugwarrior/services/github.py", line 311, in get_query
    for issue in self.client.get_query(query):
                 ~~~~~~~~~~~~~~~~~~~~~^^^^^^^
  File "/nix/store/d7pmp2f48nx8q56mwfzqqzib4rfgaic8-python3-3.13.12-env/lib/python3.13/site-packages/bugwarrior/services/github.py", line 120, in get_query
    return self._getter(url, subkey='items')
           ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^
  File "/nix/store/d7pmp2f48nx8q56mwfzqqzib4rfgaic8-python3-3.13.12-env/lib/python3.13/site-packages/bugwarrior/services/github.py", line 166, in _getter
    json_res = self.json_response(response)
  File "/nix/store/d7pmp2f48nx8q56mwfzqqzib4rfgaic8-python3-3.13.12-env/lib/python3.13/site-packages/bugwarrior/services/__init__.py", line 379, in json_response
    raise OSError(
    ...<2 lines>...
    )
OSError: Non-200 status code 422; 'https://api.github.com/search/issues?q=involves:muni-corn%20state:open&per_page=100'; '{"message":"Query must include \'is:issue\' or \'is:pull-request\'","documentation_url":"https://docs.github.com/rest/search/search#search-issues-and-pull-requests","status":"422"}'
INFO:bugwarrior.collect:Done with [github] in 0.5002522468566895.
ERROR:bugwarrior.collect:Aborted [github] due to critical error.

[muni-corn]

oh. I visited the URL in GitHub's response and it says this:

For requests made by GitHub Apps with a user access token, you can't retrieve a combination of issues and pull requests in a single query. Requests that don't include the is:issue or is:pull-request qualifier will receive an HTTP 422 Unprocessable Entity response. To get results for both issues and pull requests, you must send separate queries for issues and pull requests. For more information about the is qualifier, see "Searching only issues or pull requests."

[muni-corn]

Okay, I did some more keyboard mashing and I've figured it out!

I was getting this error using GitHub's Fine-grained access tokens. While I wouldn't get errors by using separate configurations for PRs and Issues in bugwarrior, it seemed that bugwarrior only wanted to keep the PRs and tossed the issues. I was using the same access token for both, so I don't know if that's why bugwarrior would only keep tasks imported from one of the configurations.

I tried switching to a "classic" GitHub personal access token, i.e. ghp_aaaaaaa..., and now all tasks are imported just fine with the query I've specified.

In order for bugwarrior to properly support the fine-grained access tokens, we will likely have to split queries into separate queries: one for issues and one for pull requests.

[ryneeverett]

While I wouldn't get errors by using separate configurations for PRs and Issues in bugwarrior, it seemed that bugwarrior only wanted to keep the PRs and tossed the issues. I was using the same access token for both, so I don't know if that's why bugwarrior would only keep tasks imported from one of the configurations.

If I understand you correctly, when you switched to using two separate configuration sections in your configuration file, bugwarrior closed the issues which had previously been open tasks? Are you sure your separate configuration successfully pulled the issues? Did you forget to put the issue section in your general>targets by chance?

In order for bugwarrior to properly support the fine-grained access tokens, we will likely have to split queries into separate queries: one for issues and one for pull requests.

To summarize, this means that for every api endpoint we're using which returns issues or pull requests (/search/issues, /repos/{username}/{repo}/issues, and /issues) we need to make two separate calls with is:issue and is:pull-request respectively and aggregate the results. For /search/issues we need to be sure only to do this if the query does not already contain one of these filters.