BPO "GH-5195" MouseOver ToolTip text References "GH-15194"

[sweeneyde]

At https://bugs.python.org/msg397968, I had typed GH-5194, which correctly links to python/cpython#5194.

However, the appearance of that link is suspect for two reasons:

1. When I mouse over, the tooltip text reads GitHub PR 5194: [open] bpo-21161: pdb: default: handle list comprehensions, as if to link to bpo-21161: pdb: default: handle list comprehensions cpython#15194. Note the difference of a leading 1: 15194 instead of 5194.
2. There is no strikethrough on that link, even though that pull request is merged.

This is not a major issue, but it's just a little confusing that the link is correct when the CSS implies otherwise.

[sweeneyde]

Relevant code is here.

    def _linkify_pull_request(self, match):
        """Turn a pullrequest (e.g. 'PR 123') to an HTML link"""
        template = ('<a href="%(base_url)s%(pr_no)s" %(cls)s'
                    'title="GitHub PR %(pr_no)s%(info)s">%(text)s</a>')
        pr_no = match.group('pr_no')
        text = match.group('text')
        # find title and status
        cl = self._db.pull_request
        # find all the pull_request that refer to GitHub PR pr_no,
        # with the most recently updated first
        pr_ids = cl.filter(None, dict(number=pr_no), sort=[('-', 'activity')])
        title = status = info = cls = ''
        for pr_id in pr_ids:
            if not title:
                title = cl.get(pr_id, 'title', '')
            if not status:
                status = cl.get(pr_id, 'status', '')
            if title and status:
                # once we get both, escape and add to info
                status = cgi.escape(status).replace('"', "'")
                title = cgi.escape(title).replace('"', "'")
                info = ': [%s] %s' % (status, title)
                break
        if status:
            cls = 'class="%s" ' % ('open' if status == 'open' else 'closed')
        base_url = 'https://github.com/python/cpython/pull/'
        return template % dict(base_url=base_url, pr_no=pr_no, cls=cls,
                               info=info, text=text)

In my case, this must have returned

<a href="https://github.com/python/cpython/pull/5194" class="open" title="GitHub PR 5194: [open] bpo-21161: pdb: default: handle list comprehensions">GH-5194</a>

There is nothing identifiably wrong with that HTML snippet, it's just that the wrong PR metadata was pulled.
I think there could be a rule to prioritize exact matches of number=pr_no rather than treating all substring-matches as equally valid.

This isn't a huge problem as it only occurs when referencing pull requests old enough to be four digits long.