In preparation for upgrading ts-jest, fix a couple of issues where
possibly undefined values really need to be verified *not* to be
undefined, and avoid unused variables, too.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

Suggested by `npm audit`.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

This comes in *real* handy for debugging and developing.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

It is nice and dandy if the commit-to-mail and mail-to-commit notes are
reliable, but when they are not (and they seem not to, for some reason
there was an uptick of unmatched commits recently), it becomes less nice
and less dandy.

In preparation for spending a bit more effort to identify upstream
commits, allow specifying the upstream commit explicitly.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

We just spent a lot of time trying to make sure that we find the
upstream commits corresponding to the PRs' original commits. Let's use
that information, too.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

This offers a much more robust (and much slower) way to detect which
upstream commits correspond to the original commits in the PRs.

That robustness is needed, though, as there are quite a few open PRs
that do not have the correct labels because no upstream branch is
associated with them, even if those branches exist in many cases.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

... because they hurt the reading flow.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

It declares its fields as readonly, we need them mutable (see the
next commits).

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

In the next commit, we will add some optional stdout processing that
won't work inside an async function because it needs access to the
`reject()` function of the returned Promise.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

With this option, we allow callers to specify a callback that handles
every line of the input individually.

This will come in handy when we will start processing the output of
`git log -p` in public-inbox/git, as we will be able to handle each
commit individually, together with its diff, from which we can extract
the corresponding mbox.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

We will introduce more of the same kind shortly, needing the ID of the
generated comment, so it makes sense to have all of them consistent.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

This is a bit tricky because GitHub wants a path and a position, too.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

This is super easy ;-)

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

It really is a number, not a string.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

It is all nice and dandy to submit patch series from PRs, but you cannot
really see the reviewers' comments in the PRs, making the entire
exercise a little less fun.

Let's do something about it.

This new class only helps with processing the mails, but it is not
hooked up with any command-line interface yet, and it also does not
store the state correctly yet, e.g. "what was the latest mail we
processed?" (to allow for efficient incremental processing).

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

This is what will be used in an Azure Pipeline to mirror the mails to
the PRs.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>