Similar issues in several other tests, including
bugreports_DelimiterOnLayout
name_BibliographyNameFormNeverShrinks
group_LegalWithAuthorDate
bugreports_EnvAndUrb

Also textcase_NoSpaceBeforeApostrophe has invalid xml:

            </label prefix=" (" suffix=")" form="short"/>

Is this a new problem @jgm?

I merged a PR yesterday that was intended to fix a regular expression, but perhaps it had unintended consequences.

Can you take a look at this please?

af561aa

And here's the PR:

#10

I don't know, I'm doing something new (fooling around with writing a lua citeproc), so it's not that something that was working before broke.

OK. I don't really know anything about this myself.

Maybe @fbennett can offer insight.

Not sure what is going on now!
I reset to current master HEAD.
Now when I run python processor.py --grind I get

Traceback (most recent call last):
  File "processor.py", line 421, in <module>
    params.refreshSource(force=True)
  File "processor.py", line 151, in refreshSource
    self.grindFile(hpath,filename,mp)
  File "processor.py", line 173, in grindFile
    test.parse()
  File "processor.py", line 238, in parse
    self.extract(element,required=True,is_json=False)
  File "processor.py", line 258, in extract
    raise ElementMissing(self.script,tag,self.testname)
__main__.ElementMissing: ('processor.py', 'MODE', 'ignore.txt')

and I notice that many of the .txt files are changed from their repository versions. (I wouldn't have expected this program to change the sources.)

Reverting the commit you mention doesn't affect this.

Am I doing something wrong?

If I revert to 4a549a6, then the script succeeds but again I see changes in the .txt files. (Maybe just line endings.)

The error noted above is still present with that version.

OK, that error I understand, and can fix. Give me a bit.

That error should be fixed.

I just ran the script and looked at the processor-tests/machines/name_BibliographyNameFormNeverShrinks.json. Note that I do NOT see the same output as you. I see a full CSL style for the "csl" value, with the root "style" element.

   "csl": "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n<style xmlns=\"http://purl.org/net/xbiblio/csl\" 

I'm running Python 3.8.3.

Also textcase_NoSpaceBeforeApostrophe has invalid xml:

            </label prefix=" (" suffix=")" form="short"/>

Again, on this I see what I expect: a full style; not the above.

@bdarcus -- pretty sure @jgm just pulled out that line, which is invalid XML. I've created a pull request, should be OK to merge, but I'm doing too many things at once so please take a quick look before merging.

Right; I clearly misread that one.

So @jgm - some of this is now fixed.

Great! That fixes all my xml parsing errors, thank you.

I'm still wondering why the grind script changes the source files, but that's a more minor issue.

Feel free to close this if you intend the changing .txt files.

I'll leave it open for a bit in case @adam3smith or @fbennett have anything to say on that.

But glad it's working now!

Since you're looking at the test-suite now, if you have any time, I'd welcome input on #16. I've just pushed a PR that addresses it, I think.

For citeproc-js development, I built a new test runner, citeproc-test-runner that can run in Travis-CI. It has some nice features, including a watch mode that provides dynamic updates in a terminal window as a style file is being edited, and configurable paths to the various inputs needed to run a processor. It's coded to run citeproc-js, but could probably be adapted for use with other processors.

@fbennett do you recommended we add that new test runner to this README?