`pipenv check` warns about CVE-2019-7653:

The CLI tools in RDFLib 4.2.2 can load Python modul...

Also add a unit test to verify extraction of the $-line.

The other option here is to use a <mil...

My editor started warning about comment strings which weren't
explicitly docstrings, so convert ...

Verify common state annotations propagate into the tei model
classes as notes. Unforunately seve...

Write out $ lines as TEI <note> elements, similar to what oracc
does, although they set addition...

Now that our output is good enough to avoid crashing HookTest,
add it as a development dependenc...

Instead of generating a ti:edition tag directly from
the Work object metadata, add CTS propertie...

It makes sense from the point of view of cuneiform scholarship
to consider each tranliteration, ...

Don't include the object label (tablet, seal, etc.) in the declared
reference scheme. For the va...

We were always appending a translation part, just in case the
parser found a parallel translatio...

This module is used in the convert function, but was only imported
by main, preventing atf2cts f...

This was useful for testing, but sometimes the document itself
raises and exception when it's se...

Passing the line as a separate arguments already inserts a space
so we don't need to do so manua...

These were left over from when we kept copies of all the processed
conversions.

This wasn't updated properly after convert's return type changed.

Add tei2atf script

Verify round-trip parseablitily of the SIL-034.atf example file.

Verify some features convertin...

Explicitly check the idno element release against None. For some
reason the Element object on it...

Serialize all the text inside each line, not just the immediate
element text. Also check for lin...

Reverse the transformation, converting TEI XML into ATF.

- Works well enough to restore the SI...

Apply double-P-number urn scheme

Section objects may have an optional name, `Surface a` and so on.
We use the object name as the ...

These docstrings and sketch code are no longer necessary.
Setting the urn and encodingDesc have ...

Stop creating a test umbrella text group and let the converter
fall back to generating a textgro...

For testing we've been placing all the converted documents in
a `test` textgroup. That's fine wh...

This allows better control over the formatting of the output.

We don't need three copies of this code; it's clear enough from
the function name what's happening.

Recommendations are for the edition urn to have a component
reflecting the project producing the...

The cts serialization objects are about CTS elements.

Rather than asserting, this makes it more clear what's going on
if atf2cts is called on a non-at...

We're no longer writing these into document comments, and it's
easy to miss them as the number o...

This comment string wasn't updated when I changed the name of
the loop variable so we were asser...

Improve CTS compliance

Use serialization classes to write out tei xml.

We now pass the capitains_units.cts tests except for
Duplicate passages. This is perhaps confuse...

Epidoc seems a better fit, and is what Perseus is moving to.

I don't think this is what we eventually want, but it restores
previous behaviour and matches th...

The final element of the line pattern needs to look for an <l> not
a <div> so constructing them ...

CTS guidelines specify the header should contain patterns for
constructing xpath pointers based ...

In atf there are marked by a trailing # on the sign, but in
presentation the sign should be in h...

This doesn't survive the new xml generation and needs to be
replaced code to generate an Element...

Serialize CTS metadata objects the same way we do TEI objects.
Note this changes the indent on _...

The default 8-space tabs are much too strong an indent for the
nesting of a typical tei document.

Create a class for the shared __str__ implementation so the way
xml is serialized is all in one ...

This brings the output back to near where it was before replacing
template-based xml generation ...

Set these in a more logical order.

Although generic TextPart objects don't have a language, this
lets us inherit the code in both E...

This is general tei style, not epidoc, but add it now for complete
ness. Attributes still need t...

The parser gets this from the ATF and atf2cts needs it to construct
proper metadata files.

Instead of the fancy class, just return a flag tuple. Since boolean
values are numbers these can...

The convert function spawned in parallel for handle each atf
block individually was still return...

Now that atf2tei.convert returns a tei.Document instead of an
already-serialized string, we need...

This is less confusing than remembering it's document.parts.append
vs textpart.children.append.

Simplifies the caller code since in epidoc we're pretty much
always setting this reference attri...

Support the same header information we include in the current
atf2tei template.

It would be nic...

These are single-line strings, so we don't need multi-line quotes.
I was trying to make it unifo...

This tag can't be used directly in a <sourceDesc> and needs to
be wrapped in something like a <b...

Convert documents in parallel

I added this to distinguish the progress message from other
debugging output, but there's less n...

Move the failure/success report to the end, after all atf files
given on the command line have b...

It turns out we successfully convert around 95% of the current
cdli atf blocks, so the bar didn'...

I only needed this to benchmark, and the default (the number
of cores on the machine) works fine.

Dispatch the atf conversion work through ProcessPoolExecutor,
which speeds things up by about 4x...

Update travis badge to use cdli-gh fork.

Verify tei attribute serialization.

Now that upstream is the cdli-gh repository, its status is the
more appropriate one to advertise.

Fix issues with the xml serialization class tests.

A Document sets an xml namespace attribute, ...

Cleans up a bit and makes it easier to experiment with the harness.

Hooktest requires there be a language marker. Default to English
since we don't have any languag...

Keep a list of failures and report the success ratio. I just wanted
to see how far along I was o...

Another possible punctutation sign was missing from the regex.

The regex was missing some possible trailing characters.
See for example P005984 and P292940.

Add some classes to model the major pieces which can create and
ElementTree representation for s...

These are part of the Capitains Guidelines. Just default to
English for lack of a better option,...

The _ atf markup sometimes spans multiple words, which causes
nesting problems if words are also...

Fix pycodestyle lint.

Better aligns with markup from the oracc adart conversions.

This aligns better with the oracc adart encodings.

Previously we were generating sequential line numbers so the tei
labels were regular. But since ...

This is working reliably now.

Currently pyoracc doesn't parse interlinear translations, so the
Translation object only shows u...

The triple-quote is redundant with the indent, and we don't
really need syntax highlighting, so ...

Reduces clutter in the top-level directory.

We seem to need the dummy __init__.py to make impor...

Use cts.TextGroup and cts.Work to construct and write out the
corresponding metadata files.

Not...

This encapsulates the data needed for the Canonical Text
Services file hierarchy and makes it si...

Properly handle logogram markup next to punctuation.
Issues found with P464358 Codex Hammurapi c...

This matches the lang suffix on the edition urn in the __cts__.xml
metadata file. I think this i...

The title can of course contain characters like '&'.

Calling islice is an unnecessary step here.

Verify multiple copies segment into the correct number of atf blocks.
This is redundant with `te...

Avoid repeating the name in each test.

Just verify that something comes out.

Wrapper around atf2tei which segments a multi-object atf file,
converts each segment to xml and ...

The pyoracc parser also takes a string, but this simplifies
passing in something from an externa...