Output a python-escaped version of the field values so it's clear
what triggered the warning. Of...

Python caches compiled code in this this directory. We don't want
this tracked in the source cod...

Mark commits with style nits as not passing.

This only reflects the status of the current master branch on
github, not any local checkout whe...

The continuous integration system at travis-ci.org offers some
docker support. Try instantiating...

This is the current release. Seems to work fine.

Parse command-line arguments when invoked as an executable.
Require the path to the cdli data ex...

When an extra column is present in a row, csv.DictReader returns
it as a list under the `None` k...

Reformat the check_columns loop to fit in 80 characters.
Less nesting can be easier to read and ...

Move the DictReader instance to a helper method so it can be
accessed directly by the checks whi...

Both check.py and upload.py need to open and part the catalogue
csv files. Move this and some ot...

Allow invoking as `pipenv run python upload.py /path/to/data` so
it's possible to use an existin...

This is mostly to reduce the length of the output when running in
automation.

Matches what the travis test expects.

Unfortunately there are over a thousand of the non-printable character
corruptions. Many are tab...

The upload script needs to be invoked with python explicitly.

Nicer than having to paste the url, assuming the browser will
follow a localhost link.

Don't make people look it up.

Give a basic quickstart introduction to help people try out
the demo.

Specify interaction standards for publication. Copied from
contributor-covenant.org by Coraline ...

Set a license file for publication.

invoke with `pipenv run python check.py`.

This is useful for testing when we to start with a fresh upload.

Set or update an generic 'cdli-catalogue' index alias when the
upload is complete and use that i...

Complain instead of silently altering the catalogue data in
the unlikely possibility that column...

Wrap read_catalogue in another function which adds metadata keys
to the row dictionary. This let...

This is much faster than the previous document-at-a-time indexing
upload, but still takes severa...

These are no longer necessary now that it's working.

Make div#results hide itself properly, and remove the obsolete
placeholder data.

Correctly generate a list of results. It turns out `this` isn't
bound in then functions chained ...

This ensures queries are sent to the api instance without having
to pass the correct url into th...

Use starlette's StaticFiles class to serve anything under the
vue directory.

Vue.js search page to make testing the api a little nicer.

I think this is the same as the _search?q=foo REST api.

I tried using pydanic.Schema to add the same documentation to
the paging parameters skip and lim...

Query some common fields. Finds 'Ishtar descent' but not 'K 162',
so not really doing better yet...

Fetches the catalogue data for the given CDLI id (the numerical
part of the P-number).

FastAPI-based stub for querying the index and returning search
results.

Invoke with:

pipen...

Simplest possible indexing with no tuned weights for different fields.

Generate an index name b...

This is the Python package wrapping the elasticsearch REST api.

Invoke with

pipenv install...

This is only relavent when invoked in my local config, where
a checkout of https://github.com/cd...

Build a reader to logically concatenate the two file segments,
read and parse each row, and prin...