rsundqvist / rics Goto Github PK

read_path_format = "${TEST_ROOT:tests}/imdb/{}.json"

Should match ${TEST_ROOT:tests}, but globs until the last closing bracket.

Should errors really be raised unconditionally for unknown IDs?

Remove

• ignore::pytest.PytestUnraisableExceptionWarning
• module:Jupyter is migrating its paths to use standard platformdirs:DeprecationWarning

Running

Translator(hex_fetcher, fmt="{hex}[, positive={positive}]").translate(1, names="positive_numbers")

causes exception ValueError: tuple.index(x): x not in tuple at placeholder_translations.id_pos = placeholder_translations.placeholders.index("id")

Something like

translate-ids data.csv [options]

Values and candidates currently share the same type, which is not required. I think.

Allow fetching using multiple fetchers.

Currently a dict of lists, which isn't really what is needed anywhere. SQLAlchemy natively returns an iterable of tuples which seems like a good option.

Translations defaults can only be specified per-source, even though they are often identical between sources.

When fixed, update https://rics.readthedocs.io/en/latest/translation-api.html#default-translations

No longer maintained. Replace with ruff (also replaced black and isort).

Images of tables and/or figures with/without translations

https://rics.readthedocs.io/en/latest/translation-quickstart.html#id-translation

Need to:

• Add flag for whether fetching all IDs is allowed (based on table size, configurable)
• Add logic to switch between select *, BETWEEN and IS IN in fetch_placeholders (always fetches everything now)
• Fix selecting only required placeholders (columns) in fetch_placeholders

This class is too specialized. The API has become pretty messy trying to de-specialize it.

Add translation primer to docs. Like the one for mapping.

Maybe https://pypi.org/project/flake8-unused-arguments?

Add source filtering for the fetcher; avoid certain SQL tables etc (without a blacklist).

Add support for translating keys rather than values in dict/series.

Requiring inplace=False is probably fine, if that makes it a lot easier.

⚠️ 5935cc6 sets xfail(strict=False).

Escaping doesn't work as expected.

>>> fmt = Format("{{{id}}}[, the value '{optional}' which was [optional] is here as well!]")

>>> fmt.fstring().format(id="my-id")
'{my-id} is here as well!]' 

Expected output: '{my-id}'

>>> fmt.fstring(["id", "optional"]).format(id="my-id", optional="my-optional")
"{my-id}, the value 'my-optional' which was [optional is here as well!]"

Expected output: "{my-id}, the value 'my-optional which was [optional] is here as well!"

Currently, tname(Class.property_function) returns 'property'. Should return the name of the property.

Is this a use case for pytest-benchmark?

Add Fetcher logic to verify IDs returned by implementations.

Currently, only timedeltas are supported for both, and None only for after (meaning "until next scheduled datetime").

Expands before/after with:

• 'all'; All data before/after the scheduled time, and
• 'schedule+-<num-periods>': Include data num-periods scheduled times before/after the currently scheduled time.

Do not yield folds with empty frames.

1. Improve visibility
2. Reduce log spam
3. Improve cardinality handling

Or just a performance logging framework? Would be a natural extension of rics.performance.

Add support for WHERE-clauses to limit query results. Both strings (limited functionality) and SQLAlchemy clauses should be accepted.