Packaging improvements

[lmmx]

Status - ready for review 🏁

This PR carries out some housekeeping and I have verified them all as I go.

First I ran some linting (ruff code formatter and flake8 static analyser) that put the code in a standard format and fixed some mistakes (e.g. unnamed variables) and clarified ambiguous code (e.g. made * imports explicit).

• upgraded setup.py to pyproject.toml
• bumped setuptools version
• bumped required Python version to 3.10 (as 3.7 was end of life in 2018)
• added README to the package metadata (current package has no documentation on PyPI)
• added keywords to the package metadata for discoverability
• removed transitive dependencies (blis and confection are pulled in by spacy's thinc dep)
• added optional "extra" dependency groups:
  • plots - the matplotlib and seaborn deps
  • dev - ruff and flake8
• made plotting library loading lazy and conditional (will not raise an error if not installed unless a plotting function is used, in which case it hints to install the biberplus[plots] extra)
• deleted requirements.txt which is redundant with the pyproject.toml package dependencies and pinned a spaCy model version of 3.5 which would fall behind (latest is currently 3.8)
• produced a dependency lockfile with uv (uv sync --all-extras)
• fixed syntax warnings (slashes aren't special in Python regexes and doesn't need escaping)
• removed failing test (invalid sentence-final interjection UH tag, another test already covers this)

Checklist

• Check tests pass
  • One test failure was an invalid interjection, but there is already test coverage (bad duplicate), removed this test:

FAILED tests/tagger/test_additional_features.py::TestAdditionalFunctions::test_interjections_end2 - AssertionError: 'UH' not found in ['JJ']

Click to show details on the regex syntax warnings

Regex syntax warnings

There are 2 warnings:

Regex warning 1: emoticon escape

biberplus/tagger/biber_plus_tagger.py:1229
  /home/louis/dev/biberplus/biberplus/tagger/biber_plus_tagger.py:1229: SyntaxWarning: invalid escape sequence '\/'
    emoticon_pattern = re.compile("[:;=](?:-)?[)DPp\/]")

>>> import re
>>> emoticon_pattern = re.compile(r"[:;=](?:-)?[)DPp/]")
>>> emoticon_pattern.findall("hello :) how are you ;-D")
[':)', ';-D']
>>> emoticon_pattern.findall("lol :P this is fun :-) yeah =/")
[':P', ':-)', '=/']

The pattern matches:

• [:;=] - eyes (colon, semicolon, or equals)
• (?:-)? - optional nose (dash)
• [)DPp/] - mouth (smile, big grin, tongue, or slant)

The error is from escaping the slanted mouth of :-/

>>> emoticon_pattern = re.compile("[:;=](?:-)?[)DPp\/]")
<stdin>:1: SyntaxWarning: invalid escape sequence '\/'
>>> emoticon_pattern.findall("hm :-/ um... =/")
[':-/', '=/']

We don't need the backslash

>>> emoticon_pattern = re.compile("[:;=](?:-)?[)DPp/]")
>>> emoticon_pattern.findall("hm :-/ um... =/")
[':-/', '=/']

Regex warning 2: URL escape

Similarly

>>> import biberplus
/home/louis/dev/biberplus/biberplus/tagger/biber_plus_tagger.py:1287: SyntaxWarning: invalid escape sequence '\/'
  url_pattern = "(https?:\/\/(?:www\.|(?!www))[a-zA-Z0-9][a-zA-Z0-9-]+[a-zA-Z0-9]\.[^\s]{2,}|www\.[a-zA-Z0-9][a-zA-Z0-9-]+[a-zA-Z0-9]\.[^\s]{2,}|https?:\/\/(?:www\.|(?!www))[a-zA-Z0-9]+\.[^\s]{2,}|www\.[a-zA-Z0-9]+\.[^\s]{2,})"

This one requires a raw string for the legitimate escapes \s and \. but we should just remove the backslash escaping the / (which Python doesn't need, that's Javascript-like regex)

-url_pattern = "(https?:\/\/(?:www\.|(?!www))[a-zA-Z0-9][a-zA-Z0-9-]+[a-zA-Z0-9]\.[^\s]{2,}|www\.[a-zA-Z0-9][a-zA-Z0-9-]+[a-zA-Z0-9]\.[^\s]{2,}|https?:\/\/(?:www\.|(?!www))[a-zA-Z0-9]+\.[^\s]{2,}|www\.[a-zA-Z0-9]+\.[^\s]{2,})"
+url_pattern = r"(https?://(?:www\.|(?!www))[a-zA-Z0-9][a-zA-Z0-9-]+[a-zA-Z0-9]\.[^\s]{2,}|www\.[a-zA-Z0-9][a-zA-Z0-9-]+[a-zA-Z0-9]\.[^\s]{2,}|https?://(?:www\.|(?!www))[a-zA-Z0-9]+\.[^\s]{2,}|www\.[a-zA-Z0-9]+\.[^\s]{2,})"

Specifically this removes the escaping of \/\/ -> // in the https:// parts of the regex

...and with that biberplus imports without a SyntaxWarning 🎉

[KenanA95]

This is great.

It will be difficult review with all of the linter changes included. Can you separate those out?

[lmmx]

I don't know if git will (easily) let me "undo" those commits but I went again in reverse in #9 and avoided running the ruff re-formatting (I usually use an editor that auto-formats on saves!)

I've initially stopped short of running the flake8 lints (i.e. I re-did back to commit 4 of this PR at 90db95b) because I wasn't sure if you meant lint as in just code formatters or also the flake8 ones (some of which are correctness), but I'll leave it there in case I misinterpret!

Assuming that PR is OK, remaining work is just the re-formatting and removing linter errors, please let me know how you'd want those (if you have any particular preferences) 🙂