declare abbreviations to teach inline-snapshot what it should generate

[15r10nk]

inline-snapshot should ideally create snapshots that do not need to be modified by the developer. However, there are cases where the developer may want to use specific looking snapshots.

declare_abbreviation is intended to solve this problem by allowing the developer to specify what the snapshots should look like, and allow inline-snapshot to generate the correct code in the first place.

The idea is as follows:

• inline-snapshot checks each snapshot value and inner value (dict/list/... child) to see if it can be abbreviated.
• The abbreviation with the shortest representation is used if there are several ways to abbreviate a snapshot.

Below are some examples of what this could look like:

from inline_snapshot import snapshot, declare_abbreviation, Abbreviation
from dirty_equals import IsNow

from datetime import datetime, timezone


@declare_abbreviation
def _(value, context):
    if value == IsNow(tz=timezone.utc):
        return Abbreviation(IsNow, tz=timezone.utc)


def repeated(num, v):
    return [v] * num


@declare_abbreviation
def _(value, context):
    if value == [value[0]] * len(value):
        return Abbreviation(repeated, len(value), value[0])


def test():
    # the snapshots which will be generated by inline-snapshot

    assert [0,0,0,0,0] == snapshot(repeated(5, 0))
    assert datetime.now() == snapshot(IsNow(tz=timezone.utc))

context should give the developer some information about the context of the value (what the parent object is or what the key is if the value is inside a dictionary)

These abbreviations would be declared globally (in conftest.py).

This is currently just an idea and feedback is very welcome.

[samuelcolvin]

Thanks so much for addressing this, I wondered if such a thing would be possible but hadn't gotten around to creating an issue. 🙏