FabledCurator/tests/test_gallery_dl_naming.py

"""gallery-dl naming patterns, rendered against the metadata gallery-dl really emits.

gallery-dl does not fail on a format field that names a key the extractor never
sets. It renders "None" and carries on. That is how every Discord download
landed in a `None/` directory as `<date>_None_<name>`: the patterns asked for
`{channel[name]}` and `{id}`, and the real metadata has a string `channel` and
`message_id`, with no `id` at all. The mismatch also broke sidecar pairing, so
no Discord file ever got a Post or a post date.

The fixture keeps the key set and value TYPES of a real attachment sidecar from
the operator's instance (2026-09-13), with every value invented. Rendering
through Python's own formatter raises on a missing key or a subscript into a
string, which is the loud failure gallery-dl does not give.
"""

import json
import string
from datetime import datetime
from pathlib import Path

import pytest

from backend.app.services.gallery_dl import (
    DISCORD_DIRECTORY,
    DISCORD_FILENAME,
    GalleryDLService,
    SourceConfig,
    sidecar_name_for,
)
from backend.app.utils.sidecar import find_sidecar

_FIXTURE = Path(__file__).parent / "fixtures" / "discord_attachment_sidecar.json"


@pytest.fixture
def kwdict():
    data = json.loads(_FIXTURE.read_text())
    # gallery-dl hands the formatter a datetime; the JSON sidecar stores it as text.
    data["date"] = datetime.strptime(data["date"], "%Y-%m-%d %H:%M:%S")
    return data


def _render(pattern, kwdict):
    return string.Formatter().vformat(pattern, (), kwdict)


def test_the_discord_filename_renders_from_real_keys(kwdict):
    name = _render(DISCORD_FILENAME, kwdict)
    assert "None" not in name
    assert name == "20240716_300000000000000003_01_image.png"


def test_the_discord_directory_renders_from_real_keys(kwdict):
    assert [_render(p, kwdict) for p in DISCORD_DIRECTORY] == ["nsfw-drops"]


@pytest.mark.parametrize(
    "broken", ["{channel[name]}", "{date:%Y%m%d}_{id}_{filename}.{extension}"],
)
def test_the_patterns_that_shipped_would_have_failed_here(kwdict, broken):
    """Positive control: the two patterns that produced `None/..._None_...`
    must fail this renderer, or the tests above prove nothing."""
    with pytest.raises((KeyError, TypeError)):
        _render(broken, kwdict)


def test_the_sidecar_is_named_exactly_like_its_media(kwdict, tmp_path):
    """So `find_sidecar` pairs them on its first candidate, and two attachments
    that share an original name (`image.png`) can never share a sidecar."""
    media = tmp_path / _render(DISCORD_FILENAME, kwdict)
    sidecar = tmp_path / _render(sidecar_name_for(DISCORD_FILENAME), kwdict)
    media.write_bytes(b"x")
    sidecar.write_text("{}")
    assert find_sidecar(media) == sidecar


def test_two_attachments_with_the_same_original_name_get_distinct_sidecars(kwdict):
    other = {**kwdict, "message_id": "300000000000000099"}
    pattern = sidecar_name_for(DISCORD_FILENAME)
    assert _render(pattern, kwdict) != _render(pattern, other)


def test_sidecar_name_for_needs_an_extension_suffix():
    assert sidecar_name_for("{a}_{b}.{extension}") == "{a}_{b}.json"
    assert sidecar_name_for("{a}_{b}") is None


# --- the config gallery-dl is actually given -----------------------------------


@pytest.fixture
def gdl(tmp_path):
    return GalleryDLService(images_root=tmp_path / "images", validate_files=False)


def _discord_section(gdl, **overrides):
    cfg = gdl._build_config_for_source(
        platform="discord", source_config=SourceConfig(**overrides), artist_slug="a",
    )
    return cfg["extractor"]["discord"]


def test_discord_config_carries_the_mirrored_sidecar(gdl):
    section = _discord_section(gdl)
    assert section["filename"] == DISCORD_FILENAME
    assert section["directory"] == DISCORD_DIRECTORY
    assert section["postprocessors"][0]["filename"] == sidecar_name_for(DISCORD_FILENAME)


def test_a_filename_override_keeps_the_sidecar_mirrored(gdl):
    section = _discord_section(gdl, filename_pattern="{message_id}_{num}.{extension}")
    assert section["postprocessors"][0]["filename"] == "{message_id}_{num}.json"


def test_no_metadata_means_no_discord_postprocessor(gdl):
    section = _discord_section(gdl, save_metadata=False)
    assert "postprocessors" not in section