signal-bot/tests/test_commons.py

# Copyright (c) 2026 Krzysztof Rudnicki. MIT licence.
"""The imageinfo query both Commons commands send."""

from __future__ import annotations

import pytest

from signal_bot.commons import (
    CAPTION_LIMIT,
    SUFFIXES,
    THUMB_PATH,
    caption_for,
    imageinfo_params,
    page_of,
)
from signal_bot.http_client import ImageFetchError


def test_a_file_can_be_named_by_title() -> None:
    params = imageinfo_params(640, title="File:1868 Mausefalle.jpg")
    assert params["titles"] == "File:1868 Mausefalle.jpg"
    assert "pageids" not in params
    assert params["iiurlwidth"] == "640"
    assert params["prop"] == "imageinfo"


def test_a_file_can_be_named_by_page_id() -> None:
    # The page id arrives as an int and has to reach Commons as a string;
    # sending the int would raise inside httpx rather than here.
    params = imageinfo_params(480, page_id=111419226)
    assert params["pageids"] == "111419226"
    assert "titles" not in params


@pytest.mark.parametrize(
    ("title", "page_id"),
    [(None, None), ("File:x.jpg", 1)],
)
def test_naming_a_file_neither_way_or_both_ways_is_refused(
    title: str | None, page_id: int | None
) -> None:
    # Preferring one silently would make a caller's mistake look like a 404,
    # which is the one failure shape that already exists downstream.
    with pytest.raises(ValueError, match="exactly one"):
        imageinfo_params(640, title=title, page_id=page_id)


def test_the_thumbnail_path_is_where_commons_actually_puts_it() -> None:
    # Pinned because both callers walk it and neither can see it fail until
    # a real fetch returns something unwalkable.
    assert THUMB_PATH == ("query", "pages", 0, "imageinfo", 0, "thumburl")


def test_the_suffix_allowlist_is_what_signal_can_render() -> None:
    assert ".jpg" in SUFFIXES
    assert ".svg" not in SUFFIXES


def _page(
    description: str | None = None, *, title: str = "File:Molchreuse.JPG", url: str = ""
) -> dict[str, object]:
    """One imageinfo page, shaped the way Commons shapes it."""
    info: dict[str, object] = {"thumburl": "https://cdn/x.jpg"}
    if url:
        info["descriptionurl"] = url
    if description is not None:
        info["extmetadata"] = {"ImageDescription": {"value": description}}
    return {"title": title, "imageinfo": [info]}


def test_asking_for_a_description_widens_the_query() -> None:
    plain = imageinfo_params(640, page_id=1)
    described = imageinfo_params(640, page_id=1, describe=True)
    assert plain["iiprop"] == "url"
    assert "iiextmetadatafilter" not in plain
    assert described["iiprop"] == "url|extmetadata"
    assert described["iiextmetadatafilter"] == "ImageDescription"


def test_the_caption_strips_markup_and_links_the_page() -> None:
    page = _page("Eine <b>Molchreuse</b> &amp; ein Korb", url="https://commons/File:M.JPG")
    assert caption_for(page) == "Eine Molchreuse & ein Korb\nhttps://commons/File:M.JPG"


def test_the_catalogue_list_is_cut_and_the_sentence_kept() -> None:
    # The Fisheries records append their filing fields as an HTML list.
    page = _page("Round Fyke (Bollreuse). Prussia\n<ul><li>Subject: Fishing nets</li></ul>")
    assert caption_for(page) == "Round Fyke (Bollreuse). Prussia"


@pytest.mark.parametrize("description", [None, "", "<ul><li>Subject: only filing</li></ul>"])
def test_a_description_that_says_nothing_falls_back_to_the_title(description: str | None) -> None:
    assert caption_for(_page(description)) == "Molchreuse"


def test_a_long_description_is_cut_rather_than_burying_the_photo() -> None:
    page = _page("word " * 200)
    caption = caption_for(page)
    assert caption.endswith("…")
    # rstrip() before the ellipsis means the exact length depends on where
    # the cut landed; the ceiling is what matters.
    assert len(caption) <= CAPTION_LIMIT + 1


def test_a_page_commons_could_not_find_is_an_image_fetch_error() -> None:
    # A deleted or renumbered page id comes back as {"missing": true}.
    for document in ({"query": {"pages": [{"missing": True}]}}, {"query": {"pages": []}}, {}, []):
        with pytest.raises(ImageFetchError, match="named no image"):
            page_of(document)


def test_the_page_comes_back_when_commons_did_find_it() -> None:
    assert page_of({"query": {"pages": [_page("x")]}})["title"] == "File:Molchreuse.JPG"


def test_a_template_that_is_not_a_description_falls_back_to_the_title() -> None:
    # Measured on a Maltese fish trap: the filename said more than Commons'
    # own description field did.
    page = _page(
        "This photo has been taken in the country: Malta",
        title="File:Hand woven fish trap.jpg",
    )
    assert caption_for(page) == "Hand woven fish trap"