big upgrade UwU

2026-07-27 15:16:01 +02:00
parent 9fb2bf0a74
commit 1ff979f1e5
30 changed files with 3304 additions and 183 deletions
@@ -0,0 +1,270 @@
+from __future__ import annotations
+
+import json
+from dataclasses import replace
+
+import pytest
+from telegram.error import TimedOut
+
+from sticker_downloader.config import METADATA_FILE_NAME, DownloadConfig
+from sticker_downloader.downloader import (
+    StickerDownloader,
+    primary_file_type,
+    sanitize_filename_part,
+)
+from sticker_downloader.errors import FetchError
+from sticker_downloader.results import FileStatus
+from tests.conftest import (
+    FakeBot,
+    FakeFetcher,
+    FakeSticker,
+    FakeStickerSet,
+    FakeThumbnail,
+    make_sticker,
+    no_sleep,
+)
+
+
+def build(bot, fetcher, config, **kwargs) -> StickerDownloader:
+    return StickerDownloader(
+        bot=bot, fetcher=fetcher, config=config, sleep=no_sleep, **kwargs
+    )
+
+
+async def test_downloads_stickers_thumbnails_and_pack_cover(bot, fetcher, config):
+    result = await build(bot, fetcher, config).download_pack(
+        "https://t.me/addstickers/TestPack"
+    )
+
+    assert result.ok
+    assert result.name == "TestPack"
+    assert result.title == "Test Pack"
+    assert result.total_stickers == 2
+
+    directory = config.output_dir / "TestPack"
+    written = sorted(p.name for p in directory.iterdir())
+    assert written == [
+        "001.webp",
+        "001_thumb.png",
+        "002.webp",
+        "002_thumb.png",
+        "_pack_thumbnail.webp",
+        METADATA_FILE_NAME,
+    ]
+    assert (directory / "001.webp").read_bytes() == b"sticker-bytes"
+    assert result.downloaded == 5
+    assert result.bytes_written == 5 * len(b"sticker-bytes")
+
+
+async def test_file_type_filter_limits_what_is_written(bot, fetcher, config):
+    config = replace(config, file_types=frozenset({"webp"}), pack_thumbnail=False)
+    result = await build(bot, fetcher, config).download_pack("TestPack")
+
+    directory = config.output_dir / "TestPack"
+    assert sorted(p.name for p in directory.iterdir()) == [
+        "001.webp",
+        "002.webp",
+        METADATA_FILE_NAME,
+    ]
+    assert result.downloaded == 2
+    # No thumbnails were even looked up.
+    assert bot.get_file_calls == ["file1", "file2"]
+
+
+async def test_extension_follows_sticker_kind(config, fetcher):
+    animated = FakeSticker(file_id="anim", is_animated=True)
+    video = FakeSticker(file_id="vid", is_video=True)
+    pack = FakeStickerSet(name="Mixed", stickers=[animated, video])
+    bot = FakeBot({"Mixed": pack}, extensions={"anim": "tgs", "vid": "webm"})
+
+    await build(bot, fetcher, config).download_pack("Mixed")
+
+    directory = config.output_dir / "Mixed"
+    assert (directory / "001.tgs").is_file()
+    assert (directory / "002.webm").is_file()
+
+
+async def test_pack_thumbnail_extension_comes_from_telegram(config, fetcher):
+    pack = FakeStickerSet(
+        name="Vid", stickers=[make_sticker(1)], thumbnail=FakeThumbnail("cover")
+    )
+    bot = FakeBot({"Vid": pack}, extensions={"cover": "webm"})
+
+    await build(bot, fetcher, config).download_pack("Vid")
+
+    assert (config.output_dir / "Vid" / "_pack_thumbnail.webm").is_file()
+
+
+async def test_existing_files_are_skipped_without_api_calls(bot, fetcher, config):
+    directory = config.output_dir / "TestPack"
+    directory.mkdir(parents=True)
+    (directory / "001.webp").write_bytes(b"old")
+
+    result = await build(bot, fetcher, config).download_pack("TestPack")
+
+    assert result.skipped == 1
+    assert result.downloaded == 4
+    assert (directory / "001.webp").read_bytes() == b"old"
+    assert "file1" not in bot.get_file_calls
+
+
+async def test_overwrite_replaces_existing_files(bot, fetcher, config):
+    directory = config.output_dir / "TestPack"
+    directory.mkdir(parents=True)
+    (directory / "001.webp").write_bytes(b"old")
+
+    config = replace(config, overwrite=True)
+    result = await build(bot, fetcher, config).download_pack("TestPack")
+
+    assert result.skipped == 0
+    assert (directory / "001.webp").read_bytes() == b"sticker-bytes"
+
+
+async def test_dry_run_writes_nothing(bot, fetcher, config):
+    config = replace(config, dry_run=True)
+    result = await build(bot, fetcher, config).download_pack("TestPack")
+
+    assert result.planned == 5
+    assert result.downloaded == 0
+    assert not config.output_dir.exists()
+    assert fetcher.urls == []
+    assert bot.get_file_calls == []
+
+
+async def test_invalid_reference_never_touches_the_network(bot, fetcher, config):
+    result = await build(bot, fetcher, config).download_pack("https://example.com/nope")
+
+    assert not result.ok
+    assert result.error is not None
+    assert bot.get_sticker_set_calls == []
+
+
+async def test_unknown_pack_reports_a_readable_error(bot, fetcher, config):
+    result = await build(bot, fetcher, config).download_pack("NoSuchPack")
+
+    assert not result.ok
+    assert result.error == "sticker pack not found"
+    assert not config.output_dir.exists()
+
+
+async def test_one_bad_file_does_not_sink_the_pack(bot, config):
+    fetcher = FakeFetcher(failures={"file1": FetchError("file1", 404, "Not Found")})
+    result = await build(bot, fetcher, config).download_pack("TestPack")
+
+    assert not result.ok
+    assert result.failed == 1
+    assert result.downloaded == 4
+    failure = next(o for o in result.outcomes if o.status is FileStatus.FAILED)
+    assert failure.path.name == "001.webp"
+    assert "404" in (failure.error or "")
+
+
+async def test_transient_failures_are_retried(bot, config):
+    calls = {"n": 0}
+    real_fetch = FakeFetcher().fetch
+
+    class FlakyFetcher:
+        async def fetch(self, url: str) -> bytes:
+            calls["n"] += 1
+            if calls["n"] == 1:
+                raise TimedOut
+            return await real_fetch(url)
+
+    result = await build(bot, FlakyFetcher(), config).download_pack("TestPack")
+
+    assert result.ok
+    assert result.downloaded == 5
+    assert calls["n"] == 6  # five files plus the one retry
+
+
+async def test_no_partial_files_are_left_behind(bot, config):
+    fetcher = FakeFetcher(failures={"file2": FetchError("file2", 500)})
+    config = replace(config, retries=1)
+    await build(bot, fetcher, config).download_pack("TestPack")
+
+    directory = config.output_dir / "TestPack"
+    assert [p.name for p in directory.glob("*.part")] == []
+
+
+async def test_concurrency_is_bounded(config, fetcher):
+    stickers = [make_sticker(i) for i in range(1, 11)]
+    bot = FakeBot({"Big": FakeStickerSet(name="Big", stickers=stickers)})
+    config = replace(config, concurrency=3)
+
+    await build(bot, fetcher, config).download_pack("Big")
+
+    assert fetcher.max_concurrent <= 3
+    assert len(fetcher.urls) == 20
+
+
+async def test_emoji_names_are_used_when_requested(config, fetcher):
+    pack = FakeStickerSet(name="Emo", stickers=[FakeSticker(file_id="a", emoji="🦊")])
+    bot = FakeBot({"Emo": pack})
+    config = replace(config, emoji_names=True, pack_thumbnail=False)
+
+    await build(bot, fetcher, config).download_pack("Emo")
+
+    assert (config.output_dir / "Emo" / "001_🦊.webp").is_file()
+
+
+async def test_metadata_describes_the_pack(bot, fetcher, config):
+    await build(bot, fetcher, config).download_pack("https://t.me/addstickers/TestPack")
+
+    payload = json.loads(
+        (config.output_dir / "TestPack" / METADATA_FILE_NAME).read_text(encoding="utf-8")
+    )
+    assert payload["pack"]["name"] == "TestPack"
+    assert payload["pack"]["title"] == "Test Pack"
+    assert payload["pack"]["reference"] == "https://t.me/addstickers/TestPack"
+    assert payload["sticker_count"] == 2
+    assert payload["requested_file_types"] == ["webp", "tgs", "webm", "png"]
+    first = payload["stickers"][0]
+    assert first["index"] == 1
+    assert first["emoji"] == "😀"
+    assert first["files"] == ["001.webp", "001_thumb.png"]
+
+
+async def test_metadata_can_be_disabled(bot, fetcher, config):
+    config = replace(config, write_metadata=False)
+    await build(bot, fetcher, config).download_pack("TestPack")
+    assert not (config.output_dir / "TestPack" / METADATA_FILE_NAME).exists()
+
+
+async def test_download_all_reports_every_pack(fetcher, config):
+    packs = {
+        "One": FakeStickerSet(name="One", stickers=[make_sticker(1)]),
+        "Two": FakeStickerSet(name="Two", stickers=[make_sticker(1)]),
+    }
+    bot = FakeBot(packs)
+
+    results = await build(bot, fetcher, config).download_all(["One", "Two", "Missing"])
+
+    assert [r.reference for r in results] == ["One", "Two", "Missing"]
+    assert [r.ok for r in results] == [True, True, False]
+
+
+def test_primary_file_type_prefers_animated_over_video():
+    assert primary_file_type(FakeSticker(file_id="a")) == "webp"
+    assert primary_file_type(FakeSticker(file_id="a", is_animated=True)) == "tgs"
+    assert primary_file_type(FakeSticker(file_id="a", is_video=True)) == "webm"
+
+
+@pytest.mark.parametrize(
+    ("raw", "expected"),
+    [
+        ("🦊", "🦊"),
+        ("a/b", "ab"),
+        ("..", ""),
+        ('x<>:"|?*y', "xy"),
+        ("x" * 50, "x" * 32),
+    ],
+)
+def test_sanitize_filename_part(raw: str, expected: str):
+    assert sanitize_filename_part(raw) == expected
+
+
+async def test_output_directory_is_created_lazily(bot, fetcher, tmp_path):
+    nested = tmp_path / "a" / "b" / "c"
+    config = DownloadConfig(output_dir=nested)
+    await build(bot, fetcher, config).download_pack("TestPack")
+    assert (nested / "TestPack" / "001.webp").is_file()