signal-bot/tests/test_pack_ledger.py

# Copyright (c) 2026 Krzysztof Rudnicki. MIT licence.
"""The import ledger: what the album has already taken in, and from what."""

from __future__ import annotations

import json
from typing import TYPE_CHECKING

from signal_bot import pack_ledger
from signal_bot.pack import MEDIA_DIR
from signal_bot.pack_custom import FILE_MODE
from signal_bot.pack_ledger import LEDGER_FILE, Entry
from signal_bot.pack_store import ADDS_FILE
from tests.pack_fixture import PIXEL

if TYPE_CHECKING:
    from pathlib import Path


def _album(tmp_path: Path, files: dict[str, bytes]) -> Path:
    """A pack directory holding exactly ``files`` in media/."""
    media = tmp_path / MEDIA_DIR
    media.mkdir(parents=True)
    for name, payload in files.items():
        (media / name).write_bytes(payload)
    return tmp_path


def test_digest_is_a_full_sha256() -> None:
    # Not content_name's truncated 32: a filename is a name, this is identity.
    assert len(pack_ledger.digest(b"x")) == 64


def test_a_missing_ledger_reads_as_empty(tmp_path: Path) -> None:
    assert pack_ledger.read(tmp_path) == {}


def test_a_corrupt_ledger_degrades_instead_of_raising(tmp_path: Path) -> None:
    # The whole point: pack_load raises at startup and Restart=always turns
    # that into a loop. A broken ledger must cost a duplicate, not the bot.
    (tmp_path / LEDGER_FILE).write_text("{not json", encoding="utf-8")
    assert pack_ledger.read(tmp_path) == {}


def test_a_ledger_that_is_not_an_object_reads_as_empty(tmp_path: Path) -> None:
    (tmp_path / LEDGER_FILE).write_text("[1, 2]", encoding="utf-8")
    assert pack_ledger.read(tmp_path) == {}


def test_one_malformed_row_does_not_lose_the_others(tmp_path: Path) -> None:
    (tmp_path / LEDGER_FILE).write_text(
        json.dumps(
            {
                "good.jpg": {"source": "a" * 64, "stored": "b" * 64},
                "not-a-dict.jpg": "nonsense",
                "no-stored.jpg": {"source": "c" * 64},
                "bad-source.jpg": {"source": 7, "stored": "d" * 64},
            }
        ),
        encoding="utf-8",
    )
    assert set(pack_ledger.read(tmp_path)) == {"good.jpg"}


def test_a_null_source_round_trips(tmp_path: Path) -> None:
    entries = {"chat.jpg": Entry(source=None, stored="e" * 64)}
    pack_ledger.write(tmp_path, entries)
    assert pack_ledger.read(tmp_path) == entries


def test_the_ledger_is_written_private(tmp_path: Path) -> None:
    pack_ledger.write(tmp_path, {"a.jpg": Entry(source="1" * 64, stored="2" * 64)})
    assert (tmp_path / LEDGER_FILE).stat().st_mode & 0o777 == FILE_MODE


def test_source_hashes_skips_the_unrecoverable_ones() -> None:
    entries = {
        "a.jpg": Entry(source="1" * 64, stored="2" * 64),
        "b.jpg": Entry(source=None, stored="3" * 64),
    }
    assert pack_ledger.source_hashes(entries) == {"1" * 64}
    assert pack_ledger.stored_hashes(entries) == {"2" * 64, "3" * 64}


def test_backfill_records_an_imported_file_as_its_own_source(tmp_path: Path) -> None:
    # shutil.copy2 copied these verbatim, so the bytes on disk *are* the
    # source bytes. This is the only moment that is recoverable.
    directory = _album(tmp_path, {"IMG_1.jpg": PIXEL})
    entries = pack_ledger.backfill(directory)
    assert entries["IMG_1.jpg"].source == entries["IMG_1.jpg"].stored
    assert entries["IMG_1.jpg"].stored == pack_ledger.digest(PIXEL)


def test_backfill_leaves_a_chat_upload_without_a_source(tmp_path: Path) -> None:
    # adds.json is the exact record of what came in through !add_photo, and
    # those were stripped on the way in: the source bytes never existed here.
    directory = _album(tmp_path, {"abc123.jpg": PIXEL})
    (directory / ADDS_FILE).write_text(
        json.dumps({"abc123.jpg": {"by": "someone", "at": 1}}), encoding="utf-8"
    )
    entries = pack_ledger.backfill(directory)
    assert entries["abc123.jpg"].source is None
    assert entries["abc123.jpg"].stored == pack_ledger.digest(PIXEL)


def test_backfill_persists_what_it_found(tmp_path: Path) -> None:
    directory = _album(tmp_path, {"IMG_1.jpg": PIXEL})
    pack_ledger.backfill(directory)
    assert set(pack_ledger.read(directory)) == {"IMG_1.jpg"}


def test_backfill_never_revises_an_entry_it_already_has(tmp_path: Path) -> None:
    # Re-running must not overwrite a source hash recorded before a strip
    # with the post-strip bytes, which would silently lose the origin.
    directory = _album(tmp_path, {"IMG_1.jpg": PIXEL})
    pack_ledger.write(directory, {"IMG_1.jpg": Entry(source="old", stored="older")})
    entries = pack_ledger.backfill(directory)
    assert entries["IMG_1.jpg"] == Entry(source="old", stored="older")


def test_backfill_is_idempotent(tmp_path: Path) -> None:
    directory = _album(tmp_path, {"IMG_1.jpg": PIXEL})
    assert pack_ledger.backfill(directory) == pack_ledger.backfill(directory)


def test_backfill_ignores_a_directory_inside_media(tmp_path: Path) -> None:
    directory = _album(tmp_path, {"IMG_1.jpg": PIXEL})
    (directory / MEDIA_DIR / "subdir").mkdir()
    assert set(pack_ledger.backfill(directory)) == {"IMG_1.jpg"}


def test_backfill_on_an_album_with_no_media_directory(tmp_path: Path) -> None:
    assert pack_ledger.backfill(tmp_path) == {}