bookshelf/tests/test_archives.py

"""Network integration tests for archive searcher plugins.

Each test queries a live external service for "War and Peace" by Tolstoy,
a book universally catalogued in all supported archives.

Run with:  pytest tests/ -m network
Skip with: pytest tests/ -m "not network"  (default in presubmit)
"""

import pytest

from models import CandidateRecord
from plugins.archives.html_scraper import HtmlScraperPlugin
from plugins.archives.openlibrary import OpenLibraryPlugin
from plugins.archives.rsl import RSLPlugin
from plugins.archives.sru_catalog import SRUCatalogPlugin
from plugins.rate_limiter import RateLimiter

pytestmark = pytest.mark.network

_RL = RateLimiter()
_TIMEOUT = 15


def _titles(results: list[CandidateRecord]) -> list[str]:
    return [r["title"] for r in results]


def _authors(results: list[CandidateRecord]) -> list[str]:
    return [r["author"] for r in results]


def _has_title(results: list[CandidateRecord], fragment: str) -> bool:
    """Return True if any result title contains fragment (case-insensitive)."""
    low = fragment.lower()
    return any(low in r["title"].lower() for r in results)


def _has_author(results: list[CandidateRecord], fragment: str) -> bool:
    """Return True if any result author contains fragment (case-insensitive)."""
    low = fragment.lower()
    return any(low in r["author"].lower() for r in results)


# ── OpenLibrary ───────────────────────────────────────────────────────────────


def test_openlibrary_war_and_peace() -> None:
    plugin = OpenLibraryPlugin(
        plugin_id="openlibrary",
        name="OpenLibrary",
        rate_limiter=_RL,
        rate_limit_seconds=0,
        auto_queue=True,
        timeout=_TIMEOUT,
        config={},
    )
    results = plugin.search("War and Peace Tolstoy")
    assert results, "OpenLibrary returned no results"
    assert all(r["source"] == "openlibrary" for r in results)
    assert _has_title(results, "war and peace"), f"titles={_titles(results)}"
    # OpenLibrary stores authors in their original language; accept both forms.
    assert _has_author(results, "tolstoy") or _has_author(results, "толст"), f"authors={_authors(results)}"


# ── RSL (РГБ) ─────────────────────────────────────────────────────────────────


def test_rsl_voina_i_mir() -> None:
    plugin = RSLPlugin(
        plugin_id="rsl",
        name="РГБ",
        rate_limiter=_RL,
        rate_limit_seconds=0,
        auto_queue=True,
        timeout=_TIMEOUT,
        config={},
    )
    results = plugin.search("Толстой Война и мир")
    assert results, "RSL returned no results"
    assert all(r["source"] == "rsl" for r in results)
    assert _has_title(results, "война"), f"titles={_titles(results)}"


# ── НЭБ (rusneb) ─────────────────────────────────────────────────────────────


def test_rusneb_voina_i_mir() -> None:
    plugin = HtmlScraperPlugin(
        plugin_id="rusneb",
        name="НЭБ",
        rate_limiter=_RL,
        rate_limit_seconds=0,
        auto_queue=True,
        timeout=_TIMEOUT,
        config={
            "url": "https://rusneb.ru/search/",
            "search_param": "q",
            "img_alt": True,
            "author_class": "search-list__item_subtext",
        },
    )
    results = plugin.search("Война и мир Толстой")
    assert results, "НЭБ returned no results"
    assert all(r["source"] == "rusneb" for r in results)
    assert _has_title(results, "война"), f"titles={_titles(results)}"
    assert _has_author(results, "толст"), f"authors={_authors(results)}"


# ── Alib ─────────────────────────────────────────────────────────────────────


def test_alib_voina_i_mir() -> None:
    plugin = HtmlScraperPlugin(
        plugin_id="alib_web",
        name="Alib (web)",
        rate_limiter=_RL,
        rate_limit_seconds=0,
        auto_queue=False,
        timeout=_TIMEOUT,
        config={
            "url": "https://www.alib.ru/find3.php4",
            "search_param": "tfind",
            "extra_params": {"f": "5", "s": "0"},
            "encoding": "cp1251",
            "bold_text": True,
        },
    )
    results = plugin.search("Война и мир Толстой")
    assert results, "Alib returned no results"
    assert all(r["source"] == "alib_web" for r in results)
    assert _has_title(results, "война"), f"titles={_titles(results)}"
    assert _has_author(results, "толст"), f"authors={_authors(results)}"


# ── НЛР (SRU) ────────────────────────────────────────────────────────────────
# The NLR SRU endpoint (www.nlr.ru/search/query) no longer exists (HTTP 404).


@pytest.mark.xfail(reason="nlr.ru SRU endpoint no longer available (HTTP 404)", strict=False)
def test_nlr_voina_i_mir() -> None:
    plugin = SRUCatalogPlugin(
        plugin_id="nlr",
        name="НЛР",
        rate_limiter=_RL,
        rate_limit_seconds=0,
        auto_queue=False,
        timeout=_TIMEOUT,
        config={
            "url": "http://www.nlr.ru/search/query",
            "query_prefix": "title=",
        },
    )
    results = plugin.search("Война и мир")
    assert results, "НЛР returned no results"
    assert all(r["source"] == "nlr" for r in results)
    assert _has_title(results, "война"), f"titles={_titles(results)}"


# ── ШПИЛ ─────────────────────────────────────────────────────────────────────
# The ШПИЛ IRBIS64 CGI endpoint no longer exists (HTTP 404).


@pytest.mark.xfail(reason="shpl.ru IRBIS64 CGI endpoint no longer available (HTTP 404)", strict=False)
def test_shpl_voina_i_mir() -> None:
    plugin = HtmlScraperPlugin(
        plugin_id="shpl",
        name="ШПИЛ",
        rate_limiter=_RL,
        rate_limit_seconds=0,
        auto_queue=False,
        timeout=_TIMEOUT,
        config={
            "url": "https://www.shpl.ru/cgi-bin/irbis64/cgiirbis_64.exe",
            "search_param": "S21ALL",
            "extra_params": {
                "C21COM": "S",
                "I21DBN": "BIBL",
                "P21DBN": "BIBL",
                "S21FMT": "briefWebRus",
                "Z21ID": "",
            },
            "brief_class": "brief",
        },
    )
    results = plugin.search("Война и мир")
    assert results, "ШПИЛ returned no results"
    assert all(r["source"] == "shpl" for r in results)
    assert _has_title(results, "война"), f"titles={_titles(results)}"