`tests/test_search_index.py`

63 lines
"""q= must also match repository, collection_name and publication text (seed defect: 'wichita' returned 0)."""
import time

from fastapi.testclient import TestClient

from conrad import db
from conrad.models import CartoonRecord


def _client(monkeypatch):
    for k in ("CONRAD_BASIC_USER", "CONRAD_BASIC_PASS", "CONRAD_REQUIRE_AUTH"):
        monkeypatch.delenv(k, raising=False)
    from conrad.web.app import create_app
    return TestClient(create_app())


def _seed(conn):
    db.upsert_source(conn, "wichita", "Wichita State")
    db.upsert_source(conn, "other", "Other")
    db.save_record(conn, CartoonRecord(canonical_id="w:1", identifier="1", title="Untitled drawing", year=1970,
                                       repository="Wichita State University Libraries Special Collections",
                                       collection_name="Paul Conrad Cartoons MS 87-10"), "wichita")
    db.save_record(conn, CartoonRecord(canonical_id="o:1", identifier="1", title="Nixon at sea", year=1973,
                                       publication="The Daily Iowan", repository="Elsewhere",
                                       collection_name="Box drawings"), "other")
    conn.commit()
    db.refresh_index(conn)


def test_q_matches_repository_collection_publication(tmpdb, monkeypatch):
    _seed(tmpdb)
    c = _client(monkeypatch)
    get = lambda q: c.get("/api/search", params={"q": q, "granularity": "all"}).json()["total"]
    assert get("wichita") == 1          # repository text
    assert get("ms 87-10") == 1         # collection_name text
    assert get("daily iowan") == 1      # publication text
    assert get("NIXON") == 1            # title still works, case-insensitive
    assert get("nothing-like-this") == 0


def test_index_migration_backfills_old_index(tmpdb, monkeypatch):
    _seed(tmpdb)
    tmpdb.execute("UPDATE cartoon_index SET search_text=NULL")
    tmpdb.commit()
    c = _client(monkeypatch)  # create_app must notice NULL search_text and rebuild
    assert c.get("/api/search", params={"q": "wichita", "granularity": "all"}).json()["total"] == 1


def test_search_is_fast(tmpdb, monkeypatch):
    _seed(tmpdb)
    rows = [(f"x:{i}", f"{i}") for i in range(3000)]
    for cid, ident in rows:
        db.save_record(tmpdb, CartoonRecord(canonical_id=cid, identifier=ident, title=f"t{ident}", year=1960,
                                            repository="Huntington Library", collection_name="Conrad papers"), "other")
    tmpdb.commit()
    db.refresh_index(tmpdb)
    c = _client(monkeypatch)
    c.get("/api/search", params={"q": "huntington"})
    t = time.perf_counter()
    r = c.get("/api/search", params={"q": "huntington", "granularity": "all"}).json()
    assert r["total"] == 3000
    assert time.perf_counter() - t < 0.1