143 lines
9.7 KiB
Python
143 lines
9.7 KiB
Python
"""One answer from both sources: direct first sentence, no source labels, one work once, newest first."""
|
|
import pytest
|
|
|
|
from tools.answers import unified as U
|
|
from tools.answers.context_policy import normalized
|
|
|
|
PROFILE = "https://zp.test/students/2023/ronald_zalacko"
|
|
UNITS = [
|
|
{"person": "Ronald Zalacko", "source_path": "pages/students/2023/ronald_zalacko/README.md", "kind": "supervisor",
|
|
"phase": "unknown", "work_label": None, "text": "Vedúci: Matúš Pleva", "line": 3},
|
|
{"person": "Ronald Zalacko", "source_path": "pages/students/2023/ronald_zalacko/README.md", "kind": "context",
|
|
"phase": "unknown", "work_label": "Bakalárska práca 2026", "text": "Možné ciele:", "line": 8},
|
|
{"person": "Ronald Zalacko", "source_path": "pages/students/2023/ronald_zalacko/README.md", "kind": "context",
|
|
"phase": "unknown", "work_label": "Bakalárska práca 2026", "text": "- Vytvoriť skórovací systém pre súťaže.", "line": 9},
|
|
{"person": "Ronald Zalacko", "source_path": "pages/students/2023/ronald_zalacko/README.md", "kind": "result",
|
|
"phase": "result", "work_label": "Bakalárska práca 2026", "text": "- Hotová prvá verzia aplikácie.",
|
|
"record_date": "2026-03-01", "line": 20},
|
|
]
|
|
CRZP_ROW = {"title": "Skórovací online systém pre robotické súťaže", "year": 2026, "work_type": "bachelor_thesis",
|
|
"authors": ["Zalačko, Ronald"], "supervisors": ["Pleva, Matúš"], "opponents": ["Kupcová, Eva"],
|
|
"school": "Technická univerzita v Košiciach FEI", "url": "https://opac.test/1"}
|
|
|
|
|
|
@pytest.fixture
|
|
def data(monkeypatch):
|
|
works = {"Ronald Zalacko": [{"type": "bachelor_thesis", "heading": "Bakalárska práca 2026", "year": 2026,
|
|
"title": "Skórovací online systém pre robotické súťaže", "source": "wiki",
|
|
"url": PROFILE}],
|
|
"Dominik Nagy": [{"type": "team_project", "heading": "Tímový projekt 2019", "year": 2019, "title": "Rešerš",
|
|
"source": "wiki", "url": "https://zp.test/nagy"},
|
|
{"type": "diploma_thesis", "heading": "Diplomová práca 2022", "year": 2022,
|
|
"title": "Neurónový strojový preklad", "source": "wiki", "url": "https://zp.test/nagy"}]}
|
|
people = {"Ronald Zalacko": {"name": "Ronald Zalacko", "person_kind": "student", "start_year": 2023, "url": PROFILE},
|
|
"Dominik Nagy": {"name": "Dominik Nagy", "person_kind": "student", "url": "https://zp.test/nagy"}}
|
|
monkeypatch.setattr(U, "catalog", lambda: {"people": people, "works": works})
|
|
monkeypatch.setattr(U, "registry_rows", lambda person: []) # not the local register cache
|
|
monkeypatch.setenv("NAME_FORMS_MODEL", "0") # the rules, no model call
|
|
monkeypatch.setattr("tools.answers.names.spelled", lambda name: name)
|
|
monkeypatch.setattr(U, "units", lambda person: [u for u in UNITS if u["person"] == person])
|
|
monkeypatch.setattr("tools.knowledge.sources.profile_authors", lambda: {})
|
|
monkeypatch.setattr("tools.answers.combined.same_person_work", lambda person, row: False)
|
|
|
|
|
|
def test_the_supervisor_is_one_sentence_from_both_sources(data):
|
|
works = U.person_works("Ronald Zalacko", [CRZP_ROW])
|
|
assert len(works) == 1 and works[0]["row"] is CRZP_ROW # the same work in both sources is one work
|
|
out = U.supervisors("Ronald Zalacko", works, normalized("Kto je vedúci práce Ronalda Zalačka?"))
|
|
assert out["text"] == ("Vedúcim bakalárskej práce Ronalda Zalacka „Skórovací online systém pre robotické súťaže“ "
|
|
"(2026) je Matúš Pleva.")
|
|
assert out["urls"] == [PROFILE, "https://opac.test/1"]
|
|
|
|
|
|
def test_only_the_start_year_is_said(data):
|
|
assert U.start_year("Ronald Zalacko")["text"] == "Ronald Zalacko začal študovať v roku 2023."
|
|
|
|
|
|
def test_works_are_listed_newest_first_with_a_crzp_only_work_marked(data):
|
|
other = {**CRZP_ROW, "title": "Vyučbový chatbot", "year": 2019, "authors": ["Nagy, Dominik"],
|
|
"supervisors": ["Ondáš, Stanislav"], "url": "https://opac.test/2"}
|
|
text = U.works_list("Dominik Nagy", U.person_works("Dominik Nagy", [other]))["text"]
|
|
assert text.startswith("Dominik Nagy má tieto práce:")
|
|
assert text.index("Diplomová práca (2022)") < text.index("Tímový projekt (2019)")
|
|
assert "Vyučbový chatbot" in text and "Nie je isté, že ide o tú istú osobu" in text
|
|
assert "CRZP" not in text and "ZP Wiki" not in text
|
|
|
|
|
|
def test_about_a_work_names_its_title_and_goals(data):
|
|
text = U.about("Ronald Zalacko", U.person_works("Ronald Zalacko"), normalized("O čom je práca Ronalda Zalačka?"))["text"]
|
|
assert text.startswith("Bakalárska práca Ronalda Zalacka (2026) sa volá „Skórovací online systém")
|
|
assert "- Vytvoriť skórovací systém pre súťaže." in text and "- Hotová prvá verzia aplikácie." in text
|
|
|
|
|
|
def test_a_meeting_answer_starts_with_a_sentence_and_has_plain_points():
|
|
wiki = ("**Andrii Pervashov — Diplomová práca 2027**\nPosledný záznam s úplným dátumom: 2026-10-16.\n\n"
|
|
"- Zaznamenaný výsledok: Použitý je AMI korpus.\n\n- Priebežný stav: Aplikuje sa RAG prompting.\n\n"
|
|
"- Úloha zo stretnutia: Pridať citácie.")
|
|
text = U.meeting("Andrii Pervashov", wiki, normalized("Čo riešil Andrii Pervashov na poslednom stretnutí?"), False)
|
|
assert text.startswith("Andrii Pervashov na poslednom stretnutí 16.10.2026 riešil:")
|
|
assert "- Použitý je AMI korpus." in text and "Zaznamenaný" not in text and "Priebežný" not in text
|
|
tasks = U.meeting("Andrii Pervashov", wiki, normalized("a aké úlohy dostal?"), True)
|
|
assert tasks == "Na stretnutí 16.10.2026 dostal tieto úlohy:\n\n- Pridať citácie."
|
|
|
|
|
|
def test_tools_of_one_work_are_merged_once_from_both_sources(data):
|
|
wiki = "X vo svojej bakalárskej práci „Grafová databáza“ podľa záznamov využíva: streamlit, neo4J (začína), Python."
|
|
crzp = "X vo svojej bakalárskej práci „Grafová databáza“ (2026) využíva: Neo4j, Streamlit, Cypher."
|
|
text = U.methods("Ronald Zalacko", [], wiki, crzp)["text"]
|
|
assert text == "Ronald Zalacko vo svojej bakalárskej práci „Grafová databáza“ využíva: Streamlit, Neo4j, Python, Cypher."
|
|
|
|
|
|
def test_names_in_genitive_and_source_labels_removed():
|
|
g = U.rule_genitive # the fallback without the model
|
|
assert g("Ronald Zalačko") == "Ronalda Zalačka" and g("Simona Bobrovčanová") == "Simony Bobrovčanovej"
|
|
assert g("Kristián Sokolský") == "Kristiána Sokolského" and g("Daniel Hládek") == "Daniela Hládka"
|
|
text = U.strip_sources("Odpoveď.\n\nV CRZP (Centrálny register záverečných prác):\n\n- práca\n\n**1. A** (ZP Wiki)")
|
|
assert "CRZP" not in text and "ZP Wiki" not in text
|
|
|
|
|
|
def test_inline_register_links_move_to_the_sources_line():
|
|
text = U.strip_sources("Našlo sa 2 prác:\n\n- A — „X“ (2022) · [CRZP](https://opac.test/1)\n- B — „Y“ (2022) · "
|
|
"[CRZP](https://opac.test/2)\n\nZdroje: https://opac.test/1")
|
|
assert "[CRZP]" not in text and text.endswith("Zdroje: https://opac.test/1, https://opac.test/2")
|
|
|
|
|
|
def test_the_conversation_token_is_compressed_and_still_read(monkeypatch):
|
|
from tools.answers import combined as C
|
|
monkeypatch.setenv("SEARCH_API_KEY", "k")
|
|
token = C.seal({"wiki": "w" * 20000, "entities": []})
|
|
assert token.startswith("zc2.") and len(token) < 16000 and C.unseal(token)["wiki"] == "w" * 20000
|
|
|
|
|
|
def test_name_forms_from_the_model_are_checked_and_kept(monkeypatch, tmp_path):
|
|
from tools.answers import names
|
|
monkeypatch.setenv("NAME_FORMS_MODEL", "1")
|
|
monkeypatch.setattr(names, "_file", lambda: tmp_path / "name_forms.json")
|
|
monkeypatch.setattr(names, "_cache", {})
|
|
good = names.NameForms(gender="male", genitive="Ronalda Zalačka", dative="Ronaldovi Zalačkovi",
|
|
accusative="Ronalda Zalačka", locative="Ronaldovi Zalačkovi", instrumental="Ronaldom Zalačkom")
|
|
bad = good.model_copy(update={"genitive": "pána Ronalda Zalačka"}) # an extra word
|
|
assert names.valid("Ronald Zalačko", good) and not names.valid("Ronald Zalačko", bad)
|
|
calls = []
|
|
monkeypatch.setattr("tools.answers.synthesis._call", lambda *a, **k: calls.append(1) or good)
|
|
monkeypatch.setattr(names, "spelled", lambda name: name)
|
|
assert names.case("Ronald Zalačko", "genitive") == "Ronalda Zalačka"
|
|
assert names.case("Ronald Zalačko", "instrumental") == "Ronaldom Zalačkom" and len(calls) == 1 # asked once
|
|
assert "Ronaldovi Zalačkovi" in (tmp_path / "name_forms.json").read_text(encoding="utf-8")
|
|
|
|
|
|
def test_a_name_without_diacritics_takes_the_register_spelling(monkeypatch):
|
|
from tools.answers import names
|
|
monkeypatch.setattr(names, "_spellings", lambda: {"ronald zalacko": "Ronald Zalačko", "daniel hladek": "Daniel Hládek"})
|
|
assert names.spelled("Ronald Zalacko") == "Ronald Zalačko" and names.spelled("Daniel Hladek") == "Daniel Hládek"
|
|
assert names.spelled("Ján Pták") == "Ján Pták" and names.spelled("Valerii Kutsenko") == "Valerii Kutsenko"
|
|
|
|
|
|
def test_a_respelled_or_undeclined_name_is_rejected():
|
|
from tools.answers import names
|
|
base = dict(gender="male", dative="Jakubovi Schwarcovi", accusative="Jakuba Schwarca", locative="Jakubovi Schwarcovi")
|
|
assert names.valid("Jakub Schwarc", names.NameForms(genitive="Jakuba Schwarca", instrumental="Jakubom Schwarcom", **base))
|
|
assert not names.valid("Jakub Schwarc", names.NameForms(genitive="Jakuba Schwarca", instrumental="Jakubom Schwarzom", **base))
|
|
assert not names.valid("Jakub Schwarc", names.NameForms(genitive="Jakub Schwarc", instrumental="Jakubom Schwarcom", **base))
|
|
assert names.fits("Hládek", "Hládka") and names.fits("Zalačko", "Zalačka") and not names.fits("Andrii", "Andrija")
|