tests/sync/test_anthology_source.py

from unittest.mock import MagicMock
from sync.anthology_source import iter_papers


def _make_fake_paper(full_id, title, abstract, authors, venue, year, url, bibtex="", pdf_url=""):
    parent = MagicMock()
    parent.venue_acronym = venue

    pdf = MagicMock()
    pdf.url = pdf_url

    paper = MagicMock()
    paper.full_id = full_id
    paper.parent = parent
    paper.title = title
    paper.abstract = abstract
    paper.authors = authors
    paper.year = year
    paper.web_url = url
    paper.to_bibtex.return_value = bibtex
    paper.pdf = pdf
    return paper


class FakeMarkupText:
    """Stand-in for acl_anthology.text.MarkupText, which exposes .as_text()."""

    def __init__(self, text):
        self._text = text

    def as_text(self):
        return self._text


class FakeName:
    """Stand-in for acl_anthology.people.Name, exposing .as_first_last()."""

    def __init__(self, first, last):
        self.first = first
        self.last = last

    def as_first_last(self):
        if not self.first:
            return self.last
        return f"{self.first} {self.last}"


class FakeNameSpecification:
    """Stand-in for acl_anthology.people.NameSpecification, exposing .name."""

    def __init__(self, name):
        self.name = name


def test_iter_papers_maps_fields(monkeypatch):
    fake_anthology = MagicMock()
    fake_anthology.papers.return_value = [
        _make_fake_paper("2023.acl-long.1", "Title One", "Abstract one", ["Alice", "Bob"], "ACL", "2023", "http://x/p1", "@inproceedings{one,\n  title = \"Title One\"}", "http://x/p1.pdf"),
        _make_fake_paper("2022.emnlp-main.2", "Title Two", None, [], "EMNLP", "2022", "http://x/p2"),
    ]

    monkeypatch.setattr("sync.anthology_source.Anthology", lambda datadir: fake_anthology)

    results = list(iter_papers("/fake/repo"))
    assert results[0] == {
        "id": "2023.acl-long.1", "title": "Title One", "abstract": "Abstract one",
        "authors": "Alice, Bob", "venue": "ACL", "year": 2023, "url": "http://x/p1",
        "bibtex": "@inproceedings{one,\n  title = \"Title One\"}", "pdf_url": "http://x/p1.pdf",
    }
    assert results[1]["abstract"] == ""
    assert results[1]["authors"] == ""
    assert results[1]["year"] == 2022
    assert results[1]["bibtex"] == ""
    assert results[1]["pdf_url"] == ""


def test_iter_papers_handles_real_markup_text_and_name_objects(monkeypatch):
    """The real acl_anthology package (verified against installed 1.2.0) has
    Paper.title/abstract as MarkupText objects (.as_text()) and Paper.authors
    as a tuple of NameSpecification objects exposing `.name.as_first_last()`.
    """
    fake_anthology = MagicMock()
    fake_anthology.papers.return_value = [
        _make_fake_paper(
            "2024.naacl-long.3",
            FakeMarkupText("Real Title"),
            FakeMarkupText("Real abstract text"),
            [FakeNameSpecification(FakeName("Alice", "Smith")), FakeNameSpecification(FakeName(None, "Cher"))],
            "NAACL",
            "2024",
            "http://x/p3",
            "@inproceedings{smith-2024-real}",
            "http://x/p3.pdf",
        ),
    ]

    monkeypatch.setattr("sync.anthology_source.Anthology", lambda datadir: fake_anthology)

    results = list(iter_papers("/fake/repo"))
    assert results[0] == {
        "id": "2024.naacl-long.3", "title": "Real Title", "abstract": "Real abstract text",
        "authors": "Alice Smith, Cher", "venue": "NAACL", "year": 2024, "url": "http://x/p3",
        "bibtex": "@inproceedings{smith-2024-real}", "pdf_url": "http://x/p3.pdf",
    }


def test_iter_papers_uses_datadir_subdirectory_of_local_checkout(monkeypatch):
    """anthology_path is the root of a git clone; the library expects the
    nested data/ directory, not the checkout root itself, and must not
    treat anthology_path as a repo URL to re-clone."""
    captured = {}

    def fake_anthology_ctor(datadir):
        captured["datadir"] = datadir
        anthology = MagicMock()
        anthology.papers.return_value = []
        return anthology

    monkeypatch.setattr("sync.anthology_source.Anthology", fake_anthology_ctor)

    list(iter_papers("/fake/checkout"))
    assert captured["datadir"] == "/fake/checkout/data"


def test_iter_papers_bibtex_falls_back_on_error(monkeypatch):
    """Papers whose bibkey is NO_BIBKEY (e.g. some frontmatter) raise
    ValueError from to_bibtex(); the sync must yield bibtex="" for those
    rather than crash."""
    fake_anthology = MagicMock()

    paper = _make_fake_paper("2023.acl-frontmatter.1", "Preface", "", [], "ACL", "2023", "http://x/pf")
    paper.to_bibtex.side_effect = ValueError("Cannot generate BibTeX entry without bibkey")
    fake_anthology.papers.return_value = [paper]

    monkeypatch.setattr("sync.anthology_source.Anthology", lambda datadir: fake_anthology)

    results = list(iter_papers("/fake/repo"))
    assert results[0]["bibtex"] == ""


def test_iter_papers_pdf_url_falls_back_when_no_pdf(monkeypatch):
    """Frontmatter and some older papers have no PDF reference (Paper.pdf is
    None); the sync must yield pdf_url="" for those rather than crash."""
    fake_anthology = MagicMock()

    paper = _make_fake_paper("2023.acl-frontmatter.1", "Preface", "", [], "ACL", "2023", "http://x/pf")
    paper.pdf = None
    fake_anthology.papers.return_value = [paper]

    monkeypatch.setattr("sync.anthology_source.Anthology", lambda datadir: fake_anthology)

    results = list(iter_papers("/fake/repo"))
    assert results[0]["pdf_url"] == ""