stack/tests/bib/test_format.py

"""Tests for bib.format — citation formatting functions."""

from __future__ import annotations

from bib.format import format_bibliography, format_citation
from bib.item import Download, Item, Manual, Regulation, Rule, Source

# ── Rule ─────────────────────────────────────────────────────────────


class TestFormatRule:
    def test_bluebook_full(self) -> None:
        r = Rule(
            title="Medicare and Medicaid Programs; CY 2026 PFS Final Rule",
            date_published="2025-12-31",
            fr_volume="90",
            fr_page="101174",
        )
        cite = format_citation(r, style="bluebook")
        assert "90 Fed. Reg. 101174" in cite
        assert "(Dec. 31, 2025)" in cite
        assert cite.endswith(".")

    def test_bluebook_no_page(self) -> None:
        r = Rule(title="Test Rule", fr_volume="90", date_published="2025-01-01")
        cite = format_citation(r, style="bluebook")
        assert "90 Fed. Reg. ___" in cite

    def test_apa_with_volume_and_page(self) -> None:
        r = Rule(
            title="CY 2026 PFS Final Rule",
            date_published="2025-11-15",
            fr_volume="90",
            fr_page="98452",
        )
        cite = format_citation(r)
        assert "Centers for Medicare & Medicaid Services" in cite
        assert "(2025)" in cite
        assert "CY 2026 PFS Final Rule" in cite
        assert "*Federal Register*" in cite
        assert "*90*" in cite
        assert "98452" in cite

    def test_apa_with_url(self) -> None:
        r = Rule(
            title="Test Rule",
            date_published="2025-01-01",
            fr_volume="90",
            fr_page="1000",
            url="https://example.com/rule",
        )
        cite = format_citation(r)
        assert "https://example.com/rule" in cite

    def test_apa_missing_date(self) -> None:
        r = Rule(title="Test Rule")
        cite = format_citation(r)
        assert "(n.d.)" in cite

    def test_apa_volume_only(self) -> None:
        r = Rule(title="Test", fr_volume="90", date_published="2025-01-01")
        cite = format_citation(r)
        assert "*90*" in cite


# ── Manual ───────────────────────────────────────────────────────────


class TestFormatManual:
    def test_bluebook_full(self) -> None:
        m = Manual(
            title="Physician Fee Schedule",
            manual_name="Claims Processing Manual",
            pub_number="100-04",
            chapter="12",
            date_published="2025-01-01",
        )
        cite = format_citation(m, style="bluebook")
        assert "Ctrs. for Medicare & Medicaid Servs." in cite
        assert "Claims Processing Manual" in cite
        assert "CMS Pub. 100-04" in cite
        assert "Ch. 12" in cite
        assert "(2025)" in cite

    def test_apa_full(self) -> None:
        m = Manual(
            title="Physician Fee Schedule",
            manual_name="Claims Processing Manual",
            pub_number="100-04",
            chapter="12",
            date_published="2025-01-01",
        )
        cite = format_citation(m)
        assert "Claims Processing Manual" in cite
        assert "Pub. 100-04" in cite
        assert "Ch. 12" in cite

    def test_apa_with_url(self) -> None:
        m = Manual(
            title="Test",
            manual_name="CPM",
            url="https://example.com/manual",
        )
        cite = format_citation(m)
        assert "https://example.com/manual" in cite

    def test_apa_uses_institution(self) -> None:
        m = Manual(title="Test", institution="Custom Institution")
        cite = format_citation(m)
        assert "Custom Institution" in cite


# ── Regulation ───────────────────────────────────────────────────────


class TestFormatRegulation:
    def test_bluebook_with_section(self) -> None:
        r = Regulation(
            cfr_title="42",
            cfr_section="414.22",
            effective_date="2025-01-01",
        )
        cite = format_citation(r, style="bluebook")
        assert cite == "42 C.F.R. § 414.22 (2025)."

    def test_bluebook_part_only(self) -> None:
        r = Regulation(
            cfr_title="42",
            cfr_part="414",
            effective_date="2025-01-01",
        )
        cite = format_citation(r, style="bluebook")
        assert cite == "42 C.F.R. pt. 414 (2025)."

    def test_apa_with_section(self) -> None:
        r = Regulation(
            cfr_title="42",
            cfr_part="414",
            cfr_section="414.22",
            effective_date="2025-01-01",
        )
        cite = format_citation(r)
        assert "42 C.F.R. § 414.22" in cite
        assert "(2025)" in cite

    def test_apa_part_only(self) -> None:
        r = Regulation(
            cfr_title="42",
            cfr_part="414",
            effective_date="2025-01-01",
        )
        cite = format_citation(r)
        assert "42 C.F.R. pt. 414" in cite

    def test_apa_custom_title(self) -> None:
        r = Regulation(
            cfr_title="42",
            cfr_part="414",
            title="RVU Methodology",
            effective_date="2025-01-01",
        )
        cite = format_citation(r)
        assert "*RVU Methodology*" in cite

    def test_apa_default_title_not_printed(self) -> None:
        r = Regulation(
            cfr_title="42",
            cfr_part="414",
            title="42 CFR Part 414",
            effective_date="2025-01-01",
        )
        cite = format_citation(r)
        assert "*42 CFR Part 414*" not in cite

    def test_apa_with_url(self) -> None:
        r = Regulation(
            cfr_title="42",
            cfr_part="414",
            effective_date="2025-01-01",
            url="https://ecfr.gov/title-42/part-414",
        )
        cite = format_citation(r)
        assert "https://ecfr.gov/title-42/part-414" in cite


# ── Download ─────────────────────────────────────────────────────────


class TestFormatDownload:
    def test_apa(self) -> None:
        d = Download(
            title="RVU26A",
            date_published="2026-01-01",
            url="https://example.com/rvu",
        )
        cite = format_citation(d)
        assert "Centers for Medicare & Medicaid Services" in cite
        assert "(2026)" in cite
        assert "*RVU26A*" in cite
        assert "https://example.com/rvu" in cite

    def test_custom_website_title(self) -> None:
        d = Download(
            title="Data",
            website_title="Custom Portal",
            date_published="2026-01-01",
        )
        cite = format_citation(d)
        assert "Custom Portal" in cite


# ── Source / Journal Articles ────────────────────────────────────────


class TestFormatSource:
    def test_apa_generic(self) -> None:
        s = Source(
            title="PY 2026 Financial Guarantees",
            institution="CMS Innovation Center",
            date_published="2025-06-01",
            url="https://example.com/source",
        )
        cite = format_citation(s)
        assert "CMS Innovation Center" in cite
        assert "(2025)" in cite
        assert "*PY 2026 Financial Guarantees*" in cite

    def test_unknown_institution(self) -> None:
        s = Source(title="Test")
        cite = format_citation(s)
        assert "Unknown" in cite


class TestFormatJournalArticle:
    def _make_article(self, **kwargs) -> Source:
        defaults = {
            "title": "Skin substitute outcomes in wound healing",
            "institution": "Journal of Wound Care",
            "date_published": "2023-06-15",
            "doc_type": "journal-article",
            "extra": (
                "PMID: 12345678\n"
                "DOI: 10.1234/jwc.2023.001\n"
                "Authors: Smith J; Jones K; Brown M\n"
                "Journal: Journal of Wound Care\n"
            ),
        }
        defaults.update(kwargs)
        return Source(**defaults)

    def test_apa_three_authors(self) -> None:
        s = self._make_article()
        cite = format_citation(s)
        assert "Smith et al." in cite
        assert "(2023)" in cite
        assert "Skin substitute outcomes" in cite
        assert "*Journal of Wound Care*" in cite
        assert "https://doi.org/10.1234/jwc.2023.001" in cite

    def test_apa_two_authors(self) -> None:
        s = self._make_article(extra="Authors: Smith J; Jones K\nDOI: 10.1/x\n")
        cite = format_citation(s)
        assert "Smith & Jones" in cite

    def test_apa_one_author(self) -> None:
        s = self._make_article(extra="Authors: Smith J\nDOI: 10.1/x\n")
        cite = format_citation(s)
        assert "Smith." in cite

    def test_apa_no_doi_uses_url(self) -> None:
        s = self._make_article(
            extra="PMID: 123\nAuthors: Smith J\n",
            url="https://example.com/article",
        )
        cite = format_citation(s)
        assert "https://example.com/article" in cite

    def test_jama_format(self) -> None:
        s = self._make_article()
        cite = format_citation(s, style="jama")
        assert "Smith et al." in cite
        assert "*Journal of Wound Care*" in cite
        assert "doi:10.1234/jwc.2023.001" in cite

    def test_jama_pmid_when_no_doi(self) -> None:
        s = self._make_article(extra="PMID: 99999\nAuthors: Smith J\n")
        cite = format_citation(s, style="jama")
        assert "PMID: 99999" in cite

    def test_detected_by_pmid(self) -> None:
        """Source with PMID in extra but no doc_type should still format as journal."""
        s = Source(
            title="Test Article",
            doc_type="",
            extra="PMID: 12345\nAuthors: Doe J\nJournal: Test J\n",
            date_published="2024-01-01",
        )
        cite = format_citation(s)
        assert "Doe." in cite
        assert "*Test J*" in cite


# ── Generic / Bibliography ───────────────────────────────────────────


class TestFormatGeneric:
    def test_unknown_item_type(self) -> None:
        item = Item(
            item_type="custom",
            title="Custom Item",
            institution="Test Org",
            date_published="2025-01-01",
        )
        cite = format_citation(item)
        assert "Test Org" in cite
        assert "*Custom Item*" in cite

    def test_generic_with_url(self) -> None:
        item = Item(
            item_type="custom",
            title="Custom",
            date_published="2025-01-01",
            url="https://example.com/custom",
        )
        cite = format_citation(item)
        assert "https://example.com/custom" in cite

    def test_generic_no_title(self) -> None:
        item = Item(
            item_type="custom",
            institution="Org",
            date_published="2025-01-01",
        )
        cite = format_citation(item)
        assert "Org. (2025)." in cite

    def test_generic_no_institution(self) -> None:
        item = Item(
            item_type="custom",
            title="Title",
            date_published="2025-01-01",
        )
        cite = format_citation(item)
        assert "Unknown" in cite


class TestFormatBibliography:
    def test_numbered_entries(self) -> None:
        items = [
            Rule(title="Rule A", date_published="2025-01-01"),
            Manual(title="Manual B"),
        ]
        bib = format_bibliography(items)
        lines = bib.strip().split("\n")
        assert len(lines) == 2
        assert lines[0].startswith("[1]")
        assert lines[1].startswith("[2]")

    def test_empty_list(self) -> None:
        bib = format_bibliography([])
        assert bib == ""

    def test_single_item(self) -> None:
        items = [Source(title="Test", date_published="2025-01-01")]
        bib = format_bibliography(items)
        assert bib.startswith("[1]")
        assert "\n" not in bib


# ── Date parsing edge cases (lines 82, 83, 87, 88) ────────────────


class TestParseDateEdge:
    def test_invalid_month_value_error(self) -> None:
        """Lines 82, 83: int() ValueError on non-numeric month."""
        from bib.format import _parse_date

        year, month, day = _parse_date("2025-XX-01")
        assert year == "2025"
        assert month == ""  # ValueError caught
        assert day == "1"

    def test_invalid_day_value_error(self) -> None:
        """Lines 87, 88: int() ValueError on non-numeric day."""
        from bib.format import _parse_date

        year, month, day = _parse_date("2025-01-XX")
        assert year == "2025"
        assert month == "Jan."
        assert day == ""  # ValueError caught


# ── Bluebook date variants (lines 102, 105-107) ───────────────────


class TestBluebookDate:
    def test_no_year(self) -> None:
        """Line 102: empty date → empty string."""
        from bib.format import _bluebook_date

        assert _bluebook_date("") == ""

    def test_month_and_day(self) -> None:
        """Lines 105-107: full date with month and day."""
        from bib.format import _bluebook_date

        result = _bluebook_date("2025-12-31")
        assert result == "Dec. 31, 2025"

    def test_month_only(self) -> None:
        """Line 106: date with month but no day."""
        from bib.format import _bluebook_date

        result = _bluebook_date("2025-06")
        assert result == "June 2025"


# ── Parse authors (line 121) ────────────────────────────────────────


class TestParseAuthors:
    def test_parses_authors(self) -> None:
        """Line 121: _parse_authors extracts last names."""
        from bib.format import _parse_authors

        result = _parse_authors("Authors: Smith J; Jones K\nOther: data")
        assert result == ["Smith", "Jones"]

    def test_no_authors_line(self) -> None:
        from bib.format import _parse_authors

        assert _parse_authors("No authors here") == []


# ── Bluebook rule no date (line 184) ───────────────────────────────


class TestRuleBluebookNoDate:
    def test_rule_bluebook_no_date(self) -> None:
        """Line 184: bluebook rule with no date ends with '.'."""
        r = Rule(title="Test Rule", fr_volume="90", fr_page="101174")
        cite = format_citation(r, style="bluebook")
        assert cite.endswith(".")
        assert "(" not in cite or ")" not in cite  # no date paren


# ── Source format doc_type attribute error (lines 316, 317) ─────────


class TestFormatSourceDocType:
    def test_source_without_doc_type_attr(self) -> None:
        """Lines 316, 317: Source with no doc_type still formats."""
        # Item base class doesn't have doc_type; _format_source handles
        # the AttributeError and falls to generic
        item = Item(
            item_type="source",
            title="Generic Source",
            institution="Test",
            date_published="2025-01-01",
        )
        # This should trigger the generic source path since Item
        # doesn't have doc_type attr directly accessible via format logic
        cite = format_citation(item)
        assert "Test" in cite


# ── Journal with no authors (line 341) ──────────────────────────────


class TestJournalNoAuthors:
    def test_no_authors_uses_institution(self) -> None:
        """Line 341: journal with no Authors line uses institution."""
        s = Source(
            title="Test",
            doc_type="journal-article",
            extra="PMID: 123\nJournal: Test J\n",
            institution="Fallback Journal",
            date_published="2024-01-01",
        )
        cite = format_citation(s)
        assert "Fallback Journal" in cite

    def test_no_authors_unknown(self) -> None:
        """Line 341: journal with no Authors and no institution uses 'Unknown'."""
        s = Source(
            title="Test",
            doc_type="journal-article",
            extra="PMID: 123\n",
            date_published="2024-01-01",
        )
        cite = format_citation(s)
        assert "Unknown" in cite


# ── JAMA two authors and fallback (lines 368, 372) ─────────────────


class TestJamaTwoAuthors:
    def test_two_authors_jama(self) -> None:
        """Line 368: JAMA format with two authors uses comma."""
        s = Source(
            title="Test",
            doc_type="journal-article",
            extra="Authors: Smith J; Jones K\nDOI: 10.1/x\nJournal: J\n",
            date_published="2024-01-01",
        )
        cite = format_citation(s, style="jama")
        assert "Smith, Jones" in cite

    def test_no_authors_jama_unknown(self) -> None:
        """Line 372: JAMA with no authors/institution uses 'Unknown'."""
        s = Source(
            title="Test",
            doc_type="journal-article",
            extra="PMID: 123\n",
            date_published="2024-01-01",
        )
        cite = format_citation(s, style="jama")
        assert "Unknown" in cite