stack/tests/bib/test_oig_exercise.py

"""Exercise bib.oig — extract_docs regex, classify, download_attachments."""

from __future__ import annotations

from unittest.mock import MagicMock

from bib.oig import (
    OIGDoc,
    _build_item,
    _classify_sector,
    _classify_type,
    _existing_attachment_hash,
    _extract_docs,
    _sha256,
    download_attachments,
    fetch_alerts,
    fetch_cpgs,
    ingest_all,
)


class TestClassifyTypeDeep:
    def test_sfa_by_href(self):
        assert _classify_type("Some Alert", "/special-fraud-alerts/page") == "sfa"

    def test_sab_by_href(self):
        assert _classify_type("Some Bulletin", "/special-advisory-bulletins/x") == "sab"

    def test_cpg_by_href(self):
        assert _classify_type("Some Guidance", "/compliance-guidance/x") == "cpg"

    def test_other(self):
        assert _classify_type("Random", "/random") == "other"


class TestClassifySectorDeep:
    def test_nursing_home(self):
        assert _classify_sector("Nursing Home Compliance") != ""

    def test_physician(self):
        assert _classify_sector("Individual and Small Group Physician Practices") != ""

    def test_no_match(self):
        assert _classify_sector("Unrelated Document") == ""


class TestExtractDocsDeep:
    def test_real_pattern(self):
        html = """
        <a href="/documents/compliance-guidance/cpg-hospitals.pdf">
            Compliance Program Guidance for Hospitals
        </a>
        <a href="/documents/special-fraud-alerts/sfa-2024.pdf">
            Special Fraud Alert: Kickbacks
        </a>
        """
        docs = _extract_docs(html)
        assert len(docs) == 2
        assert docs[0][1] == "Compliance Program Guidance for Hospitals"

    def test_dedup(self):
        html = """
        <a href="/documents/doc1.pdf">Doc One</a>
        <a href="/documents/doc1.pdf">Doc One Again</a>
        """
        docs = _extract_docs(html)
        assert len(docs) == 1

    def test_skip_hints(self):
        html = """
        <a href="/documents/doc1.pdf">Download</a>
        <a href="/documents/doc2.pdf">Real Title</a>
        """
        docs = _extract_docs(html)
        assert len(docs) == 1
        assert docs[0][1] == "Real Title"


class TestFetchCpgsDeep:
    def test_parses_docs(self):
        client = MagicMock()
        resp = MagicMock()
        resp.text = """
        <a href="/documents/compliance-guidance/cpg-hospitals.pdf">
            CPG for Hospitals
        </a>
        """
        resp.status_code = 200
        client.get.return_value = resp
        docs = fetch_cpgs(client)
        assert len(docs) == 1
        assert docs[0].guidance_type == "cpg"
        assert docs[0].sector == "hospitals"


class TestFetchAlertsDeep:
    def test_parses_docs(self):
        client = MagicMock()
        resp = MagicMock()
        resp.text = """
        <a href="/documents/special-fraud-alerts/sfa-2024.pdf">
            Special Fraud Alert: Laboratory Kickbacks
        </a>
        """
        resp.status_code = 200
        client.get.return_value = resp
        docs = fetch_alerts(client)
        assert len(docs) == 1
        assert docs[0].guidance_type == "sfa"


class TestBuildItemDeep:
    def test_all_types(self):
        for gtype in ("cpg", "sfa", "sab", "ea", "open_letter", "other"):
            doc = OIGDoc(
                title=f"Test {gtype}",
                url=f"https://oig.hhs.gov/{gtype}",
                guidance_type=gtype,
                sector="hospitals" if gtype == "cpg" else "",
            )
            item = _build_item(doc)
            assert "agency:oig" in item.tags
            assert f"guidance:{gtype}" in item.tags


class TestIngestAllFiltered:
    def test_cpg_only(self):
        store = MagicMock()
        store.upsert.return_value = "K1"
        client = MagicMock()
        resp = MagicMock()
        resp.text = """
        <a href="/documents/compliance-guidance/cpg1.pdf">CPG Doc</a>
        """
        resp.status_code = 200
        client.get.return_value = resp

        result = ingest_all(store, client, kinds=["cpg"])
        assert result["cpg"] >= 1
        assert result["alerts"] == 0


class TestSha256:
    def test_deterministic(self, tmp_path):
        f = tmp_path / "test.bin"
        f.write_bytes(b"hello")
        h = _sha256(f)
        assert len(h) == 64
        assert _sha256(f) == h


class TestExistingAttachmentHash:
    def test_none(self):
        store = MagicMock()
        store._con.return_value.execute.return_value.fetchone.return_value = None
        assert _existing_attachment_hash(store, "K1") is None


class TestDownloadAttachments:
    def test_downloads(self):
        store = MagicMock()
        store._db_path = "/tmp/test.sqlite"
        item = MagicMock()
        item.url = "https://oig.hhs.gov/doc.pdf"
        item.key = "K1"
        item.title = "Test Doc"
        store.list_items.return_value = [item]
        store._con.return_value.execute.return_value.fetchone.return_value = None

        client = MagicMock()
        resp = MagicMock()
        resp.content = b"pdf bytes"
        resp.status_code = 200
        client.get.return_value = resp

        n = download_attachments(store, client)
        assert n >= 1