stack/tests/bib/test_iom_exercise.py

"""Exercise bib.iom — covers fetch_chapters regex parsing, download, check_future."""

from __future__ import annotations

from unittest.mock import MagicMock

from bib.iom import (
    IOMEntry,
    _existing_attachment_hash,
    _futurepdf_anchor,
    _prune_stale,
    _sha256,
    check_future_updates,
    download_attachments,
    fetch_chapters,
    fetch_index,
    ingest_all,
    ingest_entry,
)

_ENTRY = IOMEntry(
    pub="100-04",
    title="Medicare Claims Processing Manual",
    landing_url="https://www.cms.gov/regulations-and-guidance/guidance/manuals/internet-only-manuals-ioms-items/cms018912",
)


class TestFetchChaptersRegex:
    def _client(self, html):
        client = MagicMock()
        resp = MagicMock()
        resp.text = html
        resp.status_code = 200
        client.get.return_value = resp
        return client

    def test_matches_chapters(self):
        html = """
        <a href="/regulations-and-guidance/guidance/manuals/downloads/clm104c01.pdf">
          Chapter 1 - General Billing Requirements
        </a>
        <a href="/regulations-and-guidance/guidance/manuals/downloads/clm104c02.pdf">
          Chapter 2 - Submission of Claims
        </a>
        """
        chapters = fetch_chapters(self._client(html), _ENTRY)
        assert len(chapters) == 2
        assert chapters[0].chapter == "1"
        assert "General Billing" in chapters[0].title

    def test_skips_supplements(self):
        html = """
        <a href="/regulations-and-guidance/guidance/manuals/downloads/clm104c01.pdf">
          Chapter 1 - General
        </a>
        <a href="/regulations-and-guidance/guidance/manuals/downloads/crosswalk.pdf">
          Crosswalk of Changes
        </a>
        """
        chapters = fetch_chapters(self._client(html), _ENTRY)
        assert len(chapters) == 1

    def test_whole_pub_fallback(self):
        html = """
        <a href="/regulations-and-guidance/guidance/manuals/downloads/pub100_18.pdf">
          Pub 100-18 - Medicare Prescription Drug Benefit Manual
        </a>
        """
        chapters = fetch_chapters(self._client(html), _ENTRY)
        assert len(chapters) == 1
        assert chapters[0].chapter == ""

    def test_dedup_chapters(self):
        html = """
        <a href="/regulations-and-guidance/guidance/manuals/downloads/clm104c01.pdf">
          Chapter 1 - General
        </a>
        <a href="/regulations-and-guidance/guidance/manuals/downloads/clm104c01v2.pdf">
          Chapter 1 - General (Updated)
        </a>
        """
        chapters = fetch_chapters(self._client(html), _ENTRY)
        assert len(chapters) == 1

    def test_part_chapters(self):
        html = """
        <a href="/regulations-and-guidance/guidance/manuals/downloads/ncd103c03p1.pdf">
          Chapter 3, Part 1 — Section A
        </a>
        <a href="/regulations-and-guidance/guidance/manuals/downloads/ncd103c03p2.pdf">
          Chapter 3, Part 2 — Section B
        </a>
        """
        chapters = fetch_chapters(self._client(html), _ENTRY)
        assert len(chapters) == 2
        assert chapters[0].chapter in ("3P1", "3p1")


class TestFetchIndex:
    def test_real_html_pattern(self):
        html = """
        <a href="/regulations-and-guidance/guidance/manuals/internet-only-manuals-ioms-items/cms018912"
           class="title-link">
          100-04
        </a>
        <div><label>Title</label>Medicare Claims Processing Manual</div>
        """
        client = MagicMock()
        resp = MagicMock()
        resp.text = html
        resp.status_code = 200
        client.get.return_value = resp
        entries = fetch_index(client)
        assert len(entries) == 1
        assert entries[0].pub == "100-04"


class TestPruneStale:
    def test_prunes(self):
        store = MagicMock()
        con = MagicMock()
        store._con.return_value = con
        con.execute.return_value.fetchall.return_value = [
            {"key": "K1", "url": "https://old.pdf"},
            {"key": "K2", "url": "https://live.pdf"},
        ]
        n = _prune_stale(store, "100-04", {"https://live.pdf"})
        assert n == 1
        store.delete.assert_called_once_with("K1")


class TestIngestEntryDeep:
    def test_with_chapters(self):
        store = MagicMock()
        store.upsert.return_value = "KEY1"
        store._con.return_value = MagicMock(
            execute=MagicMock(
                return_value=MagicMock(fetchall=MagicMock(return_value=[]))
            )
        )

        html = """
        <a href="/regulations-and-guidance/guidance/manuals/downloads/clm104c01.pdf">
          Chapter 1 - General
        </a>
        """
        client = MagicMock()
        resp = MagicMock()
        resp.text = html
        resp.status_code = 200
        client.get.return_value = resp

        keys = ingest_entry(store, client, _ENTRY)
        assert len(keys) == 1
        store.upsert.assert_called_once()


class TestIngestAllFiltered:
    def test_with_pub_filter(self):
        store = MagicMock()
        store.upsert.return_value = "KEY1"
        store._con.return_value = MagicMock(
            execute=MagicMock(
                return_value=MagicMock(fetchall=MagicMock(return_value=[]))
            )
        )

        index_html = """
        <a href="/regulations-and-guidance/guidance/manuals/internet-only-manuals-ioms-items/cms018912">
          100-04
        </a>
        <div><label>Title</label>Claims Processing Manual</div>
        """
        chapter_html = """
        <a href="/regulations-and-guidance/guidance/manuals/downloads/clm104c01.pdf">
          Chapter 1 - General
        </a>
        """
        client = MagicMock()
        resp1 = MagicMock(text=index_html, status_code=200)
        resp2 = MagicMock(text=chapter_html, status_code=200)
        client.get.side_effect = [resp1, resp2]

        result = ingest_all(store, client, pubs=["100-04"])
        assert isinstance(result, dict)

    def test_skips_unmatched_pubs(self):
        store = MagicMock()
        index_html = """
        <a href="/regulations-and-guidance/guidance/manuals/internet-only-manuals-ioms-items/cms018912">
          100-04
        </a>
        <div><label>Title</label>Claims Processing Manual</div>
        """
        client = MagicMock()
        resp = MagicMock(text=index_html, status_code=200)
        client.get.return_value = resp

        result = ingest_all(store, client, pubs=["100-99"])
        assert result == {}

    def test_handles_exception(self):
        store = MagicMock()
        index_html = """
        <a href="/regulations-and-guidance/guidance/manuals/internet-only-manuals-ioms-items/cms018912">
          100-04
        </a>
        <div><label>Title</label>Claims Processing Manual</div>
        """
        client = MagicMock()
        resp1 = MagicMock(text=index_html, status_code=200)
        resp2 = MagicMock()
        resp2.raise_for_status.side_effect = Exception("fail")
        client.get.side_effect = [resp1, resp2]

        result = ingest_all(store, client)
        assert result.get("100-04") == 0


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

    def test_with_file(self, tmp_path):
        f = tmp_path / "test.pdf"
        f.write_bytes(b"pdf content")
        store = MagicMock()
        con = MagicMock()
        store._con.return_value = con
        con.execute.return_value.fetchone.return_value = {"storage_path": str(f)}
        result = _existing_attachment_hash(store, "K1")
        assert result == _sha256(f)


class TestDownloadAttachments:
    def test_downloads_and_attaches(self, tmp_path):
        store = MagicMock()
        store._db_path = str(tmp_path / "bib.sqlite")
        item = MagicMock()
        item.url = "https://cms.gov/manuals/downloads/ch1.pdf"
        item.key = "K1"
        item.title = "Chapter 1"
        row = {"extra_json": '{"pub_number": "100-04"}'}
        item.to_row.return_value = row
        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

        result = download_attachments(store, client, tmp_dir=tmp_path / "dl")
        assert result.get("100-04", 0) >= 1


class TestFuturepdfAnchor:
    def test_creates_new(self):
        store = MagicMock()
        store.list_items.return_value = []
        store.upsert.return_value = "FKEY"
        key = _futurepdf_anchor(store)
        assert key == "FKEY"
        store.upsert.assert_called_once()

    def test_returns_existing(self):
        store = MagicMock()
        item = MagicMock()
        item.key = "EXISTING"
        store.list_items.return_value = [item]
        assert _futurepdf_anchor(store) == "EXISTING"


class TestCheckFutureUpdates:
    def test_changed(self, tmp_path):
        store = MagicMock()
        store._db_path = str(tmp_path / "bib.sqlite")
        store.list_items.return_value = []
        store.upsert.return_value = "FKEY"
        store._con.return_value.execute.return_value.fetchone.return_value = None

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

        changed, digest = check_future_updates(store, client, tmp_dir=tmp_path / "dl")
        assert changed is True
        assert len(digest) == 64