stack/tests/bib/test_regulations_gov.py

"""Tests for bib.regulations_gov — reg.gov v4 API client."""

from __future__ import annotations

from unittest.mock import MagicMock, patch

import pytest

from bib.regulations_gov import (
    Client,
    Comment,
    _byline,
    _docket_from_comment_id,
    _filename_from,
    _hash,
    _parse_comment,
    _reg_date,
    _slug,
    upsert_comment,
)

# ── Pure helpers ──────────────────────────────────────────────────


class TestRegDate:
    def test_iso_to_space(self):
        assert _reg_date("2017-08-30T18:35:48Z") == "2017-08-30 18:35:48"

    def test_already_space(self):
        assert _reg_date("2017-08-30 18:35:48") == "2017-08-30 18:35:48"

    def test_empty(self):
        assert _reg_date("") == ""

    def test_none(self):
        assert _reg_date("") == ""


class TestSlug:
    def test_basic(self):
        assert _slug("American Medical Association") == "american-medical-association"

    def test_special(self):
        assert _slug("Test & Co., Inc.") == "test-co-inc"

    def test_truncate(self):
        assert len(_slug("x" * 100)) <= 40


class TestFilenameFrom:
    def test_url(self):
        assert (
            _filename_from(
                "https://downloads.regulations.gov/CMS-2021-0119-26001/attachment_1.pdf"
            )
            == "attachment_1.pdf"
        )

    def test_with_query(self):
        assert _filename_from("https://x.com/file.pdf?token=abc") == "file.pdf"

    def test_empty(self):
        assert _filename_from("") == ""


class TestHash:
    def test_deterministic(self):
        assert _hash("test") == _hash("test")
        assert len(_hash("test")) == 10


class TestByline:
    def test_org(self):
        c = Comment(
            id="1",
            title="",
            posted_date="",
            received_date="",
            docket_id="",
            comment_on_id="",
            organization="AMA",
        )
        assert _byline(c) == "AMA"

    def test_name(self):
        c = Comment(
            id="1",
            title="",
            posted_date="",
            received_date="",
            docket_id="",
            comment_on_id="",
            first_name="John",
            last_name="Doe",
        )
        assert _byline(c) == "John Doe"

    def test_empty(self):
        c = Comment(
            id="1",
            title="",
            posted_date="",
            received_date="",
            docket_id="",
            comment_on_id="",
        )
        assert _byline(c) == ""


class TestDocketFromCommentId:
    def test_standard(self):
        assert _docket_from_comment_id("CMS-2025-0304-14107") == "CMS-2025-0304"

    def test_short(self):
        assert _docket_from_comment_id("AB") == "AB"


# ── _parse_comment ────────────────────────────────────────────────


class TestParseComment:
    def test_basic(self):
        row = {
            "id": "CMS-2017-0092-0002",
            "attributes": {
                "title": "Test Comment",
                "postedDate": "2017-07-18T00:00:00Z",
                "receivedDate": "2017-07-15T00:00:00Z",
                "docketId": "CMS-2017-0092",
                "commentOnId": "0900006482921ba1",
                "firstName": "Jane",
                "lastName": "Doe",
                "organization": "AMA",
                "comment": "This is the body.",
                "attachmentCount": 2,
            },
        }
        c = _parse_comment(row)
        assert c.id == "CMS-2017-0092-0002"
        assert c.title == "Test Comment"
        assert c.posted_date == "2017-07-18"
        assert c.organization == "AMA"
        assert c.comment_text == "This is the body."
        assert c.attachment_count == 2

    def test_missing_attrs(self):
        c = _parse_comment({"id": "X", "attributes": {}})
        assert c.id == "X"
        assert c.title == ""
        assert c.attachment_count == 0


# ── upsert_comment ────────────────────────────────────────────────


class TestUpsertComment:
    def test_creates_item(self):
        store = MagicMock()
        store.upsert.return_value = "KEY1"
        c = Comment(
            id="CMS-2017-0092-0002",
            title="Test",
            posted_date="2017-07-18",
            received_date="2017-07-15",
            docket_id="CMS-2017-0092",
            comment_on_id="obj123",
            organization="AMA",
        )
        key = upsert_comment(store, c, cms_id="CMS-1676-P")
        assert key == "KEY1"
        store.upsert.assert_called_once()
        item = store.upsert.call_args[0][0]
        assert "AMA" in item.title
        assert item.url == "https://www.regulations.gov/comment/CMS-2017-0092-0002"


# ── Client (mocked httpx) ────────────────────────────────────────


class TestClient:
    @patch.dict("os.environ", {"REGULATIONS_GOV_API_KEY": "test-key"})
    def test_init(self):
        c = Client(sleep=0)
        assert c._key == "test-key"
        c.close()

    @patch.dict("os.environ", {"REGULATIONS_GOV_API_KEY": ""})
    def test_init_no_key(self):
        with pytest.raises(RuntimeError, match="REGULATIONS_GOV_API_KEY"):
            Client()

    @patch.dict("os.environ", {"REGULATIONS_GOV_API_KEY": "test-key"})
    def test_get_retry_429(self):
        mock_client = MagicMock()
        resp_429 = MagicMock()
        resp_429.status_code = 429
        resp_ok = MagicMock()
        resp_ok.status_code = 200
        resp_ok.json.return_value = {"data": []}
        resp_ok.raise_for_status = MagicMock()
        mock_client.get.side_effect = [resp_429, resp_ok]

        with patch("bib.regulations_gov.time.sleep"):
            c = Client(sleep=0, client=mock_client)
            result = c._get("/test")
        assert result == {"data": []}

    @patch.dict("os.environ", {"REGULATIONS_GOV_API_KEY": "test-key"})
    def test_resolve_docket(self):
        mock_client = MagicMock()
        resp = MagicMock()
        resp.status_code = 200
        resp.json.return_value = {
            "data": [{"attributes": {"docketId": "CMS-2017-0092"}}]
        }
        resp.raise_for_status = MagicMock()
        mock_client.get.return_value = resp

        with patch("bib.regulations_gov.time.sleep"):
            c = Client(sleep=0, client=mock_client)
            result = c.resolve_docket("CMS-1676-P")
        assert result == "CMS-2017-0092"

    @patch.dict("os.environ", {"REGULATIONS_GOV_API_KEY": "test-key"})
    def test_resolve_docket_not_found(self):
        mock_client = MagicMock()
        resp = MagicMock()
        resp.status_code = 200
        resp.json.return_value = {"data": []}
        resp.raise_for_status = MagicMock()
        mock_client.get.return_value = resp

        with patch("bib.regulations_gov.time.sleep"):
            c = Client(sleep=0, client=mock_client)
            assert c.resolve_docket("NONEXISTENT") is None