"""

Tests for ingest/pollers/meetings.py

Coverage:

  • _parse_senate_xml()           -  full XML, empty identifier skipped

  • _parse_house_day_page()       -  normal row, subcommittee row, rescheduled row

  • _parse_house_event_type()     -  hearing / markup / business_meeting / unknown

  • _parse_floor_xml()            -  typical floor XML structure

  • _build_datetime_iso()         -  12h and 24h formats, missing inputs

  • _norm_senate_type()           -  all known types + unknown fallback

  • _norm_house_type()            -  prefix matching

  • _detect_status()              -  rescheduled / canceled / normal

  • _ensure_tables()              -  idempotent schema creation

  • _upsert_meeting() idempotency  -  double-upsert does not duplicate

  • poll_senate() with mocked HTTP

"""

import os

import sqlite3

import tempfile

from datetime import date

from unittest.mock import patch, MagicMock

import pytest

from ingest.db import get_connection, init_schema

from ingest.pollers.meetings import (

    _parse_senate_xml,

    _parse_house_day_page,

    _parse_house_event_type,

    _parse_floor_xml,

    _build_datetime_iso,

    _norm_senate_type,

    _norm_house_type,

    _detect_status,

    _ensure_tables,

    _upsert_meeting,

    _most_recent_monday,

    SOURCE_SENATE,

    SOURCE_HOUSE_CMT,

    SOURCE_HOUSE_FLR,

    SENATE_PAGE_URL,

)

def _make_db() -> tuple:

    """Return (conn, path) for a fresh temp DB."""

    path = tempfile.mktemp(suffix=".db")

    conn = get_connection(path)

    init_schema(conn)

    _ensure_tables(conn)

    return conn, path

@pytest.mark.parametrize("date_str,time_str,expected", [

    ("2026-06-03", "10:00 AM", "2026-06-03T10:00:00"),

    ("2026-07-15", "02:30 PM", "2026-07-15T14:30:00"),

    ("2026-07-15", "09:45 AM", "2026-07-15T09:45:00"),

    ("2026-07-15", "14:30:00", "2026-07-15T14:30:00"),

    ("2026-06-03", None, None),

    (None, "10:00 AM", None),

    ("", "10:00 AM", None),

    ("2026-06-03", "", None),

    ("2026-06-03", "bad time", None),

])

def test_build_datetime_iso(date_str, time_str, expected):

    assert _build_datetime_iso(date_str, time_str) == expected

@pytest.mark.parametrize("raw,expected", [

    ("Open Hearing", "hearing"),

    ("Closed Hearing", "hearing"),

    ("Field Hearing", "hearing"),

    ("Open Markup", "markup"),

    ("Closed Markup", "markup"),

    ("Open Business Meeting", "business_meeting"),

    ("Closed Business Meeting", "business_meeting"),

    ("Executive Session", "business_meeting"),

    ("Unknown Type", "other"),

    ("", "other"),

])

def test_norm_senate_type(raw, expected):

    assert _norm_senate_type(raw) == expected

@pytest.mark.parametrize("h1,expected", [

    ('Hearing: "The Budget Request"', "hearing"),

    ("Markup: FY 2027 NDAA", "markup"),

    ("Business Meeting to Consider Pending Business", "business_meeting"),

    ("Full Committee Hearing on Defense", "hearing"),

    ("Unknown prefix: Something", "other"),

    ("", "other"),

])

def test_norm_house_type(h1, expected):

    assert _norm_house_type(h1) == expected

@pytest.mark.parametrize("text,expected", [

    ("Rescheduled", "postponed"),

    ("Postponed", "postponed"),

    ("Canceled", "canceled"),

    ("Cancelled", "canceled"),

    ("", "scheduled"),

    ("Some other note", "scheduled"),

])

def test_detect_status(text, expected):

    assert _detect_status(text) == expected

def test_most_recent_monday_on_monday():

    mon = date(2026, 6, 22)

    assert _most_recent_monday(mon) == mon

def test_most_recent_monday_on_thursday():

    thu = date(2026, 6, 25)

    assert _most_recent_monday(thu) == date(2026, 6, 22)

def test_most_recent_monday_on_sunday():

    sun = date(2026, 6, 28)

    assert _most_recent_monday(sun) == date(2026, 6, 22)

SENATE_XML_SAMPLE = b"""<?xml version="1.0" encoding="UTF-8"?>

<css_meetings_scheduled>

  <meeting>

    <cmte_code/>

    <committee/>

    <type/>

    <date>26-JUN-2026</date>

    <date_iso_8601>2026-06-26</date_iso_8601>

    <day_of_week>Today</day_of_week>

    <time>12:00:00</time>

    <room/>

    <matter>No committee hearings scheduled</matter>

    <video_url/>

    <senate_cable_channel/>

  </meeting>

  <meeting>

    <identifier>338613</identifier>

    <last_update>06-26-2026 11:26:27 AM</last_update>

    <last_update_iso_8601>2026-06-26T11:26:27.000000Z-04:00</last_update_iso_8601>

    <cmte_code>SSHR00</cmte_code>

    <committee>Health, Education, Labor, and Pensions</committee>

    <type>Open Business Meeting</type>

    <date>15-JUL-2026 09:45 AM</date>

    <date_iso_8601>2026-07-15</date_iso_8601>

    <day_of_week>Wednesday</day_of_week>

    <time>09:45 AM</time>

    <time_iso_8601>09:45:00</time_iso_8601>

    <room>SD-430</room>

    <matter>Business meeting to consider pending calendar business.</matter>

  </meeting>

  <meeting>

    <identifier>338612</identifier>

    <last_update>06-24-2026 05:53:16 PM</last_update>

    <last_update_iso_8601>2026-06-24T17:53:16.000000Z-04:00</last_update_iso_8601>

    <cmte_code>SSAS17</cmte_code>

    <committee>Armed Services</committee>

    <sub_cmte>Personnel Subcommittee</sub_cmte>

    <type>Open Hearing</type>

    <date>15-JUL-2026 02:30 PM</date>

    <date_iso_8601>2026-07-15</date_iso_8601>

    <day_of_week>Wednesday</day_of_week>

    <time>02:30 PM</time>

    <time_iso_8601>14:30:00</time_iso_8601>

    <room>SR-222</room>

    <matter>Hearings to examine the TRICARE pharmacy program.</matter>

  </meeting>

</css_meetings_scheduled>

"""

def test_parse_senate_xml_skips_placeholder():

    """The first entry (no <identifier>) is a placeholder and must be skipped."""

    meetings = _parse_senate_xml(SENATE_XML_SAMPLE)

    assert len(meetings) == 2

def test_parse_senate_xml_business_meeting():

    meetings = _parse_senate_xml(SENATE_XML_SAMPLE)

    bm = next(m for m in meetings if m["external_id"] == "338613")

    assert bm["chamber"] == "Senate"

    assert bm["committee"] == "Health, Education, Labor, and Pensions"

    assert bm["date"] == "2026-07-15"

    assert bm["time"] == "09:45 AM"

    assert bm["datetime_iso"] == "2026-07-15T09:45:00"

    assert bm["type"] == "business_meeting"

    assert bm["location"] == "SD-430"

    assert bm["status"] == "scheduled"

    assert bm["source_url"] == SENATE_PAGE_URL

def test_parse_senate_xml_subcommittee_hearing():

    meetings = _parse_senate_xml(SENATE_XML_SAMPLE)

    h = next(m for m in meetings if m["external_id"] == "338612")

    assert h["type"] == "hearing"

    assert h["committee"] == "Personnel Subcommittee (Armed Services)"

    assert h["title"] == "Hearings to examine the TRICARE pharmacy program."

    assert h["datetime_iso"] == "2026-07-15T14:30:00"

def test_parse_senate_xml_bad_xml():

    result = _parse_senate_xml(b"not xml at all <broken>")

    assert result == []

HOUSE_DAY_HTML = """

<html><body>

<table id="MainContent_GridViewMeetings">

<tr><th>Title</th><th>Time</th><th>Location</th></tr>

<tr>

  <td>

    <a href="ByEvent.aspx?EventID=119243"

       title=""A Review of the FY2027 DHS Budget"">

       "A Review of the FY2027 DHS Budget"</a>

    <br />

    <span class="text-tiny" title="Committee on Homeland Security">

        Committee on Homeland Security

    </span>

  </td>

  <td><span class="text-small">10:00 AM</span></td>

  <td><span class="text-small">310 CHOB</span></td>

</tr>

<tr>

  <td>

    <a href="ByEvent.aspx?EventID=119340"

       title="Future of United States Maritime">

       Future of United States Maritime</a>

    <br />

    <span class="text-tiny" title="Subcommittee on Coast Guard">

        Subcommittee on Coast Guard

         (Committee on Transportation and Infrastructure)

    </span>

  </td>

  <td><span class="text-small">10:00 AM</span></td>

  <td><span class="text-small">2167 RHOB</span></td>

</tr>

<tr>

  <td>

    <a href="ByEvent.aspx?EventID=118474"

       title="Crypto Capital of the World">

       Crypto Capital of the World</a>

    <span class='muted'>(Rescheduled)</span>

    <br />

    <span class="text-tiny" title="Subcommittee on Oversight">

        Subcommittee on Oversight

         (Committee on Ways and Means)

    </span>

  </td>

  <td><span class="text-small">9:00 AM</span></td>

  <td><span class="text-small">1100 LHOB</span></td>

</tr>

</table>

</body></html>

"""

def test_parse_house_day_page_count():

    d = date(2026, 6, 3)

    meetings = _parse_house_day_page(HOUSE_DAY_HTML, d)

    assert len(meetings) == 3

def test_parse_house_day_page_basic():

    d = date(2026, 6, 3)

    meetings = _parse_house_day_page(HOUSE_DAY_HTML, d)

    m = next(m for m in meetings if m["external_id"] == "119243")

    assert m["chamber"] == "House"

    assert m["date"] == "2026-06-03"

    assert m["time"] == "10:00 AM"

    assert m["location"] == "310 CHOB"

    assert m["status"] == "scheduled"

    assert m["type"] is None

    assert "docs.house.gov" in m["source_url"]

    assert m["datetime_iso"] == "2026-06-03T10:00:00"

def test_parse_house_day_page_subcommittee():

    d = date(2026, 6, 3)

    meetings = _parse_house_day_page(HOUSE_DAY_HTML, d)

    m = next(m for m in meetings if m["external_id"] == "119340")

    assert "Subcommittee on Coast Guard" in m["committee"]

    assert "Transportation" in m["committee"]

def test_parse_house_day_page_rescheduled():

    d = date(2026, 6, 3)

    meetings = _parse_house_day_page(HOUSE_DAY_HTML, d)

    m = next(m for m in meetings if m["external_id"] == "118474")

    assert m["status"] == "postponed"

def test_parse_house_day_page_no_table():

    """Page without GridView returns empty list."""

    d = date(2026, 6, 3)

    meetings = _parse_house_day_page("<html><body>No meetings found.</body></html>", d)

    assert meetings == []

HOUSE_EVENT_HEARING = """

<html><body>

<div id="previewPanel">

  <div class="well">

    <h1>Hearing: "A Review of the FY2027 DHS Budget"

      <small><p>Committee on Homeland Security</p></small>

    </h1>

  </div>

</div>

</body></html>

"""

HOUSE_EVENT_MARKUP = """

<html><body>

<div id="previewPanel">

  <h1>Markup: FY2027 NDAA

    <small><p>Committee on Armed Services</p></small>

  </h1>

</div>

</body></html>

"""

HOUSE_EVENT_BUSINESS = """

<html><body>

<h1>Business Meeting to Consider Pending Business</h1>

</body></html>

"""

def test_parse_house_event_type_hearing():

    assert _parse_house_event_type(HOUSE_EVENT_HEARING) == "hearing"

def test_parse_house_event_type_markup():

    assert _parse_house_event_type(HOUSE_EVENT_MARKUP) == "markup"

def test_parse_house_event_type_business_meeting():

    assert _parse_house_event_type(HOUSE_EVENT_BUSINESS) == "business_meeting"

def test_parse_house_event_type_fallback():

    assert _parse_house_event_type("<html><h1>Unknown Header</h1></html>") == "other"

FLOOR_XML = b"""<?xml version="1.0" encoding="UTF-8"?>

<floorschedule congress-num="119" week-date="2026-06-22"

    create-date="2026-06-16T17:55:16.977">

  <category type="Items that may be considered under suspension of the rules"

            sort-order="1">

    <floor-items>

      <floor-item id="409612">

        <legis-num>H.R. 7396</legis-num>

        <floor-text>Native American Entrepreneurial Opportunity Act</floor-text>

      </floor-item>

      <floor-item id="409613">

        <legis-num>H.R. 7401</legis-num>

        <floor-text>Small Business Lending Fraud Prevention Act</floor-text>

      </floor-item>

    </floor-items>

  </category>

  <category type="Items considered pursuant to a rule" sort-order="2">

    <floor-items>

      <floor-item id="409620">

        <legis-num>H.R. 1</legis-num>

        <floor-text>Big Beautiful Bill Act</floor-text>

      </floor-item>

    </floor-items>

  </category>

</floorschedule>

"""

def test_parse_floor_xml_count():

    monday = date(2026, 6, 22)

    source_url = "https://docs.house.gov/floor/Default.aspx?date=2026-06-22"

    meetings = _parse_floor_xml(FLOOR_XML, monday, source_url)

    assert len(meetings) == 3

def test_parse_floor_xml_structure():

    monday = date(2026, 6, 22)

    source_url = "https://docs.house.gov/floor/Default.aspx?date=2026-06-22"

    meetings = _parse_floor_xml(FLOOR_XML, monday, source_url)

    first = meetings[0]

    assert first["chamber"] == "House"

    assert first["type"] == "floor"

    assert first["date"] == "2026-06-22"

    assert first["location"] == "House Floor"

    assert first["status"] == "scheduled"

    assert "H.R. 7396" in first["title"]

    assert first["source_url"] == source_url

    assert first["external_id"].startswith("20260622_")

def test_parse_floor_xml_external_id_unique():

    monday = date(2026, 6, 22)

    source_url = "https://docs.house.gov/floor/Default.aspx?date=2026-06-22"

    meetings = _parse_floor_xml(FLOOR_XML, monday, source_url)

    ids = [m["external_id"] for m in meetings]

    assert len(ids) == len(set(ids)), "External IDs must be unique"

def test_parse_floor_xml_bad_xml():

    result = _parse_floor_xml(b"not xml", date(2026, 6, 22), "https://docs.house.gov/floor/")

    assert result == []

def test_ensure_tables_idempotent():

    conn, path = _make_db()

    _ensure_tables(conn)

    _ensure_tables(conn)

    tables = {r[0] for r in conn.execute(

        "SELECT name FROM sqlite_master WHERE type='table'"

    ).fetchall()}

    assert "meetings" in tables

    conn.close()

    os.unlink(path)

def _sample_meeting(**kwargs) -> dict:

    base = {

        "source":       SOURCE_SENATE,

        "external_id":  "338613",

        "chamber":      "Senate",

        "committee":    "Health, Education, Labor, and Pensions",

        "date":         "2026-07-15",

        "time":         "09:45 AM",

        "datetime_iso": "2026-07-15T09:45:00",

        "title":        "Business meeting to consider pending calendar business.",

        "subject":      "Business meeting to consider pending calendar business.",

        "type":         "business_meeting",

        "location":     "SD-430",

        "status":       "scheduled",

        "source_url":   SENATE_PAGE_URL,

    }

    base.update(kwargs)

    return base

def test_upsert_meeting_inserts_row():

    conn, path = _make_db()

    m = _sample_meeting()

    _upsert_meeting(conn, m)

    conn.commit()

    row = conn.execute(

        "SELECT * FROM meetings WHERE source=? AND external_id=?",

        (m["source"], m["external_id"])

    ).fetchone()

    assert row is not None

    assert row["committee"] == "Health, Education, Labor, and Pensions"

    assert row["type"] == "business_meeting"

    conn.close()

    os.unlink(path)

def test_upsert_meeting_idempotent():

    """Double-upsert must not create duplicate rows."""

    conn, path = _make_db()

    m = _sample_meeting()

    _upsert_meeting(conn, m)

    _upsert_meeting(conn, m)

    conn.commit()

    count = conn.execute(

        "SELECT COUNT(*) FROM meetings WHERE source=? AND external_id=?",

        (m["source"], m["external_id"])

    ).fetchone()[0]

    assert count == 1

    conn.close()

    os.unlink(path)

def test_upsert_meeting_updates_status():

    """Upsert with changed status overwrites the row."""

    conn, path = _make_db()

    m = _sample_meeting(status="scheduled")

    _upsert_meeting(conn, m)

    conn.commit()

    m2 = _sample_meeting(status="canceled")

    _upsert_meeting(conn, m2)

    conn.commit()

    row = conn.execute(

        "SELECT status FROM meetings WHERE source=? AND external_id=?",

        (m["source"], m["external_id"])

    ).fetchone()

    assert row["status"] == "canceled"

    conn.close()

    os.unlink(path)

def test_upsert_meeting_house_chamber():

    conn, path = _make_db()

    m = _sample_meeting(

        source=SOURCE_HOUSE_CMT,

        external_id="119243",

        chamber="House",

        committee="Committee on Homeland Security",

        type="hearing",

        source_url="https://docs.house.gov/Committee/Calendar/ByEvent.aspx?EventID=119243",

    )

    _upsert_meeting(conn, m)

    conn.commit()

    row = conn.execute(

        "SELECT chamber, type FROM meetings WHERE source=? AND external_id=?",

        (SOURCE_HOUSE_CMT, "119243")

    ).fetchone()

    assert row["chamber"] == "House"

    assert row["type"] == "hearing"

    conn.close()

    os.unlink(path)

def test_poll_senate_mock(tmp_path):

    """poll_senate() must store 2 meetings from sample XML and be idempotent."""

    from ingest.pollers.meetings import poll_senate

    db_path = str(tmp_path / "test.db")

    conn = get_connection(db_path)

    init_schema(conn)

    _ensure_tables(conn)

    mock_resp = MagicMock()

    mock_resp.raise_for_status.return_value = None

    mock_resp.content = SENATE_XML_SAMPLE

    with patch("ingest.pollers.meetings.requests.get", return_value=mock_resp):

        count1 = poll_senate(conn)

    assert count1 == 2, f"Expected 2 new rows, got {count1}"

    with patch("ingest.pollers.meetings.requests.get", return_value=mock_resp):

        count2 = poll_senate(conn)

    assert count2 == 0, f"Expected 0 on re-run, got {count2}"

    rows = conn.execute("SELECT * FROM meetings WHERE source=?",

                        (SOURCE_SENATE,)).fetchall()

    assert len(rows) == 2

    dates = {r["date"] for r in rows}

    assert "2026-07-15" in dates

    conn.close()

def test_source_urls_are_allowlisted():

    """Confirm that all stored source_urls pass the validator."""

    from ingest.source_validator import is_allowlisted

    urls = [

        SENATE_PAGE_URL,

        "https://docs.house.gov/Committee/Calendar/ByEvent.aspx?EventID=119243",

        "https://docs.house.gov/floor/Default.aspx?date=2026-06-22",

    ]

    for url in urls:

        assert is_allowlisted(url), f"URL not allowlisted: {url}"