"""Unit tests for scraper.parsers.parse_events_page (the chamber "upcoming"
events calendar parser, schema v2 / scheduled_events forward source).

The bill queue (daily file) is a backward snapshot, so the chamber calendar is
the only forward-looking signal. These tests cover the new parser branches:
date-from-<time> attribute, kind classification, source-url resolution, header
row skipping, the <tr>-with-<time> fallback, and the empty/recess case.

The parser is the coverage-gated module, so these aim for complete branch
coverage of parse_events_page + classify_event_kind.
"""
from __future__ import annotations

import os

from scraper import parsers
from scraper.parsers import parse_events_page, classify_event_kind


# ---- a populated "views" table, matching the live assembly.ca.gov layout ----
POPULATED_HTML = """
<html><body>
<div class="view-content">
  <table class="cols-3">
    <thead><tr>
      <th class="views-field views-field-media-source-pub-date">Date</th>
      <th class="views-field views-field-title">Title</th>
      <th class="views-field">Video</th>
    </tr></thead>
    <tbody>
      <tr>
        <td class="views-field views-field-media-source-pub-date">
          <time datetime="2026-09-10T00:00:00-07:00">September 10, 2026</time>
        </td>
        <td class="views-field views-field-title">
          Assembly Floor Session</td>
        <td class="views-field">
          <a href="/media/assembly-floor-session-20260910">watch now</a></td>
      </tr>
      <tr>
        <td class="views-field views-field-media-source-pub-date">
          <time datetime="2026-09-11T00:00:00-07:00">September 11, 2026</time>
        </td>
        <td class="views-field views-field-title">
          Judiciary Committee</td>
        <td class="views-field"></td>
      </tr>
      <tr>
        <td class="views-field views-field-media-source-pub-date">
          <time datetime="2026-09-12T00:00:00-07:00">September 12, 2026</time>
        </td>
        <td class="views-field views-field-title">
          Press Conference</td>
        <td class="views-field"><a href="http://example.com/x">link</a></td>
      </tr>
    </tbody>
  </table>
</div>
</body></html>
"""

# A recess: the table is present but carries no data rows (empty view-content).
RECESS_HTML = """
<html><body>
<div class="view-content">
  <table class="cols-3">
    <thead><tr>
      <th class="views-field views-field-media-source-pub-date">Date</th>
      <th class="views-field views-field-title">Title</th>
    </tr></thead>
    <tbody></tbody>
  </table>
  <p>There are currently no upcoming committee hearings scheduled.</p>
</body></html>
"""

# Layout with no <table class="cols-3">: rows must be recovered by <time>.
NO_COLS3_HTML = """
<html><body>
  <tr>
    <td><time datetime="2026-09-13T00:00:00-07:00">September 13, 2026</time></td>
    <td>Budget Committee</td>
  </tr>
</body></html>
"""


# ---- kind classification ---------------------------------------------------

def test_classify_floor():
    assert classify_event_kind("Assembly Floor Session") == "FLOOR"


def test_classify_committee():
    assert classify_event_kind("Judiciary Committee") == "COMMITTEE"


def test_classify_other():
    assert classify_event_kind("Press Conference") == "OTHER"


def test_classify_none():
    assert classify_event_kind(None) == "OTHER"


# ---- populated page --------------------------------------------------------

def test_populated_page_parses_three_events():
    evs = parse_events_page(POPULATED_HTML, chamber="assembly",
                            source="assembly_events",
                            base_url="https://www.assembly.ca.gov/events/upcoming")
    assert len(evs) == 3
    by_date = {e["date"]: e for e in evs}
    assert by_date["2026-09-10"]["kind"] == "FLOOR"
    assert by_date["2026-09-11"]["kind"] == "COMMITTEE"
    assert by_date["2026-09-12"]["kind"] == "OTHER"


def test_dates_come_from_time_attribute():
    evs = parse_events_page(POPULATED_HTML)
    dates = sorted(e["date"] for e in evs)
    assert dates == ["2026-09-10", "2026-09-11", "2026-09-12"]


def test_chamber_normalised_to_upper():
    evs = parse_events_page(POPULATED_HTML, chamber=" Assembly ")
    assert all(e["chamber"] == "ASSEMBLY" for e in evs)


def test_relative_link_resolves_against_host():
    evs = parse_events_page(POPULATED_HTML,
                            base_url="https://www.assembly.ca.gov/events/upcoming")
    floor = next(e for e in evs if e["kind"] == "FLOOR")
    assert floor["source_url"] == \
        "https://www.assembly.ca.gov/media/assembly-floor-session-20260910"


def test_row_without_link_falls_back_to_base_url():
    evs = parse_events_page(POPULATED_HTML,
                            base_url="https://www.assembly.ca.gov/events/upcoming")
    comm = next(e for e in evs if e["kind"] == "COMMITTEE")
    assert comm["source_url"] == "https://www.assembly.ca.gov/events/upcoming"


def test_external_link_kept_verbatim():
    evs = parse_events_page(POPULATED_HTML,
                            base_url="https://www.assembly.ca.gov/events/upcoming")
    other = next(e for e in evs if e["kind"] == "OTHER")
    assert other["source_url"] == "http://example.com/x"


def test_header_row_skipped():
    evs = parse_events_page(POPULATED_HTML)
    assert all(e["title"] not in ("Date", "Title") for e in evs)


# ---- empty / recess --------------------------------------------------------

def test_recess_yields_no_events():
    assert parse_events_page(RECESS_HTML) == []


def test_empty_string_yields_no_events():
    assert parse_events_page("") == []


def test_html_with_no_time_cells_yields_no_events():
    assert parse_events_page("<html><body><p>Nothing here.</p></body></html>") == []


# ---- layout fallback + bad rows -------------------------------------------

class _Broken:
    """A soup stand-in whose find() raises, exercised via parse error path."""
    pass


def test_no_cols3_falls_back_to_time_rows():
    evs = parse_events_page(NO_COLS3_HTML, chamber="assembly")
    assert len(evs) == 1
    assert evs[0]["date"] == "2026-09-13"
    assert evs[0]["kind"] == "COMMITTEE"


def test_malformed_row_skipped_not_fatal():
    # A row with a <time> but no title cell is skipped, the good row survives.
    html = """
    <html><body><table class="cols-3"><tbody>
      <tr><td><time datetime="2026-09-14">Sep 14</time></td><td></td></tr>
      <tr><td><time datetime="2026-09-15">Sep 15</time></td>
          <td class="views-field views-field-title">Rules Committee</td></tr>
    </tbody></table></body></html>
    """
    evs = parse_events_page(html, chamber="assembly")
    assert len(evs) == 1
    assert evs[0]["title"] == "Rules Committee"


def test_broken_soup_returns_empty():
    # A document that cannot be parsed at all yields [] (never raises).
    result = parse_events_page(None, parser="lxml")
    assert result == []


# ---- live sample (regression, network-independent) -------------------------

def test_sample_past_page_parses(tmp_path):
    sample = os.path.join(os.path.dirname(os.path.dirname(__file__)),
                          "samples", "assembly_events_past.html")
    if not os.path.exists(sample):
        pytest.skip("sample not present")
    import os as _os
    html = open(sample, encoding="utf-8").read()
    evs = parse_events_page(html, chamber="assembly", source="assembly_events")
    assert len(evs) >= 2
    assert all(e["chamber"] == "ASSEMBLY" for e in evs)
    assert all(e["date"] and len(e["date"]) == 10 for e in evs)