Initial commit.

2026-03-30 19:34:46 +01:00 · 2026-03-30 19:34:46 +01:00 · a8e0bd39e5
commit a8e0bd39e5
16 changed files with 981 additions and 0 deletions
--- a/tests/test_eurostar_scraper.py
+++ b/tests/test_eurostar_scraper.py
@ -0,0 +1,82 @@
+import json
+import pytest
+from scraper.eurostar import _hhmm, _parse
+
+
+# ---------------------------------------------------------------------------
+# _hhmm
+# ---------------------------------------------------------------------------
+
+def test_hhmm_parses_datetime_string():
+    assert _hhmm('2026-03-30 09:34:00') == '09:34'
+
+def test_hhmm_none_input():
+    assert _hhmm(None) is None
+
+def test_hhmm_empty_string():
+    assert _hhmm('') is None
+
+
+# ---------------------------------------------------------------------------
+# _parse
+# ---------------------------------------------------------------------------
+
+def _make_next_data(departures: list) -> str:
+    data = {
+        'props': {
+            'pageProps': {
+                'pageData': {
+                    'liveDepartures': departures
+                }
+            }
+        }
+    }
+    return f'<script id="__NEXT_DATA__" type="application/json">{json.dumps(data)}</script>'
+
+
+def _departure(dep_dt: str, arr_dt: str) -> dict:
+    return {
+        'origin':      {'model': {'scheduledDepartureDateTime': dep_dt}},
+        'destination': {'model': {'scheduledArrivalDateTime':   arr_dt}},
+    }
+
+
+def test_parse_single_departure():
+    html = _make_next_data([_departure('2026-03-30 06:01:00', '2026-03-30 09:34:00')])
+    services = _parse(html, 'Paris Gare du Nord')
+    assert len(services) == 1
+    assert services[0] == {
+        'depart_st_pancras': '06:01',
+        'arrive_destination': '09:34',
+        'destination': 'Paris Gare du Nord',
+    }
+
+
+def test_parse_results_sorted_by_departure():
+    html = _make_next_data([
+        _departure('2026-03-30 10:00:00', '2026-03-30 13:00:00'),
+        _departure('2026-03-30 07:00:00', '2026-03-30 10:00:00'),
+    ])
+    services = _parse(html, 'Paris Gare du Nord')
+    assert services[0]['depart_st_pancras'] == '07:00'
+    assert services[1]['depart_st_pancras'] == '10:00'
+
+
+def test_parse_skips_entries_with_missing_times():
+    html = _make_next_data([
+        _departure(None, '2026-03-30 09:34:00'),
+        _departure('2026-03-30 08:00:00', None),
+        _departure('2026-03-30 09:00:00', '2026-03-30 12:00:00'),
+    ])
+    services = _parse(html, 'Paris Gare du Nord')
+    assert len(services) == 1
+    assert services[0]['depart_st_pancras'] == '09:00'
+
+
+def test_parse_no_next_data_returns_empty():
+    assert _parse('<html><body>nothing here</body></html>', 'Paris Gare du Nord') == []
+
+
+def test_parse_empty_departures():
+    html = _make_next_data([])
+    assert _parse(html, 'Paris Gare du Nord') == []
--- a/tests/test_rtt_scraper.py
+++ b/tests/test_rtt_scraper.py
@ -0,0 +1,71 @@
+import pytest
+from scraper.realtime_trains import _fmt, _parse_services
+
+
+# ---------------------------------------------------------------------------
+# _fmt
+# ---------------------------------------------------------------------------
+
+def test_fmt_four_digits():
+    assert _fmt('0830') == '08:30'
+
+def test_fmt_already_colon():
+    assert _fmt('08:30') == '08:30'
+
+def test_fmt_strips_non_digits():
+    assert _fmt('08h30') == '08:30'
+
+
+# ---------------------------------------------------------------------------
+# _parse_services
+# ---------------------------------------------------------------------------
+
+def _make_html(services: list[tuple[str, str]], time_class: str) -> str:
+    """Build a minimal servicelist HTML with (train_id, time) pairs."""
+    items = ''
+    for tid, time in services:
+        items += f'''
+        <a class="service">
+          <div class="tid">{tid}</div>
+          <div class="time plan {time_class}">{time}</div>
+        </a>'''
+    return f'<div class="servicelist">{items}</div>'
+
+
+def test_parse_services_departures():
+    html = _make_html([('1A23', '0700'), ('2B45', '0830')], 'd')
+    result = _parse_services(html, 'div.time.plan.d')
+    assert result == {'1A23': '07:00', '2B45': '08:30'}
+
+
+def test_parse_services_arrivals():
+    html = _make_html([('1A23', '0845')], 'a')
+    result = _parse_services(html, 'div.time.plan.a')
+    assert result == {'1A23': '08:45'}
+
+
+def test_parse_services_no_servicelist():
+    assert _parse_services('<html></html>', 'div.time.plan.d') == {}
+
+
+def test_parse_services_skips_missing_time():
+    html = '''
+    <div class="servicelist">
+      <a class="service"><div class="tid">1A23</div></a>
+      <a class="service"><div class="tid">2B45</div><div class="time plan d">0900</div></a>
+    </div>'''
+    result = _parse_services(html, 'div.time.plan.d')
+    assert '1A23' not in result
+    assert result == {'2B45': '09:00'}
+
+
+def test_parse_services_skips_empty_time():
+    html = '''
+    <div class="servicelist">
+      <a class="service">
+        <div class="tid">1A23</div>
+        <div class="time plan d">   </div>
+      </a>
+    </div>'''
+    result = _parse_services(html, 'div.time.plan.d')
+    assert result == {}
--- a/tests/test_trip_planner.py
+++ b/tests/test_trip_planner.py
@ -0,0 +1,131 @@
+import pytest
+from trip_planner import combine_trips, _fmt_duration
+
+DATE = '2026-03-30'
+
+
+# ---------------------------------------------------------------------------
+# _fmt_duration
+# ---------------------------------------------------------------------------
+
+def test_fmt_duration_hours_and_minutes():
+    assert _fmt_duration(95) == '1h 35m'
+
+def test_fmt_duration_exact_hours():
+    assert _fmt_duration(120) == '2h'
+
+def test_fmt_duration_minutes_only():
+    assert _fmt_duration(45) == '45m'
+
+
+# ---------------------------------------------------------------------------
+# combine_trips — basic pairing
+# ---------------------------------------------------------------------------
+
+GWR_FAST = {'depart_bristol': '07:00', 'arrive_paddington': '08:45'}   # 1h 45m
+GWR_SLOW = {'depart_bristol': '07:00', 'arrive_paddington': '09:26'}   # 2h 26m — over limit
+
+ES_PARIS = {'depart_st_pancras': '10:01', 'arrive_destination': '13:34', 'destination': 'Paris Gare du Nord'}
+ES_EARLY = {'depart_st_pancras': '09:00', 'arrive_destination': '12:00', 'destination': 'Paris Gare du Nord'}
+
+
+def test_valid_trip_is_returned():
+    trips = combine_trips([GWR_FAST], [ES_PARIS], DATE)
+    assert len(trips) == 1
+    t = trips[0]
+    assert t['depart_bristol'] == '07:00'
+    assert t['arrive_paddington'] == '08:45'
+    assert t['depart_st_pancras'] == '10:01'
+    assert t['arrive_destination'] == '13:34'
+    assert t['destination'] == 'Paris Gare du Nord'
+
+
+def test_gwr_too_slow_excluded():
+    # 2h 26m GWR journey exceeds MAX_GWR_MINUTES (110)
+    trips = combine_trips([GWR_SLOW], [ES_PARIS], DATE)
+    assert trips == []
+
+
+def test_eurostar_too_early_excluded():
+    # Eurostar departs before min connection time has elapsed
+    trips = combine_trips([GWR_FAST], [ES_EARLY], DATE)
+    assert trips == []
+
+
+def test_no_trains_returns_empty():
+    assert combine_trips([], [], DATE) == []
+
+def test_no_gwr_returns_empty():
+    assert combine_trips([], [ES_PARIS], DATE) == []
+
+def test_no_eurostar_returns_empty():
+    assert combine_trips([GWR_FAST], [], DATE) == []
+
+
+# ---------------------------------------------------------------------------
+# Connection window constraints
+# ---------------------------------------------------------------------------
+
+def test_min_connection_enforced():
+    # Arrive Paddington 08:45, need 75 min → earliest St Pancras 10:00
+    # ES at 09:59 should be excluded, 10:00 should be included
+    es_too_close = {'depart_st_pancras': '09:59', 'arrive_destination': '13:00', 'destination': 'Paris Gare du Nord'}
+    es_ok        = {'depart_st_pancras': '10:00', 'arrive_destination': '13:00', 'destination': 'Paris Gare du Nord'}
+    assert combine_trips([GWR_FAST], [es_too_close], DATE) == []
+    trips = combine_trips([GWR_FAST], [es_ok], DATE)
+    assert len(trips) == 1
+
+
+def test_max_connection_enforced():
+    # Arrive Paddington 08:45, max 140 min → latest St Pancras 11:05
+    es_ok      = {'depart_st_pancras': '11:05', 'arrive_destination': '14:00', 'destination': 'Paris Gare du Nord'}
+    es_too_late = {'depart_st_pancras': '11:06', 'arrive_destination': '14:00', 'destination': 'Paris Gare du Nord'}
+    trips = combine_trips([GWR_FAST], [es_ok], DATE)
+    assert len(trips) == 1
+    assert combine_trips([GWR_FAST], [es_too_late], DATE) == []
+
+
+# ---------------------------------------------------------------------------
+# Only earliest valid Eurostar per GWR departure
+# ---------------------------------------------------------------------------
+
+def test_only_earliest_eurostar_per_gwr():
+    es1 = {'depart_st_pancras': '10:01', 'arrive_destination': '13:34', 'destination': 'Paris Gare du Nord'}
+    es2 = {'depart_st_pancras': '11:01', 'arrive_destination': '14:34', 'destination': 'Paris Gare du Nord'}
+    trips = combine_trips([GWR_FAST], [es1, es2], DATE)
+    assert len(trips) == 1
+    assert trips[0]['depart_st_pancras'] == '10:01'
+
+
+# ---------------------------------------------------------------------------
+# Multiple GWR trains → multiple trips
+# ---------------------------------------------------------------------------
+
+def test_multiple_gwr_trains():
+    gwr2 = {'depart_bristol': '08:00', 'arrive_paddington': '09:45'}
+    es   = {'depart_st_pancras': '11:01', 'arrive_destination': '14:34', 'destination': 'Paris Gare du Nord'}
+    trips = combine_trips([GWR_FAST, gwr2], [es], DATE)
+    assert len(trips) == 2
+    assert trips[0]['depart_bristol'] == '07:00'
+    assert trips[1]['depart_bristol'] == '08:00'
+
+
+# ---------------------------------------------------------------------------
+# Duration fields
+# ---------------------------------------------------------------------------
+
+def test_gwr_duration_in_trip():
+    trips = combine_trips([GWR_FAST], [ES_PARIS], DATE)
+    assert trips[0]['gwr_duration'] == '1h 45m'
+
+
+def test_total_duration_in_trip():
+    # depart 07:00, arrive 13:34 → 6h 34m
+    trips = combine_trips([GWR_FAST], [ES_PARIS], DATE)
+    assert trips[0]['total_duration'] == '6h 34m'
+
+
+def test_connection_duration_in_trip():
+    # arrive Paddington 08:45, depart St Pancras 10:01 → 1h 16m
+    trips = combine_trips([GWR_FAST], [ES_PARIS], DATE)
+    assert trips[0]['connection_duration'] == '1h 16m'