SPlit up

pipeline/download/test_transit_network.py

@@ -7,6 +7,7 @@ from pathlib import Path
 import pytest
 
 from pipeline.download.transit_network import (
+    clean_national_rail_gtfs,
     convert_high_freq_to_frequency_based,
     validate_gtfs_feed,
 )
@@ -69,6 +70,46 @@ def test_one_based_stop_sequence_is_converted(tmp_path: Path) -> None:
     assert headway_secs == "300"
 
 
+def test_clean_national_rail_gtfs_orders_by_stop_sequence_not_file_order(
+    tmp_path: Path,
+) -> None:
+    """dtd2mysql exports happen to be ordered by stop_sequence within each
+    trip, but nothing guarantees it. Rows arriving out of order must be sorted
+    by their original stop_sequence before the backwards-time check and the
+    0-based renumbering — file order would flag the trip as backwards and drop
+    it (or scramble the stop order)."""
+    src = tmp_path / "in.zip"
+    dst = tmp_path / "out.zip"
+    with zipfile.ZipFile(src, "w") as z:
+        z.writestr(
+            "stops.txt",
+            "stop_id,stop_lat,stop_lon\nSTOP_A,51.5,-0.1\nSTOP_B,51.6,-0.1\n",
+        )
+        z.writestr("routes.txt", "route_id,route_type\nR1,2\n")
+        z.writestr("trips.txt", "trip_id,route_id,service_id\nT1,R1,S1\n")
+        # File order is seq 2 then seq 1: in file order departures look
+        # backwards (07:00 then 06:00); in sequence order they are fine.
+        z.writestr(
+            "stop_times.txt",
+            "trip_id,stop_id,stop_sequence,departure_time\n"
+            "T1,STOP_B,2,07:00:00\n"
+            "T1,STOP_A,1,06:00:00\n",
+        )
+
+    clean_national_rail_gtfs(src, dst)
+
+    with zipfile.ZipFile(dst, "r") as z:
+        stop_times = z.read("stop_times.txt").decode("utf-8").splitlines()
+        trips = z.read("trips.txt").decode("utf-8").splitlines()
+
+    assert trips == ["trip_id,route_id,service_id", "T1,R1,S1"]
+    assert stop_times == [
+        "trip_id,stop_id,stop_sequence,departure_time",
+        "T1,STOP_A,0,06:00:00",
+        "T1,STOP_B,1,07:00:00",
+    ]
+
+
 def test_raises_when_no_first_stops_found(tmp_path: Path) -> None:
     """A non-empty target trip set with unparseable stop_sequence is loud, not silent."""
     src = tmp_path / "in.zip"