From e27fdd8c2ded61f3306e4c9fd93de43e03e4e17f Mon Sep 17 00:00:00 2001
From: Alexey Zakharenkov <a-zakh@yandex.ru>
Date: Tue, 18 Oct 2022 17:39:42 +0300
Subject: [PATCH] Add to GTFS only transfers in which both stops are part of
 routes

---
 processors/gtfs.py           | 15 +++++----
 tests/test_gtfs_processor.py | 60 ++++++++++++++++++++++++++++++++++++
 2 files changed, 69 insertions(+), 6 deletions(-)
 create mode 100644 tests/test_gtfs_processor.py

diff --git a/processors/gtfs.py b/processors/gtfs.py
index a5186fe..1ae5ee7 100644
--- a/processors/gtfs.py
+++ b/processors/gtfs.py
@@ -2,6 +2,8 @@ import csv
 import io
 import zipfile
 
+from itertools import permutations
+
 from ._common import (
     DEFAULT_INTERVAL,
     format_colour,
@@ -346,19 +348,20 @@ def process(cities, transfers, filename, cache_path):
         for stoparea1 in stoparea_set:
             for stoparea2 in stoparea_set:
                 if stoparea1.id < stoparea2.id:
+                    stop1_id = f"{stoparea1.id}_st"
+                    stop2_id = f"{stoparea2.id}_st"
+                    if not {stop1_id, stop2_id}.issubset(all_stops):
+                        continue
                     transfer_time = TRANSFER_PENALTY + round(
                         distance(stoparea1.center, stoparea2.center)
                         / SPEED_ON_TRANSFER
                     )
-                    for id1, id2 in (
-                        (stoparea1.id, stoparea2.id),
-                        (stoparea2.id, stoparea1.id),
-                    ):
+                    for id1, id2 in permutations((stop1_id, stop2_id)):
                         gtfs_data["transfers"].append(
                             dict_to_row(
                                 {
-                                    "from_stop_id": f"{id1}_st",
-                                    "to_stop_id": f"{id2}_st",
+                                    "from_stop_id": id1,
+                                    "to_stop_id": id2,
                                     "transfer_type": 0,
                                     "min_transfer_time": transfer_time,
                                 },
diff --git a/tests/test_gtfs_processor.py b/tests/test_gtfs_processor.py
new file mode 100644
index 0000000..df0e2b1
--- /dev/null
+++ b/tests/test_gtfs_processor.py
@@ -0,0 +1,60 @@
+import unittest
+
+from processors.gtfs import (
+    dict_to_row,
+    GTFS_COLUMNS,
+)
+
+
+class TestGTFS(unittest.TestCase):
+    """Test processors/gtfs.py"""
+
+    def test_dict_to_row(self):
+        """Test that absent or None values in a GTFS feature item
+        are converted by dict_to_row() function to empty strings
+        in right amount.
+        """
+
+        if GTFS_COLUMNS["trips"][:3] != ["route_id", "service_id", "trip_id"]:
+            raise RuntimeError("GTFS column names/order inconsistency")
+
+        test_trips = [
+            {
+                "description": "Absent keys",
+                "trip_data": {
+                    "route_id": 1,
+                    "service_id": "a",
+                    "trip_id": "tr_123",
+                },
+            },
+            {
+                "description": "None or absent keys",
+                "trip_data": {
+                    "route_id": 1,
+                    "service_id": "a",
+                    "trip_id": "tr_123",
+                    "trip_headsign": None,
+                    "trip_short_name": None,
+                    "route_pattern_id": None,
+                },
+            },
+            {
+                "description": "None, empty-string or absent keys",
+                "trip_data": {
+                    "route_id": 1,
+                    "service_id": "a",
+                    "trip_id": "tr_123",
+                    "trip_headsign": "",
+                    "trip_short_name": "",
+                    "route_pattern_id": None,
+                },
+            },
+        ]
+
+        answer = [1, "a", "tr_123"] + [""] * (len(GTFS_COLUMNS["trips"]) - 3)
+
+        for test_trip in test_trips:
+            with self.subTest(msg=test_trip["description"]):
+                self.assertEqual(
+                    dict_to_row(test_trip["trip_data"], "trips"), answer
+                )