From 57566ec8876faf2bf5b60b9923381657d2034709 Mon Sep 17 00:00:00 2001
From: Landon Reed <landontreed@gmail.com>
Date: Wed, 19 Dec 2018 16:49:12 -0500
Subject: [PATCH] feat(validator): validate stop names and trip headsigns

refs #167
---
 .../conveyal/gtfs/error/NewGTFSErrorType.java |  4 ++
 .../gtfs/validator/NamesValidator.java        | 38 ++++++++++++++++++-
 2 files changed, 41 insertions(+), 1 deletion(-)

diff --git a/src/main/java/com/conveyal/gtfs/error/NewGTFSErrorType.java b/src/main/java/com/conveyal/gtfs/error/NewGTFSErrorType.java
index 02fc5075d..b35037642 100644
--- a/src/main/java/com/conveyal/gtfs/error/NewGTFSErrorType.java
+++ b/src/main/java/com/conveyal/gtfs/error/NewGTFSErrorType.java
@@ -44,10 +44,14 @@ public enum NewGTFSErrorType {
     SERVICE_NEVER_ACTIVE(Priority.MEDIUM, "A service code was defined, but is never active on any date."),
     SERVICE_UNUSED(Priority.MEDIUM, "A service code was defined, but is never referenced by any trips."),
     SHAPE_DIST_TRAVELED_NOT_INCREASING(Priority.MEDIUM, "Shape distance traveled must increase with stop times."),
+    STOP_DESCRIPTION_SAME_AS_NAME(Priority.LOW, "The description of a stop is identical to its name, so does not add any information."),
     STOP_LOW_POPULATION_DENSITY(Priority.HIGH, "A stop is located in a geographic area with very low human population density."),
+    STOP_NAME_MISSING(Priority.MEDIUM, "A stop does not have a name."),
     STOP_GEOGRAPHIC_OUTLIER(Priority.HIGH, "This stop is located very far from the middle 90% of stops in this feed."),
     STOP_UNUSED(Priority.MEDIUM, "This stop is not referenced by any trips."),
     TRIP_EMPTY(Priority.HIGH, "This trip is defined but has no stop times."),
+    TRIP_HEADSIGN_CONTAINS_ROUTE_NAME(Priority.LOW, "A trip headsign contains the route name, but should only contain information to distinguish it from other trips for the route."),
+    TRIP_HEADSIGN_SHOULD_DESCRIBE_DESTINATION_OR_WAYPOINTS(Priority.LOW, "A trip headsign begins with 'to' or 'towards', but should begin with destination or direction and optionally include waypoints with 'via'"),
     TRIP_NEVER_ACTIVE(Priority.MEDIUM, "A trip is defined, but its service is never running on any date."),
     ROUTE_UNUSED(Priority.HIGH, "This route is defined but has no trips."),
     TRAVEL_DISTANCE_ZERO(Priority.MEDIUM, "The vehicle does not cover any distance between the last stop and this one."),
diff --git a/src/main/java/com/conveyal/gtfs/validator/NamesValidator.java b/src/main/java/com/conveyal/gtfs/validator/NamesValidator.java
index 913b56453..889a3679b 100644
--- a/src/main/java/com/conveyal/gtfs/validator/NamesValidator.java
+++ b/src/main/java/com/conveyal/gtfs/validator/NamesValidator.java
@@ -3,6 +3,8 @@
 import com.conveyal.gtfs.error.SQLErrorStorage;
 import com.conveyal.gtfs.loader.Feed;
 import com.conveyal.gtfs.model.Route;
+import com.conveyal.gtfs.model.Stop;
+import com.conveyal.gtfs.model.Trip;
 
 import static com.conveyal.gtfs.error.NewGTFSErrorType.*;
 
@@ -41,7 +43,41 @@ public void validate() {
                 // TODO we want some additional checking for extended route types.
             }
         }
-        // TODO Check trips and all other tables.
+        // Check stops
+        for (Stop stop : feed.stops) {
+            String name = normalize(stop.stop_name);
+            String desc = normalize(stop.stop_desc);
+            // Stops must be named.
+            if (name.isEmpty()) {
+                registerError(stop, STOP_NAME_MISSING);
+            }
+            // If provided, the description of a stop should be more informative than its name.
+            if (!desc.isEmpty() && desc.equals(name)) {
+                registerError(stop, STOP_DESCRIPTION_SAME_AS_NAME, desc);
+            }
+        }
+        // Check trips
+        for (Trip trip : feed.trips) {
+            String headsign = normalize(trip.trip_headsign);
+            // TODO: check trip short name?
+//            String shortName = normalize(trip.trip_short_name);
+            Route route = feed.routes.get(trip.route_id);
+            String routeShortName = "", routeLongName = "";
+            if (route != null) {
+                routeShortName = normalize(route.route_short_name);
+                routeLongName = normalize(route.route_long_name);
+            }
+            // Trip headsign should not duplicate route name.
+            if (!headsign.isEmpty() && (headsign.contains(routeShortName) || headsign.contains(routeLongName))) {
+                registerError(trip, TRIP_HEADSIGN_CONTAINS_ROUTE_NAME, headsign);
+            }
+            // Trip headsign should not begin with "to" or "towards" (note: headsign normalized to lowercase). Headsigns
+            // should follow one of the patterns defined in the best practices: http://gtfs.org/best-practices#tripstxt
+            if (headsign.startsWith("to ") || headsign.startsWith("towards ")) {
+                registerError(trip, TRIP_HEADSIGN_SHOULD_DESCRIBE_DESTINATION_OR_WAYPOINTS, headsign);
+            }
+        }
+        // TODO Are there other tables we're not checking?
     }
 
     /** @return a non-null String that is lower case and has no leading or trailing whitespace */