Fix intermediate stops along trip headsign

gtfstidy.go

@@ -9,17 +9,18 @@ package main
 import (
 	"errors"
 	"fmt"
-	"github.com/patrickbr/gtfsparser"
-	"github.com/patrickbr/gtfsparser/gtfs"
-	"github.com/patrickbr/gtfstidy/processors"
-	"github.com/patrickbr/gtfswriter"
-	"github.com/paulmach/go.geojson"
-	flag "github.com/spf13/pflag"
 	"io/ioutil"
 	"os"
 	"path"
 	"strconv"
 	"strings"
+
+	"github.com/patrickbr/gtfsparser"
+	"github.com/patrickbr/gtfsparser/gtfs"
+	"github.com/patrickbr/gtfstidy/processors"
+	"github.com/patrickbr/gtfswriter"
+	geojson "github.com/paulmach/go.geojson"
+	flag "github.com/spf13/pflag"
 )
 
 func getGtfsPoly(poly [][][]float64) gtfsparser.Polygon {
@@ -170,6 +171,7 @@ func main() {
 	explicitCals := flag.BoolP("explicit-calendar", "", false, "add calendar.txt entry for every service, even irregular ones")
 	ensureTripHeadsigns := flag.BoolP("ensure-trip-headsigns", "", false, "write trip headsigns if missing")
 	ensureParents := flag.BoolP("ensure-stop-parents", "", false, "ensure that every stop (location_type=0) has a parent station")
+	fixTripHeadsigns := flag.BoolP("fix-trip-headsigns", "", false, "fixes trip headsigns pointing to previous stops")
 	keepColOrder := flag.BoolP("keep-col-order", "", false, "keep the original column ordering of the input feed")
 	keepFields := flag.BoolP("keep-additional-fields", "F", false, "keep all non-GTFS fields from the input")
 	dropTooFast := flag.BoolP("drop-too-fast-trips", "", false, "drop trips that are too fast to realistically occur")
@@ -641,6 +643,9 @@ func main() {
 		if *ensureTripHeadsigns {
 			minzers = append(minzers, processors.TripHeadsigner{})
 		}
+		if *fixTripHeadsigns {
+			minzers = append(minzers, processors.FixIntermediateHeadsigns{})
+		}
 
 		if *useRedTripMinimizer {
 			// to convert calendar_dates based services into regular calendar.txt services

processors/intermediateheadsign.go

@@ -0,0 +1,72 @@
+package processors
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/patrickbr/gtfsparser"
+)
+
+// FixIntermediateHeadsigns checks if the trip headsign matches an intermediate stop.
+// If so, it sets the stop_headsign for previous stops to that intermediate name
+// and updates the trip_headsign to the final destination.
+
+type FixIntermediateHeadsigns struct{}
+
+func (pro FixIntermediateHeadsigns) Run(feed *gtfsparser.Feed) {
+	fmt.Fprintf(os.Stdout, "Fixing intermediate headsigns... ")
+
+	count := 0
+
+	for _, trip := range feed.Trips {
+		if len(trip.StopTimes) < 2 {
+			continue
+		}
+
+		currentHeadsign := trip.Headsign
+		if *currentHeadsign == "" {
+			continue
+		}
+
+		lastStopIdx := len(trip.StopTimes) - 1
+		lastStop := trip.StopTimes[lastStopIdx].Stop()
+		if lastStop == nil {
+			continue
+		}
+
+		if *currentHeadsign == lastStop.Name {
+			continue
+		}
+
+		matchIndex := -1
+
+		// 1. Check if headsign is equal to a stop along the trip (except the last one)
+		for i := len(trip.StopTimes) - 2; i >= 0; i-- {
+			st := trip.StopTimes[i]
+			if st.Stop() != nil && st.Stop().Name == *currentHeadsign {
+				matchIndex = i
+				break
+			}
+		}
+
+		if matchIndex != -1 {
+			// Logic:
+			// Sequence: A -> B -> C (match) -> D -> E (last)
+			// Old Trip Headsign: C
+			// New Trip Headsign: E
+			// Stop Headsign for A, B: C
+
+			// Update trip headsign to the actual last stop
+			trip.Headsign = &lastStop.Name
+
+			// Update stop_headsign for all stops prior to the match
+			for j := 0; j < matchIndex; j++ {
+				trip.StopTimes[j].SetHeadsign(currentHeadsign)
+			}
+
+			count++
+		}
+	}
+
+	fmt.Fprintf(os.Stdout, "done. Fixed headsigns for %d trips.\n", count)
+}