Improvements to duplicate suppression, cafe stop search, and configurability.

When suppressing duplicate cafes and place names, look back a certain distance along
the route rather than just the previous point of interest.

When searching for cafes, use SearchIntersect() to return all entries in the
bounding rectangle, not just the nearest.

Remove (most) hard-coded constants and allow these to be overriden by options
to the NewGPXSummarizer() constructor.

cmd/analyze-gpx/main.go

@@ -19,8 +19,11 @@ import (
 func main() {
 	log.SetFlags(0)
 	stopNames := flag.String("stops", "", "Source for refreshment stops")
-	minDist := flag.Float64("min-dist", 0.2, "Minimum distance (km) between points of interest")
+	stopRect := flag.Float64("sr", placenames.DefaultGPXSummarizerConfig.CoffeeStopSearchRectangleSize, "Size (m) of the rectangle we search for coffee stops near the route")
-	minSettlement := flag.String("min-settlement", "Other Settlement", "Exclude populated places smaller than this (City, Town, Village, Hamlet, Other Settlement)")
+	stopDupDist := flag.Float64("sdd", placenames.DefaultGPXSummarizerConfig.CoffeeStopDuplicateDistance, "Suppress recurrences of coffee stops within this distance (km)")
+	dupDist := flag.Float64("dd", placenames.DefaultGPXSummarizerConfig.PointOfInterestDuplicateDistance, "Suppress recurrences of points of interest within this distance (km)")
+	minDist := flag.Float64("md", placenames.DefaultGPXSummarizerConfig.PointOfInterestMinimumDistance, "Minimum distance (km) between points of interest")
+	minSettlement := flag.String("ms", "Other Settlement", "Exclude populated places smaller than this (City, Town, Village, Hamlet, Other Settlement)")
 	flag.Parse()
 	if flag.NArg() != 1 {
 		log.Fatal("Usage: %s [--stops=ctccambridge|cyclingmaps] [--min-dist X] [--min-settlement S] GPX_FILE_OR_DIRECTORY")
@@ -38,12 +41,16 @@ func main() {
 			log.Fatal(err)
 		}
 	}
-	gs, err := placenames.NewGPXSummarizer()
+	gs, err := placenames.NewGPXSummarizer(
+		placenames.WithMinimumSettlement(*minSettlement),
+		placenames.WithPointOfInterestMinimumDistance(*minDist),
+		placenames.WithPointOfInterestDuplicateDistance(*dupDist),
+		placenames.WithCoffeeStopSearchRectangleSize(*stopRect),
+		placenames.WithCoffeeStopDuplicateDistance(*stopDupDist),
+	)
 	if err != nil {
 		log.Fatal(err)
 	}
-	gs.SetMinDistance(*minDist)
-	gs.SetMinSettlement(*minSettlement)
 	if info.IsDir() {
 		err = summarizeDirectory(gs, stops, inFile)
 	} else {

pkg/cafes/common.go

@@ -10,7 +10,7 @@ import (
 )
 
 // Size (in metres) of the bounding box around a stop
-const stopRectangleSize = 500
+const stopRectangleSize = 50
 
 type RefreshmentStop struct {
 	Name     string

pkg/placenames/summarize.go

@@ -9,9 +9,8 @@ import (
 
 	"github.com/dhconnelly/rtreego"
 	"github.com/fofanov/go-osgb"
-	"github.com/twpayne/go-gpx"
-
 	"github.com/ray1729/gpx-utils/pkg/cafes"
+	"github.com/twpayne/go-gpx"
 )
 
 var populatedPlaceRank = map[string]int{
@@ -22,14 +21,80 @@ var populatedPlaceRank = map[string]int{
 	"Other Settlement": 1,
 }
 
+// GPXSummarizerConfig allows override of defaults used by the search algorithm.
+type GPXSummarizerConfig struct {
+	CoffeeStopSearchRectangleSize    float64
+	CoffeeStopDuplicateDistance      float64
+	PointOfInterestDuplicateDistance float64
+	PointOfInterestMinimumDistance   float64
+	MinimumSettlementRank            int
+}
+
+var DefaultGPXSummarizerConfig = GPXSummarizerConfig{
+	CoffeeStopSearchRectangleSize:    500.0, // m
+	CoffeeStopDuplicateDistance:      2.0,   // km
+	PointOfInterestDuplicateDistance: 1.0,   // km
+	PointOfInterestMinimumDistance:   0.2,   // km
+	MinimumSettlementRank:            1,     // "Other Settlement"
+}
+
+type Option func(*GPXSummarizerConfig)
+
+// WithCoffeeStopSearchRectangleSize overrides the size (in metres) of the rectangle searched
+// for coffee stops near the route. Default 500m.
+func WithCoffeeStopSearchRectangleSize(d float64) Option {
+	return func(c *GPXSummarizerConfig) {
+		c.CoffeeStopSearchRectangleSize = d
+	}
+}
+
+// WithCoffeeStopDuplicateDistance overrides the distance (in kilometers) we look back along the
+// route when suppressing duplicate coffee stop entries. This should be at least twice the
+// CoffeeStopSearchRectangleSize. Default 2km.
+func WithCoffeeStopDuplicateDistance(d float64) Option {
+	return func(c *GPXSummarizerConfig) {
+		c.CoffeeStopDuplicateDistance = d
+	}
+}
+
+// WithPointOfInterestDuplicateDistance overrides the distance (in km) we look back along
+// the route when suppressing duplicate points of interest.
+func WithPointOfInterestDuplicateDistance(d float64) Option {
+	return func(c *GPXSummarizerConfig) {
+		c.PointOfInterestDuplicateDistance = d
+	}
+}
+
+// WithPointOfInterestMinimumDistance overrides the minimum distance (in km) between points
+// of interest (if two POI appear within this distance, the second one is suppressed). Default
+// 0km (no suppression).
+func WithPointOfInterestMinimumDistance(d float64) Option {
+	return func(c *GPXSummarizerConfig) {
+		c.PointOfInterestMinimumDistance = d
+	}
+}
+
+func WithMinimumSettlement(s string) Option {
+	rank, ok := populatedPlaceRank[s]
+	if !ok {
+		panic(fmt.Sprintf("invalid settlement type: %s", s))
+	}
+	return func(c *GPXSummarizerConfig) {
+		c.MinimumSettlementRank = rank
+	}
+}
+
 type GPXSummarizer struct {
 	poi   *rtreego.Rtree
 	trans osgb.CoordinateTransformer
-	minDist           float64
+	conf  GPXSummarizerConfig
-	minSettlementRank int
 }
 
-func NewGPXSummarizer() (*GPXSummarizer, error) {
+func NewGPXSummarizer(opts ...Option) (*GPXSummarizer, error) {
+	conf := DefaultGPXSummarizerConfig
+	for _, f := range opts {
+		f(&conf)
+	}
 	trans, err := osgb.NewOSTN15Transformer()
 	if err != nil {
 		return nil, err
@@ -38,15 +103,7 @@ func NewGPXSummarizer() (*GPXSummarizer, error) {
 	if err != nil {
 		return nil, err
 	}
-	return &GPXSummarizer{poi: rt, trans: trans, minDist: 0.2, minSettlementRank: 1}, nil
+	return &GPXSummarizer{poi: rt, trans: trans, conf: conf}, nil
-}
-
-func (gs *GPXSummarizer) SetMinSettlement(t string) {
-	gs.minSettlementRank = populatedPlaceRank[t]
-}
-
-func (gs *GPXSummarizer) SetMinDistance(d float64) {
-	gs.minDist = d
 }
 
 func distance(p1, p2 rtreego.Point) float64 {
@@ -108,7 +165,6 @@ func (gs *GPXSummarizer) SummarizeTrack(r io.Reader, stops *rtreego.Rtree) (*Tra
 	var prevPlace string
 	var prevPlacePoint rtreego.Point
 	var prevPoint rtreego.Point
-	var prevStop *cafes.RefreshmentStop
 	var start rtreego.Point
 	var dN, dE float64
 
@@ -141,25 +197,44 @@ func (gs *GPXSummarizer) SummarizeTrack(r io.Reader, stops *rtreego.Rtree) (*Tra
 				s.Distance += distance(thisPoint, prevPoint)
 				dE += thisPoint[0] - start[0]
 				dN += thisPoint[1] - start[1]
-				if nn.Contains(thisPoint) {
+				if nn.Contains(thisPoint) && populatedPlaceRank[nn.Type] >= gs.conf.MinimumSettlementRank {
 					s.Counties[nn.County]++
-					if nn.Name != prevPlace &&
+					seenRecently := false
-						distance(thisPoint, prevPlacePoint) > gs.minDist &&
+					for i := len(s.PointsOfInterest) - 1; i >= 0; i-- {
-						populatedPlaceRank[nn.Type] >= gs.minSettlementRank {
+						if i < len(s.PointsOfInterest)-1 && s.Distance-s.PointsOfInterest[i].Distance > gs.conf.PointOfInterestDuplicateDistance {
+							break
+						}
+						if nn.Name == s.PointsOfInterest[i].Name {
+							seenRecently = true
+							break
+						}
+					}
+					if !seenRecently && distance(thisPoint, prevPlacePoint) > gs.conf.PointOfInterestMinimumDistance {
 						s.PointsOfInterest = append(s.PointsOfInterest, POI{Name: nn.Name, Type: nn.Type, Distance: s.Distance})
 						prevPlace = nn.Name
 						prevPlacePoint = thisPoint
 					}
 				}
 				if stops != nil {
-					stop, ok := stops.NearestNeighbor(thisPoint).(*cafes.RefreshmentStop)
+					for _, nearbyStop := range stops.SearchIntersect(thisPoint.ToRect(gs.conf.CoffeeStopSearchRectangleSize)) {
-					if ok && stop.Contains(thisPoint) && (prevStop == nil || stop.Name != prevStop.Name) {
+						stop := nearbyStop.(*cafes.RefreshmentStop)
+						seenRecently := false
+						for i := len(s.RefreshmentStops) - 1; i >= 0; i-- {
+							if i < len(s.RefreshmentStops)-1 && s.Distance-s.RefreshmentStops[i].Distance > gs.conf.CoffeeStopDuplicateDistance {
+								break
+							}
+							if s.RefreshmentStops[i].Name == stop.Name {
+								seenRecently = true
+								break
+							}
+						}
+						if !seenRecently {
 							s.RefreshmentStops = append(s.RefreshmentStops, RefreshmentStop{
 								Name:     stop.Name,
 								Url:      stop.Url,
 								Distance: s.Distance,
 							})
-						prevStop = stop
+						}
 					}
 				}
 				prevPoint = thisPoint