Improve handling of multi-column unique checks

causes/causes.go

@@ -226,6 +226,22 @@ func (e *DuplicateIDError) Error() string {
 
 ////////////////////////////
 
+// DuplicateIDError reports when a unique ID is used more than once in a file.
+type DuplicateKeyError struct {
+	bc
+}
+
+// NewDuplicateIDError returns a new DuplicateIDErrror
+func NewDuplicateKeyError(eid string) *DuplicateKeyError {
+	return &DuplicateKeyError{bc: bc{Value: eid}}
+}
+
+func (e *DuplicateKeyError) Error() string {
+	return fmt.Sprintf("entity with fields '%s' is present more than once", e.Value)
+}
+
+////////////////////////////
+
 // DuplicateServiceExceptionError reports when a (service_id,date) value is present more than once.
 type DuplicateServiceExceptionError struct {
 	ServiceID string

copier/copier.go

@@ -243,16 +243,14 @@ func NewCopier(reader adapters.Reader, writer adapters.Writer, opts Options) (*C
 
 	// Default set of validators
 	if !opts.NoValidators {
-		copier.AddValidator(&rules.EntityDuplicateCheck{}, 0)
+		copier.AddValidator(&rules.EntityDuplicateIDCheck{}, 0)
+		copier.AddValidator(&rules.EntityDuplicateKeyCheck{}, 0)
 		copier.AddValidator(&rules.ValidFarezoneCheck{}, 0)
 		copier.AddValidator(&rules.AgencyIDConditionallyRequiredCheck{}, 0)
 		copier.AddValidator(&rules.StopTimeSequenceCheck{}, 0)
 		copier.AddValidator(&rules.InconsistentTimezoneCheck{}, 0)
 		copier.AddValidator(&rules.ParentStationLocationTypeCheck{}, 0)
 		copier.AddValidator(&rules.CalendarDuplicateDates{}, 0)
-		copier.AddValidator(&rules.DuplicateFareLegRuleCheck{}, 0)
-		copier.AddValidator(&rules.DuplicateFareTransferRuleCheck{}, 0)
-		copier.AddValidator(&rules.DuplicateFareProductCheck{}, 0)
 	}
 
 	// Default extensions

gtfs/fare_leg_rule.go

@@ -1,6 +1,8 @@
 package gtfs
 
 import (
+	"fmt"
+
 	"github.com/interline-io/transitland-lib/tt"
 )
 
@@ -33,3 +35,16 @@ func (ent *FareLegRule) TableName() string {
 func (ent *FareLegRule) GroupKey() (string, string) {
 	return "leg_group_id", ent.LegGroupID.Val
 }
+
+func (ent *FareLegRule) DuplicateKey() string {
+	key := fmt.Sprintf(
+		"fare_product_id:'%s' network_id:'%s' from_area_id:'%s' to_area_id:'%s' from_timeframe_group_id:'%s' to_timeframe_group_id:'%s'",
+		ent.FareProductID.Val,
+		ent.NetworkID.Val,
+		ent.FromAreaID.Val,
+		ent.ToAreaID.Val,
+		ent.FromTimeframeGroupID.Val,
+		ent.ToTimeframeGroupID.Val,
+	)
+	return key
+}

gtfs/fare_product.go

@@ -1,6 +1,8 @@
 package gtfs
 
 import (
+	"fmt"
+
 	"github.com/interline-io/transitland-lib/causes"
 	"github.com/interline-io/transitland-lib/tt"
 )
@@ -54,3 +56,12 @@ func (ent *FareProduct) ConditionalErrors() (errs []error) {
 	}
 	return errs
 }
+
+func (ent *FareProduct) DuplicateKey() string {
+	return fmt.Sprintf(
+		"fare_product_id:'%s' rider_category_id:'%s' fare_media_id:'%s'",
+		ent.FareProductID,
+		ent.RiderCategoryID,
+		ent.FareMediaID,
+	)
+}

gtfs/fare_rule.go

@@ -1,6 +1,8 @@
 package gtfs
 
 import (
+	"fmt"
+
 	"github.com/interline-io/transitland-lib/tt"
 )
 
@@ -23,3 +25,13 @@ func (ent *FareRule) Filename() string {
 func (ent *FareRule) TableName() string {
 	return "gtfs_fare_rules"
 }
+
+func (ent *FareRule) DuplicateKey() string {
+	return fmt.Sprintf(
+		"route_id:'%s' origin_id:'%s' destination_id:'%s' contains_id:'%s'",
+		ent.RouteID.Val,
+		ent.OriginID.Val,
+		ent.DestinationID.Val,
+		ent.ContainsID.Val,
+	)
+}

gtfs/fare_transfer_rule.go

@@ -54,3 +54,15 @@ func (ent *FareTransferRule) ConditionalErrors() (errs []error) {
 	}
 	return errs
 }
+
+func (ent *FareTransferRule) DuplicateKey() string {
+	return fmt.Sprintf(
+		"fare_product_id:'%s' from_leg_group_id:'%s' to_leg_group_id:'%s' filter_fare_product_id:'%s' transfer_count:%d duration_limit:%d",
+		ent.FareProductID.Val,
+		ent.FromLegGroupID.Val,
+		ent.ToLegGroupID.Val,
+		ent.FilterFareProductID.Val,
+		ent.TransferCount.Val,
+		ent.DurationLimit.Val,
+	)
+}

gtfs/route_network.go

@@ -20,6 +20,10 @@ func (ent *RouteNetwork) TableName() string {
 	return "gtfs_route_networks"
 }
 
-func (ent *RouteNetwork) EntityKey() string {
-	return fmt.Sprintf("%s:%s", ent.NetworkID.Val, ent.RouteID.Val)
+func (ent *RouteNetwork) DuplicateKey() string {
+	return fmt.Sprintf(
+		"network_id:'%s' route_id:'%s'",
+		ent.NetworkID.Val,
+		ent.RouteID.Val,
+	)
 }