test(search): pin the geo predicates in the parity suite

pkg/kql/error.go

@@ -39,6 +39,16 @@ func (e UnsupportedTimeRangeError) Error() string {
 	return fmt.Sprintf("unable to convert '%v' to a time range", e.Value)
 }
 
+// GeoFieldError records a geo predicate that was used on a field that is not
+// indexed as a geopoint.
+type GeoFieldError struct {
+	Key string
+}
+
+func (e GeoFieldError) Error() string {
+	return fmt.Sprintf("geo predicate on non-geo field %q", e.Key)
+}
+
 // IsValidationError reports whether err is one of the KQL parse/validation
 // errors produced by this package, i.e. the query itself is at fault and the
 // caller should treat it as a bad request.
@@ -47,9 +57,11 @@ func IsValidationError(err error) bool {
 		startsWithBinaryOperator *StartsWithBinaryOperatorError
 		namedGroupInvalidNodes   *NamedGroupInvalidNodesError
 		unsupportedTimeRange     *UnsupportedTimeRangeError
+		geoField                 *GeoFieldError
 	)
 
 	return errors.As(err, &startsWithBinaryOperator) ||
 		errors.As(err, &namedGroupInvalidNodes) ||
-		errors.As(err, &unsupportedTimeRange)
+		errors.As(err, &unsupportedTimeRange) ||
+		errors.As(err, &geoField)
 }

services/search/pkg/opensearch/internal/convert/kql_transpile.go

@@ -20,7 +20,7 @@ import (
 func geoQueryField(key string) (string, error) {
 	field, ok := query.ResolveGeoField(key)
 	if !ok {
-		return "", fmt.Errorf("geo predicate on non-geo field %q", key)
+		return "", &kql.GeoFieldError{Key: key}
 	}
 	return field, nil
 }

services/search/pkg/parity/README.md

@@ -418,6 +418,26 @@ Fixtures:
 | RANGE-07 | `mtime>2021` | no match | no match | no match | ✅ |
 | RANGE-08 | `name>100` | no match | no match | no match | ✅ |
 
+### geo
+
+Fixtures:
+
+- `vienna.txt`
+- `bruck.txt`
+- `berlin.txt`
+- `nowhere.txt`
+
+| Case | Query | expected | bleve | OpenSearch | same? |
+|---|---|---|---|---|---|
+| GEO-01 | `location:geo.distance(48.2082, 16.3738, 5km)` | vienna.txt | vienna.txt | vienna.txt | ✅ |
+| GEO-02 | `location:geo.distance(48.2082, 16.3738, 50km)` | bruck.txt, vienna.txt | bruck.txt, vienna.txt | bruck.txt, vienna.txt | ✅ |
+| GEO-03 | `location:geo.distance(48.2082, 16.3738, 1m)` | vienna.txt | vienna.txt | vienna.txt | ✅ |
+| GEO-04 | `location:geo.bbox(47.9, 16.1, 48.3, 16.5)` | vienna.txt | vienna.txt | vienna.txt | ✅ |
+| GEO-05 | `location:geo.bbox(48.3, 16.1, 47.9, 16.5)` | vienna.txt | vienna.txt | vienna.txt | ✅ |
+| GEO-06 | `location:geo.polygon(48.3 16.1, 48.3 16.5, 47.9 16.5, 47.9 16.1)` | vienna.txt | vienna.txt | vienna.txt | ✅ |
+| GEO-07 | `location:geo.distance(0, 0, 20000km)` | berlin.txt, bruck.txt, vienna.txt | berlin.txt, bruck.txt, vienna.txt | berlin.txt, bruck.txt, vienna.txt | ✅ |
+| GEO-08 | `name:geo.distance(48.2082, 16.3738, 5km)` | bad request | bad request | bad request | ✅ |
+
 ### scope
 
 Fixtures:

services/search/pkg/parity/parity_test.go

@@ -109,6 +109,7 @@ func queryGroups() []queryGroup {
 		stressGroup(),
 		everythingGroup(),
 		rangeGroup(),
+		geoGroup(),
 		scopeGroup(),
 		invalidGroup(),
 	}

services/search/pkg/parity/query_geo_test.go

@@ -0,0 +1,40 @@
+package parity
+
+import (
+	libregraph "github.com/opencloud-eu/libre-graph-api-go"
+
+	"github.com/opencloud-eu/opencloud/services/search/pkg/search"
+)
+
+func at(lat, lon float64) fixtureOption {
+	return withLocation(&libregraph.GeoCoordinates{
+		Latitude:  libregraph.PtrFloat64(lat),
+		Longitude: libregraph.PtrFloat64(lon),
+	})
+}
+
+func geoGroup() queryGroup {
+	return queryGroup{
+		name: "geo",
+		fixtures: []search.Resource{
+			// vienna, and a second point ~35km east of it
+			fixtureDoc("vienna.txt", at(48.2082, 16.3738)),
+			fixtureDoc("bruck.txt", at(48.2082, 16.8500)),
+			fixtureDoc("berlin.txt", at(52.5200, 13.4050)),
+			fixtureDoc("nowhere.txt"),
+		},
+		cases: []queryCase{
+			{id: 1, query: `location:geo.distance(48.2082, 16.3738, 5km)`, want: []string{"vienna.txt"}},
+			{id: 2, query: `location:geo.distance(48.2082, 16.3738, 50km)`, want: []string{"vienna.txt", "bruck.txt"}},
+			{id: 3, query: `location:geo.distance(48.2082, 16.3738, 1m)`, want: []string{"vienna.txt"}},
+			{id: 4, query: `location:geo.bbox(47.9, 16.1, 48.3, 16.5)`, want: []string{"vienna.txt"}},
+			// the corners may come in either order, both name the same box
+			{id: 5, query: `location:geo.bbox(48.3, 16.1, 47.9, 16.5)`, want: []string{"vienna.txt"}},
+			{id: 6, query: `location:geo.polygon(48.3 16.1, 48.3 16.5, 47.9 16.5, 47.9 16.1)`, want: []string{"vienna.txt"}},
+			// a document without a location is never a geo hit
+			{id: 7, query: `location:geo.distance(0, 0, 20000km)`, want: []string{"vienna.txt", "bruck.txt", "berlin.txt"}},
+			// geo predicates only apply to geopoint fields, both engines reject the rest
+			{id: 8, query: `name:geo.distance(48.2082, 16.3738, 5km)`, wantBadRequest: true},
+		},
+	}
+}

services/search/pkg/query/bleve/compiler.go

@@ -47,7 +47,7 @@ var bleveEscaper = strings.NewReplacer(
 func geoQueryField(key string) (string, error) {
 	field, ok := searchQuery.ResolveGeoField(key)
 	if !ok {
-		return "", fmt.Errorf("geo predicate on non-geo field %q", key)
+		return "", &kql.GeoFieldError{Key: key}
 	}
 	return field, nil
 }