fix: Fix geometry parsing

internal/service/dataset.go

@@ -33,7 +33,7 @@ type DatasetRepository interface {
 	MarkParseFailed(ctx context.Context, id uuid.UUID, reason string) error
 	MarkReady(ctx context.Context, id uuid.UUID) error
 	MarkConverted(ctx context.Context, id uuid.UUID, cogKey string, footprint []byte) error
-	SetProperties(ctx context.Context, id uuid.UUID, properties []byte) error
+	SetProperties(ctx context.Context, id uuid.UUID, properties, geometry []byte) error
 	SaveMapping(ctx context.Context, id uuid.UUID, katoColumn string, years []domain.YearColumn) (domain.Dataset, error)
 	ReplaceObservations(ctx context.Context, datasetID uuid.UUID, obs []domain.Observation) error
 	ListObservations(ctx context.Context, datasetID uuid.UUID, katoCode *string, limit, offset int) ([]domain.Observation, error)
@@ -82,11 +82,15 @@ type ColumnParser func(filename string, data []byte) ([]domain.AttributeColumn,
 // RowParser reads every attribute row from a file's raw bytes as name->value maps.
 type RowParser func(filename string, data []byte) ([]map[string]string, error)
 
-// RasterConverter converts a raster file to a Cloud-Optimized GeoTIFF and reads
-// its footprint. It operates on local file paths.
+// RasterConverter converts a raster file to a Cloud-Optimized GeoTIFF, reads its
+// footprint, and dissolves a vector file's features into a single geometry. It
+// operates on local file paths.
 type RasterConverter interface {
 	ToCOG(ctx context.Context, srcPath, dstPath string) error
 	Footprint(ctx context.Context, srcPath string) ([]byte, error)
+	// VectorGeometry returns the combined geometry of a vector file's features as
+	// GeoJSON in EPSG:4326 (nil if the file has no features).
+	VectorGeometry(ctx context.Context, srcPath string) ([]byte, error)
 }
 
 // UploadInput carries everything needed to store a new dataset.
@@ -249,10 +253,12 @@ func (s *DatasetService) Reprocess(ctx context.Context, id uuid.UUID) (domain.Da
 	return dataset, nil
 }
 
-// ExtractProperties reads a plain vector dataset's attribute table and stores it
-// (as a JSON array of row objects) in the properties column, then marks the
-// dataset ready. Invoked by the worker. Parse failures are recorded; storage
-// failures are returned for retry.
+// ExtractProperties reads a plain vector dataset's attribute table and spatial
+// geometry and stores them (the attribute table as a JSON array of row objects
+// in the properties column, the dissolved feature geometry in the geometry
+// column), then marks the dataset ready. Invoked by the worker. Parse failures
+// are recorded; storage failures are returned for retry. Geometry extraction is
+// best-effort: a failure leaves geometry unset rather than failing the job.
 func (s *DatasetService) ExtractProperties(ctx context.Context, id uuid.UUID) error {
 	dataset, err := s.repo.GetByID(ctx, id)
 	if err != nil {
@@ -278,7 +284,31 @@ func (s *DatasetService) ExtractProperties(ctx context.Context, id uuid.UUID) er
 			return err
 		}
 	}
-	return s.repo.SetProperties(ctx, id, properties)
+
+	geometry := s.vectorGeometry(ctx, dataset.Filename, data)
+	return s.repo.SetProperties(ctx, id, properties, geometry)
+}
+
+// vectorGeometry dissolves a vector file's features into a single GeoJSON
+// geometry via the converter. It writes the in-memory bytes to a temp file
+// (preserving the extension so the converter detects the format) because the
+// converter operates on file paths. Best-effort: any failure yields nil.
+func (s *DatasetService) vectorGeometry(ctx context.Context, filename string, data []byte) []byte {
+	ext := strings.ToLower(filepath.Ext(filename))
+	f, err := os.CreateTemp("", "gis-vec-*"+ext)
+	if err != nil {
+		return nil
+	}
+	defer os.Remove(f.Name())
+	if _, err := f.Write(data); err != nil {
+		f.Close()
+		return nil
+	}
+	if err := f.Close(); err != nil {
+		return nil
+	}
+	geom, _ := s.converter.VectorGeometry(ctx, f.Name())
+	return geom
 }
 
 // hasAttributeData reports whether any row carries at least one attribute.

internal/service/dataset_test.go

@@ -125,12 +125,15 @@ func (r *stubDatasetRepo) MarkConverted(_ context.Context, id uuid.UUID, cogKey
 	return nil
 }
 
-func (r *stubDatasetRepo) SetProperties(_ context.Context, id uuid.UUID, properties []byte) error {
+func (r *stubDatasetRepo) SetProperties(_ context.Context, id uuid.UUID, properties, geometry []byte) error {
 	d, ok := r.store[id]
 	if !ok {
 		return domain.ErrNotFound
 	}
 	d.Properties = properties
+	if len(geometry) > 0 {
+		d.Geometry = geometry
+	}
 	d.Status = domain.DatasetStatusReady
 	r.store[id] = d
 	return nil
@@ -192,10 +195,12 @@ func (s *stubEnqueuer) EnqueueConvert(_ context.Context, id uuid.UUID) error {
 
 // stubConverter records raster conversions.
 type stubConverter struct {
-	cogCalls    int
-	toCOGErr    error
-	footprint   []byte
-	footprintFn func(src string) ([]byte, error)
+	cogCalls     int
+	toCOGErr     error
+	footprint    []byte
+	footprintFn  func(src string) ([]byte, error)
+	vectorGeom   []byte
+	vectorGeomFn func(src string) ([]byte, error)
 }
 
 func (c *stubConverter) ToCOG(_ context.Context, _, dst string) error {
@@ -213,6 +218,13 @@ func (c *stubConverter) Footprint(_ context.Context, src string) ([]byte, error)
 	return c.footprint, nil
 }
 
+func (c *stubConverter) VectorGeometry(_ context.Context, src string) ([]byte, error) {
+	if c.vectorGeomFn != nil {
+		return c.vectorGeomFn(src)
+	}
+	return c.vectorGeom, nil
+}
+
 var (
 	noopParser    ColumnParser = func(string, []byte) ([]domain.AttributeColumn, error) { return nil, nil }
 	noopRowParser RowParser    = func(string, []byte) ([]map[string]string, error) { return nil, nil }
@@ -428,7 +440,9 @@ func TestDatasetService_ExtractProperties(t *testing.T) {
 		{"name": "Almaty", "pop": "2000"},
 	}
 	rp := RowParser(func(string, []byte) ([]map[string]string, error) { return rows, nil })
-	svc := NewDatasetService(repo, &stubStore{}, stubCategoryReader{exists: true}, &stubEnqueuer{}, noopParser, rp, &stubConverter{})
+	geom := []byte(`{"type":"GeometryCollection","geometries":[]}`)
+	conv := &stubConverter{vectorGeom: geom}
+	svc := NewDatasetService(repo, &stubStore{}, stubCategoryReader{exists: true}, &stubEnqueuer{}, noopParser, rp, conv)
 
 	if err := svc.ExtractProperties(context.Background(), id); err != nil {
 		t.Fatalf("unexpected error: %v", err)
@@ -444,6 +458,40 @@ func TestDatasetService_ExtractProperties(t *testing.T) {
 	if len(parsed) != 2 {
 		t.Fatalf("want 2 rows in properties, got %d", len(parsed))
 	}
+	if string(got.Geometry) != string(geom) {
+		t.Fatalf("want geometry %s, got %s", geom, got.Geometry)
+	}
+}
+
+func TestDatasetService_ExtractProperties_GeometryBestEffort(t *testing.T) {
+	id := uuid.New()
+	repo := newStubDatasetRepo()
+	repo.store[id] = domain.Dataset{
+		ID: id, FileType: domain.FileTypeVector,
+		Filename: "d.geojson", StorageKey: "k", Status: domain.DatasetStatusProcessing,
+	}
+	rp := RowParser(func(string, []byte) ([]map[string]string, error) {
+		return []map[string]string{{"name": "Astana"}}, nil
+	})
+	// Geometry extraction fails; the job must still succeed with properties set.
+	conv := &stubConverter{vectorGeomFn: func(string) ([]byte, error) {
+		return nil, errors.New("ogr2ogr boom")
+	}}
+	svc := NewDatasetService(repo, &stubStore{}, stubCategoryReader{exists: true}, &stubEnqueuer{}, noopParser, rp, conv)
+
+	if err := svc.ExtractProperties(context.Background(), id); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	got := repo.store[id]
+	if got.Status != domain.DatasetStatusReady {
+		t.Fatalf("want ready, got %q", got.Status)
+	}
+	if got.Geometry != nil {
+		t.Fatalf("expected nil geometry on extraction failure, got %s", got.Geometry)
+	}
+	if got.Properties == nil {
+		t.Fatalf("expected properties to be set despite geometry failure")
+	}
 }
 
 func TestDatasetService_ExtractProperties_NoTable(t *testing.T) {