"first_null_row" configuration option

internal/topdata/native.go

@@ -30,6 +30,7 @@ type nativeDataset struct {
 	Columns          []string
 	ValueEncodings   map[string]project.TopDataValueEncodingConfig
 	ValueDefaults    map[string]any
+	RowGeneration    string
 	CompareReference bool
 }
 
@@ -240,6 +241,7 @@ func buildNativeUnchecked(p *project.Project, opts NativeBuildOptions, progress
 	}
 	datasets = applyTopDataValueEncodings(datasets, p.Config.TopData.ValueEncodings)
 	datasets = applyTopDataValueDefaults(datasets, p.Config.TopData.ValueDefaults)
+	datasets = applyTopDataRowGeneration(datasets, p.Config.TopData.RowGeneration)
 	registryDatasets, err := collectGeneratedRegistryDatasets(dataDir)
 	if err != nil {
 		return BuildResult{}, err
@@ -448,6 +450,30 @@ func applyTopDataValueDefaults(datasets []nativeDataset, defaults []project.TopD
 	return out
 }
 
+func applyTopDataRowGeneration(datasets []nativeDataset, rules []project.TopDataRowGenerationConfig) []nativeDataset {
+	if len(rules) == 0 {
+		return datasets
+	}
+	out := append([]nativeDataset(nil), datasets...)
+	for index := range out {
+		for _, rule := range rules {
+			datasetName := strings.TrimSpace(rule.Dataset)
+			if datasetName == "" {
+				datasetName = strings.TrimSpace(rule.Namespace)
+			}
+			if filepath.ToSlash(datasetName) != out[index].Name {
+				continue
+			}
+			mode := strings.TrimSpace(rule.Mode)
+			if mode == "" {
+				mode = "after_base"
+			}
+			out[index].RowGeneration = mode
+		}
+	}
+	return out
+}
+
 func nativeCompileGroup(datasetName string) string {
 	head, _, ok := strings.Cut(datasetName, "/")
 	if !ok {
@@ -785,6 +811,7 @@ func collectBaseDataset(dataset nativeDataset) (nativeCollectedDataset, error) {
 	rowByID := map[int]map[string]any{}
 	rowByKey := map[string]map[string]any{}
 	usedIDs := map[int]struct{}{}
+	firstNullRowMode := dataset.RowGeneration == "first_null_row"
 	lockModified := false
 	lockAdded := 0
 	lockPruned := 0
@@ -798,8 +825,28 @@ func collectBaseDataset(dataset nativeDataset) (nativeCollectedDataset, error) {
 		}
 	}
 
+	if firstNullRowMode {
+		for key, rowID := range lockData {
+			if rowID <= baseBoundaryID {
+				continue
+			}
+			if _, ok := baseRowKeys[key]; ok {
+				continue
+			}
+			if explicitID, ok := explicitModuleIDs[key]; ok && explicitID == rowID {
+				continue
+			}
+			delete(lockData, key)
+			lockModified = true
+			lockPruned++
+		}
+	}
+
 	for key, rowID := range lockData {
 		if rowID <= baseBoundaryID {
+			if firstNullRowMode {
+				continue
+			}
 			if _, ok := baseRowKeys[key]; ok {
 				continue
 			}
@@ -823,8 +870,11 @@ func collectBaseDataset(dataset nativeDataset) (nativeCollectedDataset, error) {
 		if err != nil {
 			return nativeCollectedDataset{}, fmt.Errorf("dataset %s: %w", dataset.Name, err)
 		}
-		rows = append(rows, row)
 		rowID := row["id"].(int)
+		if firstNullRowMode && isNativeAllocationNullRow(row, columns) {
+			continue
+		}
+		rows = append(rows, row)
 		if key, ok := row["key"].(string); ok && key != "" {
 			if baseRowKeyCounts[key] > 1 {
 				if lockedID, ok := lockData[key]; !ok || lockedID != rowID {
@@ -843,8 +893,10 @@ func collectBaseDataset(dataset nativeDataset) (nativeCollectedDataset, error) {
 		}
 		usedIDs[rowID] = struct{}{}
 	}
-	for rowID := 0; rowID <= baseBoundaryID; rowID++ {
-		usedIDs[rowID] = struct{}{}
+	if !firstNullRowMode {
+		for rowID := 0; rowID <= baseBoundaryID; rowID++ {
+			usedIDs[rowID] = struct{}{}
+		}
 	}
 	for _, rowID := range explicitModuleIDs {
 		usedIDs[rowID] = struct{}{}
@@ -4457,6 +4509,18 @@ func ensureRowsExposeColumns(rows []map[string]any, columns []string) {
 	}
 }
 
+func isNativeAllocationNullRow(row map[string]any, columns []string) bool {
+	if key, ok := row["key"].(string); ok && strings.TrimSpace(key) != "" && strings.TrimSpace(key) != nullValue {
+		return false
+	}
+	for _, column := range columns {
+		if !isNullLikeValue(row[column]) {
+			return false
+		}
+	}
+	return true
+}
+
 func canonicalizeBaseRow(dataset nativeDataset, columns []string, raw map[string]any, index int) (map[string]any, error) {
 	rowID := index
 	if value, ok := raw["id"]; ok {