Fix YAML document splitting to only split on unindented --- separators (#341) (#349)

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: graza-io

kubernetes/helm_template_render.go

@@ -111,7 +111,7 @@ func getHelmRenderedTemplatesUncached(ctx context.Context, d *plugin.QueryData,
 					continue
 				}
 
-				splitManifest := strings.Split(manifest.Manifest, "---")
+				splitManifest := yamlDocSeparator.Split(manifest.Manifest, -1)
 				for _, content := range splitManifest {
 					if len(content) == 0 {
 						continue
@@ -185,7 +185,7 @@ func renderedHelmTemplateContentUncached(ctx context.Context, d *plugin.QueryDat
 		}
 		lineInfo := temp[t.Path]
 
-		for _, resource := range strings.Split(t.Data, "---") {
+		for _, resource := range yamlDocSeparator.Split(t.Data, -1) {
 			// Skip empty documents, `Decode` will fail on them
 			// Also, increment the pos to include the separator position (e.g. ---)
 			if len(resource) == 0 {
@@ -347,7 +347,7 @@ func getRawTemplateLineInfo(ctx context.Context, d *plugin.QueryData) (map[strin
 			startLine := 0
 			count := 0
 
-			for _, content := range strings.Split(templateContent, "---") {
+			for _, content := range yamlDocSeparator.Split(templateContent, -1) {
 				// Skip empty documents, `Decode` will fail on them
 				// Also, increment the pos to include the separator position (e.g. ---)
 				if len(content) == 0 {

kubernetes/utils.go

@@ -36,6 +36,12 @@ import (
 	"github.com/turbot/steampipe-plugin-sdk/v5/plugin/transform"
 )
 
+// yamlDocSeparator matches YAML document separator lines: "---" at the start
+// of a line (column 0), optionally followed by trailing whitespace. Per the
+// YAML spec, an indented "---" inside a multi-line string value is NOT a
+// document separator, so we must anchor to the beginning of the line.
+var yamlDocSeparator = regexp.MustCompile(`(?m)^---\s*$`)
+
 type SourceType string
 
 const (
@@ -727,7 +733,7 @@ func parsedManifestFileContentUncached(ctx context.Context, d *plugin.QueryData,
 
 		// Check for the start of the document
 		pos := 0
-		for _, resource := range strings.Split(string(content), "---") {
+		for _, resource := range yamlDocSeparator.Split(string(content), -1) {
 			// Skip empty documents, `Decode` will fail on them
 			// Also, increment the pos to include the separator position (e.g. ---)
 			if len(resource) == 0 {

kubernetes/yaml_split_test.go

@@ -0,0 +1,121 @@
+package kubernetes
+
+import (
+	"reflect"
+	"testing"
+)
+
+func TestYamlDocSeparator(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		expected []string
+	}{
+		{
+			name:     "simple two documents",
+			input:    "apiVersion: v1\nkind: Pod\n---\napiVersion: v1\nkind: Service",
+			expected: []string{"apiVersion: v1\nkind: Pod\n", "\napiVersion: v1\nkind: Service"},
+		},
+		{
+			name:     "separator at start of file",
+			input:    "---\napiVersion: v1\nkind: Pod",
+			expected: []string{"", "\napiVersion: v1\nkind: Pod"},
+		},
+		{
+			name:     "separator with trailing whitespace",
+			input:    "apiVersion: v1\nkind: Pod\n---  \napiVersion: v1\nkind: Service",
+			expected: []string{"apiVersion: v1\nkind: Pod\n", "\napiVersion: v1\nkind: Service"},
+		},
+		{
+			name:     "separator with trailing tab",
+			input:    "apiVersion: v1\nkind: Pod\n---\t\napiVersion: v1\nkind: Service",
+			expected: []string{"apiVersion: v1\nkind: Pod\n", "\napiVersion: v1\nkind: Service"},
+		},
+		{
+			name:  "indented separator inside multiline string is NOT a split point",
+			input: "apiVersion: v1\ndata:\n  install_info: |\n    ---\n    install_method:\n      tool: helm\nkind: ConfigMap",
+			expected: []string{
+				"apiVersion: v1\ndata:\n  install_info: |\n    ---\n    install_method:\n      tool: helm\nkind: ConfigMap",
+			},
+		},
+		{
+			name: "issue 341 exact scenario - ConfigMap with indented separator",
+			input: `---
+apiVersion: v1
+data:
+  install_info: |
+    ---
+    install_method:
+      tool: helm
+      tool_version: Helm
+      installer_version: datadog-3.135.4
+kind: ConfigMap
+metadata:
+  annotations:
+    checksum/install_info: 22bff6a15fb7a4521a3b6a06f55f1fe8ca1570dab4d8bca9f437f28e5301e89a
+  labels:
+    app.kubernetes.io/instance: datadog
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: my-service`,
+			expected: []string{
+				"",
+				"\napiVersion: v1\ndata:\n  install_info: |\n    ---\n    install_method:\n      tool: helm\n      tool_version: Helm\n      installer_version: datadog-3.135.4\nkind: ConfigMap\nmetadata:\n  annotations:\n    checksum/install_info: 22bff6a15fb7a4521a3b6a06f55f1fe8ca1570dab4d8bca9f437f28e5301e89a\n  labels:\n    app.kubernetes.io/instance: datadog\n",
+				"\napiVersion: v1\nkind: Service\nmetadata:\n  name: my-service",
+			},
+		},
+		{
+			name:     "three documents",
+			input:    "---\napiVersion: v1\nkind: Pod\n---\napiVersion: v1\nkind: Service\n---\napiVersion: v1\nkind: Deployment",
+			expected: []string{"", "\napiVersion: v1\nkind: Pod\n", "\napiVersion: v1\nkind: Service\n", "\napiVersion: v1\nkind: Deployment"},
+		},
+		{
+			name:     "no separator - single document",
+			input:    "apiVersion: v1\nkind: Pod\nmetadata:\n  name: test",
+			expected: []string{"apiVersion: v1\nkind: Pod\nmetadata:\n  name: test"},
+		},
+		{
+			name:     "dashes inside a value are not separators",
+			input:    "apiVersion: v1\nkind: ConfigMap\ndata:\n  key: ---some-value---",
+			expected: []string{"apiVersion: v1\nkind: ConfigMap\ndata:\n  key: ---some-value---"},
+		},
+		{
+			name:     "four dashes on own line are not a separator",
+			input:    "apiVersion: v1\nkind: Pod\n----\napiVersion: v1\nkind: Service",
+			expected: []string{"apiVersion: v1\nkind: Pod\n----\napiVersion: v1\nkind: Service"},
+		},
+		{
+			name:  "multiple indented separators inside strings",
+			input: "apiVersion: v1\nkind: ConfigMap\ndata:\n  a: |\n    ---\n    foo: bar\n  b: |\n    ---\n    baz: qux",
+			expected: []string{
+				"apiVersion: v1\nkind: ConfigMap\ndata:\n  a: |\n    ---\n    foo: bar\n  b: |\n    ---\n    baz: qux",
+			},
+		},
+		{
+			name:     "empty input",
+			input:    "",
+			expected: []string{""},
+		},
+		{
+			name:     "only separator",
+			input:    "---",
+			expected: []string{"", ""},
+		},
+		{
+			name:     "consecutive separators",
+			input:    "---\n---\n---",
+			expected: []string{"", "\n", "\n", ""},
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result := yamlDocSeparator.Split(tc.input, -1)
+			if !reflect.DeepEqual(result, tc.expected) {
+				t.Errorf("yamlDocSeparator.Split() mismatch\n  input:    %q\n  got:      %q\n  expected: %q", tc.input, result, tc.expected)
+			}
+		})
+	}
+}