Fix #160: Add proper text escaping in FormatHtml

bukzor · claude · bukzor · commit e35e4d88689e · 2025-11-14T11:28:47.000-06:00
HTML text nodes containing &, <, > were output without escaping, causing xq's output to be unparseable when piped back through xq -j. This commit adds: - New escapeTextContent() function for minimal entity escaping - Modified FormatHtml to escape text nodes with &amp;, &lt;, &gt; - Tests verifying the output is valid XML Example issue: echo '<html>1 &amp; 2</html>' | xq | xq -j # Before: Error - bare & in output # After: Success - properly escaped as &amp; This is a critical fix preventing data corruption when round-tripping HTML through xq. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
diff --git a/cmd/root_test.go b/cmd/root_test.go
@@ -114,6 +114,57 @@ func TestCDATASupport(t *testing.T) {
 	assert.Equal(t, expected, result)
 }
 
+func TestEscapedTextNodes(t *testing.T) {
+	// Test case 1: ampersand entity - reproduce issue #160
+	// xq outputs bare & which fails when parsed as XML (used by -j flag)
+	t.Run("ampersand entity output is valid XML", func(t *testing.T) {
+		input := "<html>1 &amp; 2</html>"
+
+		// First pass: format the HTML
+		reader1 := strings.NewReader(input)
+		var output1 bytes.Buffer
+		err := utils.FormatHtml(reader1, &output1, "", utils.ColorsDisabled)
+		assert.Nil(t, err)
+
+		result1 := strings.TrimSpace(output1.String())
+		t.Logf("First pass output: %q", result1)
+
+		// Second pass: try to parse as XML (this is what `xq -j` does)
+		reader2 := strings.NewReader(result1)
+		_, err = xmlquery.Parse(reader2)
+		assert.Nil(t, err, "xq output should be parseable as XML (for -j flag)")
+	})
+
+	// Test case 2: less-than and greater-than entities - reproduce issue #160
+	// xq outputs bare < and > which are parsed as tags
+	t.Run("less-than and greater-than entities output is valid XML", func(t *testing.T) {
+		input := "<html>is &lt;bold&gt; a valid tag?</html>"
+
+		// First pass: format the HTML
+		reader1 := strings.NewReader(input)
+		var output1 bytes.Buffer
+		err := utils.FormatHtml(reader1, &output1, "", utils.ColorsDisabled)
+		assert.Nil(t, err)
+
+		result1 := strings.TrimSpace(output1.String())
+		t.Logf("First pass output: %q", result1)
+
+		// Second pass: try to parse as XML (this is what `xq -j` does)
+		reader2 := strings.NewReader(result1)
+		doc, err := xmlquery.Parse(reader2)
+		assert.Nil(t, err, "xq output should be parseable as XML (for -j flag)")
+
+		// Verify the text content is preserved correctly
+		if doc != nil {
+			textNode := xmlquery.FindOne(doc, "//html")
+			if textNode != nil {
+				assert.Equal(t, "is <bold> a valid tag?", textNode.InnerText(),
+					"Text content should preserve the literal < and > characters")
+			}
+		}
+	})
+}
+
 func TestProcessAsJSON(t *testing.T) {
 	tests := []struct {
 		name        string
diff --git a/internal/utils/utils.go b/internal/utils/utils.go
@@ -338,6 +338,9 @@ func FormatHtml(reader io.Reader, writer io.Writer, indent string, colors int) e
 		case html.TextToken:
 			str := normalizeSpaces(string(tokenizer.Text()), indent, level)
 			hasContent = str != ""
+			if hasContent {
+				str = escapeTextContent(str)
+			}
 			_, _ = fmt.Fprint(writer, str)
 		case html.StartTagToken, html.SelfClosingTagToken:
 			if level > 0 {
@@ -585,6 +588,15 @@ func escapeText(input string) (string, error) {
 	return result, nil
 }
 
+func escapeTextContent(input string) string {
+	// Only escape the minimal set of characters needed for text content
+	// to avoid XML parsing errors: & < >
+	result := strings.ReplaceAll(input, "&", "&amp;")
+	result = strings.ReplaceAll(result, "<", "&lt;")
+	result = strings.ReplaceAll(result, ">", "&gt;")
+	return result
+}
+
 func normalizeSpaces(input string, indent string, level int) string {
 	if strings.TrimSpace(input) == "" {
 		input = ""
