🐛 Fix selector match count to show total matches

changelog.md

@@ -597,4 +597,12 @@ Simplified data extraction by merging --extract and --extract-data flags:
 - Changed --extract to always output data in YAML format
 - Removed redundant --extract-data flag
 - Maintained template support with --extract-template and config file templates
-- Improved consistency in data output formats
+- Improved consistency in data output formats
+
+# Fixed Selector Match Count
+
+Fixed the selector match count to show the total number of matches instead of the truncated sample count:
+
+- Count now shows total number of matches before sample truncation
+- Sample count limit only affects displayed samples, not the total count
+- Provides more accurate match statistics while keeping output manageable

cmd/tools/test-html-selector/main.go

@@ -35,9 +35,9 @@ type Selector struct {
 }
 
 type SimplifiedSample struct {
-	HTML    []htmlsimplifier.Document `yaml:"html"`
-	Context []htmlsimplifier.Document `yaml:"context"`
-	Path    string                    `yaml:"path"`
+	HTML    []htmlsimplifier.Document `yaml:"html,omitempty"`
+	Context []htmlsimplifier.Document `yaml:"context,omitempty"`
+	Path    string                    `yaml:"path,omitempty"`
 }
 
 type SimplifiedResult struct {
@@ -281,9 +281,9 @@ func (c *TestHTMLSelectorCommand) RunIntoWriter(
 
 	// If using extract or extract-template, process all matches without sample limit
 	if s.Extract || s.ExtractTemplate != "" {
-		extractedData := make(map[string][]string)
+		extractedData := make(map[string][]interface{})
 		for _, result := range results {
-			var matches []string
+			var matches []interface{}
 			for _, sample := range result.Samples {
 				// Process HTML content
 				htmlDocs, err := simplifier.ProcessHTML(sample.HTML)
@@ -293,6 +293,8 @@ func (c *TestHTMLSelectorCommand) RunIntoWriter(
 							matches = append(matches, doc.Text)
 						} else if doc.Markdown != "" {
 							matches = append(matches, doc.Markdown)
+						} else {
+							matches = append(matches, doc)
 						}
 					}
 				}

cmd/tools/test-html-selector/selector.go

@@ -166,6 +166,8 @@ func (st *SelectorTester) Run(ctx context.Context) ([]SelectorResult, error) {
 			return nil, fmt.Errorf("selector '%s' failed: %w", sel.Name, err)
 		}
 
+		totalCount := len(samples)
+
 		// Limit samples to configured count
 		if len(samples) > st.config.Config.SampleCount {
 			samples = samples[:st.config.Config.SampleCount]
@@ -175,7 +177,7 @@ func (st *SelectorTester) Run(ctx context.Context) ([]SelectorResult, error) {
 			Name:     sel.Name,
 			Selector: sel.Selector,
 			Type:     sel.Type,
-			Count:    len(samples),
+			Count:    totalCount,
 			Samples:  samples,
 		})
 	}