fix: support repeated markdown headers (#181)

Author: bwplotka

CHANGELOG.md

@@ -13,10 +13,12 @@ We use *breaking* word for marking changes that are not backward compatible (rel
 ### Added
 
 * [#86](https://github.com/bwplotka/mdox/pull/86) Add configuration options for sending HTTP requests (to help avoid intermittent errors).
+* [#180](https://github.com/bwplotka/mdox/pull/180) Allow ignoring/re-anchoring local links via localValidator.
 
 ### Fixed
 
 * [#84](https://github.com/bwplotka/mdox/pull/84) Allow quotes in first header.
+* [#181](https://github.com/bwplotka/mdox/pull/181) Add support for repeated markdown headers.
 
 ## [v0.9.0](https://github.com/bwplotka/mdox/releases/tag/v0.9.0)
 

pkg/mdformatter/linktransformer/link.go

@@ -485,50 +485,63 @@ func (l localLinksCache) addRelLinks(localLink string) error {
 
 	// File present, cache presence.
 	ids := make([]string, 0)
-
-	var b []byte
+	idDup := map[string]struct{}{}
 	reader := bufio.NewReader(file)
+
 	for {
-		b, err = reader.ReadBytes('\n')
+		line, err := reader.ReadBytes('\n')
 		if err != nil {
 			if err != io.EOF {
 				return fmt.Errorf("failed to read file %v: %w", localLink, err)
 			}
 			break
 		}
 
-		if bytes.HasPrefix(b, []byte(`#`)) {
-			ids = append(ids, toHeaderID(b))
+		if !bytes.HasPrefix(line, []byte("#")) {
+			if err == io.EOF {
+				break
+			}
+			continue
+		}
+
+		headerID := toHeaderID(line)
+		uniqueID := headerID
+		for i := 1; ; i++ {
+			if _, exists := idDup[uniqueID]; !exists {
+				break
+			}
+			uniqueID = fmt.Sprintf("%s-%d", headerID, i)
+		}
+		idDup[uniqueID] = struct{}{}
+		ids = append(ids, uniqueID)
+
+		if err == io.EOF {
+			break
 		}
 	}
 
 	l[localLink] = &ids
 	return nil
 }
 
+var punctuationRe = regexp.MustCompile(`[^\p{L}\p{N}\p{M}-# ]`)
+
 func toHeaderID(header []byte) string {
+	clean := punctuationRe.ReplaceAll(header, nil)
+	text := bytes.TrimLeft(bytes.ToLower(clean), "# ")
+
 	var id []byte
-	// Remove punctuation from header except '-' or '#'.
-	// '\p{L}\p{N}\p{M}' is the Unicode equivalent of '\w', https://www.regular-expressions.info/unicode.html.
-	punctuation := regexp.MustCompile(`[^\p{L}\p{N}\p{M}-# ]`)
-	header = punctuation.ReplaceAll(header, []byte(""))
-	headerText := bytes.TrimLeft(bytes.ToLower(header), "#")
-	// If header is just punctuation it comes up empty, so it cannot be linked.
-	if len(headerText) <= 1 {
-		return ""
-	}
-
-	for _, h := range headerText[1:] {
-		switch h {
+	for _, r := range string(text) {
+		switch r {
 		case '{':
 			return string(id)
-		case ' ', '-':
+		case ' ', '-', '\n', '\r':
 			id = append(id, '-')
 		default:
-			id = append(id, h)
+			id = append(id, string(r)...)
 		}
 	}
-	return string(id)
+	return strings.Trim(string(id), "-")
 }
 
 func absLocalLink(anchorDir string, docPath string, destination string) string {

pkg/mdformatter/linktransformer/link_test.go

@@ -148,6 +148,27 @@ func TestValidator_TransformDestination(t *testing.T) {
 		testutil.Ok(t, err)
 		testutil.Equals(t, 0, len(diff), diff.String())
 	})
+	t.Run("check valid link; repeated headers", func(t *testing.T) {
+		testFile := filepath.Join(tmpDir, "repo", "docs", "test", "valid-link.md")
+		testutil.Ok(t, os.WriteFile(testFile, []byte(`# Yolo
+
+# Yolo
+
+# Yolo
+
+I should link to [1](#yolo-2) and [2](#yolo-1). Also [3](#yolo).
+`), os.ModePerm))
+
+		diff, err := mdformatter.IsFormatted(context.TODO(), logger, []string{testFile})
+		testutil.Ok(t, err)
+		testutil.Equals(t, 0, len(diff), diff.String())
+
+		diff, err = mdformatter.IsFormatted(context.TODO(), logger, []string{testFile}, mdformatter.WithLinkTransformer(
+			MustNewValidator(logger, []byte(""), anchorDir, nil),
+		))
+		testutil.Ok(t, err)
+		testutil.Equals(t, 0, len(diff), diff.String())
+	})
 
 	t.Run("check valid but same link in diff files", func(t *testing.T) {
 		testFile := filepath.Join(tmpDir, "repo", "docs", "test", "valid-link.md")