fix: don't error on duplicate fields

Signed-off-by: Joseph Kato <joseph@jdkato.io>

Author: jdkato

internal/lint/metadata.go

@@ -14,22 +14,24 @@ import (
 func (l Linter) lintMetadata(f *core.File) error {
 	metadata := make(map[string]any)
 
-	_, err := frontmatter.Parse(strings.NewReader(f.Content), &metadata)
+	body, err := frontmatter.Parse(strings.NewReader(f.Content), &metadata)
 	if errors.Is(err, frontmatter.ErrNotFound) {
-		// No front matter found, return the original content.
 		return nil
 	} else if err != nil {
 		return err
 	}
 
-	seen := make(map[string]int)
+	frontmatter, fmErr := extractFrontMatter(f.Content, string(body))
+	if fmErr != nil {
+		return fmErr
+	}
+
 	for key, value := range metadata {
 		if s, ok := value.(string); ok {
-			i, line := findLineBySubstring(f.Content, s, seen)
+			i, _ := findBestLineBySubstring(frontmatter, s)
 			if i < 0 {
-				return core.NewE100(f.Path, fmt.Errorf("'%s' not found", s))
+				continue
 			}
-			seen[line] = i
 
 			scope := "text.frontmatter." + key + f.RealExt
 			block := nlp.NewLinedBlock(f.Content, s, scope, i-1)
@@ -43,3 +45,11 @@ func (l Linter) lintMetadata(f *core.File) error {
 
 	return nil
 }
+
+func extractFrontMatter(file, body string) (string, error) {
+	startIndex := strings.Index(file, body)
+	if startIndex == -1 {
+		return "", fmt.Errorf("body not found in the file")
+	}
+	return file[:startIndex], nil
+}

internal/lint/util.go

@@ -3,9 +3,38 @@ package lint
 import (
 	"strings"
 
+	"github.com/adrg/strutil"
+	"github.com/adrg/strutil/metrics"
+
 	"github.com/errata-ai/vale/v3/internal/core"
 )
 
+func findBestLineBySubstring(s, sub string) (int, string) {
+	if strings.Count(sub, "\n") > 0 {
+		sub = strings.Split(sub, "\n")[0]
+	}
+
+	bestMatchLine := -1
+	bestMatch := ""
+	bestMatchDistance := -1.0
+
+	metric := metrics.NewLevenshtein()
+	for i, line := range strings.Split(s, "\n") {
+		if !strings.Contains(line, sub) {
+			continue
+		}
+
+		distance := strutil.Similarity(line, sub, metric)
+		if bestMatchLine == -1 || distance < bestMatchDistance {
+			bestMatchDistance = distance
+			bestMatchLine = i + 1
+			bestMatch = line
+		}
+	}
+
+	return bestMatchLine, bestMatch
+}
+
 func findLineBySubstring(s, sub string, seen map[string]int) (int, string) {
 	if strings.Count(sub, "\n") > 0 {
 		sub = strings.Split(sub, "\n")[0]

testdata/features/frontmatter.feature

@@ -43,4 +43,7 @@ Feature: Frontmatter
             test.rst:81:3:vale.Annotations:'TODO' left in text
             test.rst:81:38:vale.Annotations:'XXX' left in text
             test.rst:87:10:vale.Annotations:'TODO' left in text
+            test2.md:2:8:Meta.Title:'How the SDK works' should be in title case
+            test2.md:5:3:vale.Annotations:'TODO' left in text
+            test2.mdx:2:8:Meta.Title:'How the SDK works' should be in title case
             """

testdata/fixtures/frontmatter/.vale.ini

@@ -1,6 +1,6 @@
 StylesPath = ../../styles
 MinAlertLevel = suggestion
 
-[*.{md,rst,adoc}]
+[*.{md,rst,adoc,mdx}]
 vale.Annotations = YES
 Meta.Title = YES

testdata/fixtures/frontmatter/test.md

@@ -40,14 +40,14 @@ XXX: This is a very interesting sentence that is not in a block.
 this sentence has -- dashes TODO: but it should still work.
 
 | TODO | XXX | FIXME |
-|:----:|:---:|:-----:|
+| :--: | :-: | :---: |
 | one  | two | three |
 
 > **Note**: one two FIXME three.
 
 [![Boxy Monokai ★ Predawn][img-monokai]][img-monokai]
 
-is *TODO* by
+is _TODO_ by
 
 ![TODO](/images/logo.png)
 

testdata/fixtures/frontmatter/test2.md

@@ -0,0 +1,12 @@
+---
+title: How the SDK works
+description: |
+  Understand how the Android SDK loads, tracks, and send data
+  TODO
+maintainedBy:
+  - dc.android@xx.com
+category: android
+lang: en
+---
+
+## Introduction

testdata/fixtures/frontmatter/test2.mdx

@@ -0,0 +1,11 @@
+---
+title: How the SDK works
+description: Understand how the Android SDK loads, tracks, and send data
+maintainedBy:
+  - dc.android@xx.com
+category: android
+lang: en
+---
+
+## Introduction
+