feat: smarter image url detect

utils/sensitive/api.go

@@ -34,7 +34,7 @@ type ResponseForCheck struct {
 }
 
 func CheckSensitive(params ParamsForCheck) (resp *ResponseForCheck, err error) {
-	images := detect(params.Content)
+	images, clearContent := findImagesInMarkdownContent(params.Content)
 	if len(images) != 0 {
 		for _, img := range images {
 			pass, err := checkValidUrl(img)
@@ -62,13 +62,13 @@ func CheckSensitive(params ParamsForCheck) (resp *ResponseForCheck, err error) {
 		}
 	}
 
-	params.Content = deleteImagesInMarkdown(params.Content)
-	if hasTextUrl(params.Content) {
+	if hasTextUrl(clearContent) {
 		return &ResponseForCheck{
 			Pass:   false,
 			Labels: nil,
 		}, nil
 	}
+	params.Content = clearContent
 
 	return CheckSensitiveText(params)
 }

utils/sensitive/utils.go

@@ -5,36 +5,30 @@ import (
 	"mvdan.cc/xurls/v2"
 	imageUrl "net/url"
 	"regexp"
+	"strings"
 	"treehole_next/config"
 )
 
 var imageRegex = regexp.MustCompile(
-	`!\[.*?]\(([^" )]*)`,
+	`!\[(.*?)]\(([^" )]*?)\s*(".*?")?\)`,
 )
-var deleteImageRegex = regexp.MustCompile(
-	`!\[(.*?)]\(([^" ]*)( ".*")?\)`,
-)
-
-// findImagesInMarkdown 从Markdown文本中查找所有图片链接
-func findImagesInMarkdown(markdown string) []string {
-
-	matches := imageRegex.FindAllStringSubmatch(markdown, -1)
-	images := make([]string, 0, len(matches))
-	for _, match := range matches {
-		images = append(images, match[1])
-	}
-	return images
-}
 
-func detect(markdownText string) []string {
+// findImagesInMarkdown 从Markdown文本中查找所有图片链接，并且返回清除链接之后的文本
+func findImagesInMarkdownContent(content string) (imageUrls []string, clearContent string) {
 
-	var ret []string
-	images := findImagesInMarkdown(markdownText)
-	for _, image := range images {
-		ret = append(ret, image)
-	}
-
-	return ret
+	clearContent = imageRegex.ReplaceAllStringFunc(content, func(s string) string {
+		submatch := imageRegex.FindStringSubmatch(s)
+		altText := submatch[1]
+		imageUrls = append(imageUrls, submatch[2])
+		if len(submatch) > 3 && submatch[3] != "" {
+			// If there is a title, return it along with the alt text
+			title := strings.Trim(submatch[3], "\"")
+			return altText + " " + title
+		}
+		// If there is no title, return the alt text
+		return altText
+	})
+	return
 }
 
 func checkType(params ParamsForCheck) bool {
@@ -60,17 +54,3 @@ func checkValidUrl(input string) (bool, error) {
 	}
 	return true, nil
 }
-
-func deleteImagesInMarkdown(markdown string) string {
-
-	return imageRegex.ReplaceAllStringFunc(markdown, func(s string) string {
-		submatches := deleteImageRegex.FindStringSubmatch(s)
-		altText := submatches[1]
-		if len(submatches) > 3 && submatches[3] != "" {
-			// If there is a title, return it along with the alt text
-			return altText + " " + submatches[3][2:len(submatches[3])-1]
-		}
-		// If there is no title, return the alt text
-		return altText
-	})
-}

utils/sensitive/utils_test.go

@@ -0,0 +1,46 @@
+package sensitive
+
+import (
+	"github.com/stretchr/testify/assert"
+	"testing"
+)
+
+func TestFindImagesInMarkdown(t *testing.T) {
+
+	type wantStruct struct {
+		clearContent string
+		imageUrls    []string
+	}
+	tests := []struct {
+		text string
+		want wantStruct
+	}{
+		{
+			text: `![image1](https://example.com/image1)`,
+			want: wantStruct{
+				clearContent: `image1`,
+				imageUrls:    []string{"https://example.com/image1"},
+			},
+		},
+		{
+			text: `![image1](https://example.com/image1) ![image2](https://example.com/image2)`,
+			want: wantStruct{
+				clearContent: `image1 image2`,
+				imageUrls:    []string{"https://example.com/image1", "https://example.com/image2"},
+			},
+		},
+		{
+			text: `![image1](https://example.com/image1 "title1") ![image2](https://example.com/image2 "title2")`,
+			want: wantStruct{
+				clearContent: `image1 title1 image2 title2`,
+				imageUrls:    []string{"https://example.com/image1", "https://example.com/image2"},
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		imageUrls, cleanText := findImagesInMarkdownContent(tt.text)
+		assert.EqualValues(t, tt.want.clearContent, cleanText, "cleanText should be equal")
+		assert.EqualValues(t, tt.want.imageUrls, imageUrls, "imageUrls should be equal")
+	}
+}