RF: social tags checker (#39)

* RF: social tags checker

* Add some tests

checks/checks.go

@@ -6,16 +6,18 @@ import (
 )
 
 type Checks struct {
-	Carbon *Carbon
-	Rank   *Rank
+	Carbon     *Carbon
+	Rank       *Rank
+	SocialTags *SocialTags
 }
 
 func NewChecks() *Checks {
 	client := &http.Client{
 		Timeout: 5 * time.Second,
 	}
 	return &Checks{
-		Carbon: NewCarbon(client),
-		Rank:   NewRank(client),
+		Carbon:     NewCarbon(client),
+		Rank:       NewRank(client),
+		SocialTags: NewSocialTags(client),
 	}
 }

checks/social_tags.go

@@ -0,0 +1,94 @@
+package checks
+
+import (
+	"context"
+	"net/http"
+
+	"github.com/PuerkitoBio/goquery"
+)
+
+type SocialTagsData struct {
+	Title              string `json:"title"`
+	Description        string `json:"description"`
+	Keywords           string `json:"keywords"`
+	CanonicalUrl       string `json:"canonicalUrl"`
+	OgTitle            string `json:"ogTitle"`
+	OgType             string `json:"ogType"`
+	OgImage            string `json:"ogImage"`
+	OgUrl              string `json:"ogUrl"`
+	OgDescription      string `json:"ogDescription"`
+	OgSiteName         string `json:"ogSiteName"`
+	TwitterCard        string `json:"twitterCard"`
+	TwitterSite        string `json:"twitterSite"`
+	TwitterCreator     string `json:"twitterCreator"`
+	TwitterTitle       string `json:"twitterTitle"`
+	TwitterDescription string `json:"twitterDescription"`
+	TwitterImage       string `json:"twitterImage"`
+	ThemeColor         string `json:"themeColor"`
+	Robots             string `json:"robots"`
+	Googlebot          string `json:"googlebot"`
+	Generator          string `json:"generator"`
+	Viewport           string `json:"viewport"`
+	Author             string `json:"author"`
+	Publisher          string `json:"publisher"`
+	Favicon            string `json:"favicon"`
+}
+
+func (s SocialTagsData) Empty() bool {
+	return (SocialTagsData{}) == s
+}
+
+type SocialTags struct {
+	client *http.Client
+}
+
+func NewSocialTags(client *http.Client) *SocialTags {
+	return &SocialTags{client: client}
+}
+
+func (s *SocialTags) GetSocialTags(ctx context.Context, url string) (*SocialTagsData, error) {
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
+	if err != nil {
+		return nil, err
+	}
+	resp, err := s.client.Do(req)
+	if err != nil {
+		return nil, err
+	}
+	defer resp.Body.Close()
+
+	// Parse HTML document
+	doc, err := goquery.NewDocumentFromReader(resp.Body)
+	if err != nil {
+		return nil, err
+	}
+
+	// Extract social tags metadata
+	tags := &SocialTagsData{
+		Title:              doc.Find("head title").Text(),
+		Description:        doc.Find("meta[name='description']").AttrOr("content", ""),
+		Keywords:           doc.Find("meta[name='keywords']").AttrOr("content", ""),
+		CanonicalUrl:       doc.Find("link[rel='canonical']").AttrOr("href", ""),
+		OgTitle:            doc.Find("meta[property='og:title']").AttrOr("content", ""),
+		OgType:             doc.Find("meta[property='og:type']").AttrOr("content", ""),
+		OgImage:            doc.Find("meta[property='og:image']").AttrOr("content", ""),
+		OgUrl:              doc.Find("meta[property='og:url']").AttrOr("content", ""),
+		OgDescription:      doc.Find("meta[property='og:description']").AttrOr("content", ""),
+		OgSiteName:         doc.Find("meta[property='og:site_name']").AttrOr("content", ""),
+		TwitterCard:        doc.Find("meta[name='twitter:card']").AttrOr("content", ""),
+		TwitterSite:        doc.Find("meta[name='twitter:site']").AttrOr("content", ""),
+		TwitterCreator:     doc.Find("meta[name='twitter:creator']").AttrOr("content", ""),
+		TwitterTitle:       doc.Find("meta[name='twitter:title']").AttrOr("content", ""),
+		TwitterDescription: doc.Find("meta[name='twitter:description']").AttrOr("content", ""),
+		TwitterImage:       doc.Find("meta[name='twitter:image']").AttrOr("content", ""),
+		ThemeColor:         doc.Find("meta[name='theme-color']").AttrOr("content", ""),
+		Robots:             doc.Find("meta[name='robots']").AttrOr("content", ""),
+		Googlebot:          doc.Find("meta[name='googlebot']").AttrOr("content", ""),
+		Generator:          doc.Find("meta[name='generator']").AttrOr("content", ""),
+		Viewport:           doc.Find("meta[name='viewport']").AttrOr("content", ""),
+		Author:             doc.Find("meta[name='author']").AttrOr("content", ""),
+		Publisher:          doc.Find("link[rel='publisher']").AttrOr("href", ""),
+		Favicon:            doc.Find("link[rel='icon']").AttrOr("href", ""),
+	}
+	return tags, nil
+}

checks/social_tags_test.go

@@ -0,0 +1,65 @@
+package checks
+
+import (
+	"context"
+	"net/http"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/xray-web/web-check-api/testutils"
+)
+
+func TestSocialTagsEmpty(t *testing.T) {
+	t.Parallel()
+
+	t.Run("Empty", func(t *testing.T) {
+		t.Parallel()
+
+		s := SocialTagsData{}
+		assert.True(t, s.Empty())
+	})
+
+	t.Run("Not empty", func(t *testing.T) {
+		t.Parallel()
+
+		s := SocialTagsData{
+			Title: "Example Domain",
+		}
+		assert.False(t, s.Empty())
+	})
+}
+
+func TestNewSocialTags(t *testing.T) {
+	t.Parallel()
+
+	t.Run("No social tags", func(t *testing.T) {
+		t.Parallel()
+
+		client := testutils.MockClient(testutils.Response(http.StatusOK, []byte{}))
+		tags, err := NewSocialTags(client).GetSocialTags(context.TODO(), "http://example.com")
+		assert.NoError(t, err)
+		assert.True(t, tags.Empty())
+	})
+
+	t.Run("Social tags", func(t *testing.T) {
+		t.Parallel()
+
+		var html = []byte(`
+		<html>
+			<head>
+				<title>Example Domain</title>
+				<meta name="description" content="Example description">
+				<meta property="og:title" content="Example OG Title">
+			</head>
+			<body></body>
+		</html>
+		`)
+		client := testutils.MockClient(testutils.Response(http.StatusOK, html))
+		tags, err := NewSocialTags(client).GetSocialTags(context.TODO(), "http://example.com")
+		assert.NoError(t, err)
+		assert.False(t, tags.Empty())
+		assert.Equal(t, "Example description", tags.Description)
+		assert.Equal(t, "Example Domain", tags.Title)
+		assert.Equal(t, "Example OG Title", tags.OgTitle)
+	})
+}

handlers/social_tags.go

@@ -1,120 +1,23 @@
 package handlers
 
 import (
-	"errors"
 	"net/http"
 
-	"github.com/PuerkitoBio/goquery"
+	"github.com/xray-web/web-check-api/checks"
 )
 
-type SocialTags struct {
-	Title              string `json:"title"`
-	Description        string `json:"description"`
-	Keywords           string `json:"keywords"`
-	CanonicalUrl       string `json:"canonicalUrl"`
-	OgTitle            string `json:"ogTitle"`
-	OgType             string `json:"ogType"`
-	OgImage            string `json:"ogImage"`
-	OgUrl              string `json:"ogUrl"`
-	OgDescription      string `json:"ogDescription"`
-	OgSiteName         string `json:"ogSiteName"`
-	TwitterCard        string `json:"twitterCard"`
-	TwitterSite        string `json:"twitterSite"`
-	TwitterCreator     string `json:"twitterCreator"`
-	TwitterTitle       string `json:"twitterTitle"`
-	TwitterDescription string `json:"twitterDescription"`
-	TwitterImage       string `json:"twitterImage"`
-	ThemeColor         string `json:"themeColor"`
-	Robots             string `json:"robots"`
-	Googlebot          string `json:"googlebot"`
-	Generator          string `json:"generator"`
-	Viewport           string `json:"viewport"`
-	Author             string `json:"author"`
-	Publisher          string `json:"publisher"`
-	Favicon            string `json:"favicon"`
-}
-
-func isEmpty(tags *SocialTags) bool {
-	return tags.Title == "" &&
-		tags.Description == "" &&
-		tags.Keywords == "" &&
-		tags.CanonicalUrl == "" &&
-		tags.OgTitle == "" &&
-		tags.OgType == "" &&
-		tags.OgImage == "" &&
-		tags.OgUrl == "" &&
-		tags.OgDescription == "" &&
-		tags.OgSiteName == "" &&
-		tags.TwitterCard == "" &&
-		tags.TwitterSite == "" &&
-		tags.TwitterCreator == "" &&
-		tags.TwitterTitle == "" &&
-		tags.TwitterDescription == "" &&
-		tags.TwitterImage == "" &&
-		tags.ThemeColor == "" &&
-		tags.Robots == "" &&
-		tags.Googlebot == "" &&
-		tags.Generator == "" &&
-		tags.Viewport == "" &&
-		tags.Author == "" &&
-		tags.Publisher == "" &&
-		tags.Favicon == ""
-}
-
-func HandleGetSocialTags() http.Handler {
+func HandleGetSocialTags(s *checks.SocialTags) http.Handler {
 	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		rawURL, err := extractURL(r)
 		if err != nil {
 			JSONError(w, ErrMissingURLParameter, http.StatusBadRequest)
 			return
 		}
-
-		// Fetch HTML content from the URL
-		resp, err := http.Get(rawURL.String())
+		tags, err := s.GetSocialTags(r.Context(), rawURL.String())
 		if err != nil {
+			JSONError(w, err, http.StatusInternalServerError)
 			return
 		}
-		defer resp.Body.Close()
-
-		// Parse HTML document
-		doc, err := goquery.NewDocumentFromReader(resp.Body)
-		if err != nil {
-			return
-		}
-
-		// Extract social tags metadata
-		tags := &SocialTags{
-			Title:              doc.Find("head title").Text(),
-			Description:        doc.Find("meta[name='description']").AttrOr("content", ""),
-			Keywords:           doc.Find("meta[name='keywords']").AttrOr("content", ""),
-			CanonicalUrl:       doc.Find("link[rel='canonical']").AttrOr("href", ""),
-			OgTitle:            doc.Find("meta[property='og:title']").AttrOr("content", ""),
-			OgType:             doc.Find("meta[property='og:type']").AttrOr("content", ""),
-			OgImage:            doc.Find("meta[property='og:image']").AttrOr("content", ""),
-			OgUrl:              doc.Find("meta[property='og:url']").AttrOr("content", ""),
-			OgDescription:      doc.Find("meta[property='og:description']").AttrOr("content", ""),
-			OgSiteName:         doc.Find("meta[property='og:site_name']").AttrOr("content", ""),
-			TwitterCard:        doc.Find("meta[name='twitter:card']").AttrOr("content", ""),
-			TwitterSite:        doc.Find("meta[name='twitter:site']").AttrOr("content", ""),
-			TwitterCreator:     doc.Find("meta[name='twitter:creator']").AttrOr("content", ""),
-			TwitterTitle:       doc.Find("meta[name='twitter:title']").AttrOr("content", ""),
-			TwitterDescription: doc.Find("meta[name='twitter:description']").AttrOr("content", ""),
-			TwitterImage:       doc.Find("meta[name='twitter:image']").AttrOr("content", ""),
-			ThemeColor:         doc.Find("meta[name='theme-color']").AttrOr("content", ""),
-			Robots:             doc.Find("meta[name='robots']").AttrOr("content", ""),
-			Googlebot:          doc.Find("meta[name='googlebot']").AttrOr("content", ""),
-			Generator:          doc.Find("meta[name='generator']").AttrOr("content", ""),
-			Viewport:           doc.Find("meta[name='viewport']").AttrOr("content", ""),
-			Author:             doc.Find("meta[name='author']").AttrOr("content", ""),
-			Publisher:          doc.Find("link[rel='publisher']").AttrOr("href", ""),
-			Favicon:            doc.Find("link[rel='icon']").AttrOr("href", ""),
-		}
-
-		if isEmpty(tags) {
-			JSONError(w, errors.New("no metadata found"), http.StatusBadRequest)
-			return
-		}
-
 		JSON(w, tags, http.StatusOK)
 	})
 }