fix(twitter): respect routes to non-user pages

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## Unreleased
 
+- Twitter: Blacklist special pages from being resolved as user pages. (#220)
 - Updated Facebook & Instagram endpoints to oembed v10. (#201)
 
 ## 1.2.1

internal/resolvers/twitter/check.go

@@ -16,6 +16,19 @@ func check(url *url.URL) bool {
 		return true
 	}
 
-	isTwitterUser := twitterUserRegexp.MatchString(url.String())
+	/* Simply matching the regex isn't enough for user pages. Pages like
+	   twitter.com/explore and twitter.com/settings match the expression but do not refer
+	   to a valid user page. We therefore need to check the captured name against a list
+	   of known non-user pages.
+	*/
+	m := twitterUserRegexp.FindAllStringSubmatch(url.String(), -1)
+	if len(m) == 0 || len(m[0]) == 0 {
+		return false
+	}
+	userName := m[0][1]
+
+	_, notAUser := nonUserPages[userName]
+	isTwitterUser := !notAUser
+
 	return isTwitterUser
 }

internal/resolvers/twitter/check_test.go

@@ -0,0 +1,45 @@
+package twitter
+
+import (
+	"net/url"
+	"testing"
+)
+
+func TestShouldMatch(t *testing.T) {
+	tests := []string{
+		"https://twitter.com/pajlada",
+		"http://www.twitter.com/forsen",
+	}
+
+	for _, test := range tests {
+		u, err := url.Parse(test)
+		if err != nil {
+			t.Fatalf("invalid url %s", test)
+		}
+
+		if !check(u) {
+			t.Fatalf("url %s didn't match twitter check while it should have", test)
+		}
+	}
+}
+
+func TestShouldNotMatch(t *testing.T) {
+	tests := []string{
+		"https://google.com",
+		"https://twitter.com/compose",
+		"https://twitter.com/logout",
+		"https://twitter.com/logout",
+		"https://nontwitter.com/forsen",
+	}
+
+	for _, test := range tests {
+		u, err := url.Parse(test)
+		if err != nil {
+			t.Fatalf("invalid url %s", test)
+		}
+
+		if check(u) {
+			t.Fatalf("url %s matched twitter check while it shouldn't have", test)
+		}
+	}
+}

internal/resolvers/twitter/resolver.go

@@ -9,6 +9,7 @@ import (
 	"github.com/Chatterino/api/pkg/cache"
 	"github.com/Chatterino/api/pkg/config"
 	"github.com/Chatterino/api/pkg/resolver"
+	"github.com/Chatterino/api/pkg/utils"
 )
 
 const (
@@ -35,6 +36,25 @@ var (
 	tweetRegexp       = regexp.MustCompile(`(?i)\/.*\/status(?:es)?\/([^\/\?]+)`)
 	twitterUserRegexp = regexp.MustCompile(`(?i)twitter\.com\/([^\/\?\s]+)(\/?$|(\?).*)`)
 
+	/* These routes refer to non-user pages. If the capture group in twitterUserRegexp
+	   matches any of these names, we must not resolve it as a Twitter user link.
+
+	   The pages are listed alphabetically. They were sourced by simply looking around the
+	   Twitter web page. AFAIK, there is no resource describing these "special" routes.
+	*/
+	nonUserPages = utils.SetFromSlice([]interface{}{
+		"compose",
+		"explore",
+		"home",
+		"logout",
+		"messages",
+		"notifications",
+		"search",
+		"settings",
+		"tos",
+		"privacy",
+	})
+
 	tweetTooltipTemplate = template.Must(template.New("tweetTooltip").Parse(tweetTooltip))
 
 	twitterUserTooltipTemplate = template.Must(template.New("twitterUserTooltip").Parse(twitterUserTooltip))

pkg/utils/set.go

@@ -0,0 +1,11 @@
+package utils
+
+// SetFromSlice takes a slice and returns a set (represented as a map of struct{}s)
+func SetFromSlice(slice []interface{}) map[interface{}]struct{} {
+	set := make(map[interface{}]struct{})
+	for _, v := range slice {
+		set[v] = struct{}{}
+	}
+
+	return set
+}