Add helper method to parse tweet URLs

scraper/user.go

@@ -78,6 +78,20 @@ Joined %s
     return ret
 }
 
+/**
+ * Given a tweet URL, return the corresponding user handle.
+ * If tweet url is not valid, return an error.
+ */
+func ParseHandleFromTweetUrl(tweet_url string) (UserHandle, error) {
+    r := regexp.MustCompile(`https://twitter.com/(\w+)/status/\d+`)
+    matches := r.FindStringSubmatch(tweet_url)
+    if len(matches) != 2 {  // matches[0] is the full string
+        return "", fmt.Errorf("Invalid tweet url: %s", tweet_url)
+    }
+    return UserHandle(matches[1]), nil
+}
+
+
 // Turn an APIUser, as returned from the scraper, into a properly structured User object
 func ParseSingleUser(apiUser APIUser) (ret User, err error) {
     ret.ID = UserID(apiUser.ID)

scraper/user_test.go

@@ -85,3 +85,31 @@ func TestParseSingleUser(t *testing.T) {
 		t.Errorf("Expected %q, got %q", expected_id, user.PinnedTweet)
 	}
 }
+
+/**
+ * Should extract a user handle from a tweet URL, or fail if URL is invalid
+ */
+func TestParseHandleFromTweetUrl(t *testing.T) {
+	// Test valid tweet url
+	url := "https://twitter.com/kanesays23/status/1429583672827465730"
+	expected_user := scraper.UserHandle("kanesays23")
+	result, err := scraper.ParseHandleFromTweetUrl(url)
+	if err != nil {
+		t.Errorf("Unexpected error while parsing handle: %s", err)
+	}
+	if result != expected_user {
+		t.Errorf("Expected handle %q, got %q", expected_user, result)
+	}
+
+	// Test invalid url
+	_, err = scraper.ParseHandleFromTweetUrl("awjgwekf")
+	if err == nil {
+		t.Errorf("Should have produced an error for invalid URL")
+	}
+
+	// Test empty string
+	_, err = scraper.ParseHandleFromTweetUrl("awjgwekf")
+	if err == nil {
+		t.Errorf("Should have produced an error for invalid URL")
+	}
+}