Merge pull request #84 from velour/slack_parsing

slack: import velour/chat parsing

slack/fix_text.go

@@ -0,0 +1,96 @@
+package slack
+
+import (
+	"unicode/utf8"
+)
+
+// fixText strips all of the Slack-specific annotations from message text,
+// replacing it with the equivalent display form.
+// Currently it:
+// • Replaces user mentions like <@U124356> with @ followed by the user's nick.
+// This uses the lookupUser function, which must map U1243456 to the nick.
+// • Replaces user mentions like <U123456|nick> with the user's nick.
+// • Strips < and > surrounding links.
+//
+// This was directly bogarted from velour/chat with emoji conversion removed.
+func fixText(findUser func(id string) (string, bool), text string) string {
+	var output []rune
+	for len(text) > 0 {
+		r, i := utf8.DecodeRuneInString(text)
+		text = text[i:]
+		switch {
+		case r == '<':
+			var tag []rune
+			for {
+				r, i := utf8.DecodeRuneInString(text)
+				text = text[i:]
+				switch {
+				case r == '>':
+					if t, ok := fixTag(findUser, tag); ok {
+						output = append(output, t...)
+						break
+					}
+					fallthrough
+				case len(text) == 0:
+					output = append(output, '<')
+					output = append(output, tag...)
+					output = append(output, r)
+				default:
+					tag = append(tag, r)
+					continue
+				}
+				break
+			}
+		default:
+			output = append(output, r)
+		}
+	}
+	return string(output)
+}
+
+func fixTag(findUser func(string) (string, bool), tag []rune) ([]rune, bool) {
+	switch {
+	case hasPrefix(tag, "@U"):
+		if i := indexRune(tag, '|'); i >= 0 {
+			return tag[i+1:], true
+		}
+		if findUser != nil {
+			if u, ok := findUser(string(tag[1:])); ok {
+				return []rune(u), true
+			}
+		}
+		return tag, true
+
+	case hasPrefix(tag, "#C"):
+		if i := indexRune(tag, '|'); i >= 0 {
+			return append([]rune{'#'}, tag[i+1:]...), true
+		}
+
+	case hasPrefix(tag, "http"):
+		if i := indexRune(tag, '|'); i >= 0 {
+			tag = tag[:i]
+		}
+		return tag, true
+	}
+
+	return nil, false
+}
+
+func hasPrefix(text []rune, prefix string) bool {
+	for _, r := range prefix {
+		if len(text) == 0 || text[0] != r {
+			return false
+		}
+		text = text[1:]
+	}
+	return true
+}
+
+func indexRune(text []rune, find rune) int {
+	for i, r := range text {
+		if r == find {
+			return i
+		}
+	}
+	return -1
+}

slack/slack.go

@@ -306,14 +306,13 @@ var urlDetector = regexp.MustCompile(`<(.+)://([^|^>]+).*>`)
 func (s *Slack) buildMessage(m slackMessage) msg.Message {
 	text := html.UnescapeString(m.Text)
 
-	// remove <> from URLs, URLs may also be <url|description>
-	text = urlDetector.ReplaceAllString(text, "${1}://${2}")
+	text = fixText(s.getUser, text)
 
 	isCmd, text := bot.IsCmd(s.config, text)
 
 	isAction := m.SubType == "me_message"
 
-	u := s.getUser(m.User)
+	u, _ := s.getUser(m.User)
 	if m.Username != "" {
 		u = m.Username
 	}
@@ -460,9 +459,9 @@ func (s *Slack) connect() {
 }
 
 // Get username for Slack user ID
-func (s *Slack) getUser(id string) string {
+func (s *Slack) getUser(id string) (string, bool) {
 	if name, ok := s.users[id]; ok {
-		return name
+		return name, true
 	}
 
 	log.Printf("User %s not already found, requesting info", id)
@@ -472,17 +471,17 @@ func (s *Slack) getUser(id string) string {
 	if err != nil || resp.StatusCode != 200 {
 		log.Printf("Error posting user info request: %d %s",
 			resp.StatusCode, err)
-		return "UNKNOWN"
+		return "UNKNOWN", false
 	}
 	defer resp.Body.Close()
 	var userInfo slackUserInfoResp
 	err = json.NewDecoder(resp.Body).Decode(&userInfo)
 	if err != nil {
 		log.Println("Error decoding response: ", err)
-		return "UNKNOWN"
+		return "UNKNOWN", false
 	}
 	s.users[id] = userInfo.User.Name
-	return s.users[id]
+	return s.users[id], true
 }
 
 // Who gets usernames out of a channel
@@ -513,7 +512,8 @@ func (s *Slack) Who(id string) []string {
 
 	handles := []string{}
 	for _, member := range chanInfo.Channel.Members {
-		handles = append(handles, s.getUser(member))
+		u, _ := s.getUser(member)
+		handles = append(handles, u)
 	}
 	log.Printf("Returning %d handles", len(handles))
 	return handles