Implement function to prune string keeping HTML closing tags

Fixes umputun#1587

backend/app/notify/telegram.go

@@ -3,13 +3,17 @@ package notify
 import (
 	"context"
 	"fmt"
+	"strings"
 	"time"
 
 	log "github.com/go-pkgz/lgr"
 	ntf "github.com/go-pkgz/notify"
 	"github.com/hashicorp/go-multierror"
+	"golang.org/x/net/html"
 )
 
+const comment_text_length_limit = 100
+
 // TelegramParams contain settings for telegram notifications
 type TelegramParams struct {
 	AdminChannelID       string        // unique identifier for the target chat or username of the target channel (in the format @channelusername)
@@ -75,6 +79,113 @@ func (t *Telegram) Send(ctx context.Context, req Request) error {
 	return result.ErrorOrNil()
 }
 
+type stringArr struct {
+	data []string
+	len  int
+}
+
+// Push adds element to the end
+func (s *stringArr) Push(v string) {
+	s.data = append(s.data, v)
+	s.len += len(v)
+}
+
+// Pop removes element from end and returns it
+func (s *stringArr) Pop() string {
+	l := len(s.data)
+	newData, v := s.data[:l-1], s.data[l-1]
+	s.data = newData
+	s.len -= len(v)
+	return v
+}
+
+// Unshift adds element to the start
+func (s *stringArr) Unshift(v string) {
+	s.data = append([]string{v}, s.data...)
+	s.len += len(v)
+}
+
+// Shift removes element from start and returns it
+func (s *stringArr) Shift() string {
+	v, newData := s.data[0], s.data[1:]
+	s.data = newData
+	s.len -= len(v)
+	return v
+}
+
+// String returns all strings concatenated
+func (s stringArr) String() string {
+	return strings.Join(s.data, "")
+}
+
+// Len returns total length of all strings concatenated
+func (s stringArr) Len() int {
+	return s.len
+}
+
+// pruneHTML prunes string keeping HTML closing tags
+func pruneHTML(htmlText string, maxLength int) string {
+	result := stringArr{}
+	endTokens := stringArr{}
+
+	suffix := "..."
+	suffixLen := len(suffix)
+
+	tokenizer := html.NewTokenizer(strings.NewReader(htmlText))
+	for {
+		if tokenizer.Next() == html.ErrorToken {
+			return result.String()
+		}
+		token := tokenizer.Token()
+
+		switch token.Type {
+		case html.CommentToken, html.DoctypeToken:
+			// skip tokens without content
+			continue
+
+		case html.StartTagToken:
+			// <token></token>
+			// len(token) * 2 + len("<></>")
+			totalLenToAppend := len(token.Data)*2 + 5
+			if result.Len()+totalLenToAppend+endTokens.Len()+suffixLen > maxLength {
+				return result.String() + suffix + endTokens.String()
+			}
+			endTokens.Unshift(fmt.Sprintf("</%s>", token.Data))
+
+		case html.EndTagToken:
+			endTokens.Shift()
+
+		case html.TextToken, html.SelfClosingTagToken:
+			if result.Len()+len(token.String())+endTokens.Len()+suffixLen > maxLength {
+				text := pruneStringToWord(token.String(), maxLength-result.Len()-endTokens.Len()-suffixLen)
+				return result.String() + text + suffix + endTokens.String()
+			}
+		}
+
+		result.Push((token.String()))
+	}
+}
+
+// pruneStringToWord prunes string to specified length respecting words
+func pruneStringToWord(text string, maxLength int) string {
+	if maxLength <= 0 {
+		return ""
+	}
+
+	result := ""
+
+	arr := strings.Split(text, " ")
+	for _, s := range arr {
+		if len(result)+len(s) >= maxLength {
+			return strings.TrimRight(result, " ")
+		}
+		// keep last space, it's ok
+		result += s + " "
+	}
+
+	return text
+}
+
 // buildMessage generates message for generic notification about new comment
 func (t *Telegram) buildMessage(req Request) string {
 	commentURLPrefix := req.Comment.Locator.URL + uiNav
@@ -85,7 +196,7 @@ func (t *Telegram) buildMessage(req Request) string {
 		msg += fmt.Sprintf(" -> <a href=%q>%s</a>", commentURLPrefix+req.parent.ID, ntf.EscapeTelegramText(req.parent.User.Name))
 	}
 
-	msg += fmt.Sprintf("\n\n%s", ntf.TelegramSupportedHTML(req.Comment.Text))
+	msg += fmt.Sprintf("\n\n%s", pruneHTML(ntf.TelegramSupportedHTML(req.Comment.Text), comment_text_length_limit))
 
 	if req.Comment.ParentID != "" {
 		msg += fmt.Sprintf("\n\n\"<i>%s</i>\"", ntf.TelegramSupportedHTML(req.parent.Text))

backend/app/notify/telegram_test.go

@@ -53,6 +53,15 @@ some text
 
 <b>Hello</b><i><b>World</b></i>`,
 		res)
+
+	// prune string keeping HTML closing tags
+	c = store.Comment{
+		Text: "<b>Lorem ipsum <i>dolor sit amet</i>, consectetur adipiscing <code>elit, sed do eiusmod tempor incididunt</code> ut labore et dolore magna aliqua.</b>",
+	}
+	res = tb.buildMessage(Request{Comment: c})
+	assert.Equal(t, `<a href="#remark42__comment-"></a>
+
+<b>Lorem ipsum <i>dolor sit amet</i>, consectetur adipiscing <code>elit, sed do...</code></b>`, res)
 }
 
 func TestTelegram_SendVerification(t *testing.T) {