RSSCollector/send.go

package main

import (
	"encoding/json"
	"fmt"
	"log"
	"reflect"
	"strings"
	"time"

	"github.com/boltdb/bolt"
	tgbotapi "github.com/go-telegram-bot-api/telegram-bot-api/v5"
	"golang.org/x/net/html"
)

func in_array(val interface{}, array interface{}) (exists bool) {
	exists = false

	switch reflect.TypeOf(array).Kind() {
	case reflect.Slice:
		s := reflect.ValueOf(array)

		for i := 0; i < s.Len(); i++ {
			if reflect.DeepEqual(val, s.Index(i).Interface()) == true {
				exists = true
				return
			}
		}
	}

	return
}

func NormalizeHTMLforTelegram(s string) (out string) {

	tags := []string{"br", "img", "b", "strong", "i", "em", "code", "s", "strike", "del", "u", "pre"}

	domDocTest := html.NewTokenizer(strings.NewReader(s))
	previousStartTokenTest := domDocTest.Token()
	flagendtag := false
	for {
		tt := domDocTest.Next()
		if len(out) > 2500 {
			if e := in_array(previousStartTokenTest.Data, tags) && previousStartTokenTest.Data != "img" && previousStartTokenTest.Data != "br" && !flagendtag; e {
				out += fmt.Sprintf("</%s> ...", previousStartTokenTest.Data)
			} else {
				out += " ..."
			}
			return
		}
		switch {
		case tt == html.ErrorToken:
			return
		case tt == html.StartTagToken:
			previousStartTokenTest = domDocTest.Token()
			if e := in_array(previousStartTokenTest.Data, tags); e {
				switch {
				case previousStartTokenTest.Data == "br":
					out += "\n"
				case previousStartTokenTest.Data == "img" && previousStartTokenTest.Attr[0].Key == "src":
					out += fmt.Sprintf("%s ", previousStartTokenTest.Attr[0].Val)
				// case previousStartTokenTest.Data == "a" && previousStartTokenTest.Attr[0].Key == "href":
				// 	out += fmt.Sprintf(" %s ", previousStartTokenTest.Attr[0].Val)
				default:
					out += fmt.Sprintf(" <%s>", previousStartTokenTest.Data)
				}
				flagendtag = false
			}
		case tt == html.EndTagToken:
			t := domDocTest.Token()
			if e := in_array(t.Data, tags); e {
				// switch {
				// case t.Data == "a":
				// 	out += " "
				// default:
				out += fmt.Sprintf("</%s> ", t.Data)
				// }
				flagendtag = true
			}
		case tt == html.SelfClosingTagToken:
			t := domDocTest.Token()
			if e := in_array(t.Data, tags); e {
				if t.Data == "br" {
					out += "\n"
				}
			}
		case tt == html.TextToken:
			if previousStartTokenTest.Data == "script" {
				continue
			}
			TxtContent := strings.TrimSpace(html.UnescapeString(string(domDocTest.Text())))
			if len(TxtContent) > 0 {
				out += html.EscapeString(TxtContent)
			}
		}
	}
}

func SendAndWriteToDB(send SendItems, dbpath string, rssname string, token string, chatid int64, debug bool) error {

	db, err := bolt.Open(dbpath, 0600, nil)
	if err != nil {
		log.Fatal(err)
	}
	defer db.Close()

	for i := len(send.ItemList) - 1; i >= 0; i-- {
		v := send.ItemList[i]
		log.Printf("Send to telegram post: %s", v.Title)
		bot, err := tgbotapi.NewBotAPI(token)
		if err != nil {
			log.Panic(err)
		}
		bot.Debug = debug

		s := "<i>" + rssname + "</i>\n\n" + "<b>" + string(v.Title) + "</b>\n\n" + NormalizeHTMLforTelegram(html.UnescapeString(string(v.Description))) +
			"\n\n" + v.Link
		msg := tgbotapi.NewMessage(chatid, s)
		msg.ParseMode = "Html"
		_, err = bot.Send(msg)
		if err != nil {
			log.Panic(err)

		}
		duration := time.Duration(10) * time.Second
		time.Sleep(duration)

		db.Update(func(tx *bolt.Tx) error {
			b, err := tx.CreateBucketIfNotExists([]byte(rssname))
			if err != nil {
				return err
			}
			encoded, err := json.Marshal(v)
			if err != nil {
				return err
			}
			err = b.Put([]byte(v.Link), encoded)
			if err != nil {
				return err
			}
			return nil
		})
	}
	return nil
}