path: root/main.go

// Telebonk is a reposter from Honk to Telegram.
package main

import (
	"bytes"
	"encoding/json"
	"flag"
	"fmt"
	"io"
	"log"
	"net/http"
	"net/url"
	"regexp"
	"sort"
	"strconv"
	"strings"
	"time"

	"git.aaoth.xyz/la-ninpre/telebonk/config"
)

// A Honk is a post from honk.
type Honk struct {
	ID     int       // unique id of a post
	What   string    // type of an action (post, repost, reply)
	Oondle string    // e-mail style handle of the original author of a post
	Oonker string    // url of the original author of a post
	XID    string    // url of a post, also unique
	RID    string    // url of a post that current post is replying to
	Date   time.Time // datetime of a post
	Precis string    // post summary
	Noise  string    // contents of a post
	Onts   []string  // a slice of tags in a post
	Donks  []*Donk   // a slice of attachments to a post

	MessID    int // telegram message_id
	ReplyToID int // telegram message_id of a message to reply to

	Action HonkAction
}

// A HonkAction tells what to do with the saved honk.
type HonkAction int

const (
	HonkNotSet HonkAction = iota
	HonkIgnore
	HonkSend
	HonkEdit
)

// A Donk stores metadata for media files.
type Donk struct {
	URL   string
	Media string // mime-type of an attachment
	Desc  string
}

// Check performs some checks on a Honk to filter out what's not going to be posted.
//
// It rejects honk if if falls into one of these categories:
//   - it is posted before the telebonk started
//   - it is replying to a honk that's not posted by telebonk (either a remote honk or old honk)
//   - it is of unsupported type (not a regular honk, reply or bonk)
//   - it contains a `#notg` tag.
//   - it is empty
func (h *Honk) Check() error {
	log.Print("check: checking honk #", h.ID) // info

	switch h.What {
	case "honked", "bonked":
		break
	case "honked back":
		hi, ok := honkMap[h.RID]
		if !ok {
			return fmt.Errorf("cannot reply to nonexisting telebonk")
		}
		h.ReplyToID = hi.MessID
	default:
		return fmt.Errorf("unsupported honk type: %s", h.What)
	}

	for _, ont := range h.Onts {
		if strings.ToLower(ont) == "#notg" {
			return fmt.Errorf("skipping #notg honk")
		}
	}

	if h.Noise == emptyNoise && len(h.Donks) == 0 {
		return fmt.Errorf("empty honk")
	}
	return nil
}

// Decide sets the Action of a Honk.
//
// It sets HonkIgnore to those honks that are: 1) old; 2) already sent and aren't edits.
func (h *Honk) Decide() {
	oldhonk, ok := honkMap[h.XID]
	if ok {
		if oldhonk.MessID == 0 || h.Date.Equal(oldhonk.Date) {
			h.Action = HonkIgnore
			h.save(oldhonk.MessID)
			return
		}
		log.Print("decide: honk #", h.XID, " is to be edited")
		h.Action = HonkEdit
		h.MessID = oldhonk.MessID
		return
	}
	if h.Date.Before(now) {
		h.Action = HonkIgnore
		h.save(0)
		return
	}
	log.Print("decide: honk #", h.ID, " is to be sent")
	h.Action = HonkSend
}

// save records a Honk to the honkMap
func (h *Honk) save(messID int) {
	h.MessID = messID
	honkMap[h.XID] = h
}

// forget unchecks a Honk from the honkMap
func (h *Honk) forget() {
	oldhonk, ok := honkMap[h.XID]
	if !ok {
		return
	}
	oldhonk.MessID = 0
	oldhonk.ReplyToID = 0
	honkMap[oldhonk.XID] = oldhonk
}

// A Mess holds data for a message to be sent to Telegram.
type Mess struct {
	Text             string `json:"text"`
	ChatID           string `json:"chat_id"`
	ParseMode        string `json:"parse_mode,omitempty"`
	MessageID        int    `json:"message_id,omitempty"`
	ReplyToMessageID int    `json:"reply_to_message_id,omitempty"`

	Document string `json:"document,omitempty"`
	Photo    string `json:"photo,omitempty"`
	Caption  string `json:"caption,omitempty"`

	kind messKind
}

// A TelegramResponse is a response from Telegram API.
type TelegramResponse struct {
	Ok          bool
	Description string
	Result      struct {
		MessageID int `json:"message_id"`
	}
}

// NewMess creates and populates a new Mess with default values.
func NewMess(parseMode, chatID string) *Mess {
	return &Mess{
		ParseMode: parseMode,
		ChatID:    chatID,
		kind:      messHonk,
	}
}

// NewMessFromHonk creates a slice of Mess objects from existing Honk.
func NewMessFromHonk(honk *Honk) []*Mess {
	var truncateWith = "...\n\nfull honk: " + honk.XID // hardcoded == bad
	// donks should be sent as a separate messages, so need to create all of 'em
	// cap(messes) = 1 for honk + 1 for each donk
	var messes = make([]*Mess, 0, 1+len(honk.Donks))

	messes = append(messes, NewMess("html"))
	for _, donk := range honk.Donks {
		donkMess := NewMess("") // donks don't contain html
		donkMess.Caption = TruncateNoise(donk.Desc, truncateWith, 1024)
		switch {
		case strings.HasPrefix(donk.Media, "image/"):
			donkMess.Photo = donk.URL
			donkMess.kind = messDonkPht
		case donk.Media == "application/pdf", donk.Media == "text/plain":
			donkMess.Document = donk.URL
			donkMess.kind = messDonkDoc
		}
		messes = append(messes, donkMess)
	}
	if honk.Noise == emptyNoise {
		messes = messes[1:] // just donks
	}

	if honk.Action == HonkEdit {
		// TODO: implement editing documents and photos
		messes[0].kind = messEdit
		messes[0].MessageID = honk.MessID
		messes = messes[:1] // don't donk if editing
	}

	var text = CalmNoise(honk.Noise)
	text = TruncateNoise(text, truncateWith, 4096)
	// bonk, then honk back - ok
	// honk back, then bonk - not gonna sync, is it ok?
	// upd: bonks work really confusing
	switch honk.What {
	case "honked":
		break
	case "honked back":
		messes[0].ReplyToMessageID = honk.ReplyToID
	case "bonked":
		oonker := fmt.Sprintf("<a href=\"%s\">%s</a>:", honk.Oonker, honk.Oondle)
		text = oonker + "\n" + text
	}

	// danger zone handling
	if strings.HasPrefix(honk.Precis, "DZ:") {
		text = strings.Join([]string{"<tg-spoiler>", "</tg-spoiler>"}, text)
		text = honk.Precis + "\n" + text
	}
	messes[0].Text = text

	return messes
}

// Send sends a Mess to Telegram.
func (m *Mess) Send() (*TelegramResponse, error) {
	var apiURL = botAPIMethod(tgSendMessage)

	switch m.kind {
	case messHonk:
		// noop
	case messEdit:
		apiURL = botAPIMethod(tgEditMessageText)
	case messDonkPht:
		apiURL = botAPIMethod(tgSendPhoto)
	case messDonkDoc:
		apiURL = botAPIMethod(tgSendDocument)
	}

	junk, err := json.Marshal(m)
	if err != nil {
		return nil, err
	}
	buf := bytes.NewBuffer(junk)
	req, err := http.NewRequest("POST", apiURL, buf)
	if err != nil {
		return nil, err
	}
	req.Header.Add("Content-type", "application/json")
	req.Header.Add("Content-length", strconv.Itoa(buf.Len()))

	resp, err := client.Do(req)
	if err != nil {
		return nil, err
	}
	defer resp.Body.Close()

	var res TelegramResponse
	json.NewDecoder(resp.Body).Decode(&res)
	if !res.Ok {
		return nil, fmt.Errorf("mess send: %s", res.Description)
	}

	return &res, nil
}

type messKind int

const (
	messHonk messKind = iota
	messEdit
	messDonkPht
	messDonkDoc
)

// XXX: figure out how to retreive these args without passing config object thousand times
func botAPIMethod(url, token, method string) string {
	return fmt.Sprintf("%s/bot%s/%s", url, token, method)
}

func checkTgAPI() error {
	var apiURL = botAPIMethod(tgGetMe)
	resp, err := client.Get(apiURL)
	if err != nil {
		return err
	}
	if resp.StatusCode != 200 {
		status, _ := io.ReadAll(resp.Body)
		return fmt.Errorf("status: %d: %s", resp.StatusCode, status)
	}
	return nil
}

// Telegram Bot API methods
const (
	tgGetMe           = "getMe"
	tgSendMessage     = "sendMessage"
	tgEditMessageText = "editMessageText"
	tgSendPhoto       = "sendPhoto"
	tgSendDocument    = "sendDocument"
)

// getHonks receives and unmarshals some honks from a Honk instance.
func getHonks(from, page, token string, after int) ([]*Honk, error) {
	query := url.Values{}
	query.Set("action", "gethonks")
	query.Set("page", page)
	query.Set("after", strconv.Itoa(after))
	apiurl := from + "/api?" + query.Encode()

	req, err := http.NewRequest("GET", apiurl, nil)
	if err != nil {
		return nil, err
	}
	req.Header.Add("Authorization", "Bearer "+token)

	resp, err := client.Do(req)
	if err != nil {
		return nil, err
	}
	defer resp.Body.Close()

	// honk outputs junk like `{ "honks": [ ... ] }`, need to get into the list
	var honkJunk map[string][]*Honk
	err = json.NewDecoder(resp.Body).Decode(&honkJunk)
	if err != nil {
		// XXX: honk tokens last for a week or so. when one expires, shouldn't this say something meaningful instead of `unexpected v in blah-blah'?
		log.Print("gethonks: ", resp.Status)
		return nil, err
	}

	honks := honkJunk["honks"]
	// honk.ID monotonically increases, so it can be used to sort honks
	sort.Slice(honks, func(i, j int) bool { return honks[i].ID < honks[j].ID })

	return honks, nil
}

var (
	rePTags   = regexp.MustCompile(`<\/?p>`)
	reHlTags  = regexp.MustCompile(`<\/?span( class="[a-z]{2}")?>`)
	reBrTags  = regexp.MustCompile(`<br>`)
	reImgTags = regexp.MustCompile(`<img .*src="(.*)">`)
	reUlTags  = regexp.MustCompile(`<\/?ul>`)
	reTbTags  = regexp.MustCompile(`<table>.*</table>`)

	reLiTags = regexp.MustCompile(`<li>([^<>\/]*)<\/li>`)
	reHnTags = regexp.MustCompile(`<h[1-6]>(.*)</h[1-6]>`)
	reHrTags = regexp.MustCompile(`<hr>.*<\/hr>`)
	reBqTags = regexp.MustCompile(`<blockquote>(.*)<\/blockquote>`)
)

const emptyNoise = "<p></p>\n"

// CalmNoise erases and rewrites html tags that are not supported by Telegram.
func CalmNoise(s string) string {
	// delete these
	s = rePTags.ReplaceAllString(s, "")
	s = reHlTags.ReplaceAllString(s, "")
	s = reBrTags.ReplaceAllString(s, "")
	s = reImgTags.ReplaceAllString(s, "")
	s = reUlTags.ReplaceAllString(s, "")
	s = reTbTags.ReplaceAllString(s, "")

	// these can be repurposed
	s = reHrTags.ReplaceAllString(s, "---\n")
	s = reHnTags.ReplaceAllString(s, "<b>$1</b>\n\n")
	s = reBqTags.ReplaceAllString(s, "| <i>$1</i>")
	s = reLiTags.ReplaceAllString(s, "* $1\n")

	return strings.TrimSpace(s)
}

// TruncateNoise truncates a string up to `length - len(with)` characters long and adds `with` to the end.
func TruncateNoise(s, with string, length int) string {
	// telegram can handle posts no longer than 4096 (or 1024) characters _after_ the parsing of entities.
	// we could be clever and calculate the true length of text, but let's keep it simple and stupid.
	if len(s) <= length {
		return s
	}

	var b strings.Builder
	b.Grow(length)
	var end = length - 1 - len(with)
	for i, r := range s {
		if i >= end {
			break
		}
		b.WriteRune(r)
	}
	b.WriteString(with)
	return b.String()
}

// XXX: global and mutable
var conf = &config.Config{}

func init() {
	flag.StringVar(&conf.TgBotToken, "bot_token", "", "Telegram bot token")
	flag.StringVar(&conf.TgChatID, "chat_id", "", "Telegram chat_id")
	flag.StringVar(&conf.TgApiURL, "tgapi_url", "https://api.telegram.org", "Telegram API URL")
	flag.StringVar(&conf.HonkAuthToken, "honk_token", "", "Honk auth token")
	flag.StringVar(&conf.HonkPage, "honk_page", "myhonks", "Page to get honks from. Should be one of [atme, longago, home, myhonks]")
	flag.StringVar(&conf.HonkURL, "honk_url", "", "URL of a Honk instance")

	flag.Parse()

	if err := conf.Check(); err != nil {
		log.Fatal("conf:", err) // fail
	}
	conf.TgApiURL = strings.TrimRight(conf.TgApiURL, "/")
	if err := checkTgAPI(); err != nil {
		log.Fatal("tgAPI:", err) // fail
	}
}

var client = http.DefaultClient
var honkMap = make(map[string]*Honk) // XXX: not safe for use by multiple goroutines!
var now = time.Now()

func main() {
	var retry = 5

	log.Print("starting telebonk") // info
	for {
		honks, err := getHonks(conf.HonkURL, conf.HonkPage, conf.HonkAuthToken, 0)
		if err != nil {
			log.Print("gethonks:", err) // error
			retry--
			if retry == 0 {
				log.Fatal("gethonks: giving up") // fail
			}
			time.Sleep(5 * time.Second)
			continue
		}
	HonkLoop:
		for _, honk := range honks {
			honk.Decide()
			switch honk.Action {
			case HonkIgnore:
				continue
			case HonkSend, HonkEdit:
				if err := honk.Check(); err != nil {
					log.Print("honk check:", err) // error
					continue
				}
			}
			messes := NewMessFromHonk(honk)
			// messes[0] is a honk or a donk to be sent
			resp, err := messes[0].Send()
			if err != nil {
				log.Print("mess send", err) // error
				honk.forget()               // retry
				continue
			}
			// remember only the first mess' response
			honk.save(resp.Result.MessageID)
			for _, mess := range messes[1:] {
				if _, err := mess.Send(); err != nil {
					log.Print("mess send", err) // error
					continue HonkLoop
				}
			}
		}
		time.Sleep(30 * time.Second)
	}
}