rss2world/app/app.go

package app

import (
	"bytes"
	"context"
	"log/slog"
	"os"
	"regexp"
	"strings"
	"text/template"
	"time"

	tgbotapi "github.com/go-telegram-bot-api/telegram-bot-api/v5"
	"github.com/mmcdole/gofeed"
	"gitrepo.ru/neonxp/rss2world/templates"
)

type App struct {
	config    *Config
	telegram  *tgbotapi.BotAPI
	templates *template.Template
}

func New(cfg *Config) (*App, error) {
	bot, err := tgbotapi.NewBotAPI(cfg.Telegram.BotToken)
	if err != nil {
		return nil, err
	}

	tpl, err := template.ParseFS(templates.Templates, "*.gotmpl")
	if err != nil {
		return nil, err
	}

	return &App{
		config:    cfg,
		telegram:  bot,
		templates: tpl,
	}, nil
}

func (a *App) Run(ctx context.Context) error {
	ticker := time.NewTicker(a.config.RSS.CheckInterval)
	for {
		select {
		case <-ctx.Done():
			return nil
		case <-ticker.C:
			if err := a.iteration(); err != nil {
				slog.ErrorContext(ctx, "failed iteration", slog.Any("error", err))
			}
		}
	}
}

func (a *App) iteration() error {
	seq, err := a.readSeqFile()
	if os.IsNotExist(err) {
		seq = ""
		err = nil
	}
	if err != nil {
		return err
	}

	items, err := a.findNewItems(seq)
	if err != nil || len(items) == 0 {
		return err
	}

	for i := len(items) - 1; i >= 0; i-- {
		if err := a.processItem(items[i]); err != nil {
			return err
		}
		if err := a.writeSeqFile(items[i].GUID); err != nil {
			return err
		}
	}

	return nil
}

func (a *App) processItem(item *gofeed.Item) error {
	buf := bytes.NewBufferString("")
	item.Content = trimHtml(item.Description)
	if err := a.templates.ExecuteTemplate(buf, "telegram", item); err != nil {
		return err
	}
	str := buf.String()
	for _, group := range a.config.Telegram.TargetGroups {
		msg := tgbotapi.NewMessage(group, str)
		msg.ParseMode = tgbotapi.ModeHTML
		if _, err := a.telegram.Send(msg); err != nil {
			return err
		}
	}

	return nil
}

func (a *App) readSeqFile() (string, error) {
	seqVal, err := os.ReadFile(a.config.RSS.SeqFile)
	if err != nil {
		return "", err
	}

	return string(seqVal), nil
}

func (a *App) writeSeqFile(seqVal string) error {
	return os.WriteFile(a.config.RSS.SeqFile, []byte(seqVal), 0o644)
}

func (a *App) findNewItems(from string) ([]*gofeed.Item, error) {
	fp := gofeed.NewParser()
	feed, err := fp.ParseURL(a.config.RSS.URL)
	if err != nil {
		return nil, err
	}
	out := make([]*gofeed.Item, 0, len(feed.Items))
	for _, item := range feed.Items {
		if item.GUID == from {
			break
		}
		out = append(out, item)
	}

	return out, nil
}

func trimHtml(src string) string {
	// Convert all HTML tags to lowercase
	re, _ := regexp.Compile("\\<[\\S\\s]+?\\>")
	src = re.ReplaceAllStringFunc(src, strings.ToLower)
	// Remove STYLE
	re, _ = regexp.Compile("\\<style[\\S\\s]+?\\</style\\>")
	src = re.ReplaceAllString(src, "")
	// Remove SCRIPT
	re, _ = regexp.Compile("\\<script[\\S\\s]+?\\</script\\>")
	src = re.ReplaceAllString(src, "")
	// Remove all HTML code in angle brackets and replace them with newline characters
	re, _ = regexp.Compile("\\<[\\S\\s]+?\\>")
	src = re.ReplaceAllString(src, "\n")
	// Remove consecutive newlines
	re, _ = regexp.Compile("\\s{2,}")
	src = re.ReplaceAllString(src, "\n")
	return strings.TrimSpace(src)
}