tg_exporter/internal/client_channel.go

package internal

import (
	"context"
	"mime"
	"os"

	"exporter/pkg/errorx"

	"github.com/dustin/go-humanize"
	"github.com/gotd/td/telegram/downloader"
	"github.com/gotd/td/tg"
	"github.com/pkg/errors"
	"go.uber.org/zap"
)

func (t *TClient) Channel(ctx context.Context, channel *tg.Channel, cfg *DBChannel, modeHistory bool) error {
	inputPeer := &tg.InputPeerChannel{ChannelID: channel.ID, AccessHash: channel.AccessHash}

	request := &tg.MessagesGetHistoryRequest{
		Peer:  inputPeer,
		Limit: 10,
	}

	if modeHistory { // 提供此ID供遍历历史消息
		request.OffsetID = cfg.Offset
	} else {
		request.MinID = cfg.MinID // 提供此ID供新增加的消息
	}

	// request.OffsetID = 1339
	// request.Limit = 1

	messages, err := t.Client.API().MessagesGetHistory(ctx, request)
	if err != nil {
		return errors.Wrap(err, "messages.getHistory")
	}
	if len(messages.(*tg.MessagesChannelMessages).GetMessages()) == 0 {
		logger.Info("no new message", zap.Int64("channel", channel.ID))
		return errorx.ErrNoNewMessage
	}

	for _, item := range messages.(*tg.MessagesChannelMessages).GetMessages() {
		switch item.(type) {
		case *tg.MessageEmpty:
			continue
		case *tg.MessageService:
			continue
		}

		msg, ok := item.(*tg.Message)
		if !ok {
			logger.Error("convert msg to *tg.Message failed", zap.Int64("channel", channel.ID))
			continue
		}

		channelMessage := NewChannelMessage(msg.ID, msg.GetDate())

		channelMessage.WithMessage(msg.GetMessage())
		if grpID, ok := msg.GetGroupedID(); ok {
			channelMessage.WithGroupID(grpID)
		}

		mediaClass, ok := msg.GetMedia()
		if !ok {
			continue
		}
		switch mediaClass.(type) {
		case *tg.MessageMediaDocument:
			if docClass, ok := mediaClass.(*tg.MessageMediaDocument).GetDocument(); ok {
				// logger.Warn("document",
				// 	zap.Int64("channel", channel.ID),
				// 	zap.Int("msg_id", msg.ID),
				// 	zap.String("file_name", docClass.String()),
				// )

				doc := docClass.(*tg.Document)

				if doc.GetSize() > int64(t.Config.GetMaxSize()) {
					logger.Warn(
						"document size too large",
						zap.Int64("channel", channel.ID),
						zap.Int64("size", doc.GetSize()),
						zap.String("SizeHuman", humanize.Bytes(uint64(doc.GetSize()))),
					)
					continue
				}

				data, err := t.saveDocument(ctx, cfg, doc)
				if err != nil {
					logger.Error("save document failed", zap.Error(err), zap.Int64("channel", channel.ID))
					return err
				}
				channelMessage.WithDocument(doc.GetID(), data)
			}
		case *tg.MessageMediaWebPage:
			if page, ok := mediaClass.(*tg.MessageMediaWebPage).GetWebpage().(*tg.WebPage); ok {
				channelMessage.WithWebPage(page.GetID(), page.Title, page.URL)
			} else {
				logger.Warn("web_page",
					zap.Int64("channel", channel.ID),
					zap.Int64("msg_id", page.GetID()),
					zap.String("url", mediaClass.(*tg.MessageMediaWebPage).GetWebpage().String()),
				)
			}
		case *tg.MessageMediaPhoto:
			if photoClass, ok := mediaClass.(*tg.MessageMediaPhoto).GetPhoto(); ok {
				photo := photoClass.(*tg.Photo)
				if err := t.savePhoto(ctx, cfg, photo); err != nil {
					logger.Error("save photo failed", zap.Error(err), zap.Int64("channel", channel.ID))
					return err
				}
				channelMessage.WithPhoto(photo.GetID(), "jpg")
			}
		}

		logger.Info("save message", zap.Int("id", channelMessage.ID))
		if err := cfg.SaveMessage(ctx, channelMessage); err != nil {
			logger.Error("save message failed", zap.Error(err), zap.Int64("channel", channel.ID))
			return err
		}

		logger.Info("update config", zap.Int("offset", cfg.Offset), zap.Int64("channel", channel.ID))
		if err := cfg.Update(ctx, item.GetID()); err != nil {
			logger.Error("update config failed", zap.Error(err), zap.Int64("channel", channel.ID))
			return err
		}

	}

	return nil
}

// savePhoto
func (t *TClient) savePhoto(ctx context.Context, cfg *DBChannel, photo *tg.Photo) error {
	thumbSize := ""
	if len(photo.Sizes) > 1 {
		thumbSize = photo.Sizes[len(photo.Sizes)-1].GetType()
	}

	location := &tg.InputPhotoFileLocation{
		ID:            photo.GetID(),
		AccessHash:    photo.GetAccessHash(),
		FileReference: photo.GetFileReference(),
		ThumbSize:     thumbSize,
	}

	saveTo := cfg.Asset(photo.GetID(), "jpg")
	downloader := downloader.NewDownloader()
	_, err := downloader.Download(t.Client.API(), location).ToPath(ctx, saveTo)
	if err != nil {
		os.Remove(saveTo)
		logger.Error("download failed", zap.Error(err))
		return err
	}

	logger.Info("download photo success", zap.String("location", saveTo))

	return nil
}

// saveDocument
func (t *TClient) saveDocument(ctx context.Context, cfg *DBChannel, doc *tg.Document) (ChannelMessageDocument, error) {
	location := &tg.InputDocumentFileLocation{
		ID:            doc.GetID(),
		AccessHash:    doc.GetAccessHash(),
		FileReference: doc.GetFileReference(),
	}
	exts, err := mime.ExtensionsByType(doc.GetMimeType())
	if err != nil {
		logger.Error("get extension failed", zap.Error(err), zap.String("mime_type", doc.GetMimeType()))
		return ChannelMessageDocument{}, err
	}

	if len(exts) == 0 {
		logger.Warn("no extension found", zap.String("mime_type", doc.GetMimeType()))
		switch doc.GetMimeType() {
		case "application/rar":
			exts = []string{".rar"}
		}
	}
	ext := exts[len(exts)-1]

	saveTo := cfg.Asset(doc.GetID(), ext)
	downloader := downloader.NewDownloader()
	_, err = downloader.Download(t.Client.API(), location).ToPath(ctx, saveTo)
	if err != nil {
		os.Remove(saveTo)
		logger.Error("download failed", zap.Error(err))
		return ChannelMessageDocument{}, err
	}
	docAttr := doc.GetAttributes()

	data := ChannelMessageDocument{
		Ext:      ext,
		MimeType: doc.GetMimeType(),
		Size:     doc.GetSize(),
	}
	if len(docAttr) > 0 {
		for _, attr := range docAttr {
			switch attr.(type) {
			case *tg.DocumentAttributeFilename:
				data.Filename = attr.(*tg.DocumentAttributeFilename).GetFileName()
			case *tg.DocumentAttributeVideo:
				m := attr.(*tg.DocumentAttributeVideo)
				data.Video = &ChannelMessageDocumentVideo{
					Duration: m.GetDuration(),
					Width:    m.GetW(),
					Height:   m.GetH(),
				}

			}
		}
	}

	logger.Info("download document success", zap.String("location", saveTo), zap.Any("document", data))
	return data, nil
}