linstrom/activitypub/importNote.go

package activitypub

import (
	"database/sql"
	"encoding/json"
	"errors"
	"fmt"
	"io"
	"strings"
	"time"

	"git.mstar.dev/mstar/goutils/other"
	"github.com/rs/zerolog/log"
	"gorm.io/gorm"

	"git.mstar.dev/mstar/linstrom/config"
	"git.mstar.dev/mstar/linstrom/storage-new/dbgen"
	"git.mstar.dev/mstar/linstrom/storage-new/models"
	webshared "git.mstar.dev/mstar/linstrom/web/shared"
)

const DefaultMaxImportRecursion = 100

var ErrMaxImportRecursionReached = errors.New("maximum import recursion reached")

func ImportRemoteNote(noteId string, requester *models.User) (string, error) {
	return importRemoteNoteRecursive(noteId, requester, 0)
}

func importRemoteNoteRecursive(
	noteId string,
	requester *models.User,
	recursionDepth uint,
) (string, error) {
	type NoteTag struct {
		Type string `json:"type"`
		Href string `json:"href"`
		Name string `json:"name"`
	}
	type Note struct {
		Type         string    `json:"type"`
		Id           string    `json:"id"`
		Summary      *string   `json:"summary"`
		Content      string    `json:"content"`
		MkContent    *string   `json:"_misskey_content"`
		Published    time.Time `json:"published"`
		To           []string  `json:"to"`
		Cc           []string  `json:"cc"`
		InReplyTo    *string   `json:"inReplyTo"`
		Sensitive    bool      `json:"sensitive"`
		AttributedTo string    `json:"attributedTo"`
		Tags         []NoteTag `json:"tag"`
	}
	// TODO: Decide whether the max recursion depth can be configured via config file
	if recursionDepth > DefaultMaxImportRecursion {
		return "", ErrMaxImportRecursionReached
	}
	// No need to import local notes. Ids of local notes either have the full public url as prefix
	// or no http prefix at all (internal id only)
	if strings.HasPrefix(noteId, config.GlobalConfig.General.GetFullPublicUrl()) ||
		!strings.HasPrefix(noteId, "http") {
		switch _, err := dbgen.Note.Where(dbgen.Note.ID.Eq(noteId)).First(); err {
		case nil:
			return noteId, nil
		case gorm.ErrRecordNotFound:
			return "", other.Error("activitypub", "local note doesn't exist", err)
		default:
			return "", other.Error(
				"activitypub",
				"failed to check for existence of local note",
				err,
			)
		}
	}
	res, _, err := webshared.RequestSigned("GET", noteId, nil, requester)
	if err != nil {
		return "", other.Error("activitypub", "failed to request object", err)
	}
	if res.StatusCode != 200 {
		return "", fmt.Errorf("activitypub: invalid status code: %v", res.StatusCode)
	}
	body, _ := io.ReadAll(res.Body)
	_ = res.Body.Close()
	data := Note{}
	err = json.Unmarshal(body, &data)
	if err != nil {
		return "", other.Error("activitypub", "json unmarshalling failed", err)
	}
	if data.Type != "Note" {
		return "", fmt.Errorf("activitypub: invalid object type: %q", data.Type)
	}
	// (Re-)Import account this note belongs to
	_, err = ImportRemoteAccountByAPUrl(data.AttributedTo)
	if err != nil {
		return "", other.Error("activitypub", "failed to import note author", err)
	}
	// If the note already exists in storage, update that
	dbNote, err := dbgen.Note.Where(dbgen.Note.ID.Eq(data.Id)).First()
	switch err {
	case nil:
	case gorm.ErrRecordNotFound:
		// Otherwise create a new one
		dbNote = &models.Note{
			ID:        data.Id,
			CreatorId: data.AttributedTo,
		}
	default:
		return "", other.Error("activitypub", "failed to check db for note", err)
	}
	dbNote.CreatedAt = data.Published
	if data.Summary != nil {
		dbNote.ContentWarning = sql.NullString{Valid: true, String: *data.Summary}
	} else {
		dbNote.ContentWarning = sql.NullString{Valid: false}
	}
	// Prefer raw misskey content if set since it *should*
	// have the most accurate content data
	if data.MkContent != nil {
		dbNote.RawContent = *data.MkContent
	} else {
		dbNote.RawContent = data.Content
	}
	if data.InReplyTo != nil {
		dbNote.RepliesTo = sql.NullString{Valid: true, String: *data.InReplyTo}
		defer func() {
			_, err := importRemoteNoteRecursive(*data.InReplyTo, requester, recursionDepth+1)
			switch err {
			case ErrMaxImportRecursionReached:
				log.Warn().
					Str("reply-id", data.Id).
					Str("in-reply-to-id", *data.InReplyTo).
					Msg("Hit recursion limit while importing note")
			case nil:
			default:
				log.Error().
					Err(err).
					Str("id", *data.InReplyTo).
					Msg("Failed to import note that was replied to")
			}
			// Don't return import errors of recursive imports since they don't affect this import
		}()
	} else {
		dbNote.RepliesTo = sql.NullString{Valid: false}
	}
	// TODO: Include more data like attachments
	err = dbgen.Note.Save(dbNote)
	if err != nil {
		return "", err
	}
	// Handle tags after the initial note since stored in separate tables and pings require more imports
	hashtags := []*models.NoteTag{}
	pings := []*models.NoteToPing{}
	for _, tag := range data.Tags {
		switch tag.Type {
		case "Mention":
			_, err := ImportRemoteAccountByAPUrl(tag.Href)
			if err != nil {
				return "", err
			}
			pings = append(pings, &models.NoteToPing{
				NoteId:       dbNote.ID,
				PingTargetId: tag.Href,
			})
		case "Hashtag":
			hashtags = append(hashtags, &models.NoteTag{
				NoteId: dbNote.ID,
				Tag:    tag.Name,
				TagUrl: tag.Href,
			})
		default:
			log.Warn().Str("tag-type", tag.Type).Msg("Unknown tag type")
		}
	}
	// FIXME: This is bad, what if it's a note update and not a new one?
	// For new notes this is fine, but existing ones might already have attachments.
	// In which case, you need to remove tags that don't exist anymore
	// and only create the ones not yet stored
	err = dbgen.NoteToPing.Save(pings...)
	err = dbgen.NoteTag.Save(hashtags...)
	return dbNote.ID, nil
}