kindexr/internal/nostr/parser.go

package nostr

import (
	"fmt"
	"regexp"
	"strconv"
	"strings"
	"time"

	nostrlib "github.com/nbd-wtf/go-nostr"

	"git.utn.lol/enki/kindexr/internal/db"
	"git.utn.lol/enki/kindexr/internal/torznab"
)

var infoHashRe = regexp.MustCompile(`^[0-9a-fA-F]{40}$`)

// Parse converts a kind 2003 Nostr event into a TorrentRecord ready for storage.
// Returns an error if the event is missing required fields.
func Parse(event *nostrlib.Event) (*db.TorrentRecord, error) {
	if event.Kind != nostrlib.KindTorrent {
		return nil, fmt.Errorf("wrong kind: %d", event.Kind)
	}

	r := &db.TorrentRecord{
		EventID:    event.ID,
		Pubkey:     event.PubKey,
		CreatedAt:  int64(event.CreatedAt),
		IngestedAt: time.Now().Unix(),
		RawEvent:   event.String(),
	}

	var totalSize int64
	hasTotalSize := false

	for _, tag := range event.Tags {
		if len(tag) < 2 {
			continue
		}
		switch tag[0] {
		case "x":
			hash := strings.ToLower(tag[1])
			if !infoHashRe.MatchString(hash) {
				return nil, fmt.Errorf("invalid info_hash: %q", tag[1])
			}
			r.InfoHash = hash

		case "title":
			r.Title = tag[1]

		case "file":
			path := tag[1]
			rec := db.FileRecord{Path: path}
			if len(tag) >= 3 {
				if sz, err := strconv.ParseInt(tag[2], 10, 64); err == nil {
					rec.SizeBytes = &sz
					totalSize += sz
					hasTotalSize = true
				}
			}
			r.Files = append(r.Files, rec)

		case "tracker":
			r.Trackers = append(r.Trackers, tag[1])

		case "i":
			parseITag(tag[1], r)

		case "t":
			r.Tags = append(r.Tags, tag[1])
		}
	}

	if r.InfoHash == "" {
		return nil, fmt.Errorf("missing x (info_hash) tag")
	}

	// Fall back to first line of content if no title tag.
	if r.Title == "" {
		if idx := strings.IndexByte(event.Content, '\n'); idx >= 0 {
			r.Title = strings.TrimSpace(event.Content[:idx])
		} else {
			r.Title = strings.TrimSpace(event.Content)
		}
	}
	if r.Title == "" {
		return nil, fmt.Errorf("event has no title")
	}

	r.Description = event.Content

	if hasTotalSize {
		r.SizeBytes = &totalSize
	}

	return r, nil
}

// parseITag parses a single value from an ["i", "<value>"] tag and sets
// the appropriate fields on r.
func parseITag(value string, r *db.TorrentRecord) {
	switch {
	case strings.HasPrefix(value, "newznab:"):
		if cat, err := strconv.Atoi(strings.TrimPrefix(value, "newznab:")); err == nil {
			r.NewznabCat = &cat
		}

	case strings.HasPrefix(value, "tcat:"):
		// Only use tcat if newznab: wasn't already set.
		if r.NewznabCat == nil {
			cat := torznab.TcatToNewznab(strings.TrimPrefix(value, "tcat:"))
			r.Category = strings.TrimPrefix(value, "tcat:")
			r.NewznabCat = &cat
		} else {
			r.Category = strings.TrimPrefix(value, "tcat:")
		}

	case strings.HasPrefix(value, "imdb:"):
		r.ImdbID = strings.TrimPrefix(value, "imdb:")

	case strings.HasPrefix(value, "tmdb:movie:"):
		r.TmdbID = "movie:" + strings.TrimPrefix(value, "tmdb:movie:")

	case strings.HasPrefix(value, "tmdb:tv:"):
		r.TmdbID = "tv:" + strings.TrimPrefix(value, "tmdb:tv:")

	case strings.HasPrefix(value, "tvdb:"):
		r.TvdbID = strings.TrimPrefix(value, "tvdb:")
	}
}