initial Go port of streamrip

internal/urlparse/parse.go

@@ -0,0 +1,202 @@
+package urlparse
+
+import (
+	"net/url"
+	"regexp"
+	"strings"
+)
+
+type URLKind string
+
+const (
+	KindGeneric       URLKind = "generic"
+	KindDeezerDynamic URLKind = "deezer_dynamic"
+	KindSoundcloud    URLKind = "soundcloud"
+)
+
+type ParsedURL struct {
+	OriginalURL string
+	Source      string
+	MediaType   string
+	ID          string
+	Kind        URLKind
+}
+
+var deezerDynamicRe = regexp.MustCompile(`^https?://dzr\.page\.link/`)
+
+func Parse(raw string) *ParsedURL {
+	if deezerDynamicRe.MatchString(raw) {
+		return &ParsedURL{
+			OriginalURL: raw,
+			Source:      "deezer",
+			Kind:        KindDeezerDynamic,
+		}
+	}
+
+	u, err := url.Parse(raw)
+	if err != nil || u.Host == "" {
+		return nil
+	}
+
+	host := normalizeHost(u.Host)
+	path := strings.Trim(u.EscapedPath(), "/")
+	parts := splitParts(path)
+
+	switch {
+	case isQobuzHost(host):
+		return parseQobuz(raw, parts)
+	case isTidalHost(host):
+		return parseTidal(raw, parts)
+	case isDeezerHost(host):
+		return parseDeezer(raw, parts)
+	case host == "soundcloud.com":
+		return parseSoundcloud(raw, parts)
+	default:
+		return nil
+	}
+}
+
+func parseQobuz(raw string, parts []string) *ParsedURL {
+	if len(parts) < 2 {
+		return nil
+	}
+
+	if isLocaleToken(parts[0]) {
+		parts = parts[1:]
+	}
+	if len(parts) < 2 {
+		return nil
+	}
+
+	mediaType := parts[0]
+	if !isSupportedMedia(mediaType) {
+		return nil
+	}
+	id := parts[len(parts)-1]
+	if id == "" {
+		return nil
+	}
+
+	return &ParsedURL{OriginalURL: raw, Source: "qobuz", MediaType: mediaType, ID: id, Kind: KindGeneric}
+}
+
+func parseTidal(raw string, parts []string) *ParsedURL {
+	if len(parts) < 2 {
+		return nil
+	}
+
+	if parts[0] == "browse" {
+		parts = parts[1:]
+	}
+	if len(parts) < 2 {
+		return nil
+	}
+
+	mediaType := parts[0]
+	if !isSupportedMedia(mediaType) {
+		return nil
+	}
+	id := parts[1]
+	if id == "" {
+		return nil
+	}
+
+	return &ParsedURL{OriginalURL: raw, Source: "tidal", MediaType: mediaType, ID: id, Kind: KindGeneric}
+}
+
+func parseDeezer(raw string, parts []string) *ParsedURL {
+	if len(parts) < 2 {
+		return nil
+	}
+
+	if isLangToken(parts[0]) {
+		parts = parts[1:]
+	}
+	if len(parts) < 2 {
+		return nil
+	}
+
+	mediaType := parts[0]
+	if !isSupportedMedia(mediaType) {
+		return nil
+	}
+	id := parts[1]
+	if id == "" {
+		return nil
+	}
+
+	return &ParsedURL{OriginalURL: raw, Source: "deezer", MediaType: mediaType, ID: id, Kind: KindGeneric}
+}
+
+func parseSoundcloud(raw string, parts []string) *ParsedURL {
+	if len(parts) < 2 {
+		return nil
+	}
+
+	mediaType := "track"
+	if len(parts) >= 3 && parts[1] == "sets" {
+		mediaType = "playlist"
+	}
+
+	return &ParsedURL{OriginalURL: raw, Source: "soundcloud", MediaType: mediaType, ID: raw, Kind: KindSoundcloud}
+}
+
+func splitParts(path string) []string {
+	if path == "" {
+		return nil
+	}
+	raw := strings.Split(path, "/")
+	parts := make([]string, 0, len(raw))
+	for _, p := range raw {
+		if p != "" {
+			parts = append(parts, p)
+		}
+	}
+	return parts
+}
+
+func normalizeHost(host string) string {
+	h := strings.ToLower(host)
+	return strings.TrimPrefix(h, "www.")
+}
+
+func isQobuzHost(host string) bool {
+	return host == "qobuz.com" || host == "open.qobuz.com" || host == "play.qobuz.com"
+}
+
+func isTidalHost(host string) bool {
+	return host == "tidal.com" || host == "open.tidal.com" || host == "listen.tidal.com"
+}
+
+func isDeezerHost(host string) bool {
+	return host == "deezer.com"
+}
+
+func isSupportedMedia(mediaType string) bool {
+	switch mediaType {
+	case "album", "track", "playlist", "artist", "label":
+		return true
+	default:
+		return false
+	}
+}
+
+func isLocaleToken(s string) bool {
+	if len(s) != 5 {
+		return false
+	}
+	return s[2] == '-' && isAlpha(s[:2]) && isAlpha(s[3:])
+}
+
+func isLangToken(s string) bool {
+	return len(s) == 2 && isAlpha(s)
+}
+
+func isAlpha(s string) bool {
+	for _, r := range s {
+		if (r < 'a' || r > 'z') && (r < 'A' || r > 'Z') {
+			return false
+		}
+	}
+	return true
+}