MeBox/internal/service/strm_tree_normalize.go

package service

import (
	"net/url"
	"path"
	"regexp"
	"strings"
)

var strmTreeListPrefixRE = regexp.MustCompile(`^\s*(?:[-*+•·]\s+|\d{1,4}[\.)、]\s+)`)

var strmTreeURLPathQueryKeys = []string{"path", "ref", "file", "filepath", "filename", "name"}

func normalizeSTRMTreeProvider(raw string) string {
	switch strings.ToLower(strings.TrimSpace(raw)) {
	case "115", "115pan", "pan115", "cloud115":
		return "cloud115"
	case "alist", "openlist", "webdav", "clouddrive2", "s3":
		return strings.ToLower(strings.TrimSpace(raw))
	default:
		return strings.ToLower(strings.TrimSpace(raw))
	}
}

func normalizeSTRMTreeSource(raw string) string {
	value := strings.TrimSpace(strings.Trim(raw, `"'`))
	if value == "" {
		return ""
	}
	if strings.HasPrefix(strings.ToLower(value), "cloud://") {
		parts := strmLibraryPathParts(value)
		if len(parts) == 0 {
			return ""
		}
		return "/" + path.Join(parts...)
	}
	if parsed, err := url.Parse(value); err == nil && parsed.Scheme != "" && parsed.Scheme != "file" {
		if ref := strmTreeSourceFromURL(parsed); ref != "" {
			value = ref
		}
	}
	value = trimSTRMTreeMediaMetadata(value)
	value = strings.ReplaceAll(value, "\\", "/")
	value = pathAfterWindowsDrivePrefix(value)
	value = cleanSTRMTreePathDecorations(value)
	value = strings.TrimSpace(value)
	if value == "" {
		return ""
	}
	if strmTreeHasUnsafeSegment(value) {
		return strings.Trim(value, "/")
	}
	if strings.HasPrefix(value, "/") {
		return path.Clean(value)
	}
	return strings.TrimPrefix(path.Clean("/"+value), "/")
}

func normalizeSTRMTreeSourceWithProvider(raw, fallbackProvider string) strmTreeSource {
	provider := normalizeSTRMTreeProvider(fallbackProvider)
	value := strings.TrimSpace(strings.Trim(raw, `"'`))
	if value == "" {
		return strmTreeSource{}
	}
	if info, ok := ParseCloudLibraryMount(value); ok {
		source := normalizeSTRMTreeSource(info.DisplayDir)
		if source == "" {
			return strmTreeSource{}
		}
		ref := normalizeSTRMTreeSource(info.ScanDir)
		if ref == "" {
			ref = source
		}
		return strmTreeSource{Provider: normalizeSTRMTreeProvider(info.Provider), Path: source, RefPath: ref}
	}
	if parsed, err := url.Parse(value); err == nil {
		if cloudProvider, ref := strmTreeCloudPlaySourceFromURL(parsed); cloudProvider != "" && ref != "" {
			source := normalizeSTRMTreeSource(ref)
			if source == "" {
				return strmTreeSource{}
			}
			return strmTreeSource{Provider: normalizeSTRMTreeProvider(cloudProvider), Path: source}
		}
	}
	source := normalizeSTRMTreeSource(value)
	if source == "" {
		return strmTreeSource{}
	}
	return strmTreeSource{Provider: provider, Path: source}
}

func strmTreeCloudPlaySourceFromURL(parsed *url.URL) (string, string) {
	if parsed == nil {
		return "", ""
	}
	parts := strings.Split(strings.Trim(parsed.Path, "/"), "/")
	if len(parts) < 4 || !strings.EqualFold(parts[0], "api") || !strings.EqualFold(parts[1], "cloud") || !strings.EqualFold(parts[2], "play") {
		return "", ""
	}
	provider := strings.TrimSpace(parts[3])
	if decoded, err := url.PathUnescape(provider); err == nil {
		provider = decoded
	}
	// /api/cloud/play is generated by this app with url.Values, where '+'
	// represents a space. Decode the standard query view first so re-importing
	// existing STRM links does not turn "My Movie.mkv" into "My+Movie.mkv".
	for _, value := range parsed.Query()["ref"] {
		if ref := strmTreeURLQueryMediaSource(value); ref != "" {
			return provider, ref
		}
	}
	for _, value := range strmTreeRawQueryValues(parsed.RawQuery, "ref") {
		if ref := strmTreeURLQueryMediaSource(value); ref != "" {
			return provider, ref
		}
	}
	return "", ""
}

func strmTreeSourceFromURL(parsed *url.URL) string {
	if parsed == nil {
		return ""
	}
	for _, key := range strmTreeURLPathQueryKeys {
		for _, value := range strmTreeRawQueryValues(parsed.RawQuery, key) {
			if ref := strmTreeURLQueryMediaSource(value); ref != "" {
				return ref
			}
		}
		for _, value := range parsed.Query()[key] {
			if ref := strmTreeURLQueryMediaSource(value); ref != "" {
				return ref
			}
		}
	}
	if parsed.Path != "" && strmTreeURLQueryMediaSource(parsed.Path) != "" {
		return parsed.Path
	}
	return ""
}

func strmTreeRawQueryValues(rawQuery, key string) []string {
	key = strings.ToLower(strings.TrimSpace(key))
	if rawQuery == "" || key == "" {
		return nil
	}
	values := make([]string, 0, 1)
	for _, pair := range strings.Split(rawQuery, "&") {
		if pair == "" {
			continue
		}
		rawKey, rawValue, _ := strings.Cut(pair, "=")
		decodedKey, err := url.QueryUnescape(rawKey)
		if err != nil {
			decodedKey = rawKey
		}
		if strings.ToLower(strings.TrimSpace(decodedKey)) != key {
			continue
		}
		decodedValue, err := url.PathUnescape(rawValue)
		if err != nil {
			decodedValue = rawValue
		}
		values = append(values, decodedValue)
	}
	return values
}

func strmTreeURLQueryMediaSource(raw string) string {
	value := strings.TrimSpace(strings.Trim(raw, `"'`))
	if value == "" {
		return ""
	}
	if parsed, err := url.Parse(value); err == nil && parsed.Scheme != "" && parsed.Scheme != "file" {
		value = parsed.Path
	}
	value = trimSTRMTreeMediaMetadata(value)
	if strmTreeSourceIsVideo(value) {
		return value
	}
	return ""
}

func strmTreeHasUnsafeSegment(value string) bool {
	for _, part := range strings.Split(strings.ReplaceAll(value, "\\", "/"), "/") {
		if strings.TrimSpace(part) == ".." {
			return true
		}
	}
	return false
}

func strmTreeSourceIsVideo(source string) bool {
	ext := strings.ToLower(path.Ext(trimSTRMTreeMediaMetadata(cleanSTRMTreeItemName(source))))
	if ext == ".strm" {
		return false
	}
	_, ok := videoExtensions[ext]
	return ok
}

func cleanSTRMTreePathDecorations(value string) string {
	if value == "" {
		return ""
	}
	leadingSlash := strings.HasPrefix(value, "/")
	parts := strings.Split(value, "/")
	for i, part := range parts {
		if part == "" {
			continue
		}
		parts[i] = cleanSTRMTreeItemName(part)
	}
	cleaned := strings.Join(parts, "/")
	if leadingSlash && !strings.HasPrefix(cleaned, "/") {
		cleaned = "/" + cleaned
	}
	return cleaned
}

func cleanSTRMTreeItemName(value string) string {
	value = strings.TrimSpace(strings.Trim(value, `"'`))
	for {
		before := value
		value = strings.TrimSpace(strmTreeListPrefixRE.ReplaceAllString(value, ""))
		value = strings.TrimSpace(strings.TrimLeft(value, "\ufeff📁📂📄📃🎬🎞🎥📺🗂️"))
		lower := strings.ToLower(value)
		for _, marker := range []string{
			"[file]", "[f]", "[文件]", "[檔案]", "[档案]",
			"[dir]", "[d]", "[folder]", "[目录]", "[目錄]", "[文件夹]", "[資料夾]",
			"(file)", "(f)", "(文件)", "(檔案)", "(档案)",
			"(dir)", "(d)", "(folder)", "(目录)", "(目錄)", "(文件夹)", "(資料夾)",
		} {
			if strings.HasPrefix(lower, marker) {
				value = strings.TrimLeft(strings.TrimSpace(value[len(marker):]), "-_:： ")
				break
			}
		}
		if value == before {
			return value
		}
	}
}

func trimSTRMTreeMediaMetadata(value string) string {
	value = strings.TrimSpace(value)
	lower := strings.ToLower(value)
	bestEnd := -1
	for ext := range videoExtensions {
		from := 0
		for {
			idx := strings.Index(lower[from:], ext)
			if idx < 0 {
				break
			}
			idx += from
			end := idx + len(ext)
			if strmTreeMediaExtBoundary(value, end) && end > bestEnd {
				bestEnd = end
			}
			from = idx + 1
			if from >= len(lower) {
				break
			}
		}
	}
	if bestEnd > 0 && bestEnd < len(value) {
		return strings.TrimSpace(value[:bestEnd])
	}
	return value
}

func strmTreeMediaExtBoundary(value string, end int) bool {
	if end >= len(value) {
		return true
	}
	for _, r := range value[end:] {
		if r == '(' || r == '[' || r == '{' || r == '<' || r == '（' || r == '【' || r == '，' || r == ',' {
			return true
		}
		return r == ' ' || r == '\t' || r == '\r' || r == '\n'
	}
	return true
}