bushiroad: add new entry (#3120)

data/bushiroad

@@ -0,0 +1,31 @@
+argo-bdp.com
+bang-dream.com
+bushi-navi.com
+bushimo.jp
+bushiroad-creative.com
+bushiroad-global-online-store.com
+bushiroad-move.com
+bushiroad-music.com
+bushiroad-store.com
+bushiroad-works.com
+bushiroad.co.jp
+bushiroad.com
+bushiroadgames.com
+cf-vanguard.com
+comic-growl.com
+comicbushi-web.com
+d4dj-pj.com
+dreamorder.com
+fc-buddyfight.com
+gamebiz.jp
+gekkan-bushi.com
+hibiki-cast.jp
+hikosen.co.jp
+njpw.co.jp
+njpw1972.com
+rebirth-fy.com
+revuestarlight.com
+shadowverse-evolve.com
+ws-blau.com
+ws-tcg.com
+wwr-stardom.com

data/category-cdn-!cn

@@ -0,0 +1,15 @@
+include:akamai
+include:cdn77
+include:cloudflare
+include:fastly
+include:imgix
+include:limelight
+include:pagecdn
+include:quantil
+include:stackpath
+include:vrcdn
+
+# Bunny CDN
+b-cdn.net
+bunny.net
+bunnycdn.com

data/category-cdn-cn

@@ -0,0 +1,17 @@
+include:baishancloud
+include:bootcdn
+include:cloudflare-cn
+include:dwion
+include:qiniu
+include:upai
+include:wangsu
+
+## 创世云
+chuangcache.com
+chuangcdn.com
+## FUNCDN
+funcdn.com
+## 北京知道创宇信息技术股份有限公司
+jiashule.com
+jiasule.com
+yunaq.com

data/category-entertainment

@@ -14,6 +14,7 @@ include:bahamut
 include:bandcamp
 include:bangumi
 include:boomerang
+include:bushiroad
 include:catchplay
 include:copymanga
 include:dailymotion
@@ -126,8 +127,6 @@ eroge-road.com
 famitsu.com
 ## 遊戲基地
 gamebase.com.tw
-## gamebiz
-gamebiz.jp
 ## IID, Inc.
 gamebusiness.jp
 gamespark.jp

data/geolocation-!cn

@@ -49,18 +49,10 @@ include:stripe
 include:visa
 
 # CDN companies & Services
-include:akamai
+include:category-cdn-!cn
+
 include:awempire
-include:cdn77
-include:cloudflare
-include:fastly
-include:imgix
-include:limelight
-include:pagecdn
-include:quantil
 include:scaleflex
-include:stackpath
-include:vrcdn
 include:wistia
 
 # Certificate authorities

data/geolocation-cn

@@ -52,26 +52,18 @@ lufaxcdn.com
 pospal.cn
 
 # CDN or SDWAN
+include:category-cdn-cn
+
 include:aws-cn
-include:baishancloud
-include:bootcdn
-include:cloudflare-cn
 include:ctyun
-include:dwion
 include:maocloud
 include:qingcloud
-include:qiniu
 include:snodehome
 include:ucloud
-include:upai
-include:wangsu
 include:yunfanjiasu
 
 ## 百家云
 baijiayun.com
-## 创世云
-chuangcache.com
-chuangcdn.com
 ## 123 云盘 CDN
 cjjd19.com
 ## 明源云
@@ -80,12 +72,6 @@ mingyuanyun.com
 mypaas.com
 mypaas.com.cn
 myypark.com
-## FUNCDN
-funcdn.com
-## 北京知道创宇信息技术股份有限公司
-jiashule.com
-jiasule.com
-yunaq.com
 ## 即刻雾联（北京）
 ## https://github.com/v2fly/domain-list-community/pull/1352#issuecomment-1345476790
 jikeiot.cloud

main.go

@@ -29,19 +29,6 @@ const (
 	RuleTypeInclude    string = "include"
 )
 
-var (
-	TypeChecker  = regexp.MustCompile(`^(domain|full|keyword|regexp|include)$`)
-	ValueChecker = regexp.MustCompile(`^[a-z0-9!\.-]+$`)
-	AttrChecker  = regexp.MustCompile(`^[a-z0-9!-]+$`)
-)
-
-var (
-	refMap    = make(map[string]*List)
-	plMap     = make(map[string]*ParsedList)
-	finalMap  = make(map[string]*List)
-	cirIncMap = make(map[string]bool) // Used for circular inclusion detection
-)
-
 type Entry struct {
 	Type  string
 	Value string
@@ -53,21 +40,15 @@ type List struct {
 	Entry []Entry
 }
 
-type Inclusion struct {
-	Source      string
-	MustAttrs   []*router.Domain_Attribute
-	BannedAttrs []*router.Domain_Attribute
-}
-
 type ParsedList struct {
-	Name       string
+	Name      string
-	Inclusions []Inclusion
+	Inclusion map[string]bool
-	Entry      []Entry
+	Entry     []Entry
 }
 
-func (entryList *List) toPlainText() error {
+func (l *ParsedList) toPlainText(listName string) error {
 	var entryBytes []byte
-	for _, entry := range entryList.Entry {
+	for _, entry := range l.Entry {
 		var attrString string
 		if entry.Attrs != nil {
 			for _, attr := range entry.Attrs {
@@ -78,13 +59,13 @@ func (entryList *List) toPlainText() error {
 		// Entry output format is: type:domain.tld:@attr1,@attr2
 		entryBytes = append(entryBytes, []byte(entry.Type+":"+entry.Value+attrString+"\n")...)
 	}
-	if err := os.WriteFile(filepath.Join(*outputDir, strings.ToLower(entryList.Name)+".txt"), entryBytes, 0644); err != nil {
+	if err := os.WriteFile(filepath.Join(*outputDir, listName+".txt"), entryBytes, 0644); err != nil {
 		return err
 	}
 	return nil
 }
 
-func (l *List) toProto() (*router.GeoSite, error) {
+func (l *ParsedList) toProto() (*router.GeoSite, error) {
 	site := &router.GeoSite{
 		CountryCode: l.Name,
 	}
@@ -98,6 +79,11 @@ func (l *List) toProto() (*router.GeoSite, error) {
 			})
 
 		case RuleTypeRegexp:
+			// check regexp validity to avoid runtime error
+			_, err := regexp.Compile(entry.Value)
+			if err != nil {
+				return nil, fmt.Errorf("invalid regexp in list %s: %s", l.Name, entry.Value)
+			}
 			site.Domain = append(site.Domain, &router.Domain{
 				Type:      router.Domain_Regex,
 				Value:     entry.Value,
@@ -117,72 +103,88 @@ func (l *List) toProto() (*router.GeoSite, error) {
 				Value:     entry.Value,
 				Attribute: entry.Attrs,
 			})
+
+		default:
+			return nil, fmt.Errorf("unknown domain type: %s", entry.Type)
 		}
 	}
 	return site, nil
 }
 
-func exportPlainTextList(exportFiles []string, entryList *List) {
+func exportPlainTextList(list []string, refName string, pl *ParsedList) {
-	for _, exportfilename := range exportFiles {
+	for _, listName := range list {
-		if strings.EqualFold(entryList.Name, exportfilename) {
+		if strings.EqualFold(refName, listName) {
-			if err := entryList.toPlainText(); err != nil {
+			if err := pl.toPlainText(strings.ToLower(refName)); err != nil {
-				fmt.Println("Failed to exportPlainTextList:", err)
+				fmt.Println("Failed:", err)
 				continue
 			}
-			fmt.Printf("'%s' has been generated successfully.\n", exportfilename)
+			fmt.Printf("'%s' has been generated successfully.\n", listName)
 		}
 	}
 }
 
-func parseEntry(line string) (Entry, error) {
+func removeComment(line string) string {
-	var entry Entry
+	idx := strings.Index(line, "#")
-	parts := strings.Fields(line)
+	if idx == -1 {
+		return line
+	}
+	return strings.TrimSpace(line[:idx])
+}
 
-	// Parse/Check type and value
+func parseDomain(domain string, entry *Entry) error {
-	rawTypeVal := parts[0]
+	kv := strings.Split(domain, ":")
-	kv := strings.Split(rawTypeVal, ":")
 	if len(kv) == 1 {
-		entry.Type = RuleTypeDomain // Default type
+		entry.Type = RuleTypeDomain
-		entry.Value = strings.ToLower(rawTypeVal)
+		entry.Value = strings.ToLower(kv[0])
-	} else if len(kv) == 2 {
+		return nil
+	}
+
+	if len(kv) == 2 {
 		entry.Type = strings.ToLower(kv[0])
-		if entry.Type == RuleTypeRegexp {
+
+		if strings.EqualFold(entry.Type, RuleTypeRegexp) {
 			entry.Value = kv[1]
 		} else {
 			entry.Value = strings.ToLower(kv[1])
 		}
-	} else {
+
-		return entry, fmt.Errorf("invalid format: %s", line)
+		return nil
-	}
-	if !TypeChecker.MatchString(entry.Type) {
-		return entry, fmt.Errorf("invalid type: %s", entry.Type)
-	}
-	if entry.Type == RuleTypeRegexp {
-		if _, err := regexp.Compile(entry.Value); err != nil {
-			return entry, fmt.Errorf("invalid regexp: %s", entry.Value)
-		}
-	} else if !ValueChecker.MatchString(entry.Value) {
-		return entry, fmt.Errorf("invalid value: %s", entry.Value)
 	}
 
-	// Parse/Check attributes
+	return fmt.Errorf("invalid format: %s", domain)
-	for _, part := range parts[1:] {
+}
-		if !strings.HasPrefix(part, "@") {
+
-			return entry, fmt.Errorf("invalid attribute: %s", part)
+func parseAttribute(attr string) (*router.Domain_Attribute, error) {
-		}
+	var attribute router.Domain_Attribute
-		attrKey := strings.ToLower(part[1:]) // Trim attribute prefix `@` character
+	if len(attr) == 0 || attr[0] != '@' {
-		if !AttrChecker.MatchString(attrKey) {
+		return &attribute, fmt.Errorf("invalid attribute: %s", attr)
-			return entry, fmt.Errorf("invalid attribute key: %s", attrKey)
+	}
-		}
+
-		entry.Attrs = append(entry.Attrs, &router.Domain_Attribute{
+	attribute.Key = strings.ToLower(attr[1:]) // Trim attribute prefix `@` character
-			Key:        attrKey,
+	attribute.TypedValue = &router.Domain_Attribute_BoolValue{BoolValue: true}
-			TypedValue: &router.Domain_Attribute_BoolValue{BoolValue: true},
+	return &attribute, nil
-		})
+}
+
+func parseEntry(line string) (Entry, error) {
+	line = strings.TrimSpace(line)
+	parts := strings.Split(line, " ")
+
+	var entry Entry
+	if len(parts) == 0 {
+		return entry, fmt.Errorf("empty entry")
+	}
+
+	if err := parseDomain(parts[0], &entry); err != nil {
+		return entry, err
+	}
+
+	for i := 1; i < len(parts); i++ {
+		attr, err := parseAttribute(parts[i])
+		if err != nil {
+			return entry, err
+		}
+		entry.Attrs = append(entry.Attrs, attr)
 	}
-	// Sort attributes
-	sort.Slice(entry.Attrs, func(i, j int) bool {
-		return entry.Attrs[i].Key < entry.Attrs[j].Key
-	})
 
 	return entry, nil
 }
@@ -199,13 +201,9 @@ func Load(path string) (*List, error) {
 	}
 	scanner := bufio.NewScanner(file)
 	for scanner.Scan() {
-		line := scanner.Text()
+		line := strings.TrimSpace(scanner.Text())
-		// Remove comments
+		line = removeComment(line)
-		if idx := strings.Index(line, "#"); idx != -1 {
+		if len(line) == 0 {
-			line = line[:idx]
-		}
-		line = strings.TrimSpace(line)
-		if line == "" {
 			continue
 		}
 		entry, err := parseEntry(line)
@@ -218,68 +216,99 @@ func Load(path string) (*List, error) {
 	return list, nil
 }
 
-func ParseList(refList *List) (*ParsedList, error) {
+func isMatchAttr(Attrs []*router.Domain_Attribute, includeKey string) bool {
-	pl := &ParsedList{Name: refList.Name}
+	isMatch := false
-	for _, entry := range refList.Entry {
+	mustMatch := true
-		if entry.Type == RuleTypeInclude {
+	matchName := includeKey
-			inc := Inclusion{Source: strings.ToUpper(entry.Value)}
+	if strings.HasPrefix(includeKey, "!") {
-			for _, attr := range entry.Attrs {
+		isMatch = true
-				if strings.HasPrefix(attr.Key, "-") {
+		mustMatch = false
-					inc.BannedAttrs = append(inc.BannedAttrs, &router.Domain_Attribute{
+		matchName = strings.TrimLeft(includeKey, "!")
-						Key:        attr.Key[1:], // Trim attribute prefix `-` character
+	}
-						TypedValue: &router.Domain_Attribute_BoolValue{BoolValue: true},
+
-					})
+	for _, Attr := range Attrs {
-				} else {
+		attrName := Attr.Key
-					inc.MustAttrs = append(inc.MustAttrs, attr)
+		if mustMatch {
-				}
+			if matchName == attrName {
+				isMatch = true
+				break
 			}
-			pl.Inclusions = append(pl.Inclusions, inc)
 		} else {
-			pl.Entry = append(pl.Entry, entry)
+			if matchName == attrName {
-		}
+				isMatch = false
-	}
+				break
-	return pl, nil
-}
-
-func isMatchAttrFilters(entry Entry, incFilter Inclusion) bool {
-	attrMap := make(map[string]bool)
-	for _, attr := range entry.Attrs {
-		attrMap[attr.Key] = true
-	}
-	for _, m := range incFilter.MustAttrs {
-		if !attrMap[m.Key] { return false }
-	}
-	for _, b := range incFilter.BannedAttrs {
-		if attrMap[b.Key] { return false }
-	}
-	return true
-}
-
-func ResolveList(pl *ParsedList) error {
-	//TODO: deduplicate
-	if _, pldone := finalMap[pl.Name]; pldone { return nil }
-
-	if cirIncMap[pl.Name] {
-		return fmt.Errorf("circular inclusion in: %s", pl.Name)
-	}
-	cirIncMap[pl.Name] = true
-	defer delete(cirIncMap, pl.Name)
-
-	finalList := &List{Name: pl.Name}
-	finalList.Entry = append(finalList.Entry, pl.Entry...)
-
-	for _, inc := range pl.Inclusions {
-		if err := ResolveList(plMap[inc.Source]); err != nil {
-			return err
-		}
-		for _, entry := range finalMap[inc.Source].Entry {
-			if isMatchAttrFilters(entry, inc) {
-				finalList.Entry = append(finalList.Entry, entry)
 			}
 		}
 	}
-	finalMap[pl.Name] = finalList
+	return isMatch
-	return nil
+}
+
+func createIncludeAttrEntrys(list *List, matchAttr *router.Domain_Attribute) []Entry {
+	newEntryList := make([]Entry, 0, len(list.Entry))
+	matchName := matchAttr.Key
+	for _, entry := range list.Entry {
+		matched := isMatchAttr(entry.Attrs, matchName)
+		if matched {
+			newEntryList = append(newEntryList, entry)
+		}
+	}
+	return newEntryList
+}
+
+func ParseList(list *List, ref map[string]*List) (*ParsedList, error) {
+	pl := &ParsedList{
+		Name:      list.Name,
+		Inclusion: make(map[string]bool),
+	}
+	entryList := list.Entry
+	for {
+		newEntryList := make([]Entry, 0, len(entryList))
+		hasInclude := false
+		for _, entry := range entryList {
+			if entry.Type == RuleTypeInclude {
+				refName := strings.ToUpper(entry.Value)
+				if entry.Attrs != nil {
+					for _, attr := range entry.Attrs {
+						InclusionName := strings.ToUpper(refName + "@" + attr.Key)
+						if pl.Inclusion[InclusionName] {
+							continue
+						}
+						pl.Inclusion[InclusionName] = true
+
+						refList := ref[refName]
+						if refList == nil {
+							return nil, fmt.Errorf("list not found: %s", entry.Value)
+						}
+						attrEntrys := createIncludeAttrEntrys(refList, attr)
+						if len(attrEntrys) != 0 {
+							newEntryList = append(newEntryList, attrEntrys...)
+						}
+					}
+				} else {
+					InclusionName := refName
+					if pl.Inclusion[InclusionName] {
+						continue
+					}
+					pl.Inclusion[InclusionName] = true
+					refList := ref[refName]
+					if refList == nil {
+						return nil, fmt.Errorf("list not found: %s", entry.Value)
+					}
+					newEntryList = append(newEntryList, refList.Entry...)
+				}
+				hasInclude = true
+			} else {
+				newEntryList = append(newEntryList, entry)
+			}
+		}
+		entryList = newEntryList
+		if !hasInclude {
+			break
+		}
+	}
+	pl.Entry = entryList
+
+	return pl, nil
 }
 
 func main() {
@@ -288,7 +317,7 @@ func main() {
 	dir := *dataPath
 	fmt.Println("Use domain lists in", dir)
 
-	// Generate refMap
+	ref := make(map[string]*List)
 	err := filepath.Walk(dir, func(path string, info os.FileInfo, err error) error {
 		if err != nil {
 			return err
@@ -300,7 +329,7 @@ func main() {
 		if err != nil {
 			return err
 		}
-		refMap[list.Name] = list
+		ref[list.Name] = list
 		return nil
 	})
 	if err != nil {
@@ -308,24 +337,6 @@ func main() {
 		os.Exit(1)
 	}
 
-	// Generate plMap
-	for refName, refList := range refMap {
-		pl, err := ParseList(refList)
-		if err != nil {
-			fmt.Println("Failed to ParseList:", err)
-			os.Exit(1)
-		}
-		plMap[refName] = pl
-	}
-
-	// Generate finalMap
-	for _, pl := range plMap {
-		if err := ResolveList(pl); err != nil {
-			fmt.Println("Failed to ResolveList:", err)
-			os.Exit(1)
-		}
-	}
-
 	// Create output directory if not exist
 	if _, err := os.Stat(*outputDir); os.IsNotExist(err) {
 		if mkErr := os.MkdirAll(*outputDir, 0755); mkErr != nil {
@@ -336,8 +347,13 @@ func main() {
 
 	protoList := new(router.GeoSiteList)
 	var existList []string
-	for _, siteEntries := range finalMap {
+	for refName, list := range ref {
-		site, err := siteEntries.toProto()
+		pl, err := ParseList(list, ref)
+		if err != nil {
+			fmt.Println("Failed:", err)
+			os.Exit(1)
+		}
+		site, err := pl.toProto()
 		if err != nil {
 			fmt.Println("Failed:", err)
 			os.Exit(1)
@@ -347,7 +363,7 @@ func main() {
 		// Flatten and export plaintext list
 		if *exportLists != "" {
 			if existList != nil {
-				exportPlainTextList(existList, siteEntries)
+				exportPlainTextList(existList, refName, pl)
 			} else {
 				exportedListSlice := strings.Split(*exportLists, ",")
 				for _, exportedListName := range exportedListSlice {
@@ -360,7 +376,7 @@ func main() {
 					}
 				}
 				if existList != nil {
-					exportPlainTextList(existList, siteEntries)
+					exportPlainTextList(existList, refName, pl)
 				}
 			}
 		}
@@ -373,11 +389,11 @@ func main() {
 
 	protoBytes, err := proto.Marshal(protoList)
 	if err != nil {
-		fmt.Println("Failed to Marshal:", err)
+		fmt.Println("Failed:", err)
 		os.Exit(1)
 	}
 	if err := os.WriteFile(filepath.Join(*outputDir, *outputName), protoBytes, 0644); err != nil {
-		fmt.Println("Failed to write output:", err)
+		fmt.Println("Failed:", err)
 		os.Exit(1)
 	} else {
 		fmt.Println(*outputName, "has been generated successfully.")