create category-non-domestic-cn

They are part of Chinese entities but **exclusively** serving for non-cn area

This is a continuation of https://github.com/v2fly/domain-list-community/pull/3198

data/category-bank-ru

@@ -5,10 +5,6 @@ alfabank.ru
 gazprombank.ru
 gpb.ru
 
-# Mts dengi
-dbo-dengi.online
-mtsdengi.ru
-
 # PSB Bank
 psbank.ru
 
@@ -32,9 +28,5 @@ cdn-tinkoff.ru
 tbank-online.com
 tbank.ru
 
-# Tochka bank
-tochka-tech.com
-tochka.com
-
 # VTB Bank
 vtb.ru

data/category-cryptocurrency

@@ -35,7 +35,6 @@ bitmex.com
 bitquick.co
 bitstamp.net
 bittrex.com
-blockchain.com
 blockfrost.io
 btcbox.co.jp
 cex.io

data/category-non-domestic-cn

@@ -0,0 +1,31 @@
+# Part of Chinese entities but exclusively serving for non-cn area
+
+include:alibaba @!cn
+include:anker @!cn
+include:bilibili @!cn
+include:boc @!cn
+include:bytedance @!cn
+include:ccb @!cn
+include:chinamobile @!cn
+include:chinatelecom @!cn
+include:chinaunicom @!cn
+include:citic @!cn
+include:cmb @!cn
+include:ctrip @!cn
+include:deepin @!cn
+include:dewu @!cn
+include:didi @!cn
+include:eastmoney @!cn
+include:huawei @!cn
+include:icbc @!cn
+include:ipip @!cn
+include:iqiyi @!cn
+include:jd @!cn
+include:oppo @!cn
+include:pingan @!cn
+include:sina @!cn
+include:tencent @!cn
+include:vivo @!cn
+include:xd @!cn
+include:xiaohongshu @!cn
+include:xiaomi @!cn

data/category-porn

@@ -137,7 +137,6 @@ porn
 18hmanga.com
 18insta.com
 18j.tv
-18jav.tv
 18jms.com
 18mh.co
 18mh.me
@@ -2224,8 +2223,6 @@ heavyfetish.com
 hegre.com
 heiguab.top
 heijidi.life
-heiliao.com
-heiliao88.com
 heise360181.buzz
 heise360182.buzz
 helixstudios.net
@@ -2693,7 +2690,6 @@ jasmin.com
 jav-angel.net
 jav-subtitles.com
 jav-vr.net
-jav.com.co
 jav.dog
 jav.gallery
 jav.guru
@@ -2796,7 +2792,6 @@ javpub.me
 javpush.com
 javqd.com
 javrank.com
-javrate.com
 javrave.club
 javroot.com
 javscat.net
@@ -2821,7 +2816,6 @@ javtrust.com
 javtube.cc
 javtube.com
 javtube.net
-javvideoporn.com
 javvids.com
 javxspot.com
 javxxx.me
@@ -3593,7 +3587,6 @@ ninpu.cyou
 niuc2.com
 niziero.info
 njav.tv
-njavtv.com
 nlsexfilmpjes.com
 nlt-media.com
 noc.syosetu.com
@@ -5573,6 +5566,9 @@ xgtd3.com
 xgtdr.buzz
 xh-porn.com
 xh.video
+xhamster.com
+xhamster.desi
+xhamster2.com
 xhot.pro
 xhub.tv
 xiangrikui-app.com
@@ -6161,7 +6157,6 @@ regexp:(^|\.)tqav[1-9][0-9]\.com$
 regexp:(^|\.)tt[1-2][0-9]\.tv$
 regexp:(^|\.)ttghg[1-9][0-9]\.xyz$
 regexp:(^|\.)tttv([1-9][0-9]?|100)\.com$
-regexp:(^|\.)twav[1-9]\.xyz$
 regexp:(^|\.)twseb([1-9][0-9]?)?\.com$
 regexp:(^|\.)uu[a-z][1-9][0-9]?\.com$
 regexp:(^|\.)whtdh0[1-3]\.cc$

data/category-ru

@@ -24,15 +24,10 @@ spvb.ru    # Saint-Petersburg Stock Exchange
 # Financial marketplace
 banki.ru
 finuslugi.ru
-# Investment
-sistema-capital.com
 # Mir payment system
 mirpayonline.ru
 # National Payment Card System
 nspk.ru
-# Tipping service
-netmonet.co
-tips.tips
 
 # Telecom operators
 beeline.ru

data/cn

@@ -2,3 +2,4 @@
 
 include:tld-cn
 include:geolocation-cn
+#include:category-non-domestic-cn

data/geolocation-!cn

@@ -1,5 +1,8 @@
 # This list contains domains that don't have access point in China mainland. This is opposite to geolocation-cn.
 
+# Part of Chinese entities but exclusively serving for non-cn area
+include:category-non-domestic-cn
+
 # AI Chat
 include:category-ai-!cn
 

data/missav

@@ -1,7 +1,5 @@
 missav.ai
 missav.com
-missav.live
 missav.uno
 missav.vip
 missav.ws
-missav123.com

data/xhamster

@@ -3,8 +3,6 @@ xhamster.desi
 xhamster.xxx
 xhamster18.com
 xhamster18.desi
-xhamster19.com
-xhamster3.com
 xhamsterlive.com
 xhcdn.com
 

data/yandex

@@ -55,7 +55,6 @@ yandexcom.net
 yandexmetrica.com
 yandexwebcache.org
 yastat.net
-yastatic-net.ru
 yastatic.net
 
 # Watching movies, included in the Yandex subscription

main.go

@@ -22,11 +22,18 @@ var (
 	exportLists = flag.String("exportlists", "", "Lists to be flattened and exported in plaintext format, separated by ',' comma")
 )
 
+var (
+	plMap     = make(map[string]*ParsedList)
+	finalMap  = make(map[string][]*Entry)
+	cirIncMap = make(map[string]bool) // Used for circular inclusion detection
+)
+
 type Entry struct {
 	Type  string
 	Value string
 	Attrs []string
 	Plain string
+	Affs  []string
 }
 
 type Inclusion struct {
@@ -41,12 +48,6 @@ type ParsedList struct {
 	Entries    []*Entry
 }
 
-type Processor struct {
-	plMap     map[string]*ParsedList
-	finalMap  map[string][]*Entry
-	cirIncMap map[string]bool
-}
-
 func makeProtoList(listName string, entries []*Entry) (*router.GeoSite, error) {
 	site := &router.GeoSite{
 		CountryCode: listName,
@@ -89,27 +90,29 @@ func writePlainList(listname string, entries []*Entry) error {
 	return w.Flush()
 }
 
-func parseEntry(line string) (*Entry, []string, error) {
-	entry := new(Entry)
+func parseEntry(line string) (Entry, error) {
+	var entry Entry
 	parts := strings.Fields(line)
 	if len(parts) == 0 {
-		return entry, nil, fmt.Errorf("empty line")
+		return entry, fmt.Errorf("empty line")
 	}
 
 	// Parse type and value
-	typ, val, isTypeSpecified := strings.Cut(parts[0], ":")
-	typ = strings.ToLower(typ)
-	if !isTypeSpecified { // Default RuleType
-		if !validateDomainChars(typ) {
-			return entry, nil, fmt.Errorf("invalid domain: %q", typ)
+	v := parts[0]
+	colonIndex := strings.Index(v, ":")
+	if colonIndex == -1 {
+		entry.Type = dlc.RuleTypeDomain // Default type
+		entry.Value = strings.ToLower(v)
+		if !validateDomainChars(entry.Value) {
+			return entry, fmt.Errorf("invalid domain: %q", entry.Value)
 		}
-		entry.Type = dlc.RuleTypeDomain
-		entry.Value = typ
 	} else {
+		typ := strings.ToLower(v[:colonIndex])
+		val := v[colonIndex+1:]
 		switch typ {
 		case dlc.RuleTypeRegexp:
 			if _, err := regexp.Compile(val); err != nil {
-				return entry, nil, fmt.Errorf("invalid regexp %q: %w", val, err)
+				return entry, fmt.Errorf("invalid regexp %q: %w", val, err)
 			}
 			entry.Type = dlc.RuleTypeRegexp
 			entry.Value = val
@@ -117,60 +120,57 @@ func parseEntry(line string) (*Entry, []string, error) {
 			entry.Type = dlc.RuleTypeInclude
 			entry.Value = strings.ToUpper(val)
 			if !validateSiteName(entry.Value) {
-				return entry, nil, fmt.Errorf("invalid included list name: %q", entry.Value)
+				return entry, fmt.Errorf("invalid include list name: %q", entry.Value)
 			}
 		case dlc.RuleTypeDomain, dlc.RuleTypeFullDomain, dlc.RuleTypeKeyword:
 			entry.Type = typ
 			entry.Value = strings.ToLower(val)
 			if !validateDomainChars(entry.Value) {
-				return entry, nil, fmt.Errorf("invalid domain: %q", entry.Value)
+				return entry, fmt.Errorf("invalid domain: %q", entry.Value)
 			}
 		default:
-			return entry, nil, fmt.Errorf("invalid type: %q", typ)
+			return entry, fmt.Errorf("invalid type: %q", typ)
 		}
 	}
 
 	// Parse attributes and affiliations
-	var affs []string
 	for _, part := range parts[1:] {
-		switch part[0] {
-		case '@':
-			attr := strings.ToLower(part[1:])
+		if strings.HasPrefix(part, "@") {
+			attr := strings.ToLower(part[1:]) // Trim attribute prefix `@` character
 			if !validateAttrChars(attr) {
-				return entry, affs, fmt.Errorf("invalid attribute: %q", attr)
+				return entry, fmt.Errorf("invalid attribute: %q", attr)
 			}
 			entry.Attrs = append(entry.Attrs, attr)
-		case '&':
-			aff := strings.ToUpper(part[1:])
+		} else if strings.HasPrefix(part, "&") {
+			aff := strings.ToUpper(part[1:]) // Trim affiliation prefix `&` character
 			if !validateSiteName(aff) {
-				return entry, affs, fmt.Errorf("invalid affiliation: %q", aff)
+				return entry, fmt.Errorf("invalid affiliation: %q", aff)
 			}
-			affs = append(affs, aff)
-		default:
-			return entry, affs, fmt.Errorf("invalid attribute/affiliation: %q", part)
+			entry.Affs = append(entry.Affs, aff)
+		} else {
+			return entry, fmt.Errorf("invalid attribute/affiliation: %q", part)
 		}
 	}
+	// Sort attributes
+	slices.Sort(entry.Attrs)
+	// Formated plain entry: type:domain.tld:@attr1,@attr2
+	var plain strings.Builder
+	plain.Grow(len(entry.Type) + len(entry.Value) + 10)
+	plain.WriteString(entry.Type)
+	plain.WriteByte(':')
+	plain.WriteString(entry.Value)
+	for i, attr := range entry.Attrs {
+		if i == 0 {
+			plain.WriteByte(':')
+		} else {
+			plain.WriteByte(',')
+		}
+		plain.WriteByte('@')
+		plain.WriteString(attr)
+	}
+	entry.Plain = plain.String()
 
-	if entry.Type != dlc.RuleTypeInclude {
-		slices.Sort(entry.Attrs) // Sort attributes
-		// Formated plain entry: type:domain.tld:@attr1,@attr2
-		var plain strings.Builder
-		plain.Grow(len(entry.Type) + len(entry.Value) + 10)
-		plain.WriteString(entry.Type)
-		plain.WriteByte(':')
-		plain.WriteString(entry.Value)
-		for i, attr := range entry.Attrs {
-			if i == 0 {
-				plain.WriteByte(':')
-			} else {
-				plain.WriteByte(',')
-			}
-			plain.WriteByte('@')
-			plain.WriteString(attr)
-		}
-		entry.Plain = plain.String()
-	}
-	return entry, affs, nil
+	return entry, nil
 }
 
 func validateDomainChars(domain string) bool {
@@ -206,54 +206,62 @@ func validateSiteName(name string) bool {
 	return true
 }
 
-func (p *Processor) getOrCreateParsedList(name string) *ParsedList {
-	pl, exist := p.plMap[name]
-	if !exist {
-		pl = &ParsedList{Name: name}
-		p.plMap[name] = pl
-	}
-	return pl
-}
-
-func (p *Processor) loadData(listName string, path string) error {
+func loadData(path string) ([]*Entry, error) {
 	file, err := os.Open(path)
 	if err != nil {
-		return err
+		return nil, err
 	}
 	defer file.Close()
 
-	pl := p.getOrCreateParsedList(listName)
+	var entries []*Entry
 	scanner := bufio.NewScanner(file)
 	lineIdx := 0
 	for scanner.Scan() {
+		line := scanner.Text()
 		lineIdx++
-		line, _, _ := strings.Cut(scanner.Text(), "#") // Remove comments
+		if idx := strings.Index(line, "#"); idx != -1 {
+			line = line[:idx] // Remove comments
+		}
 		line = strings.TrimSpace(line)
 		if line == "" {
 			continue
 		}
-		entry, affs, err := parseEntry(line)
+		entry, err := parseEntry(line)
 		if err != nil {
-			return fmt.Errorf("error in %q at line %d: %w", path, lineIdx, err)
+			return entries, fmt.Errorf("error in %q at line %d: %w", path, lineIdx, err)
 		}
+		entries = append(entries, &entry)
+	}
+	return entries, nil
+}
 
+func parseList(refName string, refList []*Entry) error {
+	pl, _ := plMap[refName]
+	if pl == nil {
+		pl = &ParsedList{Name: refName}
+		plMap[refName] = pl
+	}
+	for _, entry := range refList {
 		if entry.Type == dlc.RuleTypeInclude {
+			if len(entry.Affs) != 0 {
+				return fmt.Errorf("affiliation is not allowed for include:%q", entry.Value)
+			}
 			inc := &Inclusion{Source: entry.Value}
 			for _, attr := range entry.Attrs {
-				if attr[0] == '-' {
-					inc.BanAttrs = append(inc.BanAttrs, attr[1:])
+				if strings.HasPrefix(attr, "-") {
+					inc.BanAttrs = append(inc.BanAttrs, attr[1:]) // Trim attribute prefix `-` character
 				} else {
 					inc.MustAttrs = append(inc.MustAttrs, attr)
 				}
 			}
-			for _, aff := range affs {
-				apl := p.getOrCreateParsedList(aff)
-				apl.Inclusions = append(apl.Inclusions, inc)
-			}
 			pl.Inclusions = append(pl.Inclusions, inc)
 		} else {
-			for _, aff := range affs {
-				apl := p.getOrCreateParsedList(aff)
+			for _, aff := range entry.Affs {
+				apl, _ := plMap[aff]
+				if apl == nil {
+					apl = &ParsedList{Name: aff}
+					plMap[aff] = apl
+				}
 				apl.Entries = append(apl.Entries, entry)
 			}
 			pl.Entries = append(pl.Entries, entry)
@@ -288,7 +296,9 @@ func polishList(roughMap map[string]*Entry) []*Entry {
 	domainsMap := make(map[string]bool)
 	for _, entry := range roughMap {
 		switch entry.Type { // Bypass regexp, keyword and "full/domain with attr"
-		case dlc.RuleTypeRegexp, dlc.RuleTypeKeyword:
+		case dlc.RuleTypeRegexp:
+			finalList = append(finalList, entry)
+		case dlc.RuleTypeKeyword:
 			finalList = append(finalList, entry)
 		case dlc.RuleTypeDomain:
 			domainsMap[entry.Value] = true
@@ -313,11 +323,11 @@ func polishList(roughMap map[string]*Entry) []*Entry {
 			pd = "." + pd // So that `domain:example.org` overrides `full:example.org`
 		}
 		for {
-			var hasParent bool
-			_, pd, hasParent = strings.Cut(pd, ".") // Go for next parent
-			if !hasParent {
+			idx := strings.Index(pd, ".")
+			if idx == -1 {
 				break
 			}
+			pd = pd[idx+1:] // Go for next parent
 			if domainsMap[pd] {
 				isRedundant = true
 				break
@@ -334,38 +344,36 @@ func polishList(roughMap map[string]*Entry) []*Entry {
 	return finalList
 }
 
-func (p *Processor) resolveList(plname string) error {
-	if _, pldone := p.finalMap[plname]; pldone {
+func resolveList(pl *ParsedList) error {
+	if _, pldone := finalMap[pl.Name]; pldone {
 		return nil
 	}
-	pl, plexist := p.plMap[plname]
-	if !plexist {
-		return fmt.Errorf("list %q not found", plname)
+
+	if cirIncMap[pl.Name] {
+		return fmt.Errorf("circular inclusion in: %q", pl.Name)
 	}
-	if p.cirIncMap[plname] {
-		return fmt.Errorf("circular inclusion in: %q", plname)
-	}
-	p.cirIncMap[plname] = true
-	defer delete(p.cirIncMap, plname)
+	cirIncMap[pl.Name] = true
+	defer delete(cirIncMap, pl.Name)
 
 	roughMap := make(map[string]*Entry) // Avoid basic duplicates
 	for _, dentry := range pl.Entries { // Add direct entries
 		roughMap[dentry.Plain] = dentry
 	}
 	for _, inc := range pl.Inclusions {
-		if _, exist := p.plMap[inc.Source]; !exist {
-			return fmt.Errorf("list %q includes a non-existent list: %q", plname, inc.Source)
+		incPl, exist := plMap[inc.Source]
+		if !exist {
+			return fmt.Errorf("list %q includes a non-existent list: %q", pl.Name, inc.Source)
 		}
-		if err := p.resolveList(inc.Source); err != nil {
+		if err := resolveList(incPl); err != nil {
 			return err
 		}
-		for _, ientry := range p.finalMap[inc.Source] {
+		for _, ientry := range finalMap[inc.Source] {
 			if isMatchAttrFilters(ientry, inc) { // Add included entries
 				roughMap[ientry.Plain] = ientry
 			}
 		}
 	}
-	p.finalMap[plname] = polishList(roughMap)
+	finalMap[pl.Name] = polishList(roughMap)
 	return nil
 }
 
@@ -373,8 +381,8 @@ func run() error {
 	dir := *dataPath
 	fmt.Printf("using domain lists data in %q\n", dir)
 
-	// Generate plMap
-	processor := &Processor{plMap: make(map[string]*ParsedList)}
+	// Generate refMap
+	refMap := make(map[string][]*Entry)
 	err := filepath.WalkDir(dir, func(path string, d os.DirEntry, err error) error {
 		if err != nil {
 			return err
@@ -386,16 +394,23 @@ func run() error {
 		if !validateSiteName(listName) {
 			return fmt.Errorf("invalid list name: %q", listName)
 		}
-		return processor.loadData(listName, path)
+		refMap[listName], err = loadData(path)
+		return err
 	})
 	if err != nil {
 		return fmt.Errorf("failed to loadData: %w", err)
 	}
+
+	// Generate plMap
+	for refName, refList := range refMap {
+		if err := parseList(refName, refList); err != nil {
+			return fmt.Errorf("failed to parseList %q: %w", refName, err)
+		}
+	}
+
 	// Generate finalMap
-	processor.finalMap = make(map[string][]*Entry, len(processor.plMap))
-	processor.cirIncMap = make(map[string]bool)
-	for plname := range processor.plMap {
-		if err := processor.resolveList(plname); err != nil {
+	for plname, pl := range plMap {
+		if err := resolveList(pl); err != nil {
 			return fmt.Errorf("failed to resolveList %q: %w", plname, err)
 		}
 	}
@@ -404,10 +419,11 @@ func run() error {
 	if err := os.MkdirAll(*outputDir, 0755); err != nil {
 		return fmt.Errorf("failed to create output directory: %w", err)
 	}
-	// Export plaintext lists
+
+	// Export plaintext list
 	for rawEpList := range strings.SplitSeq(*exportLists, ",") {
 		if epList := strings.TrimSpace(rawEpList); epList != "" {
-			entries, exist := processor.finalMap[strings.ToUpper(epList)]
+			entries, exist := finalMap[strings.ToUpper(epList)]
 			if !exist || len(entries) == 0 {
 				fmt.Printf("list %q does not exist or is empty\n", epList)
 				continue
@@ -422,7 +438,7 @@ func run() error {
 
 	// Generate dat file
 	protoList := new(router.GeoSiteList)
-	for siteName, siteEntries := range processor.finalMap {
+	for siteName, siteEntries := range finalMap {
 		site, err := makeProtoList(siteName, siteEntries)
 		if err != nil {
 			return fmt.Errorf("failed to makeProtoList %q: %w", siteName, err)