category-electronic-cn: remove non-cn domains (#3375)

They were added in `espressif` and included in `geolocation-!cn`.

.gitignore

@@ -4,9 +4,9 @@
 /domain-list-community
 /domain-list-community.exe
 
-# Generated dat file.
+# Generated dat files.
-dlc.dat
+/*.dat
 
 # Exported plaintext lists.
-/*.yml
 /*.txt
+/*.yml

data/category-dev

@@ -54,6 +54,7 @@ include:stackexchange
 include:strikingly
 include:termux
 include:thelinuxfoundation
+include:tilda
 include:unity
 include:v8
 
@@ -152,4 +153,5 @@ wireshark.org
 x.org
 xposed.info
 yarnpkg.com
+ziglang.org
 zsh.org

data/category-electronic-cn

@@ -17,11 +17,10 @@ bouffalolab.com
 cxmt.com
 
 # 乐鑫信息科技
+#include:espressif
 esp8266.cn
-esp8266.com
 esp8266.com.cn
 espressif.cn
-espressif.com
 espressif.com.cn
 
 # 华秋电子

data/category-entertainment-ru

@@ -11,6 +11,7 @@ include:okko
 include:wink
 24h.tv
 amediateka.ru
+beeline.tv
 ivi.ru
 premier.one
 smotreshka.tv

data/category-ip-geo-detect

@@ -134,6 +134,7 @@ full:checkip.amazonaws.com
 full:ipv4-check-perf.radar.cloudflare.com
 full:ipv6-check-perf.radar.cloudflare.com
 geoip.noc.gov.ru
+ip.hetzner.com
 ip.mail.ru
 ip.nic.ru
 ip.tyk.nu

data/geolocation-cn

@@ -1625,3 +1625,6 @@ ao-x.ac.cn
 
 # 万集科技 京ICP备18036282号-2
 wanji.net.cn
+
+# 广州市雅望互联网服务有限公司
+gzyowin.com

data/qcloud

@@ -258,13 +258,14 @@ tdnsv14.net
 tdnsv15.net
 
 # myqcloud inside mainland China
-regexp:\.(.+-)?ap-beijing(-.+)?\.myqcloud\.com$             #北京
+# overrided by myqcloud.com
-regexp:\.(.+-)?ap-nanjing(-.+)?\.myqcloud\.com$             #南京
+#regexp:\.(.+-)?ap-beijing(-.+)?\.myqcloud\.com$             #北京
-regexp:\.(.+-)?ap-shanghai(-.+)?\.myqcloud\.com$            #上海
+#regexp:\.(.+-)?ap-nanjing(-.+)?\.myqcloud\.com$             #南京
-regexp:\.(.+-)?ap-guangzhou(-.+)?\.myqcloud\.com$           #广州
+#regexp:\.(.+-)?ap-shanghai(-.+)?\.myqcloud\.com$            #上海
-regexp:\.(.+-)?ap-chengdu(-.+)?\.myqcloud\.com$             #成都
+#regexp:\.(.+-)?ap-guangzhou(-.+)?\.myqcloud\.com$           #广州
-regexp:\.(.+-)?ap-chongqing(-.+)?\.myqcloud\.com$           #重庆
+#regexp:\.(.+-)?ap-chengdu(-.+)?\.myqcloud\.com$             #成都
-regexp:\.(.+-)?ap-shenzhen(-.+)?\.myqcloud\.com$            #深圳
+#regexp:\.(.+-)?ap-chongqing(-.+)?\.myqcloud\.com$           #重庆
+#regexp:\.(.+-)?ap-shenzhen(-.+)?\.myqcloud\.com$            #深圳
 
 # COS 使用到的非中国大陆的地域与可用区，参见 https://cloud.tencent.com/document/product/436/6224
 ap-hongkong.myqcloud.com @!cn       #中国香港
@@ -282,13 +283,14 @@ eu-frankfurt.myqcloud.com @!cn      #法兰克福
 eu-moscow.myqcloud.com @!cn         #莫斯科
 
 # tencentcos inside mainland China
-regexp:\.(.+-)?ap-beijing(-.+)?\.tencentcos\.(cn|com(\.cn)?)$   #北京
+# overrided by tencentcos.cn, tencentcos.com, tencentcos.com.cn
-regexp:\.(.+-)?ap-nanjing(-.+)?\.tencentcos\.(cn|com(\.cn)?)$   #南京
+#regexp:\.(.+-)?ap-beijing(-.+)?\.tencentcos\.(cn|com(\.cn)?)$   #北京
-regexp:\.(.+-)?ap-shanghai(-.+)?\.tencentcos\.(cn|com(\.cn)?)$  #上海
+#regexp:\.(.+-)?ap-nanjing(-.+)?\.tencentcos\.(cn|com(\.cn)?)$   #南京
-regexp:\.(.+-)?ap-guangzhou(-.+)?\.tencentcos\.(cn|com(\.cn)?)$ #广州
+#regexp:\.(.+-)?ap-shanghai(-.+)?\.tencentcos\.(cn|com(\.cn)?)$  #上海
-regexp:\.(.+-)?ap-chengdu(-.+)?\.tencentcos\.(cn|com(\.cn)?)$   #成都
+#regexp:\.(.+-)?ap-guangzhou(-.+)?\.tencentcos\.(cn|com(\.cn)?)$ #广州
-regexp:\.(.+-)?ap-chongqing(-.+)?\.tencentcos\.(cn|com(\.cn)?)$ #重庆
+#regexp:\.(.+-)?ap-chengdu(-.+)?\.tencentcos\.(cn|com(\.cn)?)$   #成都
-regexp:\.(.+-)?ap-shenzhen(-.+)?\.tencentcos\.(cn|com(\.cn)?)$  #深圳
+#regexp:\.(.+-)?ap-chongqing(-.+)?\.tencentcos\.(cn|com(\.cn)?)$ #重庆
+#regexp:\.(.+-)?ap-shenzhen(-.+)?\.tencentcos\.(cn|com(\.cn)?)$  #深圳
 
 # tencentcos outside mainland China
 # regexp:.+\.ap-hongkong\.tencentcos\.(cn|com(\.cn)?)$ @!cn       #中国香港

data/tilda

@@ -0,0 +1,5 @@
+tilda.cc
+tilda.ru
+tilda.ws
+tildaapi.com
+tildacdn.com

main.go

@@ -2,6 +2,7 @@ package main
 
 import (
 	"bufio"
+	"encoding/json"
 	"flag"
 	"fmt"
 	"os"
@@ -19,6 +20,7 @@ var (
 	dataPath    = flag.String("datapath", "./data", "Path to your custom 'data' directory")
 	outputName  = flag.String("outputname", "dlc.dat", "Name of the generated dat file")
 	outputDir   = flag.String("outputdir", "./", "Directory to place all generated files")
+	datProfile  = flag.String("datprofile", "", "Path of config file used to assemble custom dats")
 	exportLists = flag.String("exportlists", "", "Lists to be flattened and exported in plaintext format, separated by ',' comma")
 )
 
@@ -47,7 +49,24 @@ type Processor struct {
 	cirIncMap map[string]bool
 }
 
-func makeProtoList(listName string, entries []*Entry) (*router.GeoSite, error) {
+type GeoSites struct {
+	Sites   []*router.GeoSite
+	SiteIdx map[string]int
+}
+
+type DatTask struct {
+	Name  string   `json:"name"`
+	Mode  string   `json:"mode"`
+	Lists []string `json:"lists"`
+}
+
+const (
+	ModeAll       string = "all"
+	ModeAllowlist string = "allowlist"
+	ModeDenylist  string = "denylist"
+)
+
+func makeProtoList(listName string, entries []*Entry) *router.GeoSite {
 	site := &router.GeoSite{
 		CountryCode: listName,
 		Domain:      make([]*router.Domain, 0, len(entries)),
@@ -73,7 +92,91 @@ func makeProtoList(listName string, entries []*Entry) (*router.GeoSite, error) {
 		}
 		site.Domain = append(site.Domain, pdomain)
 	}
-	return site, nil
+	return site
+}
+
+func loadTasks(path string) ([]DatTask, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return nil, err
+	}
+	defer f.Close()
+
+	var tasks []DatTask
+	dec := json.NewDecoder(f)
+	if err := dec.Decode(&tasks); err != nil {
+		return nil, fmt.Errorf("failed to decode json: %w", err)
+	}
+	for i, t := range tasks {
+		if t.Name == "" {
+			return nil, fmt.Errorf("task[%d]: name is required", i)
+		}
+		switch t.Mode {
+		case ModeAll, ModeAllowlist, ModeDenylist:
+		default:
+			return nil, fmt.Errorf("task[%d] %q: invalid mode %q", i, t.Name, t.Mode)
+		}
+	}
+	return tasks, nil
+}
+
+func (gs *GeoSites) assembleDat(task DatTask) error {
+	datFileName := strings.ToLower(filepath.Base(task.Name))
+	geoSiteList := new(router.GeoSiteList)
+
+	switch task.Mode {
+	case ModeAll:
+		geoSiteList.Entry = gs.Sites
+	case ModeAllowlist:
+		allowedIdxes := make([]int, 0, len(task.Lists))
+		for _, list := range task.Lists {
+			if idx, ok := gs.SiteIdx[strings.ToUpper(list)]; ok {
+				allowedIdxes = append(allowedIdxes, idx)
+			} else {
+				return fmt.Errorf("list %q not found for allowlist task", list)
+			}
+		}
+		slices.Sort(allowedIdxes)
+		allowedlen := len(allowedIdxes)
+		if allowedlen == 0 {
+			return fmt.Errorf("allowlist needs at least one valid list")
+		}
+		geoSiteList.Entry = make([]*router.GeoSite, allowedlen)
+		for i, idx := range allowedIdxes {
+			geoSiteList.Entry[i] = gs.Sites[idx]
+		}
+	case ModeDenylist:
+		deniedMap := make(map[int]bool, len(task.Lists))
+		for _, list := range task.Lists {
+			if idx, ok := gs.SiteIdx[strings.ToUpper(list)]; ok {
+				deniedMap[idx] = true
+			} else {
+				fmt.Printf("[Warn] list %q not found in denylist task %q", list, task.Name)
+			}
+		}
+		deniedlen := len(deniedMap)
+		if deniedlen == 0 {
+			fmt.Printf("[Warn] nothing to deny in task %q", task.Name)
+			geoSiteList.Entry = gs.Sites
+		} else {
+			geoSiteList.Entry = make([]*router.GeoSite, 0, len(gs.Sites)-deniedlen)
+			for i, site := range gs.Sites {
+				if !deniedMap[i] {
+					geoSiteList.Entry = append(geoSiteList.Entry, site)
+				}
+			}
+		}
+	}
+
+	protoBytes, err := proto.Marshal(geoSiteList)
+	if err != nil {
+		return fmt.Errorf("failed to marshal: %w", err)
+	}
+	if err := os.WriteFile(filepath.Join(*outputDir, datFileName), protoBytes, 0644); err != nil {
+		return fmt.Errorf("failed to write file %q: %w", datFileName, err)
+	}
+	fmt.Printf("dat %q has been generated successfully\n", datFileName)
+	return nil
 }
 
 func writePlainList(listname string, entries []*Entry) error {
@@ -89,46 +192,28 @@ func writePlainList(listname string, entries []*Entry) error {
 	return w.Flush()
 }
 
-func parseEntry(line string) (*Entry, []string, error) {
+func parseEntry(typ, rule string) (*Entry, []string, error) {
-	entry := new(Entry)
+	entry := &Entry{Type: typ}
-	parts := strings.Fields(line)
+	parts := strings.Fields(rule)
 	if len(parts) == 0 {
-		return entry, nil, fmt.Errorf("empty line")
+		return entry, nil, fmt.Errorf("empty domain rule")
 	}
-
+	// Parse value
-	// Parse type and value
+	switch entry.Type {
-	typ, val, isTypeSpecified := strings.Cut(parts[0], ":")
+	case dlc.RuleTypeRegexp:
-	typ = strings.ToLower(typ)
+		if _, err := regexp.Compile(parts[0]); err != nil {
-	if !isTypeSpecified { // Default RuleType
+			return entry, nil, fmt.Errorf("invalid regexp %q: %w", parts[0], err)
-		if !validateDomainChars(typ) {
-			return entry, nil, fmt.Errorf("invalid domain: %q", typ)
 		}
-		entry.Type = dlc.RuleTypeDomain
+		entry.Value = parts[0]
-		entry.Value = typ
+	case dlc.RuleTypeDomain, dlc.RuleTypeFullDomain, dlc.RuleTypeKeyword:
-	} else {
+		entry.Value = strings.ToLower(parts[0])
-		switch typ {
+		if !validateDomainChars(entry.Value) {
-		case dlc.RuleTypeRegexp:
+			return entry, nil, fmt.Errorf("invalid domain: %q", entry.Value)
-			if _, err := regexp.Compile(val); err != nil {
-				return entry, nil, fmt.Errorf("invalid regexp %q: %w", val, err)
-			}
-			entry.Type = dlc.RuleTypeRegexp
-			entry.Value = val
-		case dlc.RuleTypeInclude:
-			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)
-			}
-		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)
-			}
-		default:
-			return entry, nil, fmt.Errorf("invalid type: %q", typ)
 		}
+	default:
+		return entry, nil, fmt.Errorf("unknown rule type: %q", entry.Type)
 	}
+	plen := len(entry.Type) + len(entry.Value) + 1
 
 	// Parse attributes and affiliations
 	var affs []string
@@ -140,6 +225,7 @@ func parseEntry(line string) (*Entry, []string, error) {
 				return entry, affs, fmt.Errorf("invalid attribute: %q", attr)
 			}
 			entry.Attrs = append(entry.Attrs, attr)
+			plen += 2 + len(attr)
 		case '&':
 			aff := strings.ToUpper(part[1:])
 			if !validateSiteName(aff) {
@@ -147,33 +233,70 @@ func parseEntry(line string) (*Entry, []string, error) {
 			}
 			affs = append(affs, aff)
 		default:
-			return entry, affs, fmt.Errorf("invalid attribute/affiliation: %q", part)
+			return entry, affs, fmt.Errorf("unknown field: %q", part)
 		}
 	}
 
-	if entry.Type != dlc.RuleTypeInclude {
+	slices.Sort(entry.Attrs) // Sort attributes
-		slices.Sort(entry.Attrs) // Sort attributes
+	// Formated plain entry: type:domain.tld:@attr1,@attr2
-		// Formated plain entry: type:domain.tld:@attr1,@attr2
+	var plain strings.Builder
-		var plain strings.Builder
+	plain.Grow(plen)
-		plain.Grow(len(entry.Type) + len(entry.Value) + 10)
+	plain.WriteString(entry.Type)
-		plain.WriteString(entry.Type)
+	plain.WriteByte(':')
-		plain.WriteByte(':')
+	plain.WriteString(entry.Value)
-		plain.WriteString(entry.Value)
+	for i, attr := range entry.Attrs {
-		for i, attr := range entry.Attrs {
+		if i == 0 {
-			if i == 0 {
+			plain.WriteByte(':')
-				plain.WriteByte(':')
+		} else {
-			} else {
+			plain.WriteByte(',')
-				plain.WriteByte(',')
-			}
-			plain.WriteByte('@')
-			plain.WriteString(attr)
 		}
-		entry.Plain = plain.String()
+		plain.WriteByte('@')
+		plain.WriteString(attr)
 	}
+	entry.Plain = plain.String()
 	return entry, affs, nil
 }
 
+func parseInclusion(rule string) (*Inclusion, error) {
+	parts := strings.Fields(rule)
+	if len(parts) == 0 {
+		return nil, fmt.Errorf("empty inclusion")
+	}
+	inc := &Inclusion{Source: strings.ToUpper(parts[0])}
+	if !validateSiteName(inc.Source) {
+		return inc, fmt.Errorf("invalid included list name: %q", inc.Source)
+	}
+
+	// Parse attributes
+	for _, part := range parts[1:] {
+		switch part[0] {
+		case '@':
+			attr := strings.ToLower(part[1:])
+			if attr[0] == '-' {
+				battr := attr[1:]
+				if !validateAttrChars(battr) {
+					return inc, fmt.Errorf("invalid ban attribute: %q", battr)
+				}
+				inc.BanAttrs = append(inc.BanAttrs, battr)
+			} else {
+				if !validateAttrChars(attr) {
+					return inc, fmt.Errorf("invalid must attribute: %q", attr)
+				}
+				inc.MustAttrs = append(inc.MustAttrs, attr)
+			}
+		case '&':
+			return inc, fmt.Errorf("affiliation is not allowed for inclusion")
+		default:
+			return inc, fmt.Errorf("unknown field: %q", part)
+		}
+	}
+	return inc, nil
+}
+
 func validateDomainChars(domain string) bool {
+	if domain == "" {
+		return false
+	}
 	for i := range domain {
 		c := domain[i]
 		if (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '.' || c == '-' {
@@ -185,9 +308,12 @@ func validateDomainChars(domain string) bool {
 }
 
 func validateAttrChars(attr string) bool {
+	if attr == "" {
+		return false
+	}
 	for i := range attr {
 		c := attr[i]
-		if (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '!' || c == '-' {
+		if (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '!' {
 			continue
 		}
 		return false
@@ -196,6 +322,9 @@ func validateAttrChars(attr string) bool {
 }
 
 func validateSiteName(name string) bool {
+	if name == "" {
+		return false
+	}
 	for i := range name {
 		c := name[i]
 		if (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || c == '!' || c == '-' {
@@ -232,26 +361,23 @@ func (p *Processor) loadData(listName string, path string) error {
 		if line == "" {
 			continue
 		}
-		entry, affs, err := parseEntry(line)
+		typ, rule, isTypeSpecified := strings.Cut(line, ":")
-		if err != nil {
+		if !isTypeSpecified { // Default RuleType
-			return fmt.Errorf("error in %q at line %d: %w", path, lineIdx, err)
+			typ, rule = dlc.RuleTypeDomain, typ
+		} else {
+			typ = strings.ToLower(typ)
 		}
-
+		if typ == dlc.RuleTypeInclude {
-		if entry.Type == dlc.RuleTypeInclude {
+			inc, err := parseInclusion(rule)
-			inc := &Inclusion{Source: entry.Value}
+			if err != nil {
-			for _, attr := range entry.Attrs {
+				return fmt.Errorf("error in %q at line %d: %w", path, lineIdx, err)
-				if attr[0] == '-' {
-					inc.BanAttrs = append(inc.BanAttrs, attr[1:])
-				} 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 {
+			entry, affs, err := parseEntry(typ, rule)
+			if err != nil {
+				return fmt.Errorf("error in %q at line %d: %w", path, lineIdx, err)
+			}
 			for _, aff := range affs {
 				apl := p.getOrCreateParsedList(aff)
 				apl.Entries = append(apl.Entries, entry)
@@ -259,7 +385,7 @@ func (p *Processor) loadData(listName string, path string) error {
 			pl.Entries = append(pl.Entries, entry)
 		}
 	}
-	return nil
+	return scanner.Err()
 }
 
 func isMatchAttrFilters(entry *Entry, incFilter *Inclusion) bool {
@@ -360,6 +486,9 @@ func (p *Processor) resolveList(plname string) error {
 			}
 		}
 	}
+	if len(roughMap) == 0 {
+		return fmt.Errorf("empty list")
+	}
 	p.finalMap[plname] = polishList(roughMap)
 	return nil
 }
@@ -387,13 +516,15 @@ func run() error {
 		return fmt.Errorf("failed to loadData: %w", err)
 	}
 	// Generate finalMap
-	processor.finalMap = make(map[string][]*Entry, len(processor.plMap))
+	sitesCount := len(processor.plMap)
+	processor.finalMap = make(map[string][]*Entry, sitesCount)
 	processor.cirIncMap = make(map[string]bool)
 	for plname := range processor.plMap {
 		if err := processor.resolveList(plname); err != nil {
 			return fmt.Errorf("failed to resolveList %q: %w", plname, err)
 		}
 	}
+	processor.plMap = nil
 
 	// Make sure output directory exists
 	if err := os.MkdirAll(*outputDir, 0755); err != nil {
@@ -403,47 +534,58 @@ func run() error {
 	for rawEpList := range strings.SplitSeq(*exportLists, ",") {
 		if epList := strings.TrimSpace(rawEpList); epList != "" {
 			entries, exist := processor.finalMap[strings.ToUpper(epList)]
-			if !exist || len(entries) == 0 {
+			if !exist {
-				fmt.Printf("list %q does not exist or is empty\n", epList)
+				fmt.Printf("[Warn] list %q does not exist\n", epList)
 				continue
 			}
 			if err := writePlainList(epList, entries); err != nil {
-				fmt.Printf("failed to write list %q: %v\n", epList, err)
+				fmt.Printf("[Error] failed to write list %q: %v\n", epList, err)
 				continue
 			}
-			fmt.Printf("list %q has been generated successfully.\n", epList)
+			fmt.Printf("list %q has been generated successfully\n", epList)
 		}
 	}
 
-	// Generate dat file
+	// Generate proto sites
-	protoList := new(router.GeoSiteList)
+	gs := &GeoSites{
-	for siteName, siteEntries := range processor.finalMap {
+		Sites:   make([]*router.GeoSite, 0, sitesCount),
-		site, err := makeProtoList(siteName, siteEntries)
+		SiteIdx: make(map[string]int, sitesCount),
-		if err != nil {
-			return fmt.Errorf("failed to makeProtoList %q: %w", siteName, err)
-		}
-		protoList.Entry = append(protoList.Entry, site)
 	}
-	// Sort protoList so the marshaled list is reproducible
+	for siteName, siteEntries := range processor.finalMap {
-	slices.SortFunc(protoList.Entry, func(a, b *router.GeoSite) int {
+		gs.Sites = append(gs.Sites, makeProtoList(siteName, siteEntries))
+	}
+	processor = nil
+	// Sort proto sites so the generated file is reproducible
+	slices.SortFunc(gs.Sites, func(a, b *router.GeoSite) int {
 		return strings.Compare(a.CountryCode, b.CountryCode)
 	})
+	for i := range sitesCount {
+		gs.SiteIdx[gs.Sites[i].CountryCode] = i
+	}
 
-	protoBytes, err := proto.Marshal(protoList)
+	// Load tasks and generate dat files
-	if err != nil {
+	var tasks []DatTask
-		return fmt.Errorf("failed to marshal: %w", err)
+	if *datProfile == "" {
+		tasks = []DatTask{{Name: *outputName, Mode: ModeAll}}
+	} else {
+		var err error
+		tasks, err = loadTasks(*datProfile)
+		if err != nil {
+			return fmt.Errorf("failed to loadTasks %q: %v", *datProfile, err)
+		}
 	}
-	if err := os.WriteFile(filepath.Join(*outputDir, *outputName), protoBytes, 0644); err != nil {
+	for _, task := range tasks {
-		return fmt.Errorf("failed to write output: %w", err)
+		if err := gs.assembleDat(task); err != nil {
+			fmt.Printf("[Error] failed to assembleDat %q: %v", task.Name, err)
+		}
 	}
-	fmt.Printf("%q has been generated successfully.\n", *outputName)
 	return nil
 }
 
 func main() {
 	flag.Parse()
 	if err := run(); err != nil {
-		fmt.Printf("Fatal error: %v\n", err)
+		fmt.Printf("[Fatal] critical error: %v\n", err)
 		os.Exit(1)
 	}
 }