Add cmd/datdump/main.go (#3213)

* Feat: add a new datdump tool

* Refactor: address code review comments

* Refactor: remove export all from main program

use datdump instead

* Refactor: allow spaces in exportlists

e.g. `--exportlists="lista, listb"`

* all: cleanup

* apply review suggestion

---------

Co-authored-by: database64128 <free122448@hotmail.com>

.github/workflows/build.yml

@@ -36,7 +36,8 @@ jobs:
       - name: Build dlc.dat and plain lists
         run: |
           cd code || exit 1
-          go run ./ --outputdir=../ --exportlists=_all_,category-ads-all,tld-cn,cn,tld-\!cn,geolocation-\!cn,apple,icloud
+          go run ./ --outputdir=../ --exportlists=category-ads-all,tld-cn,cn,tld-\!cn,geolocation-\!cn,apple,icloud
+          go run ./cmd/datdump/main.go --inputdata=../dlc.dat --outputdir=../ --exportlists=_all_
           cd ../ && rm -rf code
 
       - name: Generate dlc.dat sha256 hash

.gitignore

@@ -8,5 +8,5 @@
 dlc.dat
 
 # Exported plaintext lists.
-dlc.dat_plain.yml
+/*.yml
 /*.txt

cmd/datdump/main.go

@@ -0,0 +1,164 @@
+package main
+
+import (
+	"bufio"
+	"flag"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/v2fly/domain-list-community/internal/dlc"
+	router "github.com/v2fly/v2ray-core/v5/app/router/routercommon"
+	"google.golang.org/protobuf/proto"
+)
+
+var (
+	inputData   = flag.String("inputdata", "dlc.dat", "Name of the geosite dat file")
+	outputDir   = flag.String("outputdir", "./", "Directory to place all generated files")
+	exportLists = flag.String("exportlists", "", "Lists to be exported, separated by ',' (empty for _all_)")
+)
+
+type DomainRule struct {
+	Type  string
+	Value string
+	Attrs []string
+}
+
+type DomainList struct {
+	Name  string
+	Rules []DomainRule
+}
+
+func (d *DomainRule) domain2String() string {
+	dstring := d.Type + ":" + d.Value
+	if len(d.Attrs) != 0 {
+		dstring += ":@" + strings.Join(d.Attrs, ",@")
+	}
+	return dstring
+}
+
+func loadGeosite(path string) ([]DomainList, map[string]*DomainList, error) {
+	data, err := os.ReadFile(path)
+	if err != nil {
+		return nil, nil, fmt.Errorf("failed to read geosite file: %w", err)
+	}
+	vgeositeList := new(router.GeoSiteList)
+	if err := proto.Unmarshal(data, vgeositeList); err != nil {
+		return nil, nil, fmt.Errorf("failed to unmarshal: %w", err)
+	}
+	domainLists := make([]DomainList, len(vgeositeList.Entry))
+	domainListByName := make(map[string]*DomainList, len(vgeositeList.Entry))
+	for i, vsite := range vgeositeList.Entry {
+		rules := make([]DomainRule, 0, len(vsite.Domain))
+		for _, vdomain := range vsite.Domain {
+			rule := DomainRule{Value: vdomain.Value}
+			switch vdomain.Type {
+			case router.Domain_RootDomain:
+				rule.Type = dlc.RuleTypeDomain
+			case router.Domain_Regex:
+				rule.Type = dlc.RuleTypeRegexp
+			case router.Domain_Plain:
+				rule.Type = dlc.RuleTypeKeyword
+			case router.Domain_Full:
+				rule.Type = dlc.RuleTypeFullDomain
+			default:
+				return nil, nil, fmt.Errorf("invalid rule type: %+v", vdomain.Type)
+			}
+			for _, vattr := range vdomain.Attribute {
+				rule.Attrs = append(rule.Attrs, vattr.Key)
+			}
+			rules = append(rules, rule)
+		}
+		domainLists[i] = DomainList{
+			Name:  strings.ToUpper(vsite.CountryCode),
+			Rules: rules,
+		}
+		domainListByName[domainLists[i].Name] = &domainLists[i]
+	}
+	return domainLists, domainListByName, nil
+}
+
+func exportSite(name string, domainListByName map[string]*DomainList) error {
+	domainList, ok := domainListByName[strings.ToUpper(name)]
+	if !ok {
+		return fmt.Errorf("list '%s' does not exist", name)
+	}
+	if len(domainList.Rules) == 0 {
+		return fmt.Errorf("list '%s' is empty", name)
+	}
+	file, err := os.Create(filepath.Join(*outputDir, name+".yml"))
+	if err != nil {
+		return err
+	}
+	defer file.Close()
+	w := bufio.NewWriter(file)
+	fmt.Fprintf(w, "%s:\n", name)
+	for _, domain := range domainList.Rules {
+		fmt.Fprintf(w, "  - %q\n", domain.domain2String())
+	}
+	return w.Flush()
+}
+
+func exportAll(filename string, domainLists []DomainList) error {
+	file, err := os.Create(filepath.Join(*outputDir, filename))
+	if err != nil {
+		return err
+	}
+	defer file.Close()
+	w := bufio.NewWriter(file)
+	w.WriteString("lists:\n")
+	for _, domainList := range domainLists {
+		fmt.Fprintf(w, "  - name: %s\n", strings.ToLower(domainList.Name))
+		fmt.Fprintf(w, "    length: %d\n", len(domainList.Rules))
+		w.WriteString("    rules:\n")
+		for _, domain := range domainList.Rules {
+			fmt.Fprintf(w, "      - %q\n", domain.domain2String())
+		}
+	}
+	return w.Flush()
+}
+
+func main() {
+	flag.Parse()
+
+	// Create output directory if not exist
+	if _, err := os.Stat(*outputDir); os.IsNotExist(err) {
+		if mkErr := os.MkdirAll(*outputDir, 0755); mkErr != nil {
+			fmt.Println("Failed to create output directory:", mkErr)
+			os.Exit(1)
+		}
+	}
+
+	fmt.Printf("Loading %s...\n", *inputData)
+	domainLists, domainListByName, err := loadGeosite(*inputData)
+	if err != nil {
+		fmt.Println("Failed to loadGeosite:", err)
+		os.Exit(1)
+	}
+
+	var exportListSlice []string
+	for raw := range strings.SplitSeq(*exportLists, ",") {
+		if trimmed := strings.TrimSpace(raw); trimmed != "" {
+			exportListSlice = append(exportListSlice, trimmed)
+		}
+	}
+	if len(exportListSlice) == 0 {
+		exportListSlice = []string{"_all_"}
+	}
+
+	for _, eplistname := range exportListSlice {
+		if strings.EqualFold(eplistname, "_all_") {
+			if err := exportAll(filepath.Base(*inputData)+"_plain.yml", domainLists); err != nil {
+				fmt.Println("Failed to exportAll:", err)
+				continue
+			}
+		} else {
+			if err := exportSite(eplistname, domainListByName); err != nil {
+				fmt.Println("Failed to exportSite:", err)
+				continue
+			}
+		}
+		fmt.Printf("list: '%s' has been exported successfully.\n", eplistname)
+	}
+}

internal/dlc/dlc.go

@@ -0,0 +1,9 @@
+package dlc
+
+const (
+	RuleTypeDomain     string = "domain"
+	RuleTypeFullDomain string = "full"
+	RuleTypeKeyword    string = "keyword"
+	RuleTypeRegexp     string = "regexp"
+	RuleTypeInclude    string = "include"
+)

main.go

@@ -10,6 +10,7 @@ import (
 	"slices"
 	"strings"
 
+	"github.com/v2fly/domain-list-community/internal/dlc"
 	router "github.com/v2fly/v2ray-core/v5/app/router/routercommon"
 	"google.golang.org/protobuf/proto"
 )
@@ -21,21 +22,6 @@ var (
 	exportLists = flag.String("exportlists", "", "Lists to be flattened and exported in plaintext format, separated by ',' comma")
 )
 
-const (
-	RuleTypeDomain     string = "domain"
-	RuleTypeFullDomain string = "full"
-	RuleTypeKeyword    string = "keyword"
-	RuleTypeRegexp     string = "regexp"
-	RuleTypeInclude    string = "include"
-)
-
-var (
-	TypeChecker   = regexp.MustCompile(`^(domain|full|keyword|regexp|include)$`)
-	DomainChecker = regexp.MustCompile(`^[a-z0-9\.-]+$`)
-	AttrChecker   = regexp.MustCompile(`^[a-z0-9!-]+$`)
-	SiteChecker   = regexp.MustCompile(`^[A-Z0-9!-]+$`)
-)
-
 var (
 	refMap    = make(map[string][]*Entry)
 	plMap     = make(map[string]*ParsedList)
@@ -78,13 +64,13 @@ func makeProtoList(listName string, entries []*Entry) (*router.GeoSite, error) {
 		}
 
 		switch entry.Type {
-		case RuleTypeDomain:
+		case dlc.RuleTypeDomain:
 			pdomain.Type = router.Domain_RootDomain
-		case RuleTypeRegexp:
+		case dlc.RuleTypeRegexp:
 			pdomain.Type = router.Domain_Regex
-		case RuleTypeKeyword:
+		case dlc.RuleTypeKeyword:
 			pdomain.Type = router.Domain_Plain
-		case RuleTypeFullDomain:
+		case dlc.RuleTypeFullDomain:
 			pdomain.Type = router.Domain_Full
 		}
 		site.Domain = append(site.Domain, pdomain)
@@ -92,29 +78,10 @@ func makeProtoList(listName string, entries []*Entry) (*router.GeoSite, error) {
 	return site, nil
 }
 
-func writePlainAll(siteList *[]string) error {
-	file, err := os.Create(filepath.Join(*outputDir, *outputName + "_plain.yml"))
-	if err != nil {
-		return err
-	}
-	defer file.Close()
-	w := bufio.NewWriter(file)
-	w.WriteString("lists:\n")
-	for _, site := range *siteList {
-		fmt.Fprintf(w, "  - name: %s\n", strings.ToLower(site))
-		fmt.Fprintf(w, "    length: %d\n", len(finalMap[site]))
-		w.WriteString("    rules:\n")
-		for _, entry := range finalMap[site] {
-			fmt.Fprintf(w, "      - %s\n", entry.Plain)
-		}
-	}
-	return w.Flush()
-}
-
 func writePlainList(exportedName string) error {
 	targetList, exist := finalMap[strings.ToUpper(exportedName)]
 	if !exist || len(targetList) == 0 {
-		return fmt.Errorf("'%s' list does not exist or is empty.", exportedName)
+		return fmt.Errorf("list %q does not exist or is empty.", exportedName)
 	}
 	file, err := os.Create(filepath.Join(*outputDir, strings.ToLower(exportedName)+".txt"))
 	if err != nil {
@@ -131,41 +98,43 @@ func writePlainList(exportedName string) error {
 func parseEntry(line string) (Entry, error) {
 	var entry Entry
 	parts := strings.Fields(line)
+	if len(parts) == 0 {
+		return entry, fmt.Errorf("empty line: %q", line)
+	}
 
 	// Parse type and value
-	rawTypeVal := parts[0]
-	kv := strings.Split(rawTypeVal, ":")
-	if len(kv) == 1 {
-		entry.Type = RuleTypeDomain // Default type
-		entry.Value = strings.ToLower(rawTypeVal)
-	} else if len(kv) == 2 {
-		entry.Type = strings.ToLower(kv[0])
-		if entry.Type == RuleTypeRegexp {
-			entry.Value = kv[1]
-		} else if entry.Type == RuleTypeInclude {
-			entry.Value = strings.ToUpper(kv[1])
-		} else {
-			entry.Value = strings.ToLower(kv[1])
+	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)
 		}
 	} else {
-		return entry, fmt.Errorf("invalid format: %s", line)
+		typ := strings.ToLower(v[:colonIndex])
+		val := v[colonIndex+1:]
+		switch typ {
+		case dlc.RuleTypeRegexp:
+			if _, err := regexp.Compile(val); err != nil {
+				return entry, fmt.Errorf("invalid regexp %q: %w", val, err)
 			}
-	// Check type and value
-	if !TypeChecker.MatchString(entry.Type) {
-		return entry, fmt.Errorf("invalid type: %s", entry.Type)
+			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, fmt.Errorf("invalid include list name: %q", entry.Value)
 			}
-	switch entry.Type {
-	case RuleTypeRegexp:
-		if _, err := regexp.Compile(entry.Value); err != nil {
-			return entry, fmt.Errorf("invalid regexp: %s", entry.Value)
+		case dlc.RuleTypeDomain, dlc.RuleTypeFullDomain, dlc.RuleTypeKeyword:
+			entry.Type = typ
+			entry.Value = strings.ToLower(val)
+			if !validateDomainChars(entry.Value) {
+				return entry, fmt.Errorf("invalid domain: %q", entry.Value)
 			}
-	case RuleTypeInclude:
-		if !SiteChecker.MatchString(entry.Value) {
-			return entry, fmt.Errorf("invalid included list name: %s", entry.Value)
-		}
-	default: // `full`, `domain` and `keyword` are all (parts of) domains
-		if !DomainChecker.MatchString(entry.Value) {
-			return entry, fmt.Errorf("invalid domain: %s", entry.Value)
+		default:
+			return entry, fmt.Errorf("invalid type: %q", typ)
 		}
 	}
 
@@ -173,18 +142,18 @@ func parseEntry(line string) (Entry, error) {
 	for _, part := range parts[1:] {
 		if strings.HasPrefix(part, "@") {
 			attr := strings.ToLower(part[1:]) // Trim attribute prefix `@` character
-			if !AttrChecker.MatchString(attr) {
-				return entry, fmt.Errorf("invalid attribute key: %s", attr)
+			if !validateAttrChars(attr) {
+				return entry, fmt.Errorf("invalid attribute: %q", attr)
 			}
 			entry.Attrs = append(entry.Attrs, attr)
 		} else if strings.HasPrefix(part, "&") {
 			aff := strings.ToUpper(part[1:]) // Trim affiliation prefix `&` character
-			if !SiteChecker.MatchString(aff) {
-				return entry, fmt.Errorf("invalid affiliation key: %s", aff)
+			if !validateSiteName(aff) {
+				return entry, fmt.Errorf("invalid affiliation: %q", aff)
 			}
 			entry.Affs = append(entry.Affs, aff)
 		} else {
-			return entry, fmt.Errorf("invalid attribute/affiliation: %s", part)
+			return entry, fmt.Errorf("invalid attribute/affiliation: %q", part)
 		}
 	}
 	// Sort attributes
@@ -198,6 +167,39 @@ func parseEntry(line string) (Entry, error) {
 	return entry, nil
 }
 
+func validateDomainChars(domain string) bool {
+	for i := range domain {
+		c := domain[i]
+		if (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '.' || c == '-' {
+			continue
+		}
+		return false
+	}
+	return true
+}
+
+func validateAttrChars(attr string) bool {
+	for i := range attr {
+		c := attr[i]
+		if (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '!' || c == '-' {
+			continue
+		}
+		return false
+	}
+	return true
+}
+
+func validateSiteName(name string) bool {
+	for i := range name {
+		c := name[i]
+		if (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || c == '!' || c == '-' {
+			continue
+		}
+		return false
+	}
+	return true
+}
+
 func loadData(path string) error {
 	file, err := os.Open(path)
 	if err != nil {
@@ -206,7 +208,7 @@ func loadData(path string) error {
 	defer file.Close()
 
 	listName := strings.ToUpper(filepath.Base(path))
-	if !SiteChecker.MatchString(listName) {
+	if !validateSiteName(listName) {
 		return fmt.Errorf("invalid list name: %s", listName)
 	}
 	scanner := bufio.NewScanner(file)
@@ -238,7 +240,7 @@ func parseList(refName string, refList []*Entry) error {
 		plMap[refName] = pl
 	}
 	for _, entry := range refList {
-		if entry.Type == RuleTypeInclude {
+		if entry.Type == dlc.RuleTypeInclude {
 			if len(entry.Affs) != 0 {
 				return fmt.Errorf("affiliation is not allowed for include:%s", entry.Value)
 			}
@@ -272,18 +274,18 @@ 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 RuleTypeRegexp:
+		case dlc.RuleTypeRegexp:
 			finalList = append(finalList, entry)
-		case RuleTypeKeyword:
+		case dlc.RuleTypeKeyword:
 			finalList = append(finalList, entry)
-		case RuleTypeDomain:
+		case dlc.RuleTypeDomain:
 			domainsMap[entry.Value] = true
 			if len(entry.Attrs) != 0 {
 				finalList = append(finalList, entry)
 			} else {
 				queuingList = append(queuingList, entry)
 			}
-		case RuleTypeFullDomain:
+		case dlc.RuleTypeFullDomain:
 			if len(entry.Attrs) != 0 {
 				finalList = append(finalList, entry)
 			} else {
@@ -295,14 +297,18 @@ func polishList(roughMap *map[string]*Entry) []*Entry {
 	for _, qentry := range queuingList {
 		isRedundant := false
 		pd := qentry.Value // To be parent domain
-		if qentry.Type == RuleTypeFullDomain {
+		if qentry.Type == dlc.RuleTypeFullDomain {
 			pd = "." + pd // So that `domain:example.org` overrides `full:example.org`
 		}
 		for {
 			idx := strings.Index(pd, ".")
-			if idx == -1 { break }
+			if idx == -1 {
+				break
+			}
 			pd = pd[idx+1:] // Go for next parent
-			if !strings.Contains(pd, ".") { break } // Not allow tld to be a parent
+			if !strings.Contains(pd, ".") {
+				break
+			} // Not allow tld to be a parent
 			if domainsMap[pd] {
 				isRedundant = true
 				break
@@ -320,7 +326,9 @@ func polishList(roughMap *map[string]*Entry) []*Entry {
 }
 
 func resolveList(pl *ParsedList) error {
-	if _, pldone := finalMap[pl.Name]; pldone { return nil }
+	if _, pldone := finalMap[pl.Name]; pldone {
+		return nil
+	}
 
 	if cirIncMap[pl.Name] {
 		return fmt.Errorf("circular inclusion in: %s", pl.Name)
@@ -329,14 +337,22 @@ func resolveList(pl *ParsedList) error {
 	defer delete(cirIncMap, pl.Name)
 
 	isMatchAttrFilters := func(entry *Entry, incFilter *Inclusion) bool {
-		if len(incFilter.MustAttrs) == 0 && len(incFilter.BanAttrs) == 0 { return true }
-		if len(entry.Attrs) == 0 { return len(incFilter.MustAttrs) == 0 }
+		if len(incFilter.MustAttrs) == 0 && len(incFilter.BanAttrs) == 0 {
+			return true
+		}
+		if len(entry.Attrs) == 0 {
+			return len(incFilter.MustAttrs) == 0
+		}
 
 		for _, m := range incFilter.MustAttrs {
-			if !slices.Contains(entry.Attrs, m) { return false }
+			if !slices.Contains(entry.Attrs, m) {
+				return false
+			}
 		}
 		for _, b := range incFilter.BanAttrs {
-			if slices.Contains(entry.Attrs, b) { return false }
+			if slices.Contains(entry.Attrs, b) {
+				return false
+			}
 		}
 		return true
 	}
@@ -348,7 +364,7 @@ func resolveList(pl *ParsedList) error {
 	for _, inc := range pl.Inclusions {
 		incPl, exist := plMap[inc.Source]
 		if !exist {
-			return fmt.Errorf("list '%s' includes a non-existent list: '%s'", pl.Name, inc.Source)
+			return fmt.Errorf("list %q includes a non-existent list: %q", pl.Name, inc.Source)
 		}
 		if err := resolveList(incPl); err != nil {
 			return err
@@ -395,54 +411,51 @@ func main() {
 		}
 	}
 
-	// Generate finalMap and sorted list of site names
-	siteList := make([]string, 0 ,len(plMap))
+	// Generate finalMap
 	for _, pl := range plMap {
-		siteList = append(siteList, pl.Name)
 		if err := resolveList(pl); err != nil {
 			fmt.Println("Failed to resolveList:", err)
 			os.Exit(1)
 		}
 	}
-	slices.Sort(siteList)
 
 	// Create output directory if not exist
 	if _, err := os.Stat(*outputDir); os.IsNotExist(err) {
 		if mkErr := os.MkdirAll(*outputDir, 0755); mkErr != nil {
-			fmt.Println("Failed:", mkErr)
+			fmt.Println("Failed to create output directory:", mkErr)
 			os.Exit(1)
 		}
 	}
 
 	// Export plaintext list
-	if *exportLists != "" {
-		exportedListSlice := strings.Split(*exportLists, ",")
-		for _, exportedList := range exportedListSlice {
-			if exportedList == "_all_" {
-				if err := writePlainAll(&siteList); err != nil {
-					fmt.Println("Failed to writePlainAll:", err)
-					continue
+	var exportListSlice []string
+	for raw := range strings.SplitSeq(*exportLists, ",") {
+		if trimmed := strings.TrimSpace(raw); trimmed != "" {
+			exportListSlice = append(exportListSlice, trimmed)
 		}
-			} else {
-				if err := writePlainList(exportedList); err != nil {
+	}
+	for _, exportList := range exportListSlice {
+		if err := writePlainList(exportList); err != nil {
 			fmt.Println("Failed to write list:", err)
 			continue
 		}
-			}
-			fmt.Printf("list: '%s' has been generated successfully.\n", exportedList)
-		}
+		fmt.Printf("list %q has been generated successfully.\n", exportList)
 	}
 
 	// Generate dat file
 	protoList := new(router.GeoSiteList)
-	for _, siteName := range siteList { // So that protoList.Entry is sorted
-		site, err := makeProtoList(siteName, finalMap[siteName])
+	for siteName, siteEntries := range finalMap {
+		site, err := makeProtoList(siteName, siteEntries)
 		if err != nil {
 			fmt.Println("Failed to makeProtoList:", err)
 			os.Exit(1)
 		}
 		protoList.Entry = append(protoList.Entry, site)
 	}
+	// Sort protoList so the marshaled list is reproducible
+	slices.SortFunc(protoList.Entry, func(a, b *router.GeoSite) int {
+		return strings.Compare(a.CountryCode, b.CountryCode)
+	})
 
 	protoBytes, err := proto.Marshal(protoList)
 	if err != nil {