Checkpoint

dataset/parser/hosts.go

@@ -0,0 +1,41 @@
+package parser
+
+import (
+	"bufio"
+	"io"
+	"net"
+	"strings"
+)
+
+func init() {
+	RegisterDomainsParser(hostsParser{})
+}
+
+type hostsParser struct{}
+
+func (hostsParser) CanHandle(line string) bool {
+	part := strings.Fields(line)
+	return len(part) >= 2 && net.ParseIP(part[0]) != nil
+}
+
+func (hostsParser) ParseDomains(r io.Reader) (domains []string, ignored int, err error) {
+	scanner := bufio.NewScanner(r)
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		if isComment(line) {
+			continue
+		}
+
+		part := strings.Fields(line)
+		if len(part) >= 2 && net.ParseIP(part[0]) != nil {
+			domains = append(domains, part[1:]...)
+			continue
+		}
+
+		ignored++
+	}
+	if err = scanner.Err(); err != nil {
+		return
+	}
+	return unique(domains), ignored, nil
+}