ncdumpzone: Add mode for a URL list.

This can be used as input for a YaCy crawl job.

ncdumpzone/ncdumpzone.go

@@ -11,6 +11,7 @@ import (
 	extratypes "github.com/hlandau/ncbtcjsontypes"
 	"github.com/namecoin/ncdns/namecoin"
 	"github.com/namecoin/ncdns/ncdomain"
+	"github.com/namecoin/ncdns/rrtourl"
 	"github.com/namecoin/ncdns/tlsoverridefirefox"
 	"github.com/namecoin/ncdns/util"
 )
@@ -29,6 +30,12 @@ func dumpRR(rr dns.RR, dest io.Writer, format string) error {
 			return err
 		}
 		fmt.Fprint(dest, result)
+	case "url-list":
+		result, err := rrtourl.URLsFromRR(rr)
+		if err != nil {
+			return err
+		}
+		fmt.Fprint(dest, result)
 	}
 
 	return nil
@@ -77,7 +84,8 @@ func dumpName(item *extratypes.NameFilterItem, conn namecoin.Conn,
 // Dump extracts all domain names from conn, formats them according to the
 // specified format, and writes the result to dest.
 func Dump(conn namecoin.Conn, dest io.Writer, format string) error {
-	if format != "zonefile" && format != "firefox-override" {
+	if format != "zonefile" && format != "firefox-override" &&
+		format != "url-list" {
 		return fmt.Errorf("Invalid \"format\" argument: %s", format)
 	}
 

ncdumpzone/ncdumpzone/ncdumpzone.go

@@ -23,7 +23,8 @@ var (
 		"Namecoin RPC password")
 	formatFlag = cflag.String(flagGroup, "format", "zonefile", "Output "+
 		"format.  \"zonefile\" = DNS zone file.  "+
-		"\"firefox-override\" = Firefox cert_override.txt format.")
+		"\"firefox-override\" = Firefox cert_override.txt format.  "+
+		"\"url-list\" = URL list.")
 )
 
 var conn namecoin.Conn

rrtourl/rrtourl.go

@@ -0,0 +1,41 @@
+package rrtourl
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/miekg/dns"
+	"github.com/namecoin/ncdns/util"
+)
+
+// URLsFromRR returns a list of URL's derived from rr, which is suitable for
+// passing to a search engine crawler like YaCy.  If no such list can be
+// derived, returns an empty string.
+func URLsFromRR(rr dns.RR) (string, error) {
+	header := rr.Header()
+	if header == nil {
+		return "", fmt.Errorf("Nil RR header")
+	}
+
+	hostFQDN := header.Name
+
+	// Remove things like "_443._tcp" in TLSA records
+	for strings.HasPrefix(hostFQDN, "_") {
+		_, hostFQDN = util.SplitDomainTail(hostFQDN)
+	}
+
+	// Remove the trailing period from FQDN's
+	host := strings.TrimSuffix(hostFQDN, ".")
+
+	// Remove wildcard subdomains (later we assume that they might be "www.")
+	host = strings.TrimPrefix(host, "*.")
+
+	return "http://" + host + "/" + "\n" +
+		"http://www." + host + "/" + "\n" +
+		"https://" + host + "/" + "\n" +
+		"https://www." + host + "/" + "\n" +
+		"ftp://" + host + "/" + "\n" +
+		"ftp://www." + host + "/" + "\n" +
+		"ftps://" + host + "/" + "\n" +
+		"ftps://www." + host + "/" + "\n", nil
+}