From 4be69d35eb276266b846a8a01481894f35d244dd Mon Sep 17 00:00:00 2001
From: Ainar Garipov <a.garipov@adguard.com>
Date: Thu, 16 Dec 2021 15:58:54 +0300
Subject: [PATCH] Pull request: aghnet: do not turn bad etc/hosts entries into
 rules

Updates #3946.

Squashed commit of the following:

commit 5d632dc4c49325308570adbfbc0fe333528989b5
Author: Ainar Garipov <A.Garipov@AdGuard.COM>
Date:   Thu Dec 16 15:49:51 2021 +0300

    aghnet: imp code

commit 4da620ee625718f5cd7549277c483631f22b977b
Author: Ainar Garipov <A.Garipov@AdGuard.COM>
Date:   Thu Dec 16 15:33:39 2021 +0300

    aghnet: do not turn bad etc/hosts entries into rules
---
 internal/aghnet/hostscontainer.go  | 41 ++++++++++++++++++------------
 internal/aghnet/testdata/etc_hosts |  6 +++++
 2 files changed, 31 insertions(+), 16 deletions(-)

diff --git a/internal/aghnet/hostscontainer.go b/internal/aghnet/hostscontainer.go
index ca8631c2..e4c1678b 100644
--- a/internal/aghnet/hostscontainer.go
+++ b/internal/aghnet/hostscontainer.go
@@ -100,16 +100,16 @@ const hostsContainerPref = "hosts container"
 type HostsContainer struct {
 	// requestMatcher matches the requests and translates the rules.  It's
 	// embedded to implement MatchRequest and Translate for *HostsContainer.
+	//
+	// TODO(a.garipov, e.burkov): Consider fully merging into HostsContainer.
 	requestMatcher
 
-	// listID is the identifier for the list of generated rules.
-	listID int
-
 	// done is the channel to sign closing the container.
 	done chan struct{}
 
 	// updates is the channel for receiving updated hosts.
 	updates chan *netutil.IPMap
+
 	// last is the set of hosts that was cached within last detected change.
 	last *netutil.IPMap
 
@@ -118,8 +118,12 @@ type HostsContainer struct {
 
 	// w tracks the changes in specified files and directories.
 	w aghos.FSWatcher
+
 	// patterns stores specified paths in the fs.Glob-compatible form.
 	patterns []string
+
+	// listID is the identifier for the list of generated rules.
+	listID int
 }
 
 // ErrNoHostsPaths is returned when there are no valid paths to watch passed to
@@ -288,7 +292,7 @@ func (hp *hostsParser) parseFile(
 	s := bufio.NewScanner(r)
 	for s.Scan() {
 		ip, hosts := hp.parseLine(s.Text())
-		if ip == nil {
+		if ip == nil || len(hosts) == 0 {
 			continue
 		}
 
@@ -310,21 +314,26 @@ func (hp *hostsParser) parseLine(line string) (ip net.IP, hosts []string) {
 	}
 
 	for _, f := range fields[1:] {
-		switch hashIdx := strings.IndexByte(f, '#'); hashIdx {
-		case -1:
-			hosts = append(hosts, f)
-
-			continue
-		case 0:
-			// Go on.
-		default:
+		hashIdx := strings.IndexByte(f, '#')
+		if hashIdx == 0 {
+			// The rest of the fields are a part of the comment so return.
+			break
+		} else if hashIdx > 0 {
 			// Only a part of the field is a comment.
-			hosts = append(hosts, f[:hashIdx])
+			f = f[:hashIdx]
 		}
 
-		// The rest of the fields are a part of the comment so skip
-		// immediately.
-		break
+		// Make sure that invalid hosts aren't turned into rules.
+		//
+		// See https://github.com/AdguardTeam/AdGuardHome/issues/3946.
+		err := netutil.ValidateDomainName(f)
+		if err != nil {
+			log.Error("%s: host %q is invalid, ignoring", hostsContainerPref, f)
+
+			continue
+		}
+
+		hosts = append(hosts, f)
 	}
 
 	return ip, hosts
diff --git a/internal/aghnet/testdata/etc_hosts b/internal/aghnet/testdata/etc_hosts
index d3cabac2..15e07566 100644
--- a/internal/aghnet/testdata/etc_hosts
+++ b/internal/aghnet/testdata/etc_hosts
@@ -8,7 +8,13 @@
 # See https://github.com/AdguardTeam/AdGuardHome/issues/3846.
 1.0.0.2 a.whole lot.of aliases for.testing
 
+# See https://github.com/AdguardTeam/AdGuardHome/issues/3946.
+1.0.0.3 *
+1.0.0.4 *.com
+
 # Same for IPv6.
 ::1 simplehost
 ::  hello hello.world
 ::2 a.whole lot.of aliases for.testing
+::3 *
+::4 *.com