Pull request: 3012 idna search

Merge in DNS/adguard-home from 3012-idna-search to master

Closes #3012.

Squashed commit of the following:

commit 6a9fbfe16860df5db5982a70cfbf040967b6e6ae
Author: Eugene Burkov <e.burkov@adguard.com>
Date:   Tue Jun 29 21:28:10 2021 +0300

    querylog: add todo

commit 31292ba1aeb9e91ff4f6abae7ffdf806a87cae66
Author: Eugene Burkov <e.burkov@adguard.com>
Date:   Tue Jun 29 21:21:46 2021 +0300

    querylog: imp docs, code

commit 35757f76837cb8034f6079a351d01aa4706bfea7
Author: Eugene Burkov <e.burkov@adguard.com>
Date:   Tue Jun 29 21:01:08 2021 +0300

    queerylog: fix idn case match

commit eecfc98b6449c5c7c5a23602e80e47002034bc25
Author: Eugene Burkov <e.burkov@adguard.com>
Date:   Tue Jun 29 20:32:00 2021 +0300

    querylog: imp code, docs

commit 8aa6242fe92a9c2daa674b976595b13be96b0cf7
Author: Eugene Burkov <e.burkov@adguard.com>
Date:   Tue Jun 29 20:00:54 2021 +0300

    querylog: sup idn search

CHANGELOG.md

@@ -40,6 +40,7 @@ and this project adheres to
 
 ### Changed
 
+- Query log search now supports internationalized domains ([#3012]).
 - Internationalized domains are now shown decoded in the query log with the
   original encoded version shown in request details. ([#3013]).
 - When /etc/hosts-type rules have several IPs for one host, all IPs are now
@@ -83,6 +84,7 @@ released by then.
 [#2443]: https://github.com/AdguardTeam/AdGuardHome/issues/2443
 [#2624]: https://github.com/AdguardTeam/AdGuardHome/issues/2624
 [#2763]: https://github.com/AdguardTeam/AdGuardHome/issues/2763
+[#3012]: https://github.com/AdguardTeam/AdGuardHome/issues/3012
 [#3013]: https://github.com/AdguardTeam/AdGuardHome/issues/3013
 [#3136]: https://github.com/AdguardTeam/AdGuardHome/issues/3136
 [#3162]: https://github.com/AdguardTeam/AdGuardHome/issues/3162

internal/querylog/http.go

@@ -6,11 +6,13 @@ import (
 	"net/http"
 	"net/url"
 	"strconv"
+	"strings"
 	"time"
 
 	"github.com/AdguardTeam/AdGuardHome/internal/aghstrings"
 	"github.com/AdguardTeam/golibs/jsonutil"
 	"github.com/AdguardTeam/golibs/log"
+	"golang.org/x/net/idna"
 )
 
 type qlogConfig struct {
@@ -127,25 +129,53 @@ func getDoubleQuotesEnclosedValue(s *string) bool {
 }
 
 // parseSearchCriterion parses a search criterion from the query parameter.
-func (l *queryLog) parseSearchCriterion(q url.Values, name string, ct criterionType) (ok bool, sc searchCriterion, err error) {
+func (l *queryLog) parseSearchCriterion(q url.Values, name string, ct criterionType) (
+	ok bool,
+	sc searchCriterion,
+	err error,
+) {
 	val := q.Get(name)
-	if len(val) == 0 {
-		return false, searchCriterion{}, nil
+	if val == "" {
+		return false, sc, nil
 	}
 
-	c := searchCriterion{
+	strict := getDoubleQuotesEnclosedValue(&val)
+
+	var asciiVal string
+	switch ct {
+	case ctTerm:
+		// Decode lowercased value from punycode to make EqualFold and
+		// friends work properly with IDNAs.
+		//
+		// TODO(e.burkov):  Make it work with parts of IDNAs somehow.
+		loweredVal := strings.ToLower(val)
+		if asciiVal, err = idna.ToASCII(loweredVal); err != nil {
+			log.Debug("can't convert %q to ascii: %s", val, err)
+		} else if asciiVal == loweredVal {
+			// Purge asciiVal to prevent checking the same value
+			// twice.
+			asciiVal = ""
+		}
+	case ctFilteringStatus:
+		if !aghstrings.InSlice(filteringStatusValues, val) {
+			return false, sc, fmt.Errorf("invalid value %s", val)
+		}
+	default:
+		return false, sc, fmt.Errorf(
+			"invalid criterion type %v: should be one of %v",
+			ct,
+			[]criterionType{ctTerm, ctFilteringStatus},
+		)
+	}
+
+	sc = searchCriterion{
 		criterionType: ct,
 		value:         val,
-	}
-	if getDoubleQuotesEnclosedValue(&c.value) {
-		c.strict = true
+		asciiVal:      asciiVal,
+		strict:        strict,
 	}
 
-	if ct == ctFilteringStatus && !aghstrings.InSlice(filteringStatusValues, c.value) {
-		return false, c, fmt.Errorf("invalid value %s", c.value)
-	}
-
-	return true, c, nil
+	return true, sc, nil
 }
 
 // parseSearchParams - parses "searchParams" from the HTTP request's query string
@@ -175,15 +205,19 @@ func (l *queryLog) parseSearchParams(r *http.Request) (p *searchParams, err erro
 		p.maxFileScanEntries = 0
 	}
 
-	paramNames := map[string]criterionType{
-		"search":          ctTerm,
-		"response_status": ctFilteringStatus,
-	}
-
-	for k, v := range paramNames {
+	for _, v := range []struct {
+		urlField string
+		ct       criterionType
+	}{{
+		urlField: "search",
+		ct:       ctTerm,
+	}, {
+		urlField: "response_status",
+		ct:       ctFilteringStatus,
+	}} {
 		var ok bool
 		var c searchCriterion
-		ok, c, err = l.parseSearchCriterion(q, k, v)
+		ok, c, err = l.parseSearchCriterion(q, v.urlField, v.ct)
 		if err != nil {
 			return nil, err
 		}

internal/querylog/searchcriterion.go

@@ -11,10 +11,9 @@ import (
 type criterionType int
 
 const (
-	// ctTerm is for searching by the domain name, the client's IP
-	// address, the client's ID or the client's name.
-	//
-	// TODO(e.burkov):  Make it support IDNA while #3012.
+	// ctTerm is for searching by the domain name, the client's IP address,
+	// the client's ID or the client's name.  The domain name search
+	// supports IDNAs.
 	ctTerm criterionType = iota
 	// ctFilteringStatus is for searching by the filtering status.
 	//
@@ -47,6 +46,7 @@ var filteringStatusValues = []string{
 // searchCriterion is a search criterion that is used to match a record.
 type searchCriterion struct {
 	value         string
+	asciiVal      string
 	criterionType criterionType
 	// strict, if true, means that the criterion must be applied to the
 	// whole value rather than the part of it.  That is, equality and not
@@ -54,14 +54,16 @@ type searchCriterion struct {
 	strict bool
 }
 
-func (c *searchCriterion) ctDomainOrClientCaseStrict(
+func ctDomainOrClientCaseStrict(
 	term string,
+	asciiTerm string,
 	clientID string,
 	name string,
 	host string,
 	ip string,
 ) (ok bool) {
 	return strings.EqualFold(host, term) ||
+		(asciiTerm != "" && strings.EqualFold(host, asciiTerm)) ||
 		strings.EqualFold(clientID, term) ||
 		strings.EqualFold(ip, term) ||
 		strings.EqualFold(name, term)
@@ -98,8 +100,9 @@ func containsFold(s, substr string) (ok bool) {
 	return false
 }
 
-func (c *searchCriterion) ctDomainOrClientCaseNonStrict(
+func ctDomainOrClientCaseNonStrict(
 	term string,
+	asciiTerm string,
 	clientID string,
 	name string,
 	host string,
@@ -107,6 +110,7 @@ func (c *searchCriterion) ctDomainOrClientCaseNonStrict(
 ) (ok bool) {
 	return containsFold(clientID, term) ||
 		containsFold(host, term) ||
+		(asciiTerm != "" && containsFold(host, asciiTerm)) ||
 		containsFold(ip, term) ||
 		containsFold(name, term)
 }
@@ -127,10 +131,24 @@ func (c *searchCriterion) quickMatch(line string, findClient quickMatchClientFun
 		}
 
 		if c.strict {
-			return c.ctDomainOrClientCaseStrict(c.value, clientID, name, host, ip)
+			return ctDomainOrClientCaseStrict(
+				c.value,
+				c.asciiVal,
+				clientID,
+				name,
+				host,
+				ip,
+			)
 		}
 
-		return c.ctDomainOrClientCaseNonStrict(c.value, clientID, name, host, ip)
+		return ctDomainOrClientCaseNonStrict(
+			c.value,
+			c.asciiVal,
+			clientID,
+			name,
+			host,
+			ip,
+		)
 	case ctFilteringStatus:
 		// Go on, as we currently don't do quick matches against
 		// filtering statuses.
@@ -162,12 +180,11 @@ func (c *searchCriterion) ctDomainOrClientCase(e *logEntry) bool {
 	}
 
 	ip := e.IP.String()
-	term := strings.ToLower(c.value)
 	if c.strict {
-		return c.ctDomainOrClientCaseStrict(term, clientID, name, host, ip)
+		return ctDomainOrClientCaseStrict(c.value, c.asciiVal, clientID, name, host, ip)
 	}
 
-	return c.ctDomainOrClientCaseNonStrict(term, clientID, name, host, ip)
+	return ctDomainOrClientCaseNonStrict(c.value, c.asciiVal, clientID, name, host, ip)
 }
 
 func (c *searchCriterion) ctFilteringStatusCase(res filtering.Result) bool {