Merge pull request #338 from mrd0ll4r/fix-parsing

Fix and optimize query parsing, make parsing errors static

bittorrent/params.go

@@ -5,6 +5,8 @@ import (
 	"net/url"
 	"strconv"
 	"strings"
+
+	log "github.com/Sirupsen/logrus"
 )
 
 // Params is used to fetch (optional) request parameters from an Announce.
@@ -39,6 +41,10 @@ var ErrKeyNotFound = errors.New("query: value for the provided key does not exis
 // with invalid length.
 var ErrInvalidInfohash = ClientError("provided invalid infohash")
 
+// ErrInvalidQueryEscape is returned when a query string contains invalid
+// escapes.
+var ErrInvalidQueryEscape = ClientError("invalid query escape")
+
 // QueryParams parses a URL Query and implements the Params interface with some
 // additional helpers.
 type QueryParams struct {
@@ -88,69 +94,55 @@ func ParseURLData(urlData string) (*QueryParams, error) {
 
 // parseQuery parses a URL query into QueryParams.
 // The query is expected to exclude the delimiting '?'.
-func parseQuery(rawQuery string) (*QueryParams, error) {
+func parseQuery(query string) (q *QueryParams, err error) {
-	var (
+	// This is basically url.parseQuery, but with a map[string]string
-		keyStart, keyEnd int
+	// instead of map[string][]string for the values.
-		valStart, valEnd int
+	q = &QueryParams{
+		query:      query,
+		infoHashes: nil,
+		params:     make(map[string]string),
+	}
 
-		onKey = true
+	for query != "" {
-
+		key := query
-		q = &QueryParams{
+		if i := strings.IndexAny(key, "&;"); i >= 0 {
-			query:      rawQuery,
+			key, query = key[:i], key[i+1:]
-			infoHashes: nil,
-			params:     make(map[string]string),
-		}
-	)
-
-	for i, length := 0, len(rawQuery); i < length; i++ {
-		separator := rawQuery[i] == '&' || rawQuery[i] == ';'
-		last := i == length-1
-
-		if separator || last {
-			if onKey && !last {
-				keyStart = i + 1
-				continue
-			}
-
-			if last && !separator && !onKey {
-				valEnd = i
-			}
-
-			keyStr, err := url.QueryUnescape(rawQuery[keyStart : keyEnd+1])
-			if err != nil {
-				return nil, err
-			}
-
-			var valStr string
-
-			if valEnd > 0 {
-				valStr, err = url.QueryUnescape(rawQuery[valStart : valEnd+1])
-				if err != nil {
-					return nil, err
-				}
-			}
-
-			if keyStr == "info_hash" {
-				if len(valStr) != 20 {
-					return nil, ErrInvalidInfohash
-				}
-				q.infoHashes = append(q.infoHashes, InfoHashFromString(valStr))
-			} else {
-				q.params[strings.ToLower(keyStr)] = valStr
-			}
-
-			valEnd = 0
-			onKey = true
-			keyStart = i + 1
-
-		} else if rawQuery[i] == '=' {
-			onKey = false
-			valStart = i + 1
-			valEnd = 0
-		} else if onKey {
-			keyEnd = i
 		} else {
-			valEnd = i
+			query = ""
+		}
+		if key == "" {
+			continue
+		}
+		value := ""
+		if i := strings.Index(key, "="); i >= 0 {
+			key, value = key[:i], key[i+1:]
+		}
+		key, err = url.QueryUnescape(key)
+		if err != nil {
+			// QueryUnescape returns an error like "invalid escape: '%x'".
+			// But frontends record these errors to prometheus, which generates
+			// a lot of time series.
+			// We log it here for debugging instead.
+			log.WithFields(log.Fields{"error": err}).Debug("failed to unescape query param key")
+			return nil, ErrInvalidQueryEscape
+		}
+		value, err = url.QueryUnescape(value)
+		if err != nil {
+			// QueryUnescape returns an error like "invalid escape: '%x'".
+			// But frontends record these errors to prometheus, which generates
+			// a lot of time series.
+			// We log it here for debugging instead.
+			log.WithFields(log.Fields{"error": err}).Debug("failed to unescape query param value")
+			return nil, ErrInvalidQueryEscape
+		}
+
+		if key == "info_hash" {
+			if len(value) != 20 {
+				return nil, ErrInvalidInfohash
+			}
+			q.infoHashes = append(q.infoHashes, InfoHashFromString(value))
+		} else {
+			q.params[strings.ToLower(key)] = value
 		}
 	}
 

bittorrent/params_test.go

@@ -27,6 +27,12 @@ var (
 	InvalidQueries = []string{
 		"/announce?" + "info_hash=%0%a",
 	}
+
+	// See https://github.com/chihaya/chihaya/issues/334.
+	shouldNotPanicQueries = []string{
+		"/annnounce?" + "info_hash=" + testPeerID + "&a",
+		"/annnounce?" + "info_hash=" + testPeerID + "&=b?",
+	}
 )
 
 func mapArrayEqual(boxed map[string][]string, unboxed map[string]string) bool {
@@ -84,26 +90,40 @@ func TestParseInvalidURLData(t *testing.T) {
 	}
 }
 
+func TestParseShouldNotPanicURLData(t *testing.T) {
+	for _, parseStr := range shouldNotPanicQueries {
+		ParseURLData(parseStr)
+	}
+}
+
 func BenchmarkParseQuery(b *testing.B) {
+	announceStrings := make([]string, 0)
+	for i := range ValidAnnounceArguments {
+		announceStrings = append(announceStrings, ValidAnnounceArguments[i].Encode())
+	}
+	b.ResetTimer()
 	for bCount := 0; bCount < b.N; bCount++ {
-		for parseIndex, parseStr := range ValidAnnounceArguments {
+		i := bCount % len(announceStrings)
-			parsedQueryObj, err := parseQuery(parseStr.Encode())
+		parsedQueryObj, err := parseQuery(announceStrings[i])
-			if err != nil {
+		if err != nil {
-				b.Error(err, parseIndex)
+			b.Error(err, i)
-				b.Log(parsedQueryObj)
+			b.Log(parsedQueryObj)
-			}
 		}
 	}
 }
 
 func BenchmarkURLParseQuery(b *testing.B) {
+	announceStrings := make([]string, 0)
+	for i := range ValidAnnounceArguments {
+		announceStrings = append(announceStrings, ValidAnnounceArguments[i].Encode())
+	}
+	b.ResetTimer()
 	for bCount := 0; bCount < b.N; bCount++ {
-		for parseIndex, parseStr := range ValidAnnounceArguments {
+		i := bCount % len(announceStrings)
-			parsedQueryObj, err := url.ParseQuery(parseStr.Encode())
+		parsedQueryObj, err := url.ParseQuery(announceStrings[i])
-			if err != nil {
+		if err != nil {
-				b.Error(err, parseIndex)
+			b.Error(err, i)
-				b.Log(parsedQueryObj)
+			b.Log(parsedQueryObj)
-			}
 		}
 	}
 }