优化WAF单词匹配性能

internal/utils/runes/runes.go

@@ -2,6 +2,40 @@
 
 package runes
 
+// ContainsAnyWordRunes 直接使用rune检查字符串是否包含任一单词
+func ContainsAnyWordRunes(s string, words [][]rune, isCaseInsensitive bool) bool {
+	var allRunes = []rune(s)
+	if len(allRunes) == 0 || len(words) == 0 {
+		return false
+	}
+
+	var lastRune rune  // last searching rune in s
+	var lastIndex = -2 // -2: not started, -1: not found, >=0: rune index
+	for _, wordRunes := range words {
+		if len(wordRunes) == 0 {
+			continue
+		}
+
+		if lastIndex > -2 && lastRune == wordRunes[0] {
+			if lastIndex >= 0 {
+				result, _ := ContainsWordRunes(allRunes[lastIndex:], wordRunes, isCaseInsensitive)
+				if result {
+					return true
+				}
+			}
+			continue
+		} else {
+			result, firstIndex := ContainsWordRunes(allRunes, wordRunes, isCaseInsensitive)
+			lastIndex = firstIndex
+			if result {
+				return true
+			}
+		}
+		lastRune = wordRunes[0]
+	}
+	return false
+}
+
 // ContainsAnyWord 检查字符串是否包含任一单词
 func ContainsAnyWord(s string, words []string, isCaseInsensitive bool) bool {
 	var allRunes = []rune(s)

internal/utils/runes/runes_test.go

@@ -27,6 +27,7 @@ func TestContainsAnyWord(t *testing.T) {
 	a.IsFalse(runes.ContainsAnyWord("How are you?", []string{"how", "ok"}, false))
 	a.IsTrue(runes.ContainsAnyWord("How are you?", []string{"how"}, true))
 	a.IsTrue(runes.ContainsAnyWord("How are you?", []string{"how", "ok"}, true))
+	a.IsTrue(runes.ContainsAnyWord("How-are you?", []string{"how", "ok"}, true))
 }
 
 func TestContainsAnyWord_Sort(t *testing.T) {
@@ -100,6 +101,12 @@ func BenchmarkContainsAnyWord(b *testing.B) {
 
 	var words = strings.Split("python\npycurl\nhttp-client\nhttpclient\napachebench\nnethttp\nhttp_request\njava\nperl\nruby\nscrapy\nphp\nrust", "\n")
 	sort.Strings(words)
+
+	var wordRunes = [][]rune{}
+	for _, word := range words {
+		wordRunes = append(wordRunes, []rune(word))
+	}
+
 	b.ResetTimer()
 
 	b.RunParallel(func(pb *testing.PB) {
@@ -109,6 +116,27 @@ func BenchmarkContainsAnyWord(b *testing.B) {
 	})
 }
 
+func BenchmarkContainsAnyWordRunes(b *testing.B) {
+	runtime.GOMAXPROCS(4)
+
+	var words = strings.Split("python\npycurl\nhttp-client\nhttpclient\napachebench\nnethttp\nhttp_request\njava\nperl\nruby\nscrapy\nphp\nrust", "\n")
+	sort.Strings(words)
+
+	var wordRunes = [][]rune{}
+	for _, word := range words {
+		wordRunes = append(wordRunes, []rune(word))
+	}
+
+	b.ResetTimer()
+
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			_ = runes.ContainsAnyWordRunes("Mozilla/5.0 (Macintosh; Intel Mac OS X 10_0_0) AppleWebKit/500.00 (KHTML, like Gecko) Chrome/100.0.0.0", wordRunes, true)
+		}
+	})
+}
+
+
 func BenchmarkContainsAnyWord_Regexp(b *testing.B) {
 	runtime.GOMAXPROCS(4)
 	var reg = regexp.MustCompile("(?i)" + strings.ReplaceAll("python\npycurl\nhttp-client\nhttpclient\napachebench\nnethttp\nhttp_request\njava\nperl\nruby\nscrapy\nphp\nrust", "\n", "|"))

internal/waf/rule.go

@@ -53,6 +53,7 @@ type Rule struct {
 
 	ipRangeListValue *values.IPRangeList
 	stringValues     []string
+	stringValueRunes [][]rune
 	ipList           *values.StringList
 
 	floatValue float64
@@ -97,6 +98,11 @@ func (this *Rule) Init() error {
 			if this.Operator == RuleOperatorContainsAnyWord || this.Operator == RuleOperatorContainsAllWords || this.Operator == RuleOperatorNotContainsAnyWord {
 				sort.Strings(this.stringValues)
 			}
+
+			this.stringValueRunes = [][]rune{}
+			for _, line := range this.stringValues {
+				this.stringValueRunes = append(this.stringValueRunes, []rune(line))
+			}
 		}
 	case RuleOperatorMatch:
 		var v = this.Value
@@ -567,11 +573,11 @@ func (this *Rule) Test(value any) bool {
 		}
 		return false
 	case RuleOperatorContainsAnyWord:
-		return runes.ContainsAnyWord(this.stringifyValue(value), this.stringValues, this.IsCaseInsensitive)
+		return runes.ContainsAnyWordRunes(this.stringifyValue(value), this.stringValueRunes, this.IsCaseInsensitive)
 	case RuleOperatorContainsAllWords:
 		return runes.ContainsAllWords(this.stringifyValue(value), this.stringValues, this.IsCaseInsensitive)
 	case RuleOperatorNotContainsAnyWord:
-		return !runes.ContainsAnyWord(this.stringifyValue(value), this.stringValues, this.IsCaseInsensitive)
+		return !runes.ContainsAnyWordRunes(this.stringifyValue(value), this.stringValueRunes, this.IsCaseInsensitive)
 	case RuleOperatorContainsSQLInjection:
 		if value == nil {
 			return false