feat: improve LongestRepeatingSequence performance by reducing alloca…

…tions (#575)

Uses an optimization to keep only two arrays + other misc changes.

Benchmarks:

$ go test -bench=. -benchmem -memprofile=allocs
goos: darwin
goarch: amd64
pkg: github.com/projectdiscovery/utils/strings
cpu: VirtualApple @ 2.50GHz
BenchmarkLongestRepeatingSequence-10              171566              6612 ns/op           15584 B/op         61 allocs/op
BenchmarkLongestRepeatingSequenceNew-10              360710              3129 ns/op             704 B/op          2 allocs/op +++
PASS
ok      github.com/projectdiscovery/utils/strings       1.851s

strings/stringsutil.go

@@ -231,41 +231,48 @@ type LongestSequence struct {
 
 // LongestRepeatingSequence finds the longest repeating non-overlapping sequence in a string
 func LongestRepeatingSequence(s string) LongestSequence {
-	res := ""
-	resLength := 0
 	n := len(s)
-	lcsre := make([][]int, n+1)
-
-	for i := range lcsre {
-		lcsre[i] = make([]int, n+1)
+	if n == 0 {
+		return LongestSequence{}
 	}
 
-	idx := 0
+	// Use single row instead of full matrix
+	prev := make([]int, n+1)
+	curr := make([]int, n+1)
+
+	maxLen := 0
+	endPos := 0
+
 	for i := 1; i <= n; i++ {
 		for j := i + 1; j <= n; j++ {
-			if s[i-1] == s[j-1] && lcsre[i-1][j-1] < (j-i) {
-				lcsre[i][j] = lcsre[i-1][j-1] + 1
-				if lcsre[i][j] > resLength {
-					resLength = lcsre[i][j]
-					if i > idx {
-						idx = i
-					}
+			if s[i-1] == s[j-1] && prev[j-1] < (j-i) {
+				curr[j] = prev[j-1] + 1
+				if curr[j] > maxLen {
+					maxLen = curr[j]
+					endPos = i
 				}
 			} else {
-				lcsre[i][j] = 0
+				curr[j] = 0
 			}
 		}
+		prev, curr = curr, prev
+		for j := range curr {
+			curr[j] = 0
+		}
 	}
-	if resLength > 0 {
-		for i := idx - resLength + 1; i <= idx; i++ {
-			res += string(s[i-1])
+
+	var sequence string
+	if maxLen > 0 {
+		start := endPos - maxLen
+		if start >= 0 {
+			sequence = s[start:endPos]
 		}
 	}
-	resCount := 0
-	if res != "" {
-		resCount = strings.Count(s, res)
+
+	return LongestSequence{
+		Sequence: sequence,
+		Count:    strings.Count(s, sequence),
 	}
-	return LongestSequence{Sequence: res, Count: resCount}
 }
 
 // IsPrintable checks if the strings is made only of printable characters

strings/stringsutil_test.go

@@ -303,11 +303,20 @@ func TestLongestRepeatingSequence(t *testing.T) {
 		s        string
 		expected string
 	}{
-		{"abcdefg", ""},
-		{"abcabcabc", "abc"},
-		{"abcdefabcdef", "abcdef"},
-		{"abcdefgabcdefg", "abcdefg"},
-		{"abcabcdefdef", "abc"},
+		{s: "abcabca", expected: "abc"},
+		{s: "abcdefg", expected: ""},
+		{s: "abcabcabc", expected: "abc"},
+		{s: "abcdefabcdef", expected: "abcdef"},
+		{s: "abcdefgabcdefg", expected: "abcdefg"},
+		{s: "abcabcdefdef", expected: "abc"},
+
+		// edge cases
+		{s: "aaa", expected: "a"},
+		{s: "aaaa", expected: "aa"},
+		{s: "abababab", expected: "abab"},
+		{s: "test test test", expected: "test "},
+		{s: "AbcAbcAbc", expected: "Abc"},
+		{s: "!@#$!@#$", expected: "!@#$"},
 	}
 
 	for _, test := range tests {
@@ -467,3 +476,12 @@ func TestNormalizeWithOptions(t *testing.T) {
 		require.Equal(t, test.result, res)
 	}
 }
+
+func BenchmarkLongestRepeatingSequence(b *testing.B) {
+	s := "AAAAAAAAAAAAAAAAAAAABBBBBBBBBBBBBBBBBBBB" // 40 chars test string
+	b.ResetTimer()
+	b.ReportAllocs()
+	for i := 0; i < b.N; i++ {
+		LongestRepeatingSequence(s)
+	}
+}