Regexp Utils

[Mzack9999]

Used in projectdiscovery/wappalyzergo#135

[dwisiswant0]

afaik, using the combination of regexp2 and std regexp pkg should be sufficient, since regexep2 already support Perl5 (lookarounds & backreferences). so there’s really no need for the go-re2 engine. also it seems like the current detectEngine implementation isn’t comprehensive enough to handle non-std-regexp features properly.

• https://github.com/dlclark/regexp2#catastrophic-backtracking-and-timeouts

tip: i actually built a package called pcregexp that supports lookarounds & backreferences while aiming for compatibility with the std regexp, it uses libpcre2 API bindings under the hood with purego (so there’s no Cgo). it also has a wrapper that automatically selects the appropriate engine based on the regex features being used. it’s NOT fully stable yet, but i’m actively looking for adopters to support its continuous development and testing.

• https://github.com/dwisiswant0/pcregexp
• https://pkg.go.dev/github.com/dwisiswant0/pcregexp/pkg/regexp

go test -benchmem -run=^$ -bench . benchmark
goos: linux
goarch: amd64
pkg: benchmark
cpu: 11th Gen Intel(R) Core(TM) i9-11900H @ 2.50GHz
BenchmarkMatch/pcregexp/simple-16         	77527322	        15.09 ns/op	       0 B/op	       0 allocs/op
BenchmarkMatch/dlclark-regexp2/simple-16  	 2509765	       496.2 ns/op	      80 B/op	       1 allocs/op
BenchmarkMatch/AspieSoft-regex/simple-16  	 2244966	       552.1 ns/op	      28 B/op	       2 allocs/op
BenchmarkMatch/scorpionknifes-pcre/simple-16         	 4953229	       234.0 ns/op	       0 B/op	       0 allocs/op
BenchmarkMatch/GRbit-pcre/simple-16                  	 2469645	       522.0 ns/op	      28 B/op	       2 allocs/op
BenchmarkMatch/wasilibs-re2/simple-16                	 1229862	       961.8 ns/op	     248 B/op	       7 allocs/op
BenchmarkMatch/pcregexp/email-16                     	81896185	        14.94 ns/op	       0 B/op	       0 allocs/op
BenchmarkMatch/dlclark-regexp2/email-16              	  578250	      1836 ns/op	      64 B/op	       1 allocs/op
BenchmarkMatch/AspieSoft-regex/email-16              	 2268996	       516.1 ns/op	      16 B/op	       2 allocs/op
BenchmarkMatch/scorpionknifes-pcre/email-16          	 4037431	       287.0 ns/op	       0 B/op	       0 allocs/op
BenchmarkMatch/GRbit-pcre/email-16                   	 2544838	       489.9 ns/op	      16 B/op	       2 allocs/op
BenchmarkMatch/wasilibs-re2/email-16                 	 1000000	      1082 ns/op	     240 B/op	       7 allocs/op
BenchmarkMatch/pcregexp/backreference-16             	76438894	        15.10 ns/op	       0 B/op	       0 allocs/op
BenchmarkMatch/dlclark-regexp2/backreference-16      	 1294224	      1021 ns/op	      80 B/op	       1 allocs/op
BenchmarkMatch/AspieSoft-regex/backreference-16      	 2315284	       486.8 ns/op	      28 B/op	       2 allocs/op
BenchmarkMatch/scorpionknifes-pcre/backreference-16  	 5510665	       226.4 ns/op	       0 B/op	       0 allocs/op
BenchmarkMatch/GRbit-pcre/backreference-16           	 2588995	       460.9 ns/op	      28 B/op	       2 allocs/op
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1747392575.123307  145936 re2.cc:237] Error parsing '(\w+)\s+\1': invalid escape sequence: \1
PASS
ok  	benchmark	27.466s

package benchmark_test

import (
	"testing"

	AspieSoft "github.com/AspieSoft/go-regex/v8"
	GRbit "github.com/GRbit/go-pcre"
	dlclark "github.com/dlclark/regexp2"
	scorpionknifes "github.com/scorpionknifes/go-pcre"
	wasilibs "github.com/wasilibs/go-re2"

	"github.com/dwisiswant0/pcregexp"
)

func BenchmarkMatch(b *testing.B) {
	tests := []struct {
		name    string
		pattern string
		text    []byte
	}{
		{"simple", `p([a-z]+)ch`, []byte("peach punch pinch")},
		{"email", `\b\w+@\w+\.\w+\b`, []byte("[email protected]")},
		{"backreference", `(\w+)\s+\1`, []byte("hello hello world")},
		{"lookaround", `(?<=foo)bar`, []byte("foobar")},
	}

	for _, tt := range tests {
		r1 := pcregexp.MustCompile(tt.pattern)

		b.ResetTimer()
		b.Run("pcregexp/"+tt.name, func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				r1.Match(tt.text)
			}
		})

		r2 := dlclark.MustCompile(tt.pattern, 0)

		b.ResetTimer()
		b.Run("dlclark-regexp2/"+tt.name, func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				r2.MatchString(string(tt.text))
			}
		})

		r3, err := AspieSoft.CompTry(tt.pattern)
		if err != nil {
			b.Fatalf("r3: failed to compile pattern: %v", err)
		}

		b.ResetTimer()
		b.Run("AspieSoft-regex/"+tt.name, func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				r3.Match(tt.text)
			}
		})

		r4, err := scorpionknifes.Compile(tt.pattern, 0)
		if err != nil {
			b.Fatalf("r4: failed to compile pattern: %v", err)
		}
		r4Matcher := r4.NewMatcher()

		b.ResetTimer()
		b.Run("scorpionknifes-pcre/"+tt.name, func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				r4Matcher.Match(tt.text, 0)
			}
		})

		r5, err := GRbit.Compile(tt.pattern, 0)
		if err != nil {
			b.Fatalf("r5: failed to compile pattern: %v", err)
		}

		b.ResetTimer()
		b.Run("GRbit-pcre/"+tt.name, func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				r5.MatchWFlags(tt.text, 0)
			}
		})

		r6, err := wasilibs.Compile(tt.pattern)
		if err != nil {
			b.Skipf("r6: failed to compile pattern: %v", err)
		}

		b.ResetTimer()
		b.Run("wasilibs-re2/"+tt.name, func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				r6.MatchString(string(tt.text))
			}
		})
	}
}

latest benchstat of pcregexp v. std regexp - dwisiswant0/pcregexp#5 (comment).

[ehsandeep]

merge conflict ^