golang多行匹配与逐行匹配性能

re_multiline_test.go

package main

import (
	"fmt"
	"regexp"
	"strings"
	"testing"
)

var template = `
%s
#+begin_src bash
  cccc
  bvv
  %s
#+end_src
%s
`
var text = fmt.Sprintf(template,
	strings.Repeat("vvvvvvvvvvvvasdasdasdads\n", 100),
	strings.Repeat("vvvvvvvvvvvvasdasdasdads\n", 10),
	strings.Repeat("vvvvvvvvvvvvasdasdasdads\n", 100),
)

var (
	beginBlockRegexp = regexp.MustCompile(`(?i)^(\s*)#\+BEGIN_(\w+)(.*)`)
	endBlockRegexp   = regexp.MustCompile(`(?i)^(\s*)#\+END_(\w+)`)
)

var (
	blockRegexp = regexp.MustCompile(`(?ism:^(\s*)#\+BEGIN_(\w+)(.+?)\n(.+?)^(\s*)#\+END_(\w+))`)
)

type Block struct {
	Type       string
	Parameters []string
}

func parseSplit(lines []string) (*Block, int) {
	match := beginBlockRegexp.FindStringSubmatch(lines[0])
	if match == nil {
		return nil, 0
	}
	blockType := strings.ToUpper(match[2])

	end := len(lines)
	num := 1
	for num < end {
		if m := endBlockRegexp.FindStringSubmatch(lines[num]); m != nil && strings.ToUpper(m[2]) == blockType {
			return &Block{
				Type:       blockType,
				Parameters: strings.Split(strings.TrimSpace(match[3]), " "),
			}, num + 1
		}
		num++
	}
	return nil, 0
}

func parseRegex(lines string) (*Block, int) {
	match := blockRegexp.FindStringSubmatch(lines)
	if match == nil {
		return nil, 0
	}
	blockType := strings.ToUpper(match[2])
	b := &Block{
		Type:       blockType,
		Parameters: strings.Split(strings.TrimSpace(match[3]), " "),
	}
	return b, 0
}

func BenchmarkSplit(b *testing.B) {
	count := 0
	lines := strings.Split(text, "\n")
	for i := 0; i < b.N; i++ {
		index := 0
		end := len(lines)
		for index < end {
			if m, n := parseSplit(lines[index:]); m != nil {
				index = index + n

				if count == 0 {
					b.Log(m)
				}
				count++
				continue
			}
			index++
		}
	}
	b.Log(count)
}

func BenchmarkRegex(b *testing.B) {
	count := 0
	for i := 0; i < b.N; i++ {
		if m, _ := parseRegex(text); m != nil {
			if count == 0 {
				b.Log(m)
			}
			count++
		}
	}
	b.Log(count)
}

goos: darwin
goarch: amd64
pkg: org
cpu: Intel(R) Core(TM) i5-8279U CPU @ 2.40GHz
BenchmarkSplit
    main_test.go:84: &{SRC [bash]}
    main_test.go:92: 1
    main_test.go:84: &{SRC [bash]}
    main_test.go:92: 100
    main_test.go:84: &{SRC [bash]}
    main_test.go:92: 10000
    main_test.go:84: &{SRC [bash]}
    main_test.go:92: 47133
BenchmarkSplit-8   	   47133	     24606 ns/op	     313 B/op	       8 allocs/op
BenchmarkRegex
    main_test.go:100: &{SRC [bash]}
    main_test.go:105: 1
    main_test.go:100: &{SRC [bash]}
    main_test.go:105: 100
    main_test.go:100: &{SRC [bash]}
    main_test.go:105: 10000
    main_test.go:100: &{SRC [bash]}
    main_test.go:105: 21664
BenchmarkRegex-8   	   21664	     56515 ns/op	     293 B/op	       5 allocs/op
PASS
ok  	org	3.768s