regexp_test.go

package main

import (
	"bufio"
	"fmt"
	"io/ioutil"
	"os"
	"regexp"
	"testing"
)

func pattern_match(re *regexp.Regexp, buffer []byte) {
	_ = re.Find(buffer)
}

func pattern_match2(re *regexp.Regexp, buffer []byte) {
	_ = re.FindAll(buffer, -1)
}

func pattern_match_line_by_line(re *regexp.Regexp, scanner *bufio.Scanner) {

	for scanner.Scan() {
		_ = re.Find(scanner.Bytes())
	}
}

func pattern_match_line_by_line2(re *regexp.Regexp, scanner *bufio.Scanner) {

	for scanner.Scan() {
		_ = re.FindAll(scanner.Bytes(), -1)
	}
}

//Entire File?
func BenchmarkPattern1(b *testing.B) {
	pattern1 := "([a-zA-Z][a-zA-Z0-9]*)://([^ /]+)(/[^ ]*)?"
	rep1 := regexp.MustCompile(pattern1)
	file_path := "/Users/ssikdar/Downloads/howto2"
	content_buff, _ := ioutil.ReadFile(file_path)
	for n := 0; n < b.N; n++ {
		pattern_match(rep1, content_buff)
	}
}

func BenchmarkPattern1b(b *testing.B) {
	b.ResetTimer()
	pattern1 := "([a-zA-Z][a-zA-Z0-9]*)://([^ /]+)(/[^ ]*)?"
	rep1 := regexp.MustCompile(pattern1)
	file_path := "/Users/ssikdar/Downloads/howto2"
	content_buff, _ := ioutil.ReadFile(file_path)
	for n := 0; n < b.N; n++ {
		pattern_match2(rep1, content_buff)
	}
}

//Line By Line?
func BenchmarkPattern1LineByLine(b *testing.B) {
	pattern1 := "([a-zA-Z][a-zA-Z0-9]*)://([^ /]+)(/[^ ]*)?"
	rep1 := regexp.MustCompile(pattern1)
	file_path := "/Users/ssikdar/Downloads/howto2"

	for n := 0; n < b.N; n++ {
		in_file, _ := os.Open(file_path)
		defer in_file.Close()
		scanner := bufio.NewScanner(in_file)
		scanner.Split(bufio.ScanLines)

		pattern_match_line_by_line(rep1, scanner)

	}
}

//Line By Line?
func BenchmarkPattern1LineByLine2(b *testing.B) {
	pattern1 := "([a-zA-Z][a-zA-Z0-9]*)://([^ /]+)(/[^ ]*)?"
	rep1 := regexp.MustCompile(pattern1)
	file_path := "/Users/ssikdar/Downloads/howto2"

	for n := 0; n < b.N; n++ {
		in_file, _ := os.Open(file_path)
		defer in_file.Close()
		scanner := bufio.NewScanner(in_file)
		scanner.Split(bufio.ScanLines)

		pattern_match_line_by_line2(rep1, scanner)

	}
}

func main() {
	result := testing.Benchmark(BenchmarkPattern1)
	fmt.Printf("BenchmarkPattern1\t%v\n", result)

	result = testing.Benchmark(BenchmarkPattern1b)
	fmt.Printf("BenchmarkPattern1b\t%v\n", result)

	result = testing.Benchmark(BenchmarkPattern1LineByLine)
	fmt.Printf("BenchmarkPattern1LineByLine\t%v\n", result)

	result = testing.Benchmark(BenchmarkPattern1LineByLine2)
	fmt.Printf("BenchmarkPattern1LineByLine2\t%v\n", result)
}