paul-mannino/go-fuzzywuzzy

Port of SeatGeek's fuzzywuzzy

go-fuzzywuzzy

This is a port of SeatGeek's fuzzywuzzy, a fuzzy string matching library.

Usage

Levenshtein Edit Distance

fuzzy.EditDistance("bart", "bort")
1

Simple Ratio

fuzzy.Ratio("coolstring", "coooolstring")
91
fuzzy.Ratio("coolstring", "radstring"))
63

Partial Ratio

fuzzy.Ratio("needle", "haystackneedelhaystack")
36
fuzzy.PartialRatio("needle", "haystackneedelhaystack")
83

Token Sort Ratio

fuzzy.Ratio("several tokens arbitrary order", "order arbitrary several tokens")
50
fuzzy.TokenSortRatio("several tokens arbitrary order", "order arbitrary several tokens")
100

Token Set Ratio

fuzzy.TokenSortRatio("several tokens arbitrary order", "order order arbitrary several tokens")
91
fuzzy.TokenSetRatio("several tokens arbitrary order", "order order arbitrary several tokens")
100

Process

choices := []string{"Wayne Shorter", "Jonathan Richman", "Wayne Hancock", "Kate Bush"}
fuzzy.ExtractOne("wayne hancock", choices)
{Match:"Wayne Hancock", Score:100}
fuzzy.Extract("wayne hancock", choices, 2)
[{Match:"Wayne Hancock", Score:100}, {Match:"Wayne Shorter", Score:62}]

go-fuzzywuzzy's Issues

Can we use goroutine

When I have to find the best 50 with accepted match score 70, with 22k choices taking 2s,
It's too high
Using goroutines can we optimize it further or parallel computing?
Please suggest me possible solution.

fuzzy.Ratio() returning 100 when not exact match

Differing results between this package and python package. In this case, whenever one number is different, it still returns 100. If a letter is different, it appears to work fine:

package main

import "fmt"
import "github.com/paul-mannino/go-fuzzywuzzy"

func main(){
    a := "JOHNSMITH6211986"
    b := "JOHNSMITH6201986"
    r := fuzzy.Ratio(a,b)
    //expect 94, but get 100
    fmt.Println(r)
}

from fuzzywuzzy import fuzz

a = "JOHNSMITH6211986"
b = "JOHNSMITH6201986"
#get 94
fuzz.ratio(a,b)

fuzzy.PartialRatio function can panics with runtime error "slice bounds out of range" with certain strings

sample: https://play.golang.org/p/HdOWi9E49hM

请问中文调用fuzzy.ExtractOne匹配score都为0.有什么参数设置么？

choices := []string{"芯姐好工作（聘）", "芯姐好工作", "还能不能工作好了"}
ans,_ :=fuzzy.ExtractOne("芯姐好工作", choices)
fmt.Println("ans",ans)

{芯姐好工作（聘） 0}
{芯姐好工作 0}
{还能不能工作好了 0}
matches [0xc00011c040 0xc00011c060 0xc00011c080]
ans &{芯姐好工作（聘） 0}

this package does not support rtl language

i have tried this example but does not worked for me

package main

import (
	"fmt"

	fuzzy "github.com/paul-mannino/go-fuzzywuzzy"
)

func main() {
	choices := []string{"طماطة", "Jonathan Richman", "Wayne Hanock", "Kate Bush"}

	// words := []string{"geek", "gfg", "wh", "wl"}

	fmt.Println(fuzzy.ExtractOne("طماطة", choices))

}

Improve performance

A Code Review answer shows that go-fuzzywuzzy is much slower than necessary. The suggestions from there should probably be incorporated into the code.