133

I'm porting a library from Ruby to Go, and have just discovered that regular expressions in Ruby are not compatible with Go (google RE2). It's come to my attention that Ruby & Java (plus other languages use PCRE regular expressions (perl compatible, which supports capturing groups)), so I need to re-write my expressions so that they compile ok in Go.

For example, I have the following regex:

`(?<Year>\d{4})-(?<Month>\d{2})-(?<Day>\d{2})`

This should accept input such as:

2001-01-20

The capturing groups allow the year, month and day to be captured into variables. To get the value of each group, it's very easy; you just index into the returned matched data with the group name and you get the value back. So, for example to get the year, something like this pseudo code:

m=expression.Match("2001-01-20")
year = m["Year"]

This is a pattern I use a lot in my expressions, so I have a lot of re-writing to do.

So, is there a way to get this kind of functionality in Go regexp; how should I re-write these expressions?

Improve this question
0

8 Answers 8

Reset to default
150

how should I re-write these expressions?

Add some Ps, as defined here:

(?P<Year>\d{4})-(?P<Month>\d{2})-(?P<Day>\d{2})

Cross reference capture group names with re.SubexpNames().

And use as follows:

package main

import (
    "fmt"
    "regexp"
)

func main() {
    r := regexp.MustCompile(`(?P<Year>\d{4})-(?P<Month>\d{2})-(?P<Day>\d{2})`)
    fmt.Printf("%#v\n", r.FindStringSubmatch(`2015-05-27`))
    fmt.Printf("%#v\n", r.SubexpNames())
}
Improve this answer
Sign up to request clarification or add additional context in comments.

8 Comments

Ok great that looks encouraging, but how would I get access to the individual values, year, month and day?
Forget that last comment, I just found that answer. Its all in the ?P, as you say :)
I'm still confused by this; I'm not sure they are addressable by Year, Month, etc. I get back an array with four values and can index into it, but that's it.
@thwd Now, this is begging a question: what should happen if you named two groups in the same way? This is not a well-defined behavior, but the regex compiler doesn't complain about it. Your code throws away the first match for example, but I can imagine situations where it would make sense to throw all but first, or maybe collect all of them... Designing a language has lots of subtleties...
@VladimirBauer I'm not sure of what you are getting at. I know it's not specific to Go, I'm arguing that specifically in Go, the built-in library implementation of this feature is bad because it duplicates another simpler feature of this library, but with an additional meaningless syntactical element.
|
46

I had created a function for handling url expressions but it suits your needs too. You can check this snippet but it simply works like this:

/**
 * Parses url with the given regular expression and returns the 
 * group values defined in the expression.
 *
 */
func getParams(regEx, url string) (paramsMap map[string]string) {

    var compRegEx = regexp.MustCompile(regEx)
    match := compRegEx.FindStringSubmatch(url)

    paramsMap = make(map[string]string)
    for i, name := range compRegEx.SubexpNames() {
        if i > 0 && i <= len(match) {
            paramsMap[name] = match[i]
        }
    }
    return paramsMap
}

You can use this function like:

params := getParams(`(?P<Year>\d{4})-(?P<Month>\d{2})-(?P<Day>\d{2})`, `2015-05-27`)
fmt.Println(params)

and the output will be:

map[Year:2015 Month:05 Day:27]
Improve this answer

1 Comment

I appreciate like the completeness of the example in this answer, but it obscures the root of the solution, for me. You have a lot of setup and presentation in this code which clouds the actual answer. I'm not asking you to change your answer, it's existed a long time as it is, I just wanted to give a bit of feedback.
41

As of GO 1.15, you can simplify the process by using Regexp.SubexpIndex. You can check the release notes at https://golang.org/doc/go1.15#regexp.

Based in your example, you'd have something like the following:

re := regexp.MustCompile(`(?P<Year>\d{4})-(?P<Month>\d{2})-(?P<Day>\d{2})`)
matches := re.FindStringSubmatch("Some random date: 2001-01-20")
yearIndex := re.SubexpIndex("Year")
fmt.Println(matches[yearIndex])

You can check and execute this example at https://play.golang.org/p/ImJ7i_ZQ3Hu.

Improve this answer

Comments

29

To improve RAM and CPU usage without calling anonymous functions inside loop and without copying arrays in memory inside loop with "append" function see the next example:

You can store more than one subgroup with multiline text, without appending string with '+' and without using for loop inside for loop (like other examples posted here).

txt := `2001-01-20
2009-03-22
2018-02-25
2018-06-07`

regex := *regexp.MustCompile(`(?s)(\d{4})-(\d{2})-(\d{2})`)
res := regex.FindAllStringSubmatch(txt, -1)
for i := range res {
    //like Java: match.group(1), match.group(2), etc
    fmt.Printf("year: %s, month: %s, day: %s\n", res[i][1], res[i][2], res[i][3])
}

Output:

year: 2001, month: 01, day: 20
year: 2009, month: 03, day: 22
year: 2018, month: 02, day: 25
year: 2018, month: 06, day: 07

Note: res[i][0] =~ match.group(0) Java

If you want to store this information use a struct type:

type date struct {
  y,m,d int
}
...
func main() {
   ...
   dates := make([]date, 0, len(res))
   for ... {
      dates[index] = date{y: res[index][1], m: res[index][2], d: res[index][3]}
   }
}

It's better to use anonymous groups (performance improvement)

Using "ReplaceAllGroupFunc" posted on Github is bad idea because:

  1. is using loop inside loop
  2. is using anonymous function call inside loop
  3. has a lot of code
  4. is using the "append" function inside loop and that's bad. Every time a call is made to "append" function, is copying the array to new memory position
Improve this answer

Comments

8

Simple way to determine group names based on @VasileM answer.

Disclaimer: it's not about memory/cpu/time optimization

package main

import (
    "fmt"
    "regexp"
)

func main() {
    r := regexp.MustCompile(`^(?P<Year>\d{4})-(?P<Month>\d{2})-(?P<Day>\d{2})$`)

    res := r.FindStringSubmatch(`2015-05-27`)
    names := r.SubexpNames()
    for i, _ := range res {
        if i != 0 {
            fmt.Println(names[i], res[i])
        }
    }
}

https://play.golang.org/p/Y9cIVhMa2pU

Improve this answer

Comments

2

If you need to replace based on a function while capturing groups you can use this:

import "regexp"

func ReplaceAllGroupFunc(re *regexp.Regexp, str string, repl func([]string) string) string {
    result := ""
    lastIndex := 0

    for _, v := range re.FindAllSubmatchIndex([]byte(str), -1) {
        groups := []string{}
        for i := 0; i < len(v); i += 2 {
            groups = append(groups, str[v[i]:v[i+1]])
        }

        result += str[lastIndex:v[0]] + repl(groups)
        lastIndex = v[1]
    }

    return result + str[lastIndex:]
}

Example:

str := "abc foo:bar def baz:qux ghi"
re := regexp.MustCompile("([a-z]+):([a-z]+)")
result := ReplaceAllGroupFunc(re, str, func(groups []string) string {
    return groups[1] + "." + groups[2]
})
fmt.Printf("'%s'\n", result)

https://gist.github.com/elliotchance/d419395aa776d632d897

Improve this answer

Comments

1

You can use regroup library for that https://github.com/oriser/regroup

Example:

package main

import (
    "fmt"
    "github.com/oriser/regroup"
)

func main() {
    r := regroup.MustCompile(`(?P<Year>\d{4})-(?P<Month>\d{2})-(?P<Day>\d{2})`)
    mathces, err := r.Groups("2015-05-27")
    if err != nil {
        panic(err)
    }
    fmt.Printf("%+v\n", mathces)
}

Will print: map[Year:2015 Month:05 Day:27]

Alternatively, you can use it like this:

package main

import (
    "fmt"
    "github.com/oriser/regroup"
)

type Date struct {
    Year   int `regroup:"Year"`
    Month  int `regroup:"Month"`
    Day    int `regroup:"Day"`
}

func main() {
    date := &Date{}
    r := regroup.MustCompile(`(?P<Year>\d{4})-(?P<Month>\d{2})-(?P<Day>\d{2})`)
    if err := r.MatchToTarget("2015-05-27", date); err != nil {
        panic(err)
    }
    fmt.Printf("%+v\n", date)
}

Will print: &{Year:2015 Month:5 Day:27}

Improve this answer

Comments

-2

Function for get regexp parameters wit nil pointer checking. Returns map[] if error ocured

// GetRxParams - Get all regexp params from string with provided regular expression
func GetRxParams(rx *regexp.Regexp, str string) (pm map[string]string) {
    if !rx.MatchString(str) {
        return nil
    }
    p := rx.FindStringSubmatch(str)
    n := rx.SubexpNames()
    pm = map[string]string{}
    for i := range n {
        if i == 0 {
            continue
        }

        if n[i] != "" && p[i] != "" {
            pm[n[i]] = p[i]
        }
    }
    return
}
Improve this answer

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.