golang正则表达式以提取数量对及其单位

Question

I have a set of human readable strings expressing a duration of time. Here are four examples:

1 days 40 hrs 23 min 50 sec

3 hrs 1 min 30 sec

10 days 23 min 11 sec

52 sec

I am trying to convert these strings into number of seconds. The math to do this is quite simple once the string is broken down into its components - it's just multiplication and addition. I am having some issues however with writing the regular expression to parse the string into [<quantity>, <unit>] pairs. As an example, the output I would like for the string:

1 days 40 hrs 23 min 50 sec

is an array (or slice) like:

[[1, "days"], [40, "hrs"], [23, "min"], [50, "sec"]].

Below is the code for what I've tried so far and its output (executable at http://play.golang.org/p/iR-xfc8MVQ). segs was my first attempt, which seems to break the string down into 4 components ok but each component is just a string like 1 days rather than a 2-element array like [1, days]. segs2 was my second attempt, which seems to do something weirder where each component is repeated twice.

// time unit tokenizer
package main

import "fmt"
import "regexp"

func main() {
    s := "1 days 40 hrs 23 min 50 sec"
    re := regexp.MustCompile("(?P<quant>\\d+) (?P<unit>\\w+)+")

    segs := re.FindAllString(s, -1)
    fmt.Println("segs:", segs)
    fmt.Println(segs[0], "," ,segs[1], ",", segs[2], ",", segs[3])  
    fmt.Println("length segs:", len(segs))

    segs2 := re.FindAllStringSubmatch(s, -1)
    fmt.Println("segs2:", segs2)
    fmt.Println(segs2[0], "," ,segs2[1], ",", segs2[2], ",", segs2[3])
    fmt.Println("length segs2:", len(segs2))
}

Output:

segs: [1 days 40 hrs 23 min 50 sec]
1 days , 40 hrs , 23 min , 50 sec
length segs: 4
segs2: [[1 days 1 days] [40 hrs 40 hrs] [23 min 23 min] [50 sec 50 sec]]
[1 days 1 days] , [40 hrs 40 hrs] , [23 min 23 min] , [50 sec 50 sec]
length segs2: 4

I've written a similar regex is Python which works OK, so I'm really not sure whether I am doing something incorrect for Go's regular expression syntax or perhaps making the wrong call on the re object.

Answer 1

Regexp.FindAllStringSubmatch returns [][]string. But its contents are slightly different from the return value of the Python function re.findall (I assumed that you used re.findall in Python).

• return_value[i][0] contains whole matched string.
• return_value[i][1] contains captured group 1.
• return_value[i][2] contains captured group 2. ....

Printing return_value[i] cause all items in return_value[i] to be printed. (return_value[i][0], return_value[i][1], return_value[i][2], ..)

---

You can get what you expected by only printing captured group matches (excluding [0]) as follow:

segs2 := re.FindAllStringSubmatch(s, -1)
for i := 0; i < len(segs2); i++ {
    fmt.Println(segs2[i][1], "," ,segs2[i][2]);
}

Demo

---

Side Note

Following string literal:

"(?P<quant>\\d+) (?P<unit>\\w+)+"

can be expressed as the following raw string literals.

`(?P<quant>\d+) (?P<unit>\w+)+`

See String literals