Identify the correct hashtag indexes in tweet messages

February 28, 2022

I need to identify the correct indexes in twitter messages (various languages, emojis, etc).

I can’t find a solution that returns these positions as shown in the example below.

import (
    "regexp"
    "testing"

    "github.com/stretchr/testify/require"
)

func TestA(t *testing.T) {
    text := "🇷🇺 [URGENT] Les forces de dissuasion #nucleaire de la #Russie"

    var re = regexp.MustCompile(`#([_A-Za-z0-9]+)`)

    pos := re.FindAllStringIndex(text, -1)

    // FindAllStringIndex returns
    // [0][43,53]
    // [1][60,67]

    // These are the expected positions.

    require.Equal(t, pos[0][0], 37) 
    require.Equal(t, pos[0][1], 47)

    require.Equal(t, pos[1][0], 54)
    require.Equal(t, pos[1][1], 61)
}

>Solution :

The FindAllStringIndex() function returns the position of bytes, not runes.

You need to import "unicode/utf8" and use utf8.RuneCountInString(text[:pos[0][0]]) and so on instead of pos[0][0] to make sure you count the Unicode code points and not just bytes:

// You can edit this code!
// Click here and start typing.
package main

import (
    "regexp"
    "testing"
    "unicode/utf8"

    "github.com/stretchr/testify/require"
)

func TestA(t *testing.T) {
    text := "🇷🇺 [URGENT] Les forces de dissuasion #nucleaire de la #Russie"

    var re = regexp.MustCompile(`#\w+`)

    pos := re.FindAllStringIndex(text, -1)

    require.Equal(t, utf8.RuneCountInString(text[:pos[0][0]]), 37)
    require.Equal(t, utf8.RuneCountInString(text[:pos[0][1]]), 47)
    require.Equal(t, utf8.RuneCountInString(text[:pos[1][0]]), 54)
    require.Equal(t, utf8.RuneCountInString(text[:pos[1][1]]), 61)

}

See the Go demo.

Also, #\w+ is a a shorter pattern to match a # and then one or more letters, digits or underscores.