javascript - Split string into pairs, triplets, quadruplets and on (ngrams)? - Stack Overflow

admin管理员组

文章数量:1350125

I would like to split natural text into word pairs, triplets, quadruplets and on!

I have figured out how to split into pairs so far. I assume I will need an additional loop to acmodate the word count

Here is the code for pairs

var test = "I love you so much, but Joe said \"he doesn't\"!";
var words = test.split(" ");
var two_words = [];
for (var i = 0; i < words.length - 1; i++) {
  two_words.push(words[i] + ' ' + words[i + 1]);
}
console.log(two_words);

// Here is what I am trying

var words = test.split(" ");
var split_words = [];
var split_length = 5;
for (var l = 2; l <= split_length; l++) {
  for (var i = 0; i < words.length - (l - 1); i++) {
    var split_word;
    for (c = 0; c <= l; c++) {
      split_word += split_words[i + c];
    }
    split_words.push(split_word);
  }
}

console.log(split_words);

I would like to split natural text into word pairs, triplets, quadruplets and on!

I have figured out how to split into pairs so far. I assume I will need an additional loop to acmodate the word count

Here is the code for pairs

var test = "I love you so much, but Joe said \"he doesn't\"!";
var words = test.split(" ");
var two_words = [];
for (var i = 0; i < words.length - 1; i++) {
  two_words.push(words[i] + ' ' + words[i + 1]);
}
console.log(two_words);

// Here is what I am trying

var words = test.split(" ");
var split_words = [];
var split_length = 5;
for (var l = 2; l <= split_length; l++) {
  for (var i = 0; i < words.length - (l - 1); i++) {
    var split_word;
    for (c = 0; c <= l; c++) {
      split_word += split_words[i + c];
    }
    split_words.push(split_word);
  }
}

console.log(split_words);

Adding expected output...(an array of ngrams) sg like this

// 2grams
"I love"
"love you"
"you so"
"so much,"
"much, but"
"but Joe"
"Joe said"
"said "he"
""he doesn't"!"
//3grams
"I love you"
"love you so"
"you so much"
"so much, but"
//and on and on

• javascript
• string
• split
• nlp
• n-gram

Share Improve this question Follow edited Feb 16, 2018 at 12:54 giorgio79 asked Feb 16, 2018 at 12:18 giorgio79giorgio79 4,2391010 gold badges6363 silver badges9292 bronze badges 3

• Are you looking forward to solution which picks the pairs in sequence only? e.g. I and so is not the desired result? – Nikhil Aggarwal Commented Feb 16, 2018 at 12:25
• 1 Please share the expected output. – gurvinder372 Commented Feb 16, 2018 at 12:27
• Thx! Just added the expected result. – giorgio79 Commented Feb 16, 2018 at 13:06

Add a ment  | 

4 Answers 4

Sorted by: Reset to default 8

This is called "n-grams" and can be done in modern JavaScript using generators like this:

function* ngrams(a, n) { 
    let buf = [];

    for (let x of a) {
        buf.push(x);
        if (buf.length === n) {
            yield buf;
            buf.shift();
        }
    }
}


var test = "The quick brown fox jumps over the lazy dog";

for (let g of ngrams(test.split(' '), 3))
    console.log(g.join(' '))

Another, more concise and probably faster option:

let ngrams = (a, n) => a.slice(0, 1 - n).map((_, i) => a.slice(i, i + n));

Assuming that your desired result does not include jumbled ordered binations, you can try following

// Code goes here

var test = "I love you so much, but Joe said \"he doesn't\"!";
var arr = test.split(" ");


var words = arr.length; // total length of words
var result = [];

function process(arr, length) { // process array for number of words
  var temp = [];
  // use equal if want to include the plete string as well in the array
  if (arr.length >= length) { 
    // the check excludes any left over words which do not meet the length criteria 
    for (var i = 0; (i + length) <= arr.length; i++) { 
      temp.push(arr.slice(i, length + i).join(" "));
    }

    result.push(temp);
    process(arr, length + 1); // recursive calling
  }

}

process(arr, 2);

console.log(result);

This should do what you're looking for:

function chunkIt(str,chunk) {
  var words = str.split(" ");
  var arr = [];
  for (var i = (chunk - 1); i < words.length; i++) {
    var start = i - (chunk - 1);
    arr.push(words.slice(start, start + chunk));
  }
  return arr.map(v => v.join(" "));
}  


var test = "I love you so much, but Joe said \"he doesn't\"!";
console.log(chunkIt(test,2));
console.log(chunkIt(test,3));
console.log(chunkIt(test,4));

You can dramatically shorten your code by using a library like lodash:

var word = 'foobarbaz';
var chunks = _.chunk(word, 2).map((chunk) => chunk.join(''));
console.log(chunks); //[ 'fo', 'ob', 'ar', 'ba', 'z' ]

Then you can pass in values other than 2 to suit your needs

本文标签：