source-tokenizer

This is a utility library to tokenize any source code into syntax and manipulate on syntax tree.

Although this library is able to handle any kind of strings, it is highly recommend that only source code is processed.

This is a infrastructure reused between react-source-view and react-diff-view, in most case you will not need this library directly.

Install

With NPM:

npm install @otakustay/source-tokenizer

With Yarn:

yarn add @otakustay/source-tokenizer

Basic Usage

Use tokenize exported function to transform source string into an array containing syntax trees.

import {tokenize} from '@otakustay/source-tokenizer';

const linesOfSyntax = tokenize(source);

const dumpNode = (node, indent) => {
    if (typeof node === 'string') {
        console.log(`${'  '.repeat(indent)}- text(${node})`);
        return;
    }

    console.log(`${'  '.repeat(indent)}- node(${node.type})`);
    node.children.forEach(child => dumpNode(child, indent + 1));
};

for (let i = 0; i < linesOfSyntax.length; i++) {
    console.log(`Line ${i + 1}`);
    const tree = linesOfSyntax[i];
    tree.forEach(node => dumpNode(node, 1));
}

By inputting source as:

const a = 3;
console.log(a);

It outputs:

Line 1
  - text(const a = 3;)
Line 2
  - text(console.log(a);)

Highlight Source Code

There is no actual benefit if we only transform a string into raw syntax trees, we need highlight source code first.

import {tokenize} from '@otakustay/source-tokenizer';
import {refractor} from 'refractor';

const options = {
    highlight(source) {
        return refractor.highlight(source, 'javascript');
    },
};

const linesOfSyntax = tokenize(source, options);

After specifying a more detailed formatter, this time it outputs:

Line 1
  - node(keyword)
    - text(const)
  - text( a )
  - node(operator)
    - text(=)
  - text( )
  - node(number)
    - text(3)
  - node(punctuation)
    - text(;)
  - text()
Line 2
  - text()
  - node(console)
    - text(console)
  - node(punctuation)
    - text(.)
  - node(method)
    - text(log)
  - node(punctuation)
    - text(()
  - text(a)
  - node(punctuation)
    - text())
  - node(punctuation)
    - text(;)

We get syntax information in this tree.

Unlike native highlight tools like highlight.js or prismjs, this DOM independent data structure is easy to further manipulate and easy to customize on render.

Manipulate Syntax

We can pass enhancers array to options:

import {tokenize} from '@otakustay/source-tokenizer';

const options = {
    highlight() {
        // ...
    },
    enhancers: [
        someEnhancer,
        anotherEnhancer,
    ],
};

tokenize(source, options);

Currently 2 enhancers are natively exported.

markWord

The markWord exported enhancer is able to pick a certain character or word from each token and wrap it inside a custom syntax node:

function markWord(word: string, name: string, replacement: string = word): Enhancer;

import {markWord} from '@otakustay/source-tokenizer';

const enhancers = [
    markWord('\t', 'tab', ' '.repeat(4)),
];

This example code is to mark all \t character and replace it with 4 spaces.

We don't encourage you to replace a word with another in different length, if so, this enhancer should be placed at the last of others.

We can only mark a word inside a single token, if this word spans across multiple tokens, mark can fail.

pickRanges

Another utility is pickRanges which allows you to specify some range in each line, wrap them into a single custom node.

A range object specifies the line number, start column and range length, along with custom data attached via properties property:

import {pickRanges} from '@otakustay/source-tokenizer';

const ranges = [
    {
        type: 'reference', // tokens matching this range will transform to a node object with the same type
        line: 1, // this is 1 based
        column: 2, // this is 0 based
        length: 4,
        properties: {
            url: '/docs#foo',
        },
    },
];

const enhancers = [
    pickRanges(ranges),
];