Wondering what’s next for npm?Check out our public roadmap! »

    tokenizer2

    2.0.1 • Public • Published

    tokenizer2

    build status

    tokenize any text stream given some basic regex rules to match tokens

    NOTE This library works well, but I don't use it anymore. I just use while loops in a state machine pattern to tokenize. No library needed (or wanted). Here are some examples: one, two Yes it's stateful and verbose, but in my experience this is easier to write and maintain (using TDD of course). Just setup a test-runner and start small then grow it to tokenize everything you want. Once you get the hang of it, it's really easy to figure out how to tokenize something since you have full control of the state machine.

    Example

    var tokenizer2 = require('tokenizer2');
     
    //create a readable/writeable stream
    var token_stream = tokenizer2();
     
    //make some rules
    token_stream.addRule(/^[\s]+$/               , 'whitespace');
    token_stream.addRule(/^"([^"]|\\")*"$/       , 'string');
    token_stream.addRule(/^[-+]?[0-9]+\.?[0-9]*$/, 'number');
    token_stream.addRule(/^[^"0-9\s][^\s]*$/     , 'symbol');
     
    //write some info to the console
    token_stream.on('data', function(token){
      console.log('token:', token);
    });
    token_stream.on('end', function(){
      console.log('DONE');
    });
     
    //pipe in some data
    fs.createReadStream('./demo.txt').pipe(token_stream);

    demo.txt

    print "some multi-
    lined string"
     
    123.25 times -10

    The output

    token: {type: 'symbol'    , src: 'print',  line: 1, col:  1 }
    token: {type: 'whitespace', src: ' ',      line: 1, col:  6 }
    token: {type: 'string'    , src: '"some multi-\nlined string"', line: 1, col: 7 }
    token: {type: 'whitespace', src: '\n\n',   line: 2, col: 14 }
    token: {type: 'number'    , src: '123.25', line: 4, col:  1 }
    token: {type: 'whitespace', src: ' ',      line: 4, col:  7 }
    token: {type: 'symbol'    , src: 'times',  line: 4, col:  8 }
    token: {type: 'whitespace', src: ' ',      line: 4, col: 13 }
    token: {type: 'number'    , src: '-10',    line: 4, col: 14 }
    token: {type: 'whitespace', src: '\n',     line: 4, col: 17 }
    DONE

    What if more than one rule matches a token?

    token_stream.addRule adds rules in an order sensitive way. The first matching rule will be used.

    Why tokenizer2

    The key difference between this and tokenizer is the way it matches rules. tokenizer uses disect to do bisection on a chunk of text. This is a fast approach, however doesn't work well if your regex rule expects some specific characters at the end of the token. To solve this tokenizer2 simply starts at the beginning of the chunk, and finds the longest matching rule.

    Other differences

    • tokenizer2 wraps through2.obj so all the node stream APIs should work nicely
    • tokenizer2 uses the standard 'data' event to emit the tokens
    • tokenizer2 emits line and col numbers

    Non-streaming, synchronous API

    If, for whatever reason, you don't want to use the streaming api. There is a lighter weight, synchronous api.

    var core = require('tokenizer2/core');
     
    var t = core(function(token){
      //called synchronously on every token found
    });
     
     
    //add rules just like the streaming api
    t.addRule(/^[\s]+$/, 'whitespace');
     
    //Give it strings to tokenize
    t.onText("some text to tokenize");
    t.onText("some more text");
     
    //Call this when it's done
    t.end();//this may throw an error

    License

    MIT

    Install

    npm i tokenizer2

    DownloadsWeekly Downloads

    1,475

    Version

    2.0.1

    License

    MIT

    Unpacked Size

    7.81 kB

    Total Files

    5

    Last publish

    Collaborators

    • avatar