nounfinder/tests/nounfinder-tests.js

Extracts nouns from chunks of text, using the [Wordnik API](http://developer.wordnik.com/docs.html).

var assert = require('assert');
var createNounfinder = require('../nounfinder');
var queue = require('queue-async');
var config = require('../config');

var nounfinder = createNounfinder({
  wordnikAPIKey: config.wordnikAPIKey
});

var textsAndNouns = [
  {
    text: 'Facebook running studies to see if it can manipulate users\' emotional states by skewing their feeds.', 
    nouns: [
      'facebook',
      'it',
      'running',
      'state',
      'feed'
    ],
    interestingNouns: []
  },
  {
    text: 'During this work, the street will be closed to motorists, except for abutters.', 
    nouns: [
      'this',
      'work',
      'street',
      'will',
      'motorist',
      'abutter'
    ],
    interestingNouns: [
      'motorist',
      'abutter'
    ]
  },
  {
    text: 'Property owners will maintain access to their driveways.', 
    nouns: [
      'will',
      'property',
      'owner',
      'access',
      'driveway'
    ],
    interestingNouns: [
      'driveway'
    ]
  },
  {
    text: 'In addition to the street closure, there will be no parking on either side of the street.', 
    nouns: [
      'addition',
      'street',
      'closure',
      'will',
      'parking',
      'either',
      'side' 
    ],
    interestingNouns: [
      'closure'
    ]
  },
  {
    text: 'If voting didn\'t matter Republicans wouldn\'t be trying so hard to stop people from doing it.', 
    nouns: [
      'matter',
      'republican',
      'person',
      'doing',
      'it'
    ],
    interestingNouns: [
      'republican'
    ]
  },
  {
    text: 'Hello, dystopia. You\'re a little early, but I\'ve been expecting you.',
    nouns: [
      'dystopium',
      'you'
    ],
    interestingNouns: [
    ]
  },
  {
    text: 'So @seventeenmag\'s app has shared my stripes vs. polkadots ' + 
      '@polarpolls poll. http://polarb.com/65517  They know I\'m a dude in my forties, right?',
    nouns: [
      'app',
      'stripe',
      // 'v',
      'poll',
      'dude',
      // 'forties'
      // Should leave out v and forties.,
      'they'
    ],
    interestingNouns: [
      'stripe',
      'dude',
      // '40s'
      // should leave out 40s
    ]
  },  
  {
    text: 'So @seventeenmag\'s app has shared my stripes vs. polkadots ' + 
      '@polarpolls poll. http://polarb.com/65517  They know I\'m a dude in my 40s, right?',
    nouns: [
      'app',
      'stripe',
      // 'v',
      'poll',
      'dude',
      // '40s'
      // Should leave out v and 40.,
      'they'
    ],
    interestingNouns: [
      'stripe',
      'dude',
      // '40s'
      // should leave out 40s
    ]
  },  
  {
    text: 'PLS don\'t use pls in an exhortation',
    nouns: [
      'exhortation'
    ],
    interestingNouns: [
      'exhortation'
    ]
  },  
  {
    text: 'w/e. tl;dr capitalism ruins everything. i’m going to bed.',
    nouns: [
      'capitalism',
      'ruin',
      'everything',
      'going',
      'bed'
    ],
    interestingNouns: [
      'capitalism',
      'ruin'
    ]
  },
  {
    text: 'Back when some stuff is done. 😊 🐸📈🐸',
    nouns: [
      'back',
      'stuff',
      '😊',
      '🐸',
      '📈'
    ],
    interestingNouns: [
      '🐸',
      '📈'
    ]
  },
  {
    text: 'Read a really cool AMA with the Zodiac Killer!',
    nouns: [
      'zodiac',
      'killer'
    ],
    interestingNouns: [
      'zodiac'
    ]
  },
  {
    text: 'Alas. I seem to have missed the boat.',
    nouns: [
      'boat'
    ],
    interestingNouns: [
    ]
  }
];

var expectedNounCache = {
    facebook: 151,
    running: 1102,
    study: 880,
    user: 373,
    state: 2719,
    work: 4804,
    street: 569,
    will: 18228,
    motorist: 5,
    abutter: 0,
    property: 649,
    owner: 387,
    access: 936,
    driveway: 34,
    addition: 532,
    closure: 56,
    no: 9127,
    parking: 201,
    side: 1657,
    voting: 377,
    matter: 1382,
    republican: 51,
    person: 1675,
    doing: 2149,
    hello: 87,
    app: 163,
    stripe: 8,
    poll: 387,
    dude: 76,
    'don\'t': 5157,
    exhortation: 1,
    capitalism: 88,
    ruin: 65,
    going: 6355,
    bed: 519,
    back: 6613,
    stuff: 1107,
    zodiac: 1,
    killer: 150,
    boat: 231
  };

suite('Noun getting', function gettingSuite() {
  this.timeout(30000);

  function testTextAndNouns(text, expectedNouns, done) {
    nounfinder.getNounsFromText(text,
      function checkNouns(error, nouns) {
        assert.ok(!error);
        assert.deepEqual(nouns.sort(), expectedNouns.sort());
        done(error, nouns)
      }
    );
  }

  test('Test getting nouns from text.', function testGetNouns(testDone) {
    var q = queue(1);
    textsAndNouns.forEach(function addNounsTest(textAndNouns) {
      q.defer(testTextAndNouns, textAndNouns.text, textAndNouns.nouns);
    });
    q.awaitAll(testDone);
  });
});

suite('Noun frequencies', function frequenciesSuite() {
  this.timeout(30000);

  function testInterestingFilter(nouns, expectedNouns, done) {
    nounfinder.filterNounsForInterestingness(nouns, 100,
      function checkNouns(error, filteredNouns) {
        assert.ok(!error);
        assert.deepEqual(filteredNouns.sort(), expectedNouns.sort());
        done(error, filteredNouns)
      }
    );
  }

  test('Test filtering nouns', function testFilteringSetsOfNouns(testDone) {
    function addFilterTest(textAndNouns) {
      q.defer(testInterestingFilter, textAndNouns.nouns, 
        textAndNouns.interestingNouns
      );
    }
    var q = queue(1);
    textsAndNouns.forEach(addFilterTest);
    
    q.awaitAll(testDone);
  });
});