api/test/unit/middleware/dedupe.js

var data = require('../fixture/dedupe_elasticsearch_results');
var nonAsciiData = require('../fixture/dedupe_elasticsearch_nonascii_results');
var dedupe = require('../../../middleware/dedupe')();

module.exports.tests = {};

module.exports.tests.dedupe = function(test, common) {
  test('filter out duplicates', function(t) {
    var req = {
      clean: {
        text: 'lampeter strasburg high school',
        size: 100
      }
    };
    var res = {
      data: data
    };

    var expectedCount = 8;
    dedupe(req, res, function () {
      t.equal(res.data.length, expectedCount, 'results have fewer items than before');
      t.end();
    });
  });

  test('handle non-ascii gracefully', function(t) {
    var req = {
      clean: {
        size: 100
      }
    };
    var res = {
      data: nonAsciiData
    };

    var expectedCount = 4;
    dedupe(req, res, function () {
      t.equal(res.data.length, expectedCount, 'none were removed');
      t.end();
    });
  });

  test('truncate results based on specified size', function(t) {
    var req = {
      clean: {
        text: 'lampeter strasburg high school',
        size: 3
      }
    };
    var res = {
      data: data
    };

    dedupe(req, res, function () {
      t.equal(res.data.length, req.clean.size, 'results have fewer items than before');
      t.end();
    });
  });
};

module.exports.tests.trump = function(test, common) {
  test('whosonfirst trumps geonames, replace', function (t) {
    var req = {
      clean: {
        text: 'Lancaster',
        size: 100
      }
    };
    var res = {
      data:  [
        {
          'name': { 'default': 'Lancaster' },
          'source': 'geonames',
          'source_id': '123456',
          'layer': 'locality'
        },
        {
          'name': { 'default': 'Lancaster' },
          'source': 'whosonfirst',
          'source_id': '654321',
          'layer': 'locality'
        }
      ]
    };

    var expectedCount = 1;
    dedupe(req, res, function () {
      t.equal(res.data.length, expectedCount, 'results have fewer items than before');
      t.deepEqual(res.data[0].source, 'whosonfirst', 'whosonfirst result won');
      t.end();
    });
  });

  test('whosonfirst trumps geonames, no replace', function (t) {
    var req = {
      clean: {
        text: 'Lancaster',
        size: 100
      }
    };
    var res = {
      data:  [
        {
          'name': { 'default': 'Lancaster' },
          'source': 'whosonfirst',
          'source_id': '123456',
          'layer': 'locality'
        },
        {
          'name': { 'default': 'Lancaster' },
          'source': 'geonames',
          'source_id': '654321',
          'layer': 'locality'
        }
      ]
    };

    var expectedCount = 1;
    dedupe(req, res, function () {
      t.equal(res.data.length, expectedCount, 'results have fewer items than before');
      t.deepEqual(res.data[0].source, 'whosonfirst', 'whosonfirst result won');
      t.end();
    });
  });

  test('openstreetmap trumps whosonfirst venues', function (t) {
    var req = {
      clean: {
        text: 'Lancaster Dairy Farm',
        size: 100
      }
    };
    var res = {
      data:  [
        {
          'name': { 'default': 'Lancaster Dairy Farm' },
          'source': 'openstreetmap',
          'source_id': '123456',
          'layer': 'venue'
        },
        {
          'name': { 'default': 'Lancaster Dairy Farm' },
          'source': 'whosonfirst',
          'source_id': '654321',
          'layer': 'venue'
        }
      ]
    };

    var expectedCount = 1;
    dedupe(req, res, function () {
      t.equal(res.data.length, expectedCount, 'results have fewer items than before');
      t.deepEqual(res.data[0].source, 'openstreetmap', 'openstreetmap result won');
      t.end();
    });
  });

  test('openaddresses trumps openstreetmap', function (t) {
    var req = {
      clean: {
        text: '100 Main St',
        size: 100
      }
    };
    var res = {
      data:  [
        {
          'name': { 'default': '100 Main St' },
          'source': 'openstreetmap',
          'source_id': '123456',
          'layer': 'address'
        },
        {
          'name': { 'default': '100 Main St' },
          'source': 'openaddresses',
          'source_id': '654321',
          'layer': 'address'
        }
      ]
    };

    var expectedCount = 1;
    dedupe(req, res, function () {
      t.equal(res.data.length, expectedCount, 'results have fewer items than before');
      t.deepEqual(res.data[0].source, 'openaddresses', 'openaddresses result won');
      t.end();
    });
  });
};

module.exports.all = function (tape, common) {

  function test(name, testFunction) {
    return tape('[middleware] dedupe: ' + name, testFunction);
  }

  for( var testCase in module.exports.tests ){
    module.exports.tests[testCase](test, common);
  }
};