DomFramework/Tests/FtsTest.php

<?php

/**
 * DomFramework - Tests
 * @package domframework
 * @author Dominique Fournier <dominique@fournier38.fr>
 * @license BSD
 */

namespace Domframework\Tests;

use Domframework\Fts;

/**
 * Test the FTS
 */
class FtsTest extends \PHPUnit_Framework_TestCase
{
    public function testTokenizerSearch0()
    {
        // Empty
        $fts = new Fts();
        $fts->search("");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => [],
                                    "minuses" => []]);
    }

    public function testTokenizerSearch1()
    {
        // Too small
        $fts = new Fts();
        $fts->search("X");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => [],
                                    "minuses" => []]);
    }

    public function testTokenizerSearch2()
    {
        // One word
        $fts = new Fts();
        $fts->search("XYZ");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ"],
                                    "minuses" => [""]]);
    }

    public function testTokenizerSearch3()
    {
        // Two word
        $fts = new Fts();
        $fts->search("XYZ 123");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ", "123"],
                                    "minuses" => ["", ""]]);
    }

    public function testTokenizerSearch4()
    {
        // Three word
        $fts = new Fts();
        $fts->search("XYZ 123 ABC");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ", "123", "ABC"],
                                    "minuses" => ["", "", ""]]);
    }

    public function testTokenizerSearch5()
    {
        // Three word
        $fts = new Fts();
        $fts->search("XYZ 123 ABC KLM");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ", "123",
                                                     "ABC", "KLM"],
                                    "minuses" => ["", "", "", ""]]);
    }

    public function testTokenizerSearch6()
    {
        // Three word
        $fts = new Fts();
        $fts->search("Louis-XYZ 123 -AéBCé KLM");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["Louis-XYZ", "123",
                                                     "AéBCé", "KLM"],
                                    "minuses" => ["", "", "-", ""]]);
    }


    public function testTokenizerSentence0()
    {
        // Empty sentence
        $fts = new Fts();
        $fts->search("\"\"");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => [],
                                    "minuses" => []]);
    }

    public function testTokenizerSentence1()
    {
        // One sentence only
        $fts = new Fts();
        $fts->search("\"XYZ 123\"");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ 123"],
                                    "minuses" => [""]]);
    }

    public function testTokenizerSentence2()
    {
        // Two sentence
        $fts = new Fts();
        $fts->search("\"XYZ 123\" \"ABC KLM\"");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ 123", "ABC KLM"],
                                    "minuses" => ["", ""]]);
    }

    public function testTokenizerSentence3()
    {
        // Three sentence
        $fts = new Fts();
        $fts->search("\"XYZ 123\" -\"ABC KLM\" \"RPO YUI\"");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ 123", "ABC KLM",
                                                     "RPO YUI"],
                                    "minuses" => ["", "-", ""]]);
    }

    public function testTokenizerMixed1()
    {
        // One word and one sentence, starting by word
        $fts = new Fts();
        $fts->search("XYZ \"ABC KLM\"");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["XYZ", "ABC KLM"],
                                    "minuses" => ["", ""]]);
    }

    public function testTokenizerMixed2()
    {
        // One word and one sentence, starting by sentence
        $fts = new Fts();
        $fts->search("\"ABC KLM\" XYZ");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["ABC KLM", "XYZ"],
                                    "minuses" => ["", ""]]);
    }

    public function testTokenizerMixed3()
    {
        // One word and two sentences, starting by sentence
        $fts = new Fts();
        $fts->search("\"ABC KLM\" XYZ \"RPO YUI\"");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["ABC KLM", "XYZ",
                                                     "RPO YUI"],
                                    "minuses" => ["", "", ""]]);
    }

    public function testTokenizerMixed4()
    {
        // Two words and two sentences, starting by sentence
        $fts = new Fts();
        $fts->search("\"ABC KLM\" XYZ \"RPO YUI\" 123");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["ABC KLM", "XYZ",
                                                     "RPO YUI", "123"],
                                    "minuses" => ["", "", "", ""]]);
    }

    public function testTokenizerMixed5()
    {
        // Two words and two sentences, starting by a word
        $fts = new Fts();
        $fts->search("123 \"ABC KLM\" XYZ \"RPO YUI\"");
        $res = $fts->getTokensMin();
        $this->assertSame($res, ["tokens" => ["123", "ABC KLM",
                                                     "XYZ", "RPO YUI"],
                                    "minuses" => ["", "", "", ""]]);
    }
}