public function testPspell()
{
if (getenv('SKIP_TEST')) {
return;
}
$stemmer = new DictionaryStemmer(new PspellAdapter(), new SnowballStemmer());
$this->assertEquals("judge", $stemmer->stem("judges"));
// some times approach does not work
$this->assertNotEquals('university', $stemmer->stem("university"));
$this->assertEquals('hammock', $stemmer->stem("hammok"));
}
protected function execute(InputInterface $input, OutputInterface $output)
{
if (ftell(STDIN) === 0) {
$contents = '';
while (!feof(STDIN)) {
$contents .= fread(STDIN, 1024);
}
// filtered tokens
$tokens = array_map([$this, 'filter'], (new GeneralTokenizer())->tokenize($contents));
$tokens = array_unique($tokens);
// stem the tokens
$stemmer = new DictionaryStemmer(new PspellAdapter(), new SnowballStemmer());
$stemmedTokens = array_map(function ($token) use($stemmer) {
return $stemmer->stem($token);
}, $tokens);
// use a dictionary to catch all stemmed words that must be fixed or ignored in this data set
$stemmedTokens = array_map('mb_strtolower', $stemmedTokens);
$comparison = new LevenshteinComparison();
for ($index = 0; $index < count($tokens); $index++) {
// the stemmed word is not a word in the dictionary. The original token
// will need to be manually mapped
if (isset($stemmedTokens[$index]) && isset($tokens[$index]) && $comparison->distance($tokens[$index], $stemmedTokens[$index]) >= 4) {
echo "{$tokens[$index]},{$stemmedTokens[$index]}" . PHP_EOL;
}
}
} else {
throw new \RuntimeException("Please pipe in STDIN");
}
}