public function query(AbstractTokenizer $locale, $invoices, array $query = [])
{
$needles = !empty($query['query']) ? $locale->parseWord($query['query']) : [];
$results = [];
foreach ($invoices as $invoice) {
$tokens = $this->getTokens($locale, $invoice);
$scores = [];
// Get the number of each token type
$this->updateMedianEstimator($tokens);
// Similarity score
if (!empty($query['query'])) {
$scores['query'] = $this->getSimilarityScore($tokens, $needles);
}
// Amount found score
if (!empty($query['amount'])) {
$scores['amount'] = $this->containsAmount($tokens, $query['amount']) ? 100 : 0;
}
// Submission date
if (!empty($query['date'])) {
$scores['date'] = $this->getSubmissionDateScore($tokens, $query['date']);
}
// Invoice ID found score
if (!empty($query['id'])) {
$scores['id'] = $this->containsInvoiceId($tokens, $query['id']) ? 100 : 0;
}
$results[] = ['invoice' => $invoice, 'score' => $scores];
}
$medians = $this->computeMedians();
foreach ($results as $i => $result) {
$results[$i]['score'] = $this->computeFinalScore($i, $medians, $result['score']);
}
if ($this->cache) {
$this->cache->commit();
}
return $results;
}
public function parseWord($text)
{
$tokens = [];
// Get all words of 3+ characters
preg_match_all('/(\\w{3,})/u', $text, $matches, PREG_OFFSET_CAPTURE | PREG_SET_ORDER);
foreach ($matches as $match) {
if (!in_array(mb_strtolower($match[0][0]), $this->frequentWords)) {
$tokens[] = new Word($match[0][1], $match[0][0]);
}
}
// Merge adjacent words
$tokens = AbstractTokenizer::mergeAdjacentWords($tokens, function (Word $a, Word $b) use($text) {
$delimiterOffset = $a->getStart() + $a->getLength();
// Words are adjacent if seperated with a space
return $delimiterOffset + 1 === $b->getStart() && $text[$delimiterOffset] === ' ';
});
return $tokens;
}