SEARCHENGINE::position_of_first_word()

Complexity

Conditions	30
Paths	3445

Size

Total Lines	67
Code Lines	41

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	930

Importance

Changes

0

<?php
    public function position_of_first_word($body) {
        $lcbody = ' ' . html_entity_decode(strtolower($body)) . ' '; // spaces to make regexp mapping easier
        $pos = -1;

        // look for phrases
        foreach ($this->phrases as $phrase) {
            $phrasematch = join($phrase, '[^'.$this->wordchars.']+');

'[^' . $this->wordchars . ']+' of type string is incompatible with the type array expected by parameter $pieces of join().

            if (preg_match('/([^'.$this->wordchars.']' . $phrasematch . '[^A-Za-z0-9])/', $lcbody, $matches))
            {
                $wordpos = strpos( $lcbody, $matches[0] );
                if ($wordpos) {
                   if ( ($wordpos < $pos) || ($pos==-1) ) {
                        $pos = $wordpos;
                    }
                }
            }
        }
        if ($pos != -1) return $pos;

        $splitextract = preg_split('/([0-9,.]+|['.$this->wordcharsnodigit.']+)/', $lcbody, -1, PREG_SPLIT_DELIM_CAPTURE);
        $stemmed_words = array_map(array($this, 'stem'), $this->words);
        foreach ($splitextract as $extractword) {
            $extractword = preg_replace('/&$/', '', $extractword);
            if (!$extractword) continue;
            $wordpos = strpos($lcbody, $extractword);
            if (!$wordpos) continue;
            foreach ($stemmed_words as $word) {
                if ($word == '') continue;
                $matchword = $this->stem($extractword);
                if ($matchword == $word && ($wordpos < $pos || $pos==-1)) {
                    $pos = $wordpos;
                }
            }
        }
        // only look for earlier words if phrases weren't found
        if ($pos != -1) return $pos;

        foreach ($this->words as $word) {
            if (ctype_digit($word)) $word = '(?:'.$word.'|'.number_format($word).')';
            if (preg_match('/([^'.$this->wordchars.']' . $word . '[^'.$this->wordchars. '])/', $lcbody, $matches)) {
                $wordpos = strpos( $lcbody, $matches[0] );
                if ($wordpos) {
                    if ( ($wordpos < $pos) || ($pos==-1) ) {
                        $pos = $wordpos;
                    }
                }
            }
        }
        // only look for something containing the word (ie. something stemmed, but doesn't work all the time) if no whole word was found
        if ($pos != -1) return $pos;

        foreach ($this->words as $word) {
            if (ctype_digit($word)) $word = '(?:'.$word.'|'.number_format($word).')';
            if (preg_match('/(' . $word . ')/', $lcbody, $matches)) {
                $wordpos = strpos( $lcbody, $matches[0] );
                if ($wordpos) {
                    if ( ($wordpos < $pos) || ($pos==-1) ) {
                        $pos = $wordpos;
                    }
                }
            }
        }

        if ($pos == -1)
            $pos = 0;

        return $pos;
    }