src/infrastructure/customfield/parser/PhabricatorCustomFieldMonogramParser.php

   1 <?php
   2
   3 abstract class PhabricatorCustomFieldMonogramParser
   4   extends Phobject {
   5
   6   abstract protected function getPrefixes();
   7   abstract protected function getSuffixes();
   8   abstract protected function getInfixes();
   9   abstract protected function getMonogramPattern();
  10
  11   public function parseCorpus($corpus) {
  12     $prefixes = $this->getPrefixes();
  13     $suffixes = $this->getSuffixes();
  14     $infixes = $this->getInfixes();
  15
  16     $prefix_regex = $this->buildRegex($prefixes);
  17     $infix_regex = $this->buildRegex($infixes, true);
  18     $suffix_regex = $this->buildRegex($suffixes, true, true);
  19
  20     $monogram_pattern = $this->getMonogramPattern();
  21
  22     $pattern =
  23       '/'.
  24       '(?:^|\b)'.
  25       $prefix_regex.
  26       $infix_regex.
  27       '((?:'.$monogram_pattern.'(?:\b|$)[,\s]*)+)'.
  28       '(?:\band\s+('.$monogram_pattern.'(?:\b|$)))?'.
  29       $suffix_regex.
  30       '(?:$|\b)'.
  31       '/';
  32
  33     $matches = null;
  34     $ok = preg_match_all(
  35       $pattern,
  36       $corpus,
  37       $matches,
  38       PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
  39
  40     if ($ok === false) {
  41       throw new Exception(pht('Regular expression "%s" is invalid!', $pattern));
  42     }
  43
  44     $results = array();
  45     foreach ($matches as $set) {
  46       $monograms = array_filter(preg_split('/[,\s]+/', $set[3][0]));
  47
  48       if (isset($set[4]) && $set[4][0]) {
  49         $monograms[] = $set[4][0];
  50       }
  51
  52       $results[] = array(
  53         'match' => $set[0][0],
  54         'prefix' => $set[1][0],
  55         'infix' => $set[2][0],
  56         'monograms' => $monograms,
  57         'suffix' => idx(idx($set, 5, array()), 0, ''),
  58         'offset' => $set[0][1],
  59       );
  60     }
  61
  62     return $results;
  63   }
  64
  65   private function buildRegex(array $list, $optional = false, $final = false) {
  66     $parts = array();
  67     foreach ($list as $string) {
  68       $parts[] = preg_quote($string, '/');
  69     }
  70     $parts = implode('|', $parts);
  71
  72     $maybe_tail = $final ? '' : '\s+';
  73     $maybe_optional = $optional ? '?' : '';
  74
  75     return '(?i:('.$parts.')'.$maybe_tail.')'.$maybe_optional;
  76   }
  77
  78 }