rustemmer.module in Russian stemming 7

Russian stemming algorith provided by Dr Martin Porter.

http://snowball.tartarus.org/algorithms/russian/stemmer.html

Algorith implementation in PHP provided by Dmitry Koterov (dklab.ru): http://forum.dklab.ru/php/advises/HeuristicWithoutTheDictionaryExtractio...

Initial implementation adopted for Drupal by Algenon ([e-mail])

File

rustemmer.module

<?php

/**
 * @file
 * Russian stemming algorith provided by Dr Martin Porter.
 *
 * http://snowball.tartarus.org/algorithms/russian/stemmer.html
 *
 * Algorith implementation in PHP provided by Dmitry Koterov (dklab.ru):
 * http://forum.dklab.ru/php/advises/HeuristicWithoutTheDictionaryExtractionOfARootFromRussianWord.html
 *
 * Initial implementation adopted for Drupal by Algenon ([e-mail])
 */
define('RUSTEMMER_CHARS', '_0-9a-zA-ZабвгдежзийклмнопрстуфхцчшщьыъэюяАБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЬЫЪЭЮЯ');

/**
 * Implements hook_help().
 */
function rustemmer_help($path, $arg) {
  switch ($path) {
    case 'admin/help#rustemmer':
      return t('Improves search of Russian words by using the stemming algorithm for Russian language.');
  }
}

/**
 * Implements hook_search_preprocess().
 */
function rustemmer_search_preprocess($text) {
  $words = $text;
  $words = str_replace('ё', 'е', $words);
  $words = str_replace('Ё', 'Е', $words);

  // Split words from noise and remove apostrophes.
  $words = preg_split('/([^' . RUSTEMMER_CHARS . ']+)/u', str_replace("'", '', $words), -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);
  $stemmer = new RussianStemmer();

  // Process each word.
  $odd = TRUE;
  foreach ($words as $k => $word) {
    if ($odd) {
      $words[$k] = $stemmer
        ->stem_word($word);
    }
    $odd = !$odd;
  }

  // Put it all back together.
  return implode('', $words);
}

/**
 * Implements hook_search_admin().
 */
function rustemmer_search_admin() {
  $form = array();
  $form['indexing_settings']['rustemmer_stemcaching'] = array(
    '#type' => 'checkbox',
    '#title' => t('Stem caching'),
    '#default_value' => variable_get('rustemmer_stemcaching', 0),
    '#description' => t('Apply stem caching while performing search phrase stemming. Warning: This could lead to additional memory consumption while indexing.'),
  );
  return $form;
}

/**
 * Implements hook_search_api_processor_info().
 */
function rustemmer_search_api_processor_info() {
  $processors['search_api_russian_stemmer'] = array(
    'name' => t('Russian stemmer'),
    'description' => t('This processor improves search of Russian words by using the stemming algorithm for Russian language.'),
    'class' => 'SearchApiRussianStemmer',
  );
  return $processors;
}

Functions

Name	Description
rustemmer_help	Implements hook_help().
rustemmer_search_admin	Implements hook_search_admin().
rustemmer_search_api_processor_info	Implements hook_search_api_processor_info().
rustemmer_search_preprocess	Implements hook_search_preprocess().

Constants

Name	Description
RUSTEMMER_CHARS	@file Russian stemming algorith provided by Dr Martin Porter.