YGGo/config/sphinx.conf.txt

source common
{
  type        = mysql

  sql_host    = localhost
  sql_user    =
  sql_pass    =
  sql_db      =
  sql_port    = 3306 # optional, default is 3306
}

source hostPage : common
{
  sql_query   = \
    SELECT `hostPage`.`hostPageId`, \
           `hostPage`.`uri`, \
           `host`.`name`, \
           REGEXP_REPLACE(`hostPage`.`mime`, '^([A-z-]+)/[A-z-]+.*', '$1') AS `mime`, \
          (SELECT COUNT(*) FROM `hostPageToHostPage` \
           WHERE `hostPageToHostPage`.`hostPageIdTarget` = `hostPage`.`hostPageId` \
           AND   (SELECT `hostPageSource`.`hostId` FROM `hostPage` AS `hostPageSource` \
                  WHERE  `hostPageSource`.`hostPageId` = `hostPageToHostPage`.`hostPageIdSource`) <> `hostPage`.`hostId`) AS `rank`, \
          (SELECT GROUP_CONCAT(CONCAT_WS(' ', `hostPageDescription`.`title`, \
                                              `hostPageDescription`.`description`, \
                                              `hostPageDescription`.`keywords`)) \
           FROM  `hostPageDescription` \
           WHERE `hostPageDescription`.`hostPageId` = `hostPage`.`hostPageId`) AS `pageDescription` \
    FROM  `hostPage` \
    JOIN  `host` ON (`host`.`hostId` = `hostPage`.`hostId`) \
    WHERE `host`.`status` = '1' AND `hostPage`.`httpCode` = 200 AND `hostPage`.`timeBanned` IS NULL AND `hostPage`.`mime` IS NOT NULL \

  sql_attr_uint   = rank
  sql_attr_string = mime
}

index hostPage
{
  source            = hostPage
  morphology        = stem_enru, stem_cz, stem_ar, lemmatize_de_all # lemmatize_ru_all, lemmatize_en_all
  path              = /var/lib/sphinxsearch/data/hostPage

  min_word_len      = 2
  min_prefix_len    = 2

  html_strip        = 1

  index_exact_words = 1
}

indexer
{
  mem_limit         = 128M
  lemmatizer_cache  = 128M
}

common {
  lemmatizer_base   = /var/lib/sphinxsearch/dicts # http://sphinxsearch.com/downloads/dicts
}