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`, \
           REPLACE(REPLACE(REPLACE(REPLACE(`hostPage`.`uri`, '/', ' '), '_', ' '), '-', ' '), '.', ' ') AS `hostPageURIKeywords`, \
           `hostPage`.`rank`, \
           `host`.`name`, \
           IF (`host`.`port` IS NOT NULL, \
                CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`), \
                CONCAT(`host`.`scheme`, '://', `host`.`name`)), \
           CRC32 (IF (`host`.`port` IS NOT NULL, \
                CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`), \
                CONCAT(`host`.`scheme`, '://', `host`.`name`))), \
           IF (`host`.`port` IS NOT NULL, \
                CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`, `hostPage`.`uri`), \
                CONCAT(`host`.`scheme`, '://', `host`.`name`, `hostPage`.`uri`)), \
           CRC32 (IF (`host`.`port` IS NOT NULL, \
                CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`, `hostPage`.`uri`), \
                CONCAT(`host`.`scheme`, '://', `host`.`name`, `hostPage`.`uri`))), \
           REGEXP_REPLACE(`hostPage`.`mime`, '^([A-z-]+)/[A-z-]+.*', '$1') AS `mime`, \
          (SELECT GROUP_CONCAT(`hostPageDescription`.`title`) \
           FROM  `hostPageDescription` \
           WHERE `hostPageDescription`.`hostPageId` = `hostPage`.`hostPageId`) AS `title`, \
          (SELECT GROUP_CONCAT(`hostPageDescription`.`description`) \
           FROM  `hostPageDescription` \
           WHERE `hostPageDescription`.`hostPageId` = `hostPage`.`hostPageId`) AS `description`, \
          (SELECT GROUP_CONCAT(`hostPageDescription`.`keywords`) \
           FROM  `hostPageDescription` \
           WHERE `hostPageDescription`.`hostPageId` = `hostPage`.`hostPageId`) AS `keywords` \
    FROM  `hostPage` \
    JOIN  `host` ON (`host`.`hostId` = `hostPage`.`hostId`) \
    WHERE `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_cz, stem_ar, lemmatize_de_all, lemmatize_ru_all, lemmatize_en_all # stem_enru
  path              = /var/lib/sphinxsearch/data/hostPage

  min_word_len      = 2
  min_prefix_len    = 2

  html_strip        = 1

  index_exact_words = 1
}

indexer
{
  mem_limit         = 256M
  lemmatizer_cache  = 256M
}

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