Search Enterprise supports multiple languages. You can configure other languages using the src/main/resources/alfresco/search/elasticsearch/config/locale configuration file.

Note: Some supported languages require the installation of a plug-in and some Asian languages may also require installation of the ICU analysis plug-in Where there is a plug-in required, you must install it on every node in the cluster.

The following languages are supported:

Add language

You must update the Content Services <TOMCAT_HOME>/shared/classes/alfresco-global.properties configuration file to enable the language you want to use with Search Enterprise. For example to use French you must add -Delasticsearch.index.locale=fr, and -Dfile.encoding=utf-8.

Docker Compose

If you are installing Search Enterprise using a Docker Compose file you must instruct Alfresco that you are using another language in addition to English. For example, to use French you must add the following environment variable to the Alfresco service (alfresco):

Code copied to clipboard.
JAVA_OPTS: "
  -Delasticsearch.index.locale=fr
"

You must also instruct Alfresco and Search Enterprise to use UTF-8 encoding. This applies to the Alfresco service as well as the Elasticsearch live-indexing and Full-indexing services:

Code copied to clipboard.
JAVA_OPTS: "
  -Dfile.encoding=utf-8
"

Custom configuration

To create a custom configuration you must create a JSON file in the locale configuration directory: src/main/resources/alfresco/search/elasticsearch/config/locale. The file must follow the naming convention of xx_locale.json where xx is the two character locale code. Use the following code example to create your own configuration file. In this case the example is for the French language.

Code copied to clipboard.
{
  "settings": {
    "analysis": {
      "analyzer": {
        "locale_content": {
          "type": "french"
        },
        "locale_text_index": {
          "tokenizer": "whitespace",
          "filter": [
            "asciifolding",
            "custom_word_delimiter_graph",
            "lowercase",
            "flatten_graph",
            "french_stemmer_multiplexer"
          ]
        },
        "locale_text_query": {
          "tokenizer": "whitespace",
          "filter": [
            "asciifolding",
            "custom_word_delimiter_graph",
            "lowercase",
            "french_stemmer_multiplexer"
          ]
        }
      },
      "filter": {
        "french_stemmer": {
          "type": "stemmer",
          "language": "light_french"
        },
        "french_stemmer_multiplexer": {
          "type": "multiplexer",
          "filters": [ "french_stemmer" ]
        }
      }
    }
  }
}

The analysis section contains three analyzers and any custom defined filters. Each analyzer section must contain at least one tokenizer. You can optionally include a number of filters which can then modify the tokens, for example to convert text to lowercase for the index. locale_content is a symmetric content analyzer, which in this example is set to french. locale_text_index is an asymmetric text analyzer, which you set to use whitespace as the delimiter. It has several filters including a custom defined filter, in this instance its called french_stemmer_multiplexer. The locale_text_query section contains an asymmetric text query analyzer which is also set to use whitespace as the delimiter. It has several filters, including, in this case, one called french_stemmer_multiplexer. The filter section contains any custom defined filters that are used by the analyzers. In this case the filter is called french_stemmer_multiplexer.

Checking configuration

You can check your language configuration.

To verify that the index was created using the correct locale configuration:

curl -s '{$hostname}:9200/alfresco/_settings/' | jq

To check how your text is split into tokens:

curl -X GET "{$hostname}:9200/alfresco/_analyze?pretty" -H 'Content-Type: application/json' -d'{"analyzer" : "locale_content", "text" : "I found a dog."}'

Note: For a new environment ensure that the initial reindexing is run first.

To verify that the content of a particular node has been correctly extracted and stored in the index:

curl -s {$hostname}:9200/alfresco/_doc/{$nodeId} | jq ._source

To execute a search query:

curl -s -u admin:admin --header "Content-Type: application/json" -d '{"query":{"query":"dog","language":"afts"}}' "{$hostname}:8080/alfresco/api/-default-/public/search/versions/1/search" | jq .list.entries