search_core/Documentation/source/configuration/indexing.rst

Indexing

Holds settings regarding the indexing, e.g. of TYPO3 records, to search services.

Configured as:

plugin {
    tx_searchcore {
        settings {
            indexing {
                identifier {
                    indexer = FullyQualifiedClassname
                    // the settings
                }
            }
        }
    }
}

Where identifier is up to you, but should match table names to make TcaIndexer work.

The following settings are available. For each setting its documented which indexer consumes it.

rootLineBlacklist

Used by: TcaIndexer, PagesIndexer.

Defines a blacklist of page uids. Records below any of these pages, or subpages, are not indexed. This allows you to define areas that should not be indexed. The page attribute No Search is also taken into account to prevent indexing records from only one page without recursion.

Contains a comma separated list of page uids. Spaces are trimmed.

Example:

plugin.tx_searchcore.settings.indexing.<identifier>.rootLineBlacklist = 3, 10, 100

Also it's possible to define some behaviour for the different document types. In context of TYPO3 tables are used as document types 1:1. It's possible to configure different tables. The following options are available:

additionalWhereClause

Used by: TcaIndexer, PagesIndexer.

Add additional SQL to where clauses to determine indexable records from the table. This way you can exclude specific records like tt_content records with specific CType values or something else. E.g. you can add a new field to the table to exclude records from indexing.

Example:

plugin.tx_searchcore.settings.indexing.<identifier>.additionalWhereClause = tt_content.CType NOT IN ('gridelements_pi1', 'list', 'div', 'menu')

Attention

Make sure to prefix all fields with the corresponding table name. The selection from database will contain joins and can lead to SQL errors if a field exists in multiple tables.

abstractFields

Used by: PagesIndexer.

Define which field should be used to provide the auto generated field "search_abstract". The fields have to exist in the record to be indexed. Therefore fields like content are also possible.

Example:

# As last fallback we use the content of the page
plugin.tx_searchcore.settings.indexing.<identifier>.abstractFields := addToList(content)

Default:

abstract, description, bodytext

mapping

Used by: Elasticsearch connection while indexing.

Define mapping for Elasticsearch, have a look at the official docs: https://www.elastic.co/guide/en/elasticsearch/reference/5.2/mapping.html You are able to define the mapping for each property / columns.

Example:

plugin.tx_searchcore.settings.indexing.tt_content.mapping {
    CType {
        type = keyword
    }
}

The above example will define the CType field of tt_content as type: keyword. This makes building a facet possible.

index

Used by: Elasticsearch connection while indexing.

Define index for Elasticsearch, have a look at the official docs: https://www.elastic.co/guide/en/elasticsearch/reference/5.2/indices-create-index.html

Example:

plugin.tx_searchcore.settings.indexing.tt_content.index {
    analysis {
        analyzer {
            ngram4 {
                type = custom
                tokenizer = ngram4
                char_filter = html_strip
                filter = lowercase, asciifolding
            }
        }

        tokenizer {
            ngram4 {
                type = ngram
                min_gram = 4
                max_gram = 4
            }
        }
    }
}

char_filter and filter are a comma separated list of options.

dataProcessing

Used by: All connections while indexing.

Configure modifications on each document before sending it to the configured connection. Same as provided by TYPO3 for t3tsref:cobj-fluidtemplate through t3tsref:cobj-fluidtemplate-properties-dataprocessing.

All processors are applied in configured order. Allowing to work with already processed data.

Example:

plugin.tx_searchcore.settings.indexing.tt_content.dataProcessing {
    1 = Codappix\SearchCore\DataProcessing\CopyToProcessor
    1 {
        to = search_spellcheck
    }

    2 = Codappix\SearchCore\DataProcessing\CopyToProcessor
    2 {
        to = search_all
    }
}

The above example will copy all existing fields to the field search_spellcheck. Afterwards all fields, including search_spellcheck will be copied to search_all. E.g. used to index all information into a field for spellchecking or searching with different mapping.

The following Processor are available:

dataProcessing/CopyToProcessor dataProcessing/RemoveProcessor dataProcessing/GeoPointProcessor

The following Processor are planned:

Codappix\SearchCore\DataProcessing\ReplaceProcessor

Will execute a search and replace on configured fields.

Codappix\SearchCore\DataProcessing\RootLevelProcessor

Will attach the root level to the record.

Codappix\SearchCore\DataProcessing\ChannelProcessor

Will add a configurable channel to the record, e.g. if you have different areas in your website like "products" and "infos".

Codappix\SearchCore\DataProcessing\RelationResolverProcessor

Resolves all relations using the TCA.

Of course you are able to provide further processors. Just implement Codappix\SearchCore\DataProcessing\ProcessorInterface and use the FQCN (=Fully qualified class name) as done in the examples above.

By implementing also the same interface as necessary for TYPO3 t3tsref:cobj-fluidtemplate-properties-dataprocessing, you are able to reuse the same code also for Fluid to prepare the same record fetched from DB for your fluid.

Dependency injection is possible inside of processors, as we instantiate through extbase ObjectManager.