Search engine

Exponential has a built-in search engine that indexes the content as it is published. It searches the attributes that are marked searchable, with filters by class, attribute, subtree, section and date. A Solr-based engine, ezfind, can replace it when you need relevance ranking, highlighting or facets.

What is indexed

Each class attribute has a Searchable checkbox (Setup > Classes, edit the class). When an object is published, the text of its searchable attributes is added to the index. Some datatypes cannot be indexed; their checkbox has no effect.

Indexing happens during the publish unless it is delayed. With delayed indexing, the publish only queues the object and the cronjob part that runs cronjobs/indexcontent.php (in the default cronjob group) indexes it later:

[SearchSettings]
# enabled, disabled (the default) or classbased
DelayedIndexing=classbased
# with classbased: the classes whose objects are indexed later
DelayedIndexingClassList[]=article

Searching

• Simple search (content/search) finds objects that contain all the words typed; the order of the words does not matter.
• Advanced search (content/advancedsearch) adds an exact phrase field and filters by class, attribute, subtree, section and publishing date (yesterday, last week, last month, last three months, last year).
• The operators AND and OR are not understood; several words always mean all of them.

Settings

All in site.ini [SearchSettings]. The shipped values:

[SearchSettings]
# the search engine class; ezfind sets its own
SearchEngine=eZSearchEngine
# default: the view searches; template: the template does
SearchViewHandling=default
# record search phrases for Setup > Search statistics (they can hold personal data)
LogSearchStats=disabled
# the largest limit a visitor may ask for
MaximumSearchLimit=30
# longer search texts are cut to this many words and characters
MaximumSearchTextWords=16
MaximumSearchTextLength=1024
AllowEmptySearch=disabled
# "demo*" finds words that start with demo; costs much more time
EnableWildcard=false
MinCharacterWildcard=3
# stop words: words found in more than StopWordThresholdPercent % of the objects
# are left out of a search, once there are more than StopWordThresholdValue objects
StopWordThresholdValue=100
StopWordThresholdPercent=60
DelayedIndexing=disabled
DelayedIndexingClassList[]
ExtensionDirectories[]

With EnableWildcard=true the asterisk is accepted only at the end of a word (demo*), and the word before it must have at least MinCharacterWildcard characters.

Search statistics

Setup > Search statistics (/search/stats) lists the phrases visitors searched for, how often, how many results they found on average and which found nothing. Phrases are recorded only in siteaccesses that set LogSearchStats=enabled; the page names them. Reset statistics removes all recorded phrases.

Rebuilding the index

After changing which attributes are searchable, or after an import, rebuild the index:

# --clean removes all search data first
./console exp:updatesearchindex --clean --allow-root-user
# the same script
php bin/php/updatesearchindex.php --clean

Solr with ezfind

ezfind is not part of a new installation; Composer suggests it (composer require se7enxweb/ezfind:~3.0.11). Once it is installed and active it brings its own commands (ext:ezfind:updatesearchindexsolr, ext:ezfind:ezsolrcommit, ext:ezfind:ezsolroptimize). ./console exp:solr status controls a Solr server configured in solr.ini [SolrSettings]; see Server control commands.

Limits

• No relevance ranking and no highlighting in the built-in engine.
• Wildcards only at the end of a word, and only when switched on.
• Search statistics are off by default.

Related pages

• Links, URL aliases, wildcards and search statistics
• site.ini