Solr Search for WordPress

Search is critical for your site, but the default search for WordPress leaves a lot to be desired. Improve your user experience with the Apache Solr search engine for your WordPress website.

• Fast results, with better accuracy.
• Enables faceting on fields such as tags, categories, author, and page type.
• Indexing and faceting on custom fields.
• Drop-in support for WP_Query with the solr_integrate parameter set to true.
• Completely replaces default WordPress search, just install and configure.
• Completely integrated into default WordPress theme and search widget.
• Very developer-friendly: uses the modern Solarium library

Development

This plugin is under active development on GitHub:

https://github.com/pantheon-systems/solr-power

Please feel free to file issues there. Pull requests are also welcome! See CONTRIBUTING.md for information on contributing.

For further documentation, such as available filters and working with the SolrPower_Api class directly, please see the project wiki:

https://github.com/pantheon-systems/solr-power/wiki

You may notice there are two sets of tests running, on two different services:

• PHPUnit test suite against a Solr instance.
• The Behat test suite against a Pantheon site, to ensure the plugin’s compatibility with the Pantheon platform.

Both of these test suites can be run locally, with a varying amount of setup.

PHPUnit requires the WordPress PHPUnit test suite, and access to a database with name wordpress_test. If you haven’t already configured the test suite locally, you can run bash bin/install-wp-tests.sh wordpress_test root '' localhost. You’ll also need access to a running Solr instance, in order to run the unit tests against Solr.

Behat requires a Pantheon site with Solr enabled. Once you’ve created the site, you’ll need install Terminus, and set the TERMINUS_TOKEN, TERMINUS_SITE, and TERMINUS_ENV environment variables. Then, you can run ./bin/behat-prepare.sh to prepare the site for the test suite.

Note that dependencies are installed via Composer and the vendor directory is not committed to the repository. You will need to run composer install locally for the plugin to function. You can read more about Composer here

WP-CLI Support

This plugin has WP-CLI support.

All Solr Power related commands are grouped into the wp solr command, see an example:

$ wp solr
usage: wp solr check-server-settings
   or: wp solr delete [<id>...] [--all]
   or: wp solr index [--batch=<batch>] [--batch_size=<size>] [--post_type=<post-type>]
   or: wp solr info [--field=<field>] [--format=<format>]
   or: wp solr optimize-index
   or: wp solr repost-schema
   or: wp solr stats [--field=<field>] [--format=<format>]

See 'wp help solr <command>' for more information on a specific command.

You can see more details about the commands using wp help solr:

**NAME**

  wp solr

**DESCRIPTION**

  Perform a variety of actions against your Solr instance.

**SYNOPSIS**

  wp solr <command>

**SUBCOMMANDS**

  check-server-settings      Check server settings.
  delete                     Remove one or more posts from the index.
  index                      Index all posts for a site.
  info                       Report information about Solr Power configuration.
  optimize-index             Optimize the Solr index.
  repost-schema              Repost schema.xml to Solr.
  stats                      Report stats about indexed content.<h3>WP_Query Integration</h3>

Use Solr in a custom WP_Query instead of querying a database. Add 'solr_integrate' => true to the query arguments.

NOTE: Currently, only basic queries, tax_query, meta_query and date_query are supported. See examples/example.custom_WP_Query.php for an example.

A meta_query can use the following compare operators:

• '='
• '!='
• '>'
• '>='
• '<'
• '<='
• 'LIKE'
• 'NOT LIKE'
• 'IN'
• 'NOT IN'
• 'BETWEEN'
• 'NOT BETWEEN'
• 'EXISTS'
• 'NOT EXISTS'

('REGEXP', 'NOT REGEXP', and 'RLIKE' are not supported.)

Configuration Tips

Searching by author name

To support searching by author name (e.g. where “Pantheon” would return posts authored by the “Pantheon” user), add the following to your custom schema.xml:

  <copyField source="post_author" dest="text"/>

Boosting relevancy score by publish date

The following guidance can be used to extend the Solr index and modify boosts beyond just this example.

To support math functions on dates, you must add a custom schema.xml to Solr and reindex with the new schema.

Add the following to schema.xml:

  <!-- Add to <types> -->
  <!-- See: https://lucene.apache.org/solr/6_2_0/solr-core/org/apache/solr/schema/TrieDateField.html -->
  <fieldType name="tdate" class="solr.TrieDateField" omitNorms="true" precisionStep="6" positionIncrementGap="0"/>

  <!-- Add to <fields> -->
  <field name="post_date_iso" type="tdate" indexed="true" stored="true" required="true" />

Add the following to your functions.php file.

  <?php
  /**
   * Hooks into the document build process to add post date field in proper format.
   */
  function my_solr_build_document( $doc, $post_info ) {
        $post_time = strtotime( $post_info->post_date );
        // Matches format required for TrieDateField
        $doc->setField( 'post_date_iso', gmdate( 'c\Z', $post_time ) );
        return $doc;
  }
  add_filter( 'solr_build_document', 'my_solr_build_document', 10, 2 );

  /**
   * Hooks into query processor, Dismax, to add publish date boost.
   * See: https://www.metaltoad.com/blog/date-boosting-solr-drupal-search-results
   */
  function my_solr_dismax_query( $dismax ) {
        $dismax->setQueryParser( 'edismax' );
        $dismax->setBoostQuery( 'recip(abs(ms(NOW/HOUR,post_date_iso),3.16e-11,1,1))' );
        return $dismax;
  }
  add_filter( 'solr_dismax_query', 'my_solr_dismax_query' );

Common issues

• Failing to post the schema.xml will result in an error during indexing, “Missing post_date_iso field.”
• If you have the field and type in the schema, but don’t add the solr_build_document filter, you will get a similar error.
• If the post_date_iso field is missing from the index, Solr will ignore this boost and return regular results.
• Trying to use a regular date field for the boost query will result in an error in the request instead of results.

Explicit Commit vs Autocommit

Once solr has sent the data to the solr server, solr must COMMIT the data to the index and adjust the index and relevancy ratings accordingly before that data can appear in search results.

By default, Solr Search for WordPress has auto-commit disabled. The index is committed when the uncommitted item is two minutes old, or the cron runs. By default, the cron runs on the Pantheon platform every hour.

When autocommit is enabled, Solr Search for WordPress commits data when it sends every post. When running on Pantheon, we recommend leaving autocommit disabled to aid overall site performance.

To enable autocommit, add the following to wp-config.php or an mu-plugin.

  define( 'SOLRPOWER_DISABLE_AUTOCOMMIT', false );

To force-commit data outside of a normal cron run, from the command line, you can run the command below or simply force a cron-run.

  wp solr commit