class SearchApiSolrService in Search API Solr 7
Search service class using Solr server.
Hierarchy
- class \SearchApiAbstractService implements SearchApiServiceInterface
- class \SearchApiSolrService
Expanded class hierarchy of SearchApiSolrService
2 string references to 'SearchApiSolrService'
- search_api_solr_access_server_files in ./
search_api_solr.module - Access callback for a server's "Files" tab.
- search_api_solr_search_api_service_info in ./
search_api_solr.module - Implements hook_search_api_service_info().
File
- includes/
service.inc, line 6
View source
class SearchApiSolrService extends SearchApiAbstractService {
/**
* The date format that Solr uses, in PHP date() syntax.
*/
const SOLR_DATE_FORMAT = 'Y-m-d\\TH:i:s\\Z';
/**
* The connection class used by this service.
*
* Must implement SearchApiSolrConnectionInterface.
*
* @var string
*/
protected $connection_class = 'SearchApiSolrConnection';
/**
* A connection to the Solr server.
*
* @var SearchApiSolrConnectionInterface
*/
protected $solr;
/**
* Static cache for getFieldNames().
*
* @var array
*/
protected $fieldNames = array();
/**
* Metadata describing fields on the Solr/Lucene index.
*
* @see SearchApiSolrService::getFields().
*
* @var array
*/
protected $fields;
/**
* Saves whether a commit operation was already scheduled for this server.
*
* @var bool
*/
protected $commitScheduled = FALSE;
/**
* Request handler to use for this search query.
*
* @var string
*/
protected $request_handler = NULL;
/**
* {@inheritdoc}
*/
public function configurationForm(array $form, array &$form_state) {
if ($this->options) {
// Editing this server
$form['server_description'] = array(
'#type' => 'item',
'#title' => t('Solr server URI'),
'#description' => $this
->getServerLink(),
);
}
$options = $this->options + array(
'scheme' => 'http',
'host' => 'localhost',
'port' => '8983',
'path' => '/solr',
'http_user' => '',
'http_pass' => '',
'excerpt' => FALSE,
'retrieve_data' => FALSE,
'highlight_data' => FALSE,
'skip_schema_check' => FALSE,
'log_query' => FALSE,
'log_response' => FALSE,
'commits_disabled' => FALSE,
'solr_version' => '',
'http_method' => 'AUTO',
// Default to TRUE for new servers, but to FALSE for existing ones.
'clean_ids' => $this->options ? FALSE : TRUE,
'site_hash' => $this->options ? FALSE : TRUE,
'autocorrect_spell' => TRUE,
'autocorrect_suggest_words' => TRUE,
'highlight_prefix' => '',
'highlight_suffix' => '',
);
if (!$options['clean_ids']) {
if (module_exists('advanced_help')) {
$variables['@url'] = url('help/search_api_solr/README.txt');
}
else {
$variables['@url'] = url(drupal_get_path('module', 'search_api_solr') . '/README.txt');
}
$description = t('Change Solr field names to be more compatible with advanced features. Doing this leads to re-indexing of all indexes on this server. See <a href="@url">README.txt</a> for details.', $variables);
$form['clean_ids_form'] = array(
'#type' => 'fieldset',
'#title' => t('Clean field identifiers'),
'#description' => $description,
'#collapsible' => TRUE,
);
$form['clean_ids_form']['submit'] = array(
'#type' => 'submit',
'#value' => t('Switch to clean field identifiers'),
'#submit' => array(
'_search_api_solr_switch_to_clean_ids',
),
);
}
$form['clean_ids'] = array(
'#type' => 'value',
'#value' => $options['clean_ids'],
);
if (!$options['site_hash']) {
$description = t('If you want to index content from multiple sites on a single Solr server, you should enable the multi-site compatibility here. Note, however, that this will completely clear all search indexes (from this site) lying on this server. All content will have to be re-indexed.');
$form['site_hash_form'] = array(
'#type' => 'fieldset',
'#title' => t('Multi-site compatibility'),
'#description' => $description,
'#collapsible' => TRUE,
);
$form['site_hash_form']['submit'] = array(
'#type' => 'submit',
'#value' => t('Turn on multi-site compatibility and clear all indexes'),
'#submit' => array(
'_search_api_solr_switch_to_site_hash',
),
);
}
$form['site_hash'] = array(
'#type' => 'value',
'#value' => $options['site_hash'],
);
$form['scheme'] = array(
'#type' => 'select',
'#title' => t('HTTP protocol'),
'#description' => t('The HTTP protocol to use for sending queries.'),
'#default_value' => $options['scheme'],
'#options' => array(
'http' => 'http',
'https' => 'https',
),
);
$form['host'] = array(
'#type' => 'textfield',
'#title' => t('Solr host'),
'#description' => t('The host name or IP of your Solr server, e.g. <code>localhost</code> or <code>www.example.com</code>.'),
'#default_value' => $options['host'],
'#required' => TRUE,
);
$form['port'] = array(
'#type' => 'textfield',
'#title' => t('Solr port'),
'#description' => t('The Jetty example server is at port 8983, while Tomcat uses 8080 by default.'),
'#default_value' => $options['port'],
'#required' => TRUE,
);
$form['path'] = array(
'#type' => 'textfield',
'#title' => t('Solr path'),
'#description' => t('The path that identifies the Solr instance to use on the server. (For Solr versions 4 and above, this should include the name of the core to use.)'),
'#default_value' => $options['path'],
);
$form['http'] = array(
'#type' => 'fieldset',
'#title' => t('Basic HTTP authentication'),
'#description' => t('If your Solr server is protected by basic HTTP authentication, enter the login data here.'),
'#collapsible' => TRUE,
'#collapsed' => empty($options['http_user']),
);
$form['http']['http_user'] = array(
'#type' => 'textfield',
'#title' => t('Username'),
'#default_value' => $options['http_user'],
// This prefix with no-op text and password field will keep most browsers
// from autocompleting these fields, which is hardly ever what the user
// wants.
'#prefix' => '<input type="text" style="display:none" /><input type="password" style="display:none" />',
);
$form['http']['http_pass'] = array(
'#type' => 'password',
'#title' => t('Password'),
'#description' => t('If this field is left blank and the HTTP username is filled out, the current password will not be changed.'),
);
$form['advanced'] = array(
'#type' => 'fieldset',
'#title' => t('Advanced'),
'#collapsible' => TRUE,
'#collapsed' => TRUE,
);
$form['advanced']['excerpt'] = array(
'#type' => 'checkbox',
'#title' => t('Return an excerpt for all results'),
'#description' => t("If search keywords are given, use Solr's capabilities to create a highlighted search excerpt for each result. " . 'Whether the excerpts will actually be displayed depends on the settings of the search, though.'),
'#default_value' => $options['excerpt'],
);
// Make in-built highlighting from the solr server configurable.
$form['advanced']['highlight_prefix'] = array(
'#type' => 'textfield',
'#title' => t('Highlight prefix'),
'#default_value' => $options['highlight_prefix'],
'#description' => t('Change the prefix if using Solr to perform highlighting (not highlight processor). If empty, defaults to %tag', array(
'%tag' => '<strong>',
)),
'#states' => array(
'invisible' => array(
array(
':input[name="options[form][advanced][excerpt]"]' => array(
'checked' => FALSE,
),
),
),
),
);
$form['advanced']['highlight_suffix'] = array(
'#type' => 'textfield',
'#title' => t('Highlight suffix'),
'#default_value' => $options['highlight_suffix'],
'#description' => t('Change the suffix if using Solr to perform highlighting (not highlight processor). If empty, defaults to %tag', array(
'%tag' => '</strong>',
)),
'#states' => array(
'invisible' => array(
array(
':input[name="options[form][advanced][excerpt]"]' => array(
'checked' => FALSE,
),
),
),
),
);
$form['advanced']['retrieve_data'] = array(
'#type' => 'checkbox',
'#title' => t('Retrieve result data from Solr'),
'#description' => t('When checked, result data will be retrieved directly from the Solr server. ' . 'This might make item loads unnecessary. Only indexed fields can be retrieved. ' . 'Note also that the returned field data might not always be correct, due to preprocessing and caching issues.'),
'#default_value' => $options['retrieve_data'],
);
$form['advanced']['highlight_data'] = array(
'#type' => 'checkbox',
'#title' => t('Highlight retrieved data'),
'#description' => t('When retrieving result data from the Solr server, try to highlight the search terms in the returned fulltext fields. <strong>Note:</strong> Do not use the "Highlighting" processor for the index together with this option – use one or the other.'),
'#default_value' => $options['highlight_data'],
);
// Highlighting retrieved data only makes sense when we retrieve data.
// (Actually, internally it doesn't really matter. However, from a user's
// perspective, having to check both probably makes sense.)
$form['advanced']['highlight_data']['#states']['invisible'][':input[name="options[form][advanced][retrieve_data]"]']['checked'] = FALSE;
$form['advanced']['skip_schema_check'] = array(
'#type' => 'checkbox',
'#title' => t('Skip schema verification'),
'#description' => t('Skip the automatic check for schema-compatibility. Use this override if you are seeing an error-message about an incompatible schema.xml configuration file, and you are sure the configuration is compatible.'),
'#default_value' => $options['skip_schema_check'],
);
$form['advanced']['solr_version'] = array(
'#type' => 'select',
'#title' => t('Solr version override'),
'#description' => t('Specify the Solr version manually in case it cannot be retrieved automatically. The version can be found in the Solr admin interface under "Solr Specification Version" or "solr-spec".'),
'#options' => array(
'' => t('Determine automatically'),
'3' => '3.x',
'4' => '4.x',
'5' => '5.x',
),
'#default_value' => $options['solr_version'],
);
$form['advanced']['http_method'] = array(
'#type' => 'select',
'#title' => t('HTTP method'),
'#description' => t('The HTTP method to use for sending queries. GET will often fail with larger queries, while POST should not be cached. AUTO will use GET when possible, and POST for queries that are too large.'),
'#default_value' => $options['http_method'],
'#options' => array(
'AUTO' => t('AUTO'),
'POST' => 'POST',
'GET' => 'GET',
),
);
$form['advanced']['log_query'] = array(
'#type' => 'checkbox',
'#title' => t('Log search requests'),
'#description' => t('Log all outgoing Solr search requests.'),
'#default_value' => $options['log_query'],
);
$form['advanced']['log_response'] = array(
'#type' => 'checkbox',
'#title' => t('Log search results'),
'#description' => t('Log all search result responses received from Solr. NOTE: This may slow down your site since all response data (including possible retrieved data) will be saved in the Drupal log.'),
'#default_value' => $options['log_response'],
);
$form['advanced']['commits_disabled'] = array(
'#type' => 'checkbox',
'#title' => t('Disable explicit committing'),
'#description' => t('Do not send any commit commands to the Solr server.'),
'#default_value' => $options['commits_disabled'],
);
if (module_exists('search_api_autocomplete')) {
$form['advanced']['autocomplete'] = array(
'#type' => 'fieldset',
'#title' => t('Autocomplete'),
'#collapsible' => TRUE,
'#collapsed' => TRUE,
);
$form['advanced']['autocomplete']['autocorrect_spell'] = array(
'#type' => 'checkbox',
'#title' => t('Use spellcheck for autocomplete suggestions'),
'#description' => t('If activated, spellcheck suggestions ("Did you mean") will be included in the autocomplete suggestions. Since the used dictionary contains words from all indexes, this might lead to leaking of sensitive data, depending on your setup.'),
'#default_value' => $options['autocorrect_spell'],
);
$form['advanced']['autocomplete']['autocorrect_suggest_words'] = array(
'#type' => 'checkbox',
'#title' => t('Suggest additional words'),
'#description' => t('If activated and the user enters a complete word, Solr will suggest additional words the user wants to search, which are often found (not searched!) together. This has been known to lead to strange results in some configurations – if you see inappropriate additional-word suggestions, you might want to deactivate this option.'),
'#default_value' => $options['autocorrect_suggest_words'],
);
}
return $form;
}
/**
* {@inheritdoc}
*/
public function configurationFormValidate(array $form, array &$values, array &$form_state) {
if (isset($values['port']) && (!is_numeric($values['port']) || $values['port'] < 0 || $values['port'] > 65535)) {
form_error($form['port'], t('The port has to be an integer between 0 and 65535.'));
}
}
/**
* {@inheritdoc}
*/
public function configurationFormSubmit(array $form, array &$values, array &$form_state) {
// Since the form is nested into another, we can't simply use #parents for
// doing this array restructuring magic. (At least not without creating an
// unnecessary dependency on internal implementation.)
$values += $values['http'];
$values += $values['advanced'];
$values += !empty($values['autocomplete']) ? $values['autocomplete'] : array();
unset($values['http'], $values['advanced'], $values['autocomplete']);
// Highlighting retrieved data only makes sense when we retrieve data.
$values['highlight_data'] &= $values['retrieve_data'];
// For password fields, there is no default value, they're empty by default.
// Therefore we ignore empty submissions if the user didn't change either.
if ($values['http_pass'] === '' && isset($this->options['http_user']) && $values['http_user'] === $this->options['http_user']) {
$values['http_pass'] = $this->options['http_pass'];
}
parent::configurationFormSubmit($form, $values, $form_state);
}
/**
* {@inheritdoc}
*/
public function supportsFeature($feature) {
// First, check the features we always support.
$supported = drupal_map_assoc(array(
'search_api_autocomplete',
'search_api_between',
'search_api_facets',
'search_api_facets_operator_or',
'search_api_grouping',
'search_api_mlt',
'search_api_multi',
'search_api_service_extra',
'search_api_spellcheck',
'search_api_data_type_location',
'search_api_data_type_geohash',
'search_api_random_sort',
));
if (isset($supported[$feature])) {
return TRUE;
}
// If it is a custom data type, maybe we support it automatically via
// search_api_solr_hook_search_api_data_type_info().
if (substr($feature, 0, 21) != 'search_api_data_type_') {
return FALSE;
}
$type = substr($feature, 21);
$type = search_api_get_data_type_info($type);
// We only support it if the "prefix" key is set.
return $type && !empty($type['prefix']);
}
/**
* Overrides SearchApiAbstractService::viewSettings().
*
* Returns an empty string since information is instead added via
* getExtraInformation().
*/
public function viewSettings() {
return '';
}
/**
* {@inheritdoc}
*/
public function getExtraInformation() {
$info = array();
$info[] = array(
'label' => t('Solr server URI'),
'info' => $this
->getServerLink(),
);
if ($this->options['http_user']) {
$vars = array(
'@user' => $this->options['http_user'],
'@pass' => str_repeat('*', strlen($this->options['http_pass'])),
);
$http = t('Username: @user; Password: @pass', $vars);
$info[] = array(
'label' => t('Basic HTTP authentication'),
'info' => $http,
);
}
if ($this->server->enabled) {
// If the server is enabled, check whether Solr can be reached.
$ping = $this
->ping();
if ($ping) {
$msg = t('The Solr server could be reached (latency: @millisecs ms).', array(
'@millisecs' => $ping * 1000,
));
}
else {
$msg = t('The Solr server could not be reached. Further data is therefore unavailable.');
}
$info[] = array(
'label' => t('Connection'),
'info' => $msg,
'status' => $ping ? 'ok' : 'error',
);
if ($ping) {
try {
// If Solr can be reached, provide more information. This isn't done
// often (only when an admin views the server details), so we clear the
// cache to get the current data.
$this
->connect();
$this->solr
->clearCache();
$data = $this->solr
->getLuke();
if (isset($data->index->numDocs)) {
// Collect the stats
$stats_summary = $this->solr
->getStatsSummary();
$pending_msg = $stats_summary['@pending_docs'] ? t('(@pending_docs sent but not yet processed)', $stats_summary) : '';
$index_msg = $stats_summary['@index_size'] ? t('(@index_size on disk)', $stats_summary) : '';
$indexed_message = t('@num items !pending !index_msg', array(
'@num' => $data->index->numDocs,
'!pending' => $pending_msg,
'!index_msg' => $index_msg,
));
$info[] = array(
'label' => t('Indexed'),
'info' => $indexed_message,
);
if (!empty($stats_summary['@deletes_total'])) {
$info[] = array(
'label' => t('Pending Deletions'),
'info' => $stats_summary['@deletes_total'],
);
}
$info[] = array(
'label' => t('Delay'),
'info' => t('@autocommit_time before updates are processed.', $stats_summary),
);
$status = 'ok';
if (empty($this->options['skip_schema_check'])) {
if (substr($stats_summary['@schema_version'], 0, 10) == 'search-api') {
drupal_set_message(t('Your schema.xml version is too old. Please replace all configuration files with the ones packaged with this module and re-index you data.'), 'error');
$status = 'error';
}
elseif (substr($stats_summary['@schema_version'], 0, 9) != 'drupal-4.') {
$variables['@url'] = url('https://www.drupal.org/node/1999310');
$message = t('You are using an incompatible schema.xml configuration file. Please follow the instructions in <a href="@url">the handbook</a> for setting up Solr.', $variables);
drupal_set_message($message, 'error');
$status = 'error';
}
}
$info[] = array(
'label' => t('Schema'),
'info' => $stats_summary['@schema_version'],
'status' => $status,
);
if (!empty($stats_summary['@core_name'])) {
$info[] = array(
'label' => t('Solr Core Name'),
'info' => $stats_summary['@core_name'],
);
}
}
} catch (SearchApiException $e) {
$info[] = array(
'label' => t('Additional information'),
'info' => t('An error occurred while trying to retrieve additional information from the Solr server: @msg.', array(
'@msg' => $e
->getMessage(),
)),
'status' => 'error',
);
}
}
}
return $info;
}
/**
* Returns a link to the Solr server, if the necessary options are set.
*/
public function getServerLink() {
if (!$this->options) {
return '';
}
$host = $this->options['host'];
if ($host == 'localhost' && !empty($_SERVER['SERVER_NAME'])) {
$host = $_SERVER['SERVER_NAME'];
}
$url = $this->options['scheme'] . '://' . $host . ':' . $this->options['port'] . $this->options['path'];
return l($url, $url);
}
/**
* Create a connection to the Solr server as configured in $this->options.
*/
protected function connect() {
if (!$this->solr) {
$connection_class = $this
->getConnectionClass();
if (!class_exists($connection_class)) {
throw new SearchApiException(t('Invalid class @class set as Solr connection class.', array(
'@class' => $connection_class,
)));
}
$options = $this->options + array(
'server' => $this->server->machine_name,
);
$this->solr = new $connection_class($options);
if (!$this->solr instanceof SearchApiSolrConnectionInterface) {
$this->solr = NULL;
throw new SearchApiException(t('Invalid class @class set as Solr connection class.', array(
'@class' => $connection_class,
)));
}
}
}
/**
* {@inheritdoc}
*/
public function addIndex(SearchApiIndex $index) {
if (module_exists('search_api_multi') && module_exists('search_api_views')) {
views_invalidate_cache();
}
}
/**
* {@inheritdoc}
*/
public function fieldsUpdated(SearchApiIndex $index) {
if (module_exists('search_api_multi') && module_exists('search_api_views')) {
views_invalidate_cache();
}
// Find out if anything changed enough to need re-indexing.
$old_fields = isset($index->original->options['fields']) ? $index->original->options['fields'] : array();
$new_fields = isset($index->options['fields']) ? $index->options['fields'] : array();
if (!$old_fields && !$new_fields) {
return FALSE;
}
if (array_diff_key($old_fields, $new_fields) || array_diff_key($new_fields, $old_fields)) {
return TRUE;
}
$old_field_names = $this
->getFieldNames($index->original, TRUE);
$new_field_names = $this
->getFieldNames($index, TRUE);
return $old_field_names != $new_field_names;
}
/**
* {@inheritdoc}
*/
public function removeIndex($index) {
if (module_exists('search_api_multi') && module_exists('search_api_views')) {
views_invalidate_cache();
}
parent::removeIndex($index);
}
/**
* {@inheritdoc}
*/
public function indexItems(SearchApiIndex $index, array $items) {
$documents = array();
$ret = array();
$index_id = $this
->getIndexId($index->machine_name);
$fields = $this
->getFieldNames($index);
$languages = language_list();
$base_urls = array();
foreach ($items as $id => $item) {
$doc = new SearchApiSolrDocument();
$doc
->setField('id', $this
->createId($index_id, $id));
$doc
->setField('index_id', $index_id);
$doc
->setField('item_id', $id);
// If multi-site compatibility is enabled, add the site hash and
// language-specific base URL.
if (!empty($this->options['site_hash'])) {
$doc
->setField('hash', search_api_solr_site_hash());
$lang = $item['search_api_language']['value'];
if (empty($base_urls[$lang])) {
$url_options = array(
'absolute' => TRUE,
);
if (isset($languages[$lang])) {
$url_options['language'] = $languages[$lang];
}
$base_urls[$lang] = url(NULL, $url_options);
}
$doc
->setField('site', $base_urls[$lang]);
}
// Now add all fields contained in the item, with dynamic fields. Also,
// gather the contents of all text fields to also add them to "content".
$text_content = array();
foreach ($item as $key => $field) {
// If the field is not known for the index, something weird has
// happened. We refuse to index the items and hope that the others are
// OK.
if (!isset($fields[$key])) {
$type = search_api_get_item_type_info($index->item_type);
$vars = array(
'@field' => $key,
'@type' => $type ? $type['name'] : $index->item_type,
'@id' => $id,
);
watchdog('search_api_solr', 'Error while indexing: Unknown field @field set for @type with ID @id.', $vars, WATCHDOG_WARNING);
$doc = NULL;
break;
}
$text_content[] = $this
->addIndexField($doc, $fields[$key], $field['value'], $field['type']);
}
$doc
->setField('content', implode("\n\n", array_filter($text_content)));
if ($doc) {
$documents[] = $doc;
$ret[] = $id;
}
}
// Let other modules alter documents before sending them to solr.
drupal_alter('search_api_solr_documents', $documents, $index, $items);
$this
->alterSolrDocuments($documents, $index, $items);
if (!$documents) {
return array();
}
try {
$this
->connect();
$this->solr
->addDocuments($documents);
if (!empty($index->options['index_directly'])) {
$this
->scheduleCommit();
}
return $ret;
} catch (SearchApiException $e) {
watchdog_exception('search_api_solr', $e, "%type while indexing: !message in %function (line %line of %file).");
}
return array();
}
/**
* Creates an ID used as the unique identifier at the Solr server.
*
* This has to consist of both index and item ID. Optionally, the site hash is
* also included.
*
* @param string $index_id
* The search index's machine name.
* @param mixed $item_id
* The Search API item ID of the item.
*
* @return string
* The Solr ID to use for this item.
*
* @see search_api_solr_site_hash()
*/
public function createId($index_id, $item_id) {
$site_hash = !empty($this->options['site_hash']) ? search_api_solr_site_hash() . '-' : '';
return "{$site_hash}{$index_id}-{$item_id}";
}
/**
* Create a list of all indexed field names mapped to their Solr field names.
*
* The special fields "search_api_id" and "search_api_relevance" are also
* included. Any Solr fields that exist on search results are mapped back to
* to their local field names in the final result set.
*
* @see SearchApiSolrService::search()
*/
public function getFieldNames(SearchApiIndex $index, $reset = FALSE) {
if (!isset($this->fieldNames[$index->machine_name]) || $reset) {
// This array maps "local property name" => "solr doc property name".
$ret = array(
'search_api_id' => 'item_id',
'search_api_relevance' => 'score',
'search_api_random' => 'random',
);
// Add the names of any fields configured on the index.
$fields = isset($index->options['fields']) ? $index->options['fields'] : array();
foreach ($fields as $key => $field) {
// Generate a field name; this corresponds with naming conventions in
// our schema.xml
$type = $field['type'];
// Use the real type of the field if the server supports this type.
if (isset($field['real_type'])) {
$custom_type = search_api_extract_inner_type($field['real_type']);
if ($this
->supportsFeature('search_api_data_type_' . $custom_type)) {
$type = $field['real_type'];
}
}
$inner_type = search_api_extract_inner_type($type);
$type_info = search_api_solr_get_data_type_info($inner_type);
$pref = isset($type_info['prefix']) ? $type_info['prefix'] : '';
if (empty($type_info['always multiValued'])) {
$pref .= $type == $inner_type ? 's' : 'm';
}
if (!empty($this->options['clean_ids'])) {
$name = $pref . '_' . str_replace(':', '$', $key);
}
else {
$name = $pref . '_' . $key;
}
$ret[$key] = $name;
}
// Let modules adjust the field mappings.
drupal_alter('search_api_solr_field_mapping', $index, $ret);
$this->fieldNames[$index->machine_name] = $ret;
}
return $this->fieldNames[$index->machine_name];
}
/**
* Helper method for indexing.
*
* Adds $value with field name $key to the document $doc. The format of $value
* is the same as specified in SearchApiServiceInterface::indexItems().
*/
protected function addIndexField(SearchApiSolrDocument $doc, $key, $value, $type, $multi_valued = FALSE) {
$text_content = '';
// Don't index empty values (i.e., when field is missing).
if (!isset($value)) {
return $text_content;
}
if (search_api_is_list_type($type)) {
$type = substr($type, 5, -1);
foreach ($value as $v) {
$text_content .= $this
->addIndexField($doc, $key, $v, $type, TRUE) . "\n\n";
}
return trim($text_content);
}
switch ($type) {
case 'tokens':
foreach ($value as $v) {
$text_content .= $v['value'] . ' ';
$doc
->addField($key, $v['value']);
}
return trim($text_content);
case 'boolean':
$value = $value ? 'true' : 'false';
break;
case 'date':
$value = is_numeric($value) ? (int) $value : strtotime($value);
if ($value === FALSE) {
return $text_content;
}
$value = format_date($value, 'custom', self::SOLR_DATE_FORMAT, 'UTC');
break;
case 'integer':
$value = (int) $value;
break;
case 'decimal':
$value = (double) $value;
break;
}
if ($multi_valued) {
$doc
->addField($key, $value);
}
else {
$doc
->setField($key, $value);
}
if (search_api_is_text_type($type)) {
$text_content = $value;
}
return $text_content;
}
/**
* Applies custom modifications to indexed Solr documents.
*
* This method allows subclasses to easily apply custom changes before the
* documents are sent to Solr. The method is empty by default.
*
* @param array $documents
* An array of SearchApiSolrDocument objects ready to be indexed, generated
* from $items array.
* @param SearchApiIndex $index
* The search index for which items are being indexed.
* @param array $items
* An array of items being indexed.
*
* @see hook_search_api_solr_documents_alter()
*/
protected function alterSolrDocuments(array &$documents, SearchApiIndex $index, array $items) {
}
/**
* Implements SearchApiServiceInterface::deleteItems().
*
* This method has a custom, Solr-specific extension:
*
* If $ids is a string other than "all", it is treated as a Solr query. All
* items matching that Solr query are then deleted. If $index is additionally
* specified, then only those items also lying on that index will be deleted.
*
* It is up to the caller to ensure $ids is a valid query when the method is
* called in this fashion.
*
* @param array|string $ids
* Either an array containing the ids of the items that should be deleted,
* or 'all' if all items should be deleted. Other formats might be
* recognized by implementing classes, but these are not standardized.
* @param SearchApiIndex $index
* The index from which items should be deleted, or NULL if all indexes on
* this server should be cleared (then, $ids has to be 'all').
*
* @throws SearchApiException
* If an error occurred while trying to delete the items.
*/
public function deleteItems($ids = 'all', SearchApiIndex $index = NULL) {
$this
->connect();
if (is_array($ids)) {
$index_id = $this
->getIndexId($index->machine_name);
$solr_ids = array();
foreach ($ids as $id) {
$solr_ids[] = $this
->createId($index_id, $id);
}
$this->solr
->deleteByMultipleIds($solr_ids);
}
else {
$query = array();
if ($index) {
$index_id = $this
->getIndexId($index->machine_name);
$index_id = call_user_func(array(
$this
->getConnectionClass(),
'phrase',
), $index_id);
$query[] = "index_id:{$index_id}";
}
if (!empty($this->options['site_hash'])) {
// We don't need to escape the site hash, as that consists only of
// alphanumeric characters.
$query[] = 'hash:' . search_api_solr_site_hash();
}
if ($ids != 'all') {
$query[] = $query ? "({$ids})" : $ids;
}
$this->solr
->deleteByQuery($query ? implode(' AND ', $query) : '*:*');
}
$this
->scheduleCommit();
}
/**
* {@inheritdoc}
*/
public function search(SearchApiQueryInterface $query) {
$time_method_called = microtime(TRUE);
// Reset request handler.
$this->request_handler = NULL;
// Get field information.
$index = $query
->getIndex();
$index_id = $this
->getIndexId($index->machine_name);
$fields = $this
->getFieldNames($index);
// Get Solr connection.
$this
->connect();
$version = $this->solr
->getSolrVersion();
// Extract keys.
$keys = $query
->getKeys();
if (is_array($keys)) {
$keys = $this
->flattenKeys($keys);
}
// Set searched fields.
$options = $query
->getOptions();
$search_fields = $this
->getQueryFields($query);
// Get the index fields to be able to retrieve boosts.
$index_fields = $index
->getFields() + array(
'search_api_relevance' => array(
'type' => 'decimal',
'indexed' => TRUE,
),
'search_api_id' => array(
'type' => 'integer',
'indexed' => TRUE,
),
);
$qf = array();
foreach ($search_fields as $f) {
$boost = isset($index_fields[$f]['boost']) ? '^' . $index_fields[$f]['boost'] : '';
$qf[] = $fields[$f] . $boost;
}
// Extract filters.
$filter = $query
->getFilter();
$fq = $this
->createFilterQueries($filter, $fields, $index->options['fields']);
$fq[] = 'index_id:' . call_user_func(array(
$this
->getConnectionClass(),
'phrase',
), $index_id);
if (!empty($this->options['site_hash'])) {
// We don't need to escape the site hash, as that consists only of
// alphanumeric characters.
$fq[] = 'hash:' . search_api_solr_site_hash();
}
// Extract sort.
$sort = array();
foreach ($query
->getSort() as $field => $order) {
$f = $fields[$field];
if (substr($f, 0, 3) == 'ss_') {
$f = 'sort_' . substr($f, 3);
}
// The default Solr schema provides a virtual field named "random_SEED"
// that can be used to randomly sort the results; the field is available
// only at query-time.
if ($field == 'search_api_random') {
$params = $query
->getOption('search_api_random_sort', array());
// Random seed: getting the value from parameters or computing a new one.
$seed = !empty($params['seed']) ? $params['seed'] : mt_rand();
$f = 'random_' . $seed;
}
$order = strtolower($order);
$sort[$field] = "{$f} {$order}";
}
// Get facet fields.
$facets = $query
->getOption('search_api_facets', array());
$facet_params = $this
->getFacetParams($facets, $fields, $fq);
// Handle highlighting.
$highlight_params = $this
->getHighlightParams($query);
// Handle More Like This query.
$mlt = $query
->getOption('search_api_mlt');
if ($mlt) {
$mlt_params['qt'] = 'mlt';
// The fields to look for similarities in.
$mlt_fl = array();
// Solr 4 (before 4.6) has a bug which results in numeric fields not being
// supported in MLT queries.
$mlt_no_numeric_fields = FALSE;
if ($version == 4) {
$system_info = $this->solr
->getSystemInfo();
$mlt_no_numeric_fields = !isset($system_info->lucene->{'solr-spec-version'}) || version_compare($system_info->lucene->{'solr-spec-version'}, '4.6.0', '<');
}
foreach ($mlt['fields'] as $f) {
// Date fields don't seem to be supported at all.
if ($fields[$f][0] === 'd' || $mlt_no_numeric_fields && in_array($fields[$f][0], array(
'i',
'f',
))) {
continue;
}
$mlt_fl[] = $fields[$f];
// For non-text fields, set minimum word length to 0.
if (isset($index->options['fields'][$f]['type']) && !search_api_is_text_type($index->options['fields'][$f]['type'])) {
$mlt_params['f.' . $fields[$f] . '.mlt.minwl'] = 0;
}
}
$mlt_params['mlt.fl'] = implode(',', $mlt_fl);
$id = $this
->createId($index_id, $mlt['id']);
$id = call_user_func(array(
$this
->getConnectionClass(),
'phrase',
), $id);
$keys = 'id:' . $id;
// In (early versions of) Solr 5, facets aren't supported with MLT.
if ($version >= 5) {
$facet_params = array();
}
}
// Handle spatial filters.
if ($spatials = $query
->getOption('search_api_location')) {
foreach ($spatials as $i => $spatial) {
// Spatial options all need a field to do anything.
if (!isset($spatial['field'])) {
continue;
}
unset($radius);
$field = $fields[$spatial['field']];
$escaped_field = call_user_func(array(
$this
->getConnectionClass(),
'escapeFieldName',
), $field);
// If proper bbox coordinates were given use them to filter.
if (isset($spatial['bbox'])) {
if ($version >= 4) {
$bbox = $spatial['bbox'];
$fq[] = $escaped_field . ':[' . (double) $bbox['bottom'] . ',' . (double) $bbox['left'] . ' TO ' . (double) $bbox['top'] . ',' . (double) $bbox['right'] . ']';
}
else {
$warnings[] = t('Filtering by a bounding box is not supported in Solr versions below 4.');
}
}
// Everything other than a bounding box filter requires a point, so stop
// here (for this option) if "lat" and "lon" aren't both set.
if (!isset($spatial['lat']) || !isset($spatial['lon'])) {
continue;
}
$point = (double) $spatial['lat'] . ',' . (double) $spatial['lon'];
// Prepare the filter settings.
if (isset($spatial['radius'])) {
$radius = (double) $spatial['radius'];
}
$spatial_method = 'geofilt';
if (isset($spatial['method']) && in_array($spatial['method'], array(
'geofilt',
'bbox',
))) {
$spatial_method = $spatial['method'];
}
// Change the fq facet ranges to the correct fq.
foreach ($fq as $key => $value) {
// If the fq consists only of a filter on this field, replace it with
// a range.
$preg_field = preg_quote($escaped_field, '/');
if (preg_match('/^(?:\\{!tag[^}]+\\})?' . $preg_field . ':\\["?(\\*|\\d+(?:\\.\\d+)?)"? TO "?(\\*|\\d+(?:\\.\\d+)?)"?\\]$/', $value, $m)) {
unset($fq[$key]);
if ($m[1] && is_numeric($m[1])) {
$min_radius = isset($min_radius) ? max($min_radius, $m[1]) : $m[1];
}
if (is_numeric($m[2])) {
// Make the radius tighter accordingly.
$radius = isset($radius) ? min($radius, $m[2]) : $m[2];
}
}
}
// If either a radius was given in the option, or a filter was
// encountered, set a filter for the lowest value. If a lower boundary
// was set (too), we can only set a filter for that if the field name
// doesn't contains any colons.
if (isset($min_radius) && strpos($field, ':') === FALSE) {
$upper = isset($radius) ? " u={$radius}" : '';
$fq[] = "{!frange l={$min_radius}{$upper}}geodist({$field},{$point})";
}
elseif (isset($radius)) {
$fq[] = "{!{$spatial_method} pt={$point} sfield={$field} d={$radius}}";
}
// Change sort on the field, if set (and not already changed).
if (isset($sort[$spatial['field']]) && substr($sort[$spatial['field']], 0, strlen($field)) === $field) {
if (strpos($field, ':') === FALSE) {
$sort[$spatial['field']] = str_replace($field, "geodist({$field},{$point})", $sort[$spatial['field']]);
}
else {
$link = l(t('edit server'), 'admin/config/search/search_api/server/' . $this->server->machine_name . '/edit');
watchdog('search_api_solr', 'Location sort on field @field had to be ignored because unclean field identifiers are used.', array(
'@field' => $spatial['field'],
), WATCHDOG_WARNING, $link);
}
}
// Add parameters to fetch distance, if requested.
if (!empty($spatial['distance']) && $version >= 4) {
if (strpos($field, ':') === FALSE) {
// Add pseudofield with the distance to the result items.
$location_fields[] = '_' . $field . '_distance_:geodist(' . $field . ',' . $point . ')';
}
else {
$link = l(t('edit server'), 'admin/config/search/search_api/server/' . $this->server->machine_name . '/edit');
watchdog('search_api_solr', "Location distance information can't be added because unclean field identifiers are used.", array(), WATCHDOG_WARNING, $link);
}
}
// Change the facet parameters for spatial fields to return distance
// facets.
if (!empty($facets)) {
if (!empty($facet_params['facet.field'])) {
$facet_params['facet.field'] = array_diff($facet_params['facet.field'], array(
$field,
));
}
foreach ($facets as $delta => $facet) {
if ($facet['field'] != $spatial['field']) {
continue;
}
$steps = $facet['limit'] > 0 ? $facet['limit'] : 5;
$step = (isset($radius) ? $radius : 100) / $steps;
for ($k = $steps - 1; $k > 0; --$k) {
$distance = $step * $k;
$key = "spatial-{$delta}-{$distance}";
$facet_params['facet.query'][] = "{!{$spatial_method} pt={$point} sfield={$field} d={$distance} key={$key}}";
}
foreach (array(
'limit',
'mincount',
'missing',
) as $setting) {
unset($facet_params["f.{$field}.facet.{$setting}"]);
}
}
}
}
}
// Normal sorting on location fields isn't possible.
foreach ($sort as $field => $sort_param) {
if (substr($sort_param, 0, 3) === 'loc') {
unset($sort[$field]);
}
}
// Handle field collapsing / grouping.
$grouping = $query
->getOption('search_api_grouping');
if (!empty($grouping['use_grouping'])) {
$group_params['group'] = 'true';
// We always want the number of groups returned so that we get pagers done
// right.
$group_params['group.ngroups'] = 'true';
if (!empty($grouping['truncate'])) {
$group_params['group.truncate'] = 'true';
}
if (!empty($grouping['group_facet'])) {
$group_params['group.facet'] = 'true';
}
foreach ($grouping['fields'] as $collapse_field) {
$type = $index_fields[$collapse_field]['type'];
// Only single-valued fields are supported.
if ($version < 4) {
// For Solr 3.x, only string and boolean fields are supported.
if (search_api_is_list_type($type) || !search_api_is_text_type($type, array(
'string',
'boolean',
'uri',
))) {
$warnings[] = t('Grouping is not supported for field @field. ' . 'Only single-valued fields of type "String", "Boolean" or "URI" are supported.', array(
'@field' => $index_fields[$collapse_field]['name'],
));
continue;
}
}
else {
if (search_api_is_list_type($type) || search_api_is_text_type($type)) {
$warnings[] = t('Grouping is not supported for field @field. ' . 'Only single-valued fields not indexed as "Fulltext" are supported.', array(
'@field' => $index_fields[$collapse_field]['name'],
));
continue;
}
}
$group_params['group.field'][] = $fields[$collapse_field];
}
if (empty($group_params['group.field'])) {
unset($group_params);
}
else {
if (!empty($grouping['group_sort'])) {
foreach ($grouping['group_sort'] as $group_sort_field => $order) {
if (isset($fields[$group_sort_field])) {
$f = $fields[$group_sort_field];
if (substr($f, 0, 3) == 'ss_') {
$f = 'sort_' . substr($f, 3);
}
// The default Solr schema provides a virtual field named
// "random_SEED" that can be used to randomly sort the results;
// the field is available only at query-time.
if ($group_sort_field == 'search_api_random') {
$params = $query
->getOption('search_api_random_sort', array());
// Random seed: getting the value from parameters or computing a
// new one.
$seed = !empty($params['seed']) ? $params['seed'] : mt_rand();
$f = 'random_' . $seed;
}
$order = strtolower($order);
$group_params['group.sort'][] = $f . ' ' . $order;
}
}
if (!empty($group_params['group.sort'])) {
$group_params['group.sort'] = implode(', ', $group_params['group.sort']);
}
}
if (!empty($grouping['group_limit']) && $grouping['group_limit'] != 1) {
$group_params['group.limit'] = $grouping['group_limit'];
}
}
}
// Set defaults.
if (!$keys) {
$keys = NULL;
}
// Collect parameters.
$params = array(
'fl' => 'item_id,score',
'qf' => $qf,
'fq' => $fq,
);
if (isset($options['offset'])) {
$params['start'] = $options['offset'];
}
$params['rows'] = isset($options['limit']) ? $options['limit'] : 1000000;
if ($sort) {
$params['sort'] = implode(', ', $sort);
}
if (!empty($facet_params['facet.field'])) {
$params += $facet_params;
}
if (!empty($highlight_params)) {
$params += $highlight_params;
}
if (!empty($options['search_api_spellcheck'])) {
$params['spellcheck'] = 'true';
}
if (!empty($mlt_params['mlt.fl'])) {
$params += $mlt_params;
}
if (!empty($group_params)) {
$params += $group_params;
}
if (!empty($this->options['retrieve_data'])) {
$params['fl'] = '*,score';
}
if (!empty($location_fields)) {
$params['fl'] .= ',' . implode(',', $location_fields);
}
// Retrieve http method from server options.
$http_method = !empty($this->options['http_method']) ? $this->options['http_method'] : 'AUTO';
$call_args = array(
'query' => &$keys,
'params' => &$params,
'http_method' => &$http_method,
);
if ($this->request_handler) {
$this
->setRequestHandler($this->request_handler, $call_args);
}
try {
// Send search request.
$time_processing_done = microtime(TRUE);
drupal_alter('search_api_solr_query', $call_args, $query);
$this
->preQuery($call_args, $query);
$response = $this->solr
->search($keys, $params, $http_method);
$time_query_done = microtime(TRUE);
// Extract results.
$results = $this
->extractResults($query, $response);
// Add warnings, if present.
if (!empty($warnings)) {
$results['warnings'] = isset($results['warnings']) ? array_merge($warnings, $results['warnings']) : $warnings;
}
// Extract facets.
if ($facets = $this
->extractFacets($query, $response)) {
$results['search_api_facets'] = $facets;
}
drupal_alter('search_api_solr_search_results', $results, $query, $response);
$this
->postQuery($results, $query, $response);
// Compute performance.
$time_end = microtime(TRUE);
$results['performance'] = array(
'complete' => $time_end - $time_method_called,
'preprocessing' => $time_processing_done - $time_method_called,
'execution' => $time_query_done - $time_processing_done,
'postprocessing' => $time_end - $time_query_done,
);
return $results;
} catch (SearchApiException $e) {
throw new SearchApiException(t('An error occurred while trying to search with Solr: @msg.', array(
'@msg' => $e
->getMessage(),
)));
}
}
/**
* Extract results from a Solr response.
*
* @param object $response
* A HTTP response object.
*
* @return array
* An array with two keys:
* - result count: The number of total results.
* - results: An array of search results, as specified by
* SearchApiQueryInterface::execute().
*/
protected function extractResults(SearchApiQueryInterface $query, $response) {
$index = $query
->getIndex();
$fields = $this
->getFieldNames($index);
$field_options = $index->options['fields'];
$version = $this->solr
->getSolrVersion();
// Set up the results array.
$results = array();
$results['results'] = array();
// Keep a copy of the response in the results so it's possible to extract
// further useful information out of it, if necessary.
$results['search_api_solr_response'] = $response;
// In some rare cases (e.g., MLT query with nonexistent ID) the response
// will be NULL.
if (!isset($response->response) && !isset($response->grouped)) {
$results['result count'] = 0;
return $results;
}
// If field collapsing has been enabled for this query, we need to process
// the results differently.
$grouping = $query
->getOption('search_api_grouping');
if (!empty($grouping['use_grouping']) && !empty($response->grouped)) {
$docs = array();
$results['result count'] = 0;
foreach ($grouping['fields'] as $field) {
if (!empty($response->grouped->{$fields[$field]})) {
$results['result count'] += $response->grouped->{$fields[$field]}->ngroups;
foreach ($response->grouped->{$fields[$field]}->groups as $group) {
foreach ($group->doclist->docs as $doc) {
$docs[] = $doc;
}
}
}
}
}
else {
$results['result count'] = $response->response->numFound;
$docs = $response->response->docs;
}
$spatials = $query
->getOption('search_api_location');
// Add each search result to the results array.
foreach ($docs as $doc) {
// Blank result array.
$result = array(
'id' => NULL,
'score' => NULL,
'fields' => array(),
);
// Extract properties from the Solr document, translating from Solr to
// Search API property names. This reverses the mapping in
// SearchApiSolrService::getFieldNames().
foreach ($fields as $search_api_property => $solr_property) {
if (isset($doc->{$solr_property})) {
$value = $doc->{$solr_property};
// Date fields need some special treatment to become valid date values
// (i.e., timestamps) again.
$first_value = $value;
while (is_array($first_value)) {
$first_value = reset($first_value);
}
if (isset($field_options[$search_api_property]['type']) && search_api_extract_inner_type($field_options[$search_api_property]['type']) === 'date' && preg_match('/^\\d{4}-\\d{2}-\\d{2}T\\d{2}:\\d{2}:\\d{2}Z$/', $first_value)) {
$value = is_array($value) ? array_map('strtotime', $value) : strtotime($value);
}
$result['fields'][$search_api_property] = $value;
}
}
// We can find the item id and score in the special 'search_api_*'
// properties. Mappings are provided for these properties in
// SearchApiSolrService::getFieldNames().
$result['id'] = $result['fields']['search_api_id'];
$result['score'] = $result['fields']['search_api_relevance'];
// If location based search is enabled ensure the calculated distance is
// set to the appropriate field. If the calculation wasn't possible add
// the coordinates to allow calculation.
if ($spatials) {
foreach ($spatials as $spatial) {
if (isset($spatial['field']) && !empty($spatial['distance'])) {
if ($version >= 4) {
$doc_field = '_' . $fields[$spatial['field']] . '_distance_';
if (!empty($doc->{$doc_field})) {
$results['search_api_location'][$spatial['field']][$result['id']]['distance'] = $doc->{$doc_field};
}
}
}
}
}
$index_id = $this
->getIndexId($index->machine_name);
$solr_id = $this
->createId($index_id, $result['id']);
$excerpt = $this
->getExcerpt($response, $solr_id, $result['fields'], $fields, $this
->getQueryFields($query));
if ($excerpt) {
$result['excerpt'] = $excerpt;
}
// Use the result's id as the array key. By default, 'id' is mapped to
// 'item_id' in SearchApiSolrService::getFieldNames().
if ($result['id']) {
$results['results'][$result['id']] = $result;
}
}
// Check for spellcheck suggestions.
if (module_exists('search_api_spellcheck') && $query
->getOption('search_api_spellcheck')) {
$results['search_api_spellcheck'] = new SearchApiSpellcheckSolr($response);
}
return $results;
}
/**
* Extract and format highlighting information for a specific item from a Solr response.
*
* Will also use highlighted fields to replace retrieved field data, if the
* corresponding option is set.
*/
protected function getExcerpt($response, $id, array &$fields, array $field_mapping, array $fulltext_fields) {
if (!isset($response->highlighting->{$id})) {
return FALSE;
}
$output = '';
$highlighting = $response->highlighting->{$id};
$highlight_fields = !empty($this->options['highlight_data']);
$create_excerpt = !empty($this->options['excerpt']);
if (!$highlight_fields && !$create_excerpt) {
return FALSE;
}
// Collect highlighted field values for the excerpt and set them in the
// field values, if requested.
$excerpt_parts = array();
$field_mapping = array_flip($field_mapping);
$fulltext_fields = drupal_map_assoc($fulltext_fields);
foreach ($highlighting as $solr_property => $values) {
$values = (array) $values;
if (empty($field_mapping[$solr_property])) {
continue;
}
$search_api_property = $field_mapping[$solr_property];
// Only use fields that were actually searched for the excerpt.
if (isset($fulltext_fields[$search_api_property])) {
// Remember the highlighted value so we can use it for the excerpt, if
// requested.
$excerpt_parts = array_merge($excerpt_parts, $values);
}
if (!$highlight_fields) {
continue;
}
$values = $this
->sanitizeHighlightValue($values, $search_api_property);
// Remove highlight prefixes and suffixes so we can compare values in
// order to replace the corresponding items.
$orig_values = preg_replace('#\\[(/?)HIGHLIGHT\\]#', '', $values);
$field_values = array();
if (!empty($fields[$search_api_property])) {
$field_values = $this
->sanitizeHighlightValue($fields[$search_api_property]);
}
foreach ($field_values as $delta => $field_value) {
foreach ($orig_values as $num => $item) {
if ($item === $field_value) {
$field_values[$delta] = $this
->formatHighlighting($values[$num]);
$change = TRUE;
continue 2;
}
}
}
if (!empty($change)) {
$fields[$search_api_property] = array(
'#value' => $field_values,
'#sanitize_callback' => FALSE,
);
}
}
// Create an excerpt, if requested.
if ($create_excerpt && $excerpt_parts) {
$excerpt = array();
$excerpt_length = 0;
foreach ($excerpt_parts as $value) {
// Excerpts don't have HTML (except for the highlighting tags, of
// course).
$value = strip_tags($value);
foreach ($this
->extractHighlightingSnippets($value) as $snippet) {
$excerpt[] = $snippet;
$excerpt_length += drupal_strlen($snippet);
// Restrict ourselves to three snippets or 300 characters.
if (count($excerpt) >= 3 || $excerpt_length >= 300) {
break 2;
}
}
}
if ($excerpt) {
$output = implode(' … ', $excerpt) . ' …';
}
}
return $output;
}
/**
* Sanitizes a highlighted field value.
*
* @param string|array $value
* Either a highlighted field value, or an array of such values.
* @param string|null $field_id
* (optional) The ID of the field for which this sanitizing occurs, if any.
*
* @return string|array
* The sanitized input.
*/
protected function sanitizeHighlightValue($value, $field_id = NULL) {
if (is_array($value)) {
foreach ($value as $i => $nested_value) {
$value[$i] = $this
->sanitizeHighlightValue($nested_value, $field_id);
}
return $value;
}
return check_plain(strip_tags($value));
}
/**
* Changes highlighting tags from our custom, HTML-safe ones to HTML.
*
* @param string|string[] $snippet
* The snippet(s) to format.
*
* @return string|string[]
* The snippet(s), properly formatted as HTML.
*/
protected function formatHighlighting($snippet) {
$search = array(
'[HIGHLIGHT]',
'[/HIGHLIGHT]',
);
return str_replace($search, $this
->getHighlightingPrefixSuffix(), $snippet);
}
/**
* Returns the prefix and suffix for highlighting matches in the excerpt.
*
* @return string[]
* An array of 2 items, prefix and suffix.
*/
protected function getHighlightingPrefixSuffix() {
$prefix = '<strong>';
$suffix = '</strong>';
if (!empty($this->options['highlight_prefix'])) {
$prefix = $this->options['highlight_prefix'];
}
if (!empty($this->options['highlight_suffix'])) {
$suffix = $this->options['highlight_suffix'];
}
return array(
$prefix,
$suffix,
);
}
/**
* Extracts short snippets with highlighting from highlighted field values.
*
* @param string $value
* A highlighted field value.
*
* @return string[]
* An array of short, highlighted snippets extracted from the field value.
*/
protected function extractHighlightingSnippets($value) {
$parts = preg_split('#\\[/?HIGHLIGHT\\]#', $value);
$num_parts = count($parts);
if ($num_parts < 3) {
return array();
}
$snippets = array();
$snippet = '';
$combined_length = 0;
foreach ($parts as $i => $part) {
// Is this a match (even) or context (odd)?
if ($i % 2 === 1) {
$snippet .= '[HIGHLIGHT]' . $part . '[/HIGHLIGHT]';
continue;
}
// If there is less than 60 characters between them, we want to fuse two
// snippets.
if ($snippet && drupal_strlen($part) < 60) {
$snippet .= $part;
continue;
}
// Add a suffix context to the existing snippet.
if ($snippet) {
$space = strpos($part, ' ', 25);
// Fall back to just cutting at an arbitrary position, space or no.
if ($space === FALSE) {
$space = 30;
}
$snippet .= ' ' . substr($part, 0, $space);
$combined_length += drupal_strlen($snippet);
$snippets[] = $this
->sanitizeAndFormatExcerptSnippet($snippet);
$snippet = '';
// Restrict ourselves to three snippets or 300 characters.
if (count($snippets) >= 3 || $combined_length >= 300) {
break;
}
$part = substr($part, $space);
}
// If there are no more matches, stop.
if ($num_parts <= $i + 1) {
break;
}
// Otherwise, prepare a new prefix for the next match.
$length = drupal_strlen($part);
if ($length > 30) {
$space = strrpos(substr($part, 0, -25), ' ');
// Fall back to just cutting at an arbitrary position, space or no.
if ($space === FALSE) {
$space = $length - 30;
}
$part = substr($part, $space + 1);
}
$snippet = $part;
}
if ($snippet) {
$snippets[] = $this
->sanitizeAndFormatExcerptSnippet($snippet);
}
return $snippets;
}
/**
* Extract facets from a Solr response.
*
* @param object $response
* A response object from SolrPhpClient.
*
* @return array
* An array describing facets that apply to the current results.
*/
protected function extractFacets(SearchApiQueryInterface $query, $response) {
$facets = array();
if (!isset($response->facet_counts)) {
return $facets;
}
$index = $query
->getIndex();
$fields = $this
->getFieldNames($index);
$extract_facets = $query
->getOption('search_api_facets', array());
$facet_fields = array();
$facet_queries = array();
if (isset($response->facet_counts->facet_fields)) {
$facet_fields = $response->facet_counts->facet_fields;
}
if (isset($response->facet_counts->facet_queries)) {
$facet_queries = $response->facet_counts->facet_queries;
}
// The key for the "missing" facet (empty string in the JSON). This will
// be either "" or "_empty_", depending on the PHP version.
$empty_key = key((array) json_decode('{"":5}'));
foreach ($extract_facets as $delta => $info) {
$field = $fields[$info['field']];
// Fields that have solr facet queries set need special handling.
if (!empty($info['solr_facet_query'])) {
if (!empty($facet_queries)) {
foreach ($facet_queries as $term => $count) {
// Strip a leading local param so we can correctly detect the
// field. E.g. "{!ex=...}" might have been set in getFacetParams()
// for OR facets.
$term = preg_replace('/^{[^}]+}/', '', $term);
if (strpos($term, $field . ':') === 0) {
$term = substr($term, strlen($field) + 1);
if (!preg_match('/^(?:[(\\[][^ ]+ .*[)\\]]|".*"|!)$/', $term)) {
$term = "\"{$term}\"";
}
$facets[$delta][] = array(
'filter' => $term,
'count' => $count,
'solr_facet_query' => TRUE,
);
}
}
}
}
elseif (!empty($facet_fields->{$field})) {
$min_count = $info['min_count'];
$terms = $facet_fields->{$field};
if ($info['missing']) {
// We have to correctly incorporate the "missing" term ($empty_key).
// This will ensure that the term with the least results is dropped,
// if the limit would be exceeded.
if (isset($terms->{$empty_key})) {
if ($terms->{$empty_key} < $min_count) {
unset($terms->{$empty_key});
}
else {
$terms = (array) $terms;
arsort($terms);
if ($info['limit'] > 0 && count($terms) > $info['limit']) {
array_pop($terms);
}
}
}
}
elseif (isset($terms->{$empty_key})) {
$terms = clone $terms;
unset($terms->{$empty_key});
}
$type = isset($index->options['fields'][$info['field']]['type']) ? search_api_extract_inner_type($index->options['fields'][$info['field']]['type']) : 'string';
foreach ($terms as $term => $count) {
if ($count >= $min_count) {
if ($term === $empty_key) {
$term = '!';
}
elseif ($type == 'boolean') {
if ($term == 'true') {
$term = '"1"';
}
elseif ($term == 'false') {
$term = '"0"';
}
}
elseif ($type == 'date') {
$term = $term ? '"' . strtotime($term) . '"' : NULL;
}
else {
$term = "\"{$term}\"";
}
if ($term) {
$facets[$delta][] = array(
'filter' => $term,
'count' => $count,
);
}
}
}
if (empty($facets[$delta])) {
unset($facets[$delta]);
}
}
}
if ($facet_queries && $query
->getOption('search_api_location')) {
foreach ($facet_queries as $key => $count) {
if (!preg_match('/^spatial-(.*)-(\\d+(?:\\.\\d+)?)$/', $key, $m)) {
continue;
}
if (empty($extract_facets[$m[1]])) {
continue;
}
$facet = $extract_facets[$m[1]];
if ($count >= $facet['min_count']) {
$facets[$m[1]][] = array(
'filter' => "[* {$m[2]}]",
'count' => $count,
);
}
}
}
return $facets;
}
/**
* Flatten a keys array into a single search string.
*
* @param array $keys
* The keys array to flatten, formatted as specified by
* SearchApiQueryInterface::getKeys().
*
* @return string
* A Solr query string representing the same keys.
*/
protected function flattenKeys(array $keys) {
$k = array();
$or = $keys['#conjunction'] == 'OR';
$neg = !empty($keys['#negation']);
foreach (element_children($keys) as $i) {
$key = $keys[$i];
if (!$key) {
continue;
}
if (is_array($key)) {
$subkeys = $this
->flattenKeys($key);
if ($subkeys) {
$nested_expressions = TRUE;
// If this is a negated OR expression, we can't just use nested keys
// as-is, but have to put them into parantheses.
if ($or && $neg) {
$subkeys = "({$subkeys})";
}
$k[] = $subkeys;
}
}
else {
$key = trim($key);
$key = call_user_func(array(
$this
->getConnectionClass(),
'phrase',
), $key);
$k[] = $key;
}
}
if (!$k) {
return '';
}
// Formatting the keys into a Solr query can be a bit complex. The following
// code will produce filters that look like this:
//
// #conjunction | #negation | return value
// ----------------------------------------------------------------
// AND | FALSE | A B C
// AND | TRUE | -(A AND B AND C)
// OR | FALSE | ((A) OR (B) OR (C))
// OR | TRUE | -A -B -C
// If there was just a single, unnested key, we can ignore all this.
if (count($k) == 1 && empty($nested_expressions)) {
$k = reset($k);
return $neg ? "*:* AND -{$k}" : $k;
}
if ($or) {
if ($neg) {
return '*:* AND -' . implode(' AND -', $k);
}
return '((' . implode(') OR (', $k) . '))';
}
$k = implode(' AND ', $k);
return $neg ? "*:* AND -({$k})" : $k;
}
/**
* Transforms a query filter into a flat array of Solr filter queries, using
* the field names in $fields.
*/
protected function createFilterQueries(SearchApiQueryFilterInterface $filter, array $solr_fields, array $fields) {
$or = $filter
->getConjunction() == 'OR';
$fq = array();
$prefix = '';
foreach ($filter
->getFilters() as $f) {
if (is_array($f)) {
if (!isset($fields[$f[0]])) {
throw new SearchApiException(t('Filter term on unknown or unindexed field @field.', array(
'@field' => $f[0],
)));
}
if ($f[1] !== '') {
$fq[] = $this
->createFilterQuery($solr_fields[$f[0]], $f[1], $f[2], $fields[$f[0]]);
}
}
elseif ($f instanceof SearchApiQueryFilterInterface) {
$q = $this
->createFilterQueries($f, $solr_fields, $fields);
if ($filter
->getConjunction() != $f
->getConjunction() && count($q) > 1) {
$fq[] = '((' . implode(') ' . $f
->getConjunction() . ' (', $q) . '))';
}
else {
$fq = array_merge($fq, $q);
}
}
}
if (method_exists($filter, 'getTags')) {
foreach ($filter
->getTags() as $tag) {
$prefix = "{!tag={$tag}}";
// We can only apply one tag per filter.
break;
}
}
if ($or && count($fq) > 1) {
$fq = array(
'((' . implode(') OR (', $fq) . '))',
);
}
if ($prefix) {
foreach ($fq as $i => $filter) {
$fq[$i] = $prefix . $filter;
}
}
return $fq;
}
/**
* Create a single search query string according to the given field, value
* and operator.
*/
protected function createFilterQuery($field, $value, $operator, $field_info) {
$field = call_user_func(array(
$this
->getConnectionClass(),
'escapeFieldName',
), $field);
// Special handling for location fields.
if (isset($field_info['real_type']) && $field_info['real_type'] == 'location') {
// Empty / non-empty comparison has to take place in one of the subfields
// of the location field type. These subfields are usually generated with
// the index and the field type as name suffix.
// @TODO Do we need to handle other operators / values too?
if ($value === NULL) {
$field .= '_0___tdouble';
}
}
if ($value === NULL) {
return ($operator == '=' ? '*:* AND -' : '') . "{$field}:[* TO *]";
}
$type = search_api_extract_inner_type($field_info['type']);
if (!is_array($value)) {
$value = $this
->formatFilterValue($value, $type);
}
else {
foreach ($value as &$val) {
$val = $this
->formatFilterValue($val, $type);
}
unset($val);
}
switch (strtoupper($operator)) {
case '<>':
return "*:* AND -({$field}:{$value})";
case '<':
return "{$field}:{* TO {$value}}";
case '<=':
return "{$field}:[* TO {$value}]";
case '>=':
return "{$field}:[{$value} TO *]";
case '>':
return "{$field}:{{$value} TO *}";
case 'BETWEEN':
return "{$field}:[{$value[0]} TO {$value[1]}]";
case 'NOT BETWEEN':
return "*:* AND -{$field}:[{$value[0]} TO {$value[1]}]";
default:
return "{$field}:{$value}";
}
}
/**
* Format a value for filtering on a field of a specific type.
*/
protected function formatFilterValue($value, $type) {
switch ($type) {
case 'boolean':
$value = $value ? 'true' : 'false';
break;
case 'date':
$value = is_numeric($value) ? (int) $value : strtotime($value);
if ($value === FALSE) {
return 0;
}
$value = format_date($value, 'custom', self::SOLR_DATE_FORMAT, 'UTC');
break;
case 'text':
return '(' . call_user_func(array(
$this
->getConnectionClass(),
'escape',
), $value) . ')';
}
return call_user_func(array(
$this
->getConnectionClass(),
'phrase',
), $value);
}
/**
* Helper method for creating the facet field parameters.
*/
protected function getFacetParams(array $facets, array $fields, array &$fq = array()) {
if (!$facets) {
return array();
}
$facet_params['facet'] = 'true';
$facet_params['facet.sort'] = 'count';
$facet_params['facet.limit'] = 10;
$facet_params['facet.mincount'] = 1;
$facet_params['facet.missing'] = 'false';
foreach ($facets as $info) {
if (empty($fields[$info['field']])) {
continue;
}
// String fields have their own corresponding facet fields.
$field = $fields[$info['field']];
// Check for the "or" operator.
if (isset($info['operator']) && $info['operator'] === 'or') {
// This tag should automatically be placed on any filters created via
// this filter by the Facet API integration. We here use it to exclude
// those filters from the logic creating OR facet filters.
$tag = 'facet:' . $info['field'];
// Check whether Solr facet queries were set for this facet and use
// those.
if (!empty($info['solr_facet_query'])) {
foreach ($info['solr_facet_query'] as $expression) {
$facet_params['facet.query'][] = "{!ex={$tag}}{$field}:{$expression}";
}
}
else {
$facet_params['facet.field'][] = "{!ex={$tag}}{$field}";
}
}
elseif (!empty($info['solr_facet_query'])) {
// No tagging and excluding like above is necessary as facets are built
// with all filters applied.
foreach ($info['solr_facet_query'] as $expression) {
$facet_params['facet.query'][] = "{$field}:{$expression}";
}
}
else {
// Add the facet field.
$facet_params['facet.field'][] = $field;
}
// Set limit, unless it's the default.
if ($info['limit'] != 10) {
$facet_params["f.{$field}.facet.limit"] = $info['limit'] ? $info['limit'] : -1;
}
// Set mincount, unless it's the default.
if ($info['min_count'] != 1) {
$facet_params["f.{$field}.facet.mincount"] = $info['min_count'];
}
// Set missing, if specified.
if ($info['missing']) {
$facet_params["f.{$field}.facet.missing"] = 'true';
}
}
return $facet_params;
}
/**
* Helper method for creating the highlighting parameters.
*
* (The $query parameter currently isn't used and only here for the potential
* sake of subclasses.)
*
* @param SearchApiQueryInterface|SearchApiMultiQueryInterface $query
* The query object, either for a normal Search API query or a multi-index
* query.
*
* @return array
* An array of parameters to be added to the Solr search request.
*/
protected function getHighlightParams($query) {
$highlight_params = array();
if (!empty($this->options['excerpt']) || !empty($this->options['highlight_data'])) {
$highlight_params['hl'] = 'true';
$highlight_params['hl.fl'] = variable_get('search_api_solr_highlight_prefix', 'tm_') . '*';
$highlight_params['hl.simple.pre'] = '[HIGHLIGHT]';
$highlight_params['hl.simple.post'] = '[/HIGHLIGHT]';
$highlight_params['hl.snippets'] = 1;
$highlight_params['hl.fragsize'] = 0;
}
return $highlight_params;
}
/**
* Sets the request handler.
*
* This should also make the needed adjustments to the request parameters.
*
* @param $handler
* Name of the handler to set.
* @param array $call_args
* An associative array containing all three arguments to the
* SearchApiSolrConnectionInterface::search() call ("query", "params" and
* "method") as references.
*
* @return bool
* TRUE iff this method invocation handled the given handler. This allows
* subclasses to recognize whether the request handler was already set by
* this method.
*/
protected function setRequestHandler($handler, array &$call_args) {
if ($handler == 'pinkPony') {
$call_args['params']['qt'] = $handler;
return TRUE;
}
return FALSE;
}
/**
* Empty method called before sending a search query to Solr.
*
* This allows subclasses to apply custom changes before the query is sent to
* Solr. Works exactly like hook_search_api_solr_query_alter().
*
* @param array $call_args
* An associative array containing all three arguments to the
* SearchApiSolrConnectionInterface::search() call ("query", "params" and
* "method") as references.
* @param SearchApiQueryInterface $query
* The SearchApiQueryInterface object representing the executed search query.
*/
protected function preQuery(array &$call_args, SearchApiQueryInterface $query) {
}
/**
* Empty method to allow subclasses to apply custom changes before search results are returned.
*
* Works exactly like hook_search_api_solr_search_results_alter().
*
* @param array $results
* The results array that will be returned for the search.
* @param SearchApiQueryInterface $query
* The SearchApiQueryInterface object representing the executed search query.
* @param object $response
* The response object returned by Solr.
*/
protected function postQuery(array &$results, SearchApiQueryInterface $query, $response) {
}
//
// Autocompletion feature
//
/**
* Implements SearchApiAutocompleteInterface::getAutocompleteSuggestions().
*/
// Largely copied from the apachesolr_autocomplete module.
public function getAutocompleteSuggestions(SearchApiQueryInterface $query, SearchApiAutocompleteSearch $search, $incomplete_key, $user_input) {
$suggestions = array();
// Reset request handler
$this->request_handler = NULL;
// Turn inputs to lower case, otherwise we get case sensivity problems.
$incomp = drupal_strtolower($incomplete_key);
$index = $query
->getIndex();
$fields = $this
->getFieldNames($index);
$complete = $query
->getOriginalKeys();
// Extract keys
$keys = $query
->getKeys();
if (is_array($keys)) {
$keys_array = array();
while ($keys) {
reset($keys);
if (!element_child(key($keys))) {
array_shift($keys);
continue;
}
$key = array_shift($keys);
if (is_array($key)) {
$keys = array_merge($keys, $key);
}
else {
$keys_array[$key] = $key;
}
}
$keys = $this
->flattenKeys($query
->getKeys());
}
else {
$keys_array = drupal_map_assoc(preg_split('/[-\\s():{}\\[\\]\\\\"]+/', $keys, -1, PREG_SPLIT_NO_EMPTY));
}
if (!$keys) {
$keys = NULL;
}
// Set searched fields
$search_fields = $this
->getQueryFields($query);
$qf = array();
foreach ($search_fields as $f) {
$qf[] = $fields[$f];
}
// Extract filters
$fq = $this
->createFilterQueries($query
->getFilter(), $fields, $index->options['fields']);
$index_id = $this
->getIndexId($index->machine_name);
$fq[] = 'index_id:' . call_user_func(array(
$this
->getConnectionClass(),
'phrase',
), $index_id);
if (!empty($this->options['site_hash'])) {
// We don't need to escape the site hash, as that consists only of
// alphanumeric characters.
$fq[] = 'hash:' . search_api_solr_site_hash();
}
// Autocomplete magic
$facet_fields = array();
foreach ($search_fields as $f) {
$facet_fields[] = $fields[$f];
}
$limit = $query
->getOption('limit', 10);
$params = array(
'qf' => $qf,
'fq' => $fq,
'rows' => 0,
'facet' => 'true',
'facet.field' => $facet_fields,
'facet.prefix' => $incomp,
'facet.limit' => $limit * 5,
'facet.mincount' => 1,
'spellcheck' => !isset($this->options['autocorrect_spell']) || $this->options['autocorrect_spell'] ? 'true' : 'false',
'spellcheck.count' => 1,
);
// Retrieve http method from server options.
$http_method = !empty($this->options['http_method']) ? $this->options['http_method'] : 'AUTO';
$call_args = array(
'query' => &$keys,
'params' => &$params,
'http_method' => &$http_method,
);
if ($this->request_handler) {
$this
->setRequestHandler($this->request_handler, $call_args);
}
$second_pass = !isset($this->options['autocorrect_suggest_words']) || $this->options['autocorrect_suggest_words'];
$alter_data = array(
'search' => $search,
'query' => $query,
'incomplete_key' => $incomplete_key,
'user_input' => $user_input,
);
for ($i = 0; $i < ($second_pass ? 2 : 1); ++$i) {
try {
// Send search request
$this
->connect();
drupal_alter('search_api_solr_query', $call_args, $query);
$this
->preQuery($call_args, $query);
$response = $this->solr
->search($keys, $params, $http_method);
$alter_data['responses'][] = $response;
if (!empty($response->spellcheck->suggestions)) {
$replace = array();
foreach ($response->spellcheck->suggestions as $word => $data) {
$replace[$word] = $data->suggestion[0];
}
$corrected = str_ireplace(array_keys($replace), array_values($replace), $user_input);
if ($corrected != $user_input) {
array_unshift($suggestions, array(
'prefix' => t('Did you mean') . ':',
'user_input' => $corrected,
));
}
}
$matches = array();
if (isset($response->facet_counts->facet_fields)) {
foreach ($response->facet_counts->facet_fields as $terms) {
foreach ($terms as $term => $count) {
if (isset($matches[$term])) {
// If we just add the result counts, we can easily get over the
// total number of results if terms appear in multiple fields.
// Therefore, we just take the highest value from any field.
$matches[$term] = max($matches[$term], $count);
}
else {
$matches[$term] = $count;
}
}
}
if ($matches) {
// Eliminate suggestions that are too short or already in the query.
foreach ($matches as $term => $count) {
if (strlen($term) < 3 || isset($keys_array[$term])) {
unset($matches[$term]);
}
}
// Don't suggest terms that are too frequent (by default in more
// than 90% of results).
$result_count = $response->response->numFound;
$max_occurrences = $result_count * variable_get('search_api_solr_autocomplete_max_occurrences', 0.9);
if (($max_occurrences >= 1 || $i > 0) && $max_occurrences < $result_count) {
foreach ($matches as $match => $count) {
if ($count > $max_occurrences) {
unset($matches[$match]);
}
}
}
// The $count in this array is actually a score. We want the
// highest ones first.
arsort($matches);
// Shorten the array to the right ones.
$additional_matches = array_slice($matches, $limit - count($suggestions), NULL, TRUE);
$matches = array_slice($matches, 0, $limit, TRUE);
// Build suggestions using returned facets
$incomp_length = strlen($incomp);
foreach ($matches as $term => $count) {
if (drupal_strtolower(substr($term, 0, $incomp_length)) == $incomp) {
$suggestions[] = array(
'suggestion_suffix' => substr($term, $incomp_length),
'term' => $term,
'results' => $count,
);
}
else {
$suggestions[] = array(
'suggestion_suffix' => ' ' . $term,
'term' => $term,
'results' => $count,
);
}
}
}
}
} catch (SearchApiException $e) {
watchdog_exception('search_api_solr', $e, "%type during autocomplete Solr query: !message in %function (line %line of %file).", array(), WATCHDOG_WARNING);
}
if (count($suggestions) >= $limit) {
break;
}
// Change parameters for second query.
unset($params['facet.prefix']);
$keys = trim($keys . ' ' . $incomplete_key);
}
drupal_alter('search_api_solr_autocomplete_suggestions', $suggestions, $alter_data);
return $suggestions;
}
//
// SearchApiMultiServiceInterface methods
//
/**
* Implements SearchApiMultiServiceInterface::queryMultiple().
*/
public function queryMultiple(array $options = array()) {
return search_api_multi_query($options);
}
/**
* Implements SearchApiMultiServiceInterface::searchMultiple().
*/
public function searchMultiple(SearchApiMultiQueryInterface $query) {
$time_method_called = microtime(TRUE);
// Get field information
$solr_fields = array(
'search_api_id' => 'item_id',
'search_api_relevance' => 'score',
'search_api_multi_index' => 'index_id',
);
$fields = array(
'search_api_multi_index' => array(
'type' => 'string',
),
);
foreach ($query
->getIndexes() as $index) {
if (empty($index->options['fields'])) {
continue;
}
$prefix = $this
->getIndexId($index->machine_name) . ':';
foreach ($this
->getFieldNames($index) as $field => $key) {
if (!isset($solr_fields[$field])) {
$solr_fields[$prefix . $field] = $key;
}
}
foreach ($index->options['fields'] as $field => $info) {
$fields[$prefix . $field] = $info;
}
}
// Extract keys
$keys = $query
->getKeys();
if (is_array($keys)) {
$keys = $this
->flattenKeys($keys);
}
// Set searched fields
$search_fields = $query
->getFields();
$qf = array();
foreach ($search_fields as $f) {
$boost = isset($fields[$f]['boost']) ? '^' . $fields[$f]['boost'] : '';
$qf[] = $solr_fields[$f] . $boost;
}
// Extract filters
$filter = $query
->getFilter();
$fq = $this
->createFilterQueries($filter, $solr_fields, $fields);
// Restrict search to searched indexes.
$index_filter = array();
$indexes = array();
foreach ($query
->getIndexes() as $index) {
$index_id = $this
->getIndexId($index->machine_name);
$indexes[$index_id] = $index;
$index_filter[] = 'index_id:' . call_user_func(array(
$this
->getConnectionClass(),
'phrase',
), $index_id);
}
$fq[] = implode(' OR ', $index_filter);
if (!empty($this->options['site_hash'])) {
// We don't need to escape the site hash, as that consists only of
// alphanumeric characters.
$fq[] = 'hash:' . search_api_solr_site_hash();
}
// Extract sort
$sort = array();
foreach ($query
->getSort() as $f => $order) {
$f = $solr_fields[$f];
if (substr($f, 0, 3) == 'ss_') {
$f = 'sort_' . substr($f, 3);
}
$order = strtolower($order);
$sort[] = "{$f} {$order}";
}
// Get facet fields
$facets = $query
->getOption('search_api_facets') ? $query
->getOption('search_api_facets') : array();
$facet_params = $this
->getFacetParams($facets, $solr_fields, $fq);
// Handle highlighting.
$highlight_params = $this
->getHighlightParams($query);
// Set defaults
if (!$keys) {
$keys = NULL;
}
$options = $query
->getOptions();
// Collect parameters
$params = array(
'fl' => 'item_id,index_id,score',
'qf' => $qf,
'fq' => $fq,
);
if (isset($options['offset'])) {
$params['start'] = $options['offset'];
}
if (isset($options['limit'])) {
$params['rows'] = $options['limit'];
}
if ($sort) {
$params['sort'] = implode(', ', $sort);
}
if (!empty($facet_params['facet.field'])) {
$params += $facet_params;
}
if (!empty($highlight_params)) {
$params += $highlight_params;
}
if (!empty($this->options['retrieve_data'])) {
$params['fl'] = '*,score';
}
// Retrieve http method from server options.
$http_method = !empty($this->options['http_method']) ? $this->options['http_method'] : 'AUTO';
// Send search request
$time_processing_done = microtime(TRUE);
$this
->connect();
$call_args = array(
'query' => &$keys,
'params' => &$params,
'http_method' => &$http_method,
);
drupal_alter('search_api_solr_multi_query', $call_args, $query);
$response = $this->solr
->search($keys, $params, $http_method);
$time_query_done = microtime(TRUE);
// Extract results
$results = array();
$results['result count'] = $response->response->numFound;
$results['results'] = array();
$fulltext_fields_by_index = array();
foreach ($search_fields as $field) {
list($index_id, $field) = explode(':', $field, 2);
$fulltext_fields_by_index[$index_id][] = $field;
}
foreach ($response->response->docs as $id => $doc) {
$index_id = $doc->index_id;
if (isset($indexes[$index_id])) {
$index = $indexes[$index_id];
}
else {
$index = new SearchApiIndex(array(
'machine_name' => $index_id,
));
}
$fields = $this
->getFieldNames($index);
$field_options = $index->options['fields'];
$result = array(
'id' => NULL,
'index_id' => $index_id,
'score' => NULL,
'fields' => array(),
);
$solr_id = $this
->createId($index_id, $doc->item_id);
foreach ($fields as $search_api_property => $solr_property) {
if (isset($doc->{$solr_property})) {
$value = $doc->{$solr_property};
// Date fields need some special treatment to become valid date values
// (i.e., timestamps) again.
$first_value = $value;
while (is_array($first_value)) {
$first_value = reset($first_value);
}
if (isset($field_options[$search_api_property]['type']) && search_api_extract_inner_type($field_options[$search_api_property]['type']) === 'date' && preg_match('/^\\d{4}-\\d{2}-\\d{2}T\\d{2}:\\d{2}:\\d{2}Z$/', $first_value)) {
$value = is_array($value) ? array_map('strtotime', $value) : strtotime($value);
}
$result['fields'][$search_api_property] = $value;
}
}
$fulltext_fields = isset($fulltext_fields_by_index[$index_id]) ? $fulltext_fields_by_index[$index_id] : array();
$excerpt = $this
->getExcerpt($response, $solr_id, $result['fields'], $fields, $fulltext_fields);
if ($excerpt) {
$result['excerpt'] = $excerpt;
}
// We can find the item id and score in the special 'search_api_*'
// properties. Mappings are provided for these properties in
// SearchApiSolrService::getFieldNames().
$result['id'] = $result['fields']['search_api_id'];
$result['score'] = $result['fields']['search_api_relevance'];
$results['results'][$id] = $result;
}
// Extract facets
if (isset($response->facet_counts->facet_fields)) {
$results['search_api_facets'] = array();
$facet_fields = $response->facet_counts->facet_fields;
// The key for the "missing" facet (empty string in the JSON). This will
// be either "" or "_empty_", depending on the PHP version.
$empty_key = key((array) json_decode('{"":5}'));
foreach ($facets as $delta => $info) {
$field = $solr_fields[$info['field']];
if (!empty($facet_fields->{$field})) {
$min_count = $info['min_count'];
$terms = $facet_fields->{$field};
if ($info['missing']) {
// We have to correctly incorporate the "missing" term ($empty_key).
// This will ensure that the term with the least results is dropped,
// if the limit would be exceeded.
if (isset($terms->{$empty_key})) {
if ($terms->{$empty_key} < $min_count) {
unset($terms->{$empty_key});
}
else {
$terms = (array) $terms;
arsort($terms);
if ($info['limit'] > 0 && count($terms) > $info['limit']) {
array_pop($terms);
}
}
}
}
elseif (isset($terms->{$empty_key})) {
$terms = clone $terms;
unset($terms->{$empty_key});
}
$type = isset($fields[$info['field']]['type']) ? search_api_extract_inner_type($fields[$info['field']]['type']) : 'string';
foreach ($terms as $term => $count) {
if ($count >= $min_count) {
if ($term === $empty_key) {
$term = '!';
}
elseif ($type == 'boolean') {
if ($term == 'true') {
$term = '"1"';
}
elseif ($term == 'false') {
$term = '"0"';
}
}
elseif ($type == 'date') {
$term = $term ? '"' . strtotime($term) . '"' : NULL;
}
else {
$term = "\"{$term}\"";
}
if ($term) {
$results['search_api_facets'][$delta][] = array(
'filter' => $term,
'count' => $count,
);
}
}
}
if (empty($results['search_api_facets'][$delta])) {
unset($results['search_api_facets'][$delta]);
}
}
}
}
drupal_alter('search_api_solr_multi_search_results', $results, $query, $response);
// Compute performance
$time_end = microtime(TRUE);
$results['performance'] = array(
'complete' => $time_end - $time_method_called,
'preprocessing' => $time_processing_done - $time_method_called,
'execution' => $time_query_done - $time_processing_done,
'postprocessing' => $time_end - $time_query_done,
);
return $results;
}
//
// Additional methods that might be used when knowing the service class.
//
/**
* Ping the Solr server to tell whether it can be accessed.
*
* Uses the admin/ping request handler.
*/
public function ping() {
$this
->connect();
return $this->solr
->ping();
}
/**
* Sends a commit command to the Solr server.
*/
public function commit() {
// If committing has been disabled altogether, do nothing here.
if (!empty($this->options['commits_disabled'])) {
return;
}
try {
$this
->connect();
return $this->solr
->commit(FALSE);
} catch (SearchApiException $e) {
watchdog_exception('search_api_solr', $e, '%type while trying to commit on server @server: !message in %function (line %line of %file).', array(
'@server' => $this->server->machine_name,
), WATCHDOG_WARNING);
}
}
/**
* Schedules a commit operation for this server.
*
* The commit will be sent at the end of the current page request. Multiple
* calls to this method will still only result in one commit operation.
*/
public function scheduleCommit() {
if (!$this->commitScheduled) {
$this->commitScheduled = TRUE;
drupal_register_shutdown_function(array(
$this,
'commit',
));
}
}
/**
* Gets the Solr connection class used by this service.
*
* @return string
* The name of a class which implements SearchApiSolrConnectionInterface.
*/
public function getConnectionClass() {
return variable_get('search_api_solr_connection_class', $this->connection_class);
}
/**
* Sets the Solr connection class used by this service.
*
* @param string $class
* The name of a class which implements SearchApiSolrConnectionInterface.
*/
public function setConnectionClass($class) {
$this->connection_class = $class;
$this->solr = NULL;
}
/**
* Gets the currently used Solr connection object.
*
* @return SearchApiSolrConnectionInterface
* The solr connection object used by this server.
*/
public function getSolrConnection() {
$this
->connect();
return $this->solr;
}
/**
* Get metadata about fields in the Solr/Lucene index.
*
* @param int $num_terms
* Number of 'top terms' to return.
*
* @return array
* An array of SearchApiSolrField objects.
*
* @see SearchApiSolrConnectionInterface::getFields()
*/
public function getFields($num_terms = 0) {
$this
->connect();
return $this->solr
->getFields($num_terms);
}
/**
* Retrieves a config file or file list from the Solr server.
*
* Uses the admin/file request handler.
*
* @param string|null $file
* (optional) The name of the file to retrieve. If the file is a directory,
* the directory contents are instead listed and returned. NULL represents
* the root config directory.
*
* @return object
* A HTTP response object containing either the file contents or a file list.
*/
public function getFile($file = NULL) {
$this
->connect();
$file_servlet_name = constant($this
->getConnectionClass() . '::FILE_SERVLET');
$params['contentType'] = 'text/xml;charset=utf-8';
if ($file) {
$params['file'] = $file;
}
return $this->solr
->makeServletRequest($file_servlet_name, $params);
}
/**
* Prefixes an index ID as configured.
*
* The resulting ID will be a concatenation of the following strings:
* - If set, the "search_api_solr_index_prefix" variable.
* - If set, the index-specific "search_api_solr_index_prefix_INDEX" variable.
* - The index's machine name.
*
* @param string $machine_name
* The index's machine name.
*
* @return string
* The prefixed machine name.
*/
protected function getIndexId($machine_name) {
// Prepend per-index prefix.
$id = variable_get('search_api_solr_index_prefix_' . $machine_name, '') . $machine_name;
// Prepend environment prefix.
$id = variable_get('search_api_solr_index_prefix', '') . $id;
return $id;
}
/**
* Retrieves the effective fulltext fields from the query.
*
* Automatically translates a NULL value in the query object to all fulltext
* fields in the search index.
*
* If a specific backend supports any "virtual" fulltext fields not listed in
* the index, it should override this method to add them, if appropriate.
*
* @param SearchApiQueryInterface $query
* The search query.
*
* @return string[]
* The fulltext fields in which to search for the search keys.
*
* @see SearchApiQueryInterface::getFields()
*/
protected function getQueryFields(SearchApiQueryInterface $query) {
$fulltext_fields = $query
->getFields();
$index_fields = $query
->getIndex()
->getFulltextFields();
return $fulltext_fields === NULL ? $index_fields : array_intersect($fulltext_fields, $index_fields);
}
/**
*
*
* @param $snippet
*
* @return string|string[]
*/
protected function sanitizeAndFormatExcerptSnippet($snippet) {
// Sanitize and format the snippet.
$snippet = check_plain($snippet);
$snippet = $this
->formatHighlighting($snippet);
// The created fragments sometimes have leading or trailing punctuation.
// We remove that here for all common cases, but take care not to remove
// < or > (so HTML tags stay valid).
$snippet = trim($snippet, "\0../:;=?..@[..`");
return $snippet;
}
}Members
Name |
Modifiers | Type | Description | Overrides |
|---|---|---|---|---|
|
SearchApiAbstractService:: |
protected | property | Direct reference to the server's $options property. | |
|
SearchApiAbstractService:: |
protected | property | ||
|
SearchApiAbstractService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiServiceInterface:: |
|
|
SearchApiAbstractService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiServiceInterface:: |
|
|
SearchApiAbstractService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiServiceInterface:: |
1 |
|
SearchApiAbstractService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiServiceInterface:: |
|
|
SearchApiAbstractService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiServiceInterface:: |
|
|
SearchApiSolrService:: |
protected | property | Saves whether a commit operation was already scheduled for this server. | |
|
SearchApiSolrService:: |
protected | property | The connection class used by this service. | |
|
SearchApiSolrService:: |
protected | property | Static cache for getFieldNames(). | |
|
SearchApiSolrService:: |
protected | property | Metadata describing fields on the Solr/Lucene index. | |
|
SearchApiSolrService:: |
protected | property | Request handler to use for this search query. | |
|
SearchApiSolrService:: |
protected | property | A connection to the Solr server. | |
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
protected | function | Helper method for indexing. | |
|
SearchApiSolrService:: |
protected | function | Applies custom modifications to indexed Solr documents. | |
|
SearchApiSolrService:: |
public | function | Sends a commit command to the Solr server. | |
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
protected | function | Create a connection to the Solr server as configured in $this->options. | |
|
SearchApiSolrService:: |
protected | function | Transforms a query filter into a flat array of Solr filter queries, using the field names in $fields. | |
|
SearchApiSolrService:: |
protected | function | Create a single search query string according to the given field, value and operator. | |
|
SearchApiSolrService:: |
public | function | Creates an ID used as the unique identifier at the Solr server. | |
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::deleteItems(). Overrides SearchApiServiceInterface:: |
|
|
SearchApiSolrService:: |
protected | function | Extract facets from a Solr response. | |
|
SearchApiSolrService:: |
protected | function | Extracts short snippets with highlighting from highlighted field values. | |
|
SearchApiSolrService:: |
protected | function | Extract results from a Solr response. | |
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
protected | function | Flatten a keys array into a single search string. | |
|
SearchApiSolrService:: |
protected | function | Format a value for filtering on a field of a specific type. | |
|
SearchApiSolrService:: |
protected | function | Changes highlighting tags from our custom, HTML-safe ones to HTML. | |
|
SearchApiSolrService:: |
public | function | ||
|
SearchApiSolrService:: |
public | function | Gets the Solr connection class used by this service. | |
|
SearchApiSolrService:: |
protected | function | Extract and format highlighting information for a specific item from a Solr response. | |
|
SearchApiSolrService:: |
public | function |
Returns additional, service-specific information about this server. Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
protected | function | Helper method for creating the facet field parameters. | |
|
SearchApiSolrService:: |
public | function | Create a list of all indexed field names mapped to their Solr field names. | |
|
SearchApiSolrService:: |
public | function | Get metadata about fields in the Solr/Lucene index. | |
|
SearchApiSolrService:: |
public | function | Retrieves a config file or file list from the Solr server. | |
|
SearchApiSolrService:: |
protected | function | Returns the prefix and suffix for highlighting matches in the excerpt. | |
|
SearchApiSolrService:: |
protected | function | Helper method for creating the highlighting parameters. | |
|
SearchApiSolrService:: |
protected | function | Prefixes an index ID as configured. | |
|
SearchApiSolrService:: |
protected | function | Retrieves the effective fulltext fields from the query. | |
|
SearchApiSolrService:: |
public | function | Returns a link to the Solr server, if the necessary options are set. | |
|
SearchApiSolrService:: |
public | function | Gets the currently used Solr connection object. | |
|
SearchApiSolrService:: |
public | function |
Indexes the specified items. Overrides SearchApiServiceInterface:: |
|
|
SearchApiSolrService:: |
public | function | Ping the Solr server to tell whether it can be accessed. | |
|
SearchApiSolrService:: |
protected | function | Empty method to allow subclasses to apply custom changes before search results are returned. | |
|
SearchApiSolrService:: |
protected | function | Empty method called before sending a search query to Solr. | |
|
SearchApiSolrService:: |
public | function | Implements SearchApiMultiServiceInterface::queryMultiple(). | |
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
protected | function | ||
|
SearchApiSolrService:: |
protected | function | Sanitizes a highlighted field value. | |
|
SearchApiSolrService:: |
public | function | Schedules a commit operation for this server. | |
|
SearchApiSolrService:: |
public | function |
Executes a search on the server represented by this object. Overrides SearchApiServiceInterface:: |
|
|
SearchApiSolrService:: |
public | function | Implements SearchApiMultiServiceInterface::searchMultiple(). | |
|
SearchApiSolrService:: |
public | function | Sets the Solr connection class used by this service. | |
|
SearchApiSolrService:: |
protected | function | Sets the request handler. | |
|
SearchApiSolrService:: |
constant | The date format that Solr uses, in PHP date() syntax. | ||
|
SearchApiSolrService:: |
public | function |
Implements SearchApiServiceInterface::__construct(). Overrides SearchApiAbstractService:: |
|
|
SearchApiSolrService:: |
public | function |
Overrides SearchApiAbstractService::viewSettings(). Overrides SearchApiAbstractService:: |