SilverStripe\FullTextSearch\Search\Indexes\SearchIndex

Complexity

Total Complexity

99

Size/Duplication

Total Lines	631
Duplicated Lines	0 %

Coupling/Cohesion

Components	2
Dependencies	14

Importance

Changes

0

27 Methods

Rating	Name	Duplication	Size	Complexity
F	fieldData()	0	154	31
A	getSourceName()	0	6	1
A	__construct()	0	11	2
A	__toString()	0	4	1
A	addClass()	0	12	4
A	getClasses()	0	4	1
A	addFulltextField()	0	4	1
A	getFulltextFields()	0	4	1
A	addFilterField()	0	4	1
A	getFilterFields()	0	4	1
A	addSortField()	0	4	1
A	getSortFields()	0	4	1
B	addAllFulltextFields()	0	20	5
A	getFieldsIterator()	0	4	1
A	excludeVariantState()	0	4	1
B	variantStateExcluded()	0	17	6
A	buildDependancyList()	0	11	3
B	getDerivedFields()	0	32	5
A	getDocumentIDForState()	0	6	1
A	getDocumentID()	0	4	1
C	_getFieldValue()	0	61	12
A	warn()	0	4	1
C	getDirtyIDs()	0	78	17
	add()	0	1	?
	delete()	0	1	?
	commit()	0	1	?
	init()	0	1	?

<?php

namespace SilverStripe\FullTextSearch\Search\Indexes;

use Exception;
use Psr\Log\LoggerInterface;
use SilverStripe\Core\ClassInfo;
use SilverStripe\Core\Config\Config;
use SilverStripe\Core\Injector\Injector;
use SilverStripe\FullTextSearch\Search\SearchIntrospection;
use SilverStripe\FullTextSearch\Search\Variants\SearchVariant;
use SilverStripe\FullTextSearch\Utils\MultipleArrayIterator;
use SilverStripe\ORM\DataObject;
use SilverStripe\ORM\FieldType\DBField;
use SilverStripe\ORM\FieldType\DBString;
use SilverStripe\ORM\Queries\SQLSelect;
use SilverStripe\View\ViewableData;

/**
 * SearchIndex is the base index class. Each connector will provide a subclass of this that
 * provides search engine specific behavior.
 *
 * This class is responsible for:
 *
 * - Taking index calls adding classes and fields, and resolving those to value sources and types
 *
 * - Determining which records in this index need updating when a DataObject is changed
 *
 * - Providing utilities to the connector indexes
 *
 * The connector indexes are responsible for
 *
 * - Mapping types to index configuration
 *
 * - Adding and removing items to index
 *
 * - Parsing and converting SearchQueries into a form the engine will understand, and executing those queries
 *
 * The user indexes are responsible for
 *
 * - Specifying which classes and fields this index contains
 *
 * - Specifying update rules that are not extractable from metadata (because the values come from functions for instance)
 *
 */
abstract class SearchIndex extends ViewableData
{
    /**
     * Allows this index to hide a parent index. Specifies the name of a parent index to disable
     *
     * @var string
     * @config
     */
    private static $hide_ancestor;

The property $hide_ancestor is not used and could be removed.

    /**
     * Used to separate class name and relation name in the sources array
     * this string must not be present in class name
     * @var string
     * @config
     */
    private static $class_delimiter = '_|_';

The property $class_delimiter is not used and could be removed.

    /**
     * This is used to clean the source name from suffix
     * suffixes are needed to support multiple relations with the same name on different page types
     * @param string $source
     * @return string
     */
    protected function getSourceName($source)
    {
        $source = explode(self::config()->get('class_delimiter'), $source);

        return $source[0];
    }

    public function __construct()
    {
        parent::__construct();
        $this->init();

        foreach ($this->getClasses() as $class => $options) {
            SearchVariant::with($class, $options['include_children'])->call('alterDefinition', $class, $this);
        }

        $this->buildDependancyList();
    }

    public function __toString()
    {
        return 'Search Index ' . get_class($this);
    }

    /**
     * Examines the classes this index is built on to try and find defined fields in the class hierarchy
     * for those classes.
     * Looks for db and viewable-data fields, although can't necessarily find type for viewable-data fields.
     * If multiple classes have a relation with the same name all of these will be included in the search index
     * Note that only classes that have the relations uninherited (defined in them) will be listed
     * this is because inherited relations do not need to be processed by index explicitly
     */
    public function fieldData($field, $forceType = null, $extraOptions = [])
    {
        $fullfield = str_replace(".", "_", $field);
        $sources = $this->getClasses();

        foreach ($sources as $source => $options) {
            $sources[$source]['base'] = DataObject::getSchema()->baseDataClass($source);
            $sources[$source]['lookup_chain'] = [];
        }

        $found = [];

        if (strpos($field, '.') !== false) {
            $lookups = explode(".", $field);
            $field = array_pop($lookups);

            foreach ($lookups as $lookup) {
                $next = [];

                foreach ($sources as $source => $baseOptions) {
                    $source = $this->getSourceName($source);

                    foreach (SearchIntrospection::hierarchy($source, $baseOptions['include_children']) as $dataclass) {
                        $class = null;
                        $options = $baseOptions;
                        $singleton = singleton($dataclass);
                        $schema = DataObject::getSchema();
                        $className = $singleton->getClassName();

                        if ($hasOne = $schema->hasOneComponent($className, $lookup)) {
                            // we only want to include base class for relation, omit classes that inherited the relation
                            $relationList = Config::inst()->get($dataclass, 'has_one', Config::UNINHERITED);
                            $relationList = (!is_null($relationList)) ? $relationList : [];
                            if (!array_key_exists($lookup, $relationList)) {
                                continue;
                            }

                            $class = $hasOne;
                            $options['lookup_chain'][] = array(
                                'call' => 'method', 'method' => $lookup,
                                'through' => 'has_one', 'class' => $dataclass, 'otherclass' => $class, 'foreignkey' => "{$lookup}ID"
                            );
                        } elseif ($hasMany = $schema->hasManyComponent($className, $lookup)) {
                            // we only want to include base class for relation, omit classes that inherited the relation
                            $relationList = Config::inst()->get($dataclass, 'has_many', Config::UNINHERITED);
                            $relationList = (!is_null($relationList)) ? $relationList : [];
                            if (!array_key_exists($lookup, $relationList)) {
                                continue;
                            }

                            $class = $hasMany;
                            $options['multi_valued'] = true;
                            $options['lookup_chain'][] = array(
                                'call' => 'method', 'method' => $lookup,
                                'through' => 'has_many', 'class' => $dataclass, 'otherclass' => $class, 'foreignkey' => $schema->getRemoteJoinField($className, $lookup, 'has_many')
                            );
                        } elseif ($manyMany = $schema->manyManyComponent($className, $lookup)) {
                            // we only want to include base class for relation, omit classes that inherited the relation
                            $relationList = Config::inst()->get($dataclass, 'many_many', Config::UNINHERITED);
                            $relationList = (!is_null($relationList)) ? $relationList : [];
                            if (!array_key_exists($lookup, $relationList)) {
                                continue;
                            }

                            $class = $manyMany['childClass'];
                            $options['multi_valued'] = true;
                            $options['lookup_chain'][] = array(
                                'call' => 'method',
                                'method' => $lookup,
                                'through' => 'many_many',
                                'class' => $dataclass,
                                'otherclass' => $class,
                                'details' => $manyMany,
                            );
                        }

                        if (is_string($class) && $class) {
                            if (!isset($options['origin'])) {
                                $options['origin'] = $dataclass;
                            }

                            // we add suffix here to prevent the relation to be overwritten by other instances
                            // all sources lookups must clean the source name before reading it via getSourceName()
                            $next[$class . self::config()->get('class_delimiter') . $dataclass] = $options;
                        }
                    }
                }

                if (!$next) {

The expression $next of type array is implicitly converted to a boolean; are you sure this is intended? If so, consider using empty($expr) instead to make it clear that you intend to check for an array without elements.

                    return $next;
                } // Early out to avoid excessive empty looping
                $sources = $next;
            }
        }

        foreach ($sources as $class => $options) {
            $class = $this->getSourceName($class);
            $dataclasses = SearchIntrospection::hierarchy($class, $options['include_children']);

            while (count($dataclasses)) {
                $dataclass = array_shift($dataclasses);
                $type = null;
                $fieldoptions = $options;

                $fields = DataObject::getSchema()->databaseFields($class);

                if (isset($fields[$field])) {
                    $type = $fields[$field];
                    $fieldoptions['lookup_chain'][] = array('call' => 'property', 'property' => $field);
                } else {
                    $singleton = singleton($dataclass);

                    if ($singleton->hasMethod("get$field") || $singleton->hasField($field)) {
                        $type = $singleton->castingClass($field);
                        if (!$type) {
                            $type = 'String';
                        }

                        if ($singleton->hasMethod("get$field")) {
                            $fieldoptions['lookup_chain'][] = array('call' => 'method', 'method' => "get$field");
                        } else {
                            $fieldoptions['lookup_chain'][] = array('call' => 'property', 'property' => $field);
                        }
                    }
                }

                if ($type) {
                    // Don't search through child classes of a class we matched on. TODO: Should we?
                    $dataclasses = array_diff($dataclasses, array_values(ClassInfo::subclassesFor($dataclass)));
                    // Trim arguments off the type string
                    if (preg_match('/^(\w+)\(/', $type, $match)) {
                        $type = $match[1];
                    }
                    // Get the origin
                    $origin = isset($fieldoptions['origin']) ? $fieldoptions['origin'] : $dataclass;

                    $found["{$origin}_{$fullfield}"] = array(
                        'name' => "{$origin}_{$fullfield}",
                        'field' => $field,
                        'fullfield' => $fullfield,
                        'base' => $fieldoptions['base'],
                        'origin' => $origin,
                        'class' => $dataclass,
                        'lookup_chain' => $fieldoptions['lookup_chain'],
                        'type' => $forceType ? $forceType : $type,
                        'multi_valued' => isset($fieldoptions['multi_valued']) ? true : false,
                        'extra_options' => $extraOptions
                    );
                }
            }
        }

        return $found;
    }

    /** Public, but should only be altered by variants */

    protected $classes = array();

    protected $fulltextFields = array();

    public $filterFields = array();

    protected $sortFields = array();

    protected $excludedVariantStates = array();

    /**
     * Add a DataObject subclass whose instances should be included in this index
     *
     * Can only be called when addFulltextField, addFilterField, addSortField and addAllFulltextFields have not
     * yet been called for this index instance
     *
     * @throws Exception
     * @param string $class - The class to include
     * @param array $options - TODO: Remove
     */
    public function addClass($class, $options = array())
    {
        if ($this->fulltextFields || $this->filterFields || $this->sortFields) {

The expression $this->fulltextFields of type array is implicitly converted to a boolean; are you sure this is intended? If so, consider using ! empty($expr) instead to make it clear that you intend to check for an array without elements.

The expression $this->filterFields of type array is implicitly converted to a boolean; are you sure this is intended? If so, consider using ! empty($expr) instead to make it clear that you intend to check for an array without elements.

The expression $this->sortFields of type array is implicitly converted to a boolean; are you sure this is intended? If so, consider using ! empty($expr) instead to make it clear that you intend to check for an array without elements.

            throw new Exception('Can\'t add class to Index after fields have already been added');
        }

        $options = array_merge(array(
            'include_children' => true
        ), $options);

        $this->classes[$class] = $options;
    }

    /**
     * Get the classes added by addClass
     */
    public function getClasses()
    {
        return $this->classes;
    }

    /**
     * Add a field that should be fulltext searchable
     * @param string $field - The field to add
     * @param string $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
     * @param string $extraOptions - Dependent on search implementation
     */
    public function addFulltextField($field, $forceType = null, $extraOptions = array())
    {
        $this->fulltextFields = array_merge($this->fulltextFields, $this->fieldData($field, $forceType, $extraOptions));

It seems like $extraOptions defined by parameter $extraOptions on line 306 can also be of type string; however, SilverStripe\FullTextSearch\Search\Indexes\SearchIndex::fieldData() does only seem to accept array, maybe add an additional type check?

    }

    public function getFulltextFields()
    {
        return $this->fulltextFields;
    }

    /**
     * Add a field that should be filterable
     * @param string $field - The field to add
     * @param string $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
     * @param string $extraOptions - Dependent on search implementation
     */
    public function addFilterField($field, $forceType = null, $extraOptions = array())
    {
        $this->filterFields = array_merge($this->filterFields, $this->fieldData($field, $forceType, $extraOptions));

It seems like $extraOptions defined by parameter $extraOptions on line 322 can also be of type string; however, SilverStripe\FullTextSearch\Search\Indexes\SearchIndex::fieldData() does only seem to accept array, maybe add an additional type check?

    }

    public function getFilterFields()
    {
        return $this->filterFields;
    }

    /**
     * Add a field that should be sortable
     * @param string $field - The field to add
     * @param string $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
     * @param string $extraOptions - Dependent on search implementation
     */
    public function addSortField($field, $forceType = null, $extraOptions = array())
    {
        $this->sortFields = array_merge($this->sortFields, $this->fieldData($field, $forceType, $extraOptions));

It seems like $extraOptions defined by parameter $extraOptions on line 338 can also be of type string; however, SilverStripe\FullTextSearch\Search\Indexes\SearchIndex::fieldData() does only seem to accept array, maybe add an additional type check?

    }

    public function getSortFields()
    {
        return $this->sortFields;
    }

    /**
     * Add all database-backed text fields as fulltext searchable fields.
     *
     * For every class included in the index, examines those classes and all subclasses looking for "Text" database
     * fields (Varchar, Text, HTMLText, etc) and adds them all as fulltext searchable fields.
     */
    public function addAllFulltextFields($includeSubclasses = true)
    {
        foreach ($this->getClasses() as $class => $options) {
            $classHierarchy = SearchIntrospection::hierarchy($class, $includeSubclasses, true);

            foreach ($classHierarchy as $dataClass) {
                $fields = DataObject::getSchema()->databaseFields($dataClass);

                foreach ($fields as $field => $type) {
                    list($type, $args) = ClassInfo::parse_class_spec($type);

The assignment to $args is unused. Consider omitting it like so list($first,,$third).

                    /** @var DBField $object */
                    $object = Injector::inst()->get($type, false, ['Name' => 'test']);
                    if ($object instanceof DBString) {
                        $this->addFulltextField($field);
                    }
                }
            }
        }
    }

    /**
     * Returns an interator that will let you interate through all added fields, regardless of whether they
     * were added as fulltext, filter or sort fields.
     *
     * @return MultipleArrayIterator
     */
    public function getFieldsIterator()
    {
        return new MultipleArrayIterator($this->fulltextFields, $this->filterFields, $this->sortFields);
    }

    public function excludeVariantState($state)
    {
        $this->excludedVariantStates[] = $state;
    }

    /** Returns true if some variant state should be ignored */
    public function variantStateExcluded($state)
    {
        foreach ($this->excludedVariantStates as $excludedstate) {
            $matches = true;

            foreach ($excludedstate as $variant => $variantstate) {
                if (!isset($state[$variant]) || $state[$variant] != $variantstate) {
                    $matches = false;
                    break;
                }
            }

            if ($matches) {
                return true;
            }
        }
    }

    public $dependancyList = array();

    public function buildDependancyList()
    {
        $this->dependancyList = array_keys($this->getClasses());

        foreach ($this->getFieldsIterator() as $name => $field) {
            if (!isset($field['class'])) {
                continue;
            }
            SearchIntrospection::add_unique_by_ancestor($this->dependancyList, $field['class']);
        }
    }

    public $derivedFields = null;

    /**
     * Returns an array where each member is all the fields and the classes that are at the end of some
     * specific lookup chain from one of the base classes
     */
    public function getDerivedFields()
    {
        if ($this->derivedFields === null) {
            $this->derivedFields = array();

            foreach ($this->getFieldsIterator() as $name => $field) {
                if (count($field['lookup_chain']) < 2) {
                    continue;
                }

                $key = sha1($field['base'].serialize($field['lookup_chain']));
                $fieldname = "{$field['class']}:{$field['field']}";

                if (isset($this->derivedFields[$key])) {
                    $this->derivedFields[$key]['fields'][$fieldname] = $fieldname;
                    SearchIntrospection::add_unique_by_ancestor($this->derivedFields['classes'], $field['class']);
                } else {
                    $chain = array_reverse($field['lookup_chain']);
                    array_shift($chain);

                    $this->derivedFields[$key] = array(
                        'base' => $field['base'],
                        'fields' => array($fieldname => $fieldname),
                        'classes' => array($field['class']),
                        'chain' => $chain
                    );
                }
            }
        }

        return $this->derivedFields;
    }

    /**
     * Get the "document ID" (a database & variant unique id) given some "Base" class, DataObject ID and state array
     *
     * @param string $base - The base class of the object
     * @param integer $id - The ID of the object
     * @param array $state - The variant state of the object
     * @return string - The document ID as a string
     */
    public function getDocumentIDForState($base, $id, $state)
    {
        ksort($state);
        $parts = array('id' => $id, 'base' => $base, 'state' => json_encode($state));
        return implode('-', array_values($parts));
    }

    /**
     * Get the "document ID" (a database & variant unique id) given some "Base" class and DataObject
     *
     * @param DataObject $object - The object
     * @param string $base - The base class of the object
     * @param boolean $includesubs - TODO: Probably going away
     * @return string - The document ID as a string
     */
    public function getDocumentID($object, $base, $includesubs)
    {
        return $this->getDocumentIDForState($base, $object->ID, SearchVariant::current_state($base, $includesubs));
    }

    /**
     * Given an object and a field definition (as returned by fieldData) get the current value of that field on that object
     *
     * @param DataObject $object - The object to get the value from
     * @param array $field - The field definition to use
     * @return mixed - The value of the field, or null if we couldn't look it up for some reason
     */
    protected function _getFieldValue($object, $field)
    {
        $errorHandler = function ($no, $str) {
            throw new Exception('HTML Parse Error: ' . $str);
        };
        set_error_handler($errorHandler, E_ALL);

        try {
            foreach ($field['lookup_chain'] as $step) {
                // Just fail if we've fallen off the end of the chain
                if (!$object) {
                    return null;
                }

                // If we're looking up this step on an array or SS_List, do the step on every item, merge result
                if (is_array($object) || $object instanceof SS_List) {

The class SilverStripe\FullTextSearch\Search\Indexes\SS_List does not exist. Did you forget a USE statement, or did you not list all dependencies?

                    $next = array();

                    foreach ($object as $item) {
                        if ($step['call'] == 'method') {
                            $method = $step['method'];
                            $item = $item->$method();
                        } else {
                            $property = $step['property'];
                            $item = $item->$property;
                        }

                        if ($item instanceof SS_List) {

The class SilverStripe\FullTextSearch\Search\Indexes\SS_List does not exist. Did you forget a USE statement, or did you not list all dependencies?

                            $next = array_merge($next, $item->toArray());
                        } elseif (is_array($item)) {
                            $next = array_merge($next, $item);
                        } else {
                            $next[] = $item;
                        }
                    }

                    $object = $next;
                } else {
                    // Otherwise, just call
                    if ($step['call'] == 'method') {
                        $method = $step['method'];
                        $object = $object->$method();
                    } elseif ($step['call'] == 'variant') {
                        $variants = SearchVariant::variants();
                        $variant = $variants[$step['variant']];
                        $method = $step['method'];
                        $object = $variant->$method($object);
                    } else {
                        $property = $step['property'];
                        $object = $object->$property;
                    }
                }
            }
        } catch (Exception $e) {
            static::warn($e);
            $object = null;
        }

        restore_error_handler();
        return $object;
    }

    /**
     * Log non-fatal errors
     *
     * @param Exception $e
     */
    public static function warn($e)
    {
        Injector::inst()->get(LoggerInterface::class)->warning($e);
    }

    /**
     * Given a class, object id, set of stateful ids and a list of changed fields (in a special format),
     * return what statefulids need updating in this index
     *
     * Internal function used by SearchUpdater.
     *
     * @param  string $class
     * @param  int $id
     * @param  array $statefulids
     * @param  array $fields
     * @return array
     */
    public function getDirtyIDs($class, $id, $statefulids, $fields)
    {
        $dirty = array();

        // First, if this object is directly contained in the index, add it
        foreach ($this->classes as $searchclass => $options) {
            if ($searchclass == $class || ($options['include_children'] && is_subclass_of($class, $searchclass))) {

Due to PHP Bug #53727, is_subclass_of might return inconsistent results on some PHP versions if $searchclass can be an interface. If so, you could instead use ReflectionClass::implementsInterface.

                $base = DataObject::getSchema()->baseDataClass($searchclass);
                $dirty[$base] = array();
                foreach ($statefulids as $statefulid) {
                    $key = serialize($statefulid);
                    $dirty[$base][$key] = $statefulid;
                }
            }
        }

        $current = SearchVariant::current_state();


        // Then, for every derived field
        foreach ($this->getDerivedFields() as $derivation) {
            // If the this object is a subclass of any of the classes we want a field from
            if (!SearchIntrospection::is_subclass_of($class, $derivation['classes'])) {
                continue;
            }
            if (!array_intersect_key($fields, $derivation['fields'])) {
                continue;
            }

            foreach (SearchVariant::reindex_states($class, false) as $state) {
                SearchVariant::activate_state($state);

                $ids = array($id);

                foreach ($derivation['chain'] as $step) {
                    // Use TableName for queries
                    $tableName = DataObject::getSchema()->tableName($step['class']);

                    if ($step['through'] == 'has_one') {
                        $sql = new SQLSelect('"ID"', '"'.$tableName.'"', '"'.$step['foreignkey'].'" IN ('.implode(',', $ids).')');

'"' . $step['foreignkey'] . '" IN (' . implode(',', $ids) . ')' is of type string, but the function expects a array.

                        singleton($step['class'])->extend('augmentSQL', $sql);

                        $ids = $sql->execute()->column();
                    } elseif ($step['through'] == 'has_many') {
                        // Use TableName for queries
                        $otherTableName = DataObject::getSchema()->tableName($step['otherclass']);

                        $sql = new SQLSelect('"'.$tableName.'"."ID"', '"'.$tableName.'"', '"'.$otherTableName.'"."ID" IN ('.implode(',', $ids).')');

'"' . $otherTableName . '"."ID" IN (' . implode(',', $ids) . ')' is of type string, but the function expects a array.

                        $sql->addInnerJoin($otherTableName, '"'.$tableName.'"."ID" = "'.$otherTableName.'"."'.$step['foreignkey'].'"');
                        singleton($step['class'])->extend('augmentSQL', $sql);

                        $ids = $sql->execute()->column();
                    }

                    if (empty($ids)) {
                        break;
                    }
                }

                SearchVariant::activate_state($current);

                if ($ids) {

The expression $ids of type array is implicitly converted to a boolean; are you sure this is intended? If so, consider using ! empty($expr) instead to make it clear that you intend to check for an array without elements.

                    $base = $derivation['base'];
                    if (!isset($dirty[$base])) {
                        $dirty[$base] = array();
                    }

                    foreach ($ids as $rid) {
                        $statefulid = array('id' => $rid, 'state' => $state);
                        $key = serialize($statefulid);
                        $dirty[$base][$key] = $statefulid;
                    }
                }
            }
        }

        return $dirty;
    }

    /** !! These should be implemented by the full text search engine */

    abstract public function add($object);
    abstract public function delete($base, $id, $state);

    abstract public function commit();

    /** !! These should be implemented by the specific index */

    /**
     * Called during construction, this is the method that builds the structure.
     * Used instead of overriding __construct as we have specific execution order - code that has
     * to be run before _and/or_ after this.
     */
    abstract public function init();
}