PHRAS-2831_port-truncation_4.1

copy elastic/truncation code from 4.0
nb: from already exists
This commit is contained in:
Jean-Yves Gaulier
2019-11-19 16:45:46 +01:00
parent cc9183373e
commit 3145049247
7 changed files with 83 additions and 9 deletions

View File

@@ -45,6 +45,9 @@ class TextNode extends AbstractTermNode implements ContextAbleInterface
foreach ($context->localizeField($field) as $f) { foreach ($context->localizeField($field) as $f) {
$index_fields[] = $f; $index_fields[] = $f;
} }
foreach ($context->truncationField($field) as $f) {
$index_fields[] = $f;
}
} }
if (!$index_fields) { if (!$index_fields) {
return null; return null;

View File

@@ -90,6 +90,16 @@ class Index
// TODO Maybe replace nfkc_normalizer + asciifolding with icu_folding // TODO Maybe replace nfkc_normalizer + asciifolding with icu_folding
'filter' => ['nfkc_normalizer', 'asciifolding'] 'filter' => ['nfkc_normalizer', 'asciifolding']
], ],
'truncation_analyzer' => [
'type' => 'custom',
'tokenizer' => 'truncation_tokenizer',
'filter' => ['lowercase', 'stop', 'kstem']
],
'truncation_analyzer#search' => [
'type' => 'custom',
'tokenizer' => 'truncation_tokenizer',
'filter' => ['lowercase', 'stop', 'kstem']
],
// Lang specific // Lang specific
'fr_full' => [ 'fr_full' => [
'type' => 'custom', 'type' => 'custom',
@@ -145,6 +155,12 @@ class Index
] ]
], ],
'tokenizer' => [ 'tokenizer' => [
'truncation_tokenizer' => [
"type" => "edgeNGram",
"min_gram" => "2",
"max_gram" => "15",
"token_chars" => [ "letter", "digit", "punctuation", "symbol" ]
],
'thesaurus_path' => [ 'thesaurus_path' => [
'type' => 'path_hierarchy' 'type' => 'path_hierarchy'
] ]

View File

@@ -55,8 +55,13 @@ class StringFieldMapping extends ComplexFieldMapping
{ {
$child = new StringFieldMapping('light'); $child = new StringFieldMapping('light');
$child->setAnalyzer('general_light'); $child->setAnalyzer('general_light');
$this->addChild($child); $this->addChild($child);
$child = new StringFieldMapping('truncated');
$child->setAnalyzer('truncation_analyzer', 'indexing');
$child->setAnalyzer('truncation_analyzer#search', 'searching');
$this->addChild($child);
$this->addLocalizedChildren($locales); $this->addLocalizedChildren($locales);
return $this; return $this;

View File

@@ -9,6 +9,7 @@ use Alchemy\Phrasea\SearchEngine\Elastic\Structure\Field;
use Alchemy\Phrasea\SearchEngine\Elastic\AST\Field as ASTField; use Alchemy\Phrasea\SearchEngine\Elastic\AST\Field as ASTField;
use Alchemy\Phrasea\SearchEngine\Elastic\AST\Flag; use Alchemy\Phrasea\SearchEngine\Elastic\AST\Flag;
use Alchemy\Phrasea\SearchEngine\Elastic\Structure\Structure; use Alchemy\Phrasea\SearchEngine\Elastic\Structure\Structure;
use Alchemy\Phrasea\SearchEngine\SearchEngineOptions;
/** /**
* @todo Check for private fields and only search on them if allowed * @todo Check for private fields and only search on them if allowed
@@ -23,13 +24,23 @@ class QueryContext
private $queryLocale; private $queryLocale;
/** @var array */ /** @var array */
private $fields; private $fields;
/** @var SearchEngineOptions */
private $options;
public function __construct(Structure $structure, array $locales, $queryLocale, array $fields = null) /**
* @param SearchEngineOptions|null $options
* @param Structure $structure
* @param array $locales
* @param $queryLocale
* @param array $fields
*/
public function __construct($options, Structure $structure, array $locales, $queryLocale, array $fields = null)
{ {
$this->structure = $structure; $this->structure = $structure;
$this->locales = $locales; $this->locales = $locales;
$this->queryLocale = $queryLocale; $this->queryLocale = $queryLocale;
$this->fields = $fields; $this->fields = $fields;
$this->options = $options;
} }
public function narrowToFields(array $fields) public function narrowToFields(array $fields)
@@ -43,7 +54,7 @@ class QueryContext
} }
} }
return new static($this->structure, $this->locales, $this->queryLocale, $fields); return new static($this->options, $this->structure, $this->locales, $this->queryLocale, $fields);
} }
/** /**
@@ -131,6 +142,16 @@ class QueryContext
return $ret; return $ret;
} }
public function truncationField(Field $field)
{
if($this->options && $this->options->useTruncation()) {
return [sprintf('%s.truncated', $field->getIndexField())];
}
else {
return [];
}
}
private function localizeFieldName($field) private function localizeFieldName($field)
{ {
$fields = array(); $fields = array();

View File

@@ -23,7 +23,7 @@ class QueryContextFactory
? $this->getLimitedStructure($options) ? $this->getLimitedStructure($options)
: $this->structure; : $this->structure;
$context = new QueryContext($structure, $this->locales, $this->current_locale); $context = new QueryContext($options, $structure, $this->locales, $this->current_locale);
if ($options) { if ($options) {
$fields = $this->getSearchedFields($options); $fields = $this->getSearchedFields($options);

View File

@@ -71,6 +71,8 @@ class SearchEngineOptions
protected $i18n; protected $i18n;
/** @var bool */ /** @var bool */
protected $stemming = true; protected $stemming = true;
/** @var bool */
protected $use_truncation = false;
/** @var string */ /** @var string */
protected $sort_by; protected $sort_by;
@@ -105,7 +107,8 @@ class SearchEngineOptions
'sort_ord', 'sort_ord',
'business_fields', 'business_fields',
'max_results', 'max_results',
'first_result' 'first_result',
'use_truncation',
]; ];
/** /**
@@ -217,6 +220,29 @@ class SearchEngineOptions
return $this; return $this;
} }
/**
* Tells whether to use truncation or not
*
* @param boolean $boolean
* @return $this
*/
public function setUseTruncation($boolean)
{
$this->use_truncation = !!$boolean;
return $this;
}
/**
* Return wheter the use of truncation is enabled or not
*
* @return boolean
*/
public function useTruncation()
{
return $this->use_truncation;
}
/** /**
* Return wheter the use of stemming is enabled or not * Return wheter the use of stemming is enabled or not
* *
@@ -542,6 +568,8 @@ class SearchEngineOptions
$options->setFields($databoxFields); $options->setFields($databoxFields);
$options->setDateFields($databoxDateFields); $options->setDateFields($databoxDateFields);
$options->setUseTruncation((Boolean) $request->get('truncation'));
return $options; return $options;
} }
@@ -628,6 +656,7 @@ class SearchEngineOptions
} }
}, },
'stemming' => $optionSetter('setStemming'), 'stemming' => $optionSetter('setStemming'),
'use_truncation' => $optionSetter('setUseTruncation'),
'date_fields' => function ($value, SearchEngineOptions $options) use ($fieldNormalizer) { 'date_fields' => function ($value, SearchEngineOptions $options) use ($fieldNormalizer) {
$options->setDateFields($fieldNormalizer($value)); $options->setDateFields($fieldNormalizer($value));
}, },

View File

@@ -18,7 +18,7 @@ class QueryContextTest extends \PHPUnit_Framework_TestCase
{ {
$structure = $this->prophesize(Structure::class)->reveal(); $structure = $this->prophesize(Structure::class)->reveal();
$available_locales = ['ab', 'cd', 'ef']; $available_locales = ['ab', 'cd', 'ef'];
$context = new QueryContext($structure, $available_locales, 'fr'); $context = new QueryContext(null, $structure, $available_locales, 'fr');
$narrowed = $context->narrowToFields(['some_field']); $narrowed = $context->narrowToFields(['some_field']);
$this->assertEquals(['some_field'], $narrowed->getFields()); $this->assertEquals(['some_field'], $narrowed->getFields());
} }
@@ -33,10 +33,10 @@ class QueryContextTest extends \PHPUnit_Framework_TestCase
'bar' => $bar_field 'bar' => $bar_field
]); ]);
$context = new QueryContext($structure->reveal(), [], 'fr'); $context = new QueryContext(null, $structure->reveal(), [], 'fr');
$this->assertEquals([$foo_field, $bar_field], $context->getUnrestrictedFields()); $this->assertEquals([$foo_field, $bar_field], $context->getUnrestrictedFields());
$narrowed_context = new QueryContext($structure->reveal(), [], 'fr', ['foo']); $narrowed_context = new QueryContext(null, $structure->reveal(), [], 'fr', ['foo']);
$this->assertEquals([$foo_field], $narrowed_context->getUnrestrictedFields()); $this->assertEquals([$foo_field], $narrowed_context->getUnrestrictedFields());
} }
@@ -50,7 +50,7 @@ class QueryContextTest extends \PHPUnit_Framework_TestCase
'bar' => $bar_field 'bar' => $bar_field
]); ]);
$context = new QueryContext($structure->reveal(), [], 'fr'); $context = new QueryContext(null, $structure->reveal(), [], 'fr');
$this->assertEquals([$foo_field, $bar_field], $context->getPrivateFields()); $this->assertEquals([$foo_field, $bar_field], $context->getPrivateFields());
$narrowed_context = new QueryContext($structure->reveal(), [], 'fr', ['foo']); $narrowed_context = new QueryContext($structure->reveal(), [], 'fr', ['foo']);