[Core] Allow changing language stemmer when possible

src/module-elasticsuite-core/Block/Adminhtml/System/Config/Comment.php

@@ -0,0 +1,60 @@
+<?php
+/**
+ * DISCLAIMER
+ *
+ * Do not edit or add to this file if you wish to upgrade this module to newer versions in the future.
+ *
+ * @category  Smile
+ * @package   Smile\ElasticsuiteCore
+ * @author    Richard BAYET <richard.bayet@smile.fr>
+ * @copyright 2024 Smile
+ * @license   Open Software License ("OSL") v. 3.0
+ */
+
+namespace Smile\ElasticsuiteCore\Block\Adminhtml\System\Config;
+
+use Magento\Config\Block\System\Config\Form\Field;
+use Magento\Framework\Data\Form\Element\AbstractElement;
+
+/**
+ * System config comment block.
+ * Allows to display a label and a comment without a field label and the scope input.
+ * This allows to display a comment at default scope for fields only available at the store view level.
+ *
+ * @category Smile
+ * @package  Smile\ElasticsuiteCore
+ */
+class Comment extends Field
+{
+    /**
+     * Retrieve HTML markup for given form element.
+     *
+     * @param AbstractElement $element Form element.
+     *
+     * @return string
+     */
+    public function render(\Magento\Framework\Data\Form\Element\AbstractElement $element)
+    {
+        $html = '<div class="elasticsuite config-comment-title">' . $element->getLabel() . '</div>';
+        $html .= '<div class="elasticsuite config-comment-content">' . $element->getComment() . '</div>';
+
+        return $this->decorateRowHtml($element, $html);
+    }
+
+    /**
+     * Decorate field row html.
+     *
+     * @param AbstractElement $element Form element.
+     * @param string          $html    Field row html.
+     *
+     * @return string
+     */
+    private function decorateRowHtml(\Magento\Framework\Data\Form\Element\AbstractElement $element, $html)
+    {
+        return sprintf(
+            '<tr id="row_%s"><td colspan="3"><div class="elasticsuite config-comment">%s</div></td></tr>',
+            $element->getHtmlId(),
+            $html
+        );
+    }
+}

src/module-elasticsuite-core/Helper/IndexSettings.php

@@ -53,6 +53,16 @@ class IndexSettings extends AbstractConfiguration
      */
     const REFERENCE_ANALYZER_CONFIG_XML_PREFIX = 'reference_analyzer';
 
+    /**
+     * @var string
+     */
+    const STEMMER_USE_DEFAULT_CONFIG_XML_PATH = 'stemmer/use_default';
+
+    /**
+     * @var string
+     */
+    const STEMMER_CUSTOM_STEMMER_CONFIG_XML_PATH = 'stemmer/custom';
+
     /**
      * @var string
      */
@@ -406,6 +416,35 @@ public function getReferenceAnalyzerConfigFlag($configFlag, $store)
         return $this->scopeConfig->isSetFlag($path, ScopeInterface::SCOPE_STORE, $store);
     }
 
+
+    /**
+     * Returns true if the given store used a non-default language stemmer.
+     *
+     * @param integer|string|StoreInterface $store Store.
+     *
+     * @return bool
+     */
+    public function hasCustomLanguageStemmer($store)
+    {
+        $path = self::ANALYSIS_CONFIG_XML_PREFIX . '/' . self::STEMMER_USE_DEFAULT_CONFIG_XML_PATH;
+
+        return (false === $this->scopeConfig->isSetFlag($path, ScopeInterface::SCOPE_STORE, $store));
+    }
+
+    /**
+     * Returns the custom stemmer to use for the given store.
+     *
+     * @param integer|string|StoreInterface $store Store.
+     *
+     * @return mixed
+     */
+    public function getCustomLanguageStemmer($store)
+    {
+        $path = self::ANALYSIS_CONFIG_XML_PREFIX . '/' . self::STEMMER_CUSTOM_STEMMER_CONFIG_XML_PATH;
+
+        return $this->scopeConfig->getValue($path, ScopeInterface::SCOPE_STORE, $store);
+    }
+
     /**
      * Retrieve a configuration value by its key.
      *

src/module-elasticsuite-core/Index/Analysis/Config/Converter.php

@@ -36,6 +36,9 @@ class Converter implements \Magento\Framework\Config\ConverterInterface
     const ANALYZER_TYPE_NODE         = 'analyzer';
     const NORMALIZER_TYPE_ROOT_NODE  = 'normalizers';
     const NORMALIZER_TYPE_NODE       = 'normalizer';
+    const STEMMER_TYPE_ROOT_NODE     = 'stemmers';
+    const STEMMER_GROUP_TYPE_NODE    = 'group';
+    const STEMMER_TYPE_NODE          = 'stemmer';
     const LANGUAGE_DEFAULT           = 'default';
 
     /**
@@ -71,6 +74,11 @@ public function convert($source)
             $configuration[$language] = $this->getLanguageConfiguration($xpath, $language, $defaultConfig);
         }
 
+        $configuration['default'][self::STEMMER_TYPE_ROOT_NODE] = $this->getAllStemmersOptions(
+            $xpath,
+            $this->getAllDefaultLanguageStemmers($xpath)
+        );
+
         return $configuration;
     }
 
@@ -339,4 +347,126 @@ private function getFiltersByRef(\DOMXPath $xpath, \DomNode $rootNode, $searchPa
 
         return $filters;
     }
+
+    /**
+     * Return all default language stemmers as defined currently in config.
+     * Relieson the fact that the filter used will be of type stemmer and named stemmer.
+     *
+     * @param \DOMXPath $xpath XPath access to the document parsed.
+     *
+     * @return array
+     */
+    private function getAllDefaultLanguageStemmers(\DOMXPath $xpath)
+    {
+        $defaultStemmers = [];
+
+        $filterPath = "@type='stemmer' and @name='stemmer'";
+        $stemmerFiltersPath = sprintf(
+            "/%s/%s/%s[%s]",
+            self::ROOT_NODE_NAME,
+            self::FILTER_TYPE_ROOT_NODE,
+            self::FILTER_TYPE_NODE,
+            $filterPath
+        );
+
+        $stemmerFilterNodes = $xpath->query($stemmerFiltersPath);
+        foreach ($stemmerFilterNodes as $stemmerFilterNode) {
+            $language = $stemmerFilterNode->getAttribute('language');
+            $stemmer = false;
+            foreach ($stemmerFilterNode->childNodes as $childNode) {
+                if ($childNode instanceof \DOMElement) {
+                    if ($childNode->tagName === 'language') {
+                        $stemmer = $childNode->nodeValue;
+                        break;
+                    }
+                }
+            }
+            if (!empty($stemmer)) {
+                $defaultStemmers[$language] = $stemmer;
+            }
+        }
+
+        return $defaultStemmers;
+    }
+
+    /**
+     * Parse all stemmers options available for language that support multiple stemmers.
+     *
+     * @param \DOMXPath $xpath           XPath access to the document parsed.
+     * @param array     $defaultStemmers Default stemmers for available languages.
+     *
+     * @return array
+     */
+    private function getAllStemmersOptions(\DOMXPath $xpath, $defaultStemmers = [])
+    {
+        $stemmerOptions = [];
+
+        $searchPath = sprintf(
+            "/%s/%s/%s",
+            self::ROOT_NODE_NAME,
+            self::STEMMER_TYPE_ROOT_NODE,
+            self::STEMMER_GROUP_TYPE_NODE
+        );
+        $stemmerGroupNodes = $xpath->query($searchPath);
+        foreach ($stemmerGroupNodes as $stemmerGroupNode) {
+            $languageCode = $stemmerGroupNode->getAttribute('language');
+            $languageTitle = $stemmerGroupNode->getAttribute('title');
+            $stemmerOptions[$languageCode] = [
+                'identifier' => $languageCode,
+                'title' => $languageTitle,
+                'stemmers' => [],
+            ];
+
+            $stemmerOptions[$languageCode]['stemmers'] = $this->getLanguageStemmers(
+                $xpath,
+                $stemmerGroupNode,
+                $defaultStemmers[$languageCode] ?: null
+            );
+        }
+
+        return $stemmerOptions;
+    }
+
+    /**
+     * Parse available stemmers for a given language.
+     *
+     * @param \DOMXPath   $xpath          XPath access to the document parsed.
+     * @param \DomNode    $rootNode       Stemmers group node for a given language.
+     * @param string|null $defaultStemmer Default stemme for the given language, if defined.
+     *
+     * @return array
+     */
+    private function getLanguageStemmers(\DOMXPath $xpath, \DomNode $rootNode, $defaultStemmer = null)
+    {
+        $stemmers = [];
+
+        $searchPath = sprintf("./%s", self::STEMMER_TYPE_NODE);
+        $stemmerNodes = $xpath->query($searchPath, $rootNode);
+        foreach ($stemmerNodes as $stemmerNode) {
+            $identifier = $stemmerNode->getAttribute('identifier');
+            $stemmer    = [
+                'identifier'    => $identifier,
+                'recommended'   => $stemmerNode->getAttribute('recommended') ?: false,
+                'default'       => ($identifier === $defaultStemmer),
+            ];
+            foreach ($stemmerNode->childNodes as $childNode) {
+                if ($childNode instanceof \DOMElement) {
+                    if ($childNode->tagName === 'label') {
+                        $stemmer['label'] = $childNode->nodeValue;
+                        continue;
+                    }
+
+                    try {
+                        $stemmer[$childNode->tagName] = $this->jsonDecoder->decode($childNode->nodeValue);
+                    } catch (\Exception $exception) {
+                        $stemmer[$childNode->tagName] = $childNode->nodeValue;
+                    }
+                }
+            }
+
+            $stemmers[$identifier] = $stemmer;
+        }
+
+        return $stemmers;
+    }
 }

src/module-elasticsuite-core/Index/Analysis/Config/Reader.php

@@ -46,6 +46,8 @@ class Reader extends Filesystem
         '/analysis/filters/filter'           => ['name', 'language'],
         '/analysis/tokenizers/tokenizer'     => ['name', 'language'],
         '/analysis/analyzers/analyzer'       => ['name', 'language'],
+        '/analysis/stemmers/group'           => ['language'],
+        '/analysis/stemmers/group/stemmer'   => ['identifier'],
     ];
 
     /**

src/module-elasticsuite-core/Index/IndexSettings.php

@@ -140,6 +140,8 @@ public function getAnalysisSettings($store)
             $settings['analyzer']['reference']['filter'] = array_values($referenceFilters);
         }
 
+        $settings = $this->applyCustomLanguageStemmer($settings, $store);
+
         return $settings;
     }
 
@@ -223,4 +225,29 @@ public function getDynamicIndexSettings($store)
 
         return $settings;
     }
+
+    /**
+     * Alter analysis settings to apply possible custom language stemmer.
+     *
+     * @param array                                                 $analysisSettings Analysis settings.
+     * @param integer|string|\Magento\Store\Api\Data\StoreInterface $store            Store.
+     *
+     * @return array
+     */
+    protected function applyCustomLanguageStemmer($analysisSettings, $store)
+    {
+        if ($this->helper->hasCustomLanguageStemmer($store)
+            && is_array($analysisSettings)
+            && array_key_exists('filter', $analysisSettings)
+            && array_key_exists('stemmer', $analysisSettings['filter'])
+            && array_key_exists('language', $analysisSettings['filter']['stemmer'])
+        ) {
+            $customStemmer = $this->helper->getCustomLanguageStemmer($store) ?? false;
+            if (!empty($customStemmer)) {
+                $analysisSettings['filter']['stemmer']['language'] = $customStemmer;
+            }
+        }
+
+        return $analysisSettings;
+    }
 }