Issue #511594 by sxnc, jhodgdon, Gábor Hojtsy: Added hook_search_preprocess() needs to be language-specific.

core/modules/node/node.module

@@ -1584,7 +1584,7 @@ function node_search_execute($keys = NULL, $conditions = NULL) {
       'node' => $node,
       'extra' => $extra,
       'score' => $item->calculated_score,
-      'snippet' => search_excerpt($keys, $node->rendered),
+      'snippet' => search_excerpt($keys, $node->rendered, $item->langcode),
       'langcode' => $node->langcode,
     );
   }

core/modules/search/lib/Drupal/search/Tests/SearchPreprocessLangcodeTest.php

@@ -0,0 +1,99 @@
+<?php
+/**
+ * @file
+ * Definition of Drupal\search\Tests\SearchPreprocessLangcodeTest.
+ */
+
+namespace Drupal\search\Tests;
+
+/**
+ * Test search_simplify() on every Unicode character, and some other cases.
+ */
+class SearchPreprocessLangcodeTest extends SearchTestBase {
+
+  /**
+   * Modules to enable.
+   *
+   * @var array
+   */
+  public static $modules = array('search_langcode_test');
+
+  public static function getInfo() {
+    return array(
+      'name' => 'Search preprocess langcode',
+      'description' => 'Check that the hook_search_preprocess passes the correct langcode from the entity.',
+      'group' => 'Search',
+    );
+  }
+
+  function setUp() {
+    parent::setUp();
+
+    $web_user = $this->drupalCreateUser(array(
+      'create page content',
+      'edit own page content',
+      'search content',
+      'use advanced search',
+    ));
+    $this->drupalLogin($web_user);
+  }
+
+  /**
+   * Tests that hook_search_preprocess() returns the correct langcode.
+   */
+  function testPreprocessLangcode() {
+    // Create a node.
+    $node = $this->drupalCreateNode(array('body' => array('en' => array(array())), 'langcode' => 'en'));
+
+    // First update the index. This does the initial processing.
+    node_update_index();
+
+    // Then, run the shutdown function. Testing is a unique case where indexing
+    // and searching has to happen in the same request, so running the shutdown
+    // function manually is needed to finish the indexing process.
+    search_update_totals();
+
+    // Search for the title of the node with a POST query.
+    $edit = array('or' => $node->label());
+    $this->drupalPost('search/node', $edit, t('Advanced search'));
+
+    // Checks if the langcode has been passed by hook_search_preprocess().
+    $this->assertText('Langcode Preprocess Test: en');
+  }
+
+  /**
+   * Tests stemming for hook_search_preprocess().
+   */
+  function testPreprocessStemming() {
+    // Create a node.
+    $node = $this->drupalCreateNode(array(
+      'title' => 'we are testing',
+      'body' => array('en' => array(array())),
+      'langcode' => 'en',
+    ));
+
+    // First update the index. This does the initial processing.
+    node_update_index();
+
+    // Then, run the shutdown function. Testing is a unique case where indexing
+    // and searching has to happen in the same request, so running the shutdown
+    // function manually is needed to finish the indexing process.
+    search_update_totals();
+
+    // Search for the title of the node with a POST query.
+    $edit = array('or' => 'testing');
+    $this->drupalPost('search/node', $edit, t('Advanced search'));
+
+    // Check if the node has been found.
+    $this->assertText('Search results');
+    $this->assertText('we are testing');
+
+    // Search for the same node using a different query.
+    $edit = array('or' => 'test');
+    $this->drupalPost('search/node', $edit, t('Advanced search'));
+
+    // Check if the node has been found.
+    $this->assertText('Search results');
+    $this->assertText('we are testing');
+  }
+}

core/modules/search/search.api.php

@@ -246,7 +246,7 @@ function hook_search_execute($keys = NULL, $conditions = NULL) {
       'node' => $node,
       'extra' => $extra,
       'score' => $item->calculated_score,
-      'snippet' => search_excerpt($keys, $node->rendered),
+      'snippet' => search_excerpt($keys, $node->rendered, $item->langcode),
       'langcode' => $node->langcode,
     );
   }
@@ -305,6 +305,9 @@ function hook_search_page($results) {
  *   from between two HTML tags or from the search query. It will not contain
  *   any HTML entities or HTML tags.
  *
+ * @param $langcode
+ *   The language code of the entitiy that has been found.
+ *
  * @return
  *   The text after preprocessing. Note that if your module decides not to alter
  *   the text, it should return the original text. Also, after preprocessing,
@@ -312,8 +315,16 @@ function hook_search_page($results) {
  *
  * @ingroup search
  */
-function hook_search_preprocess($text) {
+function hook_search_preprocess($text, $langcode = NULL) {
-  // Do processing on $text
+  // If the langcode is set to 'en' then add variations of the word "testing"
+  // which can also be found during English language searches.
+  if (isset($langcode) && $langcode == 'en') {
+    // Add the alternate verb forms for the word "testing".
+    if ($text == 'we are testing') {
+      $text .= ' test tested';
+    }
+  }
+
   return $text;
 }
 

core/modules/search/search.module

@@ -430,7 +430,7 @@ function search_update_totals() {
  *
  * @see hook_search_preprocess()
  */
-function search_simplify($text) {
+function search_simplify($text, $langcode = NULL) {
   // Decode entities to UTF-8
   $text = decode_entities($text);
 
@@ -438,7 +438,7 @@ function search_simplify($text) {
   $text = drupal_strtolower($text);
 
   // Call an external processor for word handling.
-  search_invoke_preprocess($text);
+  search_invoke_preprocess($text, $langcode);
 
   // Simple CJK handling
   if (config('search.settings')->get('index.overlap_cjk')) {
@@ -523,7 +523,7 @@ function search_expand_cjk($matches) {
 /**
  * Simplifies and splits a string into tokens for indexing.
  */
-function search_index_split($text) {
+function search_index_split($text, $langcode = NULL) {
   $last = &drupal_static(__FUNCTION__);
   $lastsplit = &drupal_static(__FUNCTION__ . ':lastsplit');
 
@@ -531,7 +531,7 @@ function search_index_split($text) {
     return $lastsplit;
   }
   // Process words
-  $text = search_simplify($text);
+  $text = search_simplify($text, $langcode);
   $words = explode(' ', $text);
 
   // Save last keyword result
@@ -554,9 +554,9 @@ function _search_index_truncate(&$text) {
 /**
  * Invokes hook_search_preprocess() in modules.
  */
-function search_invoke_preprocess(&$text) {
+function search_invoke_preprocess(&$text, $langcode = NULL) {
   foreach (module_implements('search_preprocess') as $module) {
-    $text = module_invoke($module, 'search_preprocess', $text);
+    $text = module_invoke($module, 'search_preprocess', $text, $langcode);
   }
 }
 
@@ -668,7 +668,7 @@ function search_index($sid, $module, $text, $langcode) {
             $value = $linktitle;
           }
         }
-        $words = search_index_split($value);
+        $words = search_index_split($value, $langcode);
         foreach ($words as $word) {
           // Add word to accumulator
           $accum .= $word . ' ';
@@ -1109,7 +1109,7 @@ function search_data($keys, $module, $conditions = NULL) {
  * @return
  *   A string containing HTML for the excerpt.
  */
-function search_excerpt($keys, $text) {
+function search_excerpt($keys, $text, $langcode = NULL) {
   // We highlight around non-indexable or CJK characters.
   $boundary = '(?:(?<=[' . PREG_CLASS_UNICODE_WORD_BOUNDARY . PREG_CLASS_CJK . '])|(?=[' . PREG_CLASS_UNICODE_WORD_BOUNDARY . PREG_CLASS_CJK . ']))';
 
@@ -1156,7 +1156,7 @@ function search_excerpt($keys, $text) {
         $p = $match[0][1];
       }
       else {
-        $info = search_simplify_excerpt_match($key, $text, $included[$key], $boundary);
+        $info = search_simplify_excerpt_match($key, $text, $included[$key], $boundary, $langcode);
         if ($info['where']) {
           $p = $info['where'];
           if ($info['keyword']) {
@@ -1273,10 +1273,10 @@ function _search_excerpt_replace(&$text) {
  *   array with element 'where' giving the position of the match, and element
  *   'keyword' giving the actual word found in the text at that position.
  */
-function search_simplify_excerpt_match($key, $text, $offset, $boundary) {
+function search_simplify_excerpt_match($key, $text, $offset, $boundary, $langcode = NULL) {
   $pos = NULL;
-  $simplified_key = search_simplify($key);
+  $simplified_key = search_simplify($key, $langcode);
-  $simplified_text = search_simplify($text);
+  $simplified_text = search_simplify($text, $langcode);
 
   // Return immediately if simplified key or text are empty.
   if (!$simplified_key || !$simplified_text) {

core/modules/search/tests/modules/search_langcode_test/search_langcode_test.info

@@ -0,0 +1,6 @@
+name = "Test search entity langcode"
+description = "Support module for search module testing."
+package = Testing
+version = VERSION
+core = 8.x
+hidden = TRUE

core/modules/search/tests/modules/search_langcode_test/search_langcode_test.module

@@ -0,0 +1,28 @@
+<?php
+/**
+ * @file
+ * Test module setting up two tests, one for checking if the entity $langcode is
+ * being passed on and another one sets up the alternate verb forms for the
+ * stemming test.
+ */
+
+/**
+ * Implements hook_search_preprocess().
+ */
+function search_langcode_test_search_preprocess($text, $langcode = NULL) {
+  if (isset($langcode) && $langcode == 'en') {
+    // Add the alternate verb forms for the word "testing".
+    if ($text == 'we are testing') {
+      $text .= ' test tested';
+    }
+    // Prints the langcode for testPreprocessLangcode().
+    else {
+      drupal_set_message('Langcode Preprocess Test: ' . $langcode);
+    }
+  }
+  // Prints the langcode for testPreprocessLangcode().
+  elseif (isset($langcode)) {
+    drupal_set_message('Langcode Preprocess Test: ' . $langcode);
+  }
+  return $text;
+}