MDL-53579 search: Fix conflict with eDisMax parser and indexed files

search/engine/solr/classes/document.php

@@ -63,17 +63,25 @@ class document extends \core_search\document {
         'solr_fileid' => array(
             'type' => 'string',
             'stored' => true,
-            'indexed' => false
+            'indexed' => true
         ),
         'solr_filecontenthash' => array(
             'type' => 'string',
             'stored' => true,
-            'indexed' => false
+            'indexed' => true
         ),
+        // Stores the status of file indexing.
         'solr_fileindexedcontent' => array(
             'type' => 'int',
             'stored' => true,
             'indexed' => true
+        ),
+        // Field to index, but not store, file contents.
+        'solr_filecontent' => array(
+            'type' => 'text',
+            'stored' => false,
+            'indexed' => true,
+            'mainquery' => true
         )
     );
 

search/engine/solr/classes/engine.php

@@ -234,7 +234,7 @@ class engine extends \core_search\engine {
         $fields = $documentclass::get_default_fields_definition();
 
         $dismax = false;
-        if ($query instanceof SolrDisMaxQuery) {
+        if ($query instanceof \SolrDisMaxQuery) {
             $dismax = true;
         }
 
@@ -764,6 +764,11 @@ class engine extends \core_search\engine {
         // This will prevent solr from automatically making fields for every tika output.
         $url->param('uprefix', 'ignored_');
 
+        // Control how content is captured. This will keep our file content clean of non-important metadata.
+        $url->param('captureAttr', 'true');
+        // Move the content to a field for indexing.
+        $url->param('fmap.content', 'solr_filecontent');
+
         // These are common fields that matches the standard *_point dynamic field and causes an error.
         $url->param('fmap.media_white_point', 'ignored_mwp');
         $url->param('fmap.media_black_point', 'ignored_mbp');

search/engine/solr/tests/engine_test.php

@@ -165,6 +165,11 @@ class search_solr_engine_testcase extends advanced_testcase {
         $this->assertEquals($USER->id, $results[0]->get('userid'));
         $this->assertEquals(\context_system::instance()->id, $results[0]->get('contextid'));
 
+        // Do a test to make sure we aren't searching non-query fields, like areaid.
+        $querydata->q = \core_search\manager::generate_areaid('core_mocksearch', 'role_capabilities');
+        $this->assertCount(0, $this->search->search($querydata));
+        $querydata->q = 'message';
+
         sleep(1);
         $beforeadding = time();
         sleep(1);