Fix #1622 : Sanitize captions display

lib/Alchemy/Phrasea/Application.php

@@ -611,10 +611,24 @@ class Application extends SilexApplication
                 $twig->addFilter('base_from_coll', new \Twig_Filter_Function('phrasea::baseFromColl'));
                 $twig->addFilter('AppName', new \Twig_Filter_Function('Alchemy\Phrasea\Controller\Admin\ConnectedUsers::appName'));
                 $twig->addFilter(new \Twig_SimpleFilter('escapeSimpleQuote', function ($value) {
-                    $ret = str_replace("'", "\'", $value);
+                    $ret = str_replace("'", "\\'", $value);
 
                     return $ret;
                 }));
+                $twig->addFilter(new \Twig_SimpleFilter('thesaurus', function (\Twig_Environment $twig, $value) {
+                    if (!$value instanceof \ThesaurusValue) {
+                        return twig_escape_filter($twig, str_replace(array('[[em]]', '[[/em]]'), array('<em>', '</em>'), $value));
+                    }
+
+                    return "<a class=\"bounce\" onclick=\"bounce('" . $value->getField()->get_databox()->get_sbas_id() . "','"
+                        . str_replace("'", "\\'", $value->getQuery())
+                        . "', '"
+                        . str_replace("'", "\\'", $value->getField()->get_name())
+                        . "');return(false);\">"
+                        . twig_escape_filter($twig, str_replace(array('[[em]]', '[[/em]]'), array('<em>', '</em>'), $value->getValue()))
+                        . "</a>";
+                }, array('needs_environment' => true, 'is_safe' => array('html'))));
+
                 $twig->addFilter(new \Twig_SimpleFilter('escapeDoubleQuote', function ($value) {
                     return str_replace('"', '\"', $value);
                 }));

lib/Alchemy/Phrasea/SearchEngine/Phrasea/PhraseaEngine.php

@@ -641,7 +641,7 @@ class PhraseaEngine implements SearchEngineInterface
             if ($sxe && $sxe->description && $sxe->description->$name) {
                 $val = array();
                 foreach ($sxe->description->$name as $value) {
-                    $val[] = str_replace(array('[[em]]', '[[/em]]'), array('<em>', '</em>'), (string) $value);
+                    $val[] = (string) $value;
                 }
                 $separator = $field['separator'] ? $field['separator'][0] : '';
                 $val = implode(' ' . $separator . ' ', $val);

lib/Alchemy/Phrasea/SearchEngine/SphinxSearch/SphinxSearchEngine.php

@@ -552,8 +552,8 @@ class SphinxSearchEngine implements SearchEngineInterface
         }
 
         $opts = array(
-            'before_match' => "<em>",
-            'after_match'  => "</em>",
+            'before_match' => "[[em]]",
+            'after_match'  => "[[/em]]",
         );
 
         $fields_to_send = array();

lib/classes/caption/Field/Value.php

@@ -354,7 +354,7 @@ class caption_Field_Value implements cache_cacheableInterface
         }
 
         // ---------------- new code ----------------------
-        $cleanvalue = str_replace(array("<em>", "</em>", "'"), array("", "", "&apos;"), $value);
+        $cleanvalue = str_replace(array("[[em]]", "[[/em]]", "'"), array("", "", "&apos;"), $value);
 
         list($term_noacc, $context_noacc) = $this->splitTermAndContext($cleanvalue);
         $term_noacc = $this->app['unicode']->remove_indexer_chars($term_noacc);
@@ -388,21 +388,10 @@ class caption_Field_Value implements cache_cacheableInterface
         }
         if($bestnode)
         {
-            list($term, $context) = $this->splitTermAndContext($value);
-            $term = str_replace(array("<em>", "</em>"), array("", ""), $term);
-            $context = str_replace(array("<em>", "</em>"), array("", ""), $context);
-            $qjs = $term;
-            if ($context) {
-                $qjs .= " [" . $context . "]";
-            }
+            list($term, $context) = $this->splitTermAndContext(str_replace(array("[[em]]", "[[/em]]"), array("", ""), $value));
+            $qjs = $term . ($context ? '['.$context.']' : '');
 
-            $value = "<a class=\"bounce\" onclick=\"bounce('" . $databox->get_sbas_id() . "','"
-                . str_replace("'", "\'", $qjs)
-                . "', '"
-                . str_replace("'", "\'", $this->databox_field->get_name())
-                . "');return(false);\">"
-                . $bestnode->getAttribute('v')
-                . "</a>";
+            $value = new ThesaurusValue($bestnode->getAttribute('v'), $this->databox_field, $qjs);
         }
 
         return $value;

templates/web/common/caption_templates/answer.html.twig

@@ -1,6 +1,6 @@
 {% macro format_caption(record, highlight, searchEngine, includeBusiness) %}
     {% for value in record.get_caption().get_highlight_fields(highlight, null, searchEngine, includeBusiness) %}
-        <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{value.value|raw}}</div>
+        <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{ value.value | thesaurus }}</div>
     {% endfor %}
     {% if app['authentication'].getUser().getPrefs('technical_display') == 'group' %}
         <hr/>

templates/web/common/caption_templates/basket_element.html.twig

@@ -1,5 +1,5 @@
 {% macro format_caption(record, highlight, searchEngine, includeBusiness) %}
   {% for value in record.get_caption().get_highlight_fields(highlight, null, searchEngine, includeBusiness) %}
-  <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{value.value|raw}}</div>
+  <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{ value.value | thesaurus }}</div>
   {% endfor %}
 {% endmacro %}

templates/web/common/caption_templates/internal_publi.html.twig

@@ -1,6 +1,6 @@
 {% macro format_caption(record, highlight, searchEngine, includeBusiness) %}
     {% for value in record.get_caption().get_highlight_fields(highlight, null, searchEngine, includeBusiness) %}
-        <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{value.value|raw}}</div>
+        <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{ value.value | thesaurus }}</div>
     {% endfor %}
     {% if app['authentication'].getUser().getPrefs('technical_display') == 'group' %}
         <hr/>

templates/web/common/caption_templates/lazaret.html.twig

@@ -1,6 +1,6 @@
 {% macro format_caption(record, highlight, searchEngine, includeBusiness) %}
     {% for value in record.get_caption().get_highlight_fields(highlight, null, searchEngine, includeBusiness) %}
-        <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{value.value|raw}}</div>
+        <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{ value.value | thesaurus }}</div>
     {% endfor %}
     {% if app['authentication'].getUser().getPrefs('technical_display') == 'group' %}
         <hr/>

templates/web/common/caption_templates/overview.html.twig

@@ -1,5 +1,5 @@
 {% macro format_caption(record, highlight, searchEngine, includeBusiness) %}
   {% for value in record.get_caption().get_highlight_fields(highlight, null, searchEngine, includeBusiness) %}
-  <div><b>{{ value.label }}</b> : {{value.value|raw}}</div>
+  <div><b>{{ value.label }}</b> : {{ value.value }}</div>
   {% endfor %}
 {% endmacro %}

templates/web/common/caption_templates/preview.html.twig

@@ -1,5 +1,5 @@
 {% macro format_caption(record, highlight, searchEngine, includeBusiness) %}
   {% for value in record.get_caption().get_highlight_fields(highlight, null, searchEngine, includeBusiness) %}
-  <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{value.value|raw}}</div>
+  <div class="desc{% if loop.index is odd %}im{% endif %}pair"><b>{{ value.label }}</b> : {{ value.value | thesaurus }}</div>
   {% endfor %}
 {% endmacro %}

tests/Alchemy/Tests/Phrasea/SearchEngine/SearchEngineAbstractTest.php

@@ -754,7 +754,7 @@ abstract class SearchEngineAbstractTest extends \PhraseanetPHPUnitAuthenticatedA
 
         $found = false;
         foreach (self::$searchEngine->excerpt($query_string, $fields, $foundRecord) as $field) {
-            if (strpos($field, '<em>') !== false && strpos($field, '</em>') !== false) {
+            if (strpos($field, '[[em]]') !== false && strpos($field, '[[/em]]') !== false) {
                 $found = true;
                 break;
             }