[TASK] Add special wrap character option for term regex

featdd · Feb 3, 2021 · df1fdec · df1fdec
1 parent 63ce083
commit df1fdec
Show file tree

Hide file tree

Showing 4 changed files with 45 additions and 3 deletions.
diff --git a/Classes/Service/ParserService.php b/Classes/Service/ParserService.php
@@ -41,13 +41,18 @@ class ParserService implements SingletonInterface
     public const REGEX_DELIMITER = '/';
 
     /**
-     * tags to be always ignored by parsing
+     * @var string[]
      */
     public static $alwaysIgnoreParentTags = [
         'a',
         'script',
     ];
 
+    /**
+     * @var string
+     */
+    public static $additionalRegexWrapCharacters = '';
+
     /**
      * @var ContentObjectRenderer
      */
@@ -108,6 +113,14 @@ public function __construct()
                     $this->typoScriptConfiguration['persistence.']['storagePid'])
             );
 
+            $parsingSpecialWrapCharacters = GeneralUtility::trimExplode(',', $this->settings['parsingSpecialWrapCharacters'] ?? '', true);
+
+            if (0 < count($parsingSpecialWrapCharacters)) {
+                foreach ($parsingSpecialWrapCharacters as $parsingSpecialWrapCharacter) {
+                    self::$additionalRegexWrapCharacters .= '|' . preg_quote($parsingSpecialWrapCharacter);
+                }
+            }
+
             try {
                 /** @var \TYPO3\CMS\Core\Context\Context $context */
                 $context = ObjectUtility::makeInstance(Context::class);
@@ -457,9 +470,9 @@ protected function regexParser(string $text, Term $term, int &$replacements, Clo
          * i = ignores camel case
          */
         $regex = self::REGEX_DELIMITER .
-            '(^|\G|[\s\>[:punct:]]|\<br*\>)' .
+            '(^|\G|[\s\>[:punct:]]|\<br*\>' . self::$additionalRegexWrapCharacters . ')' .
             '(' . preg_quote($term->getName(), self::REGEX_DELIMITER) . ')' .
-            '($|[\s\<[:punct:]]|\<br*\>)' .
+            '($|[\s\<[:punct:]]|\<br*\>' . self::$additionalRegexWrapCharacters . ')' .
             '(?![^<]*>|[^<>]*<\/)' .
             self::REGEX_DELIMITER .
             (false === $term->isCaseSensitive() ? 'i' : '');

diff --git a/Configuration/TypoScript/constants.txt b/Configuration/TypoScript/constants.txt
@@ -21,6 +21,8 @@ plugin.tx_dpnglossary {
     parsingPids = 0
     # cat=dpn_glossary/settings/a; type=string; label=Pages (comma list) which should not be parsed for glossary terms
     parsingExcludePidList =
+    # cat=dpn_glossary/settings/a; type=string; label=Additional special characters to respect in parser regular expression
+    parsingSpecialWrapCharacters =
     # cat=dpn_glossary/settings/a; type=integer; label=Maximum replacements for each term (default: -1 = any)
     maxReplacementPerPage = -1
     # cat=dpn_glossary/settings/a; type=boolean; label=Respect replacement counter when parsing synonyms (default: off)

diff --git a/Configuration/TypoScript/setup.txt b/Configuration/TypoScript/setup.txt
@@ -27,6 +27,8 @@ plugin.tx_dpnglossary {
     parsingPids = {$plugin.tx_dpnglossary.settings.parsingPids}
     # pages where terms should NOT be searched
     parsingExcludePidList = {$plugin.tx_dpnglossary.settings.parsingExcludePidList}
+    # special characters for the regex
+    parsingSpecialWrapCharacters = {$plugin.tx_dpnglossary.settings.parsingSpecialWrapCharacters}
     # max replacemnet per page, default = -1 (any matches)
     maxReplacementPerPage = {$plugin.tx_dpnglossary.settings.maxReplacementPerPage}
     # respect max replacement when parsing synonyms

diff --git a/Documentation/Configuration/ExtensionSettings/Index.rst b/Documentation/Configuration/ExtensionSettings/Index.rst
@@ -221,4 +221,29 @@ plugin.tx_dpnglossary.settings:
   Default
     1
 
+.. container:: table-row
+
+  Constant
+    settings.priorisedSynonymParsing
+
+  Data Type
+    boolean
+
+  Description
+    Parse for synonyms before the actual term
+
+  Default
+    1
+
+.. container:: table-row
+
+  Constant
+    settings.parsingSpecialWrapCharacters
+
+  Data Type
+    string
+
+  Description
+    Comma list of special characters allowed to wrap the term
+
 .. ###### END~OF~TABLE ######