Add option fulltextregex to search page content with regex

inc/pagequery.php

@@ -831,6 +831,53 @@ function page_search($query) {
         $result = array_keys(ft_pageSearch($query, $highlight));
         return $result;
     }
+
+
+    /**
+     * use regex to search all page content
+     *
+     * @param $query
+     * @param $incl_ns
+     * @param $excl_ns
+     * @return mixed
+     */
+    function page_search_regex($query, $incl_ns, $excl_ns) {
+
+        $query = trim($query);
+        $pages = idx_get_indexer()->getPages();
+
+        // first deal with excluded namespaces, then included, order matters!
+        $pages = $this->_filter_ns($pages, $excl_ns, true);
+        $pages = $this->_filter_ns($pages, $incl_ns, false);
+
+        $cnt = count($pages);
+        for ($i = 0; $i < $cnt; $i++) {
+            $page = $pages[$i];
+
+            $text = io_readFile(wikiFN($page, '', false), false);
+
+            $matched = preg_match('/' . $query . '/iS', $text);
+            if ($matched === false) {
+                return false;
+            } elseif ($matched == 0) {
+                unset($pages[$i]);
+            }
+        }
+
+        foreach($pages as $key => $page) {
+            if ( ! page_exists($page) || isHiddenPage($page)) {
+                unset($pages[$key]);
+            }
+        }
+
+        if (count($pages) > 0) {
+            return $pages;
+        } else {
+            // we always return an array type
+            return array();
+        }
+    }
+
 
 
     /**
@@ -841,7 +888,7 @@ function page_lookup($query, $fullregex, $incl_ns, $excl_ns) {
         global $conf;
 
         $query = trim($query);
-        $pages = file($conf['indexdir'] . '/page.idx');
+        $pages = idx_get_indexer()->getPages();
 
         if ( ! $fullregex) {
             // first deal with excluded namespaces, then included, order matters!
@@ -864,7 +911,7 @@ function page_lookup($query, $fullregex, $incl_ns, $excl_ns) {
              *  and the page-exists check above
              * The @ prevents problems with invalid queries!
              */
-            $matched = @preg_match('/' . $query . '/i', $page);
+            $matched = @preg_match('/' . $query . '/iS', $page);
             if ($matched === false) {
                 return false;
             } elseif ($matched == 0) {

res/toolbar

@@ -1,6 +1,7 @@
 pagequery>	* | .* | exclude ns: ^ |-ns: | include ns: @ | ns: | . | .. | text | regex
 fulltext	Full-text search of page content
 fullregex	Search the full page id using regex
+fulltextregex Full-text search of page content using regex
 sort=column:direction	Keys to sort by, in order of sorting
 - a, ab, abc	By 1st letter, 2-letters, or 3-letters
 - name, pagename	By page name [not grouped]

syntax.php

@@ -76,6 +76,7 @@ function handle($match, $state, $pos, Doku_Handler $handler) {
         $opt['fontsize']  = '';         // base fontsize of pagequery; best to use %
         $opt['fullregex'] = false;      // power-user regex search option--file name only
         $opt['fulltext']  = false;      // search full-text; including file contents
+        $opt['fulltextregex']  = false; // search full-text; including file contents, using preg_match
         $opt['group']     = false;      // group the results based on sort headings
         $opt['hidejump']  = false;      // hide the jump to top link
         $opt['hidemsg']   = false;      // hide any error messages
@@ -99,6 +100,7 @@ function handle($match, $state, $pos, Doku_Handler $handler) {
                 case 'casesort':
 				case 'fullregex':
                 case 'fulltext':
+                case 'fulltextregex':
                 case 'group':
                 case 'hidejump':
                 case 'hidemsg':
@@ -272,8 +274,14 @@ function render($mode, Doku_Renderer $renderer, $opt) {
                 if ($query == '*') {
                     $query = '.*';
                 }
-                // search by page name or path only
-                $results = $pq->page_lookup($query, $opt['fullregex'], $incl_ns, $excl_ns);
+
+                if($opt['fulltextregex'] == true) {
+                    // search by page name or path only
+                    $results = $pq->page_search_regex($query, $incl_ns, $excl_ns);
+                } else {
+                    // search by page name or path only
+                    $results = $pq->page_lookup($query, $opt['fullregex'], $incl_ns, $excl_ns);
+                }
             }
             $results = $pq->validate_pages($results, $opt['hidestart'], $opt['maxns']);