Upgraded GeSHi to 1.0.8.6

1 files changed, 49 insertions, 38 deletions

diff --git a/inc/geshi.php b/inc/geshi.php
index abe69a2bd..14acf15a6 100644
--- a/inc/geshi.php
+++ b/inc/geshi.php
@@ -41,7 +41,7 @@
 //
 
 /** The version of this GeSHi file */
-define('GESHI_VERSION', '1.0.8.4');
+define('GESHI_VERSION', '1.0.8.6');
 
 // Define the root directory for the GeSHi code tree
 if (!defined('GESHI_ROOT')) {
@@ -207,8 +207,10 @@ define('GESHI_NUMBER_BIN_PREFIX_PERCENT', 32);   //%[01]+
 define('GESHI_NUMBER_BIN_PREFIX_0B', 64);        //0b[01]+
 /** Number format to highlight octal numbers with a leading zero */
 define('GESHI_NUMBER_OCT_PREFIX', 256);           //0[0-7]+
+/** Number format to highlight octal numbers with a prefix 0o (logtalk) */
+define('GESHI_NUMBER_OCT_PREFIX_0O', 512);           //0[0-7]+
 /** Number format to highlight octal numbers with a suffix of o */
-define('GESHI_NUMBER_OCT_SUFFIX', 512);           //[0-7]+[oO]
+define('GESHI_NUMBER_OCT_SUFFIX', 1024);           //[0-7]+[oO]
 /** Number format to highlight hex numbers with a prefix 0x */
 define('GESHI_NUMBER_HEX_PREFIX', 4096);           //0x[0-9a-fA-F]+
 /** Number format to highlight hex numbers with a suffix of h */
@@ -1084,13 +1086,14 @@ class GeSHi {
      * @param string  The style to make the escape characters
      * @param boolean Whether to merge the new styles with the old or just
      *                to overwrite them
+     * @param int     Tells the group of strings for which style should be set.
      * @since 1.0.0
      */
-    function set_strings_style($style, $preserve_defaults = false) {
+    function set_strings_style($style, $preserve_defaults = false, $group = 0) {
         if (!$preserve_defaults) {
-            $this->language_data['STYLES']['STRINGS'][0] = $style;
+            $this->language_data['STYLES']['STRINGS'][$group] = $style;
         } else {
-            $this->language_data['STYLES']['STRINGS'][0] .= $style;
+            $this->language_data['STYLES']['STRINGS'][$group] .= $style;
         }
     }
 
@@ -1132,13 +1135,14 @@ class GeSHi {
      * @param string  The style to make the numbers
      * @param boolean Whether to merge the new styles with the old or just
      *                to overwrite them
+     * @param int     Tells the group of numbers for which style should be set.
      * @since 1.0.0
      */
-    function set_numbers_style($style, $preserve_defaults = false) {
+    function set_numbers_style($style, $preserve_defaults = false, $group = 0) {
         if (!$preserve_defaults) {
-            $this->language_data['STYLES']['NUMBERS'][0] = $style;
+            $this->language_data['STYLES']['NUMBERS'][$group] = $style;
         } else {
-            $this->language_data['STYLES']['NUMBERS'][0] .= $style;
+            $this->language_data['STYLES']['NUMBERS'][$group] .= $style;
         }
     }
 
@@ -1370,6 +1374,7 @@ class GeSHi {
                 'delphi' => array('dpk', 'dpr', 'pp', 'pas'),
                 'diff' => array('diff', 'patch'),
                 'dos' => array('bat', 'cmd'),
+                'gdb' => array('kcrash', 'crash', 'bt'),
                 'gettext' => array('po', 'pot'),
                 'gml' => array('gml'),
                 'gnuplot' => array('plt'),
@@ -1966,31 +1971,33 @@ class GeSHi {
             //All this formats are matched case-insensitively!
             static $numbers_format = array(
                 GESHI_NUMBER_INT_BASIC =>
-                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])([1-9]\d*?|0)(?![0-9a-z]|\.(?!(?m:$)))',
+                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])([1-9]\d*?|0)(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_INT_CSTYLE =>
-                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])([1-9]\d*?|0)l(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])([1-9]\d*?|0)l(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_BIN_SUFFIX =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])[01]+?b(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])[01]+?[bB](?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_BIN_PREFIX_PERCENT =>
-                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])%[01]+?(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])%[01]+?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_BIN_PREFIX_0B =>
-                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])0b[01]+?(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])0b[01]+?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_OCT_PREFIX =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])0[0-7]+?(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])0[0-7]+?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
+                GESHI_NUMBER_OCT_PREFIX_0O =>
+                    '(?<![0-9a-z_\.%])(?<![\d\.]e[+\-])0o[0-7]+?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_OCT_SUFFIX =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])[0-7]+?o(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])[0-7]+?o(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_HEX_PREFIX =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])0x[0-9a-f]+?(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])0x[0-9a-fA-F]+?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_HEX_SUFFIX =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])\d[0-9a-f]*?h(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])\d[0-9a-fA-F]*?[hH](?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_FLT_NONSCI =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])\d+?\.\d+?(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])\d+?\.\d+?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_FLT_NONSCI_F =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])(?:\d+?(?:\.\d*?)?|\.\d+?)f(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])(?:\d+?(?:\.\d*?)?|\.\d+?)f(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_FLT_SCI_SHORT =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])\.\d+?(?:e[+\-]?\d+?)?(?![0-9a-z\.])',
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])\.\d+?(?:e[+\-]?\d+?)?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)',
                 GESHI_NUMBER_FLT_SCI_ZERO =>
-                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])(?:\d+?(?:\.\d*?)?|\.\d+?)(?:e[+\-]?\d+?)?(?![0-9a-z\.])'
+                    '(?<![0-9a-z_\.])(?<![\d\.]e[+\-])(?:\d+?(?:\.\d*?)?|\.\d+?)(?:e[+\-]?\d+?)?(?![0-9a-z]|\.(?:[eE][+\-]?)?\d)'
                 );
 
             //At this step we have an associative array with flag groups for a
@@ -2012,7 +2019,7 @@ class GeSHi {
                 }
 
                 $this->language_data['NUMBERS_RXCACHE'][$key] =
-                    "/(?<!<\|\/)(?<!<\|!REG3XP)(?<!<\|\/NUM!)(?<!\d\/>)($regexp)(?!\|>)(?![^\"\|\>\<]+<)/i";
+                    "/(?<!<\|\/)(?<!<\|!REG3XP)(?<!<\|\/NUM!)(?<!\d\/>)($regexp)(?!(?:<DOT>|[^\<])+>)(?![^<]*>)(?!\|>)(?!\/>)/i"; //
             }
         }
 
@@ -2033,6 +2040,10 @@ class GeSHi {
         // Start the timer
         $start_time = microtime();
 
+        // Replace all newlines to a common form.
+        $code = str_replace("\r\n", "\n", $this->source);
+        $code = str_replace("\r", "\n", $code);
+
         // Firstly, if there is an error, we won't highlight
         if ($this->error) {
             //Escape the source for output
@@ -2053,13 +2064,6 @@ class GeSHi {
             $this->build_parse_cache();
         }
 
-        // Replace all newlines to a common form.
-        $code = str_replace("\r\n", "\n", $this->source);
-        $code = str_replace("\r", "\n", $code);
-
-        // Add spaces for regular expression matching and line numbers
-//        $code = "\n" . $code . "\n";
-
         // Initialise various stuff
         $length           = strlen($code);
         $COMMENT_MATCHED  = false;
@@ -3087,7 +3091,7 @@ class GeSHi {
         $result = preg_replace('/^ /m', '&nbsp;', $result);
         $result = str_replace('  ', ' &nbsp;', $result);
 
-        if ($this->line_numbers == GESHI_NO_LINE_NUMBERS) {
+        if ($this->line_numbers == GESHI_NO_LINE_NUMBERS && $this->header_type != GESHI_HEADER_PRE_TABLE) {
             if ($this->line_ending === null) {
                 $result = nl2br($result);
             } else {
@@ -3430,7 +3434,7 @@ class GeSHi {
         //FIX for symbol highlighting ...
         if ($this->lexic_permissions['SYMBOLS'] && !empty($this->language_data['SYMBOLS'])) {
             //Get all matches and throw away those witin a block that is already highlighted... (i.e. matched by a regexp)
-            $n_symbols = preg_match_all("/<\|(?:<DOT>|[^>])+>(?:(?!\|>).*?)\|>|<\/a>|(?:" . $this->language_data['SYMBOL_SEARCH'] . ")+/", $stuff_to_parse, $pot_symbols, PREG_OFFSET_CAPTURE | PREG_SET_ORDER);
+            $n_symbols = preg_match_all("/<\|(?:<DOT>|[^>])+>(?:(?!\|>).*?)\|>|<\/a>|(?:" . $this->language_data['SYMBOL_SEARCH'] . ")+(?![^<]+?>)/", $stuff_to_parse, $pot_symbols, PREG_OFFSET_CAPTURE | PREG_SET_ORDER);
             $global_offset = 0;
             for ($s_id = 0; $s_id < $n_symbols; ++$s_id) {
                 $symbol_match = $pot_symbols[$s_id][0][0];
@@ -3969,16 +3973,16 @@ class GeSHi {
          * @todo   Document behaviour change - class is outputted regardless of whether
          *         we're using classes or not. Same with style
          */
-        $attributes = ' class="' . $this->language;
+        $attributes = ' class="' . $this->_genCSSName($this->language);
         if ($this->overall_class != '') {
-            $attributes .= " ".$this->overall_class;
+            $attributes .= " ".$this->_genCSSName($this->overall_class);
         }
         $attributes .= '"';
 
         if ($this->overall_id != '') {
             $attributes .= " id=\"{$this->overall_id}\"";
         }
-        if ($this->overall_style != '') {
+        if ($this->overall_style != '' && !$this->use_classes) {
             $attributes .= ' style="' . $this->overall_style . '"';
         }
 
@@ -4215,6 +4219,10 @@ class GeSHi {
         return strtr($string, $aTransSpecchar);
     }
 
+    function _genCSSName($name){
+        return (is_numeric($name[0]) ? '_' : '') . $name;
+    }
+
     /**
      * Returns a stylesheet for the highlighted code. If $economy mode
      * is true, we only return the stylesheet declarations that matter for
@@ -4242,11 +4250,11 @@ class GeSHi {
         // that should be used, the same for a class. Otherwise, a selector
         // of '' means that these styles will be applied anywhere
         if ($this->overall_id) {
-            $selector = '#' . $this->overall_id;
+            $selector = '#' . $this->_genCSSName($this->overall_id);
         } else {
-            $selector = '.' . $this->language;
+            $selector = '.' . $this->_genCSSName($this->language);
             if ($this->overall_class) {
-                $selector .= '.' . $this->overall_class;
+                $selector .= '.' . $this->_genCSSName($this->overall_class);
             }
         }
         $selector .= ' ';
@@ -4555,7 +4563,10 @@ class GeSHi {
         // make sure the last tokens get converted as well
         $new_entry = $this->_optimize_regexp_list_tokens_to_string($tokens);
         if (GESHI_MAX_PCRE_SUBPATTERNS && $num_subpatterns + substr_count($new_entry, '(?:') > GESHI_MAX_PCRE_SUBPATTERNS) {
-            $regexp_list[++$list_key] = $new_entry;
+            if ( !empty($regexp_list[$list_key]) ) {
+              ++$list_key;
+            }
+            $regexp_list[$list_key] = $new_entry;
         } else {
             if (!empty($regexp_list[$list_key])) {
                 $new_entry = '|' . $new_entry;