Changeset 57

Timestamp:

03/03/2009 10:34:39 AM (3 years ago)

Author:

ofer

Message:

RSS becomes translateable due to hack, removed lots of whitespaces

Location:

trunk/WordPress/plugin/transposh

Files:

• parser.php (modified) (18 diffs)
• transposh.php (modified) (44 diffs)

trunk/WordPress/plugin/transposh/parser.php

@@ -21 +21 @@
  * fetch translation and update the translated page.
  * This file should include only general purpose parser functionality while using callbacks
- * to obtain WorkdPress specific capabilities, e.g. db access. 
+ * to obtain WorkdPress specific capabilities, e.g. db access.
  */
 
@@ -27 +27 @@
 require_once("constants.php");
 
-//The language to which the current page will be translated to. 
+//The language to which the current page will be translated to.
 $lang;
- 
+
 //The html page which starts contains the content being translated
 $page;
- 
+
 //Marks the current position of the translation process within the page
 $pos = 0;
@@ -39 +39 @@
 $tags_list = array();
 
-//The translated html page 
+//The translated html page
 $tr_page;
- 
-//Points to the last character that have been copied from the original to the translated page. 
+
+//Points to the last character that have been copied from the original to the translated page.
 $tr_mark = 0;
 
-//Is the current use is in edit mode. 
+//Is the current use is in edit mode.
 $is_edit_mode = FALSE;
 
@@ -56 +56 @@
 /*
  * Parse the html page into tags, identify translateable string which
- * will be translated. 
+ * will be translated.
  */
 function process_html()
 {
-    logger("Enter " . __METHOD__, 4);
-
-    global $page, $tr_page, $pos, $tags_list, $lang;
-    $no_translate = 0;
-    $page_length = strlen($page);
-    
-    while($pos < $page_length)
-    {
-        //find beginning of next tag
-        $pos = strpos($page, '<', $pos);
-        if($pos === FALSE)
-        {
-            //logger("Error finding < in pos " . $pos . " page: " . $page);
-            break;
-        }
-        $pos++;
-
-        //Get the element identifying this tag
-        $element = get_element();
-                
-        if(should_skip_element($element))
-        {
-            //do nothing
-        }
-        else
-        {
-            //Mark tag start position
-            $tag_start = $pos;
-            
-            //skip to the '>' marking the end of the element
-            $pos = strpos($page, '>', $pos);
-
-            //Mark tag end position
-            $tag_end = $pos;
-            
-            if($page[$pos-1] == '/')
-            {
-                //single line tag - no need to update tags list
-                process_tag_init($element, $tag_start, $tag_end);
-            }
-            else if($element[0] != '/') 
-            {
-                if(!$no_translate)
-                {
-                    process_tag_init($element, $tag_start, $tag_end);
-                }
-                
-                $tags_list[] = $element;
-
-                //Look for the no translate class 
-                if(stripos($element, NO_TRANSLATE_CLASS) !== FALSE)
-                {
-                    $no_translate++;
-                }
-            }
-            else
-            {
-                $popped_element = array_pop($tags_list);
-                if(!$no_translate)
-                {
-                    process_tag_termination($element);
-                }
-                
-                //Look for the no translate class 
-                if(stripos($popped_element, NO_TRANSLATE_CLASS) !== FALSE)
-                {
-                    $no_translate--;
-                }
-            }
-
-            $pos++;
-
-            //skip processing while enclosed within a tag marked by no_translate
-            if(!$no_translate)
-            {
-                process_current_tag();
-            }
-            
-        }
-    }
-
-    if(strlen($tr_page) > 0)
-    {
-        //Some translation has been taken place. Complete the translated
-        //page up to the full contents of the original page.
-        update_translated_page(strlen($page), -1, "");
-    }
-    
-    logger("Exit " . __METHOD__, 4);
+        logger("Enter " . __METHOD__, 4);
+
+        global $page, $tr_page, $pos, $tags_list, $lang;
+        $no_translate = 0;
+        $page_length = strlen($page);
+
+        while($pos < $page_length)
+        {
+                //find beginning of next tag
+                $pos = strpos($page, '<', $pos);
+                if($pos === FALSE)
+                {
+                        //logger("Error finding < in pos " . $pos . " page: " . $page);
+                        break;
+                }
+                $pos++;
+
+                //Get the element identifying this tag
+                $element = get_element();
+
+                if(should_skip_element($element))
+                {
+                        logger ("skipping element: $element");
+                        //do nothing
+                }
+                else
+                {
+                        //Mark tag start position
+                        $tag_start = $pos;
+
+                        //skip to the '>' marking the end of the element
+                        if ($element == "!") {
+                                $pos = strpos($page, '[', $pos);
+                        }else {
+                                $pos = strpos($page, '>', $pos);
+                        }
+
+                        //Mark tag end position
+                        $tag_end = $pos;
+
+                        if($page[$pos-1] == '/')
+                        {
+                                //single line tag - no need to update tags list
+                                process_tag_init($element, $tag_start, $tag_end);
+                        }
+                        else if($element[0] != '/')
+                        {
+                                if(!$no_translate)
+                                {
+                                        process_tag_init($element, $tag_start, $tag_end);
+                                }
+
+                                $tags_list[] = $element;
+
+                                //Look for the no translate class
+                                if(stripos($element, NO_TRANSLATE_CLASS) !== FALSE)
+                                {
+                                        $no_translate++;
+                                }
+                        }
+                        else
+                        {
+                                $popped_element = array_pop($tags_list);
+                                if(!$no_translate)
+                                {
+                                        process_tag_termination($element);
+                                }
+
+                                //Look for the no translate class
+                                if(stripos($popped_element, NO_TRANSLATE_CLASS) !== FALSE)
+                                {
+                                        $no_translate--;
+                                }
+                        }
+
+                        $pos++;
+
+                        //skip processing while enclosed within a tag marked by no_translate
+                        if(!$no_translate)
+                        {
+                                process_current_tag();
+                        }
+
+                }
+        }
+
+        if(strlen($tr_page) > 0)
+        {
+                //Some translation has been taken place. Complete the translated
+                //page up to the full contents of the original page.
+                update_translated_page(strlen($page), -1, "");
+        }
+
+        logger("Exit " . __METHOD__, 4);
 }
 
@@ -155 +160 @@
  * Determine if the specified element should be skipped. If so the position
  * is moved past end of tag.
- * Return TRUE if element is skipped otherwise FALSE. 
+ * Return TRUE if element is skipped otherwise FALSE.
  */
 function should_skip_element(&$element)
 {
-    global $page, $pos;
-    $rc = TRUE;
-    
-    if(strncmp($element, "!DOCTYPE", 8) == 0)
-    {
-        $pos = strpos($page, '>', $pos);
-    }
-    else if(strncmp($element, "!--", 3) == 0)
-    {
-        $pos = strpos($page, '-->', $pos);
-    }
-    else
-    {
-        $rc = FALSE;
-    }
-
-    return $rc;
+        global $page, $pos;
+        $rc = TRUE;
+
+        if(strncmp($element, "!DOCTYPE", 8) == 0)
+        {
+                $pos = strpos($page, '>', $pos);
+        }
+        else if(strncmp($element, "!--", 3) == 0)
+        {
+                $pos = strpos($page, '-->', $pos);
+        }
+        else
+        {
+                $rc = FALSE;
+        }
+
+        return $rc;
 }
 
@@ -184 +189 @@
 function process_tag_init(&$element, $start, $end)
 {
-    switch ($element)
-    {
-        case 'a':
-            process_anchor_tag($start, $end);
-            break;
-        case 'div' :     
-        case 'span':
-            process_span_or_div_tag($element, $start, $end);
-            break;
-        case 'html':
-            process_html_tag($start, $end);
-            break;
-        case 'body':
-            global $is_in_body;
-            $is_in_body = TRUE;
-            break;
-            
-            
-    }
-    
+        switch ($element)
+        {
+                case 'a':
+                        process_anchor_tag($start, $end);
+                        break;
+                case 'div' :
+                case 'span':
+                        //case 'description':
+                        //case 'content:encoded':
+                        logger("in case : $element",1);
+                        process_span_or_div_tag($element, $start, $end);
+                        break;
+                        case 'html':
+                                process_html_tag($start, $end);
+                                break;
+                        case 'body':
+                        case 'channel':
+                                global $is_in_body;
+                                $is_in_body = TRUE;
+                                break;
+        }
+
 }
 
@@ -214 +221 @@
 function process_span_or_div_tag(&$element, $start, $end)
 {
-    
-    $cls = get_attribute($start, $end, 'class');
-
-    if($cls == NULL)
-    {
-        return;
-    }
-
-    //Look for the no translate class 
-    if(stripos($cls, NO_TRANSLATE_CLASS) === FALSE)
-    {
-        return;
-    }
-
-    //Mark the element as not translatable
-    $element .= "." . NO_TRANSLATE_CLASS;
-}
-
-
-/*
- * Process html tag. Set the direction for rtl languages. 
+
+        $cls = get_attribute($start, $end, 'class');
+
+        if($cls == NULL)
+        {
+                return;
+        }
+
+        //Look for the no translate class
+        if(stripos($cls, NO_TRANSLATE_CLASS) === FALSE)
+        {
+                return;
+        }
+
+        //Mark the element as not translatable
+        $element .= "." . NO_TRANSLATE_CLASS;
+}
+
+
+/*
+ * Process html tag. Set the direction for rtl languages.
  *
  */
 function process_html_tag($start, $end)
 {
-    global $lang, $rtl_languages;
-
-    if(!(in_array ($lang, $rtl_languages)))
-    {
-        return;
-    }
-    
-    $dir = get_attribute($start, $end, 'dir');
-
-    if($dir == NULL)
-    {
-
-        //attribute does not exist - add it
-        update_translated_page($end, -1, 'dir="rtl"');
-    }
-    else
-    {
-        $dir = 'rtl';
-
-        //rewrite url in translated page
-        update_translated_page($start, $end, $dir);
-        
-    }
-    logger(__METHOD__ . " Changed page direction to rtl");
+        global $lang, $rtl_languages;
+
+        if(!(in_array ($lang, $rtl_languages)))
+        {
+                return;
+        }
+
+        $dir = get_attribute($start, $end, 'dir');
+
+        if($dir == NULL)
+        {
+
+                //attribute does not exist - add it
+                update_translated_page($end, -1, 'dir="rtl"');
+        }
+        else
+        {
+                $dir = 'rtl';
+
+                //rewrite url in translated page
+                update_translated_page($start, $end, $dir);
+
+        }
+        logger(__METHOD__ . " Changed page direction to rtl");
 }
 
@@ -272 +279 @@
 function process_tag_termination(&$element)
 {
-    global $pos, $tags_list, $page;
-    
-    logger(__METHOD__ . " $element ". $page[$pos], 4);
-}
-
-
-/*
- * Return the element id within the current tag. 
+        global $pos, $tags_list, $page;
+
+        logger(__METHOD__ . " $element ". $page[$pos], 4);
+}
+
+
+/*
+ * Return the element id within the current tag.
  */
 function get_element()
 {
-    logger("Enter " . __METHOD__, 5);
-    global $page, $pos;
-    
-    skip_white_space();
-    
-    $start = $pos;
-    
-    //keep scanning till the first white space or the '>' mark
-    while($pos < strlen($page) && $page[$pos] != ' '&&
-          $page[$pos] != '>' && $page[$pos] != '\t')
-    {
-        $pos++;
-    }
-    
-    logger("Exit " . __METHOD__, 5);
-    return substr($page,$start, $pos - $start);
+        global $page, $pos;
+        logger("Enter " . __METHOD__. ": $pos", 5);
+
+        skip_white_space();
+
+        $start = $pos;
+
+        //keep scanning till the first white space or the '>' mark
+        // TODO - for CDATA, check '['
+        while($pos < strlen($page) && $page[$pos] != ' ' && $page[$pos] != '[' &&
+        $page[$pos] != '>' && $page[$pos] != '\t')
+        {
+                $pos++;
+        }
+
+        logger("Exit " . __METHOD__. ": $pos", 5);
+        return substr($page,$start, $pos - $start);
 }
 
@@ -306 +314 @@
  * Returns the string containing the attribute if available otherwise NULL.
  * In addition the start and end position are moved to boundaries of the
- * attribute's value. 
+ * attribute's value.
  */
 function get_attribute(&$start, &$end, $id)
 {
-    global $page;
-        
-    //look for the id within the given limits.
-    while($start < $end)
-    {
-        $index = 0;
-    
-        while($start < $end && $page[$start + $index] == $id[$index]
-              && $index < strlen($id))
-        {
-            $index++;
-        }
-
-        if($index == strlen($id))
-        {
-            //we have match
-            break;
-        }
-
-        $start++;
-    }
-
-    if($start == $end)
-    {
-        return NULL;
-    }
-    
-    //look for the " or ' marking start of attribute's value
-    while($start < $end && $page[$start] != '"' && $page[$start] != "'")
-    {
-        $start++;
-    }
-
-    $start++;
-    if($start >= $end)
-    {
-        return NULL;
-    }
-
-    $tmp = $start + 1;
-    //look for the " or ' marking the end of attribute's value
-    while($tmp < $end && $page[$tmp] != '"' && $page[$tmp] != "'")
-    {
-        $tmp++;
-    }
-    
-    $end = $tmp - 1;
-    
-        
-    return substr($page, $start, $end - $start + 1);
+        global $page;
+
+        //look for the id within the given limits.
+        while($start < $end)
+        {
+                $index = 0;
+
+                while($start < $end && $page[$start + $index] == $id[$index]
+                && $index < strlen($id))
+                {
+                        $index++;
+                }
+
+                if($index == strlen($id))
+                {
+                        //we have match
+                        break;
+                }
+
+                $start++;
+        }
+
+        if($start == $end)
+        {
+                return NULL;
+        }
+
+        //look for the " or ' marking start of attribute's value
+        while($start < $end && $page[$start] != '"' && $page[$start] != "'")
+        {
+                $start++;
+        }
+
+        $start++;
+        if($start >= $end)
+        {
+                return NULL;
+        }
+
+        $tmp = $start + 1;
+        //look for the " or ' marking the end of attribute's value
+        while($tmp < $end && $page[$tmp] != '"' && $page[$tmp] != "'")
+        {
+                $tmp++;
+        }
+
+        $end = $tmp - 1;
+
+
+        return substr($page, $start, $end - $start + 1);
 }
 
@@ -369 +377 @@
 function process_current_tag()
 {
-    global $page, $pos, $tags_list, $is_in_body;
-
-    $current_tag = end($tags_list);
-    
-    logger("Enter " . __METHOD__  ." : $current_tag", 4);
-
-    //translate only elements within the body or title
-    if($is_in_body || $current_tag == 'title')
-    {
-        skip_white_space();
-        $start = $pos;
-        $page_length =  strlen($page);
-
-        while($pos < $page_length && $page[$pos] != '<')
-        {
-            //will break translation unit when one of the following characters is reached: ., 
-            if(is_sentence_breaker($pos))
-             {
-                 translate_text($start);
-                 $pos++;
-                 $start = $pos;
-             }
-            else if(($end_of_entity = is_html_entity($pos)))
-             {
-                 translate_text($start);
-                 $pos++;
-                 $start = $end_of_entity;
-             }
-             else
-             {
-                 $pos++;
-             }
-        }
-        
-        if($pos > $start)
-        {
-            translate_text($start);
-        }
-    }
-    logger("Exit" .  __METHOD__ . " : $current_tag" , 4);
+        global $page, $pos, $tags_list, $is_in_body;
+
+        $current_tag = end($tags_list);
+
+        logger("Enter " . __METHOD__  ." : $current_tag", 4);
+
+        //translate only elements within the body or title
+        if($is_in_body || $current_tag == 'title')
+        {
+                skip_white_space();
+                $start = $pos;
+                $page_length =  strlen($page);
+
+                while($pos < $page_length && $page[$pos] != '<')
+                {
+                        //will break translation unit when one of the following characters is reached: .,
+                        if(is_sentence_breaker($pos))
+                        {
+                                translate_text($start);
+                                $pos++;
+                                $start = $pos;
+                        }
+                        else if(($end_of_entity = is_html_entity($pos)))
+                        {
+                                translate_text($start);
+                                $pos++;
+                                $start = $end_of_entity;
+                        }
+                        else
+                        {
+                                $pos++;
+                        }
+                }
+
+                if($pos > $start)
+                {
+                        translate_text($start);
+                }
+        }
+        logger("Exit" .  __METHOD__ . " : $current_tag" , 4);
 }
 
@@ -419 +427 @@
 function is_sentence_breaker($position)
 {
-    global $page;
-    $rc = FALSE;
-
-    if($page[$position] == '.' || $page[$position] == '-')
-    {
-        //Only break if the next character is a white space,
-        //in order to avoid breaks on cases like this: (hello world.)
-        if(is_white_space($position + 1) || $page[$position + 1] == '<')
-        {
-            $rc = TRUE;
-        }
-    }
-    else if($page[$position] == ',' || $page[$position] == '?' ||
-       $page[$position] == '(' || $page[$position] == ')' ||
-       $page[$position] == '[' || $page[$position] == ']' ||
-       $page[$position] == '"' || $page[$position] == '!' ||
-       $page[$position] == ':' || $page[$position] == '|')
-    {
-        //break the sentence into segments regardless of the next character.
-        $rc = TRUE;
-    }
-    
-    return $rc;
+        global $page;
+        $rc = FALSE;
+
+        if($page[$position] == '.' || $page[$position] == '-')
+        {
+                //Only break if the next character is a white space,
+                //in order to avoid breaks on cases like this: (hello world.)
+                if(is_white_space($position + 1) || $page[$position + 1] == '<')
+                {
+                        $rc = TRUE;
+                }
+        }
+        else if($page[$position] == ',' || $page[$position] == '?' ||
+        $page[$position] == '(' || $page[$position] == ')' ||
+        $page[$position] == '[' || $page[$position] == ']' ||
+        $page[$position] == '"' || $page[$position] == '!' ||
+        $page[$position] == ':' || $page[$position] == '|')
+        {
+                //break the sentence into segments regardless of the next character.
+                $rc = TRUE;
+        }
+
+        return $rc;
 }
 
@@ -448 +456 @@
  * entity. E.g &amp;
  * Return 0 if not an html entity otherwise return the position past this
- *          entity. 
+ *          entity.
  *
  */
 function is_html_entity($position)
 {
-    global $page;
-    if($page[$position] == "&" )
-    {
-        $end_pos = $position + 1;
-        
-        while($page[$end_pos] == "#" ||
-              is_digit($end_pos) || is_a_to_z_character($end_pos))
-        {
-            $end_pos++;
-        }
-
-        if($page[$end_pos] == ';')
-        {
-            $entity = substr($page, $position, $end_pos - $position + 1);
-
-            //Don't break on ` so for our use we don't consider it an entity
-            //e.g. Jack`s apple
-            if($entity ==  "&#8217;" || $entity == "&apos;")
-            {
-                return 0;
-            }
-            
-            //It is an html entity.
-            return $end_pos + 1;
-        }
-    }
-
-    return 0;
+        global $page;
+        if($page[$position] == "&" )
+        {
+                $end_pos = $position + 1;
+
+                while($page[$end_pos] == "#" ||
+                is_digit($end_pos) || is_a_to_z_character($end_pos))
+                {
+                        $end_pos++;
+                }
+
+                if($page[$end_pos] == ';')
+                {
+                        $entity = substr($page, $position, $end_pos - $position + 1);
+
+                        //Don't break on ` so for our use we don't consider it an entity
+                        //e.g. Jack`s apple
+                        if($entity ==  "&#8217;" || $entity == "&apos;")
+                        {
+                                return 0;
+                        }
+
+                        //It is an html entity.
+                        return $end_pos + 1;
+                }
+        }
+
+        return 0;
 }
 
@@ -493 +501 @@
 function is_a_to_z_character($position)
 {
-    global $page;
-    
-    if(($page[$position] >= 'a' && $page[$position] <= 'z') ||
-        ($page[$position] >= 'A' && $page[$position] <= 'Z'))
-    {
-        return TRUE;
-    }
-
-    return FALSE;
+        global $page;
+
+        if(($page[$position] >= 'a' && $page[$position] <= 'z') ||
+        ($page[$position] >= 'A' && $page[$position] <= 'Z'))
+        {
+                return TRUE;
+        }
+
+        return FALSE;
 }
 
@@ -510 +518 @@
 function is_digit($position)
 {
-    global $page;
-    
-    if($page[$position] >= '0' && $page[$position] <= '9')
-    {
-        return TRUE;
-    }
-
-    return FALSE;
-}
-    
-/*
- * Determine if the current position in buffer is a white space. 
+        global $page;
+
+        if($page[$position] >= '0' && $page[$position] <= '9')
+        {
+                return TRUE;
+        }
+
+        return FALSE;
+}
+
+/*
+ * Determine if the current position in buffer is a white space.
  * return TRUE if current position marks a white space otherwise FALSE.
- */ 
+ */
 function is_white_space($position)
 {
-    global $page;
-    
-    if($page[$position] == " "  || $page[$position] ==  ""    ||
-       $page[$position] == "\t" || $page[$position] == "\r"   ||
-       $page[$position] == "\n" || $page[$position] == "\x0B" ||
-       $page[$position] == "\0")
-    {
-        return TRUE;
-    }
+        global $page;
+
+        if($page[$position] == " "  || $page[$position] ==  ""    ||
+        $page[$position] == "\t" || $page[$position] == "\r"   ||
+        $page[$position] == "\n" || $page[$position] == "\x0B" ||
+        $page[$position] == "\0")
+        {
+                return TRUE;
+        }
 }
 
@@ -540 +548 @@
  * Skip within buffer past unreadable characters , i.e. white space
  * and characters considred to be a sentence breaker. Staring from the specified
- * position going either forward or backward. 
- * param forward - indicate direction going either backward of forward. 
+ * position going either forward or backward.
+ * param forward - indicate direction going either backward of forward.
  */
 function skip_unreadable_chars(&$index, $forward=TRUE)
 {
-    global $page, $pos;
-
-    if(!isset($index))
-    {
-        //use $pos as the default position if not specified otherwise
-        $index = &$pos;
-    }
-    $start = $index;
-
-    while($index < strlen($page) && $index > 0 &&
-          (is_white_space($index) || is_sentence_breaker($index)))
-    {
-        ($forward ? $index++ : $index--);
-    }
-    
-    return $index;
+        global $page, $pos;
+
+        if(!isset($index))
+        {
+                //use $pos as the default position if not specified otherwise
+                $index = &$pos;
+        }
+        $start = $index;
+
+        while($index < strlen($page) && $index > 0 &&
+        (is_white_space($index) || is_sentence_breaker($index)))
+        {
+                ($forward ? $index++ : $index--);
+        }
+
+        return $index;
 }
 
 /*
  * Skip within buffer past white space characters , Staring from the specified
- * position going either forward or backward. 
- * param forward - indicate direction going either backward of forward. 
+ * position going either forward or backward.
+ * param forward - indicate direction going either backward of forward.
  */
 function skip_white_space(&$index, $forward=TRUE)
 {
-    global $page, $pos;
-
-    if(!isset($index))
-    {
-        //use $pos as the default position if not specified otherwise
-        $index = &$pos;
-    }
-
-    while($index < strlen($page) && $index > 0 && is_white_space($index))
-    {
-        ($forward ? $index++ : $index--);
-    }
-
-    return $index;
+        global $page, $pos;
+
+        if(!isset($index))
+        {
+                //use $pos as the default position if not specified otherwise
+                $index = &$pos;
+        }
+
+        while($index < strlen($page) && $index > 0 && is_white_space($index))
+        {
+                ($forward ? $index++ : $index--);
+        }
+
+        return $index;
 }
 
 /**
  * Translate the text between the given start position and the current
- * position (pos) within the buffer. 
+ * position (pos) within the buffer.
  */
 function translate_text($start)
 {
-    logger("Enter " . __METHOD__  . " : $start", 4);
-    global $page, $pos, $is_edit_mode;
-
-    //trim white space from the start position going forward
-    skip_white_space($start);
-
-    //Set the end position of the string to one back from current position
-    //(i.e. current position points to '<' or a breaker '.') and then trim
-    //white space from the right backwards
-    $end = $pos - 1;
-    $end = skip_white_space($end, $forward=FALSE);
-    
-    if($start >= $end)
-    {
-        //empty string - nothing to do
-        return;
-    }
-
-    $original_text = substr($page, $start, $end - $start + 1);
-
-    //Cleanup and prepare text
-    $original_text = scrub_text($original_text);
-    if($original_text == NULL)
-    {
-        //nothing left from the text
-        return;
-    }
-
-    $translated_text = fetch_translation($original_text);
-
-    insert_translation($original_text, $translated_text, $start, $end);
+        logger("Enter " . __METHOD__  . " : $start", 4);
+        global $page, $pos, $is_edit_mode;
+
+        //trim white space from the start position going forward
+        skip_white_space($start);
+
+        //Set the end position of the string to one back from current position
+        //(i.e. current position points to '<' or a breaker '.') and then trim
+        //white space from the right backwards
+        $end = $pos - 1;
+        $end = skip_white_space($end, $forward=FALSE);
+
+        if($start >= $end)
+        {
+                //empty string - nothing to do
+                return;
+        }
+
+        $original_text = substr($page, $start, $end - $start + 1);
+
+        //Cleanup and prepare text
+        $original_text = scrub_text($original_text);
+        if($original_text == NULL)
+        {
+                //nothing left from the text
+                return;
+        }
+
+        $translated_text = fetch_translation($original_text);
+
+        insert_translation($original_text, $translated_text, $start, $end);
 }
 
@@ -634 +642 @@
 function insert_translation(&$original_text, &$translated_text, $start, $end)
 {
-    global $segment_id, $is_edit_mode, $tags_list;
-    
-    $is_translated = FALSE;
-
-    if(!$is_edit_mode || !in_array('body', $tags_list))
-    {
-        if($translated_text != NULL)
-        {
-            update_translated_page($start, $end, $translated_text);
-        }
-    }
-    else
-    {
-        $span = "<span id=\"" . SPAN_PREFIX . "$segment_id\">";
-         
-        if($translated_text == NULL)
-        {
-            $span .= $original_text . '</span>';
-        }
-        else
-        {
-            $span .= $translated_text . "</span>";
-            $is_translated = TRUE;
-        }
-        
-        //Insert text (either original or translated) marked by a <span>
-        update_translated_page($start, $end, $span);
-
-                
-        //Insert image to allow editing this segment
-        $img = get_img_tag($original_text, $translated_text, $segment_id, $is_translated);
-        update_translated_page($end + 1, - 1, $img);
-
-        //Increment only after both text and image are generated so they
-        //will be the same for each translated segement
-        $segment_id++;
-        
-    }
-
-    logger("Exit " . __METHOD__  . " : $original_text" , 4);
+        global $segment_id, $is_edit_mode, $tags_list;
+
+        $is_translated = FALSE;
+
+        if(!$is_edit_mode || !in_array('body', $tags_list))
+        {
+                if($translated_text != NULL)
+                {
+                        update_translated_page($start, $end, $translated_text);
+                }
+        }
+        else
+        {
+                $span = "<span id=\"" . SPAN_PREFIX . "$segment_id\">";
+
+                if($translated_text == NULL)
+                {
+                        $span .= $original_text . '</span>';
+                }
+                else
+                {
+                        $span .= $translated_text . "</span>";
+                        $is_translated = TRUE;
+                }
+
+                //Insert text (either original or translated) marked by a <span>
+                update_translated_page($start, $end, $span);
+
+
+                //Insert image to allow editing this segment
+                $img = get_img_tag($original_text, $translated_text, $segment_id, $is_translated);
+                update_translated_page($end + 1, - 1, $img);
+
+                //Increment only after both text and image are generated so they
+                //will be the same for each translated segement
+                $segment_id++;
+
+        }
+
+        logger("Exit " . __METHOD__  . " : $original_text" , 4);
 }
 
@@ -684 +692 @@
 function scrub_text(&$text)
 {
-    //skip strings like without any readable characters (i.e. ".")
-    //Todo: need a broader defintion for non-ascii characters as well
-    if(preg_match("/^[.?!|\(\)\[\],0-9]+$/", $text))
-    {
-        return NULL;
-    }
-
-    //replace multi space chars with a single space
-    $text = preg_replace("/\s\s+/", " ", $text);
-
-    return $text;
+        //skip strings like without any readable characters (i.e. ".")
+        //Todo: need a broader defintion for non-ascii characters as well
+        if(preg_match("/^[.?!|\(\)\[\],0-9]+$/", $text))
+        {
+                return NULL;
+        }
+
+        //replace multi space chars with a single space
+        $text = preg_replace("/\s\s+/", " ", $text);
+
+        return $text;
 }
 
@@ -704 +712 @@
  * param start - marks the starting position of the replaced string in the original page.
  * param end - marks the end position of the replaced string in the original page.
-               Use -1 to do insert instead of replace.
+ Use -1 to do insert instead of replace.
  * param translated_text - text to be inserted.
  */
 function update_translated_page($start, $end, $translated_text)
 {
-    global $page, $tr_page, $tr_mark;
-    
-    //Bring the translated up to date up to the start position.
-    while($tr_mark < $start)
-    {
-        $tr_page .= $page[$tr_mark++];
-    }
-
-    $tr_page .= $translated_text;
-
-    if($end > $start)
-    {
-        //Move mark to correlate the posistion between the two pages.
-        //Only do this when some content has been replaced, i.e. not
-        //an insert.
-        $tr_mark = $end + 1;
-    }
-    
+        global $page, $tr_page, $tr_mark;
+
+        //Bring the translated up to date up to the start position.
+        while($tr_mark < $start)
+        {
+                $tr_page .= $page[$tr_mark++];
+        }
+
+        $tr_page .= $translated_text;
+
+        if($end > $start)
+        {
+                //Move mark to correlate the posistion between the two pages.
+                //Only do this when some content has been replaced, i.e. not
+                //an insert.
+                $tr_mark = $end + 1;
+        }
+
 }
 

trunk/WordPress/plugin/transposh/transposh.php

@@ -62 +62 @@
 //The url pointing to the base of the plugin
 $plugin_url;
-                             
+
 //Error message displayed for the admin in case of failure
 $admin_msg;
@@ -70 +70 @@
  * Called when the buffer containing the original page is flused. Triggers the
  * translation process.
- * 
- */ 
+ *
+ */
 function process_page(&$buffer) {
-    
+
     global $wp_query, $tr_page, $page, $pos, $lang, $plugin_url, $is_edit_mode, $wpdb,
            $table_name;
-    
+
     $start_time = microtime(TRUE);
-    
+
     if (!isset($wp_query->query_vars[LANG_PARAM]))
     {
         //No language code - avoid further processing.
         return $buffer;
-        
+
     }
 
@@ -94 +94 @@
         return $buffer;
     }
-    
-    
+
+
     $page = $buffer;
 
@@ -105 +105 @@
         //user has the required permissions
         $editable_langs = get_option(EDITABLE_LANGS);
-        
+
         if(is_translator() && strstr($editable_langs, $lang))
         {
             $is_edit_mode = TRUE;
         }
-        
-    }
-    
+
+    }
+
     logger("translating " . $_SERVER['REQUEST_URI'] . " to: $lang", 1);
-    
+
     //translate the entire page
     process_html();
-    
+
     $end_time = microtime(TRUE);
 
     logger("Translation completed in " . ($end_time - $start_time) . " seconds", 1);
-    
+
     //return the translated page unless it is empty, othewise return the original
     return (strlen($tr_page) > 0 ? $tr_page : $page);
@@ -128 +128 @@
 /*
  * Fix links on the page. href needs to be modified to include
- * lang specifier and editing mode. 
+ * lang specifier and editing mode.
  *
  */
@@ -134 +134 @@
 {
     global $home_url, $home_url_quoted, $lang, $is_edit_mode, $wp_rewrite;
-    
+
     $href = get_attribute($start, $end, 'href');
 
@@ -149 +149 @@
 
     $use_params = FALSE;
-    
-    //Only use params if permalinks are not enabled. 
+
+    //Only use params if permalinks are not enabled.
     //don't fix links pointing to real files as it will cause that the
     //web server will not be able to locate them
-    if(!$wp_rewrite->using_permalinks() ||   
+    if(!$wp_rewrite->using_permalinks() ||
        stripos($href, '/wp-admin') !== FALSE   ||
        stripos($href, '/wp-content') !== FALSE ||
@@ -171 +171 @@
 
 /*
- * Update the given url to include language params. 
+ * Update the given url to include language params.
  * param url - the original url to rewrite
  * param lang - language code
@@ -186 +186 @@
         $use_params_only = TRUE;
     }
-    
+
     if($is_edit)
     {
         $params = EDIT_PARAM . '=1&';
-        
+
     }
 
@@ -210 +210 @@
         //Cleanup extra &
         $url = preg_replace("/&&+/", "&", $url);
-            
+
             //Cleanup extra ?
         $url = preg_replace("/\?\?+/", "?", $url);
@@ -219 +219 @@
 
 /*
- * Fetch translation from db or cache. 
+ * Fetch translation from db or cache.
  * Returns the translated string or NULL if not available.
  */
@@ -226 +226 @@
     global $wpdb, $lang, $table_name;
     $translated = NULL;
-    
-    logger("Enter " . __METHOD__ . " $original", 4);
+
+    logger("Enter " . __METHOD__ . ": $original", 4);
     if(ENABLE_APC && function_exists('apc_fetch'))
     {
@@ -233 +233 @@
         if($rc === TRUE)
         {
-            return $cached;
-        }
-    }
-    
-    $query = "SELECT * FROM $table_name WHERE original = '$original' and lang = '$lang' "; 
+                logger("Exit from cache " . __METHOD__ . ": $cached", 4);
+                return $cached;
+        }
+    }
+
+    $query = "SELECT * FROM $table_name WHERE original = '$original' and lang = '$lang' ";
     $row = $wpdb->get_row($query);
-    
+
     if($row !== FALSE)
     {
         $translated = $row->translated;
         $translated = stripslashes($translated);
-        
+
         logger("db result for $original >>> $translated ($lang)" , 3);
     }
-    
-    
+
+
     if(ENABLE_APC && function_exists('apc_store'))
     {
@@ -257 +258 @@
             $cache_entry = "";
         }
-        
+
         //update cache
         $rc = apc_store($original . $lang, $cache_entry, 3600);
@@ -265 +266 @@
         }
     }
-    
-    logger("Exit " . __METHOD__ . " $translated", 4);
+
+    logger("Exit " . __METHOD__ . ": $translated", 4);
     return $translated;
 }
@@ -278 +279 @@
 {
     global $plugin_url;
-    
+
     $overlib_dir = "$plugin_url/js/overlibmws";
-    
+
     $js = "\n<script type=\"text/javascript\" src=\"$overlib_dir/overlibmws.js\"></script>";
     $js .= "\n<script type=\"text/javascript\" src=\"$overlib_dir/overlibmws_filter.js\"></script>";
@@ -297 +298 @@
 /*
  * Return the img tag that will added to enable editing a translatable
- * item on the page. 
+ * item on the page.
  * param segement_id The id (number) identifying this segment. Needs to be
          placed within the img tag for use on client side operation (jquery)
@@ -305 +306 @@
     global $plugin_url, $lang, $home_url;
     $url = $home_url . '/index.php';
-    
-    //For use in javascript, make the following changes: 
+
+    //For use in javascript, make the following changes:
     //1. Add slashes to escape the inner text
     //2. Convert the html special characters
@@ -318 +319 @@
     }
 
-    $img = "<img src=\"$plugin_url/translate$add_img.png\" alt=\"translate\" id=\"" . IMG_PREFIX . "$segment_id\" 
-           onclick=\"translate_dialog('$original','$translation','$lang','$url', '$segment_id'); return false;\" 
-           onmouseover=\"hint('$original'); return true;\" 
+    $img = "<img src=\"$plugin_url/translate$add_img.png\" alt=\"translate\" id=\"" . IMG_PREFIX . "$segment_id\"
+           onclick=\"translate_dialog('$original','$translation','$lang','$url', '$segment_id'); return false;\"
+           onmouseover=\"hint('$original'); return true;\"
            onmouseout=\"nd()\" />";
-    
+
     return $img;
 }
@@ -346 +347 @@
         return;
     }
-  
-    //include the transposh.css 
+
+    //include the transposh.css
         echo "<link rel=\"stylesheet\" href=\"$plugin_url/transposh.css\" type=\"text/css\" />";
 
@@ -359 +360 @@
 {
     global $home_url, $home_url_quoted, $plugin_url, $table_name, $wpdb;
-    
+
     $home_url = get_option('home');
     $local_dir = preg_replace("/.*\//", "", dirname(__FILE__));
-    
+
     $plugin_url= $home_url . "/wp-content/plugins/$local_dir";
     $home_url_quoted = preg_quote($home_url);
@@ -372 +373 @@
 
 /*
- * A new translation has been posted, update the translation database. 
+ * A new translation has been posted, update the translation database.
  *
  */
@@ -378 +379 @@
 {
     global $wpdb, $table_name;
-    
+
     $ref=getenv('HTTP_REFERER');
     $original = $_POST['original'];
     $translation = $_POST['translation'];
     $lang = $_POST['lang'];
-    
+
     if(!isset($original) || !isset($translation) || !isset($lang))
     {
@@ -397 +398 @@
     }
 
-    //Decode & remove already escaped character to avoid double escaping 
+    //Decode & remove already escaped character to avoid double escaping
     $original    = $wpdb->escape(stripslashes(urldecode($original)));
     $translation = $wpdb->escape(htmlspecialchars(stripslashes(urldecode($translation))));
-        
+
     $update = "REPLACE INTO  $table_name (original, translated, lang)
                 VALUES ('" . $original . "','" . $translation . "','" . $lang . "')";
@@ -455 +456 @@
         logger("Error !!! failed to update transaction log:  $loguser, $original ,$translation, $lang" , 0);
     }
-    
+
 }
 
@@ -467 +468 @@
 {
     global $languages;
-    
+
     $default = get_option(DEFAULT_LANG);
     if(!$languages[$default])
@@ -500 +501 @@
 
 /*
- * Page generation completed - flush buffer. 
+ * Page generation completed - flush buffer.
  */
 function on_shutdown()
@@ -520 +521 @@
         return $rule;
     }
-    
+
     $newRules = array();
     $lang_prefix="([a-z]{2,2}(\-[a-z]{2,2})?)/";
@@ -526 +527 @@
     $lang_parameter= "&" . LANG_PARAM . '=$matches[1]';
 
-    //catch the root url 
+    //catch the root url
     $newRules[$lang_prefix."?$"] = "index.php?lang=\$matches[1]";
     logger("\t" . $lang_prefix."?$" . "  --->  " . "index.php?lang=\$matches[1]");
@@ -533 +534 @@
         $original_key = $key;
         $original_value = $value;
-        
+
         $key = $lang_prefix . $key;
 
@@ -544 +545 @@
 
         $value .= $lang_parameter;
-        
+
         logger("\t" . $key . "  --->  " . $value);
-                
+
 
         $newRules[$key] = $value;
@@ -571 +572 @@
 
 /*
- * Setup the translation database. 
- *
- */
-function setup_db() 
+ * Setup the translation database.
+ *
+ */
+function setup_db()
 {
 
@@ -584 +585 @@
     if($wpdb->get_var("SHOW TABLES LIKE '$table_name'") != $table_name)
     {
-        logger("Attempting to create table $table_name", 0); 
+        logger("Attempting to create table $table_name", 0);
         $sql = "CREATE TABLE " . $table_name . " (original VARCHAR(256) NOT NULL,
                                                   lang CHAR(5) NOT NULL,
                                                   translated VARCHAR(256),
                                                   PRIMARY KEY (original, lang)) ";
-        
-                                     
+
+
         dbDelta($sql);
 
@@ -598 +599 @@
 
         $result = $wpdb->query($insert);
-        
+
         if($result === FALSE)
         {
-            logger("Error failed to create $table_name !!!", 0); 
+            logger("Error failed to create $table_name !!!", 0);
         }
         else
         {
-            logger("Table $table_name was created successfuly", 0); 
+            logger("Table $table_name was created successfuly", 0);
             add_option(TRANSPOSH_DB_VERSION, DB_VERSION);
         }
@@ -614 +615 @@
     if($wpdb->get_var("SHOW TABLES LIKE '$table_name'") != $table_name)
     {
-        logger("Attempting to create table $table_name", 0); 
+        logger("Attempting to create table $table_name", 0);
         $sql = "CREATE TABLE " . $table_name . " (original VARCHAR(256) NOT NULL,
                                                   lang CHAR(5) NOT NULL,
@@ -621 +622 @@
                                                   timestamp TIMESTAMP,
                                                   PRIMARY KEY (original, lang, timestamp)) ";
-        
-                                     
+
+
         dbDelta($sql);
     }
-    
+
     logger("Exit " . __METHOD__  );
 }
@@ -645 +646 @@
     if(get_option(ANONYMOUS_TRANSLATION))
     {
-        //if anonymous translation is allowed - let anyone enjoy it 
+        //if anonymous translation is allowed - let anyone enjoy it
         return TRUE;
     }
-    
+
     return FALSE;
 }
-         
-
-/*
- * Plugin activated. 
+
+
+/*
+ * Plugin activated.
  *
  */
@@ -666 +667 @@
     add_filter('rewrite_rules_array', 'update_rewrite_rules');
     $wp_rewrite->flush_rules();
-        
+
     logger("plugin_activate exit: " . dirname(__FILE__));
 }
@@ -681 +682 @@
     remove_filter('rewrite_rules_array', 'update_rewrite_rules');
     $wp_rewrite->flush_rules();
-    
+
     logger("plugin_deactivate exit: " . dirname(__FILE__));
 }
 
 /*
- * Callback from admin_notices - display error message to the admin. 
+ * Callback from admin_notices - display error message to the admin.
  *
  */
@@ -693 +694 @@
     global $admin_msg;
     logger("Enter " . __METHOD__, 0);
-    
+
     echo '<div class="updated"><p>';
     echo 'Error has occured in the installation process of the translation plugin: <br>';
-    
+
     echo $admin_msg;
-    
+
     if (function_exists('deactivate_plugins') ) {
         deactivate_plugins("transposh/translate.php", "translate.php");
@@ -711 +712 @@
  * Callback when all plugins have been loaded. Serves as the location
  * to check that the plugin loaded successfully else trigger notification
- * to the admin and deactivate plugin. 
+ * to the admin and deactivate plugin.
  *
  */
@@ -722 +723 @@
     {
         $admin_msg = "Failed to locate the translation table  <em> " . TRANSLATIONS_TABLE . "</em> in local database. <br>";
-        
+
         logger("Messsage to admin: $admin_msg", 0);
         //Some error occured - notify admin and deactivate plugin
@@ -733 +734 @@
     {
         $admin_msg = "Translation database version ($db_version) is not comptabile with this plugin (". DB_VERSION . ")  <br>";
-        
+
         logger("Messsage to admin: $admin_msg", 0);
         //Some error occured - notify admin and deactivate plugin
@@ -740 +741 @@
 }
 
-//Register callbacks 
+//Register callbacks
 add_action('wp_head', 'add_custom_css');
 add_filter('query_vars', 'parameter_queryvars' );