<?
/*
NOTE: In PHP4, you do not need to first decode cdata from utf8. PHP5 assumes the source is in utf8, so you must first
decode the cdata.
*/

include_once("schema_loader.php5");

class db_modifier {
    
    public $parser;
    public $out;

    public $cur_data;
    public $count;
    public $table;

    public $symbols;
    public $substitutes;
    public $first_chars;
    public $pattern;
    public $alpha_symbols;
    public $alpha_substitutes;
    public $alpha_first_chars;
    public $alpha_pattern;

    public $sort_fields;
    public $has_sort_fields;
    public $alpha;
    public $cur_alpha_index;
    public $head_tag;

    public $produce_stripped_version;

    public $strippable_fields;
    public $fields;
    public $fields_to_strip;

    public $in_entry;
    public $group_names;

    function __construct() 
    {
	$this->fields_to_strip = array();
	$this->check_all_fields = false;
	$this->count = 0;
	$this->cur_data = "";
	$this->symbols = array();
	$this->substitutes = array();
	$this->pattern = "";

	$this->alpha_symbols = array();
	$this->alpha_substitutes = array();
	$this->alpha_first_chars = array();
	$this->alpha_pattern = "";

	$this->sort_fields = array();
	$this->has_sort_fields = false;
	$this->alpha = "";
	$this->cur_alpha_index = 10;

	$this->strippable_fields = array();
	$this->fields = array();
	$this->fields_to_strip = array();

	$this->in_entry = false;
    }

    function get_sort_fields($sort_fields) {
	$fields = explode(" ",$sort_fields);
	foreach($fields as $field) {
	    $this->sort_fields[] = $field;
	}
    }

    function get_fields_to_strip($fields) {
	$in = fopen($fields,"r");
	while($line = fgets($in)) {
	    $field = trim($line);
	    $fields_to_strip[] = $field;
	}
	return $fields_to_strip;
    }

    function make_alpha_symbols_table($textarea) {
	$textarea = stripslashes($textarea);
	$textarea = trim(utf8_decode($textarea));
	$lines = explode("\n",$textarea);	
	foreach($lines as $line) {
	    $vals = explode(" = ",$line);
	    $substitute = trim($vals[1]);
	    if($substitute == "''") {
		$substitute = "";
	    }
	    $symbols = preg_split('//', $vals[0], -1, PREG_SPLIT_NO_EMPTY);
	    foreach($symbols as $symbol) {
		$this->alpha_symbols[] = $symbol;
		$this->alpha_substitutes[] = $substitute;
	    }
	}
	$symbols_string = implode("",$this->alpha_symbols);
	if(substr($symbols_string,-1,1) == "'") {
	    $symbols_string = "'".substr($symbols_string,0,-1);
	}
	if($symbols_string != "") {
	    $this->alpha_pattern = "[".$symbols_string."]";
	}
    }

    function make_alpha_first_chars_table($textarea) {
	$textarea = trim(stripslashes($textarea));
	$textarea = utf8_decode($textarea);
	$lines = explode("\n",$textarea);
	foreach($lines as $line) {
	    $vals = explode(" = ",$line);
	    $substitute = trim($vals[1]);
	    if($substitute == "''") {
		$substitute = "";
	    }
	    $symbols = preg_split('//', $vals[0], -1, PREG_SPLIT_NO_EMPTY);
	    foreach($symbols as $symbol) {
		$this->alpha_first_chars[$symbol] = $substitute;
	    }
	}
	$first_chars_string = trim(implode("",array_keys($this->alpha_first_chars)));
	if(substr($first_chars_string,-1,1) == "'") {
	    $first_chars_string = "'".substr($first_chars_string,0,-1);
	}
	if($first_chars_string != "") {
	    if($this->alpha_pattern != "") {
		$this->alpha_pattern .= "|";
	    }
	    $this->alpha_pattern .= "^[".$first_chars_string."]";
	}
    }
    


    function make_symbols_table($textarea) {
	$textarea = stripslashes($textarea);
	$textarea = trim(utf8_decode($textarea));
	$lines = explode("\n",$textarea);	
	foreach($lines as $line) {
	    $vals = explode(" = ",$line);
	    $substitute = trim($vals[1]);
	    if($substitute == "''") {
		$substitute = "";
	    }
	    $symbols = preg_split('//', $vals[0], -1, PREG_SPLIT_NO_EMPTY);
	    foreach($symbols as $symbol) {
		$this->symbols[] = $symbol;
		$this->substitutes[] = $substitute;
	    }
	}
	$symbols_string = implode("",$this->symbols);
	if(substr($symbols_string,-1,1) == "'") {
	    $symbols_string = "'".substr($symbols_string,0,-1);
	}
	if($symbols_string != "") {
	    $this->pattern = "[".$symbols_string."]";
	}
    }
    
    function make_first_chars_table($textarea) {
	$textarea = trim(stripslashes($textarea));
	$textarea = utf8_decode($textarea);
	$lines = explode("\n",$textarea);
	foreach($lines as $line) {
	    $vals = explode(" = ",$line);
	    $substitute = trim($vals[1]);
	    if($substitute == "''") {
		$substitute = "";
	    }
	    $symbols = preg_split('//', $vals[0], -1, PREG_SPLIT_NO_EMPTY);
	    foreach($symbols as $symbol) {
		$this->first_chars[$symbol] = $substitute;
	    }
	}
	$first_chars_string = trim(implode("",array_keys($this->first_chars)));
	if(substr($first_chars_string,-1,1) == "'") {
	    $first_chars_string = "'".substr($first_chars_string,0,-1);
	}
	if($first_chars_string != "") {
	    if($this->pattern != "") {
		$this->pattern .= "|";
	    }
	    $this->pattern .= "^[".$first_chars_string."]";
	}
    }
    

    function alpha_strip($value) {
	$stripped = str_replace($this->alpha_symbols,$this->alpha_substitutes,$value);
	$first_char = substr($stripped,0,1);
	while(@array_key_exists($first_char,$this->alpha_first_chars)) {
	    $stripped = $this->alpha_first_chars[$first_char].substr($stripped,1);
	    $first_char = substr($stripped,0,1);
	}
	return $stripped;
    }

    function strip($value) {
	$stripped = str_replace($this->symbols,$this->substitutes,$value);
	$first_char = substr($stripped,0,1);
	while(@array_key_exists($first_char,$this->first_chars)) {
	    $stripped = $this->first_chars[$first_char].substr($stripped,1);
	    $first_char = substr($stripped,0,1);
	}
	return $stripped;
    }


    function startHandler($xp, $element, $attribs) {	
	$element = strtolower($element);
	if($element == $this->head_tag) {
	    $this->in_entry = true;	    
	}
	if($this->produce_stripped_version) {
	    if($this->in_entry) {
		if(in_array($element,$this->fields)) {
		    fwrite($this->out,"<$element>");
		}
	    }
	    else {
		fwrite($this->out,"<$element>\n");
	    }
	}
    }


    function endHandler($xp, $element) {	
	$element = strtolower($element);
	if($this->produce_stripped_version) {
	    if($this->has_sort_fields && in_array($element, $this->sort_fields)) {
		$index = array_search($element,$this->sort_fields);
		if($index < $this->cur_alpha_index) {
		    $this->alpha = $this->alpha_strip($this->cur_data);
		    $this->cur_alpha_index = $index;
		}
	    }
	    if(($element == $this->head_tag) && $this->has_sort_fields) {
		fwrite($this->out,"<alpha>$this->alpha</alpha>\n");
		$this->alpha = "";
		$this->cur_alpha_index = count($this->sort_fields);	    
	    }
	    if($this->in_entry) {
		if(in_array($element,$this->fields)) {
		    fwrite($this->out,"$this->cur_data</$element>\n");
		}
	    }
	    else {
		fwrite($this->out,"$this->cur_data</$element>\n");
	    }
	}
	else {
	    if($this->in_entry) {
		if(!in_array($element,$this->fields)) {
		    $this->fields[] = $element;
		}
	    }
	}

	if($this->pattern != "") {
	    if(ereg($this->pattern,$this->cur_data)) {
		if($this->produce_stripped_version) {
		    if(in_array($element,$this->fields_to_strip)) {
			$stripped_value = $this->strip($this->cur_data);
			fwrite($this->out,"<".$element."_s>$stripped_value</".$element."_s>\n");
		    }
		}
		else {
		    if(!in_array($element,$this->strippable_fields)) {
			$this->strippable_fields[] = $element;
		    }
		}
	    }
	}
	    
	if($element == $this->head_tag) {
	    $this->in_entry = false;
	}	    

	$this->cur_data = "";

    }


    function cDataHandler($xp, $data) {
	$data = trim($data);
	if($data != "")
	{
	    $data = ereg_replace("&","&amp;",$data);
	    $data = ereg_replace("<","&lt;",$data);
	    $data = ereg_replace(">","&gt;",$data);
	    $this->cur_data .= $data;
	}
    }

    function unhtmlentities ($string)  {
	$trans_tbl = get_html_translation_table (HTML_ENTITIES);
	$trans_tbl = array_flip ($trans_tbl);
	$ret = strtr ($string, $trans_tbl);
	return  preg_replace('/\&\#([0-9]+)\;/me',
			     "chr('\\1')",$ret);
    }


    function get_fields($xml,$head_tag,$symbols,$first_chars) {
	$this->parser = xml_parser_create();
	xml_set_object($this->parser,$this);
	xml_set_element_handler($this->parser,"startHandler","endHandler");
	xml_set_character_data_handler($this->parser,"cDataHandler");

	$this->head_tag = $head_tag;

	if(trim($symbols) != "") {
	    $this->make_symbols_table($symbols);
	}
	if(trim($first_chars) != "") {
	    $this->make_first_chars_table($first_chars);
	}
	$this->produce_stripped_version = false;

	$in = fopen($xml,"r");
	while($line = fgets($in)) {
	    xml_parse($this->parser,$line,false);
	}

	fclose($in);
	sort($this->strippable_fields);	
	sort($this->fields);
    }


    function modify($xml,$new_xml_name, $head_tag, $sort_fields,
		    $alpha_symbols, $alpha_first_chars,
		    $symbols, $first_chars, $fields, 
		    $fields_to_strip) {
	
	$this->produce_stripped_version = true;

	$this->parser = xml_parser_create();
	xml_set_object($this->parser,$this);
	xml_set_element_handler($this->parser,"startHandler","endHandler");
	xml_set_character_data_handler($this->parser,"cDataHandler");

	$this->fields = $fields;
	$this->fields_to_strip = $fields_to_strip;

	$this->head_tag = $head_tag;
	if($sort_fields != "") {
	    $this->get_sort_fields($sort_fields);
	    $this->has_sort_fields = true;
	    $this->fields[] = "alpha";
	}
	else {
	    $this->has_sort_fields = false;
	}
	
	if(trim($symbols) != "") {
	    $this->make_symbols_table($symbols);
	}
	if(trim($first_chars) != "") {
	    $this->make_first_chars_table($first_chars);
	}

	if(trim($alpha_symbols) != "") {
	    $this->make_alpha_symbols_table($alpha_symbols);
	}
	if(trim($alpha_first_chars) != "") {
	    $this->make_alpha_first_chars_table($alpha_first_chars);
	}
	$this->out = fopen($new_xml_name,"w");
	fwrite($this->out,"<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>\n");
	$in = fopen($xml,"r");
	while($line = fgets($in)) {
	    xml_parse($this->parser,$line,false);
	}
	fclose($in);
	fclose($this->out);

	xml_parser_free($this->parser);
    }
}

/*
//NOTE: you must remove utf8_encode from the make_symbol_table functions when testing from the command line
set_time_limit(0);
$converter = new db_modifier();
$sort_fields = "lxam lxoa";
$symbols = "á = a\né = e\ní = i\nó = o\nú = u\nÁ = A\nÉ = E\nÍ = I\nÓ = O\nÚ = U\nÑ = N\nñ = n\nü = u";
$first_chars = "-' = ''";
$alpha_symbols = "á = a";
$alpha_first_chars = "-' = ''";
$fields_not_to_strip = array();
$converter->get_fields("test.xml","refgroup",$symbols,$first_chars);
print_r($converter->fields);



$fields_to_strip = array_diff($converter->strippable_fields,$fields_not_to_strip);
$converter->modify("test.xml","with_stripped.xml","refgroup",
		   $sort_fields,$alpha_symbols,$alpha_first_chars,$symbols,$first_chars,
		   $converter->fields,$fields_to_strip);
*/

?>