Viewing file:  Translate.php (22.64 KB) -rw-rw-rw-
Select action/file-type:
 (+) |  (+) |  (+) | Code (+) | Session (+) |  (+) | SDB (+) |  (+) |  (+) |  (+) |  (+) |  (+) |
<?php
//
// +----------------------------------------------------------------------+
// | PHP Version 4 |
// +----------------------------------------------------------------------+
// | Copyright (c) 1997-2003 The PHP Group |
// +----------------------------------------------------------------------+
// | This source file is subject to version 2.02 of the PHP license, |
// | that is bundled with this package in the file LICENSE, and is |
// | available at through the world-wide-web at |
// | http://www.php.net/license/2_02.txt. |
// | If you did not receive a copy of the PHP license and are unable to |
// | obtain it through the world-wide-web, please send a note to |
// | license@php.net so we can mail you a copy immediately. |
// +----------------------------------------------------------------------+
// | Authors: Wolfram Kriesing <wolfram@kriesing.de> |
// +----------------------------------------------------------------------+
//
// $Id: Translate.php 113824 2003-01-28 19:20:04Z cain $
// we have to move this to some more common place in PEAR
// this is just a quick hack here :-)
require_once( 'Tree/OptionsDB.php' ); // this contains all the methods like setOption, getOption, etc.
/**
* Description
*
* @package Language
* @access public
* @author Wolfram Kriesing <wolfram@kriesing.de>
* @version 2001/12/29
*/
class I18N_Messages_Translate extends Tree_OptionsDB
{
var $options = array( 'tablePrefix' => 'translate_' // the DB-table name prefix, at the end we add the lang-string passed to the method
,'sourceLanguage'=> 'en' // the source language, the language used to retrieve the strings to translate from
// its also the table which is used to retreive the source string
// be case senstivie by default since not all languages write nouns and verbs etc.
// in lower case translating from german might fail if the case is not considered
,'caseSensitive'=> true
,'translatorUrl' => '' // the url to a translator tool, only used if given
);
/**
* Those are the delimiters surrounding translatable text.
* This way we prevent from translating each HTML-tag, which definetly wouldnt work :-)
* those delimiters might also work on any other markup language, like xml - but not tested
*
* NOTE: if you have php inside such a tag then you have to use an extra filter
* since <a href="">< ?=$var? ></a> would find the php-tags and see them as delimiters
* which results in < ?=$var? > can not be translated, see sf.net/projects/simpletpl, there
* is a filter in 'SimpleTemplate/Filter/Basic::applyTranslateFunction' which solves this
* it wraps a given function/method around it so that it finally will be:
* <a href="">< ?=translateThis($var)? ></a>
*
* @var array $possibleMarkUpDelimiters
*/
var $possibleMarkUpDelimiters = array(
// cant use this
// '>[^<]*' ,'[^>]*<', // this mostly applies, that a text is inbetween '>' and '<'
// because it would translate 'class="..."' too, if we have the word 'as' to be translated :-(
// but this also means we have to handle stuff like of others specials chars, that dont start
// and end with a < or > somehow ... i dont know how yet :-(
// this mostly applies, that a text is inbetween '>' and '<'
array('>\s*', '\s*<')
// same as above,
// only that this leaves before and after the text outside html-tags as they are ...
//
// actually we need a more common thing here, which also takes care of other chars which
// we dont want to bother the translation with
// ,array('>[\s* ]*', '[\s* ]*<')
//the above is not really secure, since [] means any char inside
// so '>nbsp translate this<' would be translated too
// this is for input button's values
//FIXXXME this fails if the order of type and value attrbs is switched :-(
,array('<input type=["\']?(submit|button)["\']? [^>]*value=["\']?\s*' , '\s*["\']?[^>]*>' , 1)
// ,array('<\s*input .*type=submit .*value=["\']?\s*','\s*["\']?.*>')
// array('<\s*input.*type=[\'"]?(button|submit)[\'"]?.*value=["\']?\s*','\s*["\']?.*>',1)
// ,array('<\s*input.*value=["\']?\s*','\s*["\']?.*>')
// all the reg-exps here are limited, the ones below would work perfect on:
// <a title="translate this" ...> or <img alt="translate this" ....>
// but with the following they would have problems:
// <a href="< ?php echo §foor ? >" title="translate this">
// since there is a < before the title-attribute, and if we build the regexp
// with .* instead of [^>]* before the attribute it might return ambigious results
// SOLUTION for now: put the attrbutes you want translated first inside the tag, like in the example above
,array('<img [^>]*alt=["\']?\s*' , '\s*["\']?.*>')
,array('<a [^>]*title=["\']?\s*' , '\s*["\']?.*>')
);
/**
* @var array this contains the content from the DB, to prevent from multiple DB accesses
*/
var $_translated = array('destLanguage'=>'','strings'=>array());
/**
* @var array this array contains the translated strings but with the difference to $_translated
* that the source strings is the index, so a lookup if a translation exists is much faster
*/
var $_sourceStringIndexed = array();
/**
*
*
* @access public
* @author
* @version
*/
function __construct( $dsn , $options )
{
parent::Tree_OptionsDB( $dsn , $options );
// FIXXME pass a resource to the constructor which can be used to translate the
// string, it should be possible to use XML, DB, or whatever
// currently, as you can see there is only a DB interface hardcoded in here
// this will be removed soon
}
/**
* for pre-ZE2 compatibility
*
* @access public
* @author
* @version
*/
function I18N_Messages_Translate( $dsn , $options=array() )
{
return $this->__construct( $dsn , $options );
}
/**
* tries to translate a given string, but only exactly the string as it is in the DB
*
* @access public
* @author Wolfram Kriesing <wolfram@kriesing.de>
* @version 01/12/29
* @param string $string the string that shall be translated
* @param string $lang iso-string for the destination language
* @return string the translated string
*/
function simpleTranslate( $string , $lang )
{
if( $lang == $this->getOption('sourceLanguage') ) // we dont need to translate a string from the source language to the source language
return $string;
if( sizeof($this->_translated['strings'])>0 && // this checks if the DB content had been read already
$this->_translated['destLanguage'] == $lang ) // for this language
{
if( sizeof($this->_sourceStringIndexed) == 0 )
{
foreach( $this->_translated['strings'] as $aSet)
$this->_sourceStringIndexed[$aSet['string']] = $aSet['translated'];
}
if( isset($this->_sourceStringIndexed[$string]) )
return $this->_sourceStringIndexed[$string];
return $string;
}
// FIXXME may be it would be better just reading the entire DB-content once
// and using this array then ??? this uses up a lot of RAM and that for every user ... so i guess not OR?
// or use PEAR::Cache
$query = sprintf( "SELECT d.string FROM %s%s s,%s%s d WHERE s.string=%s AND s.id=d.id",
$this->getOption('tablePrefix'),$this->getOption('sourceLanguage'), // build the source language name
$this->getOption('tablePrefix'),$lang,
$this->dbh->quote($string) ); // build the destination language name
$res = $this->dbh->getOne( $query );
if( DB::isError($res) )
{
// return $this->raiseError('...');
return $string; // return the actual string on failure
}
if( !$res ) // if no translation was found return the source string
return $string;
return $res;
}
/**
* tries to translate a given string, it also tries using the regexp's which might be in the DB
*
* @access public
* @author Wolfram Kriesing <wolfram@kriesing.de>
* @version 01/12/29
* @param string $string the string that shall be translated
* @param string $lang iso-string for the destination language
* @return string the translated string
*/
function translate( $string , $lang )
{
//FIXXME extract the reg-exp thingy from the translateMarkup method and call
// it explicitly here, and once we have found a translation stop the process,
// so we dont translate stuff that is already translated again,
// i.e. 'Data saved' is translated into 'Daten gespeichert' and if we continue
// as we do now it will translate 'Date' into 'Datum' which results in 'Datumn gespeichert'
// which is big bullshit
// and the second thing: do only translate full words, then the above wouldnt happen neither
$res = $this->simpleTranslate( $string , $lang );
// if the select didnt translate the string we need to go thru all the strings
// which contain parts of regular expressions, so we can leave out all
// the pure strings, this should save some time
if( $res == $string )
{
$temp = $this->possibleMarkUpDelimiters; // remember the delimiters
$this->possibleMarkUpDelimiters = array(''=>''); // dont use any delimiters
// may be better using a property like 'useMarkupDelimiters'
$res = $this->translateMarkUpString( $string , $lang ); // translate
$this->possibleMarkUpDelimiters = $temp; // set delimiters properly again
}
return $res;
}
/**
* returns the DB content for the source and the destination language given as paramter $lang
*
* @access public
* @author Wolfram Kriesing <wolfram@kriesing.de>
* @version 02/01/08
* @param string $lang iso-string for the destination language
* @return array
*/
function getAll( $lang )
{
if( sizeof($this->_translated['strings'])==0 || // this checks if the DB content had been read already
$this->_translated['destLanguage'] != $lang ) // for this language
{
//print "read again<br>";
$this->_translated['destLanguage'] = $lang;
}
else
{
return $this->_translated['strings'];
}
// for the translation API, we need to have the long sentences at first, since translating a single word
// might screw up the entire content, like translating 'i move to germany' and starting to tranlate the
// word 'move' makes it impossible to properly translate the entire phrase
// even though this problem can not really happen, since we check for delimiters around the string that
// shall be translated see $posDelimiters
$query = sprintf( 'SELECT d.string as translated,d.*,s.* '. // d.string shall be named 'translated'
// but we still need all the rest from the destination language table
// and s.* overwrites d.string but we dont need it we have it in 'translated'
'FROM %s%s s,%s%s d WHERE s.id=d.id '.
'ORDER BY LENGTH(s.string) DESC', // sort the results by the length of the strings, so we translate
// sentences first and single words at last
$this->getOption('tablePrefix'),$this->getOption('sourceLanguage'), // build the source language name
$this->getOption('tablePrefix'),$lang ); // build the destination language name
$res = $this->dbh->getAll( $query );
if( DB::isError($res) )
{
// return $this->raiseError('...');
echo sprintf('ERROR - Translate::getAll<br>QUERY:%s<br>%s<br><br>',$query,$res->message);
return false;
}
$this->_translated['destLanguage'] = $lang;
$this->_translated['strings'] = $res;
return $this->_translated['strings'];
}
/**
* translates all the strings that match any of the source language-string
* the input is mostly an HTML-file, and it is filtered so only real text
* is translated, at least i try it as good as i can :-)
*
* @access public
* @author Wolfram Kriesing <wolfram@kriesing.de>
* @version 02/01/08
* @param string $input the string that shall be translated, mostly an entire HTML-page
* @param string $lang iso-string for the destination language
* @return string iso-string for the language
*
*/
function translateMarkUpString( $input , $lang )
{
if( $lang == $this->getOption('sourceLanguage') ) // we dont need to translate a string from the source language to the source language
{
// this would be a cool feature, if i get it done :-)
$url=$this->getOption('translatorUrl');
if( $url )
{
$this->getAll( $lang );
return $this->addTranslatorLinks( $input , $url );
}
return $input;
}
$this->getAll( $lang ); // get all the possible strings from the DB
$addModifier = $this->getOption('caseSensitive') ? '' : 'i';
// FIXXME replace all spaces by something like this: (\s*|<br>|<br/>|<font.*>|</font>|<i>|</i>|<b>|</b>| )
// simply all those formatting tags which dont really cancel the phrase that should be translated
// and put them back in the translated string
// by filling $x in the right place and updating $lastSubpattern
// then it will be really cool and the text to translate will be recognized with any kind of space inbetween
// *NO*
// we dont want to do the above thing, since the formatting inside a text needs to be taken care of
// by the translator, this method here has no chance to know that 'this is a <b>user</b> from germany'
// has to become 'dies ist ein <b>Benutzer</b> aus Deutschland', the grammar of languages might be so different,
// that the marked part can be in a totally different place in the destination language string!
if(is_array($this->_translated['strings']) && sizeof($this->_translated['strings']))
foreach( $this->_translated['strings'] as $aString ) // search for each single string and try to translate it
{
$lastSubpattern = 2;
// we use 2 strings that we search for, one is the real text as from the db
// the second $htmlSourceString is the source string but with all non html characters
// translated using htmlentities, in case someone has been programming proper html :-)
// shall the translated string be converted to HTML, or does it may be contain HTML?
if( isset($aString['convertToHtml']) && $aString['convertToHtml'] )
$translated = htmlentities($aString['translated']);
else
$translated = $aString['translated'];
if( $aString['numSubPattern'] )// if the string is a regExp, we need to update $lastSubpattern
{
// we dont preg_quote the strings which contain regExps's
// if chars like '.' or alike which also appear in regExps they have to be
// escaped by the person who enters the source-string (may be we do that better one day)
$sourceString = $aString['string'];
$htmlSourceString = htmlentities($aString['string']);// we should not preg_quote the string
$lastSubpattern = 2 + $aString['numSubPattern']; // set $lastSubpattern properly
// in the DB the spaceholders start with $1, but here we need it
// to start with $2, that's what the following does
preg_match_all ( '/\$(\d)/' , $translated , $res );
$res[0] = array_reverse($res[0]); // reverse the arrays, since we replace $1 by $2 and then $2 by $3 ...
$res[1] = array_reverse($res[1]); // ... if we wouldnt reverse all would become $<lastNumber>
foreach( $res[0] as $index=>$aRes )
{
$aRes = preg_quote($aRes);
$translated = preg_replace( '/'.$aRes.'/' , '\$'.($res[1][$index]+1) , $translated );
}
}
else
{
// in none regExp's source strings we better quote the chars which could be
// mistakenly seen as regExp chars
$sourceString = preg_quote(trim($aString['string']));
$htmlSourceString = preg_quote(htmlentities(trim($aString['string'])));
// escape all slashes, since preg_quote doenst do that :-(
$sourceString = str_replace('/','\/',$sourceString);
$htmlSourceString = str_replace('/','\/',$htmlSourceString);
}
foreach( $this->possibleMarkUpDelimiters as $delimiters ) // go thru all the delimiters and try to translate the strings
{
// FIXXME there might be a major problem:
// <td
// {if($currentPageIndex==$key)}
// class="naviItemSelected" this line will also be tried to translated, since the line before and the one after
// will start/end with php tags, which also start/end with a < or > which are possible delimtier :-(
// {else}
// class="naviItem"
// nowrap>
//
//
$numSubPatterns = array(0,0);
$begin = $delimiters[0];
$end = $delimiters[1];
if( isset($delimiters[2]) ) $numSubPatterns[0] = $delimiters[2];
if( isset($delimiters[3]) ) $numSubPatterns[1] = $delimiters[3];
// replace all spaces in the source string by \s* so that there can be spaces
// as many as one wants
// and even newlines (the modifier s in the preg_replace takes care of that)
$sourceString = preg_replace('/\s+/','\\s*',$sourceString);
$htmlSourceString = preg_replace('/\s+/s','\\s*',$htmlSourceString);
$_hashCode = md5($input);
$input = preg_replace( '/('.$begin.')'.$sourceString.'('.$end.')/sU'.$addModifier ,
'$1'.$translated.'$'.($lastSubpattern+$numSubPatterns[0]) ,
$input );
// if the regExp above didnt have no effect try this one with all html characters translated
// if we wouldnt check this i had the effect that something was translated twice ...
// dont know exactly why but it did :-)
if( $_hashCode == md5($input) )
{
// try also to translate the string with all non-HTML-characters translated using htmlentities
// may be someone was creating proper html :-)
$input = preg_replace( '/('.$begin.')'.$htmlSourceString.'('.$end.')/sU'.$addModifier ,
'$1'.$translated.'$'.($lastSubpattern+$numSubPatterns[0]) ,
$input );
}
}
}
return $input;
}
/**
*
*
* @access public
* @author Wolfram Kriesing <wolfram@kriesing.de>
* @version 02/04/14
* @param string the url to a translation tool
* @return
*/
/* function addTranslatorLinks( $input , $url )
{
$linkBegin = '<a href="#" onClick="javascript:window.open(\''.$url.'?string=';
$linkEnd = '\',\'translate\',\'left=100,top=100,width=400,height=200\')" '.
'style="background-color:red; color:white; font-style:Courier; font-size:12px;"> T </a>';
foreach( $this->_translated['strings'] as $aString ) // search for each single string and try to translate it
{
$englishString = preg_quote($aString['string']);
if( $aString['numSubPattern'] ) // if the string is a regExp, we need to update $lastSubpattern
{
$englishString = $aString['string'];// we should not preg_quote the string
$lastSubpattern = '$'.( 2 + $aString['numSubPattern'] ); // set $lastSubpattern properly
}
$link = $linkBegin.urlencode($englishString).$linkEnd;
$input = preg_replace( '/(\s*>\s*)('.$englishString.')(\s*<\/a>)/isU' , '$1$2$3'.$link , $input );
$input = preg_replace( '/(<option.*>\s*)('.$englishString.')(.*<\/select>)/isU' , '$1$2$3'.$link , $input );
$input = preg_replace( '/(<input[^>]*type=.?(button|submit|reset)[^>]*value=.?\s*)'.
'('.$englishString.')([^>]*>)/isU' , '$1$3$4'.$link , $input );
}
return $input;
}
# '>\s*' => '\s*<', // this mostly applies, that a text is inbetween '>' and '<'
# '<\s*input .*value=["\']?\s*' => '\s*["\']?.*>' // this is for input button's values
*/
} // end of class
?>
|