4.5 accent in flash chart

[kboughanem]

Hi,

We've finished the french translation for SugarCRM 4.5.0 but there's a problem with accent in Flash charts

How can we do to resolve this problem...

Thanks

[Markku]

We have the same problem, any ideas?

[clint]

A bug fix was submitted to us today by our friends at Synolia. We're going to try it out, but I wanted to see if you folks could also see if it works for you.


In file "*/include/charts/charts.php/*" on line 86 the test for value $app_strings['LBL_CHARSET'] is:
if ($app_strings['LBL_CHARSET'] != "UTF-8") ....

It should be:
if ($app_strings['LBL_CHARSET'] == "UTF-8")

[Markku]

Works great, thanks for the info and kudos to Synolia guys!

line 86 in Charts.php states now

if ($app_strings['LBL_CHARSET'] == "UTF-8") {

[aserrano]

Hi, I had to revert that change.

For the Spanish language pack, it should be:

if ($app_strings['LBL_CHARSET'] != "UTF-8") {

instead of:

if ($app_strings['LBL_CHARSET'] == "UTF-8") {


The problem is that, when you set LBL_CHARSET to UTF-8 in include/language/xx_xx.lang.php, you use that character set for your localized PHP files. And if you do so, using utf8_encode on already UTF-8 encoded strings will mess up the results.

Given the fact that most strings used in the charts (maybe all strings?) come from localized files (_and_ therefore UTF-8 encoded when LBL_CHARSET is UTF-8), perhaps we should use utf8_encode not with the whole file, but just with the strings that don't come from localized files, if there are any of such strings.

If there are not such strings, I can't see the point on using utf8_encode when LBL_CHARSET is UTF-8. Only if the localized xx_xx.lang.php files used a differente encoding and LBL_CHARSET is UTF-8 and, in that case, the charts would display properly but not the rest of the HTML page.

AFAIK, utf8_encoded should be used just when LBL_CHARSET is not UTF-8, because the flash generation code requires, with its current configuration, that its input xml file is UTF-8 encoded, and that's ok.

IMHO, if the charts just use string literals from the localized language files, you should revert that patch, and encourage people to save their localized PHP files in the same character set they establish in LBL_CHARSET (time to update http://www.sugarcrm.com/wiki/index.p...Language_Packs ?)

UPDATE: This issue has been reported in bugs #9502 and #9200.

[dricrm]

I would change the whole if/else section with something like this:

if (function_exists('mb_convert_encoding')) {
		//best way, if available
		$xml_file = mb_convert_encoding($xml_file, 'UTF-8');
	}
	elseif ($app_strings['LBL_CHARSET'] != 'UTF-8') {
		if (function_exists('iconv')) {
			$xml_file = iconv($app_strings['LBL_CHARSET'], 'UTF-8', $xml_file);
		}
		else {
			$xml_file = utf8_encode($xml_file);
		}
	}
	if (fwrite($handle, $xml_file) === FALSE) {
		$GLOBALS['log']->debug("Cannot write to file ($filename)");
		return false;
	}

in this way:
1. it would try to use first the mb encoder, which doesnt harm already utf8 encoded strings
2. if mb is not available, but charset is not utf8, then we have to translate it for flash, so we try to use iconv, if available
3. if we dont have iconv, we try to use utf8 encode, but remember, it only encodes Latin1 correctly, so this is the worst case
4. then we write out the string, the writing, and the encoding code is now separated

regards,
Adam Maschek@DRI