vcard folding problem with utf-8 database

[koessler]

The method rfc2425_fold in class rcube_vcard splits multibyte characters in the middle. When the database table is defined with character set UTF-8 the contact cannot be stored.

Create a contact with last name:

12345679801234567980123456798012345679801234567980123456798012345679801ä

and save the contact. You will get error message: "invalid byte sequence for encoding "UTF8": 0xc30a"

A possible correction would be:
1) be sure that php environment 'mbstring.internal_encoding' is set correctly to the internal coding of the vcard (probably UTF-8)
2) then you may use the following method rfc2425_fold_callback() in rcube_vcard.php, line 466:

  private static function rfc2425_fold_callback($matches)
  {
    return ":" . mb_ereg_replace('(.{72})', "\\1\n ", $matches[1]);
  }

[alec]

Isn't mb_ereg* deprecated? This can be done also with a loop and mb_substr(). Also we should add checking if these mb_* functions exists.

[koessler]

According to ​http://www.php.net/manual/de/function.mb-ereg-replace.php this method is not deprecated....

[alec]

I've found in rcube_vcard class comments that it was intended to store vcard unfolded. Attached patch fixes the DB error issue by disabling folding. However, we need still to fix folding.

[thomasb]

Fixed in [bf80b5a2]

[koessler]

Your fix is not working as desired. Pls read the description carefully.

Function mb_ereg_replace() works only when environment value "mbstring.internal_encoding" is set. Currently this variable is not set and therefore the method still splits the € sign inbetween the 3 bytes representation.
Before using mb_ereg_replace() be sure to set the variable "mbstring.internal_encoding" to UTF-8 (respectively to the same character coding which the database expects and which rcube_vcard uses internally)

Please reopen the ticket.

[alec]

It is set in inlude/iniset.php file.

[koessler]

Yes, you are right.
I checked trunk 4655, it's not working there. Strange, I will have closer look.........

[koessler]

Found the problem:
See ​http://www.php.net/manual/en/function.mb-ereg-replace.php:

Note:

The internal encoding or the character encoding specified by mb_regex_encoding() will be used as the character encoding for this function.

Solution: you also have to call mb_regex_encoding(RCMAIL_CHARSET) in iniset.php then it works.

[alec]

Fixed in [5c4c0666].