CiviCRM Community Forums (archive)

CiviCRM doesn't treat newlines correctly in iCal files. Rather than encoding newlines in text, it mistakenly strips them out, and tries to 'fold' the line instead. I have included an analysis and a patch for this problem.

Example

For example, the LOCATION ...

CQ Function Centre\n113 Queen St\nMelbourne, VIC 3000\n

... contains newline characters (\n) and is encoded by CiviCRM as...

LOCATION:CQ Function Centre
 113 Queen St
 Melbourne\, VIC 3000
 

... which an iCal reader will interpret as ...

LOCATION:CQ Function Centre113 Queen StMelbourne, VIC 3000

Note that the newlines are missing.

Analysis

RTF 2445 http://www.ietf.org/rfc/rfc2445.txt says that newline characters in text should be encoded as "\\n" (a backslash followed by an 'n'). It also specifies that lines longer than 75 bytes should be 'folded' to keep lines from getting too long. A line is 'folded' by inserting the character sequence "\n " (a newline character followed by a whitespace character). The reader of an iCal file must 'unfold' lines by removing these sequences before any processing is done.

CiviCRM confuses the encoding of newlines with folding of lines. Here is a patch to fix the code on lines 55-65 of CRM/Utils/ICalendar.php ...

     static function formatText( $text )
     {
         $text = strip_tags($text);
         $text = str_replace("\"", "DQUOTE", $text);
         $text = str_replace("\\", "\\\\", $text);
         $text = str_replace(",", "\,", $text);
         $text = str_replace(":", "\":\"", $text);
         $text = str_replace(";", "\;", $text);
-        $text = str_replace("\n", "\n ", $text);
+        $text = str_replace("\n", "\\n", $text);
+        $text = implode("\n ", str_split($text, 50));
         return $text;
     }
The line to be deleted does not encode the newline, but replaces it with a 'fold' sequence. Not only are newlines stripped out, but the folding is not done correctly: the 75 byte limit is not respected.

The lines to be added fix this by encoding the newlines, and then inserting the fold sequences. Note that '50' is suggested rather than '75' ...

• The first line may already contain text when formatText is called. In iCal.tpl the longest parameter name is "ORGANIZER:MAILTO:" which is 17 characters. A value of '50' means such a line will respect the RTF
• The RFC says "75 octets" rather than "75 characters". As the encoding may be UTF-8 '50' allows some contingency.
• A programmer would use a constant rather than a hard-wired number

Ken

Yashodha,

I've found another problem with that function. Colons are being quoted unnecessarily. The following patch should be applied to lines 55- 66 CRM/Utils/ICalendar.php ...

     static function formatText( $text )
     {
         $text = strip_tags($text);
         $text = str_replace("\"", "DQUOTE", $text);
         $text = str_replace("\\", "\\\\", $text);
         $text = str_replace(",", "\,", $text);
-        $text = str_replace(":", "\":\"", $text);
         $text = str_replace(";", "\;", $text);
         $text = str_replace("\n", "\\n", $text);
         $text = implode("\n ", str_split($text, 50));
         return $text;
     }
This is what http://www.ietf.org/rfc/rfc2445.txt says about escaping characters...

     text       = *(TSAFE-CHAR / ":" / DQUOTE / ESCAPED-CHAR)
     ; Folded according to description above

     ESCAPED-CHAR = "\\" / "\;" / "\," / "\N" / "\n")
        ; \\ encodes \, \N or \n encodes newline
        ; \; encodes ;, \, encodes ,

     TSAFE-CHAR = %x20-21 / %x23-2B / %x2D-39 / %x3C-5B
                  %x5D-7E / NON-US-ASCII
        ; Any character except CTLs not needed by the current
        ; character set, DQUOTE, ";", ":", "\", ","

Note that the colon character is not escaped.

Ken