From 5570ad60b496974607790dab49fc80cce8f5c700 Mon Sep 17 00:00:00 2001
From: thomascube <thomas@roundcube.net>
Date: Tue, 26 Jan 2010 02:19:56 -0500
Subject: [PATCH] Improved charset detection in vcard import + added unit tests for it
---
program/include/rcube_vcard.php | 64 ++++++++++++++++++++++---------
1 files changed, 45 insertions(+), 19 deletions(-)
diff --git a/program/include/rcube_vcard.php b/program/include/rcube_vcard.php
index dde14c5..320607b 100644
--- a/program/include/rcube_vcard.php
+++ b/program/include/rcube_vcard.php
@@ -47,10 +47,10 @@
/**
* Constructor
*/
- public function __construct($vcard = null)
+ public function __construct($vcard = null, $charset = RCMAIL_CHARSET)
{
if (!empty($vcard))
- $this->load($vcard);
+ $this->load($vcard, $charset);
}
@@ -59,18 +59,22 @@
*
* @param string vCard string to parse
*/
- public function load($vcard)
+ public function load($vcard, $charset = RCMAIL_CHARSET)
{
$this->raw = self::vcard_decode($vcard);
+
+ // resolve charset parameters
+ if ($charset == null)
+ $this->raw = $this->charset_convert($this->raw);
// find well-known address fields
- $this->displayname = $this->raw['FN'][0];
+ $this->displayname = $this->raw['FN'][0][0];
$this->surname = $this->raw['N'][0][0];
$this->firstname = $this->raw['N'][0][1];
$this->middlename = $this->raw['N'][0][2];
- $this->nickname = $this->raw['NICKNAME'][0];
- $this->organization = $this->raw['ORG'][0];
- $this->business = ($this->raw['X-ABShowAs'][0] == 'COMPANY') || (join('', (array)$this->raw['N'][0]) == '' && !empty($this->organization));
+ $this->nickname = $this->raw['NICKNAME'][0][0];
+ $this->organization = $this->raw['ORG'][0][0];
+ $this->business = ($this->raw['X-ABSHOWAS'][0][0] == 'COMPANY') || (join('', (array)$this->raw['N'][0]) == '' && !empty($this->organization));
foreach ((array)$this->raw['EMAIL'] as $i => $raw_email)
$this->email[$i] = is_array($raw_email) ? $raw_email[0] : $raw_email;
@@ -106,7 +110,7 @@
switch ($field) {
case 'name':
case 'displayname':
- $this->raw['FN'][0] = $value;
+ $this->raw['FN'][0][0] = $value;
break;
case 'firstname':
@@ -118,11 +122,11 @@
break;
case 'nickname':
- $this->raw['NICKNAME'][0] = $value;
+ $this->raw['NICKNAME'][0][0] = $value;
break;
case 'organization':
- $this->raw['ORG'][0] = $value;
+ $this->raw['ORG'][0][0] = $value;
break;
case 'email':
@@ -156,6 +160,28 @@
return $result;
}
+
+
+ /**
+ * Convert a whole vcard (array) to UTF-8.
+ * Each member value that has a charset parameter will be converted.
+ */
+ private function charset_convert($card)
+ {
+ foreach ($card as $key => $node) {
+ foreach ($node as $i => $subnode) {
+ if (is_array($subnode) && $subnode['charset'] && ($charset = $subnode['charset'][0])) {
+ foreach ($subnode as $j => $value) {
+ if (is_numeric($j) && is_string($value))
+ $card[$key][$i][$j] = rcube_charset_convert($value, $charset);
+ }
+ unset($card[$key][$i]['charset']);
+ }
+ }
+ }
+
+ return $card;
+ }
/**
@@ -168,11 +194,14 @@
{
$out = array();
+ // check if charsets are specified (usually vcard version < 3.0 but this is not reliable)
+ if (preg_match('/charset=/i', substr($data, 0, 2048)))
+ $charset = null;
// detect charset and convert to utf-8
- $encoding = self::detect_encoding($data);
- if ($encoding && $encoding != RCMAIL_CHARSET) {
- $data = rcube_charset_convert($data, $encoding);
+ else if (($charset = self::detect_encoding($data)) && $charset != RCMAIL_CHARSET) {
+ $data = rcube_charset_convert($data, $charset);
$data = preg_replace(array('/^[\xFE\xFF]{2}/', '/^\xEF\xBB\xBF/', '/^\x00+/'), '', $data); // also remove BOM
+ $charset = RCMAIL_CHARSET;
}
$vcard_block = '';
@@ -184,7 +213,7 @@
if (trim($line) == 'END:VCARD') {
// parse vcard
- $obj = new rcube_vcard(self::cleanup($vcard_block));
+ $obj = new rcube_vcard(self::cleanup($vcard_block), $charset);
if (!empty($obj->displayname))
$out[] = $obj;
@@ -217,9 +246,6 @@
// Remove cruft like item1.X-AB*, item1.ADR instead of ADR, and empty lines
$vcard = preg_replace(array('/^item\d*\.X-AB.*$/m', '/^item\d*\./m', "/\n+/"), array('', '', "\n"), $vcard);
- // remove vcard 2.1 charset definitions
- $vcard = preg_replace('/;CHARSET=[^:;]+/', '', $vcard);
-
// if N doesn't have any semicolons, add some
$vcard = preg_replace('/^(N:[^;\R]*)$/m', '\1;;;;', $vcard);
@@ -269,7 +295,7 @@
foreach($regs2[1] as $attrid => $attr) {
if ((list($key, $value) = explode('=', $attr)) && $value) {
- $value = trim($value);
+ $value = trim($value);
if ($key == 'ENCODING') {
// add next line(s) to value string if QP line end detected
while ($value == 'QUOTED-PRINTABLE' && preg_match('/=$/', $lines[$i]))
@@ -286,7 +312,7 @@
}
$entry = array_merge($entry, (array)self::vcard_unquote($line[2]));
- $data[$field][] = count($entry) > 1 ? $entry : $entry[0];
+ $data[$field][] = $entry;
}
}
--
Gitblit v1.9.1