Improved charset detection in vcard import + added unit tests for it

5 files changed, 161 insertions, 22 deletions

diff --git a/program/include/rcube_vcard.php b/program/include/rcube_vcard.php
index dde14c541..320607b6e 100644
--- a/program/include/rcube_vcard.php
+++ b/program/include/rcube_vcard.php
@@ -47,10 +47,10 @@ class rcube_vcard
   /**
    * 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 @@ class rcube_vcard
    *
    * @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] : $raw_email;
@@ -106,7 +110,7 @@ class rcube_vcard
     switch ($field) {
       case 'name':
       case 'displayname':
-        $this->raw['FN'][0] = $value;
+        $this->raw['FN'][0][0] = $value;
         break;
         
       case 'firstname':
@@ -118,11 +122,11 @@ class rcube_vcard
         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 @@ class rcube_vcard
     
     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 @@ class rcube_vcard
   {
     $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 @@ class rcube_vcard
 
       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 @@ class rcube_vcard
     // 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 @@ class rcube_vcard
 
         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 @@ class rcube_vcard
         }
 
         $entry = array_merge($entry, (array)self::vcard_unquote($line[2]));
-        $data[$field][] = count($entry) > 1 ? $entry : $entry[0];
+        $data[$field][] = $entry;
       }
     }
 
diff --git a/tests/mailfunc.php b/tests/mailfunc.php
index ae35c5d77..0e9363a0b 100644
--- a/tests/mailfunc.php
+++ b/tests/mailfunc.php
@@ -51,11 +51,11 @@ class rcube_test_mailfunc extends UnitTestCase
     $this->assertPattern('#background="./program/blocked.gif"#', $html, "Replace external background image");
     $this->assertNoPattern('/ex3.jpg/', $html, "No references to external images");
     $this->assertNoPattern('/<meta [^>]+>/', $html, "No meta tags allowed");
-    $this->assertNoPattern('/<style [^>]+>/', $html, "No style tags allowed");
+    //$this->assertNoPattern('/<style [^>]+>/', $html, "No style tags allowed");
     $this->assertNoPattern('/<form [^>]+>/', $html, "No form tags allowed");
     $this->assertPattern('/Subscription form/', $html, "Include <form> contents");
-    $this->assertPattern('/<!-- input not allowed -->/', $html, "No input elements allowed");
-    $this->assertPattern('/<!-- link not allowed -->/', $html, "No external links allowed");
+    $this->assertPattern('/<!-- input ignored -->/', $html, "No input elements allowed");
+    $this->assertPattern('/<!-- link ignored -->/', $html, "No external links allowed");
     $this->assertPattern('/<a[^>]+ target="_blank">/', $html, "Set target to _blank");
     $this->assertTrue($GLOBALS['REMOTE_OBJECTS'], "Remote object detected");
     
diff --git a/tests/src/apple.vcf b/tests/src/apple.vcf
new file mode 100644
index 000000000..856eaf328
--- /dev/null
+++ b/tests/src/apple.vcf
@@ -0,0 +1,49 @@
+BEGIN:VCARD

+VERSION:3.0

+N:;;;;

+FN:Apple Computer AG

+ORG:Apple Computer AG;

+item1.ADR;type=WORK;type=pref:;;Birgistrasse 4a;Wallisellen-Zürich;;8304;Switzerland

+item1.X-ABADR:ch

+item2.URL;type=pref:http\://www.apple.ch

+item2.X-ABLabel:_$!<HomePage>!$_

+PHOTO;BASE64:

+  /9j/4AAQSkZJRgABAQAAAQABAAD/7QAcUGhvdG9zaG9wIDMuMAA4QklNBAQAAAAAAAD/2wBDAAEB
+  AQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEB
+  AQEBAQH/2wBDAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEB
+  AQEBAQEBAQEBAQEBAQEBAQH/wAARCAAwADADAREAAhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAA
+  AAECAwQFBgcICQoL/8QAtRAAAgEDAwIEAwUFBAQAAAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEI
+  I0KxwRVS0fAkM2JyggkKFhcYGRolJicoKSo0NTY3ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlq
+  c3R1dnd4eXqDhIWGh4iJipKTlJWWl5iZmqKjpKWmp6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW
+  19jZ2uHi4+Tl5ufo6erx8vP09fb3+Pn6/8QAHwEAAwEBAQEBAQEBAQAAAAAAAAECAwQFBgcICQoL
+  /8QAtREAAgECBAQDBAcFBAQAAQJ3AAECAxEEBSExBhJBUQdhcRMiMoEIFEKRobHBCSMzUvAVYnLR
+  ChYkNOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElKU1RVVldYWVpjZGVmZ2hpanN0dXZ3eHl6goOE
+  hYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3uLm6wsPExcbHyMnK0tPU1dbX2Nna4uPk5ebn
+  6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwD+/igAoAKAPmH43ftT+CfgzqNt4bNjeeLvGV2IHXw7
+  pVxDbLZx3LBbdtU1GVLhbN7jIMFvHa3VzIpWRoY4mWQ9dDCTrrmuoQ/mavfvZXV7dW2jkr4ynQfL
+  Zzn1inZL1lZ6+ST87H0lp1zLe6fY3k9s1nNd2dtczWjv5j2ss8KSyWzybU3tA7mJn2JuKk7Vzgcr
+  Vm1e9m1fvrv8zqi+aKbVm0nbtdXt8i5SGFAHHeOfH3hH4b6DP4k8Z61a6JpUBCCW4LPNdTsCUtbK
+  1iD3F5dSYO2C3jd8AuwVFZhdOnOrLlhFyf4Lzb6IzqVYUo81SSiundvslu3/AEz5i0n9u74Fanqo
+  064m8UaNbvII49X1TRUGnnORvmFje3t5BGTjDtatwcyCPBrreArpX9xvqlJ3/FJP7zlWYUHKzVRL
+  +ZxVvnaTa+4+QLvVPgJ4U+LutfFXx78QJfi3q954iuvEOieHvBmkyzaVbO1x5ulPruq6pLawTvp1
+  uLdIrK08xFmgXzl2oI67LV50o0qdP2K5VGUptKXnyxi29XfVtb6HDehGtKpUm6zcnNKCfK9brnlK
+  3pZJ37pH3/8ACj9qb4T/ABd1FdD0TUr3SPETozwaJ4ht47C5vQgLONOnjnuLS8dVBYwJOt1tDMIN
+  oLV51XCVqK5pJSj1cW3b1uk1vvqvM9Kji6VZ8qbjJ7KVlf0abTf4n0dXMdQUAfhf+2J8UNW8ffFz
+  W9Cnlki0XwDqOp+GdNsFdhB9qsr6a31DUDHu2tcXbwojSEbhFCka4Uc+9hKUadGMl8VRKcn11V0v
+  RJng4urKpWkntTlKCXTRtN+re58n11HKFAH6PfsGfBvQfEl3rHxR8Q2y38vhrUrfT/DNpIT5FvqY
+  iF1carIgI8ye2R4Y7RXykbySTbS6xlfOx9aUVGlF2503N+W1vnrc9HAUYzlKrJX5GuVf3t7+dvu3
+  vc/WKvIPXCgD8FP2s/BF/wCC/jd4wlu0It/Fmp6h4usJf4JINZ1G7ndVbu0UpZZBztY446V9BhZq
+  dCnZ/DFQfrFJHz+Kg4V6l/tSc15qTb/rzPmqug5woA++v2J/j74d+HN9rHgLxndrpmjeJr62vtJ1
+  mbP2Sw1dY/s0ltfOM+Rb30Yh8u5ZTHFPEFlZEk3Dgx2HlVUZw1lBNOPVp9vNa6X22137sFiI0ZSh
+  N2jNpqXSLSe/k+/R+p+w1eMe0FAHxz+2P8DLr4r+CIPEPhy2Nx4y8FJdXVnaxrmfWdGlAk1DSosK
+  WkuozEt5p8fHmTLNADuuQR24KuqU3GTtCpbXtLo32T2b9GcWNw7qwU4q84X06yi9Wl531XfXyPxK
+  ME4nNsYZRciUwG3MbicTh/LMJix5glD/ACGPbv3/AC4zxXtniFvUNJ1XSmjXVNM1DTWmUvCuoWVz
+  ZtKgxloxcRxmRRkZZcgZGTzSTT2afo7hqtz6w/ZB+BV78UPHtl4n1eykHgfwdewajfXE0bCDVtVt
+  mE+n6PAzAJOBOkdxqAUssdsnlSDNwoPLi66pU3FP95NNJdk95P06d35XOvCUHWqJtfu4NOT7vdR8
+  7vfsvVH7gV4R7oUAFAHKReA/A8Gsy+IofBnhSHxBO7ST67F4d0iPWZnb7zy6mlmL2R2/iZ5yT3NX
+  7Spbl9pPl/l55W+69iPZUubm9nDmvfm5I81+97XuX9a8MeGvEtsLLxF4e0PX7MMGFprWk2Gq2wZe
+  jCC+t54tw7HZkdqUZzg7wlKL7xk4v700OUIT0lCMl2lFS/NMuaXpOlaHZQ6Zoumafo+nW4It9P0u
+  yttPsoATkiG1tI4oIgTyQkagnmk5Sk7ybk+7bb+96jjGMVaMVFdopJfcjQpDP//Z

+X-ABShowAs:COMPANY

+X-ABUID:2E4CB084-4767-4C85-BBCA-805B1DCB1C8E\:ABPerson

+END:VCARD

diff --git a/tests/src/johndoe.vcf b/tests/src/johndoe.vcf
new file mode 100644
index 000000000..67b649df3
--- /dev/null
+++ b/tests/src/johndoe.vcf
@@ -0,0 +1,11 @@
+BEGIN:VCARD

+VERSION:2.1

+N;CHARSET=windows-1252:Doë;John;;;

+FN;CHARSET=windows-1252:John Doë

+ORG:roundcube.net;

+EMAIL;INTERNET;WORK:inbox@roundcube.net

+EMAIL;INTERNET;HOME;TYPE=pref:roundcube@gmail.com

+TEL;WORK:+123456789

+ADR;WORK:;;The street;Hometown;;5555;Cayman Islands

+NOTE:The notes...

+END:VCARD

diff --git a/tests/vcards.php b/tests/vcards.php
new file mode 100644
index 000000000..72c687484
--- /dev/null
+++ b/tests/vcards.php
@@ -0,0 +1,53 @@
+<?php
+
+/**
+ * Unit tests for class rcube_vcard
+ *
+ * @package Tests
+ */
+class rcube_test_vcards extends UnitTestCase
+{
+
+  function __construct()
+  {
+    $this->UnitTestCase('Vcard encoding/decoding tests');
+  }
+  
+  function _srcpath($fn)
+  {
+    return realpath(dirname(__FILE__) . '/src/' . $fn);
+  }
+  
+  function test_parse_one()
+  {
+    $vcard = new rcube_vcard(file_get_contents($this->_srcpath('apple.vcf')));
+    
+    $this->assertEqual(true, $vcard->business, "Identify as business record");
+    $this->assertEqual("Apple Computer AG", $vcard->displayname, "FN => displayname");
+    $this->assertEqual("", $vcard->firstname, "No person name set");
+  }
+
+  function test_parse_two()
+  {
+    $vcard = new rcube_vcard(file_get_contents($this->_srcpath('johndoe.vcf')), null);
+    
+    $this->assertEqual(false, $vcard->business, "Identify as private record");
+    $this->assertEqual("John Doë", $vcard->displayname, "Decode according to charset attribute");
+    $this->assertEqual("roundcube.net", $vcard->organization, "Test organization field");
+    $this->assertEqual(2, count($vcard->email), "List two e-mail addresses");
+    $this->assertEqual("roundcube@gmail.com", $vcard->email[0], "Use PREF e-mail as primary");
+  }
+  
+  function test_import()
+  {
+    $input = file_get_contents($this->_srcpath('apple.vcf'));
+    $input .= file_get_contents($this->_srcpath('johndoe.vcf'));
+    
+    $vcards = rcube_vcard::import($input);
+    
+    $this->assertEqual(2, count($vcards), "Detected 2 vcards");
+    $this->assertEqual("Apple Computer AG", $vcards[0]->displayname, "FN => displayname");
+    $this->assertEqual("John Doë", $vcards[1]->displayname, "Displayname with correct charset");
+  }
+  
+}