Zend Framework Migration Notes

Zend Framework 1.10

Lors de la migration d'un version précédente vers Zend Framework 1.10 ou plus récent vous devriez prendre note de ce qui suit.

Zend_Controller_Front

A wrong behaviour was fixed, when there was no module route and no route matched the given request. Previously, the router returned an unmodified request object, so the front controller just displayed the default controller and action. Since Zend Framework 1.10, the router will correctly as noted in the router interface, throw an exception if no route matches. The error plugin will then catch that exception and forward to the error controller. You can then test for that specific error with the constant :

  1. /**
  2. * Before 1.10
  3. */
  4.     public function errorAction()
  5.     {
  6.         $errors = $this->_getParam('error_handler');
  7.  
  8.         switch ($errors->type) {
  9.             case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_CONTROLLER:
  10.             case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_ACTION:
  11.     // ...
  12.  
  13. /**
  14. * With 1.10
  15. */
  16.     public function errorAction()
  17.     {
  18.         $errors = $this->_getParam('error_handler');
  19.  
  20.         switch ($errors->type) {
  21.             case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_ROUTE:
  22.             case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_CONTROLLER:
  23.             case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_ACTION:
  24.     // ...

Zend_Feed_Reader

With the introduction of Zend Framework 1.10, Zend_Feed_Reader's handling of retrieving Authors and Contributors was changed, introducing a break in backwards compatibility. This change was an effort to harmonise the treatment of such data across the RSS and Atom classes of the component and enable the return of Author and Contributor data in more accessible, usable and detailed form. It also rectifies an error in that it was assumed any author element referred to a name. In RSS this is incorrect as an author element is actually only required to provide an email address. In addition, the original implementation applied its RSS limits to Atom feeds significantly reducing the usefulness of the parser with that format.

The change means that methods like getAuthors() and getContributors no longer return a simple array of strings parsed from the relevant RSS and Atom elements. Instead, the return value is an ArrayObject subclass called Zend_Feed_Reader_Collection_Author which simulates an iterable multidimensional array of Authors. Each member of this object will be a simple array with three potential keys (as the source data permits). These include: name, email and uri.

The original behaviour of such methods would have returned a simple array of strings, each string attempting to present a single name, but in reality this was unreliable since there is no rule governing the format of RSS Author strings.

The simplest method of simulating the original behaviour of these methods is to use the Zend_Feed_Reader_Collection_Author's getValues() which also returns a simple array of strings representing the "most relevant data", for authors presumed to be their name. Each value in the resulting array is derived from the "name" value attached to each Author (if present). In most cases this simple change is easy to apply as demonstrated below.

  1. /**
  2. * Before 1.10
  3. */
  4.  
  5. $feed = Zend_Feed_Reader::import('http://example.com/feed');
  6. $authors = $feed->getAuthors();
  7.  
  8. /**
  9. * With 1.10
  10. */
  11. $feed = Zend_Feed_Reader::import('http://example.com/feed');
  12. $authors = $feed->getAuthors()->getValues();

Zend_File_Transfer

Security change

For security reasons Zend_File_Transfer does no longer store the original mimetype and filesize which is given from the requesting client into its internal storage. Instead the real values will be detected at initiation.

Additionally the original values within $_FILES will be overridden within the real values at initiation. This makes also $_FILES secure.

When you are in need of the original values you can eighter store them before initiating Zend_File_Transfer or use the disableInfos option at initiation. Note that this option is useless when its given after initiation.

Count validation

Before release 1.10 the MimeType validator used a wrong naming. For consistency the following constants have been changed:

Changed Validation Messages
Old New Value
TOO_MUCH TOO_MANY Too many files, maximum '%max%' are allowed but '%count%' are given
TOO_LESS TOO_FEW Too few files, minimum '%min%' are expected but '%count%' are given

When you are translating these messages within your code then use the new constants. As benefit you don't need to translate the original string anymore to get a correct spelling.

Zend_Filter_HtmlEntities

In order to default to a more secure character encoding, Zend_Filter_HtmlEntities now defaults to UTF-8 instead of ISO-8859-1.

Additionally, because the actual mechanism is dealing with character encodings and not character sets, two new methods have been added, setEncoding() and getEncoding(). The previous methods setCharSet() and setCharSet() are now deprecated and proxy to the new methods. Finally, instead of using the protected members directly within the filter() method, these members are retrieved by their explicit accessors. If you were extending the filter in the past, please check your code and unit tests to ensure everything still continues to work.

Zend_Filter_StripTags

Zend_Filter_StripTags contains a flag, commentsAllowed, that, in previous versions, allowed you to optionally whitelist HTML comments in HTML text filtered by the class. However, this opens code enabling the flag to XSS attacks, particularly in Internet Explorer (which allows specifying conditional functionality via HTML comments). Starting in version 1.9.7 (and backported to versions 1.8.5 and 1.7.9), the commentsAllowed flag no longer has any meaning, and all HTML comments, including those containing other HTML tags or nested commments, will be stripped from the final output of the filter.

Zend_Translate

Xliff adapter

In past the Xliff adapter used the source string as message Id. According to the Xliff standard the trans-unit Id should be used. This behaviour was corrected with Zend Framework 1.10. Now the trans-unit Id is used as message Id per default.

But you can still get the incorrect and old behaviour by setting the useId option to FALSE.

  1. $trans = new Zend_Translate('xliff', '/path/to/source', $locale, array('useId' => false));

Zend_Validate

Adaptateurs personnels

Lorsqu'une erreur apparait dans un adaptateur crée de toute pièce, _error() doit être appelée. Avant Zend Framework 1.10, il était possible d'appeler cette méthode sans aucun paramètre. Le premier template de message d'erreur était alors utilisé.

Ce comportement est problématique lorsque vous avez des validateurs retournant plusieurs messages. Aussi, étendre un validateur peut mener à des comportements inattendus dans une telle situation, comme par exemple l'apparition du mauvais message d'erreur.

  1. My_Validator extends Zend_Validate_Abstract
  2. {
  3.     public isValid($value)
  4.     {
  5.         ...
  6.         $this->_error(); // Résultat inattendu
  7.         ...
  8.     }
  9. }

Pour éviter ces problèmes _error() doit desormais prendre obligatoirement un paramètre.

  1. My_Validator extends Zend_Validate_Abstract
  2. {
  3.     public isValid($value)
  4.     {
  5.         ...
  6.         $this->_error(self::MY_ERROR); // Ok, erreur définie
  7.         ...
  8.     }
  9. }

Simplification dans le validateur des dates

Avant Zend Framework 1.10, 2 messages identiques étaient envoyés dans le validateur des dates. NOT_YYYY_MM_DD et FALSEFORMAT. Depuis Zend Framework 1.10, seul FALSEFORMAT sera retourné lorsque la date passée ne correspond pas au format demandé.

Corrections dans Alpha, Alnum et Barcode

Avant Zend Framework 1.10, les messages dans les 2 validateurs barcode, le Alpha et le Alnum étaient identiques. Des problèmes pouvaient alors faire surface avec des messages personnalisés, des traducteurs ou des instances multiples des validateurs.

Depuis Zend Framework 1.10, les valeurs des constantes ont changé pour être uniques. Si vous utilisiez les constantes comme le manuel le recommande, aucun changement n'est nécessaire. Mais si vous utilisiez les messages d'erreurs, alors il faudra les changer. Voici les changements opérés:

Messages de validation disponibles
Validateur Constante Valeur
Alnum STRING_EMPTY alnumStringEmpty
Alpha STRING_EMPTY alphaStringEmpty
Barcode_Ean13 INVALID ean13Invalid
Barcode_Ean13 INVALID_LENGTH ean13InvalidLength
Barcode_UpcA INVALID upcaInvalid
Barcode_UpcA INVALID_LENGTH upcaInvalidLength
Digits STRING_EMPTY digitsStringEmpty

Zend Framework Migration Notes