Filtering for Rules about validation

[septianw]

This is good feature, currently what i do is "filter it first and validate it." this feature will make my life simpler.

[ramon-to]

It's a nice feature!

But we already have a bug =P

<?php
  $myPureInt = v::int()->filter("OnlyWords"); // int(0)
  v::int()->validate($myPureInt); // true

I think will be better if we work in the other features, and release a stable version before we do this feature.

[alganet]

Not sure if its a bug yet, its behavior is currently not defined. What behavior do you expect with the sample you provided?

[ramon-to]

I expect the filter returns false:

<?php
  $myPureInt = v::int()->filter("OnlyWords"); // bool false
  v::int()->validate($myPureInt); // false

If the filter never returns false, what is the proposal of the validate function?

Maybe I not understand right the idea of the filter... =/

[alganet]

Perhaps the int was a bad idea for a first prototype.

I believe in filters as sanitizers. The validators are there just to tell if the data is right, but they don't alter anything. Filters are there to actively make the data right.

For example, The NoWhitespace as a validator returns false when it finds white space, NoWhitespace as a filter would probably strip out any whitespace it founds. Anything like that.

[ramon-to]

I get the concept, but if the filter can't make the data right?

<?php
$dirty = "Foo Bar"; 
$validator = v::string()->noWhitespace()->length(7,10);
$validator->filter($dirty);

And if the filter will make the data right, why we should validade?

<?php
   $myPureInt = v::int()->filter("12 pigs"); // int(12)
   (bool)$myPureInt; // true

   /* Invalid data  */

   $myPureInt = v::int()->filter("One pig"); // bool false
   (bool)$myPureInt; // false

[alganet]

Filtering is more like a "Fix whatever you can for me, I want clean data". Sometimes its impossible to make data right, thats why we should always validate it.

Perhaps some filters should always return null when they can't make things right, so the validator can catch those edge cases.

In some cases the filtering is before validation, in some cases its after validation. "12 " (twelve plus whitespace) is a valid integer number for the Int Rule, but isnt really a good practice to store this on a database with that trailing space. In this case, I would probably apply a post-validation filter before using that data.

Maybe we could split this feature into filters and sanitizers. Sanitizers would clean up data and filters would make them null if they don't follow the rules:

<?php
 $dirty = v::digits()->sanitize("12a34"); // string(4) "1234"
  v::digits()->validate($dirty); // true

 $dirty = v::digits()->filter("12a34"); // NULL
  v::digits()->validate($dirty); // false

[ramon-to]

Hm'...

I like the idea, but I guess it's better if filter work this way:

<?php
 $dirty = v::digits()->sanitize("12a34"); // string(4) "1234"
  v::digits()->validate($dirty); // true

  $dirty = v::digits()->filter("12a34"); // string(4) "1234"
  v::digits()->validate($dirty); // true


 $dirty = v::digits()->sanitize("foo"); // string(1) "0" 
  v::digits()->validate($dirty); // true

  $dirty = v::digits()->filter("foo"); // NULL
  v::digits()->validate($dirty); // false

The diference between filter and sanitize is:

• Filter will always return a "valid" data
• Sanitizers filter and validate the data

[alganet]

Didn't quite get your idea. Is the samples you provided right?

[ramon-to]

Sorry, I fix the sample...

[alganet]

The second sample could be:

<?php
 $dirty = v::digits()->sanitize("foo"); // string(0) "" 
  v::digits()->validate($dirty); // true

  $dirty = v::digits()->filter("foo"); // NULL
  v::digits()->validate($dirty); // false

The Digits rule accept an empty string as input.

The idea seems nice. Sanitizers could be implemented directly in AbstractRule. They're after all a standardized call to both filter() and validate().

We could raise up some real use cases for these guys, just to validate the concept a little further.

[ramon-to]

Prevent SQL Injection

Let's assume we have a product page, and we get a id parameter:

product.php?id=10

And some kid wanna hack our site:

product.php?id=10 some sql injection exploit --

Let's filter the parameter...

<?php
  $dirtyId = $_GET["id"];  // "10 some sql injection exploit --"
  $clearId = v::int()->filter($dirtyId); // int(10)
  if($clearId){
   /* Make a  Select in DB like:*/
   $sql = "SELECT title, price, photo FROM products WHERE id="+$clearId;
     /* Rest of the code */
   }else{
      echo "FBI will get you!";
  }

Help with parsing XML files

Get only the date of an node in a xml file:

<?xml version="1.0" encoding="utf-8" ?>
<post>
   <date>Posted in 13/10/2011</date>
   <title>FooBaar </title>
</post>

<?php
   $xml = simplexml_load_file('post.xml');
   $dirtyDate = $xml->date; // Posted in 13/10/2011
   $clearDate = v::date('d-m-Y')->sanitize($dirtyDate); // 13/10/2011

[alganet]

That seems enough for starting a better implementation. Gonna work on it soon!

[alganet]

Improved the current prototype here: 41286de

[alganet]

Me and @augustohp decided to drop this for now. Seems that no one used or needed. If you want this, please re-open the issue!