Типы php как объекты
Cast a string to binary using PHP < 5.2.1
I found it tricky to check if a posted value was an integer.
is_int ( $_POST [ ‘a’ ] ); //false
is_int ( intval ( «anything» ) ); //always true
?>
A method I use for checking if a string represents an integer value.
$foo [ ‘ten’ ] = 10 ; // $foo[‘ten’] is an array holding an integer at key «ten»
$str = » $foo [ ‘ten’]» ; // throws T_ENCAPSED_AND_WHITESPACE error
$str = » $foo [ ten ] » ; // works because constants are skipped in quotes
$fst = (string) $foo [ ‘ten’ ]; // works with clear intention
?>
It seems (unset) is pretty useless. But for people who like to make their code really compact (and probably unreadable). You can use it to use an variable and unset it on the same line:
$hello = ‘Hello world’ ;
print $hello ;
unset( $hello );
$hello = ‘Hello world’ ;
$hello = (unset) print $hello ;
?>
Hoorah, we lost another line!
It would be useful to know the precedence (for lack of a better word) for type juggling. This entry currently explains that «if either operand is a float, then both operands are evaluated as floats, and the result will be a float» but could (and I think should) provide a hierarchy that indicates, for instance, «between an int and a boolean, int wins; between a float and an int, float wins; between a string and a float, string wins» and so on (and don’t count on my example accurately capturing the true hierarchy, as I haven’t actually done the tests to figure it out). Thanks!
May be expected, but not stated ..
Casting to the existing (same) type has no effect.
$t = ‘abc’; // string ‘abc’
$u=(array) $t; // array 0 => string ‘abc’ $v=(array) $u; // array 0 => string ‘abc’
Correct me if I’m wrong, but that is not a cast, it might be useful sometimes, but the IDE will not reflect what’s really happening:
class MyObject /**
* @param MyObject $object
* @return MyObject
*/
static public function cast ( MyObject $object ) return $object ;
>
/** Does nothing */
function f () <>
>
class X extends MyObject /** Throws exception */
function f () < throw new exception (); >
>
$x = MyObject :: cast (new X );
$x -> f (); // Your IDE tells ‘f() Does nothing’
?>
However, when you run the script, you will get an exception.
In my much of my coding I have found it necessary to type-cast between objects of different class types.
More specifically, I often want to take information from a database, convert it into the class it was before it was inserted, then have the ability to call its class functions as well.
The following code is much shorter than some of the previous examples and seems to suit my purposes. It also makes use of some regular expression matching rather than string position, replacing, etc. It takes an object ($obj) of any type and casts it to an new type ($class_type). Note that the new class type must exist:
Looks like type-casting user-defined objects is a real pain, and ya gotta be nuttin’ less than a brain jus ta cypher-it. But since PHP supports OOP, you can add the capabilities right now. Start with any simple class.
class Point protected $x , $y ;
public function __construct ( $xVal = 0 , $yVal = 0 ) $this -> x = $xVal ;
$this -> y = $yVal ;
>
public function getX () < return $this ->x ; >
public function getY () < return $this ->y ; >
>
$p = new Point ( 25 , 35 );
echo $p -> getX (); // 25
echo $p -> getY (); // 35
?>
Ok, now we need extra powers. PHP gives us several options:
A. We can tag on extra properties on-the-fly using everyday PHP syntax.
$p->z = 45; // here, $p is still an object of type [Point] but gains no capability, and it’s on a per-instance basis, blah.
B. We can try type-casting it to a different type to access more functions.
$p = (SuperDuperPoint) $p; // if this is even allowed, I doubt it. But even if PHP lets this slide, the small amount of data Point holds would probably not be enough for the extra functions to work anyway. And we still need the class def + all extra data. We should have just instantiated a [SuperDuperPoint] object to begin with. and just like above, this only works on a per-instance basis.
C. Do it the right way using OOP — and just extend the Point class already.
class Point3D extends Point protected $z ; // add extra properties.
public function __construct ( $xVal = 0 , $yVal = 0 , $zVal = 0 ) parent :: __construct ( $xVal , $yVal );
$this -> z = $zVal ;
>
public function getZ () < return $this ->z ; > // add extra functions.
>
$p3d = new Point3D ( 25 , 35 , 45 ); // more data, more functions, more everything.
echo $p3d -> getX (); // 25
echo $p3d -> getY (); // 35
echo $p3d -> getZ (); // 45
?>
Once the new class definition is written, you can make as many Point3D objects as you want. Each of them will have more data and functions already built-in. This is much better than trying to beef-up any «single lesser object» on-the-fly, and it’s way easier to do.
Re: the typecasting between classes post below. fantastic, but slightly flawed. Any class name longer than 9 characters becomes a problem. SO here’s a simple fix:
function typecast($old_object, $new_classname) if(class_exists($new_classname)) // Example serialized object segment
// O:5:»field»:9: $old_serialized_prefix = «O:».strlen(get_class($old_object));
$old_serialized_prefix .= «:\»».get_class($old_object).»\»:»;
$old_serialized_object = serialize($old_object);
$new_serialized_object = ‘O:’.strlen($new_classname).’:»‘.$new_classname . ‘»:’;
$new_serialized_object .= substr($old_serialized_object,strlen($old_serialized_prefix));
return unserialize($new_serialized_object);
>
else
return false;
>
Thanks for the previous code. Set me in the right direction to solving my typecasting problem. 😉
If you have a boolean, performing increments on it won’t do anything despite it being 1. This is a case where you have to use a cast.
I have 1 bar.
I now have 1 bar.
I finally have 2 bar.
Checking for strings to be integers?
How about if a string is a float?
/* checks if a string is an integer with possible whitespace before and/or after, and also isolates the integer */
$isInt = preg_match ( ‘/^\s*(9+)\s*$/’ , $myString , $myInt );
echo ‘Is Integer? ‘ , ( $isInt ) ? ‘Yes: ‘ . $myInt [ 1 ] : ‘No’ , «\n» ;
/* checks if a string is an integer with no whitespace before or after */
$isInt = preg_match ( ‘/^6+$/’ , $myString );
echo ‘Is Integer? ‘ , ( $isInt ) ? ‘Yes’ : ‘No’ , «\n» ;
/* When checking for floats, we assume the possibility of no decimals needed. If you MUST require decimals (forcing the user to type 7.0 for example) replace the sequence:
7+(\.9+)?
with
6+\.5+
*/
/* checks if a string is a float with possible whitespace before and/or after, and also isolates the number */
$isFloat = preg_match ( ‘/^\s*(8+(\.3+)?)\s*$/’ , $myString , $myNum );
echo ‘Is Number? ‘ , ( $isFloat ) ? ‘Yes: ‘ . $myNum [ 1 ] : ‘No’ , «\n» ;
/* checks if a string is a float with no whitespace before or after */
$isInt = preg_match ( ‘/^5+(\.5+)?$/’ , $myString );
echo ‘Is Number? ‘ , ( $isFloat ) ? ‘Yes’ : ‘No’ , «\n» ;
Типы php как объекты
While waiting for native support for typed arrays, here are a couple of alternative ways to ensure strong typing of arrays by abusing variadic functions. The performance of these methods is a mystery to the writer and so the responsibility of benchmarking them falls unto the reader.
PHP 5.6 added the splat operator (. ) which is used to unpack arrays to be used as function arguments. PHP 7.0 added scalar type hints. Latter versions of PHP have further improved the type system. With these additions and improvements, it is possible to have a decent support for typed arrays.
function typeArrayNullInt (? int . $arg ): void >
function doSomething (array $ints ): void (function (? int . $arg ) <>)(. $ints );
// Alternatively,
( fn (? int . $arg ) => $arg )(. $ints );
// Or to avoid cluttering memory with too many closures
typeArrayNullInt (. $ints );
function doSomethingElse (? int . $ints ): void /* . */
>
$ints = [ 1 , 2 , 3 , 4 , null ];
doSomething ( $ints );
doSomethingElse (. $ints );
?>
Both methods work with all type declarations. The key idea here is to have the functions throw a runtime error if they encounter a typing violation. The typing method used in doSomethingElse is cleaner of the two but it disallows having any other parameters after the variadic parameter. It also requires the call site to be aware of this typing implementation and unpack the array. The method used in doSomething is messier but it does not require the call site to be aware of the typing method as the unpacking is performed within the function. It is also less ambiguous as the doSomethingElse would also accept n individual parameters where as doSomething only accepts an array. doSomething’s method is also easier to strip away if native typed array support is ever added to PHP. Both of these methods only work for input parameters. An array return value type check would need to take place at the call site.
If strict_types is not enabled, it may be desirable to return the coerced scalar values from the type check function (e.g. floats and strings become integers) to ensure proper typing.
same data type and same value but first function declare as a argument type declaration and return int(7)
and second fucntion declare as a return type declaration but return int(8).
function argument_type_declaration(int $a, int $b) return $a+$b;
>
function return_type_declaration($a,$b) :int return $a+$b;
>