In Python (and others), you can incrementally process large volumes of data by using the 'yield' operator in a function. What would be the similar way to do so in PHP?
For example, lets say in Python, if I wanted to read a potentially very large file, I could work on each line one at a time like so (this example is contrived, as it is basically the same thing as 'for line in file_obj'):
def file_lines(fname):
f = open(fname)
for line in f:
yield line
f.close()
for line in file_lines('somefile'):
#process the line
What I'm doing right now (in PHP) is I'm using a private instance variable to keep track of state, and acting accordingly each time the function is called, but it seems like there must be a better way.
PHP has a direct equivalent called generators.
Old (pre php 5.5 answer):
Unfortunately, there isn't a language equivalent. The easiest way is to either to what you're already doing, or to create a object that uses instance variables to maintain state.
There is however a good option if you want to use the function in conjunction with the foreach-statement: SPL Iterators. They can be used to achieve something quite similar to python generators.
There may not be an equivalent operator, but the following code is equivalent in function and overhead:
function file_lines($file) {
static $fhandle;
if ( is_null($fhandle) ) {
$fhandle = fopen($file, 'r');
if ( $fhandle === false ) {
return false;
}
}
if ( ($line = fgets($fhandle))!== false ) {
return $line;
}
fclose($fhandle);
$fhandle = null;
}
while ( $line = file_lines('some_file') ) {
// ...
}
That looks about right. Sorry, I haven't tested it.
I prototype everything in Python before implementing in any other languages, including PHP. I ended up using callbacks to achieve what I would with the yield
.
function doSomething($callback)
{
foreach ($something as $someOtherThing) {
// do some computations that generates $data
call_user_func($callback, $data);
}
}
function myCallback($input)
{
// save $input to DB
// log
// send through a webservice
// etc.
var_dump($input);
}
doSomething('myCallback');
This way each $data
is passed to the callback function and you can do what you want.
Extending @Luiz's answer - another cool way is to use anonymous functions:
function iterator($n, $cb)
{
for($i=0; $i<$n; $i++) {
call_user_func($cb, $i);
}
}
$sum = 0;
iterator(10,
function($i) use (&$sum)
{
$sum += $i;
}
);
print $sum;
There is a rfc at https://wiki.php.net/rfc/generators adressing just that, which might be included in PHP 5.5.
In the mean time, check out this proof-of-concept of a poor mans "generator function" implemented in userland.
namespace Functional;
error_reporting(E_ALL|E_STRICT);
const BEFORE = 1;
const NEXT = 2;
const AFTER = 3;
const FORWARD = 4;
const YIELD = 5;
class Generator implements \Iterator {
private $funcs;
private $args;
private $key;
private $result;
public function __construct(array $funcs, array $args) {
$this->funcs = $funcs;
$this->args = $args;
}
public function rewind() {
$this->key = -1;
$this->result = call_user_func_array($this->funcs[BEFORE],
$this->args);
$this->next();
}
public function valid() {
return $this->result[YIELD] !== false;
}
public function current() {
return $this->result[YIELD];
}
public function key() {
return $this->key;
}
public function next() {
$this->result = call_user_func($this->funcs[NEXT],
$this->result[FORWARD]);
if ($this->result[YIELD] === false) {
call_user_func($this->funcs[AFTER], $this->result[FORWARD]);
}
++$this->key;
}
}
function generator($funcs, $args) {
return new Generator($funcs, $args);
}
/**
* A generator function that lazily yields each line in a file.
*/
function get_lines_from_file($file_name) {
$funcs = array(
BEFORE => function($file_name) { return array(FORWARD => fopen($file_name, 'r')); },
NEXT => function($fh) { return array(FORWARD => $fh, YIELD => fgets($fh)); },
AFTER => function($fh) { fclose($fh); },
);
return generator($funcs, array($file_name));
}
// Output content of this file with padded linenumbers.
foreach (get_lines_from_file(__FILE__) as $k => $v) {
echo str_pad($k, 8), $v;
}
echo "
";
The same sentence 'yield' exists now on PHP 5.5: