PHP Australia Conference 2015

MongoCollection::group

(PECL mongo >=0.9.2)

MongoCollection::groupPerforms an operation similar to SQL's GROUP BY command

Description

public array MongoCollection::group ( mixed $keys , array $initial , MongoCode $reduce [, array $options = array() ] )

Parameters

keys

Fields to group by. If an array or non-code object is passed, it will be the key used to group results.

1.0.4+: If keys is an instance of MongoCode, keys will be treated as a function that returns the key to group by (see the "Passing a keys function" example below).

initial

Initial value of the aggregation counter object.

reduce

A function that takes two arguments (the current document and the aggregation to this point) and does the aggregation.

options

Optional parameters to the group command. Valid options include:

  • "condition"

    Criteria for including a document in the aggregation.

  • "finalize"

    Function called once per unique key that takes the final output of the reduce function.

  • "maxTimeMS"

    Specifies a cumulative time limit in milliseconds for processing the operation (does not include idle time). If the operation is not completed within the timeout period, a MongoExecutionTimeoutException will be thrown.

Return Values

Returns an array containing the result.

Changelog

Version Description
1.5.0 Added "maxTimeMS" option.
1.2.11 Emits E_DEPRECATED when options is scalar.

Examples

Example #1 MongoCollection::group() example

This groups documents by category and creates a list of names within that category.

<?php

$collection
->insert(array("category" => "fruit""name" => "apple"));
$collection->insert(array("category" => "fruit""name" => "peach"));
$collection->insert(array("category" => "fruit""name" => "banana"));
$collection->insert(array("category" => "veggie""name" => "corn"));
$collection->insert(array("category" => "veggie""name" => "broccoli"));

$keys = array("category" => 1);

$initial = array("items" => array());

$reduce "function (obj, prev) { prev.items.push(obj.name); }";

$g $collection->group($keys$initial$reduce);

echo 
json_encode($g['retval']);

?>

The above example will output something similar to:

[{"category":"fruit","items":["apple","peach","banana"]},{"category":"veggie","items":["corn","broccoli"]}]

Example #2 MongoCollection::group() example

This example doesn't use any key, so every document will be its own group. It also uses a condition: only documents that match this condition will be processed by the grouping function.

<?php

$collection
->save(array("a" => 2));
$collection->save(array("b" => 5));
$collection->save(array("a" => 1));

// use all fields
$keys = array();

// set intial values
$initial = array("count" => 0);

// JavaScript function to perform
$reduce "function (obj, prev) { prev.count++; }";

// only use documents where the "a" field is greater than 1
$condition = array('condition' => array("a" => array( '$gt' => 1)));

$g $collection->group($keys$initial$reduce$condition);

var_dump($g);

?>

The above example will output something similar to:

array(4) {
  ["retval"]=>
  array(1) {
    [0]=>
    array(1) {
      ["count"]=>
      float(1)
    }
  }
  ["count"]=>
  float(1)
  ["keys"]=>
  int(1)
  ["ok"]=>
  float(1)
}

Example #3 Passing a keys function

If you want to group by something other than a field name, you can pass a function as the first parameter of MongoCollection::group() and it will be run against each document. The return value of the function will be used as its grouping value.

This example demonstrates grouping by the num field modulo 4.

<?php

$c
->group(new MongoCode('function(doc) { return {mod : doc.num % 4}; }'),
     array(
"count" => 0),
     new 
MongoCode('function(current, total) { total.count++; }'));

?>
add a note add a note

User Contributed Notes 3 notes

up
0
ronan dot guilloux at gmail dot com
1 year ago
Since the $options argument is no longer a scalar (deprecated since the 1.2.11 version of the Mongo's PHP driver), you now must compose your $options as an associative array, otherwise you'll rise a "Implicitly passing condition as $options will be removed in the future" alert.

Example:

<?php

$m
= new Mongo();
$db = $m->selectDB('test');
$collection = new MongoCollection($db, 'FooBar');

// grouping results by categories, where foo is 'bar'
$keys = array('categorie'=>true, 'foo'=>true); // the fields list we want to return
$initial = array('count' => 0); // gets a subtotal for each categorie
$reduce = "function(obj,prev) { prev.count += 1; }"; // yes, this is js code
$conditions = array('foo'=> 'bar');
$grouped = $myColl::group($keys, $initial, $reduce, array('condition'=>$conditions));
$result = $grouped['retval'];

?>
up
-1
nick at pitchinteractive dot com
2 years ago
Here's my code to do an equivalent of a GROUP BY and a SUM

<?php
$contributionCol
= $db->customers->contribution;
$group = $contributionCol->group(array('date' => true), array('sum' => 0), "function (obj, prev) { prev['sum'] += obj.amount; }");
?>

This groups by the 'date' column and sums over the 'amount' column. In my testing this was much slower than querying all rows and doing the Group by code with PHP. It could just be my particular setup and data set.

Also at first my amount column was a string, which caused the results to be concatenated rather than arithmetic addition, something else to watch out for.

[red.: Instead of this, please use the new aggregation framework. See the documentation for MongoCollection::aggregate]
up
-1
Evgeniy Abduzhapparov
3 years ago
Here I am posting how I get tags from my documents. Documents should have 'tags' field which is array of strings:
{'tags':['php', 'mongo']}
<?php
$keys
= array();
$initial = array('tags'=>array(), 'count'=>0);
$reduce = 'function (doc, total) { if (doc.tags.length) {doc.tags.forEach(function (e) {total.tags[e]=total.tags[e]||0; total.tags[e]++; total.count++;});} }';
$criteria = array(
   
'condition' => array(
       
'tags' => array('$exists'=>true)
    ),
);
?>
To Top