无法使聚合函数起作用

Question

I'm completely new to Mongo, and I'm having some serious trouble wrapping my head around the Aggregation Framework. I have a document that looks like this:

record {
    _id: '2013-04-22/tacos',
    first: {
        something: {0: 1, 1:4},
        somethingelse: {0: 18, 1:22}
    },
    second: {
        0: [11, 18, 7],
        1: [17, 9, 22]
    },
    metadata: {
        date: '2013-04-22'
    }
}

What I want to do is get sums (since this will need to work with multiple documents) of each item in first.something, each item in second, and then sums of all items in first.something , second.<whatever>.0 , second.<whatever>.1 , and second.<whatever>.2 . So I'm looking for a result like this (pardon my formatting):

sum(first.something.0):      1
sum(first.something.1):      4
sum(second.0.0):            11
sum(second.0.1):            18
sum(second.0.2):             7
sum(second.1.0):            17
sum(second.1.1):             9
sum(second.1.2):            22
sum(first.something.all):    5
sum(second.<whatever>.0):   28
sum(second.<whatever>.1):   27
sum(second.<whatever>.2):   29

Right now I'm just trying to get the first two working, on a single record, and I've got this:

// $db is my collection, and is verified to work
$aggregate = array(
    '$group' => array(
        '_id' => '$metadata.date',
        'first_zero' => array('$sum' => 'first.something.0'),
        'first_one' => array('$sum' => 'first.something.1')
    ),
    '$match' => array(
        "metadata.date" => '2013-04-22'
    )
);
$cursor = $db->aggregate($aggregate);
var_dump($cursor);

However my output tells me exception: A pipeline stage specification object must contain exactly one field. . But if I leave out the '$match' part, it returns the record with int(0) for first_zero and first_one as shown:

array(2) {
  ["result"]=>
  array(1) {
    [0]=>
    array(3) {
      ["_id"]=>
      string(10) "2013-04-22"
      ["clicks_zero"]=>
      int(0)
      ["clicks_one"]=>
      int(0)
    }
  }
  ["ok"]=>
  float(1)
}

So I can't even get the simple part working. What am I doing wrong here? And is the second.<whatever>.0 even a thing I can do?

Answer 1

I'm no expert in Php but should the aggregation query look like:

  $aggregate = array(
      array('$group' => array(
          '_id' => '$metadata.date',
          'first_zero' => array('$sum' => 'first.something.0'),
          'first_one' => array('$sum' => 'first.something.1')
      )),
      array('$match' => array(
          "metadata.date" => '2013-04-22'
      ))
  );

Each pipeline operator should be inside a hash/array.

Edit

However, the above would give you no result because of the $match pipeline at the end. There is no metadata.date field that is created in the previous pipeline, so no results will match your query. The right solution would be:

  $aggregate = array(
      array('$match' => array(
          "metadata.date" => '2013-04-22'
      )),
      array('$group' => array(
          '_id' => '$metadata.date',
          'first_zero' => array('$sum' => 'first.something.0'),
          'first_one' => array('$sum' => 'first.something.1')
      ))
  );

Which will give you:

{
"result" : [
    {
        "_id" : "2013-04-22",
        "first_zero" : 1,
        "first_one" : 4
    }
],
"ok" : 1
}