Unpack Cassandra Map Type

Question

I have created some data using Cassandra DB 2.0.1 (CQL 3)

CREATE TABLE fans (id text PRIMARY KEY, fans map<text, text>);
INSERT INTO fans (id, fans) VALUES ('John', {'fruit' : 'apple', 'band' : 'Beatles'});
UPDATE fans SET fans = fans + {'movie' : 'Cassablanca'} WHERE id = 'John';

It work's fine.

cqlsh:testdb> SELECT * FROM fans;

 id   | fans
------+---------------------------------------------------------------
 John | {'band': 'Beatles', 'fruit': 'apple', 'movie': 'Cassablanca'}

(1 rows)

Now I'm trying to get data with PHP (thobbs/phpcassa v1.1.0).

include_once ("/include/autoload.php");
$pool = new phpcassa\Connection\ConnectionPool('testdb');
$connection = $pool->get();
$rows = $connection->client->execute_cql3_query("SELECT id, fans FROM fans", cassandra\Compression::NONE, cassandra\ConsistencyLevel::ONE);
var_dump($rows->rows);
$pool->return_connection($connection);
unset($connection);
$pool->close();

It also work's fine.

array (size=1)
  0 => 
    object(cassandra\CqlRow)[10]
      public 'key' => string '' (length=0)
      public 'columns' => 
        array (size=2)
          0 => 
            object(cassandra\Column)[11]
              public 'name' => string 'id' (length=2)
              public 'value' => string 'John' (length=4)
              public 'timestamp' => null
              public 'ttl' => null
          1 => 
            object(cassandra\Column)[12]
              public 'name' => string 'fans' (length=4)
              public 'value' => string '��band�Beatles�fruit�apple�movie�Cassablanca' (length=51)
              public 'timestamp' => null
              public 'ttl' => null

The problem is how to unpack the value that represented as a map? I can see

��band�Beatles�fruit�apple�movie�Cassablanca

and I know it showld be

{'band': 'Beatles', 'fruit': 'apple', 'movie': 'Cassablanca'}

Is there any internal function to deserialize or unpack that encoded string into a map or array?

I wroute a function that reads non-printable symbols:

function unistr_to_ords($str, $encoding = 'UTF-8') {
    $str = mb_convert_encoding($str, 'UCS-4BE', $encoding);
    $ords = array();
    for ($i = 0; $i < mb_strlen($str, 'UCS-4BE'); $i++) {
        $s2 = mb_substr($str, $i, 1, 'UCS-4BE');
        $val = unpack('N', $s2);
        $ords[] = $val[1];
    }
    return($ords);
}

And when I try it with that value I see the following result:

array (size=51)
  0 => int 0
  1 => int 3
  2 => int 0
  3 => int 4
  4 => int 98
  5 => int 97
  6 => int 110
  7 => int 100
  8 => int 0
  9 => int 7
  10 => int 66
  11 => int 101
  12 => int 97
  13 => int 116
  14 => int 108
  15 => int 101
  16 => int 115
  17 => int 0
  18 => int 5
  19 => int 102
  20 => int 114
  21 => int 117
  22 => int 105
  23 => int 116
  24 => int 0
  25 => int 5
  26 => int 97
  27 => int 112
  28 => int 112
  29 => int 108
  30 => int 101
  31 => int 0
  32 => int 5
  33 => int 109
  34 => int 111
  35 => int 118
  36 => int 105
  37 => int 101
  38 => int 0
  39 => int 11
  40 => int 67
  41 => int 97
  42 => int 115
  43 => int 115
  44 => int 97
  45 => int 98
  46 => int 108
  47 => int 97
  48 => int 110
  49 => int 99
  50 => int 97

As I understood 0 (zero) is a splitter, after 0 is a length, e.g. first 03 means 3 items in the map. Then 04 means 4 is a length of word 'band', then 07 means new word with length 7 for word 'Beatles' on so on.

But is any internal built-in method or function to extract map, list or set?

Answer 1

Somebody may have hacked something together, but for now the official answer is that phpcassa doesn't support CQL3 and hence doesn't support CQL3-only features like column collections (maps, sets, and lists).

Answer 2

I was hoping to just find something to do this, but couldn't so I hacked this up. Works for a map of <text,text>. Which is what I'm using. Thought I'd put it on here for anyone else looking.

function cql3_map_to_array($map) {
    $byte_array = unpack('C*', $map);
    $map_array = array();

    if(sizeof($byte_array) > 2) {
        $pos = 1;
        $size = (intval($byte_array[$pos]) * 256) + intval($byte_array[$pos + 1]);
        $pos += 2;

        while($size > 0) {
            $length = (intval($byte_array[$pos]) * 256) + intval($byte_array[$pos + 1]);
            $pos += 2;
            $key = substr($map, $pos - 1, $length);
            $pos += $length;

            $length = (intval($byte_array[$pos]) * 256) + intval($byte_array[$pos + 1]);
            $pos += 2;
            $value = substr($map, $pos - 1, $length);
            $pos += $length;

            $map_array[$key] = $value;

            $size--;
        }
    }

    return $map_array;
}