The package provides generic data abstractions. The aim is to hide storage aspect from the operations of reading, writing and processing data.
Each data consists of items. Each item has multiple named fields. All items in a data set have the same structure.
The library provides interfaces for reading, writing and processing such data sets.
Data reader aim is to read data from a storage such as database, array or API and convert it to simple array of field => value items.
$reader = new MyDataReader(...);
$result = $reader->read();
Note that result is iterable
so you can use foreach
on it but need to prepare it if you need to use it as array:
// using is foreach
foreach ($result as $item) {
// ...
}
// preparing array
$dataArray = $result instanceof \Traversable ? iterator_to_array($result, true) : (array)$result;
Number of items returned can be limited:
$reader = (new MyDataReader(...))->withLimit(10);
In order to know total number of items in a data provider implementing CountableDataInterface
:
$reader = new MyDataReader(...);
$total = count($reader);
In order to filter data in a data provider implementing FilterableDataInterface
you need to supply filter to
withFilter()
method:
$filter = new All(
new GreaterThan('id', 3),
new Like('name', 'agent')
);
$reader = (new MyDataReader(...))
->withFilter($filter);
$data = $reader->read();
Filter could be composed with:
All
Any
Equals
GreaterThan
GreaterThanOrEqual
In
LessThan
LessThanOrEqual
Like
Not
The All
and Any
filters have a withFiltersArray()
method, which allows you to define filters with arrays.
$dataReader->withFilter((new All())->withFiltersArray([
['=', 'id', 88],
[
'or',
[
['=', 'color', 'red'],
['=', 'state', 1],
]
]
]));
In order to have your own filter:
- Implement at least
FilterInterface
, which includes:getOperator()
method that returns a string that represents a filter operationtoArray()
method that returns an array with filtering parameters.
- If you want to create a filter processor for a specific data reader type, then you need to implement at least
FilterProcessorInterface
. It has a singlegetOperator()
method that returns a string representing a filter operation. In addition, each data reader specifies an extended interface required for processing or building the operation. For example,IterableDataFilter
definesIterableProcessorInterface
, which contains additionalmatch()
method to execute a filter on PHP variables.
You can add your own filter processors to the data reader using the withFilterProcessors()
method. You can add any filter
processor to Reader. If reader is not able to use a filter, filter is ignored.
// own filter for filtering
class OwnNotTwoFilter implenents FilterInterface
{
private $field;
public function __construct($field)
{
$this->field = $field;
}
public static function getOperator(): string
{
return 'my!2';
}
public function toArray(): array
{
return [static::getOperator(), $this->field];
}
}
// own iterable filter processor for matching
class OwnIterableNotTwoFilterProcessor implements
{
public function getOperator(): string
{
return OwnNotTwoFilter::getOperator();
}
public function match(array $item, array $arguments, array $filterProcessors): bool
{
[$field] = $arguments;
return $item[$field] != 2;
}
}
// and using it on a data reader
$filter = new All(
new LessThan('id', 8),
new OwnNotTwoFilter('id'),
);
$reader = (new MyDataReader(...))
->withFilter($filter)
->withFilterProcessors(
new OwnIterableNotTwoFilterProcessor()
new OwnSqlNotTwoFilterProcessor() // for SQL
// and for any supported readers...
);
$data = $reader->read();
In order to sort data in a data provider implementing SortableDataInterface
you need to supply a sort object to
withSort()
method:
$sorting = Sort::only([
'id',
'name'
]);
$sorting = $sorting->withOrder(['name' => 'asc']);
// or $sorting = $sorting->withOrderString('name');
$reader = (new MyDataReader(...))
->withSort($sorting);
$data = $reader->read();
The goal of the Sort
is to map logical fields sorting to real data set fields sorting and form a criteria for the data
reader. Logical fields are the ones user operates with. Real fields are the ones actually present in a data set.
Such a mapping helps when you need to sort by a single logical field that, in fact, consists of multiple fields
in underlying the data set. For example, we provide a user with a username which consists of first name and last name
fields in actual data set.
To get a Sort
instance, you can use either Sort::only()
or Sort::any()
. Sort::only()
ignores user-specified order
for logical fields that have no configuration. Sort::any()
uses user-specified logical field name and order directly
for fields that have no configuration.
Either way you pass a config array that specifies which logical fields should be order-able and, optionally, details on how these should map to real fields order.
The current order to apply is specified via withOrder()
where you supply an array with keys corresponding to logical
field names and values correspond to order (asc
or desc
). Alternatively withOrderString()
can be used. In this case
ordering is represented as a single string containing comma separate logical field names. If the name is prefixed by -
,
ordering direction is set to desc
.
In case you need to skip some items from the beginning of data reader implementing OffsetableDataInterface
:
$reader = (new MyDataReader(...))->withOffset(10);
In order to have your own data reader you need to implement at least DataReaderInteface
. It has a single read()
method that returns iterable representing a set of items.
Additional interfaces could be implemented in order to support different pagination types, ordering and filtering:
CountableDataInterface
- allows getting total number of items in data provider.FilterableDataInterface
- allows returning subset of items based on criteria.SortableDataInterface
- allows sorting by one or multiple fields.OffsetableDataInterface
- allows to skip first N items when reading data.
Note that when implementing these, methods, instead of modifying data, should only define criteria that is later used
in read()
to affect what data is returned.
Pagination allows to obtain a limited subset of data that is both handy for displaying items page by page and for getting acceptable performance on big data sets.
There are two types of pagination provided: traditional offset pagination and keyset pagination.
Offset pagination is a common pagination method that selects OFFSET + LIMIT items and then skips OFFSET items.
Advantages:
- Total number of pages is available
- Can get to specific page
- Data can be unordered
Disadvantages:
- Performance degrades with page number increase
- Insertions or deletions in the middle of the data are making results inconsistent
Usage is the following:
$reader = (new MyDataReader(...));
$paginator = (new OffsetPaginator($dataReader))
->withPageSize(10)
->withCurrentPage(2);
$total = $paginator->getTotalPages();
$data = $paginator->read();
Keyset pagination is alternative pagination method that is good for infinite scrolling and "load more". It is selecting LIMIT items that have key field greater or lesser (depending on the sorting) than value specified.
Advantages:
- Performance does not depend on page number
- Consistent results regardless of insertions and deletions
Disadvantages:
- Total number of pages is not available
- Can not get to specific page, only "previous" and "next"
- Data cannot be unordered
Usage is the following:
$sort = Sort::only(['id', 'name'])->withOrderString('id');
$dataReader = (new MyDataReader(...))
->withSort($sort);
$paginator = (new KeysetPaginator($dataReader))
->withPageSize(10)
->withNextPageToken('13');
When displaying first page ID (or another field name to paginate by) of the item displayed last is used with withNextPageToken()
to obtain next page.
$writer = new MyDataWriter(...);
$writer->write($arrayOfItems);
$processor = new MyDataProcessor(...);
$processor->process($arrayOfItems);
The package is tested with PHPUnit. To run tests:
./vendor/bin/phpunit
The package tests are checked with Infection mutation framework. To run it:
./vendor/bin/infection
The code is statically analyzed with Psalm. To run static analysis:
./vendor/bin/psalm
The Yii Data is free software. It is released under the terms of the BSD License.
Please see LICENSE
for more information.
Maintained by Yii Software.