z3c.batching 2.2

Simple Batching

This module implements a simple batching mechanism that allows you to split a large sequence into smaller batches. Let’s start by creating a simple list, which will be our full sequence:

Batch on empty root:

>>> from z3c.batching.batch import Batch
>>> batch = Batch([], size=3)
>>> len(batch)
0
>>> bool(batch)
False
>>> batch.firstElement
Traceback (most recent call last):
...
IndexError: ...

>>> batch.lastElement
Traceback (most recent call last):
...
IndexError: ...

>>> batch[0]
Traceback (most recent call last):
...
IndexError: ...

>>> batch.next is None
True

>>> batch.previous is None
True

>>> sequence = ['one', 'two', 'three', 'four', 'five', 'six', 'seven',
...             'eight', 'nine', 'ten', 'eleven', 'twelve', 'thirteen']

We can now create a batch for this sequence. Let’s make our batch size 3:

>>> batch = Batch(sequence, size=3)

The first argument to the batch is always the full sequence. If no start element is specified, the batch starts at the first element:

>>> list(batch)
['one', 'two', 'three']

The start index is commonly specified in the constructor though:

>>> batch = Batch(sequence, start=6, size=3)
>>> list(batch)
['seven', 'eight', 'nine']

Note that the start is an index and starts at zero. If the start index is greater than the largest index of the sequence, an index error is raised:

>>> Batch(sequence, start=15, size=3)
Traceback (most recent call last):
...
IndexError: start index key out of range

A batch implements the finite sequence interface and thus supports some standard methods. For example, you can ask the batch for its length:

>>> len(batch)
3

Note that the length returns the true size of the batch, not the size we asked for:

>>> len(Batch(sequence, start=12, size=3))
1

Like any sequence, a non-empty batch is true-ish in a boolean context:

>>> bool(batch)
True

You can also get an element by index, which is relative to the batch:

>>> batch[0]
'seven'
>>> batch[1]
'eight'
>>> batch[2]
'nine'

Slicing:

>>> batch[:1]
['seven']

>>> batch[1:2]
['eight']

>>> batch[1:]
['eight', 'nine']

>>> batch[:]
['seven', 'eight', 'nine']

>>> batch[10:]
[]

If you ask for index that is out of range, an index error is raised:

>>> batch[3]
Traceback (most recent call last):
...
IndexError: batch index out of range

You can also iterate through the batch:

>>> iterator = iter(batch)
>>> next(iterator)
'seven'
>>> next(iterator)
'eight'
>>> next(iterator)
'nine'

Batch also implement some of IReadSequence interface:

>>> 'eight' in batch
True

>>> 'ten' in batch
False

>>> batch == Batch(sequence, start=6, size=3)
True

>>> batch != Batch(sequence, start=6, size=3)
False

>>> batch != Batch(sequence, start=3, size=3)
True

Besides all of those common API methods, there are several properties that were designed to make your life simpler. The start and size are specified:

>>> batch.start
6
>>> batch.size
3

The end index of the batch is immediately computed:

>>> batch.end
8

The UI often requires that the number of the batch and the total number of batches is computed:

>>> batch.number
3
>>> batch.total
5

You can also ask for the next batch:

>>> batch.next
<Batch start=9, size=3>

If the current batch is the last one, the next batch is None:

>>> Batch(sequence, start=12, size=3).next is None
True

The previous batch shows the previous batch:

>>> batch.previous
<Batch start=3, size=3>

If the current batch is the first one, the previous batch is None:

>>> Batch(sequence, start=0, size=3).previous is None
True

The final two properties deal with the elements within the batch. They ask for the first and last element of the batch:

>>> batch.firstElement
'seven'

>>> batch.lastElement
'nine'

Total batches:

>>> batch = Batch(sequence[:-1], size=3)
>>> batch.total
4

We can have access to all batches:

>>> len(batch.batches)
4

>>> batch.batches[0]
<Batch start=0, size=3>

>>> batch.batches[3]
<Batch start=9, size=3>

>>> batch.batches[4]
Traceback (most recent call last):
...
IndexError: ...

>>> batch.batches[-1]
<Batch start=9, size=3>

>>> batch.batches[-2]
<Batch start=6, size=3>

Slicing:

>>> batch.batches[:1]
[<Batch start=0, size=3>]

>>> batch.batches[:]
[<Batch start=0, size=3>, <Batch start=3, size=3>, <Batch start=6, size=3>, <Batch start=9, size=3>]

>>> batch.batches[1:2]
[<Batch start=3, size=3>]

>>> batch.batches[1:]
[<Batch start=3, size=3>, <Batch start=6, size=3>, <Batch start=9, size=3>]

>>> batch.batches[10:]
[]

>>> batch.batches[2:50]
[<Batch start=6, size=3>, <Batch start=9, size=3>]

Batch neighbourhood of a large batch list

When the full list of batches is too large to be displayed in a user interface, we want to display only a subset of all the batches. A helper function is provided for that purpose:

First build a large sequence of batches (or anything else):

>>> batches = range(100)

Then extract only the first and last items, as well as the neighbourhood of the 46th item (index = 45). We want 3 neighbours at the left, 5 at the right:

>>> from z3c.batching.batch import first_neighbours_last
>>> first_neighbours_last(batches, 45, 3, 5)
[0, None, 42, 43, 44, 45, 46, 47, 48, 49, 50, None, 99]

‘None’ can be used to display a separator in a user interface (see z3c.table)

Subset Batching

>>> from z3c.batching.subset import SubsetBatch

Sometimes (for performance reasons), even though the user needs a batched UI, we want to limit the computation to the subset of values actually shown to the user.

Because we initialize the batch with a subset of data, we also need to provide explicitly the length of the full data set.

Let’s create a subset of data:

>>> data = range(20, 30)

We use it as part of a longer data set:

>>> batch = SubsetBatch(data, length=50, start=20, size=10)

Full API check:

>>> batch.firstElement
20
>>> batch.lastElement
29
>>> batch.index
2
>>> batch.number
3
>>> batch.total
5
>>> batch[2]
22
>>> len(batch)
10
>>> batch[-1] == batch.lastElement
True
>>> [item for item in batch]
[20, 21, 22, 23, 24, 25, 26, 27, 28, 29]

>>> batch.next
<EmptyBatch start=30, size=10>
>>> batch.previous
<EmptyBatch start=10, size=10>
>>> batch.next.previous == batch
True
>>> 22 in batch
True
>>> 10 in batch
False
>>> batch[5:8]
[25, 26, 27]

You have seen above that the contiguous batches are instances of the EmptyBatch class. As those instances hold no data, we raise errors to ensure that no batch provider tries to display item data:

>>> empty = batch.next
>>> empty
<EmptyBatch start=30, size=10>
>>> empty.firstElement
Traceback (most recent call last):
...
ValueError: EmptyBatch holds no item
>>> empty.lastElement
Traceback (most recent call last):
...
ValueError: EmptyBatch holds no item
>>> empty[0]
Traceback (most recent call last):
...
ValueError: EmptyBatch holds no item
>>> [item for item in empty]
Traceback (most recent call last):
...
ValueError: EmptyBatch holds no item

2.2 (2018-10-20)

• Add support for Python 3.6, 3.7 and PyPy3.

• Drop support for Python 2.6 and 3.3.

2.1.0 (2016-06-05)

• Support Python 3.3 through 3.5.

2.0.1 (2015-11-09)

• Standardize namespace __init__

2.0.0 (2013-02-25)

• New feature: Subset batch. Sometimes (for performance reasons), even though the user needs a batched UI, we want to limit the computation to the subset of values actually shown to the user.

• Register batch.Batch as named ("z3c.batching.batch") factory.

1.1.0 (2008-11-12)

• Added a function to build a small neighbourhood list of the current batch, from a large batch list. (extracted from z3c.table)

• Really fixed the bug with batches slicing

1.0.1 (2008-09-09)

• Fixed bug with batches slicing.

1.0.0 (2008-02-18)

• Initial release.