sections 0.0.0

Flexible tree data structures for organizing lists and dicts into sections.

Flexible tree data structures for organizing lists and dicts into sections.

sections is designed to be:

• Intuitive: Start quickly and spend less time reading the docs.

• Scalable: Grow arbitrarily complex trees as your problem scales.

• Flexible: Rapidly build nodes with any custom attributes, properties, and methods on the fly.

• Fast: Made with performance in mind - access lists and sub-lists/dicts in as little as Θ(1) time in many cases. See the Performance section for the full details.

• Reliable: Contains an exhaustive test suite and 100% code coverage.

See the full documentation at: https://sections.readthedocs.io/

Usage

pip install sections

import sections

menu = sections(
    'Breakfast', 'Dinner',
    mains=['Bacon&Eggs', 'Burger'],
    sides=['HashBrown', 'Fries'],
)
# Resulting structure's API and the expected results
assert menu.mains == ['Bacon&Eggs', 'Burger']
assert menu.sides == ['HashBrown', 'Fries']
assert menu['Breakfast'].main == 'Bacon&Eggs'
assert menu['Breakfast'].side == 'HashBrown'
assert menu['Dinner'].main == 'Burger'
assert menu['Dinner'].side == 'Fries'
assert menu('sides', list) == ['HashBrown', 'Fries']
assert menu('sides', dict) == {'Breakfast': 'HashBrown', 'Dinner': 'Fries'}
assert isinstance(menu, sections.Section)  # root section/node
assert isinstance(menu['Breakfast'], sections.Section)  # child section/node
assert isinstance(menu['Dinner'], sections.Section)     # child section/node

Attrs: Plural/singular hybrid attributes and more

Spend less time deciding between using the singular or plural form for an attribute name:

tasks = sections('pay bill', 'clean', status=['completed', 'started'])
assert tasks.statuses == ['completed', 'started']
assert tasks['pay bill'].status == 'completed'
assert tasks['clean'].status == 'started'

When an attribute is not found in a Section node, both the plural and singular forms of the word are then checked to see if the node contains the attribute under those forms of the word. If they are still not found, the node will recursively repeat the same search on each of its children, concatenating the results into a list or dict.

Properties: Easily add on the fly

Properties and methods are automatically added to a Section class instance when passed as keyword arguments:

schedule = sections(
    'Weekdays', 'Weekend',
    hours_per_day=[[8, 8, 6, 10, 8], [4, 6]],
    hours=property(lambda self: sum(self.hours_per_day)),
)
assert schedule['Weekdays'].hours == 40
assert schedule['Weekend'].hours == 10
assert schedule.hours == 50

Each sections() call returns a structure containing nodes of a unique class created in a factory function, where the class definition contains no logic except that it inherits from the Section class. This allows properties added to one structure creation to not affect the class instances in other structures.

Construction: Build gradually or all at once

Construct section-by-section, section-wise, attribute-wise, or other ways:

def demo_different_construction_techniques():
    """Example construction techniques for producing the same structure."""

    # Building section-by-section
    books = sections()
    books['LOTR'] = sections(topic='Hobbits', author='JRR Tolkien')
    books['Harry Potter'] = sections(topic='Wizards', author='JK Rowling')
    demo_resulting_object_api(books)

    # Section-wise construction
    books = sections(
        sections('LOTR', topic='Hobbits', author='JRR Tolkien'),
        sections('Harry Potter', topic='Wizards', author='JK Rowling')
    )
    demo_resulting_object_api(books)

    # Attribute-wise construction
    books = sections(
        'LOTR', 'Harry Potter',
        topics=['Hobbits', 'Wizards'],
        authors=['JRR Tolkien', 'JK Rowling']
    )
    demo_resulting_object_api(books)

    # setattr post-construction
    books = sections(
        'LOTR', 'Harry Potter',
    )
    books.topics = ['Hobbits', 'Wizards']
    books['LOTR'].author = 'JRR Tolkien'
    books['Harry Potter'].author = 'JK Rowling'
    demo_resulting_object_api(books)

def demo_resulting_object_api(books):
    """Example Sections API and expected results."""
    assert books.names == ['LOTR', 'Harry Potter']
    assert books.topics == ['Hobbits', 'Wizards']
    assert books.authors == ['JRR Tolkien', 'JK Rowling']
    assert books['LOTR'].topic == 'Hobbits'
    assert books['LOTR'].author == 'JRR Tolkien'
    assert books['Harry Potter'].topic == 'Wizards'
    assert books['Harry Potter'].author == 'JK Rowling'

demo_different_construction_techniques()

Details

Section names

The non-keyword arguments passed into a sections() call define the section names and are accessed through the attribute name. The names are used like keys in a dict to access each child section of the root Section node:

books = sections(
    'LOTR', 'Harry Potter',
    topics=['Hobbits', 'Wizards'],
    authors=['JRR Tolkien', 'JK Rowling']
)
assert books.names == ['LOTR', 'Harry Potter']
assert books['LOTR'].name == 'LOTR'
assert books['Harry Potter'].name == 'Harry Potter'

Names are optional, and by default, children will be given integer values corresponding to indices in an array, while a root has a default keyvalue of sections.NoneValue:

sect = sections(x=['a', 'b'])
assert sect.sections.names == [0, 1]
assert sect.name is sections.NoneValue

# the string representation of sections.NoneValue is 'section'
assert str(sect.name) == 'section'

Parent names and attributes

A parent section name can optionally be provided as the first argument in a list or Section instantiation by defining it in a set (surrounding it with curly brackets). This strategy avoids an extra level of braces when instantiating Sections. This idea applies also for defining parent attributes:

library = sections(
    {"Trevor's Bookshelf"},
    [{'Fantasy'}, 'LOTR', 'Harry Potter'],
    [{'Academic'}, 'Advanced Mathematics', 'Physics for Engineers'],
    topics=[{'All my books'},
            [{'Imaginary things'}, 'Hobbits', 'Wizards'],
            [{'School'}, 'Numbers', 'Forces']],
)
assert library.name == "Trevor's Bookshelf"
assert library.sections.names == ['Fantasy', 'Academic']
assert library['Fantasy'].sections.names == ['LOTR', 'Harry Potter']
assert library['Academic'].sections.names == [
    'Advanced Mathematics', 'Physics for Engineers'
]
assert library['Fantasy']['Harry Potter'].name == 'Harry Potter'
assert library.topic == 'All my books'
assert library['Fantasy'].topic == 'Imaginary things'
assert library['Academic'].topic == 'School'

Return attributes as a list, dict, or iterable

Access the data in different forms with the gettype argument in Sections.__call__() as follows:

menu = sections('Breakfast', 'Dinner', sides=['HashBrown', 'Fries'])

# return as list always, even if a single element is returned
assert menu('sides', list) == ['HashBrown', 'Fries']
assert menu['Breakfast']('side', list) == ['HashBrown']

# return as dict
assert menu('sides', dict) == {'Breakfast': 'HashBrown', 'Dinner': 'Fries'}
assert menu['Breakfast']('side', dict) == {'Breakfast': 'HashBrown'}

# return as iterator over elements in list (fastest method, theoretically)
for i, value in enumerate(menu('sides', iter)):
    assert value == ['HashBrown', 'Fries'][i]
for i, value in enumerate(menu['Breakfast']('side', iter)):
    assert value == ['HashBrown'][i]

See the __call__ method in the References section of the docs for more options: https://sections.readthedocs.io/

Set the default return type when accessing structure attributes by changing Section.default_gettype as follows:

menu = sections('Breakfast', 'Dinner', sides=['HashBrown', 'Fries'])

menu['Breakfast'].default_gettype = dict  # set for only 'Breakfast' node
assert menu.sides == ['HashBrown', 'Fries']
assert menu['Breakfast']('side') == {'Breakfast': 'HashBrown'}

menu.cls.default_gettype = dict           # set for all nodes in `menu`
assert menu('sides') == {'Breakfast': 'HashBrown', 'Dinner': 'Fries'}
assert menu['Breakfast']('side') == {'Breakfast': 'HashBrown'}

sections.Section.default_gettype = dict   # set for all structures
tasks1 = sections('pay bill', 'clean', status=['completed', 'started'])
tasks2 = sections('pay bill', 'clean', status=['completed', 'started'])
assert tasks1('statuses') == {'pay bill': 'completed', 'clean': 'started'}
assert tasks2('statuses') == {'pay bill': 'completed', 'clean': 'started'}

The above will also work for accessing attributes in the form object.attr but only if the node does not contain the attribute attr, otherwise it will return the non-iterable raw value for attr. Therefore, for consistency, access attributes using Section.__call__() like above if you wish always receive an iterable form of the attributes.

Printing

Section structures can be visualized through the Section.deep_str() method as follows:

menu = sections(
    'Breakfast', 'Dinner',
    mains=['Bacon&Eggs', 'Burger'],
    sides=['HashBrown', 'Fries'],
)
print(menu.deep_str())

Output:

###############################################################################
<class 'sections.Sections.UniqueSection.<locals>.Section'>: root, parent
children                      : ['Breakfast', 'Dinner']
name                          : 'section'
<class 'sections.Sections.UniqueSection.<locals>.Section'>: child, leaf
name                          : 'Breakfast'
parent                        : 'section'
mains                         : 'Bacon&Eggs'
sides                         : 'HashBrown'
<class 'sections.Sections.UniqueSection.<locals>.Section'>: child, leaf
name                          : 'Dinner'
parent                        : 'section'
mains                         : 'Burger'
sides                         : 'Fries'
###############################################################################

See the References section of the docs for more printing options: https://sections.readthedocs.io/.

Subclassing

Inheriting Section is easy, the only requirement is to call super().__init__(**kwds) at some point in __init__ like below if you override that method:

class Library(sections.Section):
    def __init__(price="Custom default value", **kwds):
        super().__init__(**kwds)

    @property
    def genres(self):
        if self.isroot:
            return self.sections
        else:
            raise AttributeError('This library has only 1 level of genres')

    @property
    def books(self): return self.leaves

    @property
    def titles(self): return self.names

    def critique(self, impression="Haven't read it yet", rating=0):
        self.review = impression
        self.price = rating * 2

library = Library(
    [{'Fantasy'}, 'LOTR', 'Harry Potter'],
    [{'Academic'}, 'Advanced Math.', 'Physics for Engineers']
)
assert library.genres.names == ['Fantasy', 'Academic']
assert library.books.titles == [
    'LOTR', 'Harry Potter', 'Advanced Math.', 'Physics for Engineers'
]
library.books['LOTR'].critique(impression='Good but too long', rating=7)
library.books['Harry Potter'].critique(
    impression="I don't like owls", rating=4)
assert library.books['LOTR'].price == 14
assert library.books['Harry Potter'].price == 8
import pytest
with pytest.raises(AttributeError):
    this_should_raise_error = library['Fantasy'].genres

Section.____init__() assigns the kwds values passed to it to the object attributes, and the passed kwds are generated during instantiation by a metaclass.

Performance

Each non-leaf Section node keeps a cache containing quickly readable references to an attribute dict previously parsed from manual traversing through descendant nodes in a previous read. The caches are invalidated accordingly when the tree structure or node attribute values change. The caches allow instant reading of sub-lists/dicts in Θ(1) time and can often make structure attribute reading faster by 5x and even much more. The downside is that it also increases memory usage by roughly 5x as well. This is not a concern on a general-purpose computer for structures containing less than 1000 - 10,000 nodes. For clarity, converting a list with 10,000 elements would create 10,001 nodes (1 root plus 10,000 children). After 1000 - 10,000 nodes, it may be recommended to consider changing the node or structure’s class attribute use_cache to False. This can be done as follows:

sect = sections([[[[[42] * 10] * 10] * 10] * 10])
sect.use_cache = False              # turn off for just the root node
sect.cls.use_cache = False          # turn off for all nodes in `sect`
sections.Section.use_cache = False  # turn off for all structures

The dict option for gettype in the Section.__call__() method is currently slower than the other options. For performance-critical uses, if a dict is required just for visual printing purposes, it is recommended to use the faster ‘full_dict’ option for gettype instead of dict. See the Section.__call__() method in the References section of the docs for more details on the gettype options: https://sections.readthedocs.io/.

Changelog

0.0.0 (2021-06-14)

• First release on PyPI.