Python iterables的一个轻量级查询引擎：lifter

jopen 9年前

lifter

Python iterables的一个轻量级查询引擎，受Django ORM启发。

  • Free software: BSD license

Warning: This package is still in alpha state and a lot of work is still needed to make queries faster and efficient. Contributions are welcome :)

The query api will probably be under a lot of changes soon. If you want to be part of the discussion and contribute, please refer to the corresponding issue.

Contents

Features

  • Operates on plain objects or mapping (such as dictionaries)
  • API similar to Django querysets
  • Lightweight: absolutely no dependencies
  • Tested and working on Python 2.7 to Python 3.5

Why lifter?

Because making queries against regular Python iterables (lists, tuples, etc.) is painful.

Consider the following list of users:

users = [      {          "id": "56c4e39a3e05a86e9f759ba8",          "is_active": True,          "number_of_children": 0,          "age": 35,          "eye_color": "brown",          "name": "Bernard",          "gender": "male",          "has_beard": False,          "email": "bernard@blackbooks.com",          "own_bookshop": True,          "company": {              "name": "blackbooks"          }      },      {          "id": "56c4e39a3d08fcd553cb73b9",          "is_active": True,          "number_of_children": 0,          "age": 34,          "eye_color": "brown",          "name": "Manny",          "gender": "male",          "has_beard": True,          "email": "manny@blackbooks.com",          "own_bookshop": False,          "company": {              "name": "blackbooks"          }      },      {          "id": "56c4e39a1f9b6f64db8a1b98",          "is_active": True,          "number_of_children": 0,          "age": 35,          "eye_color": "brown",          "name": "Fran",          "gender": "female",          "has_beard": False,          "email": "fran@blackbooks.com",          "own_bookshop": False,          "company": {              "name": "blackbooks"          }      },      # And so on ...  ]

Now, imagine you have to extract data from this list:

  • Getting all active 26 years old users
  • Getting names and emails of inactive users
  • Getting all active users except the one with brown eyes and sort them by age
  • Getting average and minimum age of women
  • etc.

Of course it's totally doable in plain python. A for loop, some if statements, maybe a list comprehension and you're done:

# Getting all active 26 years old users  under_26 = [      user for user in users      if user['age'] == 26 and user['is_active']  ]    # Getting names and emails of inactive users  inactive_mail_and_names = [      (user['name'], user['email']) for user in users      if not user['is_active']  ]    # Getting all active users except the one with brown eyes and sort them by age  active_without_brown_eyes = [      user for user in users      if user['is_active'] and not user['eye_color'] == 'brown'  ]  active_without_brown_eyes_sorted = sorted(active_without_brown_eyes, key=lambda v: v['age'])    # minimum and average women age  from statistics import mean  women_ages = [      user['age'] for user in users      if user['gender'] == 'female'  ]  women_average_age = mean(women_ages)  minimum_woman_age = min(women_ages)

But, as you can see, plain Python code is quite redundant and not especially readable. It would be even longer without list comprehensions.

Let's see if we can do better using lifter:

import lifter    # We load all users into lifter so we can run queries  manager = lifter.load(users)    # Getting all active 26 years old users  under_26 = manager.filter(age=26, is_active=True)    # Getting names and emails of inactive users  inactive_mail_and_names = manager.filter(is_active=False).values_list('name', 'email')    # Getting all active users except the one with brown eyes and sort them by age  active_without_brown_eyes_sorted = (      manager      .filter(is_active=True)      .exclude(eye_color='brown')      .order_by('age')  )    # average women age  women_average_age, women_min_age = (      manager.filter(gender='female')              .aggregate(lifter.Avg('age'), lifter.Min('age'), flat=True)  )

Better, isn't it?

Installation

At the command line:

$ pip install lifter

Usage

To use lifter in a project:

import lifter

Before running queries, you need to load your data inside a manager:

manager = lifter.load(my_iterable)

If you want to use the same data as the examples provided in this documentation, copy-paste the content of tests/fake_data.py inside your python interpreter then run:

manager = lifter.load(fake)

Note

All examples use a list of dictionaries as source data, but lifter works exactly the same if you feed your manager with a list of regular objects. Lifter will seamlessly lookup both object attributes and dictionary keys.

About querysets

Just like Django, lifter is based on querysets. Basically, a queryset in lifter is an object containing values with functions to refine these values.

You can chain most queryset methods, which wil give you enough flexibility to build complex queries:

results = manager.all().filter(is_active=True).exclude(age=42).order_by('age')    for result in results:      # do something with the results

Note

Unless stated otherwise, all queryset methods behave just like Django querysets

Warning

At the moment, lifter querysets are not lazy, which mean they are applied immediately when called.

filter

One of the most basic query method is filter. Use it if you want to retrieve objects that match a set of criteria. Example:

manager.filter(name='Manny')

The previous example will return a QuerySet instance containing all users whose name equals Manny. It's absolutely okay to provide multiple arguments at once:

# these two queries have the same effect  manager.filter(name='Manny', has_beard=True)  manager.filter(name='Manny').filter(has_beard=True)

This time, we'll only get users named Manny AND with a beard.

get

get returns a single object that match a set of criteria, raising an exception if no value is found or if multiple values are found:

manager.get(name='Fran', gender='female')

You can catch these exceptions as follow:

try:      manager.get(name='Hodor')  except lifter.DoesNotExist:      print('Wrong show dude')    try:      manager.get(gender='male')  except lifter.MultipleObjectsReturned:      print('Bernard or Manny, you have to choose')

And, finally, you can chain get after other querysets to reduce available choices:

# the following will look for a single male among users without beard  manager.filter(has_beard=False).get(gender='male')

exclude

This method is the exact opposite of filter. Use it if you want to retrieve objects that do not match a set of criteria. Example:

manager.exclude(name='Bernard')

The previous example will return a QuerySet instance containing all users not named Bernard. Contrary to filter, providing multiple arguments at once and chaining do not achieve the same thing:

# This will exclude only objects with name == 'Bernard' AND own_bookshop == True  manager.exclude(name='Bernard', own_bookshop=True)    # This will exclude objects with name == 'Bernard' OR own_bookshop == True  manager.exclude(name='Bernard').exclude(own_bookshop=True)

order_by

Note

By default, order of provided data is preserved accross all subsequent querysets, unless you explicitly call order_by at some point.

Use this method to change results' order based on a given attribute:

# will return younger users first  manager.all().order_by('age')

You can prefix the attribute with - to reverse the ordering:

# will return older users first  manager.all().order_by('-age')
# will return random ordered queryset  manager.all().order_by('?')

count

A simple method that returns the number of object inside the queryset:

manager.filter(has_beard=False).count()

exists

A simple method that return True if a queryset contains at least one result, returning False otherwise:

# return True  manager.filter(has_beard=False).exists()

first

A shortcut that returns the first result or None if the query has no results:

manager.all().first()

last

Same as first, but return the last result.

values

Use values if you don't want to access original objects but only a subset of specific values:

# will return a list of dictionaries as follow:  # [  #     {'name': 'Bernard', 'email': 'bernard@blackbooks.com'},  #     {'name': 'Manny', 'email': 'manny@blackbooks.com'},  # ]  manager.all().values('name', 'email')

values_list

This method behaves as values, but return a list of tuples instead of a list of dictionaries:

# will return a list of tuples as follow:  # [  #     ('Bernard', 'bernard@blackbooks.com')  #     ('Manny', 'manny@blackbooks.com')  # ]  manager.all().values_list('name', 'email')

Additionaly, if you only want a single value without nested tuples, you can provide the optional flat parameter:

# will return a list as follow:  # ['Bernard', 'Manny']  manager.all().values_list('name', flat=True)

distinct

distinct remove duplicate entries in a queryset:

# will return ['blue', 'brown', 'green', 'purple']  manager.order_by('eye_color').values_list('eye_color', flat=True).distinct()

Spanning lookups

If you want to access attributes from nested objects, you can use the following lookup syntax:

# will filter users with a company whose name is "blackbooks"  manager.filter(company__name='blackbooks')    # return a list of all companies names, without duplicates  manager.values_list('company__name', flat=True).distinct()

Complex lookups

Most of the time, simple lookups using equality in filter/exclude clauses will be enough. If it's not the case, you can user built-in lookups to build more complex queries:

# return all users older than 37  manager.filter(age=lifter.gt(37))    # exclude all users under 43  manager.exclude(age=lifter.lt(43))    # return all users between 21 and 27 years old  manager.exclude(age=lifter.value_range(21, 27))    # return users with brown or green eyes  manager.filter(eye_color=lifter.value_in(['brown', 'green']))

Finally, if you need a lookup that is not provided, you can provide a callable to filter and exclude:

# Leave only users whose age is odd  manager.exclude(age=lambda v: v % 2 == 0)

Note that such callables must return a boolean.

Available lookups:

  • gt: greater than
  • gte: greater than or equal
  • lt: less than
  • lte: less than or equal
  • startswith: case sensitive startswith
  • istartswith: case insensitive startswith
  • endswith: case sensitive endswith
  • iendswith: case insensitive endswith
  • contains: case sensitive search
  • icontains: case insensitive search
  • value_in: value is present in given iterable
  • value_range: value is between given range

Aggregation

If you want to extract global data instead of returning results, you can use aggregation:

# return the total number of children of all users combined, like this:  # {'number_of_children__sum': 267}    manager.all().aggregate(lifter.Sum('number_of_children'))

You can bind the aggregate to a custom key:

# {'children': 267}  manager.all().aggregate(children=lifter.Sum('number_of_children'))

Additionaly, you can return multiple aggregates at once:

manager.all().aggregate(lifter.Sum('number_of_children'), lifter.Avg('age'))

If you would rather have a flat list of values returned, use the flat keyword:

# [267]    manager.all().aggregate(children=lifter.Sum('number_of_children'), flat=True)

Available lookups are:

  • Sum: sums the values of the given field
  • Min: return the lowest value
  • Max: return the greatest value
  • Avg: return the average value

Contributing

Bug reports, feature requests and pull requests, are welcome, but before sumitting anything, please read CONTRIBUTING.rst.

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.

项目地址： https://github.com/EliotBerriot/lifter