Skip to main content

DjangoQL: Advanced search language for Django

Project description

https://github.com/ivelum/djangoql/workflows/Tests/badge.svg

Advanced search language for Django, with auto-completion. Supports logical operators, parenthesis, table joins, and works with any Django model. Tested on Python 2.7, 3.5 - 3.10, Django 1.8 - 4.0. The auto-completion feature has been tested in Chrome, Firefox, Safari, IE9+.

See a video: DjangoQL demo

https://raw.githubusercontent.com/ivelum/djangoql/master/djangoql/static/djangoql/img/completion_example_scaled.png

DjangoQL is used by:

logo1 logo2 logo3 logo4

Is your project using DjangoQL? Please submit a PR and let us know!

Contents

Installation

$ pip install djangoql

Add 'djangoql' to INSTALLED_APPS in your settings.py:

INSTALLED_APPS = [
    ...
    'djangoql',
    ...
]

Add it to your Django admin

Adding DjangoQLSearchMixin your model admin will replace the standard Django search functionality with DjangoQL search. Example:

from django.contrib import admin

from djangoql.admin import DjangoQLSearchMixin

from .models import Book


@admin.register(Book)
class BookAdmin(DjangoQLSearchMixin, admin.ModelAdmin):
    pass

Language reference

DjangoQL is shipped with comprehensive Syntax Help, which can be found in Django admin (see the Syntax Help link in auto-completion popup). Here’s a quick summary:

DjangoQL’s syntax resembles Python’s, with some minor differences. Basically you just reference model fields as you would in Python code, then apply comparison and logical operators and parenthesis. DjangoQL is case-sensitive.

  • model fields: exactly as they are defined in Python code. Access nested properties via ., for example author.last_name;

  • strings must be double-quoted. Single quotes are not supported. To escape a double quote use \";

  • boolean and null values: True, False, None. Please note that they can be combined only with equality operators, so you can write published = False or date_published = None, but published > False will cause an error;

  • logical operators: and, or;

  • comparison operators: =, !=, <, <=, >, >= - work as you expect;

  • string-specific comparison operators: startswith, not startswith, endswith, not endswith - work as you expect. Test whether or not a string contains a substring: ~ and !~ (translated into __icontains under the hood). Example: name endswith "peace" or author.last_name ~ "tolstoy";

  • date-specific comparison operators, compare by date part: ~ and !~. Example: date_published ~ "2021-11" - find books published in Nov, 2021;

  • test a value vs. list: in, not in. Example: pk in (2, 3).

DjangoQL Schema

Schema defines limitations - what you can do with a DjangoQL query. If you don’t specify any schema, DjangoQL will provide a default schema for you. This will walk recursively through all model fields and relations and include everything it finds in the schema, so users would be able to search through everything. Sometimes this is not what you want, either due to DB performance or security concerns. If you’d like to limit search models or fields, you should define a schema. Here’s an example:

class UserQLSchema(DjangoQLSchema):
    exclude = (Book,)
    suggest_options = {
        Group: ['name'],
    }

    def get_fields(self, model):
        if model == Group:
            return ['name']
        return super(UserQLSchema, self).get_fields(model)


@admin.register(User)
class CustomUserAdmin(DjangoQLSearchMixin, UserAdmin):
    djangoql_schema = UserQLSchema

In the example above we created a schema that does 3 things:

  • excludes the Book model from search via exclude option. Instead of exclude you may also use include, which limits a search to listed models only;

  • limits available search fields for Group model to only the name field , in the .get_fields() method;

  • enables completion options for Group names via suggest_options.

An important note about suggest_options: it looks for the choices model field parameter first, and if it’s not specified - it will synchronously pull all values for given model fields, so you should avoid large querysets there. If you’d like to define custom suggestion options, see below.

Custom search fields

Deeper search customization can be achieved with custom search fields. Custom search fields can be used to search by annotations, define custom suggestion options, or define fully custom search logic. In djangoql.schema, DjangoQL defines the following base field classes that you may subclass to define your own behavior:

  • IntField

  • FloatField

  • StrField

  • BoolField

  • DateField

  • DateTimeField

  • RelationField

Here are examples for common use cases:

Search by queryset annotations:

from djangoql.schema import DjangoQLSchema, IntField


class UserQLSchema(DjangoQLSchema):
    def get_fields(self, model):
        fields = super(UserQLSchema, self).get_fields(model)
        if model == User:
            fields += [IntField(name='groups_count')]
        return fields


@admin.register(User)
class CustomUserAdmin(DjangoQLSearchMixin, UserAdmin):
    djangoql_schema = UserQLSchema

    def get_queryset(self, request):
        qs = super(CustomUserAdmin, self).get_queryset(request)
        return qs.annotate(groups_count=Count('groups'))

Let’s take a closer look at what’s happening in the example above. First, we add groups_count annotation to the queryset that is used by Django admin in the CustomUserAdmin.get_queryset() method. It would contain the number of groups a user belongs to. As our queryset now pulls this column, we can filter by it. It just needs to be included in the schema. In UserQLSchema.get_fields() we define a custom integer search field for the User model. Its name should match the name of the column in our queryset.

Custom suggestion options

from djangoql.schema import DjangoQLSchema, StrField


class GroupNameField(StrField):
    model = Group
    name = 'name'
    suggest_options = True

    def get_options(self, search):
        return super(GroupNameField, self)\
            .get_options(search)\
            .annotate(users_count=Count('user'))\
            .order_by('-users_count')


class UserQLSchema(DjangoQLSchema):
    def get_fields(self, model):
        if model == Group:
            return ['id', GroupNameField()]
        return super(UserQLSchema, self).get_fields(model)


@admin.register(User)
class CustomUserAdmin(DjangoQLSearchMixin, UserAdmin):
    djangoql_schema = UserQLSchema

In this example we’ve defined a custom GroupNameField that sorts suggestions for group names by popularity (no. of users in a group) instead of default alphabetical sorting.

Custom search lookup

DjangoQL base fields provide two basic methods that you can override to substitute either search column, search value, or both - .get_lookup_name() and .get_lookup_value(value):

class UserDateJoinedYear(IntField):
    name = 'date_joined_year'

    def get_lookup_name(self):
        return 'date_joined__year'


class UserQLSchema(DjangoQLSchema):
    def get_fields(self, model):
        fields = super(UserQLSchema, self).get_fields(model)
        if model == User:
            fields += [UserDateJoinedYear()]
        return fields


@admin.register(User)
class CustomUserAdmin(DjangoQLSearchMixin, UserAdmin):
    djangoql_schema = UserQLSchema

In this example we’ve defined the custom date_joined_year search field for users, and used the built-in Django __year filter option in .get_lookup_name() to filter by date year only. Similarly you can use .get_lookup_value(value) hook to modify a search value before it’s used in the filter.

Fully custom search lookup

.get_lookup_name() and .get_lookup_value(value) hooks cover many simple use cases, but sometimes they’re not enough and you want a fully custom search logic. In such cases you can override main .get_lookup() method of a field. Example below demonstrates User age search:

from djangoql.schema import DjangoQLSchema, IntField


class UserAgeField(IntField):
    """
    Search by given number of full years
    """
    model = User
    name = 'age'

    def get_lookup_name(self):
        """
        We'll be doing comparisons vs. this model field
        """
        return 'date_joined'

    def get_lookup(self, path, operator, value):
        """
        The lookup should support with all operators compatible with IntField
        """
        if operator == 'in':
            result = None
            for year in value:
                condition = self.get_lookup(path, '=', year)
                result = condition if result is None else result | condition
            return result
        elif operator == 'not in':
            result = None
            for year in value:
                condition = self.get_lookup(path, '!=', year)
                result = condition if result is None else result & condition
            return result

        value = self.get_lookup_value(value)
        search_field = '__'.join(path + [self.get_lookup_name()])
        year_start = self.years_ago(value + 1)
        year_end = self.years_ago(value)
        if operator == '=':
            return (
                Q(**{'%s__gt' % search_field: year_start}) &
                Q(**{'%s__lte' % search_field: year_end})
            )
        elif operator == '!=':
            return (
                Q(**{'%s__lte' % search_field: year_start}) |
                Q(**{'%s__gt' % search_field: year_end})
            )
        elif operator == '>':
            return Q(**{'%s__lt' % search_field: year_start})
        elif operator == '>=':
            return Q(**{'%s__lte' % search_field: year_end})
        elif operator == '<':
            return Q(**{'%s__gt' % search_field: year_end})
        elif operator == '<=':
            return Q(**{'%s__gte' % search_field: year_start})

    def years_ago(self, n):
        timestamp = now()
        try:
            return timestamp.replace(year=timestamp.year - n)
        except ValueError:
            # February 29
            return timestamp.replace(month=2, day=28, year=timestamp.year - n)


class UserQLSchema(DjangoQLSchema):
    def get_fields(self, model):
        fields = super(UserQLSchema, self).get_fields(model)
        if model == User:
            fields += [UserAgeField()]
        return fields


@admin.register(User)
class CustomUserAdmin(DjangoQLSearchMixin, UserAdmin):
    djangoql_schema = UserQLSchema

Can I use it outside of Django admin?

Sure. You can add DjangoQL search functionality to any Django model using DjangoQLQuerySet:

from django.db import models

from djangoql.queryset import DjangoQLQuerySet


class Book(models.Model):
    name = models.CharField(max_length=255)
    author = models.ForeignKey('auth.User')

    objects = DjangoQLQuerySet.as_manager()

With the example above you can perform a search like this:

qs = Book.objects.djangoql(
    'name ~ "war" and author.last_name = "Tolstoy"'
)

It returns a normal queryset, so you can extend it and reuse if necessary. The following code works fine:

print(qs.count())

Alternatively you can add DjangoQL search to any existing queryset, even if it’s not an instance of DjangoQLQuerySet:

from django.contrib.auth.models import User

from djangoql.queryset import apply_search

qs = User.objects.all()
qs = apply_search(qs, 'groups = None')
print(qs.exists())

Schemas can be specified either as a queryset option, or passed to .djangoql() queryset method directly:

class BookQuerySet(DjangoQLQuerySet):
    djangoql_schema = BookSchema


class Book(models.Model):
    ...

    objects = BookQuerySet.as_manager()

# Now, Book.objects.djangoql() will use BookSchema by default:
Book.objects.djangoql('name ~ "Peace")  # uses BookSchema

# Overriding default queryset schema with AnotherSchema:
Book.objects.djangoql('name ~ "Peace", schema=AnotherSchema)

You can also provide schema as an option for apply_search()

qs = User.objects.all()
qs = apply_search(qs, 'groups = None', schema=CustomSchema)

Using completion widget outside of Django admin

The completion widget is not tightly coupled to Django admin, so you can easily use it outside of the admin if you want. The widget is available on npm as a standalone package. See the source code and the docs in the djangoql-completion repo on GitHub.

The completion widget is also bundled with the djangoql Python package on PyPI. If you’re not using Webpack or another JavaScript bundler, you can use the pre-built version that ships with the Python package. Here is an example:

Template code, completion_demo.html:

{% load static %}
<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8">
  <title>DjangoQL completion demo</title>
  <link rel="stylesheet" type="text/css" href="{% static 'djangoql/css/completion.css' %}" />
  <script src="{% static 'djangoql/js/completion.js' %}"></script>
</head>
<body>

  <form action="" method="get">
    <p style="color: red">{{ error }}</p>
    <textarea name="q" cols="40" rows="1" autofocus>{{ q }}</textarea>
  </form>

  <ul>
  {% for item in search_results %}
    <li>{{ item }}</li>
  {% endfor %}
  </ul>

  <script>
    DjangoQL.DOMReady(function () {
      new DjangoQL({
        // either JS object with a result of DjangoQLSchema(MyModel).as_dict(),
        // or an URL from which this information could be loaded asynchronously
        introspections: {{ introspections|safe }},

        // css selector for query input or HTMLElement object.
        // It should be a textarea
        selector: 'textarea[name=q]',

        // optional, you can provide URL for Syntax Help link here.
        // If not specified, Syntax Help link will be hidden.
        syntaxHelp: null,

        // optional, enable textarea auto-resize feature. If enabled,
        // textarea will automatically grow its height when entered text
        // doesn't fit, and shrink back when text is removed. The purpose
        // of this is to see full search query without scrolling, could be
        // helpful for really long queries.
        autoResize: true
      });
    });
  </script>
</body>
</html>

And in your views.py:

import json

from django.contrib.auth.models import Group, User
from django.shortcuts import render_to_response
from django.views.decorators.http import require_GET

from djangoql.exceptions import DjangoQLError
from djangoql.queryset import apply_search
from djangoql.schema import DjangoQLSchema
from djangoql.serializers import DjangoQLSchemaSerializer


class UserQLSchema(DjangoQLSchema):
    include = (User, Group)
    suggest_options = {
        Group: ['name'],
    }


@require_GET
def completion_demo(request):
    q = request.GET.get('q', '')
    error = ''
    query = User.objects.all().order_by('username')
    if q:
        try:
            query = apply_search(query, q, schema=UserQLSchema)
        except DjangoQLError as e:
            query = query.none()
            error = str(e)
    # You may want to use SuggestionsAPISerializer and an additional API
    # endpoint (see in djangoql.views) for asynchronous suggestions loading
    introspections = DjangoQLSchemaSerializer().serialize(
      UserQLSchema(query.model),
    )
    return render_to_response('completion_demo.html', {
        'q': q,
        'error': error,
        'search_results': query,
        'introspections': json.dumps(introspections),
    })

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

djangoql-0.17.0.tar.gz (228.1 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page