django-ajax-datatable is a Django app (previously named morlandi/django-datatables-view) which provides advanced integration for a Django project with the jQuery Javascript library DataTables.net, when used in server-side processing mode.

In this context, the rendering of the table is the result of a serie of Ajax requests to the server following user interactions (i.e. when paging, ordering, searching, etc.).

With django-ajax-datatable, basically you have to provide a AjaxDatatableView-derived view to describe the desired table content and behaviour, and the app manages the interaction with DataTables.net by reacting to the ajax requests with suitable responses.

Notes:

Since someone asked ...

• I use this app for my own projects, and improve it from time to time as new needs arises.
• I received so much from the Django community, so I'm more than happy to share something hopefully useful for others. The app is intended to be opensource; feel free to use it we no restrictions at all. I added a MIT Licence file to the github repo, to make this more explicit.
• Since v4.0.0, the package has been renamed from django-datatables-view to django-ajax-datatable to avoid a conflict on PyPI
• Unfortunately I only have a few unit tests, and didn't bother (yet) to add a TOX procedure to run then with different Python/Django versions. Having said this, I can confirm that I do happen to use it with no problems in projects based on Django 2.x. However, most recent improvements have been tested mainly with Django 3. As far as I know, no Django3-specific features have been applied. In case, please open an issue, and I will fix it.
• I'm not willing to support Python 2.x and Django 1.x any more; in case, use a previous release (tagged as v2.x.x); old releases will be in place in the repo forever

Features:

• Pagination
• Column ordering
• Global generic search
• Global date-range search over "get_latest_by" column
• Column specific filtering
• Foreign key fields can be used, using the "model1__model2__field" notation
• Customizable rendering of table rows
• and more ...

Inspired from:

https://github.com/monnierj/django-datatables-server-side

A very minimal working Django project which uses django-ajax-datatable can be found in the folder example_minimal.

A more realistic solution, with a frontend based on Bootstrap4, can be found in example, and is published as a demo site at the address: http://django-ajax-datatable-demo.brainstorm.it/.

Install the package by running:

pip install django-ajax-datatable

or:

pip install git+https://github.com/morlandi/[email protected]

then add 'ajax_datatable' to your INSTALLED_APPS:

INSTALLED_APPS = [
    ...
    'ajax_datatable',
]

Optional dependencies (for better debug tracing):

• sqlparse
• termcolor
• pygments

Your base template should include what required by datatables.net, plus:

• /static/ajax_datatable/css/style.css
• /static/ajax_datatable/js/utils.js

Example (plain jQuery from CDN):

{% block extrastyle %}

    <link href="{% static 'ajax_datatable/css/style.css' %}" rel="stylesheet" />
    <link href="//cdn.datatables.net/1.10.22/css/jquery.dataTables.min.css" />

{% endblock extrastyle %}

{% block extrajs %}

    <script src="https://code.jquery.com/jquery-3.5.1.min.js"></script>
    <script type="text/javascript" src="{% static 'ajax_datatable/js/utils.js' %}"></script>
    <script src="//cdn.datatables.net/1.10.22/js/jquery.dataTables.min.js"></script>

{% endcompress %}

Example (with Bootstrap4 support):

{% block extrastyle %}

    <link href="{% static 'ajax_datatable/css/style.css' %}" rel="stylesheet" />
    <!-- link rel='stylesheet' href="{% static 'datatables.net-bs/css/dataTables.bootstrap.min.css' %}" -->
    <link rel='stylesheet' href="{% static 'datatables.net-bs4/css/dataTables.bootstrap4.min.css' %}">
    <link rel='stylesheet' href="{% static 'datatables.net-buttons-bs/css/buttons.bootstrap.min.css' %}">

{% endblock extrastyle %}

{% block extrajs %}

    <script type="text/javascript" src="{% static 'ajax_datatable/js/utils.js' %}"></script>

    <script src="{% static 'datatables.net/js/jquery.dataTables.min.js' %}"></script>
    <!-- script src="{% static 'datatables.net-bs/js/dataTables.bootstrap.min.js' %}"></script -->
    <script src="{% static 'datatables.net-bs4/js/dataTables.bootstrap4.min.js' %}"></script>
    <script src="{% static 'datatables.net-buttons/js/dataTables.buttons.min.js' %}"></script>
    <script src="{% static 'datatables.net-buttons/js/buttons.print.min.js' %}"></script>
    <script src="{% static 'datatables.net-buttons/js/buttons.html5.min.js' %}"></script>
    <script src="{% static 'datatables.net-buttons-bs/js/buttons.bootstrap.min.js' %}"></script>
    <script src="{% static 'jszip/dist/jszip.min.js' %}"></script>
    <script src="{% static 'pdfmake/build/pdfmake.min.js' %}"></script>
    <script src="{% static 'pdfmake/build/vfs_fonts.js' %}"></script>

{% endcompress %}

To provide server-side rendering of a Django Model, you will need:

1. an ordinary view which will render an HTML page containing:

   • an empty HTML <table> element
   • a javascript code which links this HTML table to the (AjaxDatatableView-derived) second view

2. a specific view derived from AjaxDatatableView() which will be called multiple times via Ajax during data navigation; this second view has two duties:

   • render the initial table layout based on specified columns
   • respond to datatables.net requests, as a consequence of the user interaction with the table

Example:

We start by rendering an HTML page from this template:

file permissions_list.html

<table id="datatable_permissions">
</table>

or:

<div class="table-responsive">
    <table id="datatable_permissions" width="100%" class="table table-striped table-bordered dt-responsive compact nowrap">
    </table>
</div>

...

<script language="javascript">

    $(document).ready(function() {
        AjaxDatatableViewUtils.initialize_table(
            $('#datatable_permissions'),
            "{% url 'ajax_datatable_permissions' %}",
            {
                // extra_options (example)
                processing: false,
                autoWidth: false,
                full_row_select: true,
                scrollX: false
            }, {
                // extra_data
                // ...
            },
        );
    });

</script>

Here, "{% url 'ajax_datatable_permissions' %}" is the endpoint to the specialized view:

file urls.py

from django.urls import path
from . import ajax_datatable_views

app_name = 'frontend'

urlpatterns = [
    ...
    path('ajax_datatable/permissions/', ajax_datatable_views.PermissionAjaxDatatableView.as_view(), name="ajax_datatable_permissions"),
]

The javascript helper AjaxDatatableViewUtils.initialize_table(element, url, extra_options={}, extra_data={}) connects the HTML table element to the "server-size table rendering" machinery, and performs a first call (identified by the action=initialize parameter) to the AjaxDatatableView-derived view.

This in turn populates the HTML empty table with a suitable layout, while subsequent calls to the view will be performed to populate the table with real data.

This strategy allows the placement of one or more dynamic tables in the same page.

I often keep all AjaxDatatableView-derived views in a separate "ajax_datatable_views.py" source file, to make it crystal clear that you should never call them directly:

file ajax_datatable_views.py

from ajax_datatable.views import AjaxDatatableView
from django.contrib.auth.models import Permission


class PermissionAjaxDatatableView(AjaxDatatableView):

    model = Permission
    title = 'Permissions'
    initial_order = [["app_label", "asc"], ]
    length_menu = [[10, 20, 50, 100, -1], [10, 20, 50, 100, 'all']]
    search_values_separator = '+'

    column_defs = [
        AjaxDatatableView.render_row_tools_column_def(),
        {'name': 'id', 'visible': False, },
        {'name': 'codename', 'visible': True, },
        {'name': 'name', 'visible': True, },
        {'name': 'app_label', 'foreign_field': 'content_type__app_label', 'visible': True, },
        {'name': 'model', 'foreign_field': 'content_type__model', 'visible': True, },
    ]

In the previous example, row id is included in the first column of the table, but hidden to the user.

AjaxDatatableView will serialize the required data during table navigation.

This is the resulting table:

You can use common CSS style to customize the final rendering:

AjaxDatatableViewUtils.initialize_table() parameters are:

element

table element

url

action (remote url to be called via Ajax)

extra_options={}

custom options for dataTable()

extra_data={}

extra parameters to be sent via ajax for global "initial queryset" filtering; see: Provide "extra data" to narrow down the initial queryset

Required:

• model
• column_defs

Optional:

• initial_order = [[1, "asc"], [5, "desc"]] # positions can also be expressed as column names: [['surname', 'asc'], ]
• length_menu = [[10, 20, 50, 100], [10, 20, 50, 100]]
• latest_by = None
• show_date_filters = None
• show_column_filters = None
• disable_queryset_optimization = False
• disable_queryset_optimization_only = False
• disable_queryset_optimization_select_related = False
• disable_queryset_optimization_prefetch_related = False
• table_row_id_prefix = 'row-'
• table_row_id_fieldname = 'id'
• render_row_details_template_name = "render_row_details.html"
• search_values_separator = ''
• sort_field: None

or override the following methods to provide attribute values at run-time, based on request:

def get_column_defs(self):
    return self.column_defs

def get_initial_order(self):
    return self.initial_order

def get_length_menu(self):
    return self.length_menu

def get_latest_by(self, request):
    """
    Override to customize based on request.

    Provides the name of the column to be used for global date range filtering.
    Return either '', a fieldname or None.

    When None is returned, in model's Meta 'get_latest_by' attributed will be used.
    """
    return self.latest_by

def get_show_date_filters(self, request):
    """
    Override to customize based on request.

    Defines whether to use the global date range filter.
    Return either True, False or None.

    When None is returned, will'll check whether 'latest_by' is defined
    """
    return self.show_date_filters

def get_show_column_filters(self, request):
    """
    Override to customize based on request.

    Defines whether to use the column filters.
    Return either True, False or None.

    When None is returned, check if at least one visible column in searchable.
    """
    return self.show_column_filters

def get_table_row_id(self, request, obj):
    """
    Provides a specific ID for the table row; default: "row-ID"
    Override to customize as required.
    """
    result = ''
    if self.table_row_id_fieldname:
        try:
            result = self.table_row_id_prefix + str(getattr(obj, self.table_row_id_fieldname))
        except:
            result = ''
    return result

Example:

column_defs = [{
    'name': 'currency',                 # required
    'data': None,
    'title': 'Currency',                # optional: default = field verbose_name or column name
    'visible': True,                    # optional: default = True
    'searchable': True,                 # optional: default = True if visible, False otherwise
    'orderable': True,                  # optional: default = True if visible, False otherwise
    'foreign_field': 'manager__name',   # optional: follow relation
    'm2m_foreign_field': 'manager__name',   # optional: follow m2m relation
    'placeholder': False,               # ???
    'className': 'css-class-currency',  # optional class name for cell
    'defaultContent': '<h1>test</h1>',  # ???
    'width': 300,                       # optional: controls the minimum with of each single column
    'choices': None,                    # see `Filtering single columns` below
    'initialSearchValue': None,         # see `Filtering single columns` below
    'autofilter': False,                # see `Filtering single columns` below
    'boolean': False,                   # treat calculated column as BooleanField
    'max_length': 0,                    # if > 0, clip result longer then max_length
    'lookup_field': '__icontains',      # used for searches; default: '__iexact' for columns with choices, '__icontains' in all other cases
}, {
    ...

Notes:

• title: if not supplied, the verbose name of the model column (when available) or name will be used
• width: for this to be effective, you need to add table-layout: fixed; style to the HTML table, but in some situations this causes problems in the computation of the table columns' widths (at least in the current version 1.10.19 of Datatables.net)

Sometimes you might need to restrict the initial queryset based on the context.

To that purpose, you can provide a dictionary of additional filters during table initialization; this dictionary will be sent to the View, where you can use it for queryset filtering.

Provide as many key as required; assign either constant values or callables. The special keys 'date_from' and 'date_to' may be used to override values collected by the optional global date range filter (format: 'YYYY-MM-DD').

Example:

AjaxDatatableViewUtils.initialize_table(
    element,
    url,
    {
        // extra_options (example)
        processing: false,
        autoWidth: false,
        full_row_select: false,
        scrollX: true,
        bFilter: false
    }, {
        // extra_data
        client_id: '{{client.id}}',
        date_from: function() { return date_input_to_isoformat('#date_from'); },
        date_to: function() { return date_input_to_isoformat('#date_to'); }
    }
);

then:

class SampleAjaxDatatableView(AjaxDatatableView):

    ...

    def get_initial_queryset(self, request=None):

        if not request.user.is_authenticated:
            raise PermissionDenied

        # We accept either GET or POST
        if not getattr(request, 'REQUEST', None):
            request.REQUEST = request.GET if request.method=='GET' else request.POST

        queryset = self.model.objects.all()

        if 'client_id' in request.REQUEST:
            client_id = int(request.REQUEST.get('client_id'))
            queryset = queryset.filter(client_id=client_id)

        return queryset

Sometimes you need to provide complex or very specific filters to let the user control the content of the table in an advanced manner.

In those cases, the global or column filters provided by AjaxDatatableView, which are based on simple <input> and <select> widgets, may not be enought.

Still, you can easily add a sidebar with custom filters, and apply to them the concepts explained in the previous paragraph (Provide "extra data" to narrow down the initial queryset).

An example of this technique has been added to the Example project; the result and a detailed explanation is presented here:

http://django-ajax-datatable-demo.brainstorm.it/side_filters/

Starting from v3.2.0, each table row is characterized with a specific ID on each row (tipically, the primary key value from the queryset)

The default behaviour is to provide the string "row-ID", where:

• "row-" is retrieved from self.table_row_id_prefix
• "ID" is retrieved from the row object, using the field with name self.table_row_id_fieldname (default: "id")

Note that, for this to work, you are required to list the field "id" in the column list (maybe hidden).

This default behaviour can be customized by either:

• replacing the values for table_row_id_fieldname and/or table_row_id_prefix, or
• overriding def get_table_row_id(self, request, obj)

Sorting is managed the by the overridable method sort_queryset(), and fully delegated to the database for better performances.

For each orderable column, the column name will be used, unless a sort_field has been specified; in which case, the latter will be used instead.

DatatableView.show_column_filters (or DatatableView.get_show_column_filters(request)) defines whether to show specific filters for searchable columns as follows:

• None (default): show if at least one visible column in searchable
• True: always show
• False: always hide

By default, a column filter for a searchable column is rendered as a text input box; you can instead provide a select box using the following attributes:

choices

• None (default) or False: no choices (use text input box)

• True: use Model's field choices;

  • failing that, we might use "autofilter"; that is: collect the list of distinct values from db table
  • or, for BooleanField columns, provide (None)/Yes/No choice sequence
  • calculated columns with attribute 'boolean'=True are treated as BooleanFields

• ((key1, value1), (key2, values), ...) : use supplied sequence of choices

autofilter

• default = False
• when set: if choices == True and no Model's field choices are available, collects distinct values from db table (much like Excel "autofilter" feature)

For the first rendering of the table:

initialSearchValue

• optional initial value for column filter

Note that initialSearchValue can be a value or a callable object. If callable it will be called every time a new object is created.

For example:

class MyAjaxDatatableView(AjaxDatatableView):

    def today():
        return datetime.datetime.now().date()

    ...

    column_defs = [
        ...
        {
            'name': 'created',
            'choices': True,
            'autofilter': True,
            'initialSearchValue': today
        },
        ...
    ]

Searching on multiple values can be obtained by assigning a "search value separator" as in the following example:

search_values_separator = '+'

In this case, if the user inputs "aaa + bbb", the following search will be issued:

Q("aaa") | Q("bbb")

This works for text search on both global and columns filters.

TODO: test with dates, choices and autofilter.

You can insert placeholder columns in the table, and feed their content with arbitrary HTML.

Example:

@method_decorator(login_required, name='dispatch')
class RegisterAjaxDatatableView(AjaxDatatableView):

    model = Register
    title = _('Registers')

    column_defs = [
        {
            'name': 'id',
            'visible': False,
        }, {
            'name': 'created',
        }, {
            'name': 'dow',
            'title': 'Day of week',
            'placeholder': True,
            'searchable': False,
            'orderable': False,
            'className': 'highlighted',
        }, {
            ...
        }
    ]

    def customize_row(self, row, obj):
        days = ['monday', 'tuesday', 'wednesday', 'thyrsday', 'friday', 'saturday', 'sunday']
        if obj.created is not None:
            row['dow'] = '<b>%s</b>' % days[obj.created.weekday()]
        else:
            row['dow'] = ''
        return

Sometimes you might want to clip results up to a given maximum length, to control the column width.

This can be obtained by specifying a positive value for the max_length column_spec attribute.

Results will be clipped in both the column cells and in the column filter.

Clipped results are rendered as html text as follows:

def render_clip_value_as_html(self, long_text, short_text, is_clipped):
    """
    Given long and shor version of text, the following html representation:
        <span title="long_text">short_text[ellipsis]</span>

    To be overridden for further customisations.
    """
    return '<span title="{long_text}">{short_text}{ellipsis}</span>'.format(
        long_text=long_text,
        short_text=short_text,
        ellipsis='&hellip;' if is_clipped else ''
    )

You can customise the rendering by overriding render_clip_value_as_html()

The following table events are broadcasted to your custom handlers, provided you subscribe them:

• initComplete(table)
• drawCallback(table, settings)
• rowCallback(table, row, data)
• footerCallback(table, row, data, start, end, display)

Please note the the first parameter of the callback is always the event, and next parameters are additional data:

.trigger('foo', [1, 2]);

.on('foo', function(event, one, two) { ... });

More events triggers sent directly by DataTables.net are listed here:

https://datatables.net/reference/event/

Example:

<div class="table-responsive">
    <table id="datatable" width="100%" class="table table-striped table-bordered dataTables-log">
    </table>
</div>

<script language="javascript">
    $(document).ready(function() {

        // Subscribe "rowCallback" event
        $('#datatable').on('rowCallback', function(event, table, row, data ) {
            //$(e.target).show();
            console.log('rowCallback(): table=%o', table);
            console.log('rowCallback(): row=%o', row);
            console.log('rowCallback(): data=%o', data);
        });

        // Initialize table
        AjaxDatatableViewUtils.initialize_table(
            $('#datatable'),
            "{% url 'frontend:object-datatable' model|app_label model|model_name %}",
            extra_option={},
            extra_data={}
        );
    });
</script>

Provides the queryset to work with; defaults to self.model.objects.all()

Example:

def get_initial_queryset(self, request=None):
    if not request.user.view_all_clients:
        queryset = request.user.related_clients.all()
    else:
        queryset = super().get_initial_queryset(request)
    return queryset

When collecting data for autofiltering in a "foreign_field" column, we need some data source for doing the lookup.

The default implementation is as follows:

def get_foreign_queryset(self, request, field):
    queryset = field.model.objects.all()
    return queryset

You can override it for further reducing the resulting list.

Called every time a new data row is required by the client, to let you further customize cell content

Example:

def customize_row(self, row, obj):
    # 'row' is a dictionary representing the current row, and 'obj' is the current object.
    row['code'] = '<a class="client-status client-status-%s" href="%s">%s</a>' % (
        obj.status,
        reverse('frontend:client-detail', args=(obj.id,)),
        obj.code
    )
    if obj.recipe is not None:
        row['recipe'] = obj.recipe.display_as_tile() + ' ' + str(obj.recipe)
    return

Renders an HTML fragment to show table row content in "detailed view" fashion, as previously explained later in the Add row tools as first column section. Having "pk" in your column_defs list is needed to have the script get the object to render.

See also: row details customization

Example:

def render_row_details(self, pk, request=None):
    client = self.model.objects.get(pk=pk)
    ...
    return render_to_string('frontend/pages/includes/client_row_details.html', {
        'client': client,
        ...
    })

OR you can have your own callback called instead (thanks to PetrDlouhy):

AjaxDatatableViewUtils.initialize_table(
    element,
    url,
    {
        // extra_options
        ...
        detail_callback: function(data, tr) {
            console.log('tr: %o', tr);
            console.log('data: %o', data);

            // for example: open a Bootstrap3 modal
            $('.modal-body').html(data, 'details');
            $('.modal').modal();
        }
    }, {
        // extra_data
        ...
    },
);

You can annotate the table footer with a custom message by overridding the following View method.

def footer_message(self, qs, params):
    """
    Overriden to append a message to the bottom of the table
    """
    return None

Example:

def footer_message(self, qs, params):
    return 'Selected rows: %d' % qs.count()

<style>
    .dataTables_wrapper .dataTables_extraFooter {
        border: 1px solid blue;
        color: blue;
        padding: 8px;
        margin-top: 8px;
        text-align: center;
    }
</style>

Same as footer_message() but appends message to toolbar:

def footer_message(self, qs, params):
    return 'Selected rows: %d' % qs.count()

Renders clipped results as html span tag, providing the non-clipped value as title:

def render_clip_value_as_html(self, long_text, short_text, is_clipped):
    """
    Given long and shor version of text, the following html representation:
        <span title="long_text">short_text[ellipsis]</span>

    To be overridden for further customisations.
    """
    return '<span title="{long_text}">{short_text}{ellipsis}</span>'.format(
        long_text=long_text,
        short_text=short_text,
        ellipsis='&hellip;' if is_clipped else ''
    )

Override to customise the rendering of clipped cells.

As the purpose of this module is all about querysets rendering, any chance to optimize data extractions from the database is more then appropriate.

Starting with v2.3.0, AjaxDatatableView tries to burst performances in two ways:

1. by using only to limit the number of columns in the result set
2. by using select_related to minimize the number of queries involved

The parameters passed to only() and select_related() are inferred from column_defs.

Should this cause any problem, you can disable queryset optimization in two ways:

• globally: by activating the AJAX_DATATABLE_DISABLE_QUERYSET_OPTIMIZATION setting
• per table: by setting to True the value of the disable_queryset_optimization attribute

Alternatively, you can selectively disable the only or select_related optimization with the following flags:

• self.disable_queryset_optimization_only
• self.disable_queryset_optimization_select_related

1. Plain queryset:

   SELECT "tasks_devicetesttask"."id",
          "tasks_devicetesttask"."description",
          "tasks_devicetesttask"."created_on",
          "tasks_devicetesttask"."created_by_id",
          "tasks_devicetesttask"."started_on",
          "tasks_devicetesttask"."completed_on",
          "tasks_devicetesttask"."job_id",
          "tasks_devicetesttask"."status",
          "tasks_devicetesttask"."mode",
          "tasks_devicetesttask"."failure_reason",
          "tasks_devicetesttask"."progress",
          "tasks_devicetesttask"."log_text",
          "tasks_devicetesttask"."author",
          "tasks_devicetesttask"."order",
          "tasks_devicetesttask"."appliance_id",
          "tasks_devicetesttask"."serial_number",
          "tasks_devicetesttask"."program_id",
          "tasks_devicetesttask"."position",
          "tasks_devicetesttask"."hidden",
          "tasks_devicetesttask"."is_duplicate",
          "tasks_devicetesttask"."notes"
   FROM "tasks_devicetesttask"
   WHERE "tasks_devicetesttask"."hidden" = FALSE
   ORDER BY "tasks_devicetesttask"."created_on" DESC
   

   [sql] (233ms) 203 queries with 182 duplicates

2. With select_related():

   SELECT "tasks_devicetesttask"."id",
          "tasks_devicetesttask"."description",
          "tasks_devicetesttask"."created_on",
          "tasks_devicetesttask"."created_by_id",
          "tasks_devicetesttask"."started_on",
          "tasks_devicetesttask"."completed_on",
          "tasks_devicetesttask"."job_id",
          "tasks_devicetesttask"."status",
          "tasks_devicetesttask"."mode",
          "tasks_devicetesttask"."failure_reason",
          "tasks_devicetesttask"."progress",
          "tasks_devicetesttask"."log_text",
          "tasks_devicetesttask"."author",
          "tasks_devicetesttask"."order",
          "tasks_devicetesttask"."appliance_id",
          "tasks_devicetesttask"."serial_number",
          "tasks_devicetesttask"."program_id",
          "tasks_devicetesttask"."position",
          "tasks_devicetesttask"."hidden",
          "tasks_devicetesttask"."is_duplicate",
          "tasks_devicetesttask"."notes",
          "backend_appliance"."id",
          "backend_appliance"."description",
          "backend_appliance"."hidden",
          "backend_appliance"."created",
          "backend_appliance"."created_by_id",
          "backend_appliance"."updated",
          "backend_appliance"."updated_by_id",
          "backend_appliance"."type",
          "backend_appliance"."rotation",
          "backend_appliance"."code",
          "backend_appliance"."barcode",
          "backend_appliance"."mechanical_efficiency_min",
          "backend_appliance"."mechanical_efficiency_max",
          "backend_appliance"."volumetric_efficiency_min",
          "backend_appliance"."volumetric_efficiency_max",
          "backend_appliance"."displacement",
          "backend_appliance"."speed_min",
          "backend_appliance"."speed_max",
          "backend_appliance"."pressure_min",
          "backend_appliance"."pressure_max",
          "backend_appliance"."oil_temperature_min",
          "backend_appliance"."oil_temperature_max",
          "backend_program"."id",
          "backend_program"."description",
          "backend_program"."hidden",
          "backend_program"."created",
          "backend_program"."created_by_id",
          "backend_program"."updated",
          "backend_program"."updated_by_id",
          "backend_program"."code",
          "backend_program"."start_datetime",
          "backend_program"."end_datetime",
          "backend_program"."favourite"
   FROM "tasks_devicetesttask"
   LEFT OUTER JOIN "backend_appliance" ON ("tasks_devicetesttask"."appliance_id" = "backend_appliance"."id")
   LEFT OUTER JOIN "backend_program" ON ("tasks_devicetesttask"."program_id" = "backend_program"."id")
   WHERE "tasks_devicetesttask"."hidden" = FALSE
   ORDER BY "tasks_devicetesttask"."created_on" DESC
   

   [sql] (38ms) 3 queries with 0 duplicates

3. With select_related() and only():

   SELECT "tasks_devicetesttask"."id",
          "tasks_devicetesttask"."started_on",
          "tasks_devicetesttask"."completed_on",
          "tasks_devicetesttask"."status",
          "tasks_devicetesttask"."failure_reason",
          "tasks_devicetesttask"."author",
          "tasks_devicetesttask"."order",
          "tasks_devicetesttask"."appliance_id",
          "tasks_devicetesttask"."serial_number",
          "tasks_devicetesttask"."program_id",
          "tasks_devicetesttask"."position",
          "backend_appliance"."id",
          "backend_appliance"."code",
          "backend_program"."id",
          "backend_program"."code"
   FROM "tasks_devicetesttask"
   LEFT OUTER JOIN "backend_appliance" ON ("tasks_devicetesttask"."appliance_id" = "backend_appliance"."id")
   LEFT OUTER JOIN "backend_program" ON ("tasks_devicetesttask"."program_id" = "backend_program"."id")
   WHERE "tasks_devicetesttask"."hidden" = FALSE
   ORDER BY "tasks_devicetesttask"."created_on" DESC
   

   [sql] (19ms) 3 queries with 0 duplicates

AJAX_DATATABLE_MAX_COLUMNS = 30
AJAX_DATATABLE_TRACE_COLUMNDEFS = False               #  enables debug tracing of applied column defs
AJAX_DATATABLE_TRACE_QUERYDICT = False                #  enables debug tracing of datatables requests
AJAX_DATATABLE_TRACE_QUERYSET = False                 #  enables debug tracing of applied query
AJAX_DATATABLE_TEST_FILTERS = False                   # trace results for each individual filter, for debugging purposes
AJAX_DATATABLE_DISABLE_QUERYSET_OPTIMIZATION = False  # all queryset optimizations are disabled
AJAX_DATATABLE_STRIP_HTML_TAGS = True                 # string HTML tags when rendering the table

You can insert AjaxDatatableView.render_row_tools_column_def() as the first element in column_defs to obtain some tools at the beginning of each table row.

If full_row_select=true is specified as extra-option during table initialization, row details can be toggled by clicking anywhere in the row.

datatables_views.py

from django.contrib.auth.decorators import login_required
from django.utils.decorators import method_decorator

from ajax_datatable.views import AjaxDatatableView
from backend.models import Register


@method_decorator(login_required, name='dispatch')
class RegisterAjaxDatatableView(AjaxDatatableView):

    model = Register
    title = 'Registers'

    column_defs = [
        AjaxDatatableView.render_row_tools_column_def(),
        {
            'name': 'id',
            'visible': False,
        }, {
        ...

By default, these tools will provide an icon to show and hide a detailed view below each table row.

The tools are rendered according to the template ajax_datatable/row_tools.html, which can be overridden.

Row details are automatically collected via Ajax by calling again the views with a specific ?action=details parameters, and will be rendered by the method:

def render_row_details(self, pk, request=None)

which you can further customize when needed.

The default behaviour provided by the base class if shown below:

The default implementation of render_row_details() tries to load a template in the following order:

• ajax_datatable/<app_label>/<model_name>/<render_row_details_template_name>
• ajax_datatable/<app_label>/<render_row_details_template_name>
• ajax_datatable/<render_row_details_template_name>

(where the default value for <render_row_details_template_name> is "render_row_details.html")

and, when found, uses it for rendering.

The template receives the following context:

html = template.render({
    'model': self.model,
    'model_admin': self.get_model_admin(),
    'object': obj,
    'extra_data': [extra_data dict retrieved from request]
}, request)

model_admin, when available, can be used to navigate fieldsets (if defined) in the template, much like django's admin/change_form.html does.

If no template is available, a simple HTML table with all field values is built instead.

In all cases, the resulting HTML will be wrapped in the following structure:

<tr class="details">
    <td class="details">
        <div class="row-details-wrapper" data-parent-row-id="PARENT-ROW-ID">
            ...

When a latest_by column has been specified and show_date_filter is active, a global date range filtering widget is provided, based on jquery-ui.datepicker:

The header of the column used for date filtering is decorated with the class "latest_by"; you can use it to customize it's rendering.

You can fully replace the widget with your own by providing a custom fn_daterange_widget_initialize() callback at Module's initialization, as in the following example, where we use bootstrap.datepicker:

AjaxDatatableViewUtils.init({
    search_icon_html: '<i class="fa fa-search"></i>',
    language: {
    },
    fn_daterange_widget_initialize: function(table, data) {
        var wrapper = table.closest('.dataTables_wrapper');
        var toolbar = wrapper.find(".toolbar");
        toolbar.html(
            '<div class="daterange" style="float: left; margin-right: 6px;">' +
            '{% trans "From" %}: <input type="text" class="date_from" autocomplete="off">' +
            '&nbsp;&nbsp;' +
            '{% trans "To" %}: <input type="text" class="date_to" autocomplete="off">' +
            '</div>'
        );
        var date_pickers = toolbar.find('.date_from, .date_to');
        date_pickers.datepicker();
        date_pickers.on('change', function(event) {
            // Annotate table with values retrieved from date widgets
            var dt_from = toolbar.find('.date_from').data("datepicker");
            var dt_to = toolbar.find('.date_to').data("datepicker");
            table.data('date_from', dt_from ? dt_from.getFormattedDate("yyyy-mm-dd") : '');
            table.data('date_to', dt_to ? dt_to.getFormattedDate("yyyy-mm-dd") : '');
            // Redraw table
            table.api().draw();
        });
    }
});

In case of errors, Datatables.net shows an alert popup:

You can change it to trace the error in the browser console, insted:

// change DataTables' error reporting mechanism to throw a Javascript
// error to the browser's console, rather than alerting it.
$.fn.dataTable.ext.errMode = 'throw';

All details of Datatables.net requests can be logged to the console by activating these setting:

AJAX_DATATABLE_TRACE_COLUMNDEFS = True
AJAX_DATATABLE_TRACE_QUERYDICT = True

The resulting query (before pagination) can be traced as well with:

AJAX_DATATABLE_TRACE_QUERYSET = True

Debugging traces for date range filtering, column filtering or global filtering can be displayed by activating this setting:

AJAX_DATATABLE_TEST_FILTERS

Chances are you might want to supply a standard user interface for listing several models.

In this case, it is possible to use a generic approach and avoid code duplications, as detailed below.

First, we supply a generic view which receives a model as parameter, and passes it to the template used for rendering the page:

file frontend/datatables_views.py:

@login_required
def object_list_view(request, model, template_name="frontend/pages/object_list.html"):
    """
    Render the page which contains the table.
    That will in turn invoke (via Ajax) object_datatable_view(), to fill the table content
    """
    return render(request, template_name, {
        'model': model,
    })

In the urlconf, link to specific models as in the example below:

file frontend/urls.py:

path('channel/', datatables_views.object_list_view, {'model': backend.models.Channel, }, name="channel-list"),

The template uses the model received in the context to display appropriate verbose_name and verbose_name_plural attributes, and to extract app_label and model_name as needed; unfortunately, we also had to supply some very basic helper templatetags, as the _meta attribute of the model is not directly visible in this context.

{% extends 'frontend/base.html' %}
{% load static datatables_view_tags i18n %}

{% block breadcrumbs %}
    <li>
        <a href="{% url 'frontend:index' %}">{% trans 'Home' %}</a>
    </li>
    <li class="active">
        <strong>{{model|model_verbose_name_plural}}</strong>
    </li>
{% endblock breadcrumbs %}

{% block content %}

    {% testhasperm model 'view' as can_view_objects %}
    {% if not can_view_objects %}
        <h2>{% trans "Sorry, you don't have the permission to view these objects" %}</h2>
    {% else %}

        <div>
            <h5>{% trans 'All' %} {{ model|model_verbose_name_plural }}</h5>
            {% ifhasperm model 'add' %}
                <a href="#">{% trans 'Add ...' %}</a>
            {% endifhasperm %}
        </div>
        <div class="table-responsive">
            <table id="datatable" width="100%" class="table table-striped table-bordered table-hover dataTables-example">
            </table>
        </div>

        {% ifhasperm model 'add' %}
            <a href="#">{% trans 'Add ...' %}</a>
        {% endifhasperm %}

    {% endif %}

{% endblock content %}


{% block extrajs %}
    <script language="javascript">

        $(document).ready(function() {
            AjaxDatatableViewUtils.initialize_table(
                $('#datatable'),
                "{% url 'frontend:object-datatable' model|app_label model|model_name %}",
                extra_option={},
                extra_data={}
            );
        });

    </script>
{% endblock %}

app_label and model_name are just strings, and as such can be specified in an url.

The connection with the Django backend uses the following generic url:

{% url 'frontend:object-datatable' model|app_label model|model_name %}

from urls.py:

# List any Model
path('datatable/<str:app_label>/<str:model_name>/', datatables_views.object_datatable_view, name="object-datatable"),

object_datatable_view() is a lookup helper which navigates all AjaxDatatableView-derived classes in the module and selects the view appropriate for the specific model in use:

file frontend/datatables_views.py:

import inspect

def object_datatable_view(request, app_label, model_name):

    # List all AjaxDatatableView in this module
    datatable_views = [
        klass
        for name, klass in inspect.getmembers(sys.modules[__name__])
        if inspect.isclass(klass) and issubclass(klass, AjaxDatatableView)
    ]

    # Scan AjaxDatatableView until we find the right one
    for datatable_view in datatable_views:
        model = datatable_view.model
        if (model is not None and (model._meta.app_label, model._meta.model_name) == (app_label, model_name)):
            view = datatable_view
            break

    return view.as_view()(request)

which for this example happens to be:

@method_decorator(login_required, name='dispatch')
class ChannelAjaxDatatableView(BaseAjaxDatatableView):

    model = Channel
    title = 'Channels'

    column_defs = [
        AjaxDatatableView.render_row_tools_column_def(),
        {
            'name': 'id',
            'visible': False,
        }, {
            'name': 'description',
        }, {
            'name': 'code',
        }
    ]

setTimeout(function () {
    AjaxDatatableViewUtils.adjust_table_columns();
}, 200);

or maybe better:

var table = element.DataTable({
    ...
    "initComplete": function(settings) {
        setTimeout(function () {
            AjaxDatatableViewUtils.adjust_table_columns();
        }, 200);
    }

where:

function adjust_table_columns() {
    // Adjust the column widths of all visible tables
    // https://datatables.net/reference/api/%24.fn.dataTable.tables()
    $.fn.dataTable
        .tables({
            visible: true,
            api: true
        })
        .columns.adjust();
}

$.fn.dataTable.tables({
    api: true
}).draw();

table = $(element).closest('table.dataTable');
$.ajax({
    type: 'GET',
    url: ...
}).done(function(data, textStatus, jqXHR) {
    table.DataTable().ajax.reload(null, false);
});

table.DataTable().row(tr).invalidate().draw(false);

Working example:

{% get_current_language as LANGUAGE_CODE %}

function onToggleQueueStatus(event) {

    // The link is a table cell
    event.preventDefault();
    let td = $(event.target).closest('td');

    // Retrieve the table row and the record id
    let tr = td.closest('tr');
    // Es: "row-692255dc-7eaa-4150-be19-a555a8b34188"
    let row_id = tr.attr('id').substr(4);

    // Call the server via AJAX to process the record
    let url = sprintf('/{{LANGUAGE_CODE}}/j/product_order/%s/toggle_queue_status/', row_id);
    FrontendForms.overlay_show(tr);
    var promise = $.ajax({
        type: 'POST',
        url: url,
        data: null,
        cache: false,
        crossDomain: false,
        headers: {
            'X-CSRFToken': FrontendForms.getCookie('csrftoken')
        }
    }).done(function(data, textStatus, jqXHR) {
        //console.log('OK; data=%o', data);
    }).fail(function(jqXHR, textStatus, errorThrown) {
        console.log('ERROR: ' + jqXHR.responseText);
        Frontend.display_server_error_ex(jqXHR);
    }).always(function() {

        // Since the record has been changed, we need to update the table row;
        // Redraw the row holding the current paging position
        let table = $(tr).closest('table.dataTable');
        table.DataTable().row(tr).invalidate().draw(false);

    });
    return promise;
}

Another (very old) Example:

var table = $(element).closest('table.dataTable');
var table_row_id = table.find('tr.shown').attr('id');
$.ajax({
    type: 'POST',
    url: ...
}).done(function(data, textStatus, jqXHR) {
    table.DataTable().ajax.reload(null, false);

    // Since we've update the record via Ajax, we need to redraw this table row
    var tr = table.find('#' + table_row_id);
    var row = table.DataTable().row(tr)
    row.invalidate().draw();

    // Hack: here we would like to enhance the updated row, by adding the 'updated' class;
    // Since a callback is not available upon draw completion,
    // let's use a timer to try later, and cross fingers
    setTimeout(function() {
        table.find('#' + table_row_id).addClass('updated');
    }, 200);
    setTimeout(function() {
        table.find('#' + table_row_id).addClass('updated');
    }, 1000);

});

// change DataTables' error reporting mechanism to throw a Javascript
// error to the browser's console, rather than alerting it.
$.fn.dataTable.ext.errMode = 'throw';

• AjaxDatatableViewUtils.init(options)
• AjaxDatatableViewUtils.initialize_table(element, url, extra_options={}, extra_data={})
• AjaxDatatableViewUtils.after_table_initialization(table, data, url)
• AjaxDatatableViewUtils.adjust_table_columns()
• AjaxDatatableViewUtils.redraw_all_tables()
• AjaxDatatableViewUtils.redraw_table(element)

You can provide localized messages by initializing the AjaxDatatableViewUtils JS module as follow (example in italian):

AjaxDatatableViewUtils.init({
    search_icon_html: '<i class="fa fa-search" style="font-size: 16px"></i>',
    language: {
        "decimal":        "",
        "emptyTable":     "Nessun dato disponibile",
        "info":           "Visualizzate da _START_ a _END_ di _TOTAL_ righe",
        "infoEmpty":      "",
        "infoFiltered":   "(filtered from _MAX_ total entries)",
        "infoPostFix":    "",
        "thousands":      ",",
        "lengthMenu":     "Visualizza _MENU_ righe per pagina",
        "loadingRecords": "Caricamento in corso ...",
        "processing":     "Elaborazione in corso ...",
        "search":         "Cerca:",
        "zeroRecords":    "Nessun record trovato",
        "paginate": {
            "first":      "Prima",
            "last":       "Ultima",
            "next":       ">>",
            "previous":   "<<"
        },
        "aria": {
            "sortAscending":  ": activate to sort column ascending",
            "sortDescending": ": activate to sort column descending"
        }
    }
});

You can do this, for example, in your "base.html" template, and it will be in effect for all subsequent instantiations:

<script language="javascript">
    $(document).ready(function() {
        AjaxDatatableViewUtils.init({
            ...
        });
    });
</script>

Since the list of table columns is controlled by the library, based on column_defs list specified in the AjaxDatatableView class, you can't insert a custom column "javascript-side".

However, you can easily do it "python-side":

class ArtistAjaxDatatableView(AjaxDatatableView):

    ...

    column_defs = [
        ...
        {'name': 'edit', 'title': 'Edit', 'placeholder': True, 'searchable': False, 'orderable': False, },
        ...
    ]

    def customize_row(self, row, obj):
        row['edit'] = """
            <a href="#" class="btn btn-info btn-edit"
               onclick="var id=this.closest('tr').id.substr(4); alert('Editing Artist: ' + id); return false;">
               Edit
            </a>
        """
        ...

In the snippet above, we added an 'edit' column, customizing it's content via customize_row().

Note how we retrieved the object id from the "row-NNN" table row attribute in the "onclick" handler.

import jsonfield
from ajax_datatable.views import AjaxDatatableView
from .utils import json_prettify


class MyAjaxDatatableView(AjaxDatatableView):

    ...

    def render_row_details(self, pk, request=None):

        obj = self.model.objects.get(pk=pk)
        fields = [f for f in self.model._meta.get_fields() if f.concrete]
        html = '<table class="row-details">'
        for field in fields:
            value = getattr(obj, field.name)
            if isinstance(field, jsonfield.JSONField):
                value = json_prettify(value)
            html += '<tr><td>%s</td><td>%s</td></tr>' % (field.name, value)
        html += '</table>'
        return html

where:

import json
from pygments import highlight
from pygments.lexers import JsonLexer
from pygments.formatters import HtmlFormatter
from django.utils.safestring import mark_safe


def json_prettify_styles():
    """
    Used to generate Pygment styles (to be included in a .CSS file) as follows:
        print(json_prettify_styles())
    """
    formatter = HtmlFormatter(style='colorful')
    return formatter.get_style_defs()


def json_prettify(json_data):
    """
    Adapted from:
    https://www.pydanny.com/pretty-formatting-json-django-admin.html
    """

    # Get the Pygments formatter
    formatter = HtmlFormatter(style='colorful')

    # Highlight the data
    json_text = highlight(
        json.dumps(json_data, indent=2),
        JsonLexer(),
        formatter
    )

    # # remove leading and trailing brances
    # json_text = json_text \
    #     .replace('<span class="p">{</span>\n', '') \
    #     .replace('<span class="p">}</span>\n', '')

    # Get the stylesheet
    #style = "<style>" + formatter.get_style_defs() + "</style>"
    style = ''

    # Safe the output
    return mark_safe(style + json_text)

First, we mark the relevant info with a specific CSS class, so we can search for it later

column_defs = [
    ...
    }, {
        'name': 'error_counter',
        'title': 'errors',
        'className': 'error_counter',
    }, {
    ...
]

Have a callback called after each table redraw

var table = element.DataTable({
    ...
});

table.on('draw.dt', function(event) {
    onTableDraw(event);
});

then change the rendered table as needed

var onTableDraw = function (event) {

    var html_table = $(event.target);
    html_table.find('tr').each(function(index, item) {

        try {
            var row = $(item);
            text = row.children('td.error_counter').first().text();
            var error_counter = isNaN(text) ? 0 : parseInt(text);

            if (error_counter > 0) {
                row.addClass('bold');
            }
            else {
                row.addClass('grayed');
            }
        }
            catch(err) {
        }

    });
}

or use a rowCallback as follows:

// Subscribe "rowCallback" event
$('#datatable').on('rowCallback', function(event, table, row, data ) {
    $(row).addClass(data.read ? 'read' : 'unread');
}

This works even if the 'read' column we're interested in is actually not visible.

Row details customization:

def render_row_details(self, pk, request=None):

    obj = self.model.objects.get(pk=pk)
    html = '<table class="row-details">'
    html += "<tr><td>alarm status:</td><td>"
    for choice in BaseTask.ALARM_STATUS_CHOICES:
        # Lo stato corrente lo visualizziamo in grassetto
        if choice[0] == obj.alarm:
            html += '<b>%s</b>&nbsp;' % (choice[1])
        else:
            # Se non "unalarmed", mostriamo i link per cambiare lo stato
            # (tutti tranne "unalarmed")
            if obj.alarm != BaseTask.ALARM_STATUS_UNALARMED and choice[0] != BaseTask.ALARM_STATUS_UNALARMED:
                html += '<a class="set-alarm" href="#" onclick="set_row_alarm(this, \'%s\', %d); return false">%s</a>&nbsp;' % (
                    str(obj.pk),
                    choice[0],
                    choice[1]
                )
    html += '</td></tr>'

Client-side code:

<script language="javascript">

    function set_row_alarm(element, task_id, value) {
        $("body").css("cursor", "wait");
        //console.log('set_row_alarm: %o %o %o', element, task_id, value);
        table = $(element).closest('table.dataTable');
        //console.log('table id: %o', table.attr('id'));

        $.ajax({
            type: 'GET',
            url: sprintf('/set_alarm/%s/%s/%d/', table.attr('id'), task_id, value),
            dataType: 'html'
        }).done(function(data, textStatus, jqXHR) {
            table.DataTable().ajax.reload(null, false);
        }).always(function( data, textStatus, jqXHR) {
            $("body").css("cursor", "default");
        });
    }

Server-side code:

urlpatterns = [
    ...
    path('set_alarm/<str:table_id>/<uuid:task_id>/<int:new_status>/',
        views.set_alarm,
        name="set_alarm"),
]

@login_required
def set_alarm(request, table_id, task_id, new_status):

    # Retrieve model from table id
    # Example table_id:
    #   'datatable_walletreceivetransactionstask'
    #   'datatable_walletcheckstatustask_summary'
    model_name = table_id.split('_')[1]
    model = apps.get_model('tasks', model_name)

    # Retrieve task
    task = get_object_by_uuid_or_404(model, task_id)

    # Set alarm value
    task.set_alarm(request, new_status)

    return HttpResponse('ok')

• Table row selection
• Export table data to excel of pdf
• Responsive table

• specific rendering for boolean columns

Currently, an exact match is applied; a date-range selection would be better; references:

• https://datatables.net/plug-ins/filtering/row-based/range_dates
• https://datatables.net/extensions/select/
• https://github.com/RobinDev/jquery.dataTables.columnFilter.js?files=1

• https://github.com/yourlabs/django-autocomplete-light
• https://github.com/crucialfelix/django-ajax-selects

A checkbox or a select