django-simple-stats 0.7.0

A django package for creating simple stats from a query

django-simple-stats

A django package for creating stats from a query. This package should be compatible with all Django versions > 3.x

You can use something like this:

class TravelStats(StatSet):
    total = QueryAggregateSingleStat(label='Total')
    amount = QueryAggregateSingleStat(label='Total amount', method='sum')
    port__name = QueryAggregateStat(label='By port', )

to get something like this:

Installation

Install it from pip:

pip install django-simple-stats

or the latest version from git:

pip install git+https://github.com/spapas/django-simple-stats

No other installation is needed.

Stat kinds

• The query_aggregate_single / QueryAggregateSingleStat will run the aggregate function on a field and return a single value. For example you can get the total number of rows of your query or the sum of all fields.

• The query_aggregate / QueryAggregateStat will run the aggregate function on a field and return the list of values. You can run the aggregate function on a different field by passing aggr_field (so you can group by a field and return the sum of another field for each group). This is mainly useful for foreign keys and if you’ve got distinct values in your queries. For example count the number of rows per user. Also it is useful for booleans for example to get the number of rows that have a flag turned on and off.

• The choice_aggregate / ChoiceAggregateStat is similar to the query_aggregate but will use a choices attribute to return better looking values. This will not return Null values

• The choice_aggregate_with_null / ChoiceAggregateNullStat is the same as choice_aggregate but will return Null values (so you can add a (None, "Empty") choice to your choices)

• The query_aggregate_date / QueryAggregateDateStat is similar to the query_aggregate but will return the aggregates on a specific date field; use what to pass year, month, day.

• The query_aggregate_datetime / QueryAggregateDateTimeStat is similar to the query_aggregate_date but will return the aggregates on time also.

• The query_aggregate_extract_date / QueryAggregateExtractDateStat is similar to query_aggregate_date but will use Extract for the date instead of Trunc. This is useful if you want to group by the month/day/hour etc as a specific value, i.e this will group all rows of June on the same row while query_aggregate_date will differntiate between June 21 and June 22 and June 23.

• Finally, the query_aggregate_buckets // QueryAggregateBucketsStat is used to create buckets of values. You’ll pass the list of buckets and the query will return the results that belong in each bucket. The stats module will run individual queries with field__gte for each value. So for example if you pass [100, 50, 10] and you have a field price it will run price__gte=100, price__gte=50, price__gte=10 and return the results.

Usage

Both a declarative (similar to how Django models/forms work) and a functional API can be used.

For the declarative API, you need to declare a class inheriting from simple_stats.StatSet and declare the stats fields in it using the corresponding stats classes like simple_stats.QueryAggregateStat. Then you’ll initialize this class in your Django view passing it the query. For example:

from simple_stats import StatSet, QueryAggregateSingleStat

class MyStats(StatSet):
    id = QueryAggregateSingleStat(label='Total number')

# Then in your view
stats = MyStats(query=MyModel.objects.all())

The stats classes take the following initialization parameters:

• field (optional): The field that the aggreate will run on; use __ for joins i.e fiedld1__field2

• label (optional): The textual description of this statistic (will be filled with the field if not passed)

• method (optional): The aggregate method. Can be one of count, sum, max, min, avg. Default is count

• what (optional): Only required for query_aggregate_date, it is eithed year, month, day

• choices (optional): Only required for choice_aggregate and choice_aggregate_with_null, it must be a django choices list

• buckets (optional): Only required for query_aggregate_buckets. Must be a list from the biggest to the lowest value.

• aggr_field (optional): This field is optional and can be used for query_aggregate, query_aggregate_date, choice_aggregate and choice_aggregate_with_null. It denotes a field that would run the aggregate function on.

• formatter (optional): A callback that can be used to format the value, it should get a value and return a rendered value (i.e lambda v: "${}".format(v))

Please notice that if you don’t pass the field parameter then the name of the attribute will be used (i.e. it will be field=id in the example above).

The StatSet instance will be an enumerable returning a list of dictionaries with the following attributes:

• label: Same as the label in the configuration

• value: Will have a value if you use the query_aggregate_single, else will be None

• values: Will be empty for query_aggregate_single else will be a list of tuples. Each tuple will have two elements, (label, value)

On the other hand, the only supported method in the functional API the simple_stats.get_stats. It expects a django query and the stats configuration (list of dicts). Each element of the configuration list is a dictionary that has the same attributes as the init parameters for the class. There are only two differences:

• The field is now required

• We must pass the kind of aggregate we need (similar to the class we used on the declarative API). Choices here are: query_aggregate_single, query_aggregate, choice_aggregate, choice_aggregate_with_null, query_aggregate_date, query_aggregate_buckets.

See below for a complete example.

The response will be a list of dictionaries with the same attributes as the StatSet instance.

Please notice that the declarative api will create a dictionary and actually call the get_stats function so in both cases the result will be exactly the same.

Example declarative

Please remember in the example below that if you don’t pass the field parameter then the name of the attribute will be used. Also by default the method is count.

from simple_stats import from simple_stats import (
    StatSet,
    QueryAggregateStat,
    QueryAggregateSingleStat,
    ChoiceAggregateStat,
    QueryAggregateDateStat,
    QueryAggregateBucketsStat,
)

class MyStats(StatSet):
    id = QueryAggregateSingleStat(label='Total number')
    price = QueryAggregateSingleStat(label='Total price', method='sum')
    pilot_authority__name = QueryAggregateStat(label='Per authority')
    pilot_authority__name = QueryAggregateStat(label='Per authority by price', aggr_field='price')
    status = ChoiceAggregateStat(label='Per status', choices=MyModel.STATUS_CHOICES)
    status_price = ChoiceAggregateStat(
        label='Per status by price',
        choices=MyModel.STATUS_CHOICES,
        field='status',
        aggr_field='price'
        formatter=lambda v: "€ {}".format(v) if v else '-'
    )
    year = QueryAggregateDateStat(label='Per year', what='year', field='created_on')
    year_price = QueryAggregateDateStat(label='Per year by price', what='year', aggr_field='price', field='created_on')
    buckets = QueryAggregateBucketsStat(label='Buckets', buckets=[100, 50, 10])

def my_view(request):
    qs = TestModel.objects.all()

    stats = MyStats(qs)
    return render(request, 'my_template.html', {'stats': stats})

the stats result will be an enumerable similar to this one:

[
  {'label': 'Total', 'values': [], 'value': 1216},
  {'label': 'Total price', 'values': [], 'value': 323.16},
  {'label': 'Per authority', 'values': [('Authority 1', 200), ('Authority 2', 9),   ], 'value': None},
  {'label': 'Per authority by price', 'values': [('Authority 1', 123.23), ('Authority 2', 42.12),   ], 'value': None},
  {'label': 'Per status', 'values': [('New', 200), ('Cancel', 0)], 'value': None},
  {'label': 'Per status by price', 'values': [('New', '€ 32.01'), ('Cancel', '€ 44.23')], 'value': None},
  {'label': 'Per year', 'values': [(2021, 582), (2022, 634)], 'value': None}
  {'label': 'Per year by price', 'values': [(2021, 5.82), (2022, 6.34)], 'value': None}
  {'label': 'Per price', 'values': [('> 5000', 1), ('> 1000', 29), ('> 500', 86), ('> 0', 305)], 'value': None}
]

You can display this in your template using something like this (using bootstrap):

<div class='row'>
  {% for s in stats %}
  <div class='col-md-4 mb-5' style='max-height: 500px; overflow: auto;'>
      <h4>{{ s.label }}</h4>
      {% if s.values %}
          <table class='table table-condensed table-striped small table-sm'>
              {% for v in s.values %}
                  <tr>
                      <td>{{ v.0 }}</td>
                      <td>{{ v.1 }}</td>
                  </tr>
              {% endfor %}
          </table>
      {% else %}
          <b>{{ s.value }}</b>
      {% endif %}
  </div>
  {% endfor %}
</div>

Example functional

from simple_stats import get_stats

STATS_CFG = cfg = [
        {
            'kind': 'query_aggregate_single',
            'label': 'Total',
            'field': 'id',
        }, {
            'kind': 'query_aggregate_single',
            'label': 'Total price',
            'method': 'sum',
            'field': 'price',
        }, {
            'kind': 'query_aggregate',
            'label': 'Per authority',
            'field': 'pilot_authority__name',
        }, {
            'kind': 'query_aggregate',
            'label': 'Per authority by price',
            'field': 'pilot_authority__name',
            'aggr_field': 'price',
            'formatter': lambda v: "€ {}".format(v) if v else '-'
        }, {
            'kind': 'choice_aggregate',
            'label': 'Per status',
            'field': 'status',
            'choices': models.STATUS_CHOICES,
        }, {
            'kind': 'choice_aggregate',
            'label': 'Per status by price',
            'field': 'status',
            'aggr_field': 'price',
            'choices': models.STATUS_CHOICES,
        }, {
            'kind': 'query_aggregate_date',
            'label': 'Per year',
            'field': 'created_on',
            'what': 'year',
        }, {
            'kind': 'query_aggregate_date',
            'label': 'Per year by price',
            'field': 'created_on',
            'what': 'year',
            'aggr_field': 'price',
        }, {
            'kind': 'query_aggregate_buckets',
            'label': 'Per price',
            'field': 'price',
            'buckets': [100_00, 50_00, 1_000, 500, 0]
        }
    ]

def my_view(request):
    qs = TestModel.objects.all()

    stats = get_stats(qs, STATS_CFG)
    return render(request, 'my_template.html', {'stats': stats})

The stats will be an array of dictionaries, similar to the declarative example.

Exporting the stats

You can easily export these stats in xls using the xlwt (https://pypi.org/project/xlwt/) library and this function:

import xlwt

def create_xls_resp(stats, response):
    context = self.get_context_data()
    import xlwt
    wb = xlwt.Workbook(encoding="utf-8")
    for stat in stats:
        ws = wb.add_sheet(stat["label"][:31])
        ws.write(0,0,stat["label"], xlwt.easyxf('font: name Calibri, bold on', ))
        if stat["value"]:
            ws.write(0,1,stat["value"], xlwt.easyxf('font: name Calibri, bold on', ))

        for i, val in enumerate(stat["values"], start=2):
            for j,v in enumerate(val, start=0):
                ws.write(i,j,v)
    wb.save(response)

Now you can call it like this from your view:

from django.http import HttpResponse

def my_export_view(request):
    qs = TestModel.objects.all()

    stats = get_stats(qs, STATS_CFG)
    response = HttpResponse(content_type="application/ms-excel")
    response["Content-Disposition"] = "attachment; filename=export.xls"
    create_xls_resp(response)
    return response

Changelog

• v.0.7.0: Add query_aggregate_extract_date and QueryAggregateExtractDateStat

• v.0.6.0: Add tests!

• v.0.5.1: Allow adding a formatter for the values

• v.0.5.0: Add declarative API

• v.0.4.0: Allow the aggregate function to run on a different field using aggr_field

• v.0.3.1: Fix small bug with choice_aggregate_with_null

• v.0.3.0: Add choice_aggregate_with_null and throw if stat kind is not found

• v.0.2.1: Fix small bug with column aliases

• v.0.2.0: Changed API; use query_aggregate_datetime for a datetime field and query_aggregate_date for a date field

• v.0.1.0: Initial version