Typeform ETL

A python class and main function to extract and continuously sync data from a Typeform.com account.

Installation

Requirements

You should prefer Python 3 packages from your operating system.

So under Fedora Linux:

dnf install python3-requests python3-sqlalchemy python3-pandas python3-configobj python3-mysql

Under Red Hat Enterprise Linux 8:

dnf install python3-requests python3-sqlalchemy python3-configobj mariadb-connector-c-devel mariadb-connector-c gcc
pip3 install mysqlclient --user

This MariaDB connector is the one that works with SQLAlchemy (used by the module). Other connectors as PyMySQL failed our tests.

Installing mysqlclient with pip requires compilers and MariaDB development framework pre-installed in the system, as shown in the Red Hat Enterprise Linux section.

Install the module

pip3 install TypeformETL --user

All unsatisfied dependencies (such as Pandas on RHEL) will be installed along.

Or, to upgrade:

pip3 install -U TypeformETL --no-deps --user

Remove the --no-deps if you wish to upgrade also dependent modules as pandas and numpy.

Usage

Get a personal access token from Typeform's admin portal. We'll call it ___API_KEY___ from now on.

With plain command line

python3 -m TypeformETL --typeform "___API_KEY___" --database 'mysql://user:password@host/dbname'

Add --restart to get and sync all data from Typeform, not just last updates.

Add --updatedb to do everything except update database. Good for tests.

Add --debug to be more verbose.

Config file

If you want to not pass arguments through command line, you can also use a config file. Use --config to pass a config file name. By default, the main module will look for a file name called syncFromTypeform.conf in the current directory. Arguments from config file and command line will be combined giving priority to the ones passed in command line. So if a database URL is passed both in command line and config file, the command line will be used.

With CRON

I have these 2 entries together on my crontab and then I don't need a config file:

@hourly    python3 -m TypeformETL --typeform "___API_KEY___" --database 'mysql://user:password@host/dbname' --tableprefix 'tf_'
30 3 * * 0 python3 -m TypeformETL --typeform "___API_KEY___" --database 'mysql://user:password@host/dbname' --tableprefix 'tf_' --restart

Which will run a sync every hour. And once a week will reset data tables and bring all data from scratch.

Into a Python program

from TypeformETL import TypeformETL

tf = TypeformETL(
	token='___API_KEY___',
	dburl='mysql://user:password@host/dbname',
	restart=False,     # True to reset data tables and bring all data from scratch
	dbupdate=True,     # Wether to simulate or actually write in database
	tableprefix='tf_'  # To better organize your tables
)


# Read Typeform updates and write to DB
tf.sync()

Database

A SQL database must exist. MySQL and MariaDB tested. Use the examples/datamodel.sql script to create the database.

Having “tf_” as a table prefix, these are the objects (tables and views) that will be created and updated:

The module makes INSERT, UPDATE, CREATE TABLE, DROP TABLE, TRUNCATE operations. Make sure the database connection user has granted permission to all these operations.

SQL definition for all these tables and views can be found in examples/datamodel.sql.

Net Promoter Score

A common use of Typeform service is to measure user satisfaction through questions like “From 0 to 10, what is the chance of recommending this web site/app/service to a friend?”.

Once all answers are structured in a database its easy to calculate, from answers to a specific form field, both current NPS and NPS evolution as time series.

SQL views tf_nps and tf_nps_daily respectively will deliver this information. The tf_nps_daily is quite advanced and only possible (in one view) when using advanced SQL Window functions:

create view tf_nps_daily as 
select
	a.form as form_id,
	fi.name as field_name,
	r.submitted as date,
	fi.type as type,
	f.title as form_title,
	fi.title as field_title,

 	((count(case when a.answer>=9 then 1 else NULL end) over day)-(count(case when a.answer<7 then 1 else NULL end) over day))/(count(a.answer) over day) as NPS_ofdate,
	count(case when a.answer<7 then 1 else NULL end) over day as detractors,
	count(case when a.answer between 7 and 8 then 1 else NULL end) over day as passives,
	count(case when a.answer>=9 then 1 else NULL end) over day as promoters,
	count(a.answer) over day as total,

	((count(case when a.answer>=9 then 1 else NULL end) over untilday)-(count(case when a.answer<7 then 1 else NULL end) over untilday))/(count(a.answer) over untilday) as NPS_cumulative,
	count(case when a.answer<7 then 1 else NULL end) over untilday as detr_cumulative,
	count(case when a.answer between 7 and 8 then 1 else NULL end) over untilday as pass_cumulative,
	count(case when a.answer>=9 then 1 else NULL end) over untilday as prom_cumulative,
	count(a.answer) over untilday as totl_cumulative
from
	tf_answers a,
	tf_responses r,
	tf_form_items fi,
	tf_forms f
where
	a.data_type_hint in ('number')
	and r.submitted is not NULL
	and r.id = a.response
	and fi.id = a.field
	and fi.form = a.form
	and f.id = a.form
window
	day as (partition by date(r.submitted), a.form, a.field),
	untilday as (partition by a.form, a.field order by r.submitted asc rows unbounded preceding)
order by
	a.form, a.field, r.submitted asc

The tf_nps view is simpler:

CREATE OR REPLACE VIEW tf_nps AS
select
	a.form as form_id,
	fi.name as field_name,
	min(r.submitted) as first,
	max(r.submitted) as last,
	fi.type as type,
	f.title as form_title,
	fi.title as field_title,
 	((count(case when a.answer>=9 then 1 else NULL end))-(count(case when a.answer<7 then 1 else NULL end)))/(count(a.answer)) as NPS,
	count(case when a.answer<7 then 1 else NULL end) as detractors,
	count(case when a.answer between 7 and 8 then 1 else NULL end) as passives,
	count(case when a.answer>=9 then 1 else NULL end) as promoters,
	count(a.answer) as total,
	avg(a.answer) as average,
	std(a.answer) as std_deviation
from
	tf_answers a,
	tf_responses r,
	tf_form_items fi,
	tf_forms f
where
	a.data_type_hint in ('number')
	and r.submitted is not NULL
	and r.id = a.response
	and fi.id = a.field
	and fi.form = a.form
	and f.id = a.form
group by
	form_id, field_name
order by
	a.form, a.field, r.submitted asc

Once set, you can query the NPS for a specific field as a time series (tf_nps_daily) or its last value (tf_nps) like this:

SELECT * FROM tf_nps_daily WHERE form_id='to6xfp' AND field_name='353eb07c-15bf-4669-9793-9c0ec33f818a'

Resulting data, when graphed, looks like this:

This graph was generated by the examples/NPS Analysis.ipynb notebook.