API to typeform data sets

Typeform is a company that specializes in online form building. This R package allows users to download their form results through the exposed API (V2).

** The rtypeform package now uses V2. This is a breaking change from the previous version.**

Installation

The package can be installed from CRAN

install.packages("rtypeform")

and loaded in the usual way.

library("rtypeform")

Using this package

This package can be used with either a typeform personal access token or by setting up an application and creating an OAuth access token.

A personal access token gives you full access to all of the typeform API for your typeforms and results. Note anyone with your personal access token can retrieve, update and delete your typeforms and data. To access your typeform data with a personal access token see the Personal Access Token section below.

When creating an application with an OAuth access token, explicit permission for different functionality (scopes) must be granted. See the section below on OAuth access.

If you have previously used the version 1 API this is now entirely removed. You will need to generate new tokens.

Personal Access Token

To use this package with a personal access token you need to first obtain one. It is fairly easy to obtain one. See typeform’s help page. The token will look something like

943af478d3ff3d4d760020c11af102b79c440513

OAuth Access

When you create an application that authenticates using OAuth you will use scopes to define the extent of access to a users data. This way your app can request a users permission to undertake actions on that users behalf.

This link will get you started with registering a new application on your account.

Once you have your client id and client secret you can use the rtypeform package to set these as options.

rtypeform_set_client_id(my_client_id)
rtypeform_set_client_secret(my_client_secret)

As with the personal access token. Anyone with these details can impersonate you to obtain, update and remove data, they should always be kept safe.

Having set the client id and secret, before we can obtain an access token we also need to define the scope of our application. rtypeform_set_scope takes as argument a character vector of allowed access scopes. For more information see the scopes section below.

rtypeform_set_scope("forms:read")

We can then generate a new token with

api = make_new_token()

This will open a web browser prompting the user to give permission. The token can be cached in a local .httr-oauth file between sessions.

Scopes

You define the scope at the time that the access token is generated. To discover what each scope allows access to, see here.

Using the package

Once you have this key, (either personal access token, or an oauth token created by make_new_token()) we can extract data from typeform

api = "XXXXX"
# Was get_typeforms() in V1 of the package
forms = get_forms(api)

The forms object is also contains attributes containing the total number of forms.

attr(forms, "total_items")
#> [1] 135

If you don’t pass your api token as an argument, it will attempt to read the variable typeform_api2 from your .Renviron file, via Sys.getenv("typeform_api2"). If this variable is set correctly, then you can omit the api argument

# See ?get_forms for further details
forms = get_forms()

To set the access token for the current session you can use

rtypeform_set_token(api)

set (see Efficient R programming Chapter 2 for more details).

You can download data from a particular typeform via

# Most recent typeform
form_id = forms$form_id[1]
q = get_responses(form_id, completed = TRUE)

The object q is a list. The first element is meta that contain details on the user, such as, their platform and user_agent. The other list elements are responses to each question.

There are a number of options for downloading the data. For example

q = get_responses(form_id, completed = TRUE, page_size = 100)

See the ?get_responses() help page for other options.

Looking at the responses

Since the responses is list, we get to perform lots of map operations. I find using purrr and the tidyverse make this a bit easier. To see the question types we can use string a few map() commands together

library("tidyverse")
question_types = q[-1] %>% # Remove the meta
   map(~select(.x, type)) %>%
   map_df(~slice(.x, 1)) %>%
   pull()

Example: Multiple Filters / Order

Imagine we only want:

completed results, so we add the parameter completed = TRUE.
a maximum of 5 results, so we add the parameter page_size = 5.
results since 2018-01-01 11:00:00.

since = "2018-01-01 11:00:00"
# convert to date-time 
since = lubridate::ymd_hms(since)
q = get_responses(form_id, completed = TRUE, 
                  page_size = 5, since = since)

Other information

If you have any suggestions or find bugs, please use the github issue tracker.
Feel free to submit pull requests.

Development of this package was supported by Jumping Rivers

Via SO thread

I think I'd consider returning separate components for and 100% completed and less than 100% completed answers:

library(purrr)
library(dplyr)

res <- httr::GET("https://api.typeform.com/v1/form/JlCM2J",
           query=list(key="ecab3f590a2af4ca55468adc95686a043bbf6c9a"))

httr::content(res, as="text") %>%
  jsonlite::fromJSON(simplifyVector=FALSE) -> tmp

http_status <- tmp$http_status

stats <- tmp$stats

map_df(tmp$questions, flatten_df) -> questions

keep(tmp$responses, ~.$completed ==1) %>%
  map_df(flatten_df) -> completed

glimpse(completed)
## Observations: 1
## Variables: 35
## $ completed                          <chr> "1"
## $ token                              <chr> "b84567f64cedf2b0050eb0c209...
## $ browser                            <chr> "default"
## $ platform                           <chr> "other"
## $ date_land                          <chr> "2016-12-17 12:21:06"
## $ date_submit                        <chr> "2016-12-17 12:22:29"
## $ user_agent                         <chr> "Mozilla/5.0 (Windows NT 10...
## $ referer                            <chr> "https://gowtham11.typeform...
## $ network_id                         <chr> "fdebc44ea7"
## $ textfield_38991412                 <chr> "A"
## $ dropdown_38991418                  <chr> "Accounting"
## $ textarea_38991420                  <chr> "A"
## $ textfield_38991413                 <chr> "A"
## $ textarea_38991421                  <chr> "A"
## $ listimage_38991426_choice          <chr> "Company"
## $ textfield_38991414                 <chr> "A"
## $ website_38991435                   <chr> "http://A.com"
## $ textarea_38991422                  <chr> "A"
## $ listimage_38991427_choice          <chr> "Sincere"
## $ listimage_38991428_choice          <chr> "Male"
## $ list_38991436_choice               <chr> "17 or younger"
## $ list_38991437_choice               <chr> "Upper class"
## $ listimage_38991429_choice_49501105 <chr> "Store"
## $ listimage_38991430_choice          <chr> "Product"
## $ textarea_38991423                  <chr> "A"
## $ listimage_38991431_choice          <chr> "Techy"
## $ listimage_38991432_choice_49501124 <chr> "Fuchsia Rose"
## $ listimage_38991433_choice          <chr> "Classic"
## $ list_38991438_choice               <chr> "$3,000 or less"
## $ listimage_38991434_choice_49501140 <chr> "Brand Design"
## $ textarea_38991424                  <chr> "A"
## $ textfield_38991415                 <chr> "A"
## $ dropdown_38991419                  <chr> "Afghanistan"
## $ email_38991439                     <chr> "[email protected]"
## $ textarea_38991425                  <chr> "A"

keep(tmp$responses, ~.$completed ==0) %>%
  map_df(flatten_df) -> not_completed

glimpse(not_completed)
## Observations: 1
## Variables: 9
## $ completed   <chr> "0"
## $ token       <chr> "27181b2b95f69f0024f198194ec48561"
## $ browser     <chr> "default"
## $ platform    <chr> "other"
## $ date_land   <chr> "2016-12-16 04:01:40"
## $ date_submit <chr> "0000-00-00 00:00:00"
## $ user_agent  <chr> "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWe...
## $ referer     <chr> "https://gowtham11.typeform.com/to/JlCM2J"
## $ network_id  <chr> "8c7fd4d289"

You could return the 5 components as a list():

list(
  http_status=tmp$http_status,
  stats=tmp$stats,
  questions=tmp$questions,
  completed_responses=completed,
  uncompleted_responses=not_completed
)

and give it a class name which wld enable having a special print.class() method for it to show the summary data and the responses in a tad nicer format.

Not sure I'd spend more time on it unless you see lots of folks using typeform for their forms.

(Just realized I hadn't logged out of the work github acct to post this. Rest assured it's @hrbrmstr)

csgillespie / rtypeform Goto Github PK

rtypeform's Introduction

API to typeform data sets

Installation

Using this package

Personal Access Token

OAuth Access

Scopes

Using the package

Looking at the responses

Example: Multiple Filters / Order

Other information

rtypeform's People

Contributors

Stargazers

Watchers

Forkers

rtypeform's Issues

Recommend Projects

Recommend Topics

Recommend Org