Converting Python REST API Calls to Power Query

Question

I am working with a cloud-based service that is exposed using OAuth REST API.

I have a working Python script, with which I am able to first post username, password and api token to get an authentication token. Then I use the api and authentication tokens to get data. Here is my Python script:

import requests

base_url = 'https://example.com'
user_name = 'john_doe'
password = 'pssw0rd'
api_token = '4r-eueyta-dh7cuq-26'
timeout_min = 1

headers = {'Accept': 'application/json'}

# get authentication token
auth_url = base_url + '/api/authenticate'
data = {'Username': user_name, 
        'Password': password, 
        'ApiToken': api_token, 
        'AuthorizationTokenTimeoutMinutes': timeout_min}
r = requests.post(auth_url, headers=headers, data=data)
# the entire string response is the auth token
auth_token = r.text

# get data
proj_url = base_url + '/api/project/active?' + \
             'api-token={0}&authentication-token={1}'.format(api_token, auth_token)
r = requests.get(proj_url, headers=headers)
print(r.text)

This is working fine.

---

Next I want to use the same approach inside Microsoft Power BI to configure this data as a data source. I followed the approach outlined at http://angryanalyticsblog.azurewebsites.net/index.php/2016/05/16/api-strategies-with-power-bi/

As mentioned in the post, I set the variables

• base_url
• user_name
• password
• api_token
• auth_token_timeout_min

as parameters and set their values. Then got the following power query script in advanced editor:

 let
 // get auth token
 auth_url = #"base_url" & "/api/authenticate",
 post_data = "Username=" & #"user_name" & "&Password=" & #"password" & "&ApiToken=" & #"api_token" & "&AuthorizationTokenTimeoutMinutes=" & #"auth_token_timeout_min",
 AuthJsonQuery = Web.Contents(auth_url,
                              [Headers = [#"Accept"="application/json"], 
                               Content = Text.ToBinary(post_data)]),
 auth_token = Json.Document(AuthJsonQuery),

 // get data
 proj_url = #"base_url" & "/api/project/active?api-token=" & #"api_token" & "&authentication-token=" & auth_token,
 GetJsonQuery = Web.Contents(proj_url,
                             [Headers = [#"Accept"="application/json"]]),
 FormatAsJsonQuery = Json.Document(GetJsonQuery),
  in
  #"FormatAsJsonQuery"

This script gives the following error:

DataSource.Error: Web.Contents failed to get contents from 'https://example.com/api/authenticate' (415): Unsupported Media Type
Details:
    DataSourceKind=Web
    DataSourcePath=https://example.com/api/authenticate
    Url=https://example.com/api/authenticate

The raw response from the first POST request is exactly the authentication token and I am missing something in correctly decoding it.

If I manually configure the authentication token directly in proj_url then I am able to get the data, so I am doing something wrong with getting the authentication token.

---

UPDATE: My goal was to somehow make the data available inside Power BI. I took an alternative approach with using R as a data source. (You will need to install R and the httr package in R for this.) Posting that solution here:

library(httr)

base_url <- 'https://example.com'
user_name <- 'john_doe'
password <- 'pwws0rd'
api_token <- '4r-eueyta-dh7cuq-26'
timeout_min <- 1

# get authentication token
auth_url <- paste0(base_url, '/api/authenticate')
data <- list('Username' = user_name,
             'Password' = password,
             'ApiToken' = api_token,
             'AuthorizationTokenTimeoutMinutes' = timeout_min)
resp <- POST(auth_url,
             body = data, 
             encode = 'json')
auth_token <- content(resp, "parsed", "application/json")

# get data
proj_url <- paste0(base_url, '/api/project/active?', 
                   'api-token=', api_token, 
                   '&authentication-token=', auth_token)
resp <- GET(proj_url, encode = 'json')

# convert the response to data frame
results <- content(resp, "parsed", "application/json")
results <- lapply(results, function(x) {
                    x[sapply(x, is.null)] <- NA
                    unlist(x)})
projects <- do.call("rbind", results) # matrix
projects <- data.frame(projects, stringsAsFactors = FALSE)

projects is then available for importing into Power BI.

Answer 1

I know it is too late to answer, but I'm guessing it is because the header lacks "Content-Type=application/x-www-form-urlencoded".

Python requests library automatically adds this header part, while Power Query does not.