Python to PHP - Azure Machine Learning - php

Question

Hi im trying to translate this python script to php. I don't have much knowledge of Python and limited for PHP.
The python script is:
import urllib2
import json
data = {
"Inputs": {
"input1": {
"ColumnNames": ["Client_ID"],
"Values": [ [ "0" ], [ "0" ], ]
},
},
"GlobalParameters": {}
}
body = str.encode(json.dumps(data))
url = 'https://ussouthcentral.services.azureml.net/workspaces/3e1515433b9d477f8bd02b659428cddc/services/cb1b14b17422435984943d51b5957ec7/execute?api-version=2.0&details=true'
api_key = 'abc123'
headers = {'Content-Type':'application/json', 'Authorization':('Bearer '+ api_key)}
req = urllib2.Request(url, body, headers)
try:
response = urllib2.urlopen(req)
result = response.read()
print(result)
except urllib2.HTTPError, error:
print("The request failed with status code: " + str(error.code))
print(error.info())
print(json.loads(error.read()))
In a bid to try and convert it myself, here is what I have done so far:
<?php
error_reporting(E_ALL);
ini_set('display_errors', 1);
$data = array(
'Inputs'=> array(
'input1'=> array(
'ColumnNames' => ["Client_ID"],
'Values' => [ [ "0" ], [ "0" ], ]
),
),
'GlobalParameters'=> array()
);
$body = json_encode($data);
$url = 'https://ussouthcentral.services.azureml.net/workspaces/3e1515433b9d477f8bd02b659428cddc/services/cb1b14d17425435984943d41a5957ec7/execute?api-version=2.0&details=true';
$api_key = 'abc123';
$headers = array('Content-Type'=>'application/json', 'Authorization'=>('Bearer '+ $api_key));
$curl = curl_init();
curl_setopt($curl, CURLOPT_URL,$url);
curl_setopt($curl, CURLOPT_POST, 1);
curl_setopt($curl, CURLOPT_POSTFIELDS, $body);
curl_setopt($curl, CURLOPT_HTTPHEADER, $headers);
$result = curl_exec($curl);
var_dump($result);
im sure I have got lots wrong but would appreciate the help.
thanks

Answer 1

I just had to do this myself and though I'd provide an answer for you. If you're going to talk to Azure's ML platform using php, you need to build your CURL call like this:
$data = array(
'Inputs'=> array(
'input1'=> array(
'ColumnNames' => array("header1", "header2", "header3"),
'Values' => array( array("value1" , "value2" , "value3"))
),
),
'GlobalParameters'=> null
);
$body = json_encode($data);
$url = 'your-endpoint-url';
$api_key = 'your-api-key';
$headers = array('Content-Type: application/json', 'Authorization:Bearer ' . $api_key, 'Content-Length: ' . strlen($body));
$this->responseArray['body'] = $body;
$curl = curl_init($url);
curl_setopt($curl, CURLOPT_CUSTOMREQUEST, "POST");
curl_setopt($curl, CURLOPT_POSTFIELDS, $body);
curl_setopt($curl, CURLOPT_HTTPHEADER, $headers);
curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
$result = curl_exec($curl);
Of all the places to get hung up, it was on the GlobalParameters for me, and is for you, too. You need this instead:
GlobalParameters => null
This generates the following JSON
GlobalParameters: {}
whereas
GlobalParameters => array()
gives
GlobalParameters: []
it's a subtle distinction, but enough to make Azure throw a hissy fit.
I didn't test using your curl_setopt functions and instead used what I've included in my example. I'm assuming it will work using the curl_setopts you have, but I don't know for sure.

Answer 2

I had some trouble adapting this perfectly, and I wanted to be able to easily work with JSON & Guzzle. Below is the solution that I built.
First is the function for making the actual call to Azure. Note that Guzzle wants your URL to be split into the domain and URI pieces.
This should all be in your .env file (if you're using Laravel at least).
AZURE_BASE=https://ussouthcentral.services.azureml.net
AZURE_URL=/workspaces/[[YOUR_STUFF_HERE]]/services/[[YOUR_STUFF_HERE]]/execute?api-version=2.0&format=swagger
AZURE_PRIMARY_KEY=[[YOUR_KEY]]
use GuzzleHttp\Client;
public function learn () {
$client = new Client([
'base_uri' => env('AZURE_BASE'),
'timeout' => 2.0,
]);
$headers = [
'Authorization' => 'Bearer ' .env('AZURE_PRIMARY_KEY'),
'Accept' => 'application/json',
'Content-Type' => 'application/json'
];
$data = $this->test_data();
$body = json_encode($data);
$response = $client->request('POST', env('AZURE_URL'), [
'headers' => $headers,
'body' => $body
]);
return $response;
}
As other answers have noted, the data setup is very touchy. new \stdClass is the key here, as we need to end up with a JSON Object ({}) not an array ([]). stdClass creates that empty object for us.
function test_data () {
return array(
'Inputs'=> array(
'input1'=> array(
[
'DESC' => "",
'2-week-total'=> "1",
'last-week'=> "1",
'this-week'=> "1",
'delta'=> "1",
'normalized delta'=> "1",
'delta-percent'=> "1",
'high-total-delta'=> "1",
'high-total-amt'=> "1",
'role'=> ""
]
),
),
'GlobalParameters'=> new \stdClass,
);
}
Now when you call ->learn(), you'll get back some nice JSON to do what you need.