Develop Reference

How to refer a local file for Azure emotion api

I am trying to refer a local jpg file for using in Azure Emotion API.
To do this, I refer my file through "file:///" like below.
body = "{'url': 'file:///Users/jonghkim/dev_jhk/Research/Crowdfunding/Face_Analysis/me.jpg'}"
But the response says "Invalid image URL." How could I fix it?
{"error":{"code":"InvalidUrl","message":"Invalid image URL."}}
Whole code looks like below.
########### Python 2.7 #############
import httplib, urllib, base64
headers = {
# Request headers. Replace the placeholder key below with your subscription key.
'Content-Type': 'application/json',
'Ocp-Apim-Subscription-Key': '***********************',
}
params = urllib.urlencode({
})
# Replace the example URL below with the URL of the image you want to analyze.
body = "{'url': 'file:///Users/jonghkim/dev_jhk/Research/Crowdfunding/Face_Analysis/me.jpg'}"
try:
conn = httplib.HTTPSConnection('westus.api.cognitive.microsoft.com')
conn.request("POST", "/emotion/v1.0/recognize?%s" % params, body, headers)
response = conn.getresponse()
data = response.read()
print(data)
conn.close()
except Exception as e:
print("[Errno {0}] {1}".format(e.errno, e.strerror))

I solved this problem. The true reason was two fold. At first, when we refer local file, we should use 'Content-Type': 'application/octet-stream' in a header.
The second problem is that the image should satisfy the condition of Azure (learn.microsoft.com/ko-kr/azure/cognitive-services/emotion/fa‌​q).
Full code is here:
########### Python 2.7 #############
import httplib, urllib, base64
headers = {
# Request headers. Replace the placeholder key below with your subscription key.
'Content-Type': 'application/octet-stream',
'Ocp-Apim-Subscription-Key': '**************************',
}
params = urllib.urlencode({
})
# Replace the example URL below with the URL of the image you want to analyze.
body = open('test.jpg','rb').read()
conn = httplib.HTTPSConnection('westus.api.cognitive.microsoft.com')
conn.request("POST", "/emotion/v1.0/recognize?%s" % params, body, headers)
response = conn.getresponse()
data = response.read()
print(data)
conn.close()

You're executing the emotion API within Cognitive Services - just take a look at the URI. This code is not being executed locally. It's a service. Being run somewhere else.
So, when the service gets the URL (via url in the body), it then needs to reach out to that resource, which is impossible to do if the resource is on your computer. And file:// is going to be an invalid scheme because the service won't be reading from its own file system.
You'll need to have your resource in an accessible place (e.g. a public or SAS-signed blob, an image link from a website, etc).

API access to trading platform using Python

I'm new to getting data using API and Python. I want to pull data from my trading platform. They've provided the following instructions:
http://www.questrade.com/api/documentation/getting-started
I'm ok up to step 4 and have an access token. I need help with step 5. How do I translate this request:
GET /v1/accounts HTTP/1.1
Host: https://api01.iq.questrade.com
Authorization: Bearer C3lTUKuNQrAAmSD/TPjuV/HI7aNrAwDp
into Python code? I've tried
import requests
r = requests.get('https://api01.iq.questrade.com/v1/accounts', headers={'Authorization': 'access_token myToken'})
I tried that after reading this: python request with authentication (access_token)
Any help would be appreciated. Thanks.

As you point out, after step 4 you should have received an access token as follows:
{
“access_token”: ”C3lTUKuNQrAAmSD/TPjuV/HI7aNrAwDp”,
“token_type”: ”Bearer”,
“expires_in”: 300,
“refresh_token”: ”aSBe7wAAdx88QTbwut0tiu3SYic3ox8F”,
“api_server”: ”https://api01.iq.questrade.com”
}
To make subsequent API calls, you will need to construct your URI as follows:
uri = [api_server]/v1/[rest_operation]
e.g.
uri = "https://api01.iq.questrade.com/v1/time"
Note: Make sure you use the same [api_server] that you received in your json object from step 4, otherwise your calls will not work with the given access_token
Next, construct your headers as follows:
headers = {'Authorization': [token_type] + ' ' + [access_token]}
e.g.
headers = {'Authorization': 'Bearer C3lTUKuNQrAAmSD/TPjuV/HI7aNrAwDp'}
Finally, make your requests call as follows
r = requests.get(uri, headers=headers)
response = r.json()
Hope this helps!
Note: You can find a Questrade API Python wrapper on GitHub which handles all of the above for you.
https://github.com/pcinat/QuestradeAPI_PythonWrapper

Improving a bit on Peter's reply (Thank you Peter!)
start by using the token you got from the QT website to obtain an access_token and get an api_server assigned to handle your requests.
# replace XXXXXXXX with the token given to you in your questrade account
import requests
r = requests.get('https://login.questrade.com/oauth2/token?grant_type=refresh_token&refresh_token=XXXXXXXX')
access_token = str(r.json()['access_token'])
refresh_token= str(r.json()['refresh_token']) # you will need this refresh_token to obtain another access_token when it expires
api_server= str(r.json()['api_server'])
token_type= str(r.json()['token_type'])
api_server= str(r.json()['api_server'])
expires_in = str(r.json()['expires_in'])
# uri = api_server+'v1/'+[action] - let's try checking the server's time:
uri = api_server+'v1/'+'time'
headers = {'Authorization': token_type +' '+access_token}
# will look sth like this
# headers will look sth like {'Authorization': 'Bearer ix7rAhcXx83judEVUa8egpK2JqhPD2_z0'}
# uri will look sth like 'https://api05.iq.questrade.com/v1/time'
# you can test now with
r = requests.get(uri, headers=headers)
response = r.json()
print(response)

Using Python to test HTTP APIs

I'm fairly new to Python programming and I don't know all the libraries needed for the following.
I would like to use Python to test some HTTP APIs. Mainly I want to use OAuth and make a few JSON calls. The APIs in question can be found on: https://developers.trustpilot.com/authentication and the generate product review link (I can only use one link)
I want to authenticate myself and then generate a product review link in one step. So far I've been using the Advanced REST client (ARC) to make these calls individually. I could also use .arc files if you think it's easier.
The idea would be make these calls successively in one go. So it would be something along the lines:
1) Make the authentication call.
The HTTP Method looks like this:
https://api.trustpilot.com/v1/oauth/oauth-business-users-for-applications/accesstoken
Method Post:
Header
Authorization: Basic Base64encode(APIkey:Secret)
Content-Type: application/x-www-form-urlencoded
Payload:
grant_type=password&username=user#mail.com&password=SomePass
Translate this bit into Python basically.
1.a) Add a header to the call
Header Authorization: base64encode hash Content-Type: application/x-www-form-urlencoded
1.b) Add a payload to the call
Payload: grant_type=password&username
4) Receive the token from call made in step 1) (Result is format)
"access token": Auth_token
5) Take the token and use it in creating a product review.
5.a) Add the token in the header
Header: Authorization: Bearer Auth_token
6.a) Add a JSON payload to the call made in step 5.
Here's the code I have so far:
Import requests
header = {'Authorization: Basic NnNrQUprTWRHTU5VSXJGYXBVRGxack1oT01oTUFRZHI6QTFvOGJjRUNDdUxBTmVqUQ==}','Content-Type: application/x-www-form-urlencoded'}
payload = {'grant_type=password&username=email#address.com&password=SomePassword'}
r = requests.post('https://api.trustpilot.com/v1/oauth/oauth-business-users-for-applications/accesstoken', headers=header, params=payload )
Ideally I want to create the requests.post(url, header, payload) and then return what the server answers in JSON format. I think that print r.text would do the last part.
So this is the code I have writtent (that works now):
import requests
import getpass
import json
from requests.auth import HTTPBasicAuth
header = {'grant_type':'password' , 'username':'mail#maildomain.com', 'password':'YourPassword'}
username= "YOURAPIKEY" #APIKey
password= "YOURSECRET" #Secret
res = requests.post(
'URL/v1/oauth/oauth-business-users-for-applications/accesstoken',
auth=HTTPBasicAuth(username, password), # basic authentication
data=header)
#print(res.content) #See content of the call result.
data = res.json() # get response as parsed json (will return a dict)
auth_token = data.get('access_token')

requests can do all what you ask without any work from your part.
See the doc for authentication, parameters, json output, json input
Make the authentication call.
import requests
import getpass
from requests.auth import HTTPBasicAuth
username = raw_input('Username: ')
password = getpass.getpass('Password: ')
res = requests.post(
'https://api.trustpilot.com/v1/oauth/oauth-business-users-for-applications/accesstoken',
auth=HTTPBasicAuth(username, password), # basic authentication
params={ # url parameters
'grant_type': 'password',
'username': 'email#address.com',
'password': 'SomePassword'
})
Receive the token from call made in step 1) (Result is format)
# res = requests.post.....
data = res.json() # get response as parsed json (will return a dict)
auth_token = data.get('access token')
Take the token and use it in creating a product review.
request.post(
'.../product_review',
headers={
'Authorization': 'Bearer ' + auth_token
},
json={'my': 'payload'}) # send data as json

How to use a refresh_token to get a new access_token (using Flask-OAuthLib)?

I'm building a website + backend with the FLask Framework in which I use Flask-OAuthlib to authenticate with google. After authentication, the backend needs to regularly scan the user his Gmail. So currently users can authenticate my app and I store the access_token and the refresh_token. The access_token expires after one hour, so within that one hour I can get the userinfo like so:
google = oauthManager.remote_app(
'google',
consumer_key='xxxxxxxxx.apps.googleusercontent.com',
consumer_secret='xxxxxxxxx',
request_token_params={
'scope': ['https://www.googleapis.com/auth/userinfo.email', 'https://www.googleapis.com/auth/gmail.readonly'],
'access_type': 'offline'
},
base_url='https://www.googleapis.com/oauth2/v1/',
request_token_url=None,
access_token_method='POST',
access_token_url='https://accounts.google.com/o/oauth2/token',
authorize_url='https://accounts.google.com/o/oauth2/auth'
)
token = (the_stored_access_token, '')
userinfoObj = google.get('userinfo', token=token).data
userinfoObj['id'] # Prints out my google id
Once the hour is over, I need to use the refresh_token (which I've got stored in my database) to request a new access_token. I tried replacing the_stored_access_token with the_stored_refresh_token, but this simply gives me an Invalid Credentials-error.
In this github issue I read the following:
regardless of how you obtained the access token / refresh token (whether through an authorization code grant or resource owner password credentials), you exchange them the same way, by passing the refresh token as refresh_token and grant_type set to 'refresh_token'.
From this I understood I had to create a remote app like so:
google = oauthManager.remote_app(
'google',
# also the consumer_key, secret, request_token_params, etc..
grant_type='refresh_token',
refresh_token=u'1/xK_ZIeFn9quwvk4t5VRtE2oYe5yxkRDbP9BQ99NcJT0'
)
But this leads to a TypeError: __init__() got an unexpected keyword argument 'refresh_token'. So from here I'm kinda lost.
Does anybody know how I can use the refresh_token to get a new access_token? All tips are welcome!

This is how I get a new access_token for google:
from urllib2 import Request, urlopen, URLError
from webapp2_extras import json
import mimetools
BOUNDARY = mimetools.choose_boundary()
def refresh_token()
url = google_config['access_token_url']
headers = [
("grant_type", "refresh_token"),
("client_id", <client_id>),
("client_secret", <client_secret>),
("refresh_token", <refresh_token>),
]
files = []
edata = EncodeMultiPart(headers, files, file_type='text/plain')
headers = {}
request = Request(url, headers=headers)
request.add_data(edata)
request.add_header('Content-Length', str(len(edata)))
request.add_header('Content-Type', 'multipart/form-data;boundary=%s' % BOUNDARY)
try:
response = urlopen(request).read()
response = json.decode(response)
except URLError, e:
...
EncodeMultipart function is taken from here:
https://developers.google.com/cloud-print/docs/pythonCode
Be sure to use the same BOUNDARY

Looking at the source code for OAuthRemoteApp. The constructor does not take a keyword argument called refresh_token. It does however take an argument called access_token_params which is an optional dictionary of parameters to forward to the access token url.
Since the url is the same, but the grant type is different. I imagine a call like this should work:
google = oauthManager.remote_app(
'google',
# also the consumer_key, secret, request_token_params, etc..
grant_type='refresh_token',
access_token_params = {
refresh_token=u'1/xK_ZIeFn9quwvk4t5VRtE2oYe5yxkRDbP9BQ99NcJT0'
}
)

flask-oauthlib.contrib contains an parameter named auto_refresh_url / refresh_token_url in the remote_app which does exactely what you wanted to wanted to do. An example how to use it looks like this:
app= oauth.remote_app(
[...]
refresh_token_url='https://www.douban.com/service/auth2/token',
authorization_url='https://www.douban.com/service/auth2/auth',
[...]
)
However I did not manage to get it running this way. Nevertheless this is possible without the contrib package. My solution was to catch 401 API calls and redirect to a refresh page if a refresh_token is available.
My code for the refresh endpoint looks as follows:
#app.route('/refresh/')
def refresh():
data = {}
data['grant_type'] = 'refresh_token'
data['refresh_token'] = session['refresh_token'][0]
data['client_id'] = CLIENT_ID
data['client_secret'] = CLIENT_SECRET
# make custom POST request to get the new token pair
resp = remote.post(remote.access_token_url, data=data)
# checks the response status and parses the new tokens
# if refresh failed will redirect to login
parse_authorized_response(resp)
return redirect('/')
def parse_authorized_response(resp):
if resp is None:
return 'Access denied: reason=%s error=%s' % (
request.args['error_reason'],
request.args['error_description']
)
if isinstance(resp, dict):
session['access_token'] = (resp['access_token'], '')
session['refresh_token'] = (resp['refresh_token'], '')
elif isinstance(resp, OAuthResponse):
print(resp.status)
if resp.status != 200:
session['access_token'] = None
session['refresh_token'] = None
return redirect(url_for('login'))
else:
session['access_token'] = (resp.data['access_token'], '')
session['refresh_token'] = (resp.data['refresh_token'], '')
else:
raise Exception()
return redirect('/')
Hope this will help. The code can be enhanced of course and there surely is a more elegant way than catching 401ers but it's a start ;)
One other thing: Do not store the tokens in the Flask Session Cookie. Rather use Server Side Sessions from "Flask Session" which I did in my code!

This is how i got my new access token.
from urllib2 import Request, urlopen, URLError
import json
import mimetools
BOUNDARY = mimetools.choose_boundary()
CRLF = '\r\n'
def EncodeMultiPart(fields, files, file_type='application/xml'):
"""Encodes list of parameters and files for HTTP multipart format.
Args:
fields: list of tuples containing name and value of parameters.
files: list of tuples containing param name, filename, and file contents.
file_type: string if file type different than application/xml.
Returns:
A string to be sent as data for the HTTP post request.
"""
lines = []
for (key, value) in fields:
lines.append('--' + BOUNDARY)
lines.append('Content-Disposition: form-data; name="%s"' % key)
lines.append('') # blank line
lines.append(value)
for (key, filename, value) in files:
lines.append('--' + BOUNDARY)
lines.append(
'Content-Disposition: form-data; name="%s"; filename="%s"'
% (key, filename))
lines.append('Content-Type: %s' % file_type)
lines.append('') # blank line
lines.append(value)
lines.append('--' + BOUNDARY + '--')
lines.append('') # blank line
return CRLF.join(lines)
def refresh_token():
url = "https://oauth2.googleapis.com/token"
headers = [
("grant_type", "refresh_token"),
("client_id", "xxxxxx"),
("client_secret", "xxxxxx"),
("refresh_token", "xxxxx"),
]
files = []
edata = EncodeMultiPart(headers, files, file_type='text/plain')
#print(EncodeMultiPart(headers, files, file_type='text/plain'))
headers = {}
request = Request(url, headers=headers)
request.add_data(edata)
request.add_header('Content-Length', str(len(edata)))
request.add_header('Content-Type', 'multipart/form-data;boundary=%s' % BOUNDARY)
response = urlopen(request).read()
print(response)
refresh_token()
#response = json.decode(response)
#print(refresh_token())

With your refresh_token, you can get a new access_token like:
from google.oauth2.credentials import Credentials
from google.auth.transport import requests
creds = {"refresh_token": "<goes here>",
"token_uri": "https://accounts.google.com/o/oauth2/token",
"client_id": "<YOUR_CLIENT_ID>.apps.googleusercontent.com",
"client_secret": "<goes here>",
"scopes": ["https://www.googleapis.com/auth/userinfo.email"]}
cred = Credentials.from_authorized_user_info(creds)
cred.refresh(requests.Request())
my_new_access_token = cred.token

No Content-Disposition header in response from mechanize

I have to download the csv file from AdWords Billing page during the Celery task. And I have no idea what's wrong with my implementation, so need your help.
Log in:
browser = mechanize.Browser()
browser.open('https://accounts.google.com/ServiceLogin')
browser.select_form(nr=0)
browser['Email'] = g_email
browser['Passwd'] = g_password
browser.submit()
browser.set_handle_robots(False)
billing_resp = browser.open('https://adwords.google.com/')
It's OK, I'm on the billing page now. Next, I've parsed the result page for token and ids, analyzed request headers and action url in Chrome debugger and now I want to make POST request and receive my csv file. Response headers (in Chrome) are:
content-disposition:attachment; filename="myclientcenter.csv.gz"
content-length:307479
content-type:application/x-gzip; charset=UTF-8
With mechanize:
data = {
'__u': effectiveUserId,
'__c': customerId,
'token': token,
}
browser.addheaders = [
('accept', 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8'),
('content-type', 'application/x-www-form-urlencoded'),
("accept-encoding", "gzip,deflate,sdch"),
('user-agent', "Mozilla/5.0"),
('referer', "https://adwords.google.com/mcm/Mcm?__u=8183865359&__c=3069937889"),
('origin', "https://adwords.google.com"),
]
browser.set_handle_refresh(True)
browser.set_debug_responses(True)
browser.set_debug_redirects(True)
browser.set_handle_referer(True)
browser.set_debug_http(True)
browser.set_handle_equiv(True)
browser.set_handle_gzip(True)
response = browser.open(
'https://adwords.google.com/mcm/file/ClientSummary/',
data='&'.join(['='.join(pair) for pair in data.items()]),
)
BUT! The Content-Length header is 0 and no Content-Disposition in this response. Why? And what can I do to make it work?
Had a try to use Requests, but could not even pass the login stage...

I have the answer on my own question now (thanks for my team lead).
The main mistake is in this incorrect request data:
data = {
'__u': effectiveUserId,
'__c': customerId,
'token': token,
}
Let's try again, with proper solution.
# Open Google login page and log in.
browser = mechanize.Browser()
try:
browser.open('https://accounts.google.com/ServiceLogin')
browser.select_form(nr=0)
browser['Email'] = 'email#adwords.login'
browser['Passwd'] = 'password'
browser.submit()
except HTTPError:
raise AdWordsException("Can't find the Google login form")
We are logged in now and can go deeper.
try:
browser.set_handle_robots(False)
billing_resp = browser.open('https://adwords.google.com/')
except HTTPError:
raise AdWordsException("Can't open AdWords dashboard page")
# Welcome to the AdWords billing dashboard. We can get
# session-unique token from this page for the further POST-request
token_re = re.search(r"token:\'(.{41})\'", billing_resp.read())
if token_re is None:
raise AdWordsException("Can't parse the token")
# It's time for some magic now. We have to construct proper mcsSelector
# serialized data structure. This is GWT-RPC wire protocol hell.
# Paste your specific version from web debugger.
MCS_TEMPLATE = (
"7|0|49|https://adwords.google.com/mcm/gwt/|18FBB090A5C26E56AC16C9DF0689E720|"
"com.google.ads.api.services.common.selector.Selector/1054041135|"
"com.google.ads.api.services.common.date.DateRange/1118087507|"
"com.google.ads.api.services.common.date.Date/373224763|"
"java.util.ArrayList/4159755760|java.lang.String/2004016611|ClientName|"
"ExternalCustomerId|PrimaryUserLogin|PrimaryCompanyName|IsManager|"
"SalesChannel|Tier|AccountSettingTypes|Labels|Alerts|CostWithCurrency|"
"CostUsd|Clicks|Impressions|Ctr|Conversions|ConversionRate|SearchCtr|"
"ContentCtr|BudgetAmount|BudgetStartDate|BudgetEndDate|BudgetPercentSpent|"
"BudgetType|RemainingBudget|ClientDateTimeZoneId|"
"com.google.ads.api.services.common.selector.OrderBy/524388450|"
"SearchableData|"
"com.google.ads.api.services.common.sorting.SortOrder/2037387810|"
"com.google.ads.api.services.common.pagination.Paging/363399854|"
"com.google.ads.api.services.common.selector.Predicate/451365360|"
"SeedObfuscatedCustomerId|"
"com.google.ads.api.services.common.selector.Predicate$Operator/2293561107|"
"java.util.Arrays$ArrayList/2507071751|[Ljava.lang.String;/2600011424|"
"3069937889|ExcludeSeeds|true|ClientTraversal|DIRECT|"
"com.google.ads.api.services.common.selector.Summary/3224078220|included|1|"
"2|3|4|5|"
"{report_date}|5|{report_date}" # take a note of this
"|6|26|7|8|7|9|7|10|7|11|7|12|7|13|7|14|7|15|7|16|7|17|7|18|7|19|7|20|7|21|"
"7|22|7|23|7|24|7|25|7|26|7|27|7|28|7|29|7|30|7|31|7|32|7|33|6|0|0|0|6|2|34|"
"35|36|0|34|9|-35|37|100|0|6|0|6|3|38|39|40|2|41|42|1|43|38|44|40|0|41|42|1|"
"45|38|46|-45|41|42|1|47|0|0|6|0|6|1|48|6|0|49|6|0|0|"
)
# To take stats for today
report_date = datetime.date.today()
mcs_selector = MCS_TEMPLATE.format(
report_date='%s|%s|%s' % (
report_date.day,
report_date.month,
report_date.year
),
)
data = urllib.urlencode({
'token': token_re.group(1),
'mcsSelector': mcs_selector,
})
# And... it finally works! Token and proper mcsSelector is all we need.
# POST-request with this data returns zipped csv file for us with
# current balance state and another info that's not available via AdWords API
zipped_csv = browser.open(
'https://adwords.google.com/mcm/file/ClientSummary',
data=data
)
# Unpack it and use as you wish.
with gzip.GzipFile(mode='r', fileobj=zipped_csv) as csv_io:
try:
csv = StringIO.StringIO(csv_io.read())
except IOError:
raise AdWordsException("Can't get CSV file from response")
finally:
browser.close()