Thursday, November 6, 2014

Authorized Google API access from Python (part 2 of 2)

Listing your files with the Google Drive API

NOTE: You can also watch a video walkthrough of the common code covered in this blogpost here.

UPDATE (Mar 2020): You can build this application line-by-line with our codelab (self-paced, hands-on tutorial) introducing developers to G Suite APIs. The deprecated auth library comment from the previous update below is spelled out in more detail in the green sidebar towards the bottom of step/module 5 (Install the Google APIs Client Library for Python). Also, the code sample is now maintained in a GitHub repo which includes a port to the newer auth libraries so you have both versions to refer to.

UPDATE (Apr 2019): In order to have a closer relationship between the GCP and G Suite worlds of Google Cloud, all G Suite Python code samples have been updated, replacing some of the older G Suite API client libraries with their equivalents from GCP. NOTE: using the newer libraries requires more initial code/effort from the developer thus will seem "less Pythonic." However, we will leave the code sample here with the original client libraries (deprecated but not shutdown yet) to be consistent with the video.

UPDATE (Aug 2016): The code has been modernized to use oauth2client.tools.run_flow() instead of the deprecated oauth2client.tools.run_flow(). You can read more about that change here.

UPDATE (Jun 2016): Updated to Python 2.7 & 3.3+ and Drive API v3.

Introduction

In this final installment of a (currently) two-part series introducing Python developers to building on Google APIs, we'll extend from the simple API example from the first post (part 1) just over a month ago. Those first snippets showed some skeleton code and a short real working sample that demonstrate accessing a public (Google) API with an API key (that queried public Google+ posts). An API key however, does not grant applications access to authorized data.

Authorized data, including user information such as personal files on Google Drive and YouTube playlists, require additional security steps before access is granted. Sharing of and hardcoding credentials such as usernames and passwords is not only insecure, it's also a thing of the past. A more modern approach leverages token exchange, authenticated API calls, and standards such as OAuth2.

In this post, we'll demonstrate how to use Python to access authorized Google APIs using OAuth2, specifically listing the files (and folders) in your Google Drive. In order to better understand the example, we strongly recommend you check out the OAuth2 guides (general OAuth2 info, OAuth2 as it relates to Python and its client library) in the documentation to get started.

The docs describe the OAuth2 flow: making a request for authorized access, having the user grant access to your app, and obtaining a(n access) token with which to sign and make authorized API calls with. The steps you need to take to get started begin nearly the same way as for simple API access. The process diverges when you arrive on the Credentials page when following the steps below.

Google API access

In order to Google API authorized access, follow these instructions (the first three of which are roughly the same for simple API access):
  • Go to the Google Developers Console and login.
    • Use your Gmail or Google credentials; create an account if needed
  • Click "Create a Project" from pulldown under your username (at top)
    • Enter a Project Name (mutable, human-friendly string only used in the console)
    • Enter a Project ID (immutable, must be unique and not already taken)
  • Once project has been created, enable APIs you wish to use
  • Select "Credentials" in left-nav
    • Click "Create credentials" and select OAuth client ID
    • In the new dialog, select your application type — we're building a command-line script which is an "Installed application"
    • In the bottom part of that same dialog, specify the type of installed application; choose "Other" (cmd-line scripts are not web nor mobile)
    • Click "Create Client ID" to generate your credentials
  • Finally, click "Download JSON" to save the new credentials to your computer... perhaps choose a shorter name like "client_secret.json" or "client_id.json"
NOTEs: Instructions from the previous blogpost were to get an API key. This time, in the steps above, we're creating and downloading OAuth2 credentials. You can also watch a video walkthrough of this app setup process of getting simple or authorized access credentials in the "DevConsole" here.

    Accessing Google APIs from Python

    In order to access authorized Google APIs from Python, you still need the Google APIs Client Library for Python, so in this case, do follow those installation instructions from part 1.

    We will again use googleapiclient.discovery.build(), which is required to create a service endpoint for interacting with an API, authorized or otherwise. However, for authorized data access, we need additional resources, namely the httplib2 and oauth2client packages. Here are the first five lines of the new boilerplate code for authorized access:

    from __future__ import print_function
    
    from googleapiclient import discovery
    from httplib2 import Http
    from oauth2client import file, client, tools
    
    SCOPES = # one or more scopes (strings)
    
    SCOPES is a critical variable: it represents the set of scopes of authorization an app wants to obtain (then access) on behalf of user(s). What's does a scope look like?

    Each scope is a single character string, specifically a URL. Here are some examples:
    • 'https://www.googleapis.com/auth/plus.me' — access your personal Google+ settings
    • 'https://www.googleapis.com/auth/drive.metadata.readonly' — read-only access your Google Drive file or folder metadata
    • 'https://www.googleapis.com/auth/youtube' — access your YouTube playlists and other personal information
    You can request one or more scopes, given as a single space-delimited string of scopes or an iterable (list, generator expression, etc.) of strings.  If you were writing an app that accesses both your YouTube playlists as well as your Google+ profile information, your SCOPES variable could be either of the following:
    SCOPES = 'https://www.googleapis.com/auth/plus.me https://www.googleapis.com/auth/youtube'

    That is space-delimited and made tiny by me so it doesn't wrap in a regular-sized browser window; or it could be an easier-to-read, non-tiny, and non-wrapped tuple:

    SCOPES = (
        'https://www.googleapis.com/auth/plus.me',
        'https://www.googleapis.com/auth/youtube',
    )

    Our example command-line script will just list the files on your Google Drive, so we only need the read-only Drive metadata scope, meaning our SCOPES variable will be just this:
    SCOPES = 'https://www.googleapis.com/auth/drive.metadata.readonly'
    The next section of boilerplate represents the security code:
    store = file.Storage('storage.json')
    creds = store.get()
    if not creds or creds.invalid:
        flow = client.flow_from_clientsecrets('client_secret.json', SCOPES)
        creds = tools.run_flow(flow, store)
    
    Once the user has authorized access to their personal data by your app, a special "access token" is given to your app. This precious resource must be stored somewhere local for the app to use. In our case, we'll store it in a file called "storage.json". The lines setting the store and creds variables are attempting to get a valid access token with which to make an authorized API call.

    If the credentials are missing or invalid, such as being expired, the authorization flow (using the client secret you downloaded along with a set of requested scopes) must be created (by client.flow_from_clientsecrets()) and executed (by tools.run_flow()) to ensure possession of valid credentials. The client_secret.json file is the credentials file you saved when you clicked "Download JSON" from the DevConsole after you've created your OAuth2 client ID.

    If you don't have credentials at all, the user much explicitly grant permission — I'm sure you've all seen the OAuth2 dialog describing the type of access an app is requesting (remember those scopes?). Once the user clicks "Accept" to grant permission, a valid access token is returned and saved into the storage file (because you passed a handle to it when you called tools.run_flow()).

    Note: tools.run() deprecated by tools.run_flow()
    You may have seen usage of the older tools.run() function, but it has been deprecated by tools.run_flow(). We explain this in more detail in another blogpost specifically geared towards migration.

    Once the user grants access and valid credentials are saved, you can create one or more endpoints to the secure service(s) desired with googleapiclient.discovery.build(), just like with simple API access. Its call will look slightly different, mainly that you need to sign your HTTP requests with your credentials rather than passing an API key:

    DRIVE = discovery.build(API, VERSION, http=creds.authorize(Http()))

    In our example, we're going to list your files and folders in your Google Drive, so for API, use the string 'drive'. The API is currently on version 3 so use 'v3' for VERSION:

    DRIVE = discovery.build('drive', 'v3', http=creds.authorize(Http()))

    If you want to get comfortable with OAuth2, what it's flow is and how it works, we recommend that you experiment at the OAuth Playground. There you can choose from any number of APIs to access and experience first-hand how your app must be authorized to access personal data.

    Going back to our working example, once you have an established service endpoint, you can use the list() method of the files service to request the file data:

    files = DRIVE.files().list().execute().get('files', [])

    If there's any data to read, the response dict will contain an iterable of files that we can loop over (or default to an empty list so the loop doesn't fail), displaying file names and types:

    for f in files:
        print(f['name'], f['mimeType'])

    Conclusion

    To find out more about the input parameters as well as all the fields that are in the response, take a look at the docs for files().list(). For more information on what other operations you can execute with the Google Drive API, take a look at the reference docs and check out the companion video for this code sample. Don't forget the codelab and this sample's GitHub repo. That's it!

    Below is the entire script for your convenience:
    '''
    drive_list.py -- Google Drive API demo; maintained at:
        http://github.com/googlecodelabs/gsuite-apis-intro
    '''
    from __future__ import print_function
    
    from googleapiclient import discovery
    from httplib2 import Http
    from oauth2client import file, client, tools
    
    SCOPES = 'https://www.googleapis.com/auth/drive.readonly.metadata'
    store = file.Storage('storage.json')
    creds = store.get()
    if not creds or creds.invalid:
        flow = client.flow_from_clientsecrets('client_secret.json', SCOPES)
        creds = tools.run_flow(flow, store)
    
    DRIVE = discovery.build('drive', 'v3', http=creds.authorize(Http()))
    files = DRIVE.files().list().execute().get('files', [])
    for f in files:
        print(f['name'], f['mimeType'])
    
    When you run it, you should see pretty much what you'd expect, a list of file or folder names followed by their MIMEtypes — I named my script drive_list.py:
    $ python3 drive_list.py
    Google Maps demo application/vnd.google-apps.spreadsheet
    Overview of Google APIs - Sep 2014 application/vnd.google-apps.presentation
    tiresResearch.xls application/vnd.google-apps.spreadsheet
    6451_Core_Python_Schedule.doc application/vnd.google-apps.document
    out1.txt application/vnd.google-apps.document
    tiresResearch.xls application/vnd.ms-excel
    6451_Core_Python_Schedule.doc application/msword
    out1.txt text/plain
    Maps and Sheets demo application/vnd.google-apps.spreadsheet
    ProtoRPC Getting Started Guide application/vnd.google-apps.document
    gtaskqueue-1.0.2_public.tar.gz application/x-gzip
    Pull Queues application/vnd.google-apps.folder
    gtaskqueue-1.0.1_public.tar.gz application/x-gzip
    appengine-java-sdk.zip application/zip
    taskqueue.py text/x-python-script
    Google Apps Security Whitepaper 06/10/2010.pdf application/pdf
    
    Obviously your output will be different, depending on what files are in your Google Drive. But that's it... hope this is useful. You can now customize this code for your own needs and/or to access other Google APIs. Thanks for reading!

    EXTRA CREDIT: To test your skills, add functionality to this code that also displays the last modified timestamp, the file (byte)size, and perhaps shave the MIMEtype a bit as it's slightly harder to read in its entirety... perhaps take just the final path element? One last challenge: in the output above, we have both Microsoft Office documents as well as their auto-converted versions for Google Apps... perhaps only show the filename once and have a double-entry for the filetypes!

    4 comments:

    1. I just tried this, but get: Instance of 'Resource' has no 'files' member

      ReplyDelete
      Replies
      1. Usually an error like this occurs when running PyLint or any other type of analyzer that attempts to look for issues with your code. If you just execute the script, and obtain the proper permissions, this error shouldn't occur.

        Delete
    2. Hello,
      I am the beginner on the google API and I have no knowledge of this application and more software Source code, python ....

      To date, I have developed a simple procedure unsing google API and HTML: User input screen and then update the User signature on the signature parameter.

      In my procedure, if I put "me" with "people API" (to fetch a user Google photo) and the "setting" (for updating the signature format HTML), it works during my test.

      Then I developed a procedure for sending HTML format to all users, it doesn't work! I had an error message "delegation ...." This should be avoided because the number of my colleagues is more than 40 people! ...
      How to do ?

      If you have a simple procedure as an example, would it be possible to email it me privately? because, I can't really find a simple solution !!!...and more, I read in the site internet " GITHUB". there is many different files ( Cs, ymd, etc ... I don't know these extensions )...
      it's too complicated !

      thank you in advance for this helpness

      ReplyDelete
      Replies
      1. Unfortunately I'm not familar with the Google People API. Please check its documentation to look for a similar "getting started" sample.

        Delete