Just finished another Chrome extension

Posted May 18th, 2017 in Javascript by Florentin

Clippy is a Chrome extension which allows users to tag any url in a web page so we can later recognize them.
You may read more about it here Clippy

Object attribute lookup in Python

Posted September 4th, 2015 in Uncategorized by Florentin



  • https://docs.python.org/2/howto/descriptor.html
  • http://stackoverflow.com/questions/3798835/understanding-get-and-set-and-python-descriptors

python 2 docs

  • https://docs.python.org/2/reference/datamodel.html?highlight=metaclass

great insights

  • http://blog.ionelmc.ro/2015/02/09/understanding-python-metaclasses/

__mro__ (Method Resolution Order)

  • http://stackoverflow.com/questions/2010692/what-does-mro-do-in-python
  • http://stackoverflow.com/questions/1848474/method-resolution-order-mro-in-new-style-python-classes

simulate a __getattribute__ call

  • http://codereview.stackexchange.com/questions/92497/simulate-object-getattribute-in-python


  • Everything in Python is an object (i.e. classes, modules, the numbers, the strings, etc)
  • A class is also an object
  • Every object is an instance of a class (example: isinstance(5, int) )
  • Because of that, every class is an instance of a special kind of class called metaclass
  • An instance is created by calling a class object
  • A non-data descriptor is an object following the data descriptor protocol as described in the docs
  • A data descriptor is a descriptor which defined both the __set__ AND __get__ methods
  • __mro__ is a tuple of classes that are considered when looking for base classes during method resolution

Code snippet

demo code

Instance attribute look up

Manual excerpt

The implementation works through a precedence chain that gives data descriptors priority over instance variables, instance variables priority over non-data descriptors, and assigns lowest priority to getattr() if provided.

Given a Class “C” and an Instance “c” where “c = C(…)”, calling “c.name” means looking up an Attribute “name” on the Instance “c” like this:

  • Get the Class from Instance
  • Call the Class’s special method __getattribute__. All objects have a default __getattribute__

Inside __getattribute__

  • Get the Class’s __mro__ as ClassParents
  • For each ClassParent in ClassParents
    • If the Attribute is in the ClassParent’s __dict__
    • If is a data descriptor
      • Return the result from calling the data descriptor’s special method __get__()
    • Break the for each (do not continue searching the same Attribute any further)
  • If the Attribute is in Instance’s __dict__
    • Return the value as it is (even if the value is a data descriptor)
  • For each ClassParent in ClassParents
    • If the Attribute is in the ClassParent’s __dict__
    • If is a non-data descriptor
      • Return the result from calling the non-data descriptor’s special method __get__()
    • If it is NOT a descriptor
      • Return the value
  • If Class has the special method __getattr__
    • Return the result from calling the Class’s special method __getattr__.
  • Raises an AttributeError

Things to remember (from the manual)

  • descriptors are invoked by the getattribute() method
  • overriding getattribute() prevents automatic descriptor calls
  • getattribute() is only available with new style classes and objects
  • object.getattribute() and type.getattribute() make different calls to get().
  • data descriptors always override instance dictionaries.
  • non-data descriptors may be overridden by instance dictionaries.

Class attribute look up

Given a MetaClass “M” and a Class “C” instance of the Metaclass “M”, calling “C.name” means looking up an Attribute “name” on the Class “C” like this:

  • Get the Metaclass from Class
  • Call the Metaclass’s special method __getattribute__

Inside __getattribute__

  • Get the Metaclass’s __mro__ as MetaParents
  • For each MetaParent in MetaParents
    • If the Attribute is in the MetaParent’s __dict__
    • If is a data descriptor
      • Return the result from calling the data descriptor’s special method __get__()
    • Break the for each
  • Get the Class’s __mro__ as ClassParents
  • For each ClassParent in ClassParents
    • If the Attribute is in the ClassParent’s __dict__
    • If is a (data or non-data) descriptor
      • Return the result from calling the descriptor’s special method __get__()
    • Else
      • Return the value
  • For each MetaParent in MetaParents
    • If the Attribute is in the MetaParent’s __dict__
    • If is a non-data descriptor
      • Return the result from calling the non-data descriptor’s special method __get__()
    • If it is NOT a descriptor
      • Return the value
  • If MetaClass has the special method __getattr__
    • Return the result from calling the MetaClass’s special method __getattr__.
  • Raises an AttributeError

A guide to using Google’s API Client Library for JavaScript with Sheets API

Posted August 12th, 2015 in Javascript, Uncategorized by Florentin

Google allow users to interact with Google Apps (like Calendar, Contacts, Drive) though APIs.
Working with the API Client Library for JavaScript in a browser environment presents it’s challenges as described below.


I’m making a single page application (one html file) to test the Javascript library capabilities for using it with Google Sheets Api.
Wish to demo the following:
* create a new spreadsheet
* search for the newly created file by name
* read rows and columns
* insert new rows

Working with private data requires setting up a project at https://console.developers.google.com/project
Read the chapter “Get access keys for your application” from https://developers.google.com/api-client-library/javascript/start/start-js
Make sure to add “Drive API” here https://console.developers.google.com/project/PROJECT_KEY/apiui/apis/enabled
After “Create new client id” steps, write down the “Client ID” and make sure that JavaScript origins matches the demo server (in my case it was http://localhost:8888)

I’ll start building the demo with the help of these resources:



The demo

This depends on “client.js” which is a single line js file like this: window.clientId = ‘aaa-bbbcccddd.apps.googleusercontent.com';

Here’s the file on GITHUB: https://github.com/florentin/google-sheets-api

Please read the code comments, they are important.

<!DOCTYPE html>
	  <meta charset='utf-8' />
		<h1>Hello Google Sheets API</h1>
		<em>Open the console and watch for errors and debug messages.</em>
		<div id="step1">
		    <h2>Step 1: Authorize this app with your Google account</h2>
		    <span id="authorize-status"></span>
		    <button id="authorize-button" style="visibility: hidden">Authorize</button>
		<div id="step2">
            <h2>Step 2: Create a Spreadsheet document in Google Drive</h2>
            <span id="create-status"></span>
            <button id="create-button">Create file in Drive</button>
            <em>Add a file named "blocky" to Google Drive</em>
        <div id="step3">
            <h2>Step 3: Search spreadsheet files by the name "blocky"</h2>
            <span id="list-status"></span>
            <button id="list-button">Search files (json)</button>
        <div id="step4">
            <h2>Step 4: Find the first file named "blocky" and retrieve the worksheets</h2>
            <span id="worksheets-status"></span>
            <button id="worksheets-button">Retrieve worksheets</button>
        <div id="step5">
            <h2>Step 5: Add rows to the first worksheet from the spreadsheet "blocky"</h2>
            <span id="rows-status"></span>
            <button id="rows-button">Add rows</button>
		<script src="client.js"></script>
        <script type="text/javascript">
		var scopes = 'https://spreadsheets.google.com/feeds',
		    headers = {
				accept_json: {'Accept': 'application/json'},
				accept_xml: {'Accept': 'application/atom+xml'},
				send_json: {'Content-Type': 'application/json'},
				send_xml: {'Content-Type': 'application/atom+xml'}
		    authorizeButton = document.getElementById('authorize-button'),
		    authorizeStatus = document.getElementById('authorize-status'),
	        createButton = document.getElementById('create-button'),
	        listButton = document.getElementById('list-button'),
	        worksheetsButton = document.getElementById('worksheets-button'),
	        rowsButton = document.getElementById('rows-button');

		function handleClientLoad() {
			createButton.onclick = handleCreateClick;
			listButton.onclick = handleListClick;
			worksheetsButton.onclick = handleWorksheetsClick;
			rowsButton.onclick = handleRowClick;
		function checkAuth() {
		    gapi.auth.authorize({client_id: clientId, scope: scopes, immediate: true}, handleAuthResult);
		function handleAuthResult(authResult) {
			if (authResult && !authResult.error) {
				// this app is authorised
				authorizeButton.style.visibility = 'hidden';
				authorizeStatus.innerHTML = 'Is Authorised';
			} else {
				authorizeButton.style.visibility = '';
				authorizeButton.onclick = handleAuthClick;
				authorizeStatus.innerHTML = 'Not Authorised';
		function handleAuthClick(event) {
			gapi.auth.authorize({client_id: clientId, scope: scopes, immediate: false}, handleAuthResult);
		    return false;
		function request(method, url, post_data, sucess_call, headers) {
			var oauthToken = gapi.auth.getToken(),
			    xhr = new XMLHttpRequest();
			console.info('request url', url);
			xhr.open(method, url);
			xhr.setRequestHeader('Authorization', 'Bearer ' + oauthToken.access_token);
			//xhr.setRequestHeader("GData-Version", "3.0");
			if(headers !== undefined) {
				for(var key in headers) {
					xhr.setRequestHeader(key, headers[key]);
			xhr.onload = function() {
			    console.info('response xhr', xhr);
			    if(typeof sucess_call === "function")
			xhr.onerror = function(err) {
			    console.error('Get Woops', err);
		function get_request(url, success_call, headers) {
			return request('GET', url, null, success_call, headers);
		function post_request(url, data, success_call, headers) {
            return request('POST', url, data, success_call, headers);
		function json_url(url) {
		    return url+'?alt=json';
		function serialize(obj) {
            return '?'+Object.keys(obj).reduce(function(a,k){a.push(k+'='+encodeURIComponent(obj[k]));return a},[]).join('&')
        function xhr_json_response(xhr) {
            return JSON.parse(xhr.responseText);
		function handleCreateClick(event) {
			To upload a file, check the Javascript example
			from https://developers.google.com/drive/v2/reference/files/insert
			var body = {
				    'mimeType': 'application/vnd.google-apps.spreadsheet',
	                'title': 'blocky'},
	            url = 'https://www.googleapis.com/drive/v2/files';
			post_request(url, JSON.stringify(body), null, send_json_header);
		function handleListClick(event, callback) {
			find the Drive Api reference here


			var url = 'https://www.googleapis.com/drive/v2/files'+serialize({'q': 'title="blocky" and trashed=false'});   
			get_request(url, function(xhr) {
				var obj = xhr_json_response(xhr);
				console.info('documents', obj, obj.items);
				if(typeof callback === "function") {
		function handleWorksheetsClick(events) {
			Google Sheets Api has no "reference" page and i 
			couldn't figure out a way to find a spreadsheet based on it's id.


			Make sure that the "blocky" spreadsheet has the following rows:
	        key   content url
	        a1    a2      a3
	        b1    b2      b3
			handleListClick(null, function(obj) {
				// This is the endpoint for getting the worksheets of a spreadsheet identified by "id"
                var url = 'https://spreadsheets.google.com/feeds/worksheets/'+encodeURIComponent(obj.items[0].id)+'/private/full';
				=== TROUBLE AHEAD ===
                Google Sheets Api by default returns and accepts "application/atom+xml" content type - that's XML strings
                // Let's get an "ATOM+XML" response of the worksheets for our "blocky" spreadsheet
                get_request(url, function(xhr) {
                	console.info('Request atom+xml', xhr);
                }, headers.accept_xml);
                The request doesn't get thought because of this error:
                No 'Access-Control-Allow-Origin' header is present on the requested resource. Origin 'http://localhost:8888' is therefore not allowed access.
                If you check the Network Headers in Chrome Developer tools, 
                Chrome first does a "OPTIONS" request to the "url" and gets:
                All seems fine until it makes the actual "GET" request where the "access-control-allow-origin" header dissapears 
                and Google Chrome denies the cross origin request. 
                Let's alter the "url" to request a "JSON" feed as described here: 


                get_request(url+'?alt=json', function(xhr) {
                    console.info('Request json', xhr);
                    var obj = xhr_json_response(xhr);
                    console.log('json', obj, obj.feed.entry[0]);
                    Using a JSON request (?alt=json) allows getting the data out of the worksheets.
                    Manually add some data to the "blocky" spreadsheet in Google Drive app to see results
                    var url = obj.feed.entry[0].link[0].href;
                    get_request(url+'?alt=json', function(xhr) {
                    	var obj = xhr_json_response(xhr);
                    	console.info('Spreadsheet data', obj, obj.feed.entry);
                }, headers.accept_json);
		function handleRowClick(events) {
			handleListClick(null, function(obj) {
		        Adding new rows as described in the chapter "Add a list row" from 


		        is not working

				var url = 'https://spreadsheets.google.com/feeds/worksheets/'+encodeURIComponent(obj.items[0].id)+'/private/full';
				get_request(url+'?alt=json', function(xhr) {
                    console.info('Request json', xhr);
                    var obj = xhr_json_response(xhr);
                    //console.log('worksheet', obj)
                    var url = obj.feed.entry[0].link[0].href;
                    var post_data = '<entry xmlns="http://www.w3.org/2005/Atom" xmlns:gsx="http://schemas.google.com/spreadsheets/2006/extended"><gsx:key>x1</gsx:key><gsx:content>x2</gsx:content><gsx:url>x3</gsx:url></entry>';
                    post_request(url, post_data, function(xhr) {
                    	console.log('post data', xhr);
                    	No 'Access-Control-Allow-Origin' header is present on the requested resource. Origin 'http://localhost:8888' is therefore not allowed access.
                    }, headers.send_xml);
    <script src="https://apis.google.com/js/auth.js?onload=handleClientLoad"></script>


  • Google Sheets API is using the old “gdata API”. Read more https://developers.google.com/gdata/docs/directory and here https://developers.google.com/gdata/docs/json
  • Reading operations seem to work fine with CORS (https://developers.google.com/api-client-library/javascript/features/cors) requests as long as you request JSON feeds (add ?alt=json to the requested url)
  • I wasn’t able to “POST” data using CORS requests, using either “atom+xml” or “json” content types.
  • Sending “POST” requests from “curl” or Python worked fine.
  • All these basically mean that you can’t build Single-Page Apps for changing data in Google services (unless you use a back-end request proxy)

Blocky – the Chrome Extension that enables tagging blocks of html

Posted July 30th, 2015 in Javascript by Florentin

This small Chrome Extension i’ve created is good for tagging divs or other blocks of html.

Let’s say you see a web page listing the latest 50 news. Using blocky extension you would be able to tag the news you like with “green”, the ones you don’t like with “red” and the ones you didn’t read with “yellow”. On the following visit you’d see the marked items.

Please check the extension here:


A better diigo bookmarklet (diigolet)

Posted March 29th, 2013 in Uncategorized by Florentin

What is the ‘diigolet’

The diigolet is a Diigo.com tool used to manage a user’s bookmarks. It is a bookmarklet, a bookmark stored in a browser that contains Javascript.
The original diigolet is here:


How is BetterDiigolet different ?

BetterDiigolet by default opens the “bookmark” menu from the diigolet toolbar, it fills in the description and clicks all recommended tags. All is left for the user is to click “Save Bookmark”.
If it’s used on a already bookmarked url, it will color the tags previously used in ‘green’. The new recommended tags will remain ‘blue’.
This is an easy way for the user to find out if the url has been bookmarked before.
By using better_diigolet.js one can save time with clicking though all the tags and filling in the bookmark’s description field.

How it’s built

Diigo’s diigolet is loading a js file named “diigolet_b_h_b.js” which in turns loads “diigolet.js”. diigolet.js contains jQuery v1.4.3.
BetterDiigolet loads everything in better_diigolet.js, jQuery has been replaced with v1.8.3
The downside is the missing “Loading Diigolet…” tooltip.

How to install it in Firefox

  1. Make sure the “Bookmarks Toolbar” is visible. If it is not, go to menu View > Toolbars.
  2. Drag this button: BetterDiigolet up to your Bookmarks Toolbar.

Django reusable apps

Posted October 31st, 2012 in Python by Florentin

Warning: this article is still a draft

Explaining the Django reusable apps in an agile way

  • as a developer
    • i want to easily plug the app into the project so that i can improve the productivity
    • i want to customize and extend existing Django apps so that they fit the requirements
  • some of the things that need customization:
    • models
      • e.g. make the app work with a different model class than the ones provided by the app
      • e.g. create a custom model based on the one supplied by the app and add extra fields or modify the existing field definition
    • forms
      • e.g. the app should use a custom form
      • e.g. a custom form which adds extra fields or change the definition of the existing ones
    • template names
    • success or redirect urls
    • template variables (extra_context)

1 James Bennett’s approach – the custom objects are passed as arguments to the views


def register(request, success_url=None,

2 Use the settings to define the custom objects


  • in settings.py, define the values for models/forms/etc which will be later used by the app
  • it’s useful to prefix all the names required by a specific app, i.e. BLOGAPP_MODEL, BLOGAPP_FORM etc
  • example


from django.conf import settings
from .models import AppModel

def register(request):

my_model = getattr(settings, ‘MY_APP_MODEL’, AppModel)

What the current approaches don’t offer

  • an easy way to modify or extend the urls. The urlpatterns is just a list, it’s difficult to delete/edit/insert url definitions.
  • models placed in models.py are used by Django in different db operations (i.e. syncdb).

One might want to ignore those models and let Django use customized/child Models.

  • the Model and Form fields are not easy to alter

Ideas for a new solution

  • make everything a class. Forms and Models are classes but the urls (or maybe the views) are not.
  • move the Model definitions from models.py and allow the developer to use it’s own Models (i.e. though getters or factories)
    • instead one can play with abstract models but without any concrete Models the app won’t work out of the box
  • define the Form and Model fields inside a method so that the derived classes can modify the definitions


Django project skeleton

Posted October 24th, 2011 in Python by Florentin

If you need a Django project skeleton to base your work upon, please check this one:


And now a Django app skeleton as well


Django settings

Posted October 24th, 2011 in Python by Florentin

I’m going to talk here about various techniques regarding the Django’s settings.

Let’s assume the following project layout:
├── __init__.py
├── manage.py
├── settings
└── urls.py

How do I use custom settings or overwrite existing ones on my development machine ?

Answer 1

Create your custom settings file under the same parent as the existing settings.py. Let’s name the new file settings_local.py
At the bottom of settings.py, add the following:

    from settings_local import *
except ImportError:


  • no need to change the manage.py or the wsgi file in order to apply the new settings


  • hard/impossible to use different settings for different environment

 Answer 2 (preferred)

Create a new directory called “settings” and move the existing setting file there. Then make a different file for each type of environment.
├── __init__.py
├── manage.py
├── settings
│   ├── __init__.py
│   ├── development.py
│   ├── production.py
│   └── settings.py
└── urls.py

settings.py will include the default Django settings, probably the file created by django-admin.py
development.py will hold settings/overwrites needed for the development environment.
The extra settings files (production.py, development.py, etc) will extend the existing settings.py (or another parent file which in turn extends settings.py) and add their own settings customizations.
This could be your development.py file:

from .production import *
DEBUG = True
# APP: debug_toolbar

Where production.py is:
from .settings import *
import os.path
PROJECT_ROOT = os.path.join(os.path.abspath(os.path.dirname(__file__)), '../')
DEBUG = False

Once your settings setup is in place, all you have to do is change manage.py and your WSGI file.
The manage.py file could now look like this:

#!/usr/bin/env python
from django.core.management import execute_manager
import sys, os
PROJECT_ROOT = os.path.abspath(os.path.dirname(__file__))
sys.path.insert(0, PROJECT_ROOT)
import settings.development # Assumed to be in the same directory.
except ImportError, e:
import sys
sys.stderr.write("Error: Can't find the file 'settings.py' in the directory containing %r. It appears you've customized things. You'll have to run django-admin.py, passing it your settings module. (If the file settings.py does indeed exist, it's causing an ImportError somehow.) " % __file__)
if __name__ == "__main__":

In the same time, your WSGI file would use settings.production:

os.environ[“DJANGO_SETTINGS_MODULE”] = “settings.production”

  • easy to create settings for each of your environments (production, development, etc)
  • it’s a great way to keep your settings organized, easy to find and edit.
  • easier to reuse your settings for other projects. For example, we could use the same development.py file (as shown above) with other production.py settings.


  • you have to change the manage.py and the WSGI file, which might not be possible on your client server.

 Other Django settings tips

  • do not use your project name in the settings, i.e. use ROOT_URLCONF = ‘urls’ and not ROOT_URLCONF = ‘myproject.urls’, use INSTALLED_APPS = (“appname”,) and not INSTALLED_APPS = (“myproject.appname”, ). You will then be able to easily move settings and applications between one project to another
  • use calculated paths for TEMPLATE_DIR, MEDIA_ROOT, STATIC_ROOT etc, i.e. PROJECT_ROOT = os.path.abspath(os.path.dirname(__file__)) and then MEDIA_ROOT = os.path.join(PROJECT_ROOT, “media”)

This means business

Posted August 23rd, 2011 in Funny by Florentin

This is how it goes …

Father: Son, I would like to be the one who chooses your future wife.

Son: No way

Tata: That girl is Bill Gates’s doughter

Son: AAA, then it is ok.

Father goes to Bill Gates

Father: I want my son to marry your daughter.

Bill Gates: No chance

Dad: My son is CEO of the World Bank

Bill Gates: aa, then it is ok.,

Father goes to World Bank President

Tata: I would like to offer my son as CEO of the World Bank

Presenintele: No chance

Dad: My son is the future husband of Bill Gates’s doughter

President: Oh, that’s ok.

This means BUSINESS