Review: git, Python, JSON

Some of us have attended in-house training on using git and Python developer training, with some exposure to both the Flask and Django frameworks.

Git

Make sure you know how to git clone, checkout, add, commit, push

Python & Django

MVC: Model-View-Controller (and Django's weird terminology)

How to start a Django project and an app within the project: django-admin startproject and django-admin startapp.

Database persistance: models.py

Examples will follow.

Javascript Object Notation (JSON)

JSON is the machine-readable de-facto standard for serializing data over the web, supplanting XML, HTML, etc.

• Serialization of objects (for transmission across the net or sharing as documents).
• Simpler than XML: maps directly to programming language data structures.
• Human readable/writeable.

JSON data types

All JSON data types are represented as strings in a JSON-encoded (serialized) document but they map into common programming language datatypes (such as in Javascript).

• strings "hello, world"
• numbers 1, 5.6, 1.3E22
• boolean - true/false
• null
• object - unordered set of key/value pairs, where the value can be of any type:

  {"given": "Alan", "surname": "Crosswell", "age": 59, "likes": ["pizza", "xiao long bao"]}
  

• array - ordered list of any types:

  [1, 2, "three", {"color": "red"}, ["a", "b", null, true]]
  

(De)serializing JSON (to)from Python variables: a simple Python app

#!/usr/bin/env python
import json
from pprint import pprint

# json_serialized: "on the wire" string representation:
json_serialized = '[1,2,"three",{"color":"red"},["a","b",null,true]]'
print("JSON serialized:", json_serialized)
# json_deserialized: converted to Python-native variables
json_deserialized = json.loads(json_serialized)
print("Python object:")
pprint(json_deserialized)
for item in json_deserialized:
    print("item:",item)
print(json_deserialized[3]['color'])

$ python3 json-example.py
JSON serialized: [1,2,"three",{"color":"red"},["a","b",null,true]]
Python object:
[1, 2, 'three', {'color': 'red'}, ['a', 'b', None, True]]
item: 1
item: 2
item: three
item: {'color': 'red'}
item: ['a', 'b', None, True]
red

Footnote: Geezer history of serialization

Data in "machine" format is represented in unique ways depending on the hardware architecture. For instance, integers can have an 8, 16, 32, or 64 bit 2's-complement representation. Floating point numbers can be 32, 64 or 128-bits using representations that are vendor-proprietary or perhaps IEEE floating point. Furthermore, the order of bytes within words can be big- or little-endian.

When early computer networks were created among heterogenous systems, their architectural differences were significant and standards were created to move these data "across the wire" -- as a serial stream of bits -- and reconstruct them on the other end. Some of these early versions include ntohs and so on which converted between network and host byte order.

Later on, higher-level representations such as ASN.1 were developed to similarly serialize/deserialize more complex objects using a variety of encoding rules such as BER.

The problem with all these binary encodings is that a human can't decode them by just looking at them. As storage and networks became cheaper and faster, plain text string serializations, based on text markup languages, became popular, including XML (maybe not so easy:-) and, later, JSON. One of the things that has made JSON so popular is it is really easy for a person to read and write.

Of course, the pendulum swings back, the amounts of data exchanged over the network become massive, and some compact binary serialization formats are back such as Avro -- which looks a lot like ASN.1 BER, DER and so on!