Posts

Mario Meir-Huber
, , ,

Python for Spark Tutorial – Dynamically creating classes in Python

In our previous tutorial, we had a look at how to (de) serialise objects from and to JSON in Python. Now, let’s have a look into how to dynamically create and extend classes in Python. Basically, we are using the library that Python itself is using. This is the dynamic type function in Python. This function takes several parameters, we will only focus on three relevant one’s for our sample.

How to use the dynamic type function in Python

Basically, this function takes several parameters. We utilize 3 parameters. These are:

type(CLASS_NAME, INHERITS, PARAMETERS)

These parameters have the following meaning:

  • CLASS_NAME: The name of the new class
  • INHERITS: from which the new type should inherit
  • PARAMETERS: new methods or parameters added to the class

In our following example, we want to extend the Person class with a new attribute called “location”. We call our new class “PersonNew” and instruct Python to inherit from “Person”, which we have created some tutorials earlier. Strange is that it is passed as an array, even there can only be one inheritance hierarchy in Python. Last, we specify the method “location” as key-value pair. Our sample looks like the following:

pn = type("PersonNew", (Person,), {"location": "Vienna"})
pn.age = 35
pn.name = "Mario"

If you test the code, it will just work like expected. All other objects such as age and name can also be retrieved. Now, let’s make it a bit more complex. We extend our previous sample with the JSON serialisation to be capable of dynamically creating a JSON object from a string.

Dynamically creating a class in Python from JSON

We therefore create a new function that takes the object to serialise and takes all values out of that. In addition, we add one more key-value pair, which we call “__class__” in order to store the name of the class. getting the class-name is a bit more complex, since it is written like “class ‘main.PersonNew'”. Therefore, we first split the object name with a “.”, take the last entry and again split it by the ‘ and take the first one. There are more elegant ways for this, but I want to keep it simple. Once we have the classname, we store it in the dictionary and return the dictionary. The complex sample is here:

def map_proxy(obj):
    dict = {}
    for k in obj.__dict__.keys():
        dict.update({k : obj.__dict__.get(k)})
    cls_name = str(obj).split(".")[1].split("'")[0]
    dict.update({"__class__" : cls_name})
    return dict

We can now use the json.dumps method and call the map_proxy function to return the JSON string:

st_pn = json.dumps(map_proxy(pn))
print(st_pn)

Now, we are ready to dynamically create a new class with the “type” method. We name the method after the class name that was provided above. This can be retrieved with “__class__”. We let it inherit from Person and pass the parameters from the entire object into it, since it is already a key/value pair:

def dyn_create(obj):
    return type(obj["__class__"], (Person, ), obj)

We can now also invoke the json.loads method to dynamically create the class:

obj = json.loads(st_pn, object_hook=dyn_create)
print(obj)
print(obj.location)

And the output should be like that:

{"location": "Vienna", "__module__": "__main__", "__doc__": null, "age": 35, "name": "Mario", "__class__": "PersonNew"}
<class '__main__.PersonNew'>
Vienna

As you can see, it is very easy to dynamically create new classes in Python. We could largely improve this code, but i’ve created this tutorial for explanatory reasons rather than usability ;).

In our next tutorial, we will have a look at logging.

If you are not yet familiar with Spark, have a look at the Spark Tutorial i created here. Also, I will create more tutorials on Python and Machine Learning in the future, so make sure to check back often to the Big Data & Data Science tutorial overview. I hope you liked this tutorial. If you have any suggestions and what to improve, please feel free to get in touch with me! If you want to learn more about Python, I also recommend you the official page.

/0 Comments/by
Mario Meir-Huber
, , ,

Python for Spark Tutorial – Python serialization with Objects and JSON

One important aspect of working with Data is serialisation. Basically, this means that classes can be persisted to a storage (e.g. the file system, HDFS or S3). With Spark, a lot of file formats are possible. However, in this tutorial we will have a look on how to deal with JSON, a very popular file format and often used in Spark. Now we will have a look at Python serialization.

What is it and how does Python serialization work?

JSON stands for “Java Script Object Notation” and was usually developed for Client-Server applications with JavaScript as main user of it. It was built to have less overhead than XML.

First, let’s start with copying objects. Basically, Python knows two ways: normal copies and deep copies. The difference is that with normal copies, references to objects within the copied object are built. This is relevant when using objects as classes. In a deep copy, no references are built but every value is copied to the new object. This means that you can now use it independent from the previous one.

To copy objects to another, you only need to import copy and call the copy or deepcopy function. The following code shows how this works.

import copy
ps1 = Person("Mario", 35)
pss = copy.copy(ps1)
psd = copy.deepcopy(ps1)
ps1.name = "Meir-Huber"
print(ps1.name)
print(pss.name)
print(psd.name)

And the output should be this:

Meir-Huber
Mario
Mario

JSON serialization in Python

Now, let’s look at how we can serialise an object with the use of JSON. Basically, you need to import “json”. An object that you want to serialise needs to be serialise-able. A lot of classes in Python already implement that. However, when we want to serialise our own object (e.g. the “Person” class that we have created in this tutorial), we need to implement the serialise-function or a custom serialiser. However, Python is great and provides us the possibility to access all variables in an object via the “__dict__” dictionary. This means that we don’t have to write our own serialiser and can do this via an easy call to “dumps” of “json”:

import json
js = json.dumps(ps1.__dict__)
print(js)

The above function creates a JSON representation of the entire class

{"name": "Meir-Huber", "age": 35}

We might want to add more information to the JSON string – e.g. the class name that it was originally stored in. We can do this by calling a custom function in the “dumps” method. This method gets the object to be serialised as only parameter. We then only pass the original object (Person) and the function we want to execute. We name this function “make_nice”. In the function, we create a dictionary and add the name of the class as first parameter. We give this the key “obj_name”. We then join the dictionary of the object into the new dictionary and return it.

Finishing the serialization

Another parameter added to the “dumps” function is “indent”. The only thing it does is printing it pretty – by adding line breaks and indents. This is just for improved readability. The method and call looks like this:

def make_nice(obj):
    dict = {
        "obj_name": obj.__class__.__name__
    }
    dict.update(obj.__dict__)
    return dict
js_pretty = json.dumps(ps1, default=make_nice,indent=3)
print(js_pretty)

And the result should now look like the following:

{
   "obj_name": "Person",
   "name": "Meir-Huber",
   "age": 35
}

Now, we know how we can serialise an object to a JSON string. Basically, you can now store this string to a file or an object on S3. The only thing that we haven’t discussed yet is how to get back an object from a string. We therefore take the JSON object we “dumps” before. Our goal now is to create a Person object from it. This can be done via the call “loads” from the json-object. We also define a method to do the casting via the “object_hook” parameter. This object_hook method has one argument – the JSON object itself. We access each of the parameters from the object with named indexers and return the new object.

str_json = "{\"name\": \"Meir-Huber\", \"age\": 35}"
def create(obj):
    print(obj)
    return Person(obj["name"], obj["age"])
obj = json.loads(str_json, object_hook=create)
print(obj)

The output should now look like this.

{'name': 'Meir-Huber', 'age': 35}
<__main__.Person object at 0x7fb84831ddd8>

Now we know how to create JSON serialisers and how to get them back from a string value. In the next tutorial, we will have a look on how to improve this and make it more dynamic – by dynamic class creation in Python.

If you are not yet familiar with Spark, have a look at the Spark Tutorial i created here. Also, I will create more tutorials on Python and Machine Learning in the future, so make sure to check back often to the Big Data & Data Science tutorial overview. I hope you liked this tutorial. If you have any suggestions and what to improve, please feel free to get in touch with me! If you want to learn more about Python, I also recommend you the official page.

/0 Comments/by
Mario Meir-Huber
, , ,

Python for Spark Tutorial – String manipulations in Python

In the last tutorials, we already worked a lot with Strings and even manipulated some of them. Now, it is about time to have a look at the theory behind it. Basically, formatting strings is very easy. The only thing you need is the “format” method appended to a string with a variable amount of data. If you add numbers, the str() function is executed on them by itself, so no need to convert them. This tutorial is about String manipulations in Python.

String manipulations in Python

Basically, the annotation is very similar to the one from other string formatters you are used to. One really nice thing though is that you don’t need to provide the positional arguments. Python assumes that the positions are in-line with the parameters you provide. An easy sample is this:

str01 = "This is my string {} and the value is {}".format("Test", 11)
print(str01)

And the output should look like this:

This is my string Test and the value is 11

You can also use classes for this. Therefore, we define a class “Person”:

class Person:
    def __init__(self, name, age):
        self.name = name
        self.age = age
p = Person("Mario Meir-Huber", 35)
str02 = "The author \"{}\" is {} years old".format(p.name, p.age)
print(p.name)
print(str02)

The output for this should look like this:

Mario Meir-Huber
The author "Mario Meir-Huber" is 35 years old

The difflib in Python

One nice thing in Python is the difflib. This library enables us to easily check two array of strings for differences. One use-case would be to check my lastname for differences. Note that my lastname is one of the most frequent lastname combinations in the german speaking countries and thus allows different ways to write it.

To work with difflib, simply import it and call the difflib context_diff function. This prints the differences detected with “!”.

import difflib
arr01 = ["Mario", "Meir", "Huber"]
arr02 = ["Mario", "Meier", "Huber"]
for line in difflib.context_diff(arr01, arr02):
    print(line)

Below you can see the output. One difference was spotted. You can easily use this for spotting differences in datasets and creating golden records from it. 

***
---
***************
*** 1,3 ****
  Mario
! Meir
  Huber
--- 1,3 ----
  Mario
! Meier
  Huber

Textwrap in Python

Another nice feature in Python is the usage of textwrap. This library has some basic features for text “prettyfying”. Basically, in the following sample, we use 5 different things:

  • Indent: creates an indent to a text, e.g. a tab before the text
  • Wrap: wraps the text into an array of strings in case it is longer than the maximum width. This is useful to split text into a maximum number of arrays
  • Fill: does the same as Wrap, but creates new lines out of it
  • Shorten: shortens the text with a specified maximum number. This is written like “[…]” and you might use it to add a “read more” around it
  • Detent: deletes any whitespace before or after the text

The functions are used in simple statements:

from textwrap import *
print(indent("Mario Meir-Huber", "\t"))
print(wrap("Mario Meir-Huber", width=10))
print(fill("Mario Meir-Huber", width=10))
print(shorten("Mario Meir-Huber Another", width=15))
print(dedent(" Mario Meir-Huber "))

And the output should look like this:

	Mario Meir-Huber
['Mario', 'Meir-Huber']
Mario
Meir-Huber
Mario [...]
Mario Meir-Huber

Today’s tutorial was more of a “housekeeping” since we used it already. In the next tutorial, I will write about object serialisation with JSON, as this is also very useful.

If you are not yet familiar with Spark, have a look at the Spark Tutorial i created here. Also, I will create more tutorials on Python and Machine Learning in the future, so make sure to check back often to the Big Data & Data Science tutorial overview. I hope you liked this tutorial. If you have any suggestions and what to improve, please feel free to get in touch with me! If you want to learn more about Python, I also recommend you the official page.

/1 Comment/by
Mario Meir-Huber
, , ,

Python for Spark Tutorial – Python Async and await functionality

In the last tutorials, we had a look at methods, classes and deorators. Now, let’s have a brief look at asynchronous operations in Python. Most of the time, this is anyway abstracted for us via Spark, but it is nevertheless relevant to have some basic understanding of it. In this Tutorial, we will look at Python async and await functionality.

Python Async and await functionality

Basically, you define a method to be asynchronous by simply adding “async” as keyword ahead of the method definition. This is written like that:

async def FUNCTION_NAME():
    
    FUNCTION-BLOCK

Another keyword in that context is “await”. Basically, every function that is doing something asynchronous is awaitable. When adding “await”, nothing else happens until the asynchronous function has finished. This means that you might loose the benefit of asynchronous execution but get better handling when working with web data. In the following code, we create an async function that sleeps some seconds (between 1 and 10). We call the function twice with the “await” operator.

import asyncio
import random
async def func():
    tim = random.randint(1,10)
    await asyncio.sleep(tim)
    print(f"Function finished after {tim} seconds")
await func()
await func()

In the output, you can see that it was first waited for the first function to finish and only then the second one was executed. Basically, all of the execution happened sequentially, not in parallel. 

Function finished after 9 seconds
Function finished after 9 seconds

Python also knows parallel execution. This is done via Tasks. We use the Method “create_task” from the asyncio library in order to execute a function in parallel. In order to see how this works, we invoke the function several times and add a print-statement at the end of the code.

Parallel execution in Python async

asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
asyncio.create_task(func())
print("doing something else ...")

This now looks very different to the previous sample. The print statement is the first to show up, and all code path finish after 9 seconds max. This is due to the fact that (A) the first execution finishes after 1 second – thus the print statement is the first to be shown, since it is executed immediately. (B) Everything is executed in parallel and the maximum sleep interval is 9 seconds.

doing something else ...
Function finished after 1 seconds
Function finished after 1 seconds
Function finished after 3 seconds
Function finished after 4 seconds
Function finished after 5 seconds
Function finished after 7 seconds
Function finished after 7 seconds
Function finished after 7 seconds
Function finished after 8 seconds
Function finished after 10 seconds
Function finished after 10 seconds
Function finished after 10 seconds

However, there are also some issues with async operations. You can never say how long it takes a task to execute. It could finish fast or it could also take forever, due to a weak network connection or an overloaded server. Therefore, you might want to specify a timeout, which is the maximum an operation should be waited for. In Python, this is done via the “wait_for” method. It basically takes the function to execute and the timeout in seconds. In case the call runs into a timeout, a “TimeoutError” is raised. This allows us to surround it with a try-block.

Dealing with TimeoutError in Python

try:
    await asyncio.wait_for(func(), timeout=3.0)
except asyncio.TimeoutError:
    print("Timeout occured")

In two third of the cases, our function will run into a timeout. The function should return this:

Timeout occured

Each task that should be executed can also be controlled. Whenever you call the “create_task” function, it returns a Task-object. A task can either be done, cancelled or contain an error. In the next sample, we create a new task and wait for it’s completion. We then check if the task was done or cancelled. You could also check for an error and retrieve the error message from it.

Create_Task in Python

task = asyncio.create_task(func())
print("running task")
await task
if task.done():
    print("Task was done")
elif task.cancelled():
    print("Task was cancelled")

In our case, no error should have occurred and thus the output should be the following:

running task
Function finished after 8 seconds
Task was done

Now we know how to work with async operations in Python. In our next tutorial, we will have a deeper look into how to work with Strings.

If you are not yet familiar with Spark, have a look at the Spark Tutorial i created here. Also, I will create more tutorials on Python and Machine Learning in the future, so make sure to check back often to the Big Data & Data Science tutorial overview. I hope you liked this tutorial. If you have any suggestions and what to improve, please feel free to get in touch with me! If you want to learn more about Python, I also recommend you the official page.

/0 Comments/by