2019-02-15

How to Export Data to XLSX Files

A while ago I wrote an article about exporting data to different spreadsheet formats. As recently I was reimplementing export to Excel for the 1st things 1st project, I noticed that the API changed a little, so it's time to blog about that again.

For Excel export I am using the XLSX file format which is a zipped XML-based format for spreadsheets with formatting support. XLSX files can be opened with Microsoft Excel, Apache OpenOffice, Apple Numbers, LibreOffice, Google Drive, and a handful of other applications. For building the XLSX file I am using openpyxl library.

Installing openpyxl

You can install openpyxl to your virtual environment the usual way with pip:

(venv) pip install openpyxl==2.6.0

Simplest Export View

To create a function exporting data from a QuerySet to XLSX file, you would need to create a view that returns a response with a special content type and file content as an attachment. Plug that view to URL rules and then link it from an export button in a template.

Probably the simplest view that generates XLSX file out of Django QuerySet would be this:

# movies/views.py
from datetime import datetime
from datetime import timedelta
from openpyxl import Workbook
from django.http import HttpResponse

from .models import MovieCategory, Movie

def export_movies_to_xlsx(request):
    """
    Downloads all movies as Excel file with a single worksheet
    """
    movie_queryset = Movie.objects.all()
    
    response = HttpResponse(
        content_type='application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
    )
    response['Content-Disposition'] = 'attachment; filename={date}-movies.xlsx'.format(
        date=datetime.now().strftime('%Y-%m-%d'),
    )
    workbook = Workbook()
    
    # Get active worksheet/tab
    worksheet = workbook.active
    worksheet.title = 'Movies'

    # Define the titles for columns
    columns = [
        'ID',
        'Title',
        'Description',
        'Length',
        'Rating',
        'Price',
    ]
    row_num = 1

    # Assign the titles for each cell of the header
    for col_num, column_title in enumerate(columns, 1):
        cell = worksheet.cell(row=row_num, column=col_num)
        cell.value = column_title

    # Iterate through all movies
    for movie in movie_queryset:
        row_num += 1
        
        # Define the data for each cell in the row 
        row = [
            movie.pk,
            movie.title,
            movie.description,
            movie.length_in_minutes,
            movie.rating,
            movie.price,
        ]
        
        # Assign the data for each cell of the row 
        for col_num, cell_value in enumerate(row, 1):
            cell = worksheet.cell(row=row_num, column=col_num)
            cell.value = cell_value

    workbook.save(response)

    return response

If you try this, you will notice, that there is no special formatting in it, all columns are of the same width, the value types are barely recognized, the header is displayed the same as the content. This is enough for further data export to CSV or manipulation with pandas. But if you want to present the data for the user in a friendly way, you need to add some magic.

Creating More Worksheets

By default, each Excel file has one worksheet represented as a tab. You can access it with:

worksheet = workbook.active
worksheet.title = 'The New Tab Title'

If you want to create tabs dynamically with data from the database of Python structures, you can at first delete the current tab and add the others with:

workbook.remove(workbook.active)

for index, category in enumerate(category_queryset):
    worksheet = workbook.create_sheet(
        title=category.title,
        index=index,
    )

Although not all spreadsheet applications support this, you can set the background color of the worksheet tab with:

worksheet.sheet_properties.tabColor = 'f7f7f9'

Working with Cells

Each cell can be accessed by its 1-based indexes for the rows and for the columns:

top_left_cell = worksheet.cell(row=1, column=1)
top_left_cell.value = "This is good!"

Styles and formatting are applied to individual cells instead of rows or columns. There are several styling categories with multiple configurations for each of them. You can find some available options from the documentation, but even more by exploring the source code.

from openpyxl.styles import Font, Alignment, Border, Side, PatternFill

top_left_cell.font = Font(name='Calibri', bold=True)
top_left_cell.alignment = Alignment(horizontal='center')
top_left_cell.border = Border(
    bottom=Side(border_style='medium', color='FF000000'),
)
top_left_cell.fill = PatternFill(
    start_color='f7f7f9',
    end_color='f7f7f9',
    fill_type='solid',
)

If you are planning to have multiple styled elements, instantiate the font, alignment, border, fill options upfront and then assign the instances to the cell attributes. Otherwise, you can get into memory issues when you have a lot of data entries.

Setting Column Widths

If you want to have some wider or narrower width for some of your columns, you can do this by modifying column dimensions. They are accessed by column letter which can be retrieved using a utility function:

from openpyxl.utils import get_column_letter

column_letter = get_column_letter(col_num)
column_dimensions = worksheet.column_dimensions[column_letter]
column_dimensions.width = 40

The units here are some relative points depending on the width of the letters in the specified font. I would suggest playing around with the width value until you find what works for you.

When defining column width is not enough, you might want to wrap text into multiple lines so that everything can be read by people without problems. This can be done with the alignment setting for the cell as follows:

from openpyxl.styles import Alignment

wrapped_alignment = Alignment(vertical='top', wrap_text=True)
cell.alignment = wrapped_alignment

Data Formatting

Excel automatically detects text or number types and aligns text to the left and numbers to the right. If necessary that can be overwritten.

There are some gotchas on how to format cells when you need a percentage, prices, or time durations.

Percentage

For percentage, you have to pass the number in float format from 0.0 till 1.0 and style should be 'Percent' as follows:

cell.value = 0.75
cell.style = 'Percent'

Currency

For currency, you need values of Decimal format, the style should be 'Currency', and you will need a special number format for currency other than American dollars, for example:

from decimal import Decimal
cell.value = Decimal('14.99')
cell.style = 'Currency'
cell.number_format = '#,##0.00 €'

Durations

For time duration, you have to pass timedelta as the value and define special number format:

from datetime import timedelta

cell.value = timedelta(minutes=90)
cell.number_format = '[h]:mm;@'

This number format ensures that your duration can be greater than '23:59', for example, '140:00'.

Freezing Rows and Columns

In Excel, you can freeze rows and columns so that they stay fixed when you scroll the content vertically or horizontally. That's similar to position: fixed in CSS.

To freeze the rows and columns, locate the top-left cell that is below the row that you want to freeze and is on the right from the column that you want to freeze. For example, if you want to freeze one row and one column, the cell would be 'B2'. Then run this:

worksheet.freeze_panes = worksheet['B2']

Fully Customized Export View

So having the knowledge of this article now we can build a view that creates separate sheets. for each movie category. Each sheet would list movies of the category with titles, descriptions, length in hours and minutes, rating in percent, and price in Euros. The tabs, as well as the headers, can have different background colors for each movie category. Cells would be well formatted. Titles and descriptions would use multiple lines to fully fit into the cells.

# movies/views.py
from datetime import datetime
from datetime import timedelta
from openpyxl import Workbook
from openpyxl.styles import Font, Alignment, Border, Side, PatternFill
from openpyxl.utils import get_column_letter
from django.http import HttpResponse

from .models import MovieCategory, Movie

def export_movies_to_xlsx(request):
    """
    Downloads all movies as Excel file with a worksheet for each movie category
    """
    category_queryset = MovieCategory.objects.all()
    
    response = HttpResponse(
        content_type='application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
    )
    response['Content-Disposition'] = 'attachment; filename={date}-movies.xlsx'.format(
        date=datetime.now().strftime('%Y-%m-%d'),
    )
    workbook = Workbook()
    
    # Delete the default worksheet
    workbook.remove(workbook.active)

    # Define some styles and formatting that will be later used for cells
    header_font = Font(name='Calibri', bold=True)
    centered_alignment = Alignment(horizontal='center')
    border_bottom = Border(
        bottom=Side(border_style='medium', color='FF000000'),
    )
    wrapped_alignment = Alignment(
        vertical='top',
        wrap_text=True
    )

    # Define the column titles and widths
    columns = [
        ('ID', 8),
        ('Title', 40),
        ('Description', 80),
        ('Length', 15),
        ('Rating', 15),
        ('Price', 15),
    ]
    
    # Iterate through movie categories
    for category_index, category in enumerate(category_queryset):
        # Create a worksheet/tab with the title of the category
        worksheet = workbook.create_sheet(
            title=category.title,
            index=category_index,
        )
        # Define the background color of the header cells
        fill = PatternFill(
            start_color=category.html_color,
            end_color=category.html_color,
            fill_type='solid',
        )
        row_num = 1
        
        # Assign values, styles, and formatting for each cell in the header
        for col_num, (column_title, column_width) in enumerate(columns, 1):
            cell = worksheet.cell(row=row_num, column=col_num)
            cell.value = column_title
            cell.font = header_font
            cell.border = border_bottom
            cell.alignment = centered_alignment
            cell.fill = fill
            # set column width
            column_letter = get_column_letter(col_num)
            column_dimensions = worksheet.column_dimensions[column_letter]
            column_dimensions.width = column_width

        # Iterate through all movies of a category
        for movie in category.movie_set.all():
            row_num += 1
            
            # Define data and formats for each cell in the row
            row = [
                (movie.pk, 'Normal'),
                (movie.title, 'Normal'),
                (movie.description, 'Normal'),
                (timedelta(minutes=movie.length_in_minutes), 'Normal'),
                (movie.rating / 100, 'Percent'),
                (movie.price, 'Currency'),
            ]
            
            # Assign values, styles, and formatting for each cell in the row
            for col_num, (cell_value, cell_format) in enumerate(row, 1):
                cell = worksheet.cell(row=row_num, column=col_num)
                cell.value = cell_value
                cell.style = cell_format
                if cell_format == 'Currency':
                    cell.number_format = '#,##0.00 €'
                if col_num == 4:
                    cell.number_format = '[h]:mm;@'
                cell.alignment = wrapped_alignment

        # freeze the first row
        worksheet.freeze_panes = worksheet['A2']
        
        # set tab color
        worksheet.sheet_properties.tabColor = category.html_color

    workbook.save(response)

    return response

The Takeaways

  • Spreadsheet data can be used for further mathematical processing with pandas.
  • XLSX file format allows quite a bunch of formatting options that can make your spreadsheet data more presentable and user-friendly.
  • To see Excel export in action, go to 1st things 1st, log in as a demo user, and navigate to project results where you can export them as XLSX. Feedback is always welcome.

Cover photo by Tim Evans.

2019-02-02

Equivalents in Python and JavaScript. Bonus

From time to time I google for the right syntax how to process lists and dictionaries in Python or arrays and objects in JavaScript. So I decided to extend my series of equivalents with those functions. After all, it's me too, who will be using the information I provide here.

All truthful elements

Sometimes we need to check from a list of conditions if all of them are true, or from a list of elements if all of them are not empty.

This can be checked with the following in Python:

items = [1, 2, 3]
all_truthy = all(items)
# True

And here is an equivalent in JavaScript:

items = [1, 2, 3];
all_truthy = items.every(Boolean);
// true

Any truthful elements

Similarly, we can check if at least one of the conditions is true, or there is at least one non-empty element in a list.

It Python we would do that with:

items = [0, 1, 2, 3]
some_truthy = any(items)
# True

And in JavaScript we would check it like this:

items = [0, 1, 2, 3];
some_truthy = items.some(Boolean);
// true

Iterate through each element and its index

Here is an example of how to iterate through a list of items and also check their indices in Python. It is useful for verbose console output when creating different command line tools that process data:

items = ['a', 'b', 'c', 'd']
for index, element in enumerate(items):
    print(f'{index}: {element};')

In JavaScript an analogous way to do the same would be using the forEach() method. The usual for loop is also an option, but I find the forEach() more elegant and clear.

items = ['a', 'b', 'c', 'd'];
items.forEach(function(element, index) {
    console.log(`${index}: ${element};`);
});

Map elements to the results of a function

To process all elements of a list, you can either iterate through them with the for loop and create a new list with modifications, or you can do that in one step by mapping the list items to a modification function. In Python this can be done with the map() function:

items = [0, 1, 2, 3]
all_doubled = list(map(lambda x: 2 * x, items))
# [0, 2, 4, 6]

In JavaScript the map() is a method of an array:

items = [0, 1, 2, 3];
all_doubled = items.map(x => 2 * x);
// [0, 2, 4, 6]

Filter elements by a function

When you need to search for some elements in a list or array and want to avoid for loops, you can use the filtering functionality. In Python that is doable with the filter() function that accepts the filtering function and the list and returns a new filtered list.

items = [0, 1, 2, 3]
only_even = list(filter(lambda x: x % 2 == 0, items))
# [0, 2]

In JavaScript there is a filter() method of the array for that.

items = [0, 1, 2, 3];
only_even = items.filter(x => x % 2 === 0);
// [0, 2]

In both cases, the filtering function checks each item if it is matching the filter criteria and returns true in that case.

Reduce elements by a function to a single value

When you want to apply some function to a list of items to get a single result in one go, you can use the reduce function. It works for summing, multiplying, ORing, ANDing, or checking maximums and minimums.

In Python there is a reduce() function for that.

from functools import reduce
items = [1, 2, 3, 4]
total = reduce(lambda total, current: total + current, items)
# 10

In JavaScript there is a reduce() method of the array.

items = [1, 2, 3, 4];
total = items.reduce((total, current) => total + current);
// 10

Merge dictionaries

There are multiple ways to merge dictionaries in Python or objects in JavaScript. But these are probably the simplest ones.

In Python it's decomposing dictionaries to tuples of keys and arrays, joining them, and creating a new dictionary.

d1 = {'a': 'A', 'b': 'B'}
d2 = {'a': 'AAA', 'c': 'CCC'}
merged = dict(list(d1.items()) + list(d2.items()))
# {'a': 'AAA', 'b': 'B', 'c': 'CCC'}

Analogously, in JavaScript it's spreading two objects into a new object:

d1 = {a: 'A', b: 'B'}
d2 = {a: 'AAA', c: 'CCC'}
merged = {...d1, ...d2};
// {a: 'AAA', b: 'B', c: 'CCC'}

The Takeaways

  • In both languages, you can traverse through lists of items without explicitly incrementing and referencing an index.
  • For processing list items, you don't necessarily need a loop. The dedicated methods or functions all() / every(), any() / some(), map(), filter(), and reduce() are there to help you.
  • In both languages, you can merge multiple dictionaries into one. If the same key appears in several dictionaries, the latest one will be used in the merged dictionary.

Of course, I also updated the cheat sheet with the full list of equivalents in Python and JavaScript that you saw here described. This cheat sheet helps me with a good overview next to my laptop, so I believe that it would be helpful to you too. The new revision 10 is with syntax highlighting, so it makes it even better to explore and understand.

Get the Ultimate Cheat Sheet of Equivalents in Python and JavaScript

Use it for good!


Cover photo by Darren Chan.

2019-01-11

How to Create PDF Documents with Django in 2019

If you've read my Web Development with Django Cookbook, you might remember a recipe for creating PDF documents using Pisa xhtml2pdf. Well, this library does its job, but it supports only a subset of HTML and CSS features. For example, for multi-column layouts, you have to use tables, like it's 1994.

I needed some fresh and flexible option to generate donation receipts for the donation platform www.make-impact.org and reports for the strategic planner 1st things 1st I have been building. After a quick research I found another much more suitable library. It's called WeasyPrint. In this article, I will tell you how to use it with Django and what's valuable in it.

Features

WeasyPrint uses HTML and CSS 2.1 to create pixel-perfect, or let's rather say point-perfect, PDF documents. WeasyPrint doesn't use WebKit or Gecko but has its own rendering engine. As a proof that it works correctly, it passes the famous among web developers Acid2 test which was created back in the days before HTML5 to check how compatible browsers are with CSS 2 standards.

All supported features (and unsupported exceptions) are listed in the documentation. But my absolute favorites are these:

  • Layouts with floated elements. You don't have to use tables anymore if you want to have the recipient address on the left side and the sender information on the right side in a letter, or if you want to have the main content and the side notes in an exercise book. Just use floated elements.
  • Working links. The generated document can have clickable links to external URLs and internal anchors. You can straightforwardly create a clickable table of contents or a banner that leads back to your website.
  • Support for web fonts. With the wide variety of embeddable web fonts, your documents don't need to look boring anymore. Why not write titles in elegant cursive or in bold western letters?
  • Background images. By default, when you print an HTML page, all foreground images get printed, but the backgrounds are skipped. When you generate a PDF document for printing, you can show background images anywhere, even in the margins of the printed page.
  • SVG kept as vector images. When you have diagrams and graphics in a PDF document, you usually want to preserve the quality of the lines. Even if they look good on the screen, raster images might be not what you want, because on a printed page the resolution will differ and the quality can be lost. WeasyPrint keeps SVG images as vector images, so you have the highest possible quality in the prints.

Important Notes

WeasyPrint needs Python 3.4 or newer. That's great for new Django projects, but might be an obstacle if you want to integrate it into an existing website running on Python 2.7. Can it be the main argumentation for you to upgrade your old Django projects to the new Python version?

WeasyPrint is dependent on several OS libraries: Pango, GdkPixbuf, Cairo, and Libffi. In the documentation, there are understandable one-line instructions how to install them on different operating systems. You can have a problem only if you don't have full control of the server where you are going to deploy your project.

If you need some basic headers and footers for all pages, you can use @page CSS selector for that. If you need extended headers and footers for each page, it's best to combine the PDF document out of separate HTML documents for each page. Examples follow below.

The fun fact, Emojis are drawn using some weird raster single-color font. I don't recommend using them in your PDFs unless you replace them with SVG images.

Show Me the Code

A technical article is always more valuable when it has some quick code snippets to copy and paste. Here you go!

Simple PDF View

This snippet generates a donation receipt and shows it directly in the browser. Should the PDF be downloadable immediately, change content disposition from inline to attachment.

# -*- coding: UTF-8 -*-
from __future__ import unicode_literals

from django.http import HttpResponse
from django.template.loader import render_to_string
from django.utils.text import slugify
from django.contrib.auth.decorators import login_required

from weasyprint import HTML
from weasyprint.fonts import FontConfiguration

from .models import Donation

@login_required
def donation_receipt(request, donation_id):
    donation = get_object_or_404(Donation, pk=donation_id, user=request.user)
    response = HttpResponse(content_type="application/pdf")
    response['Content-Disposition'] = "inline; filename={date}-{name}-donation-receipt.pdf".format(
        date=donation.created.strftime('%Y-%m-%d'),
        name=slugify(donation.donor_name),
    )
    html = render_to_string("donations/receipt_pdf.html", {
        'donation': donation,
    })

    font_config = FontConfiguration()
    HTML(string=html).write_pdf(response, font_config=font_config)
    return response

Page Configuration Using CSS

Your PDF document can have a footer with an image and text on every page, using background-image and content properties:

{% load staticfiles i18n %}
<link href="https://fonts.googleapis.com/css?family=Playfair+Display:400,400i,700,700i,900" rel="stylesheet" />
<style>
@page {
    size: "A4";
    margin: 2.5cm 1.5cm 3.5cm 1.5cm;
    @bottom-center {
        background: url({% static 'site/img/logo-pdf.svg' %}) no-repeat center top;
        background-size: auto 1.5cm;
        padding-top: 1.8cm;
        content: "{% trans "Donation made via www.make-impact.org" %}";
        font: 10pt "Playfair Display";
        text-align: center;
        vertical-align: top;
    }
}
</style>

Pagination

You can show page numbers in the footer using CSS as follows.

@page {
    margin: 3cm 2cm;
    @top-center {
        content: "Documentation";
    }
    @bottom-right {
        content: "Page " counter(page) " of " counter(pages); 
    }
}

Horizontal Page Layout

You can rotate the page to horizontal layout with size: landscape.

@page {
    size: landscape;
}

HTML-based Footer

Another option to show an image and text in the header or footer on every page is to use an HTML element with position: fixed. This way you have more flexibility about formatting, but the element on all your pages will have the same content.

<style>
footer {
    position: fixed;
    bottom: -2.5cm;
    width: 100%;
    text-align: center;
    font-size: 10pt;
}
footer img {
    height: 1.5cm;
}
</style>
<footer>
    {% with website_url="https://www.make-impact.org" %}
        <a href="{{ website_url }}">
            <img alt="" src="{% static 'site/img/logo-contoured.svg' %}" />
        </a><br />
        {% blocktrans %}Donation made via <a href="{{ website_url }}">www.make-impact.org</a>{% endblocktrans %}
    {% endwith %}
</footer>

Document Rendering from Page to Page

When you need to have a document with complex unique headers and footers, it is best to render each page as a separate HTML document and then to combine them into one. This is how to do that:

def letter_pdf(request, letter_id):
    letter = get_object_or_404(Letter, pk=letter_id)
    response = HttpResponse(content_type='application/pdf')
    response['Content-Disposition'] = (
        'inline; '
        f'filename={letter.created:%Y-%m-%d}-letter.pdf'
    )
    COMPONENTS = [
        'letters/pdf/cover.html',
        'letters/pdf/page01.html',
        'letters/pdf/page02.html',
        'letters/pdf/page03.html',
    ]
    documents = []
    font_config = FontConfiguration()
    for template_name in COMPONENTS:
        html = render_to_string(template_name, {
            'letter': letter,
        })
        document = HTML(string=html).render(font_config=font_config)
        documents.append(document)

    all_pages = [page for document in documents for page in document.pages]
    documents[0].copy(all_pages).write_pdf(response)

    return response

Final Thoughts

I believe that WeasyPrint could be used not only for invoices, tickets, or booking confirmations but also for online magazines and small booklets. If you want to see PDF rendering with WeasyPrint in action, make a donation to your chosen organization at www.make-impact.org (when it's ready) and download the donation receipt. Or check the demo account at my.1st-things-1st.com and find the button to download the results of a prioritization project as PDF document.


Cover photo by Daniel Korpai.

2018-12-19

What's New in the Third Edition of Web Development with Django Cookbook?

A couple of months ago the third release of Django Cookbook was published under the title Django 2 Web Development Cookbook - Third Edition. This edition was thoroughly and progressively re-written by Jake Kronika, the guy who had reviewed my second edition and had added a lot of value to it. I was sure that he wouldn't disappoint the readers, so I invited him to write the update. In this article, I will guide you through the main new highlights of over 500 pages of this new book.

Up to Date

Just like William S. Vincent's books, Django 2 Web Development Cookbook - Third Edition is adapted to Django 2.1 and Python 3.6. So you will be dealing with the state of the art technologies building your Django projects. Unicode strings, f-strings, super() without parameters, HTML5 tags, and object-oriented JavaScript to mention a few are used all over the book. The code is carefully generalized and even more adapted to the Don't-Repeat-Yourself (DRY) principle.

Working with Docker

Docker is one of the most popular deployment technologies and Jake gives a good compact introduction how to use it with Django.

Using Environment Variables for Configuration

12-factor app guidelines suggest saving app configuration in environment variables. In the book, there is a practical example of how to use it.

Multilingual Fields even with Region-specific Language Codes

I introduced multilingual fields in previous editions of the book, but there they had a limitation, that region-specific languages like Australian English or Swiss German were not supported. Now they are!

Using Precisely Semantic Markup with schema.org Microdata

Schema.org Microdata allows you to define the context of the content more specifically so that the content is more machine-readable. This was new to me and I still don't know the exact practical value of it, but I guess it is related to accessibility, new ways of presenting data via plugins, and Artificial Intelligence.

Defining Custom Templates for the Default Django Form Fields

Since Django 1.11 form fields are rendered using templates instead of Python code and those templates can be customized. There is a recipe that shows you how to do that.

Providing Responsive Images

HTML5 has the <picture> tag with <source> children that can be used in combination with the sorl-thumbnail Python package to generate different versions of the image based on your viewport size: load small image on the mobile, middle image on the tablet, and big image on the desktop or smart TV.

Uploading Images and Deleting them by Ajax

In my previous editions, I only showed how to upload a file by Ajax and attach it to a Django model. In Jake's update, it is shown how you can also delete the image.

Validating Passwords with Special Requirements

Since Django 1.11 you can define special requirements for the passwords of your users, for example, have a mix of small and big letters or include at least 1 number and 3 special characters, etc. There is a practical recipe how to do that.

Adding Watermarks to Images

When it comes to branding or copyright protection, it is common to add special watermarks, semitransparent images on top of your normal pictures. Jake added an example, how to do that and it was very interesting to me.

Authenticating with Auth0

In one recipe it is shown how to login to a Django website using Auth0, which seems to be a passwordless authentication system with integrations of popular connection services like OpenID Connect, Facebook, Google, Github, LinkedIn, PayPal, Yahoo!, and others. I haven't tried that myself yet, but it can be an interesting option for a social website.

Using Redis for Caching

It is common to cache websites using Memcached service, but a good alternative is caching with Redis and django-redis. Moreover, you can easily save user sessions there.

Creating Hierarchies with django-treebeard

In the previous editions, I introduced django-mptt for creating hierarchical structures. However, recently many projects are moving towards its alternative - django-treebeard which has more stability and writing speed. In the book, Jake shows you how to work with it.

Conclusion

There was a lot of new things to learn. For example, for me personally Docker usage was new, and I haven't heard of schema.org microdata and Auth0 which were introduced in this book. All in all, I think, Jake Kronika did an enormous job with this update and it's really worth purchasing this book, especially as there is a winter-holidays sale where you can get the EPUB, MOBI, and PDF with the code examples just for ~ 5 €.

Have a nice Christmas time and come back to this blog next year!


Cover photo by chuttersnap.

2018-07-09

Equivalents in Python and JavaScript. Part 4

In the last three parts of the series of articles about analogies in Python and JavaScript, we explored lots of interesting concepts like serializing to JSON, error handling, using regular expressions, string interpolation, generators, lambdas, and many more. This time we will delve into function arguments, creating classes, using class inheritance, and defining getters and setters of class properties.

Function arguments

Python is very flexible with argument handling for functions: you can set default values there, allow a flexible amount of positional or keyword arguments (*args and **kwargs). When you pass values to a function, you can define by name to which argument that value should be assigned. All that in a way is now possible in JavaScript too.

Default values for function arguments in Python can be defined like this:

from pprint import pprint

def report(post_id, reason='not-relevant'):
    pprint({'post_id': post_id, 'reason': reason})
    
report(42)
report(post_id=24, reason='spam')

In JavaScript that can be achieved similarly:

function report(post_id, reason='not-relevant') {
    console.log({post_id: post_id, reason: reason});
}

report(42);
report(post_id=24, reason='spam');

Positional arguments in Python can be accepted using the * operator like this:

from pprint import pprint

def add_tags(post_id, *tags):
    pprint({'post_id': post_id, 'tags': tags})
    
add_tags(42, 'python', 'javascript', 'django')

In JavaScript positional arguments can be accepted using the ... operator:

function add_tags(post_id, ...tags) {
    console.log({post_id: post_id, tags: tags});
}

add_tags(42, 'python', 'javascript', 'django');    

Keyword arguments are often used in Python when you want to allow a flexible amount of options:

from pprint import pprint

def create_post(**options):
    pprint(options)

create_post(
    title='Hello, World!', 
    content='This is our first post.',
    is_published=True,
)
create_post(
    title='Hello again!',
    content='This is our second post.',
)

A common practice to pass multiple optional arguments to a JavaScript function is through a dictionary object, for example, options.

function create_post(options) {
    console.log(options);
}

create_post({
    'title': 'Hello, World!', 
    'content': 'This is our first post.',
    'is_published': true
});
create_post({
    'title': 'Hello again!', 
    'content': 'This is our second post.'
});

Classes and inheritance

Python is an object-oriented language. Since ECMAScript 6 standard support, it's also possible to write object-oriented code in JavaScript without hacks and weird prototype syntax.

In Python you would create a class with the constructor and a method to represent its instances textually like this:

class Post(object):
    def __init__(self, id, title):
        self.id = id
        self.title = title
        
    def __str__(self):
        return self.title

post = Post(42, 'Hello, World!')
isinstance(post, Post) == True
print(post)  # Hello, World!

In JavaScript to create a class with the constructor and a method to represent its instances textually, you would write:

class Post {
    constructor (id, title) {
        this.id = id;
        this.title = title;
    }
    toString() {
        return this.title;
    }
}

post = new Post(42, 'Hello, World!');
post instanceof Post === true;
console.log(post.toString());  // Hello, World!

Now we can create two classes Article and Link in Python that will extend the Post class. Here you can also see how we are using super to call methods from the base Post class.

class Article(Post):
    def __init__(self, id, title, content):
        super(Article, self).__init__(id, title)
        self.content = content

class Link(Post):
    def __init__(self, id, title, url):
        super(Link, self).__init__(id, title)
        self.url = url
        
    def __str__(self):
        return '{} ({})'.format(
            super(Link, self).__str__(),
            self.url,
        )
           
article = Article(1, 'Hello, World!', 'This is my first article.')
link = Link(2, 'DjangoTricks', 'https://djangotricks.blogspot.com')
isinstance(article, Post) == True
isinstance(link, Post) == True
print(link)
# DjangoTricks (https://djangotricks.blogspot.com)

In JavaScript the same is also doable by the following code:

class Article extends Post {
    constructor (id, title, content) {
        super(id, title);
        this.content = content;
    }
}

class Link extends Post {
    constructor (id, title, url) {
        super(id, title);
        this.url = url;
    }
    toString() {
        return super.toString() + ' (' + this.url + ')';
    }
}

article = new Article(1, 'Hello, World!', 'This is my first article.');
link = new Link(2, 'DjangoTricks', 'https://djangotricks.blogspot.com');
article instanceof Post === true;
link instanceof Post === true;
console.log(link.toString());
// DjangoTricks (https://djangotricks.blogspot.com)

Class properties: getters and setters

In object oriented programming, classes can have attributes, methods, and properties. Properties are a mixture of attributes and methods. You deal with them as attributes, but in the background they call special getter and setter methods to process data somehow before setting or returning to the caller.

The basic wireframe for getters and setters of the slug property in Python would be like this:

class Post(object):
    def __init__(self, id, title):
        self.id = id
        self.title = title
        self._slug = ''
        
    @property
    def slug(self):
        return self._slug
        
    @slug.setter
    def slug(self, value):
        self._slug = value
            
post = new Post(1, 'Hello, World!')
post.slug = 'hello-world'
print(post.slug)

In JavaScript getters and setters for the slug property can be defined as:

class Post {
    constructor (id, title) {
        this.id = id;
        this.title = title;
        this._slug = '';
    }
    
    set slug(value) {
        this._slug = value;
    }
    
    get slug() {
        return this._slug;
    }
}

post = new Post(1, 'Hello, World!');
post.slug = 'hello-world';
console.log(post.slug);

The Takeaways

  • In both languages, you can define default argument values for functions.
  • In both languages, you can pass a flexible amount of positional or keyword arguments for functions.
  • In both languages, object-oriented programming is possible.

As you might have noticed, I am offering a cheat sheet with the full list of equivalents in Python and JavaScript that you saw here described. At least for me, it is much more convenient to have some printed sheet of paper with valuable information next to my laptop, rather than switching among windows or tabs and scrolling to get the right piece of snippet. So I encourage you to get this cheat sheet and improve your programming!

Get the Ultimate Cheat Sheet of Equivalents in Python and JavaScript

Use it for good!


Cover photo by Andre Benz

2018-07-06

Equivalents in Python and JavaScript. Part 3

This is the 3rd part of 4-article series about analogies in Python and JavaScript. In the previous parts we covered a large part of the traditional Python 2.7 and JavaScript based on the ECMAScript 5 standard. This time we will start looking into Python 3.6 and JavaScript based on the ECMAScript 6 standard. ECMAScript 6 standard is pretty new and supported only the newest versions of browsers. For older browsers you will need Babel to compile your next-generation JavaScript code to the cross-browser-compatible equivalents. It opens the door to so many interesting things to explore. We will start from string interpolation, unpacking lists, lambda functions, iterations without indexes, generators, and sets!

Variables in strings

The old and inefficient way to build strings with values from variables is this concatenation:

name = 'World'
value = 'Hello, ' + name + '!\nWelcome!'

This can get very sparse and difficult to read. Also it is very easy to miss whitespaces in the sentence around variables.

Since Python version 3.6 and JavaScript based on the ECMAScript 6 standard, you can use so called string interpolation. These are string templates which are filled in with values from variables.

In Python they are also called f-string, because their notation starts with letter "f":

name = 'World'
value = f"""Hello, {name}!
Welcome!"""

price = 14.9
value = f'Price: {price:.2f} €'  # 'Price: 14.90 €'

In JavaScript string templates start and end with backticks:

name = 'World';
value = `Hello, ${name}!
Welcome!`;

price = 14.9;
value = `Price ${price.toFixed(2)} €`;  // 'Price: 14.90 €'

Note that string templates can be of a single line as well as of multiple lines. For f-strings in Python you can pass the format for variables, but you can't call methods of a variable unless they are properties and call getter methods.

Unpacking lists

Python and now JavaScript has an interesting feature to assign items of sequences into separate variables. For example, we can read the three values of a list into variables a, b, and c with the following syntax:

[a, b, c] = [1, 2, 3]

For tuples the parenthesis can be omitted. The following is a very popular way to swap values of two variables in Python:

a = 1
b = 2
a, b = b, a  # swap values

With the next generation JavaScript this can also be achieved:

a = 1;
b = 2;
[a, b] = [b, a];  // swap values

In Python 3.6 if we have an unknown number of items in a list or tuple, we can assign them to a tuple of several values while also unpacking the rest to a list:

first, second, *the_rest = [1, 2, 3, 4]
# first == 1
# second == 2
# the_rest == [3, 4]

This can also be done with JavaScript (ECMAScript 6):

[first, second, ...the_rest] = [1, 2, 3, 4];
// first === 1
// last === 2
// the_rest === [3, 4]

Lambda functions

Python and JavaScript have a very neat functionality to create functions in a single line. These functions are called lambdas. Lambdas are very simple functions that take one or more arguments and return some calculated value. Usually lambdas are used when you need to pass a function to another function as a callback or as a function to manipulate every separate elements in a sequence.

In Python, you would define a lambda using the lambda keyword, like this:

sum = lambda x, y: x + y
square = lambda x: x ** 2

In JavaScript lambdas use the => notation. If there are more than one arguments, they have to be in parenthesis:

sum = (x, y) => x + y;
square = x => Math.pow(x, 2);

Iteration without indexes

Many programming languages allow iterating through a sequence only by using indexes and incrementing their values. Then to get an item at some position, you would read it from an array, for example:

for (i=0; i<items.length; i++) {
    console.log(items[i]);
}

This is not a nice syntax and is very technical - it doesn't read naturally. What we really want is just to grab each value from the list. And Python has a very neat possibility just to iterate through the elements:

for item in ['A', 'B', 'C']:
    print(item)

In the modern JavaScript this is also possible with the for..of operator:

for (let item of ['A', 'B', 'C']) {
    console.log(item);
}

You can also iterate through a string characters in Python:

for character in 'ABC':
    print(character)

And in the modern JavaScript:

for (let character of 'ABC') {
    console.log(character);
}

Generators

Python and modern JavaScript has a possibility to define special functions through which you can iterate. With each iteration they return the next generated value in a sequence. These functions are called generators. With generators you can get numbers in a range, lines from a file, data from different paginated API calls, fibonacci numbers, and any other dynamicly generated sequences.

Technically generators are like normal functions, but instead of returning a value, they yield a value. This value will be returned for one iteration. And this generation happens as long as the end of the function is reached.

To illustrate that, the following Python code will create a generator countdown() which will return numbers from the given one back to 1, (like 10, 9, 8, ..., 1):

def countdown(counter):
    while counter > 0:
        yield counter
        counter -= 1
        
for counter in countdown(10):
    print(counter)

Exactly the same can be achieved in modern JavaScript, but notice the asterisk at the function statement. It defines that it's a generator:

function* countdown(counter) {
    while (counter > 0) {
        yield counter;
        counter--;
    }
}
for (let counter of countdown(10)) {
    console.log(counter);
}

Sets

We already had a look at lists, tuples and arrays. But here is another type of data - sets. Sets are groups of elements that ensure that each element there exists only once. Set theory also specifies set operations like union, intersection, and difference, but we won't cover them here today.

This is how to create a set, add elements to it, check if a value exists, check the total amount of elements in a set, and iterate through its values, and remove a value using Python:

s = set(['A'])
s.add('B'); s.add('C')
'A' in s
len(s) == 3
for elem in s:
    print(elem)
s.remove('C')

Here is how to achieve the same in modern JavaScript:

s = new Set(['A']);
s.add('B').add('C');
s.has('A') === true;
s.size === 3;
for (let elem of s.values()) {
    console.log(elem);
}
s.delete('C')

The Takeaways

  • String interpolation or literal templates allows you to have much cleaner and nicer code even with a possibility to have multiple lines of text.
  • You can iterate through elements in a sequence or group without using indexes.
  • Use generators when you have a sequence of nearly unlimited elements.
  • Use sets when you want to ensure fast check if data exists in a collection.
  • Use lambdas when you need short and clear single-line functions.

As you know from the previous parts, I am offering a cheat sheet with the whole list of equivalents in Python and JavaScript, both, time honored and future proof. To have something printed in front of your eyes is much more convenient than switching among windows or scrolling up and down until you find what you exactly were searching for. So I suggest you to get the cheat sheet and use it for good!

Get the Ultimate Cheat Sheet of Equivalents in Python and JavaScript

In the next and last part of the series, we will have a look at function arguments, classes, inheritance, and properties. Stay tuned!


Cover photo by Alex Knight

2018-07-03

Equivalents in Python and JavaScript. Part 2

Last time we started a new series of articles about analogies in Python and JavaScript. We had a look at lists, arrays, dictionaries, objects, and strings, conditional assignments, and parsing integers. This time we will go through more interesting and more complex things like serializing dictionaries and lists to JSON, operations with regular expressions, as well as raising and catching errors.

JSON

When working with APIs it is very usual to serialize objects to JSON format and be able to parse JSON strings.

In Python it is done with the json module like this:

import json
json_data = json.dumps(dictionary, indent=4)
dictionary = json.loads(json_data)

Here we'll indent the nested elements in the JSON string by 4 spaces.

In JavaScript there is a JSON object that has methods to create and parse JSON strings:

json_data = JSON.stringify(dictionary, null, 4);
dictionary = JSON.parse(json_data);

Splitting strings by regular expressions

Regular expressions are multi-tool that once you master, you can accomplish lots of things.

In the last article, we saw how one can join lists of strings into a single string. But how can you split a long string into lists of strings? What if the delimiter can be not a single character as the comma, but a range of possible variations? This can be done with regular expressions and the split() method.

In Python, the split() method belongs to the regular expression pattern object. This is how you could split a text string into sentences by punctuation marks:

import re

# One or more characters of "!?." followed by whitespace
delimiter = re.compile(r'[!?\.]+\s*')

text = "Hello!!! What's new? Follow me."
sentences = delimiter.split(text)
# sentences == ['Hello', "What's new", 'Follow me', '']

In JavaScript the split() method belongs to the string:

// One or more characters of "!?." followed by whitespace
delimiter = /[!?\.]+\s*/;

text = "Hello!!! What's new? Follow me.";
sentences = text.split(delimiter)
// sentences === ["Hello", "What's new", "Follow me", ""]

Matching regular expression patterns in strings

Regular expressions are often used to validate data from the forms.

For example, to validate if the entered email address is correct, you would need to match it against a regular expression pattern. In Python that would look like this:

import re

# name, "@", and domain
pattern = re.compile(r'([\w.+\-]+)@([\w\-]+\.[\w\-.]+)')

match = pattern.match('hi@example.com')
# match.group(0) == 'hi@example.com'
# match.group(1) == 'hi'
# match.group(2) == 'example.com'

If the text matches the pattern, it returns a match object with the group() method to read the whole matched string, or separate captures of the pattern that were defined with the parenthesis. 0 means getting the whole string, 1 means getting the match in the first group, 2 means getting the match in the second group, and so on. If the text doesn't match the pattern, the None value will be returned.

In JavaScript the match() method belongs to the string and it returns either a match object, or null. Pretty similar:

// name, "@", and domain
pattern = /([\w.+\-]+)@([\w\-]+\.[\w\-.]+)/;

match = 'hi@example.com'.match(pattern);
// match[0] === 'hi@example.com'
// match[1] === 'hi'
// match[2] === 'example.com'

The match object in JavaScript acts as an array. Its value at the zeroth position is the whole matched string. The other indexes correspond to the captures of the pattern defined with the parenthesis.


Moreover, sometimes you need to search if a specific value exists in a string and at which letter position it will be found. That can be done with the search() method.

In Python this method belongs to the regular expression pattern and it returns the match object. The match object has the start() method telling at which letter position the match starts:

text = 'Say hi at hi@example.com'
first_match = pattern.search(text)
if first_match:
    start = first_match.start()  # start == 10

In JavaScript the search() method belongs to the string and it returns just an integer telling at which letter position the match starts. If nothing is found, -1 is returned:

text = 'Say hi at hi@example.com';
first_match = text.search(pattern);
if (first_match > -1) {
    start = first_match;  // start === 10
}

Replacing patterns in strings using regular expressions

Replacing with regular expressions usually happen when cleaning up data, or adding additional features. For example, we could take some text and make all email addresses clickable.

Python developers would use the sub() method of the regular expression pattern:

html = pattern.sub(
    r'<a href="mailto:\g<0>">\g<0></a>',
    'Say hi at hi@example.com',
)
# html == 'Say hi at <a href="mailto:hi@example.com">hi@example.com</a>'

JavaScript developers would use the replace() method of the string:

html = 'Say hi at hi@example.com'.replace(
    pattern, 
    '<a href="mailto:$&">$&</a>',
);
// html === 'Say hi at <a href="mailto:hi@example.com">hi@example.com</a>'

In Python the captures, also called as "backreferences", are accessible in the replacement string as \g<0>, \g<1>, \g<2>, etc. In JavaScript the same is accessible as $&, $1, $2, etc. Backreferences are usually used to wrap some strings or to switch places of different pieces of text.


It is also possible to replace a match with a function call. This can be used to do replacements within replacements or to count or collect some features of a text. For example, using replacements with function calls in JavaScript, I once wrote a fully functional HTML syntax highlighter.

Here let's change all email addresses in a text to UPPERCASE.

In Python, the replacement function receives the match object. We can use its group() method to do something with the matched text and return a text as a replacement:

text = pattern.sub(
    lambda match: match.group(0).upper(), 
    'Say hi at hi@example.com',
)
# text == 'Say hi at HI@EXAMPLE.COM'

In JavaScript the replacement function receives the whole match string, the first capture, the second capture, etc. We can do what we need with those values and then return some string as a replacement:

text = 'Say hi at hi@example.com'.replace(
    pattern,
    function(match, p1, p2) {
        return match.toUpperCase();
    }
);
// text === 'Say hi at HI@EXAMPLE.COM'

Error handling

Contrary to Python, client-side JavaScript normally isn't used for saving or reading files or connecting to remote databases. So try..catch blocks are quite rare in JavaScript compared to try..except analogy in Python.

Anyway, error handling can be used with custom user errors implemented and raised in JavaScript libraries and caught in the main code.

The following example in Python shows how to define a custom exception class MyException, how to raise it in a function, and how to catch it and handle in a try..except..finally block:

class MyException(Exception):
    def __init__(self, message):
        self.message = message
        
    def __str__(self):
        return self.message
        
def proceed():
    raise MyException('Error happened!')

try:
    proceed()
except MyException as err:
    print('Sorry! {}'.format(err))
finally:
    print('Finishing')    

The following example in JavaScript does exactly the same: here we define a MyException class, throw it in a function, and catch it and handle in the try..catch..finally block.

function MyException(message) {
   this.message = message;
   this.toString = function() {
       return this.message;
   }
}

function proceed() {
    throw new MyException('Error happened!');
}

try {
    proceed();
} catch (err) {
    if (err instanceof MyException) {
        console.log('Sorry! ' + err);
    }
} finally {
    console.log('Finishing');
}

The MyException class in both languages has a parameter message and a method to represent itself as a string using the value of the message.

Of course, exceptions should be raised/thrown just in the case of errors. And you define what is an error in your module design.

The Takeaways

  • Serialization to JSON is quite straightforward in both, Python and JavaScript.
  • Regular expressions can be used as multi-tools when working with textual data.
  • You can do replacements with function calls in both languages.
  • For more sophisticated software design you can use custom error classes.

As I mentioned last time, you can grab a side-by-side comparison of Python and JavaScript that I compiled for you (and my future self). Side by side you will see features from traditional list, array, dictionary, object, and string handling to modern string interpolation, lambdas, generators, sets, classes, and everything else. Use it for good.

Get the Ultimate Cheat Sheet of Equivalents in Python and JavaScript

In the next part of the series, we will have a look at textual templates, list unpacking, lambda functions, iteration without indexes, generators, and sets. Stay tuned!


Cover photo by Benjamin Hung.

2018-06-30

Equivalents in Python and JavaScript. Part 1

Although Python and JavaScript are quite different languages, there are some analogies which full stack Python developers should know when developing web projects. In this series of 4 parts, I will explore what is similar in each of those languages and what are the common ways to solve common problems. This is not meant to be a reference and I will skip the basics like primitive variable types, conditions, and loops. But I will dig into more complex structures and data operations using both, Python and JavaScript. Also, I will try to focus on the practical use cases. This series should be interesting for the developers of Django, Flask, or another Python framework who want to get a grasp of traditional and modern vanilla JavaScript. On the other hand, it will be useful for the front-enders who want to better understand how the backend is working and maybe even start their own Django website.

Parsing integers

We'll begin with integer parsing.

In Python that's straightforward:

number = int(text)

But in JavaScript you have to explain what number system you expect: decimal, octal, hexadecimal, or binary:

number = parseInt(text, 10);

To use the "normal" decimal number system we are passing number 10 as the second parameter of the parseInt() function. 8 goes for octal, 16 for hexadecimal, or 2 – for binary. If the second parameter is missing, the number in text starts with zero, and you are using a slightly older browser, the number in the text will be interpreted as octal. For example,

parseInt('012') == 10  // in some older browsers
parseInt('012', 10) == 12

And that can really mess up your calculations.

Conditional assignment

For conditional assignment, Python and JavaScript have different syntaxes, but conditional assignments are quite popular in both languages. That's popular, because it's just a single statement to have a condition checking, the true-case value, and the false-case value.

Since Python 2.7 you can write conditional assignments like this:

value = 'ADULT' if age >= 18 else 'CHILD'

In JavaScript conditional assignments are done using ternary operator ?:, similar to the ones in C, C++, C#, Java, Ruby, PHP, Perl, Swift, and ActionScript:

value = age >= 18? 'ADULT': 'CHILD';

Object attribute value by attribute name

The normal way to access an object's attribute is by the dot notation in both, Python and JavaScript:

obj.color = 'YELLOW'

But what if you want to refer to an attribute by its name saved as a string? For example, the attribute name could be coming from a list of attributes or the attribute name is combined from two strings like 'title_' + lang_code.

For that reason, in Python, there are functions getattr() and setattr(). I use them a lot.

attribute = 'color'
value = getattr(obj, attribute, 'GREEN')
setattr(obj, attribute, value)

In JavaScript you can treat an object like a dictionary and pass the attribute name in square brackets:

attribute = 'color';
value = obj[attribute] || 'GREEN';
obj[attribute] = value;

To retrieve a default value when an object has no such attribute, in Python, getattr() has the third parameter. In JavaScript, if obj attribute doesn't exist, it will return the undefined value. Then it can be OR-ed with the default value that you want to assign. That's a common practice in JavaScript that you can find in many JavaScript libraries and frameworks.

Dictionary value by key

This is similar to the previous one. The normal way to assign a dictionary's value by key in both languages is using the square brackets:

dictionary = {}
dictionary['color'] = 'YELLOW'

To read a value in Python you can use the square-bracket notation, but it will fail on non-existing keys with KeyError. The more flexible way is to use the get() method which returns None for non-existing keys. Also you can pass an optional default value as the second parameter:

key = 'color'
value = dictionary.get(key, 'GREEN')

In JavaScript you would use the same trick as with object attributes, because dictionaries and objects are the same there:

key = 'color';
value = dictionary[key] || 'GREEN';

Slicing lists and strings

Python has the slice [:] operator to get parts of lists, tuples, and similar more complex structures, for example Django QuerySets:

items = [1, 2, 3, 4, 5]
first_two = items[:2]      # [1, 2]
last_two = items[-2:]      # [4, 5]
middle_three = items[1:4]  # [2, 3, 4]

In JavaScript arrays have the slice() method with the same effect and similar usage:

items = [1, 2, 3, 4, 5];
first_two = items.slice(0, 2);     // [1, 2] 
last_two = items.slice(-2);        // [4, 5]
middle_three = items.slice(1, 4);  // [2, 3, 4]

But don't mix it up with the splice() method which modifies the original array!


The [:] slice operator in Python also works for strings:

text = 'ABCDE'
first_two = text[:2]      # 'AB'
last_two = text[-2:]      # 'DE'
middle_three = text[1:4]  # 'BCD'

In JavaScript strings just like arrays have the slice() method:

text = 'ABCDE';
first_two = text.slice(0, 2);    // 'AB'
last_two = text.slice(-2);       // 'DE'
middle_three = text.slice(1, 4); // 'BCD'

Operations with list items

In programming it is very common to collect and analyze sequences of elements. In Python that is usually done with lists and in JavaScript with arrays. They have similar syntax and operations, but different method names to add and remove values.

This is how to concatenate two lists, add one value to the end, add one value to the beginning, get and remove a value from the beginning, get and remove a value from the end, and delete a certain value by index in Python:

items1 = ['A']
items2 = ['B']
items = items1 + items2  # items == ['A', 'B']
items.append('C')        # ['A', 'B', 'C']
items.insert(0, 'D')     # ['D', 'A', 'B', 'C']
first = items.pop(0)     # ['A', 'B', 'C']
last = items.pop()       # ['A', 'B']
items.delete(0)          # ['B']

This is how to do exactly the same with arrays in JavaScript:

items1 = ['A'];
items2 = ['B'];
items = items1.concat(items2);  // items === ['A', 'B']
items.push('C');                // ['A', 'B', 'C']
items.unshift('D');             // ['D', 'A', 'B', 'C']
first = items.shift();          // ['A', 'B', 'C']
last = items.pop();             // ['A', 'B']
items.splice(0, 1);             // ['B']

Joining lists of strings

It is very common after having a list or array of strings, to combine them into one string by a separator like comma or new line.

In Python that is done by the join() method of a string where you pass the list or tuple. Although it might feel unnatural, you start with the separator there. But I can assure that you get used to it after several times of usage.

items = ['A', 'B', 'C']
text = ', '.join(items)  # 'A, B, C'

In JavaScript the array has the join() method where you pass the separator:

items = ['A', 'B', 'C'];
text = items.join(', ');  // 'A, B, C'

The Takeaways

  • List and tuples in Python are similar to arrays in JavaScript.
  • Dictionaries in Python are similar to objects in JavaScript.
  • Strings in Python are similar to strings in JavaScript.
  • Numbers in JavaScript should be parsed with care.
  • Single-line conditional assignments exist in both languages.
  • Joining sequences of strings in Python is confusing, but you can quickly get used to it.

I compiled the whole list of equivalents of Python and JavaScript to a cheat sheet that you can print out and use for good. Side by side it compares traditional Python 2.7 and JavaScript based on ECMAScript 5 standard, as well as newer Python 3.6 and JavaScript based on ECMAScript 6 standard with such goodies as string interpolation, lambdas, generators, classes, etc.

Get the Ultimate Cheat Sheet of Equivalents in Python and JavaScript

In the next part of the series, we will have a look at JSON creation and parsing, operations with regular expressions, and error handling. Stay tuned!


Cover photo by Benjamin Hung.

2018-06-17

Data Filtering in a Django Website using Elasticsearch

In my Web Development with Django Cookbook section Forms and Views there is a recipe Filtering object lists. It shows you how to filter a Django QuerySet dynamically by different filter parameters selected in a form. From practice, the approach is working well, but with lots of data and complex nested filters, the performance might get slow. You know - because of all those INNER JOINS in SQL, the page might take even 12 seconds to load. And that is not preferable behavior. I know that I could denormalize the database or play with indices to optimize SQL. But I found a better way to increase the loading speed. Recently we started using Elasticsearch for one of the projects and its data filtering performance seems to be enormously faster: in our case, it increased from 2 to 16 times depending on which query parameters you choose.

What is Elasticsearch?

Elasticsearch is java-based search engine which stores data in JSON format and allows you to query it using special JSON-based query language. Using elasticsearch-dsl and django-elasticsearch-dsl, I can bind my Django models to Elasticsearch indexes and rewrite my object list views to use Elasticsearch queries instead of Django ORM. The API of Elasticsearch DSL is chainable like with Django QuerySets or jQuery functions, and we'll have a look at it soon.

The Setup

At first, let's install Elasticsearch server. Elasticsearch is quite a complex system, but it comes with convenient configuration defaults.

On macOS you can install and start the server with Homebrew:

$ brew install elasticsearch
$ brew services start elasticsearch

For other platforms, the installation instructions are also quite clear.

Then in your Django project's virtual environment install django-elasticsearch-dsl. I guess, "DSL" stands for "domain specific language".

With pipenv it would be the following from the project's directory:

$ pipenv install django-elasticsearch-dsl

If you are using just pip and virtual environment, then you would do this with your project's environment activated.

(venv)$ pip install django-elasticsearch-dsl

This, in turn, will install related lower level client libraries: elasticsearch-dsl and elasticsearch-py.

In the Django project settings, add 'django_elasticsearch_dsl' to INSTALLED_APPS.

Finally, add the lines defining default connection configuration there:

ELASTICSEARCH_DSL={
    'default': {
        'hosts': 'localhost:9200'
    },
}

Elasticsearch Documents for Django Models

For the illustration how to use Elasticsearch with Django, I'll create Author and Book models, and then I will create Elasticsearch index document for the books.

models.py

# -*- coding: UTF-8 -*-
from __future__ import unicode_literals

from django.db import models
from django.utils.translation import ugettext_lazy as _
from django.utils.encoding import python_2_unicode_compatible


@python_2_unicode_compatible
class Author(models.Model):
    first_name = models.CharField(_("First name"), max_length=200)
    last_name = models.CharField(_("Last name"), max_length=200)
    author_name = models.CharField(_("Author name"), max_length=200)

    class Meta:
        verbose_name = _("Author")
        verbose_name_plural = _("Authors")
        ordering = ("author_name",)

    def __str__(self):
        return self.author_name


@python_2_unicode_compatible
class Book(models.Model):
    title = models.CharField(_("Title"), max_length=200)
    authors = models.ManyToManyField(Author, verbose_name=_("Authors"))
    publishing_date = models.DateField(_("Publishing date"), blank=True, null=True)
    isbn = models.CharField(_("ISBN"), blank=True, max_length=20)

    class Meta:
        verbose_name = _("Book")
        verbose_name_plural = _("Books")
        ordering = ("title",)

    def __str__(self):
        return self.title

Nothing fancy here. Just an Author model with fields id, first_name, last_name, author_name, and a Book model with fields id, title, authors, publishing_date, and isbn. Let's go to the documents.

documents.py

In the same directory of your app, create documents.py with the following content:

# -*- coding: UTF-8 -*-
from __future__ import unicode_literals

from django_elasticsearch_dsl import DocType, Index, fields
from .models import Author, Book

# Name of the Elasticsearch index
search_index = Index('library')
# See Elasticsearch Indices API reference for available settings
search_index.settings(
    number_of_shards=1,
    number_of_replicas=0
)


@search_index.doc_type
class BookDocument(DocType):
    authors = fields.NestedField(properties={
        'first_name': fields.TextField(),
        'last_name': fields.TextField(),
        'author_name': fields.TextField(),
        'pk': fields.IntegerField(),
    }, include_in_root=True)

    isbn = fields.KeywordField(
        index='not_analyzed',
    )

    class Meta:
        model = Book # The model associated with this DocType

        # The fields of the model you want to be indexed in Elasticsearch
        fields = [
            'title',
            'publishing_date',
        ]
        related_models = [Author]

    def get_instances_from_related(self, related_instance):
        """If related_models is set, define how to retrieve the Book instance(s) from the related model."""
        if isinstance(related_instance, Author):
            return related_instance.book_set.all()

Here we defined a BookDocument which will have fields: title, publishing_date, authors, and isbn.

The authors will be a list of nested dictionaries at the BookDocument. The isbn will be a KeywordField which means that it will be not tokenized, lowercased, nor otherwise processed and handled the whole as is.

The values for those document fields will be read from the Book model.

Using signals, the document will be automatically updated either when a Book instance or Author instance is added, changed, or deleted. In the method get_instances_from_related(), we tell the search engine which books to update when an author is updated.

Building the Index

When the index document is ready, let's build the index at the server:

(venv)$ python manage.py search_index --rebuild

Django QuerySets vs. Elasticsearch Queries

The concepts of SQL and Elasticsearch queries are quite different. One is working with relational tables and the other works with dictionaries. One is using queries that are kind of human-readable logical sentences and another is using nested JSON structures. One is using the content verbosely and another does string processing in the background and gives search relevance for each result.

Even when there are lots of differences, I will try to draw analogies between Django ORM and elasticsearch-dsl API as close as possible.

1. Query definition

Django QuerySet:

queryset = MyModel.objects.all()

Elasticsearch query:

search = MyModelDocument.search()

2. Count

Django QuerySet:

queryset = queryset.count()

Elasticsearch query:

search = search.count()

3. Iteration

Django QuerySet:

for item in queryset:
    print(item.title)

Elasticsearch query:

for item in search:
    print(item.title)

4. To see the generated query:

Django QuerySet:

>>> queryset.query

Elasticsearch query:

>>> search.to_dict()

5. Filter by single field containing a value

Django QuerySet:

queryset = queryset.filter(my_field__icontains=value)

Elasticsearch query:

search = search.filter('match_phrase', my_field=value)

6. Filter by single field equal to a value

Django QuerySet:

queryset = queryset.filter(my_field__exact=value)

Elasticsearch query:

search = search.filter('match', my_field=value)

If a field type is a string, not a number, it has to be defined as KeywordField in the index document:

my_field = fields.KeywordField()

7. Filter with either of the conditions (OR)

Django QuerySet:

from django.db import models
queryset = queryset.filter(
    models.Q(my_field=value) |
    models.Q(my_field2=value2)
)

Elasticsearch query:

from elasticsearch_dsl.query import Q
search = search.query(
    Q('match', my_field=value) |
    Q('match', my_field2=value2)
)

8. Filter with all of the conditions (AND)

Django QuerySet:

from django.db import models
queryset = queryset.filter(
    models.Q(my_field=value) &
    models.Q(my_field2=value2)
)

Elasticsearch query:

from elasticsearch_dsl.query import Q
search = search.query(
    Q('match', my_field=value) & 
    Q('match', my_field2=value2)
)

9. Filter by values less than or equal to certain value

Django QuerySet:

from datetime import datetime

queryset = queryset.filter(
    published_at__lte=datetime.now(),
)

Elasticsearch query:

from datetime import datetime

search = search.filter(
    'range',
    published_at={'lte': datetime.now()}
)

10. Filter by a value in a nested field

Django QuerySet:

queryset = queryset.filter(
    category__pk=category_id,
)

Elasticsearch query:

search = search.filter(
    'nested', 
    path='category', 
    query('match', category__pk=category_id)
)

11. Filter by one of many values in a related model

Django QuerySet:

queryset = queryset.filter(
    category__pk__in=category_ids,
)

Elasticsearch query:

from django.utils.six.moves import reduce
from elasticsearch_dsl.query import Q

search = search.query(
    reduce(operator.ior, [
        Q(
            'nested', 
            path='category', 
            query('match', category__pk=category_id),
        )
        for category_id in category_ids
    ])
)

Here the reduce() function combines a list of Q() conditions using the bitwise OR operator (|).

12. Ordering

Django QuerySet:

queryset = queryset.order_by('-my_field', 'my_field2')

Elasticsearch query:

search = search.sort('-my_field', 'my_field2')

13. Creating query dynamically

Django QuerySet:

import operator
from django.utils.six.moves import reduce

filters = []
if value1:
    filters.append(models.Q(
        my_field1=value1,
    ))
if value2:
    filters.append(models.Q(
        my_field2=value2,
    ))
queryset = queryset.filter(
    reduce(operator.iand, filters)
)

Elasticsearch query:

import operator
from django.utils.six.moves import reduce
from elasticsearch_dsl.query import Q

queries = []
if value1:
    queries.append(Q(
        'match',
        my_field1=value1,
    ))
if value2:
    queries.append(Q(
        'match',
        my_field2=value2,
    ))
search = search.query(
    reduce(operator.iand, queries)
)

14. Pagination

Django QuerySet:

from django.core.paginator import (
    Paginator, Page, EmptyPage, PageNotAnInteger
)

paginator = Paginator(queryset, paginate_by)
page_number = request.GET.get('page')
try:
    page = paginator.page(page_number)
except PageNotAnInteger:
    page = paginator.page(1)
except EmptyPage:
    page = paginator.page(paginator.num_pages)

Elasticsearch query:

from django.core.paginator import (
    Paginator, Page, EmptyPage, PageNotAnInteger
)
from django.utils.functional import LazyObject

class SearchResults(LazyObject):
    def __init__(self, search_object):
        self._wrapped = search_object

    def __len__(self):
        return self._wrapped.count()

    def __getitem__(self, index):
        search_results = self._wrapped[index]
        if isinstance(index, slice):
            search_results = list(search_results)
        return search_results

search_results = SearchResults(search)

paginator = Paginator(search_results, paginate_by)
page_number = request.GET.get('page')
try:
    page = paginator.page(page_number)
except PageNotAnInteger:
    page = paginator.page(1)
except EmptyPage:
    page = paginator.page(paginator.num_pages)

ElasticSearch doesn't work with Django's pagination by default. Therefore, we have to wrap the search query with lazy SearchResults class to provide the necessary functionality.

Example

I built an example with books written about Django. You can download it from Github and test it.

Takeaways

  • Filtering with Elasticsearch is much faster than with SQL databases.
  • But it comes at the cost of additional deployment and support time.
  • If you have multiple websites using Elasticsearch on the same server, configure a new cluster and node for each of those websites.
  • Django ORM can be in a way mapped to Elasticsearch DSL.
  • I summarized the comparison of Django ORM and Elasticsearch DSL, mentioned in this article, into a cheat sheet. You can get it for a symbolic fee. Print it on a single sheet of paper and use it as a reference for your developments.

Buy Django ORM vs. Elasticsearch DSL Cheat Sheet


Cover photo by Karl Fredrickson.

2018-05-27

QuerySet Filters on Many-to-many Relations

Django ORM (Object-relational mapping) makes querying the database so intuitive, that at some point you might forget that SQL is being used in the background.

This year at the DjangoCon Europe Katie McLaughlin was giving a talk and mentioned one thing that affects the SQL query generated by Django ORM, depending on how you call the QuerySet or manager methods. This particularity is especially relevant when you are creating your QuerySets dynamically. Here it is. When you have a many-to-many relationship, and you try to filter objects by the fields of the related model, every new filter() method of a QuerySet creates a new INNER JOIN clause. I won't discuss whether that's a Django bug or a feature, but these are my observations about it.

The Books and Authors Example

Let's create an app with books and authors, where each book can be written by multiple authors.

# -*- coding: UTF-8 -*-
from __future__ import unicode_literals

from django.db import models
from django.utils.translation import ugettext_lazy as _
from django.utils.encoding import python_2_unicode_compatible


@python_2_unicode_compatible
class Author(models.Model):
    first_name = models.CharField(_("First name"), max_length=200)
    last_name = models.CharField(_("Last name"), max_length=200)
    author_name = models.CharField(_("Author name"), max_length=200)

    class Meta:
        verbose_name = _("Author")
        verbose_name_plural = _("Authors")
        ordering = ("author_name",)

    def __str__(self):
        return self.author_name


@python_2_unicode_compatible
class Book(models.Model):
    title = models.CharField(_("Title"), max_length=200)
    authors = models.ManyToManyField(Author, verbose_name=_("Authors"))
    publishing_date = models.DateField(_("Publishing date"), blank=True, null=True)

    class Meta:
        verbose_name = _("Book")
        verbose_name_plural = _("Books")
        ordering = ("title",)

    def __str__(self):
        return self.title

The similar app with sample data can be found in this repository.

Inefficient Filter

With the above models, you could define the following QuerySet to select books which author is me, Aidas Bendoraitis.

queryset = Book.objects.filter(
    authors__first_name='Aidas',
).filter(
    authors__last_name='Bendoraitis',
)

We can check what SQL query it would generate with str(queryset.query) (or queryset.query.__str__()).

The output would be something like this:

SELECT `libraryapp_book`.`id`, `libraryapp_book`.`title`, `libraryapp_book`.`publishing_date`
FROM `libraryapp_book`
INNER JOIN `libraryapp_book_authors` ON ( `libraryapp_book`.`id` = `libraryapp_book_authors`.`book_id` )
INNER JOIN `libraryapp_author` ON ( `libraryapp_book_authors`.`author_id` = `libraryapp_author`.`id` )
INNER JOIN `libraryapp_book_authors` T4 ON ( `libraryapp_book`.`id` = T4.`book_id` )
INNER JOIN `libraryapp_author` T5 ON ( T4.`author_id` = T5.`id` )
WHERE (`libraryapp_author`.`first_name` = 'Aidas' AND T5.`last_name` = 'Bendoraitis')
ORDER BY `libraryapp_book`.`title` ASC;

Did you notice, that the database table libraryapp_author was attached through the libraryapp_book_authors table to the libraryapp_book table TWICE where just ONCE would be enough?

Efficient Filter

On the other hand, if you are defining query expressions in the same filter() method like this:

queryset = Book.objects.filter(
    authors__first_name='Aidas',
    authors__last_name='Bendoraitis',
)

The generated SQL query will be much shorter and (theoretically) would perform faster:

SELECT `libraryapp_book`.`id`, `libraryapp_book`.`title`, `libraryapp_book`.`publishing_date`
FROM `libraryapp_book`
INNER JOIN `libraryapp_book_authors` ON ( `libraryapp_book`.`id` = `libraryapp_book_authors`.`book_id` )
INNER JOIN `libraryapp_author` ON ( `libraryapp_book_authors`.`author_id` = `libraryapp_author`.`id` )
WHERE (`libraryapp_author`.`first_name` = 'Aidas' AND `libraryapp_author`.`last_name` = 'Bendoraitis')
ORDER BY `libraryapp_book`.`title` ASC;

The same SQL query can be achieved using the Q() objects:

queryset = Book.objects.filter(
    models.Q(authors__first_name='Aidas') &
    models.Q(authors__last_name='Bendoraitis')
)

The Q() objects add a lot of flexibility to filters allowing to OR, AND, and negate query expressions.

Dynamic Filtering

So to have faster performance, when creating QuerySets dynamically, DON'T use filter() multiple times:

queryset = Book.objects.all()
if first_name:
    queryset = queryset.filter(
        authors__first_name=first_name,
    )
if last_name:
    queryset = queryset.filter(
        authors__last_name=last_name,
    )

DO this instead:

filters = models.Q()
if first_name:
    filters &= models.Q(
        authors__first_name=first_name,
    )
if last_name:
    filters &= models.Q(
        authors__last_name=last_name,
    )
queryset = Book.objects.filter(filters)

Here the empty Q() doesn't have any impact for the generated SQL query, so you don't need the complexity of creating a list of filters and then joining all of them with the bitwise AND operator, like this:

import operator
from django.utils.six.moves import reduce

filters = []
if first_name:
    filters.append(models.Q(
        authors__first_name=first_name,
    ))
if last_name:
    filters.append(models.Q(
        authors__last_name=last_name,
    ))
queryset = Book.objects.filter(reduce(operator.iand, filters))

Profiling

In DEBUG mode, you can check how long the previously executed SQL queries took by checking django.db.connection.queries:

>>> from django.db import connection
>>> connection.queries
[{'sql': 'SELECT …', 'time': '0.001'}, {'sql': 'SELECT …', 'time': '0.004'}]

The Takeaways

  • When querying many-to-many relationships, avoid using multiple filter() methods, make use of Q() objects instead.
  • You can check the SQL query of a QuerySet with str(queryset.query).
  • Check the performance of recently executed SQL queries with django.db.connection.queries.
  • With small datasets, the performance difference is not so obvious. For your specific cases you should do the benchmarks yourself.

Cover photo by Tobias Fischer.