blog/static/pages/Deliver-customized-newsletters-from-RSS-feeds-with-Platypush.md

15 KiB
Raw Blame History

Ive always been a supporter of well-curated newsletters. They give me an opportunity to get a good overview of what happened in the fields I follow within a span of a day, a week or a month. However, not all the newsletters fit this category. Some dont think three times before selling email addresses to 3rd-parties — and within the blink of an eye your mailbox can easily get flooded with messages that you didnt request. Others may sign up your address for other services or newsletters as well, and often they dont often much granularity to configure which communications you want to receive. Even in the best-case scenario, the most privacy-savvy user may still think twice before signing up for a newsletter — youre giving your personal email address to someone else you dont necessarily trust, implying “yes, this is my address and Im interested in this subject”. Additionally, most of the newsletters spice up their URLs with tracking parameters, so they can easily measure user engagement — something you may not necessarily be happy with. Moreover, the customization junkie may also have a valid use case for a more finely tuned selection of content in his newsletter — you may want to group some sources together into the same daily/weekly email, or you may be interested only in some particular subset of the subjects covered by a newsletter, filtering out those that arent relevant, or customize the style of the digest that gets delivered. Finally, a fully automated way to deliver newsletters through 5 lines of code and the tuning of a couple of parameters is the nirvana for many companies of every size out there.

Feed up the newsletter

Those who read my articles in the past may know that Im an avid consumer of RSS feeds. Despite being a 21-year-old technology, they do their job very well when it comes to deliver the information that matters without all the noise and trackers, and they provide a very high level of integration being simple XML documents. However, in spite of all the effort I put to be up-to-date with all my sources, a lot of potentially interesting content inevitably slips through — and thats where newsletters step in, as they filter and group together all the content that was generated in a given time frame and periodically deliver it to your inbox.

My ideal solution would be something that combines the best aspects of both the worlds: the flexibility of an RSS subscription, combined with a flexible way of filtering and aggregating content and sources, and get the full package delivered at my door in whichever format I like (HTML, PDF, MOBI…). In this article Im going to show how to achieve this goal with a few tools:

  • One or more sources that you want to track and that support RSS feeds (in this example Ill use the MIT Technology Review RSS feed, but the procedure works for any RSS feed).

  • An email address.

  • Platypush to do the heavy-lifting job — monitor the RSS sources at custom intervals, trigger events when a source has some new content, create a digest out of the new content, and deliver the full package to a list of email addresses.

Lets cover these points step by step.

Installing and configuring Platypush

Well be using the http.poll backend configured with one or more RssUpdate objects to poll our RSS sources at regular intervals and create the digests, and either the mail.smtp plugin or the google.mail plugin to send the digests to our email.

You can install Platypush on any device where you want to run your logic — a RaspberryPi, an old laptop, a cloud node, and so on. We will install the base package with the rss module. Optionally, you can install it with the pdf module as well (if you want to export your digests also to PDF) or the google module (if you want to send the newsletter from a GMail address instead of an SMTP server).

The first option is to install the latest stable version through pip:

[sudo] pip install 'platypush[rss,pdf,google]'

The other option is to install the latest git version:

git clone https://git.platypush.tech/platypush/platypush
cd platypush
[sudo] pip install '.[rss,pdf,google]'

Monitoring your RSS feeds

Once the software is installed, create the configuration file ~/.config/platypush/config.yaml if it doesn't exist already and add the configuration for the RSS monitor:

# Generic HTTP endpoint monitor
backend.http.poll:
    requests:
        # Add a new RSS feed to the pool
        - type: platypush.backend.http.request.rss.RssUpdates
          # URL to the RSS feed
          url: https://www.technologyreview.com/feed/
          # Title of the feed (shown in the head of the digest)
          title: MIT Technology Review
          # How often we should monitor this source (24*60*60 secs = once a day)
          poll_seconds: 86400
          # Format of the digest (HTML or PDF)
          digest_format: html

You can also add more sources to the http.poll requests object, each with its own configuration. Also, you can customize the style of your digest by passing some valid CSS to these configuration attributes:

# Style of the body element
body_style: 'font-size: 20px; font-family: "Merriweather", Georgia, "Times New Roman", Times, serif'

# Style of the main title
title_style: 'margin-top: 30px'

# Style of the subtitle
subtitle_style: 'margin-top: 10px; page-break-after: always'

# Style of the article titles
article_title_style: 'font-size: 1.6em; margin-top: 1em; padding-top: 1em; border-top: 1px solid #999'

# Style of the article link
article_link_style: 'color: #555; text-decoration: none; border-bottom: 1px dotted font-size: 0.8em'

# Style of the article content
article_content_style: 'font-size: 0.8em'

The digest_format attribute determines the output format of your digest - you may want to choose html if you want to deliver a summary of the articles in a newsletter, or pdf if you want instead to deliver the full content of each item as an attachment to an email address. Bonus point: since you can send PDFs to a Kindle if you configured an email address, this mechanism allows you to deliver the full digest of your RSS feeds to your Kindle's email address.

The RssUpdates object also provides native integration with the Mercury Parser API to automatically scrape the content of a web page - I covered some of these concepts in my previous article on how to parse RSS feeds and send the PDF digest to your e-reader. The same mechanism works well for newsletters too. If you want to parse the content of the newsletter as well, all you have to do is configure the http.webpage Platypush plugin. Since the Mercury API doesn't provide a Python binding, this requires a couple of JavaScript dependencies:

# Install Node and NPM, e.g. on Debian:
apt-get install nodejs npm

# Install the Mercury Parser API
npm install [-g] @postlight/mercury-parser

# Make sure that the Platypush PDF module dependencies
# are installed if you plan HTML->PDF conversion
pip install 'platypush[pdf]'

Then, if you want to parse the full content of the items and generate a PDF digest out of them, change your http.poll configuration to something like this:

backend.http.poll:
    requests:
        - type: platypush.backend.http.request.rss.RssUpdates
          url: https://www.technologyreview.com/feed/
          title: MIT Technology Review
          poll_seconds: 86400
          # PDF digest format
          digest_format: pdf
          # Extract the full content of the items
          extract_content: True

WARNING: Extracting the full content of the articles in an RSS feed has two limitations — a practical one and a legal one:

  • Some websites may require user login before displaying the full content of an article. Some websites perform such checks client-side — and the parser API can usually circumvent them, especially if the full content of an article is actually just hidden behind a client-side paywall. Some websites, however, implement their user checks server-side too before sending the content to the client — and in those cases the parser API may return only a part of the content or no content at all.

  • Always keep in mind that parsing the full content of an article behind a paywall may represent a violation of intellectual property under some jurisdictions, so make sure to do it only for content that is either free or that you have to permission to scrape.

Configuring the mail delivery

When new content is published on a subscribed RSS feed Platypush will generate a NewFeedEvent and it should create a copy of the digest under ~/.local/share/platypush/feeds/cache/{date:time}_{feed-title}.[html|pdf]. The NewFeedEvent in particular is the link you need to create your custom logic that sends an email to a list of addresses when new content is available.

First, configure the Platypush mail plugin you prefer. When it comes to sending emails you primarily have two options:

  • The mail.smtp plugin — if you want to send emails directly through an SMTP server. Platypush configuration:
mail.smtp:
    username: you@gmail.com
    password: your-pass
    server: smtp.gmail.com
    port: 465
    ssl: True
  • The google.mail plugin — if you want to use the native GMail API to send emails. If that is the case then first make sure that you have the dependencies for the Platypush Google module installed:
[sudo] pip install 'platypush[google]'

In this case youll also have to create a project on the Google Developers console and download the OAuth credentials:

  • Click on “Credentials” from the context menu > OAuth Client ID.

  • Once generated, you can see your new credentials in the “OAuth 2.0 client IDs” section. Click on the “Download” icon to save them to a JSON file.

  • Copy the file to your Platypush device/server under e.g. ~/.credentials/google/client_secret.json.

  • Run the following command on the device to authorize the application:

python -m platypush.plugins.google.credentials \
    "https://www.googleapis.com/auth/gmail.modify" \
    ~/.credentials/google/client_secret.json \
    --noauth_local_webserver

At this point the GMail delivery is ready to be used by your Platypush automation.

Connecting the dots

Now that both the RSS parsing logic and the mail integration are in place, we can glue them together through the NewFeedEvent event. The new advised way to configure events in Platypush is through native Python scripts - the custom YAML-based syntax for events and procedure was becoming too cumbersome to maintain and write (although its still supported), and I feel like going back to a clean and simple Python API may be a better option.

Create and initialize the Platypush scripts directory, if it doesnt exist already:

mkdir -p ~/.config/platypush/scripts
cd ~/.config/platypush/scripts

# Make sure that the scripts module is initialized
touch __init__.py

Then, create a new hook on NewFeedEvent:

$EDITOR rss_news.py
import os
from typing import List

from platypush.event.hook import hook
from platypush.message.event.http.rss import NewFeedEvent
from platypush.utils import run

# Path to your mailing list - a text file with one address per line
maillist = os.path.expanduser('~/.mail.list')

def get_addresses() -> List[str]:
    with open(maillist, 'r') as f:
        return [addr.strip() for addr in f.readlines()
                if addr.strip() and not addr.strip().startswith('#')]


# This hook matches:
# - event_type=NewFeedEvent
# - digest_format='html'
# - source_title='MIT Technology Review'
@hook(NewFeedEvent, digest_format='html', source_title='MIT Technology Review')
def send_mit_rss_feed_digest(event: NewFeedEvent, **_):
    # The digest output file is stored in event.args['digest_filename']
    with open(event.args['digest_filename'], 'r') as f:
        run(action='mail.smtp.send',
            from_='you@yourdomain.com',
            to=get_addresses(),
            subject=f'{event.args.get("source_title")} feed digest',
            body=f.read(),
            body_type='html')

# Or, if you opted for the native GMail plugin you may want to go for:

@hook(NewFeedEvent, digest_format='html', source_title='MIT Technology Review')
def send_mit_rss_feed_digest(event: NewFeedEvent, **_):
    # The digest output file is stored in event.args['digest_filename']
    with open(event.args['digest_filename'], 'r') as f:
        run(action='google.mail.compose',
            sender='you@gmail.com',
            to=get_addresses(),
            subject=f'{event.args.get("source_title")} feed digest',
            body=f.read())

# If instead you want to send the digest in PDF format as an attachment:

@hook(NewFeedEvent, digest_format='html', source_title='MIT Technology Review')
def send_mit_rss_feed_digest(event: NewFeedEvent, **_):
    # mail.smtp plugin case
    run(action='mail.smtp.send',
        from_='you@yourdomain.com',
        to=get_addresses(),
        subject=f'{event.args.get("source_title")} feed digest',
        body='',
        attachments=[event.args['digest_filename']])

    # google.mail case
    run(action='google.mail.compose',
        sender='you@gmail.com',
        to=get_addresses(),
        subject=f'{event.args.get("source_title")} feed digest',
        body='',
        files=[event.args['digest_filename']])

Finally, create your ~/.mail.list file with one destination email address per line and start platypush either from the command line or as a service. You should receive your email with the first batch of articles shortly after startup, and you'll receive more items if a new batch is available after the poll_seconds configured period.