track changes to the news, where news is anything with an RSS feed
diffengine is a utility for watching RSS feeds to see when story content changes. When new content is found a snapshot is saved at the Internet Archive, and a diff is generated for sending to social media. The hope is that it can help draw attention to the way news is being shaped on the web. It also creates a database of changes over time that can be useful for research purposes.
diffengine draws heavily on the inspiration of NYTDiff and NewsDiffs which almost did what we wanted. NYTdiff is able to create presentable diff images and tweet them, but was designed to work specifically with the NYTimes API. NewsDiffs provides a comprehensive framework for watching changes on multiple sites (Washington Post, New York Times, CNN, BBC, etc) but you need to be a programmer to add a parser module for a website that you want to monitor. It is also a full-on website which involves some commitment to install and run.
With the help of feedparser, diffengine takes a different approach by working with any site that publishes an RSS feed of changes. This covers many news organizations, but also personal blogs and organizational websites that put out regular updates. And with the readability module, diffengine is able to automatically extract the primary content of pages, without requiring special parsing to remove boilerplate material. And like NYTDiff, instead of creating another website for people to watch, diffengine pushes updates out to social media (via Twitter or email) where people are already, while also building a local database of diffs that can be used for research purposes.
pip3 install diffengine
In order to run diffengine you need to pick a directory location where you can store the diffengine configuration, database and diffs. For example I have a directory in my home directory, but you can use whatever location you want, you just need to be able to write to it.
The first time you run diffengine it will prompt you to enter an RSS or Atom feed URL to monitor. You will the be asked to provide the credentials to publish the diffs in social media.
% diffengine /home/ed/.diffengineWhat RSS/Atom feed would you like to monitor? https://inkdroid.org/feed.xml
Would you like to set up tweeting edits? [Y/n] Y
Go to https://apps.twitter.com and create an application.
What is the consumer key?
What is the consumer secret?
Log in to https://twitter.com as the user you want to tweet as and hit enter.
Visit https://api.twitter.com/oauth/authorize?oauth_token=NRW9BQAAAAAAyqBnAAXXYYlCL8g
What is your PIN: 8675309
Saved your configuration in /home/ed/.diffengine/config.yaml
Would you like to set up emailing edits with Sendgrid? [Y/n] y
Go to https://app.sendgrid.com/ and get an API key.
What is the API key?
What email address is sending the email?
Who are the recipients of the emails?
Fetching initial set of entries.
Done!
After that you just need to put diffengine in your crontab to have it run regularly, or you can run it manually at your own intervals if you want. Here's my crontab to run every 30 minutes to look for new content.
0,30 * * * * /usr/local/bin/diffengine /home/ed/.diffengine
You can examine your config file at any time and add/remove feeds as needed. It is the
config.yamlfile that is stored relative to the storage directory you chose, so in my case
/home/ed/.diffengine/config.yaml.
Logs can be found in
diffengine.login the storage directory, for example
/home/ed/.diffengine/diffengine.log.
Checkout Ryan Baumann's "diffengine" Twitter list for a list of known diffengine Twitter accounts that are out there.
By default the database is configured for Sqlite and the file
./diffengine.dbthrough the
dbconfig prop
db: sqlite:///diffengine.db
This value responds to the database URL connection string format.
For instance, you can co˚nnect to your postgresql database using something like this.
db: postgresql://postgres:[email protected]:5432/my_database
In case you store your database url connection into an environment var, like in Heroku. You can simply do as follows.
db: "${DATABASE_URL}"
If you are setting multiple accounts, and multiple feeds if may be helpful to setup a directory for each account. For example:
/home/nruest/.torontosun
/home/nruest/.torontostar
/home/nruest/.globemail
/home/nruest/.canadaland
/home/nruest/.cbc
Then you will configure a cron entry for each account:
0,15,30,45 * * * * /usr/bin/flock -xn /tmp/globemail.lock -c "/usr/local/bin/diffengine /home/nruest/.globemail" 0,15,30,45 * * * * /usr/bin/flock -xn /tmp/torontosun.lock -c "/usr/local/bin/diffengine /home/nruest/.torontosun" 0,15,30,45 * * * * /usr/bin/flock -xn /tmp/cbc.lock -c "/usr/local/bin/diffengine /home/nruest/.cbc" 0,15,30,45 * * * * /usr/bin/flock -xn /tmp/lapresse.lock -c "/usr/local/bin/diffengine /home/nruest/.lapresse" 0,15,30,45 * * * * /usr/bin/flock -xn /tmp/calgaryherald.lock -c "/usr/local/bin/diffengine /home/nruest/.calgaryherald"
If there are multiple feeds for an account, you can setup the
config.ymllike so:
- name: The Globe and Mail - Report on Business twitter: access_token: ACCESS_TOKEN access_token_secret: ACCESS_TOKEN_SECRET sendgrid: sender: FROM_ADDRESS recipients: TO_ADDRES1, TO_ADDRESS2 url: http://www.theglobeandmail.com/report-on-business/?service=rss - name: The Globe and Mail - Opinion twitter: access_token: ACCESS_TOKEN access_token_secret: ACCESS_TOKEN_SECRET sendgrid: sender: FROM_ADDRESS2 recipients: TO_ADDRES3, TO_ADDRESS4 url: http://www.theglobeandmail.com/opinion/?service=rss - name: The Globe and Mail - News twitter: access_token: ACCESS_TOKEN access_token_secret: ACCESS_TOKEN_SECRET url: http://www.theglobeandmail.com/news/?service=rss twitter: consumer_key: CONSUMER_KEY consumer_secret: CONSUMER_SECRET sendgrid: api_token: API_TOKEN
You can also keep an entry if matches with a regular expression pattern. This is useful for avoid the "subscribe now" pages. This is configured per feed like so:
- name: The Globe and Mail - Report on Business skip_pattern: "you have access to only \\d+ articles" twitter: access_token: ACCESS_TOKEN access_token_secret: ACCESS_TOKEN_SECRET url: http://www.theglobeandmail.com/report-on-business/?service=rss
In this example, if the page says contains the text "you have access to only 10 articles" will skip it. the same if says any number of articles as it's a regular expression. The
skip_patternperforms a
re.searchoperation and uses the flags for
case insensitiveand
multiline.
Look for the docs for more information about Regular Expressions and the search operation.
By default, the tweeted diff will include the article's title and the archive diff url, like this.
You change this by tweeting what's changed: the url, the title and/or the summary. For doing so, you need to specify all the following
langkeys:
lang: change_in: "Change in" the_url: "the URL" the_title: "the title" and: "and" the_summary: "the summary"
Only if all the keys are defined, the tweet will include what's changed on its content, followed by the
diff.url. Some examples:
And so on with all the possible combinations between url, title and summary
The configuration file has support for environment variables. This is useful if you want to keeping your credentials secure when deploying to Heroku, Vercel (former ZEIT Now), AWS, Azure, Google Cloud or any other similar services. The environment variables are defined on the app of the platform you use or directly in a dotenv file, which is the usual case when coding locally.
For instance, say you want to keep your Twitter credentials safe. You'd keep a reference to it in the
config.yamlthis way:
twitter: consumer_key: "${MY_CONSUMER_KEY_ENV_VAR}" consumer_secret: "${MY_CONSUMER_SECRET_ENV_VAR}"
Then you would define your environment variables
MY_CONSUMER_KEY_ENV_VARand
MY_CONSUMER_SECRET_ENV_VARin your
.envfile:
MY_CONSUMER_KEY_ENV_VAR="CONSUMER_KEY" MY_CONSUMER_SECRET_ENV_VAR="CONSUMER_SECRET"
Done! You can use diffengine as usual and keep your credentials safe.
You can use the following command for adding Twitter accounts to the config file.
$ diffengine --addLog in to https://twitter.com as the user you want to tweet as and hit enter. Visit https://api.twitter.com/oauth/authorize?oauth_token=QKGAqgAAAAABDsonAAABcbfQfFw in your browser and hit enter. What is your PIN: 1234567
These are your access token and secret. DO NOT SHARE THEM WITH ANYONE!
ACCESS_TOKEN xxxxxxxxxxx-yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy
ACCESS_TOKEN_SECRET zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz
Then you would use the
ACCESS_TOKENand the
ACCESS_TOKEN_SECRETinside the config like this
feeds: - name: My new feed url: http://www.mynewfeed.com/feed/ twitter: access_token: "${ACCESS_TOKEN}" access_token_secret: "${ACCESS_TOKEN_SECRET}"
Diffengine has support for
geckodriverand
chromedriver.
You can configure this in the
config.yaml. The keys are the following ones.
yaml webdriver: engine: executable_path: binary_location:
The
geckodriveris properly defined by default. In case you need to configure it, then:
webdriver: engine: "geckodriver" executable_path: null (this config has no use with geckodriver) binary_location: null (the same as above with this one)
If you want to use
chromedriverlocally, then you should leave the config this way:
webdriver: engine: "chromedriver" executable_path: null ("chromedriver" by default) binary_location: null ("" by default)
If you use Heroku, then you have to add the Heroku chromedriver buildpack. And then use the environment vars provided automatically by it.
webdriver: engine: "chromedriver" executable_path: "${CHROMEDRIVER_PATH}" binary_location: "${GOOGLE_CHROME_BIN}"
By default, the script will log everyhintg to
./diffengine.log. Anyway, you can disable the file logger and/or enable the console logger as well. You can modify the log filename, too.
If no present, the default values will be the following ones.
yaml log: diffengine.log logger: file: true console : false
Logging to the console could be useful to see what's happening if the app lives in services like Heroku.
Here's how to get started hacking on diffengine with pipenv:
% git clone https://github.com/docnow/diffengine % cd diffengine % pipenv install % pytest ============================= test session starts ============================== platform linux -- Python 3.5.2, pytest-3.0.5, py-1.4.32, pluggy-0.4.0 rootdir: /home/ed/Projects/diffengine, inifile: collected 5 itemstest_diffengine.py .....
=========================== 5 passed in 8.09 seconds ===========================
Last, you need to install the pre-commit hooks to be run before any commit
pre-commit install
This way, Black formatter will be executed every time.
We recommend you to to configure it in your own IDE here.