[DEPRECATED] pull_facebook_data_for_good

---

⛔ [DEPRECATED] ⛔

This library has been deprecated by changes to the portal for downloading data from Facebook Data for Good and enhanced security mechanisms in the new download portal. The recommended method for downloading data is from the newly available "Partner Portal". This library is compatible with the previous "Geoinsights Portal" and will no longer be maintained.

---

Imitate an API for downloading data from Facebook Data for Good.

This library uses selenium webdriver to imitate the behaviour of an API for downloading the full timeseries of a data collection.

This library is developed and tested in Python 3.8.

Disclaimer: This download routine will only work for those with access to the Facebook Geoinsights platform, and will only function for datasets to which the user has been granted access. This tool is not developed by or associated with Facebook, it is simply a utility to automate downloading data from the Geoinsights platform.

Installation

From a clone:

To develop this project locally, clone it onto your machine:

git clone https://github.com/hamishgibbs/pull_facebook_data_for_good.git

Enter the project directory:

cd pull_facebook_data_for_good

Install the package with:

pip install .

From GitHub:

To install the package directly from GitHub run:

pip install git+https://github.com/hamishgibbs/pull_facebook_data_for_good.git

Usage

Currently functional for TileMovement and TilePopulation datasets only.

Use the CLI from the directory where you would like data to be downloaded:

cd path/to/downloaded/data

The CLI follows the format:

pull_fb --dataset_name --area

For example, to pull the TileMovement dataset for Britain:

pull_fb --dataset_name TileMovement --area Britain

or:

pull_fb --d TileMovement --a Britain

The country name must exactly match the name stored in the .config file. For multi-word names, each word will be separated by '_'. ie. New_Zealand

For a full API reference, please see the Reference.

Please Note:

If the .config file is missing variables for a given dataset, please alter the .config file and open a pull request to share with others.

Chrome Web Driver

To download data, this library relies on selenium and ChromeDriver.

This requires a chromedriver executable which can be downloaded here. Make sure that your Chrome version is the same as your chromedriver version.

pull_facebook_data_for_good assumes that the chromedriver executable is located at Applications/chromedriver. To supply a different path, use the argument --driver_path or -driver from the command line.

Credentials

Credentials must be input manually on each download.

Credentials are not stored on your computer and are passed directly to the Facebook login page by the web driver.

Reference

--dataset_name (-d)

Name of the data collection to be downloaded (i.e. "TileMovement", "TilePopulation").

--area (-a)

Name of the area of the data collection (i.e. "Britain").

--outdir (-o)

Directory where datasets will be downloaded (default: current directory - os.getcwd()).

--end_date (-e)

Dataset end date (default: datetime.datetime.now()).

--frequency (-f)

Dataset update frequency in hours (default: 8).

--driver_path (-driver)

Path to ChromeDriver (see download options here).

--config_path (-config)

Path to .config file (default: config in remote repo. Also accepts local paths).

--username (-user)

Facebook username (or registered email address). To run without prompting for login credentials.

--password (-pass)

Facebook password. To run without prompting for login credentials.

--driver_flags (-driver_flags)

Flags passed to chromedriver (allows multiple flags). Flags are passed with selenium.webdriver.ChromeOptions.add_argument.

--driver_prefs (-driver_prefs)

Preferences passed to chromedriver as dict. Preferences are passed with selenium.webdriver.ChromeOptions.add_experimental_option.

Tests

This project is tested with tox.

To run unit tests:

tox

Contributions

Issues:

To request a feature or report an issue with this tool, please open an issue.

Adding a dataset:

Dataset attributes are stored in the .config file.

Each time you use the library, pull_facebook_data_for_good will look for dataset configuration variables here unless you specify a path to another .config file.

To add the ability to download another dataset, alter the .config file with two pieces of information:

1. The dataset id, embedded in the url of the Geoinsights download page. For example, the dataset ID for the collection stored at https://www.facebook.com/geoinsights-portal/downloads/?id=243071640406689 is 243071640406689.

2. The date origin of the dataset, the earliest date of data publication, in the format: year_month_day(_hour). i.e. 2020_01_01_00.

Please open a pull request to share the config variables for a new dataset with everyone.

Other contributions:

Other contributions are welcome.

Please look for open issues with the Help Wanted tag.