Build a relationship graph of a target user Facebook friendships.
Partially reconstruct hidden friendlists by using the "mutual friends" functionality. Requires knowing at least another account having at least one mutual friend with the target.
The auxiliary script fint.py can retrieve users who have interacted with a given target profile and can be used as an input to ffff.py
Here is a detailed blog post about the tools, including a practical example.
Scraping without written permission is against Facebook Terms of Services: therefore you should not use any scraping tool without proper written authorization. All the information provided in this and related articles are for educational purposes only. The authors or are in no way responsible for any misuse of the information or the code provided.
Facebook offers a function to list some friends two accounts have in common, called "mutual friends". It works only if at least one of the accounts' lists is visible to the observer. In order to partially reconstruct a hidden friend list, this function can be applied recursively over all mutual friends found at each step.
It is also possible to keep track of the relationships between users along the way, by connecting them in a graph, assigning weights based on how many times a relationship has been observed during the procedure. In this way it is possible to identify communities within the network.
The mutual friends functionality only works if at least one of the two users compared have their friendlists public. Furthermore, Facebook limits the number of mutual friends that are shown. It seems that using different accounts sometimes lead to different results. It might be worth giving a try - maybe using accounts from different locations.
The script requires Python 3 and the libraries selenium, argparse and networkx.
You can download the repository as a zip file or clone it:
git clone https://github.com/sowdust/ffff.git
Install the requirements:
pip3 install -r requirements.txt
As of now the script supports only Firefox webdriver. The geckodriver executable must be downloaded and stored locally. I suggest you store it inside the ffff folder. Support for more webdrivers can be easily added if necessary - just ask!
To run the script ffff the following information is required:
- Valid Facebook credentials (user/password)
- Target account Facebook id
- One or more "pivot" account ids (if target friend list is hidden) - you can use fint.py for this
- Path to the Geckodriver file
The script works with Facebook numeric account ids. There are several ways to find out an account numeric id, including many online services such as findmyfbid.com or graph.tips.
A set of valid credentials (username and password) of a Facebook account to be used for collecting the results is required.
Warning: Accounts used for scraping are not allowed without permission, and can therefore be locked by Facebook anti-scraping mechanisms. If you have a developer account, you could try using a test account.
If the target account friend list is hidden, it is also necessary to provide the script at least one "pivot" account, ie.: one account that has their friend list public and at least one mutual friend with the target.
A common way to find a suitable pivot account is to look at users who reacted to content published by the target.
Obviously the more pivots are provided, the better.
If there are many pivots, they can be stored in a file, one per line.
In order to check if mutual friends between two users are visible, visit the url https://www.facebook.com/browse/mutual_friends/?uid=USER1&node=USER2, where USER1 and USER2 are the two users' numeric ids.
One way to find pivot accounts is to use the auxiliary script fint.py. This script will provide a list of users who have interacted with the target.
Run fint.py -h to have a list of command line options. This script will produce a txt file with a list of potential pivots; this file can later be fed to ffff.py by using the -P (--pivots-file) option.
If the friend list is public, no pivots are necessary: just provide the target account id also as a pivot.
It is also necessary to know the local path in which you have stored the geckodriver executable.
Some options (the facebook credentials and the the driver executables) can be hardcoded inside the script source code:
FB_USR = '[email protected]'
FB_PWD = 'password'
GECKO_PATH = 'C:\geckodriver.exe'
Otherwise, they can be provided as a command line option.
Pivot accounts can be provided either as a list of integers separated by a space via command line (with the option --pivots), or one per row in a file (--pivots-file).
A list of accounts to ignore as pivots can be provided as well (--ignore-file).
The session can be interrupted (Ctrl+C) and resumed at a later time. This is also possible in case an exception occurs during the program run (e.g.: a request timeout).
usage: ffff.py [-h] [-fu USERNAME] [-fp PASSWORD] [-t TARGET]
[-p PIVOT [PIVOT ...]] [-o OUTPUTFILE] [-g GRAPHFILE]
[-P PIVOTSFILE] [-I IGNOREFILE] [-r SESSION] [-q] [-w]
[-d EXECUTABLE]
Recursively build friends list of Facebook users by exploiting the "mutual
friends" utility.
optional arguments:
-h, --help show this help message and exit
-fu USERNAME, --user USERNAME
Username of the Facebook account that will be used for
scraping
-fp PASSWORD, --password PASSWORD
Username of the Facebook account that will be used for
scraping
-t TARGET, --target TARGET
Numeric id of the target Facebook account
-p PIVOT [PIVOT ...], --pivots PIVOT [PIVOT ...]
Numeric id(s) of the Facebook accounts known to have
mutual friends with the target. Their friends list
must be public. Can be a single value or a list of
values separated by a space.
-o OUTPUTFILE, --output OUTPUTFILE
Specify where to store list of friends found in CSV
format
-g GRAPHFILE, --graph-output GRAPHFILE
Specify where to store graph in GEXF (Graph Exchange
XML Format) format
-P PIVOTSFILE, --pivots-file PIVOTSFILE
Load a list of Facebook ids to use as pivot accounts
from file PIVOTSFILE. Numeric ids must be one per
line. This option can be used together or in place of
--pivots.
-I IGNOREFILE, --ignore-file IGNOREFILE
Load a list of Facebook ids NOT to use as pivot
accounts from file IGNOREFILE. Numeric ids must be one
per line.
-r SESSION, --resume SESSION
Resume a previous session from file SESSION
-q, --headless Run browser in headless mode. No browser window will
be shown.
-w, --store-weights Assign weights based on how many times a relationship
is observed.
-d EXECUTABLE, --driver-path EXECUTABLE
Path to geckodriver executable
The following example usage tries to build a subset of the friends list of target account id "00000000", pivoting on account "00000001", and finally storing the results in CSV file "results.csv" and the Gephi graph file in "graph.gexf".
The account used for scraping is "[email protected]", protected by the password "Passw0rd"
Adding the -q option makes the webdriver run in headless mode: no browser window should be shown.
In the example, the Windows geckodriver executable is stored in the current directory. Geckodriver executables can be downloaded from Mozilla repository. If you are using Linux or Mac and have downloaded the geckodriver file in the current directory, use --driver-path ./geckodriver.
python3 ffff.py -fu [email protected] -fp Passw0rd -t 00000000 -p 00000001 -o results.csv -g graph.gexf -q --driver-path geckodriver.exe
The session can be stopped (Ctrl+C) and resumed at a later time:
^CSession stored to file session-00000000-20190221232552. Use "--resume session-00000000-20190221232552" to resume from here
Use fint.py to find potential pivot accounts for target user 111111111, using the most recent 10 stories and 5 photos published, using a maximum of 100 comments and 1000 reactions:
python fint.py -fu [email protected] -fp fbpassword -d geckodriver.exe -t 111111111 -ls 10 -lp 5 -lc 100 -lr 1000
Build the community graph of target user 111111111 with public friend list, showing the browser window (without using option -q). Geckodriver file is stored in C::
python3 ffff.py -fu [email protected] -fp Passw0rd -t 111111111 -p 111111111 --driver-path C:\geckodriver.exe
Provide a file with a list of pivot accounts as well as a a file with a list of accounts not to be used as pivots if found among the target's friend. Files pivots.txt and ignore.txt must contain one Facebook id per line. Operating system is linux and the geckodriver executable is in the user's Desktop:
python3 ffff.py -fu [email protected] -fp Passw0rd -t 12345678 -P pivots.txt -I ignore.txt -d /home/user/Desktop/geckodriver
Resume an old session (either terminated or interrupted), adding three new pivots to be processed:
python3 ffff.py -fu [email protected] -fp Passw0rd --resume session-00000000-20190418122352 -p 111111111 222222222 333333333
The scripts ffff.py produces three output files:
- one .csv file containing the list of friends found, in the form
account id, account name, account url - one .gexf file containing the annotated graph that can be opened using Gephi
- one session file that can be used to resume the work (e.g.: in case more pivots are found at a later time)
The script fint.py produces two output files:
- one .txt file containing a list of account ids that can be used as a list of pivots for ffff
- one .csv file (optional) containing a list of potential pivots in the form
id,name,urlIt also prints out statistics on who are the users who have interacted the most.
This code is ffffree