fddp: Facebook Downloaded Data Processor

Facebook allows you to download "a copy of what you've shared on Facebook". This includes a lot of Facebook data, including Messenger (chat) messages (what we are interested in)

fddp is an experiment about doing some data mining (or other things) on Facebook messages from your own account.

Getting your Facebook archive

You can acquire your archive on Facebook settings page, it is at the end under "Download a copy of your Facebook data", then follow the steps to save it.

Once you have this zip file, open it and place the messages.htm inside the personal folder

Samples

I have crafted samples for you to use to test. These are identical to how Facebook's archives distribute their messages.htm.

This is an easier way to get started then download your FB archive but is way less populated (only a few messages).

Setup

fddp uses godep for dependency management. It is not required to run.

Install

• Setup your $GOPATH

go get github.com/Cretezy/fddp && cd $GOPATH/src/github.com/Cretezy/fddp
go build

• Enjoy! Check if everything works with ./fddp. You must be in the fddp directory to run commands (server).

Usage

Web UI

• To start the Web UI, run ./fddp server
• Visit http://localhost:3000/

You can switch the port using the PORT environment variable.

Commandline

Convert

Converts a HTML message file (ex: Facebook's messages.htm) to JSON.

You must convert your HTML message file to JSON before doing anything with it. It will also clean it.

./fddp convert personal/messages.htm personal/messages.json

This will turn personal/messages.htm to JSON format and save it (under personal/messages.json).

You can use -i (or --indent) to indent (pretty print). This is not recommended on big data set as it adds useless storage bulk (see Example File Size for increase).

Count

Counts threads/messages/words in a data set.

You must input a JSON file (use convert command first). You may use many flags at the same time.

./fddp count [flags] input.json

Compare

Shows the difference between 2 data sets (in count, not data).

You must input 2 JSON file (use convert command first).

./fddp compare samples/sample.json samples/sample-indent.json

List

List tops people you have messaged.

You must input a JSON file (use convert command first).

./fddp list samples/sample.json

Default shows top 50 but using -c (or --count) followed by a custom number you may change the number of threads displayed.

Notes

Purpose

The purpose of this project is to:

• Gain experience (Go, etc)
• Have a better interface to read past Facebook messages
• Analyse Facebook messages as "big data"

Example File Size

Sample size of this is from my Facebook, ~350k messages. Running on an average-high end desktop CPU (4770K) and SSD.

Note: I think that reading the actual file (~36MB) from disk is the main bottleneck for these stats. I estimate that reading from file (and parsing the JSON) actually takes around 800ms or so, except the convert which takes a much longer time (parsing HTML/messages is a lot slower) and compare, which needs to open 2 files.