https://github.com/atajti/fbpack

R package for user friendly Facebook scraping
https://github.com/atajti/fbpack

Last synced: 3 days ago
JSON representation

R package for user friendly Facebook scraping

• Host: GitHub
• URL: https://github.com/atajti/fbpack
• Owner: atajti
• Created: 2015-05-14T20:14:42.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2017-07-06T15:12:13.000Z (over 7 years ago)
• Last Synced: 2024-08-02T06:03:42.193Z (3 months ago)
• Language: R
• Homepage:
• Size: 27.3 KB
• Stars: 10
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# fbpack

R package for user frienly Facebook scraping

## Purpose

Facebook provides programmatically collectable data through its API,

and R has a package (RFacebook) which can connect to it. Why did I start this project?

First, when I tried it in november, 2014, the RFacebook package did not worked as I expected. One of the reasons

was that since it then-last update, Facebook modified its API rules: one could not collect information about

friendships which are available for the user through the browser when logged in. This gave me the motivation,

but I lacked tools. At the Budapest R meetup in january, I saw the demonstration of the RSelenium package.

The RSelenium packge connects R to the Selenium Web Driver, which allows the user to navigate and use a browser

programmatically as a real person. This became an opportunity to simulate a user browsing Facebook,

and gather the data displayed in the browser.

## Installing

Installing the package can be done with Hadley Wickham's devtools package:

```{r "package installation"}

if(!require(devtools)){

  install.packages("devtools")

}

devtools::install_github("atajti/fbpack")

```

Systems running Windows in a non-English locale may run into problems with Rtools if its path contains letters with accents. If this is the case, one can download the files then run `install.packages("path/to/files", repos=NULL, type="source")`.

Also, the Selenium Web Driver should be on the computer. The RSelenium package takes care of this if one runs the

`RSeleium::checkForServer()` function.

## Usage

The Selenium server have to be started as all of the functions use it.

```{r "package usage"}

library(fbpack)

startServer()

```

After the `startServer()` function called, one should wait a few seconds for Selenium to start. After that,

`fbpack`'s two function(`search_fb()`, `get_user_friends()`) should run properly. For further information,

see `?search_fb` and `?ger_user_friends`, or contact me.

## Draw backs, TO DOs

The package is still under development. Its biggest problem the **dependence on a stable internet connection**,

as timeouts can break the functions.

These functions are still perform poorly in **terms** of speed and **memory-efficiency**. The first issue is hard -

at least for me - as the stability and speed of loading is limiting wait times as well as robustness.

The second problem is the reason why every call opens a new browser instead of continuing in the opened one.

The Firefox browserconsumes much memory after a certain time and page load count. The most efficient solution I

found is to create a command line R script, and calling it from an other R session, each time start and close the

Selenium server and clean the memory. This workaround also slows the workflow.

Another issue is password **encryption**, because the

functions need the password as unencrypted strings. One can use interactive login (login in the browser), but it is needed for every new call, so is not scalable. One solution would need functions which are working without

new browsers, or some solutions to store and use cookies in an R session. Another solution would be the encryprion

of the password string, but I could not find any function for this task.