Extract Links from Webpage using R

Question

The two posts below are great examples of different approaches of extracting data from websites and parsing it into R.

Scraping html tables into R data frames using the XML package

How can I use R (Rcurl/XML packages ?!) to scrape this webpage

I am very new to programming, and am just starting out with R, so I am hoping this question is pretty basic, but given those posts above, I imagine that it is.

All I am looking to do is extract links that match a given pattern. I feel like I could probably use RCurl to read in the web pages and extract them brute force method using string expressions. That said, if the webpage is fairly well formed, how would I go about doing so using the XML package.

As I learn more, I like to "look" at the data as I work through the problem. The issue is that some of these approaches generate lists of lists of lists, etc., so it is hard for someone that is new (like me) to walk through where I need to go.

Again, I am very new to all that is programming, so any help or code snippets will be greatly appreciated.

Answer 1

The iMacros wiki has a nice chapter on web scraping. Mabe that gives you some ideas.

Answer 2

The documentation for htmlTreeParse shows one method. Here's another:

> url <- "http://stackoverflow.com/questions/3746256/extract-links-from-webpage-using-r"
> doc <- htmlParse(url)
> links <- xpathSApply(doc, "//a/@href")
> free(doc)

(You can drop the "href" attribute from the returned links by passing "links" through "as.vector".)

My previous reply:

One approach is to use Hadley Wickham's stringr package, which you can install with install.packages("stringr", dep=TRUE).

> url <- "http://stackoverflow.com/questions/3746256/extract-links-from-webpage-using-r"
> html <- paste(readLines(url), collapse="\n")
> library(stringr)
> matched <- str_match_all(html, "<a href=\"(.*?)\"")

(I guess some people might not approve of using regexp's here.)

matched is a list of matrixes, one per input string in the vector html -- since that has length one here, matched just has one element. The matches for the first capture group are in column 2 of this matrix (and in general, the ith group would appear in column (i + 1)).

> links <- matched[[1]][, 2]
> head(links)
[1] "/users/login?returnurl=%2fquestions%2f3746256%2fextract-links-from-webpage-using-r"
[2] "http://careers.stackoverflow.com"                                                  
[3] "http://meta.stackoverflow.com"                                                     
[4] "/about"                                                                            
[5] "/faq"                                                                              
[6] "/"