R: Using rvest package instead of XML package to get links from URL

Question

I use XML package to get the links from this url.

# Parse HTML URL
v1WebParse <- htmlParse(v1URL)
# Read links and and get the quotes of the companies from the href
t1Links <- data.frame(xpathSApply(v1WebParse, '//a', xmlGetAttr, 'href'))

While this method is very efficient, I've used rvest and seems faster at parsing a web than XML. I tried html_nodes and html_attrs but I can't get it to work.

Answer 1

Despite my comment, here's how you can do it with rvest. Note that we need to read in the page with htmlParse first since the site has the content-type set to text/plain for that file and that tosses rvest into a tizzy.

library(rvest)
library(XML)

pg <- htmlParse("http://www.bvl.com.pe/includes/empresas_todas.dat")
pg %>% html_nodes("a") %>% html_attr("href")

##   [1] "/inf_corporativa71050_JAIME1CP1A.html" "/inf_corporativa10400_INTEGRC1.html"  
##   [3] "/inf_corporativa66100_ACESEGC1.html"   "/inf_corporativa71300_ADCOMEC1.html"  
## ...
## [273] "/inf_corporativa64801_VOLCAAC1.html"   "/inf_corporativa58501_YURABC11.html"  
## [275] "/inf_corporativa98959_ZNC.html"  

That further illustrates rvest's XML package underpinnings.

UPDATE

rvest::read_html() can handle this directly now:

pg <- read_html("http://www.bvl.com.pe/includes/empresas_todas.dat")

Answer 2

I know you're looking for an rvest answer, but here's another way using the XML package that might be more efficient than what you're doing.

Have you seen the getLinks() function in example(htmlParse)? I use this modified version from the examples to get href links. It's a handler function so we can collect the values as they are read, saving on memory and increasing efficiency.

links <- function(URL) 
{
    getLinks <- function() {
        links <- character()
        list(a = function(node, ...) {
                links <<- c(links, xmlGetAttr(node, "href"))
                node
             },
             links = function() links)
        }
    h1 <- getLinks()
    htmlTreeParse(URL, handlers = h1)
    h1$links()
}

links("http://www.bvl.com.pe/includes/empresas_todas.dat")
#  [1] "/inf_corporativa71050_JAIME1CP1A.html"
#  [2] "/inf_corporativa10400_INTEGRC1.html"  
#  [3] "/inf_corporativa66100_ACESEGC1.html"  
#  [4] "/inf_corporativa71300_ADCOMEC1.html"  
#  [5] "/inf_corporativa10250_HABITAC1.html"  
#  [6] "/inf_corporativa77900_PARAMOC1.html"  
#  [7] "/inf_corporativa77935_PUCALAC1.html"  
#  [8] "/inf_corporativa77600_LAREDOC1.html"  
#  [9] "/inf_corporativa21000_AIBC1.html"     
#  ...
#  ...

Answer 3

# Option 1
library(RCurl)
getHTMLLinks('http://www.bvl.com.pe/includes/empresas_todas.dat')

# Option 2
library(rvest)
library(pipeR) # %>>% will be faster than %>%
html("http://www.bvl.com.pe/includes/empresas_todas.dat")%>>% html_nodes("a") %>>% html_attr("href")

Answer 4

Richard's answer works for HTTP pages but not the HTTPS page I needed (Wikipedia). I substituted RCurl's getURL function as below:

library(RCurl)

links <- function(URL) 
{
  getLinks <- function() {
    links <- character()
    list(a = function(node, ...) {
      links <<- c(links, xmlGetAttr(node, "href"))
      node
    },
    links = function() links)
  }
  h1 <- getLinks()
  xData <- getURL(URL)
   htmlTreeParse(xData, handlers = h1)
  h1$links()
}