image-archive-complete.R

# Year: 2012
# http://www.runtwolf.com/cc-2012-photo-gallery.html

# clear directory, call libraries
rm(list=ls())
library(rvest)
library(dplyr)

# declare functions
substrRight <- function(x, n){substr(x, nchar(x)-n+1, nchar(x))}
substrLeft <- function(x, n){substr(x, 1, n)}

# get page with list of album names
url = 'http://www.runtwolf.com/cc-2012-photo-gallery.html'
download.file(url, destfile = 'page.html', quiet=TRUE)
content <- read_html('page.html')

# clean list of strings
albums <- content %>% html_nodes('a') %>% html_attr('href')
albums <- albums[grepl('old.runtwolf.com/CC2012/', albums)]
albums <- substr(albums, gregexpr(pattern ='/CC2012/', albums), gregexpr(pattern ='/index.html', albums))
albums <- substr(albums, 9, nchar(albums)-1)
album_urls <- paste('http://www.old.runtwolf.com/cc2012/', albums, '/thumbnails.html', sep = "")

# Session 1
albums <- albums[1:30]
album_urls <- album_urls[1:30]

# Session 2 
#albums <- albums[31:60]
#album_urls <- album_urls[31:60]

# Session 3
#albums <- albums[61:length(albums)]
#album_urls <- album_urls[61:length(album_urls)]

##### DOWNLOAD PHOTOS FROM EACH ALBUM ##### 

mainDir <- '/Users/erikgregorywebb/Documents/Python/runtwolf/R/2012/XC'
for (i in 1:length(album_urls)) {
  setwd(file.path(mainDir))
  Sys.sleep(1)
  download.file(album_urls[i], destfile = 'page.html', quiet=TRUE)
  content <- read_html('page.html')
  images <- content %>% html_nodes('img') %>% html_attr('src')
  images <- substrRight(images, 12)
  
  subDir = albums[i]
  dir.create(file.path(mainDir, subDir))
  setwd(file.path(mainDir, subDir))
  
  for (j in 1:length(images)) {
    Sys.sleep(1)
    url = paste('http://old.runtwolf.com/CC2012/', albums[i], '/images/', images[j], sep = "")
    print(url)
    try(download.file(url, destfile = paste(paste(albums[i], j, sep = "-"), ".jpg", sep = ""), mode = "wb", quiet = T))
  }
}