blog_img.R

library(httr)
library(rvest)
library(dplyr)
library(magrittr)
library(stringr)
library(jsonlite)
library(purrr)
library(tibble)
library(openssl)
library(crul)

get_links <- function(tar){
  GET(tar) %>%
    content() %>%
    str_split(',"pagingHtml') %>%
    .[[1]] %>%
    .[1] %>%
    str_c(.,"}") %>%
    fromJSON()
}

size <- 30
root <- "https://blog.naver.com/PostTitleListAsync.nhn?blogId=kdh4548&viewdate=&currentPage="
seq2 <- str_c("&categoryNo=0&parentCategoryNo=&countPerPage=", size)

str_c(root, 1, seq2) %>%
  get_links() ->
  dat

tcnt <- as.integer(dat$totalCount)

1:(as.integer(tcnt/size) + 1) %>%
  str_c(root, ., seq2) %>%
  map( ~ get_links(.x) %>%
         .$postList %>%
         .$logNo) %>%
  flatten() %>% 
  unlist() ->
  links

root <- "https://blog.naver.com/kdh4548/"

links %>%
  str_c(root, .) %>%
  # .[1:10] %>% 
  map(
    ~ slowly(read_html)(.x) %>%
      html_nodes("iframe#mainFrame") %>%
      html_attr("src")
  ) %>%
  flatten() %>% 
  unlist() ->
  in_frame

root <- "https://blog.naver.com"

in_frame %>%
  str_c(root, .) ->
  tar

dir.create("img")
for (i in 1:length(tar)) {
  cat("now : ",i," / ", length(tar), "\n")
  tar[i] %>%
    read_html() %>%
    html_nodes("img") %>%
    html_attr("src") %>%
    enframe(name = NULL) %>%
    filter(str_detect(value, "postfiles")) %>%
    pull(value) %>%
    str_split("type=") %>%
    map(
      ~ {
        tem <- c()
        tem[1] <- .x %>% .[1] %>% str_c("type=w580")
        tem[2] <- .x %>% .[1] %>% str_split("/") %>% .[[1]] %>% .[5] %>% str_replace(fixed(".kdh4548"),"")
        tem
      }
    ) ->
    tar_img
  print(length(tar_img))
  if (length(tar_img) == 0) {
    next
  }
  
  cnt <- length(tar_img)
  lst <- cnt %/% 100 + 1
  if (cnt %% 100 == 0) {
    lst <- cnt %/% 100
  }
  
  if (length(tar_img) == 1) {
    GET(tar_img[[1]][1], write_disk(str_c("img/", tar_img[[1]][2])))
    next
  }
  
  for (j in 1:lst) {
    str <- 1 + 100 * (j - 1)
    if (j * 100 > cnt) {
      end <- cnt
    } else {
      end <- 100 * j
    }
    cc <- Async$new( urls = tar_img[str:end] %>% transpose() %>% .[[1]] %>% unlist() )
    res <- cc$get(disk = str_c("img/", tar_img[str:end] %>% transpose() %>% .[[2]]), verbose = F)
  }
}