Prep

Packages

# devtools::install_github("johndharrison/wdman")
# devtools::install_github("johndharrison/binman")
# devtools::install_github("ropensci/RSelenium")
# devtools::install_github("johndharrison/seleniumPipes")
# devtools::install_github("systats/hrscraper")
pacman::p_load(dplyr, ggplot2, rvest, xml2, stringr, purrr, RSelenium, wdman, seleniumPipes, hrscraper)

Init Drivers

# port <- sample(4000L:5000L, 1)
# 
# pJS <- wdman::phantomjs(
#   port = port
# )
# pJS$process

rD <- RSelenium::rsDriver(verbose = F)
remDr <- rD$client

Base Url

# base_url <- "https://login.xing.com/"
# remDr$navigate(base_url)
#remDr$screenshot()

Search

Login

remDr$open()
post_login(
  "nomis.roth@gmx.net",
  "Toshiba32"
)
#remDr$screenshot(display = T)
#remDr$getCurrentUrl()

Stage search query

keywords <- c("tetec")
tetec_search <- scrape_main(keywords)
tetec_search %>% glimpse
# save(tetec_base, file = "tetec_base.Rdata")
post_login(
  "msautter1991@gmail.com",
  "msautter1234"
)
#get_profile(tetec_search$name_id[1])
get_profile_progress <- tidyMBO::progressively(purrr::safely(get_profile), nrow(tetec_search))

tetec <- tetec_search %>% 
  .$name_id %>% 
  purrr::map(get_profile_progress) %>% 
  purrr::map("result") %>% 
  purrr::reduce(bind_rows) %>%
  mutate(time_stamp = Sys.Date())

tetec <- tetec %>% 
  left_join(tetec_search)

tetec %>%
  glimpse


systats/hrscraper documentation built on May 28, 2019, 9:52 a.m.