rvest final tutorial 4

.r

# select movie name
imdb %>%
  html_nodes(".lister-item-header a") %>%
  html_text() -> movie_names
  
# select year
imdb %>%
  html_nodes(".text-muted.unbold") %>%
  html_text() %>%
  # extract numeric year
  stringr::str_extract('[0-9]+') %>% 
  as.integer() -> years

# select rating
imdb %>%
  html_nodes(".ratings-imdb-rating strong") %>%
  html_text() %>%
  as.numeric() -> ratings
  
# create dataframe
imdb_data <- data.frame(movie_names, years, ratings)

# export csv file
write.csv(imdb_data, "imdb_data.csv")