spotify_scraper.R

library(spotifyr)
library(tidyverse)


Sys.setenv(SPOTIFY_CLIENT_ID = '')
Sys.setenv(SPOTIFY_CLIENT_SECRET = '')

xml <- read_html("https://developer.spotify.com/documentation/general/guides/scopes/")
scopes <- xml %>% html_nodes("code") %>% html_text() %>% unique()
auth <- get_spotify_authorization_code(scope = scopes)
#auth <- get_spotify_access_token()
#get_spotify_authorization_token()

#access_token <- get_spotify_access_token()

#get_spotify_authorization_code()

get_my_recently_played(limit = 5, authorization = auth) 


### get all albums from artist
input_artist_id <- "7AGSJihqYPhYy5QzMcwcQT"
#artist_name <- "Headhunterz"
# get artist name
input_artist_name <- get_artist(id = input_artist_id)$name


pgs <- 49 * c(0:10)

sp_albums <- pgs %>% 
  purrr::map(~ get_artist_albums(id = input_artist_id, offset = ., limit = 50))

sp_albums_df <- sp_albums %>% 
  bind_rows() %>% 
  as_tibble() %>% 
  distinct(id, name, release_date) %>% 
  arrange(release_date) %>% 
  rename_all(~ str_c("album_", .x))

albums_ids <- unique(sp_albums_df$album_id)


### get all tracks from albums
get_album_tracks_include_album <- function(...){
  args <- list(...)
  out <- get_album_tracks(...)
  
  if (inherits(out, "data.frame")){
    if (nrow(out) > 0){
      out$album_id <- args[[1]]
    }
  }
  out
}

sp_tracks <- albums_ids %>% 
  map(get_album_tracks_include_album, limit = 50)


sp_tracks_df <- sp_tracks %>% 
  discard(is.null) %>% 
  bind_rows() %>% 
  as_tibble() %>% 
  mutate(artists = map(artists, ~ rename_all(.x, ~ str_c("artist_", .x)))) %>% 
  unnest(artists)


sp_tracks_df_clean <- sp_tracks_df %>% 
  filter(artist_id == input_artist_id) %>% 
  select(track_id = id, track_name = name, track_uri = uri, album_id) %>% 
  # clean track names
  mutate(track_name_cleaned = str_squish(str_to_lower(track_name))) %>% 
  # filter out tracks with same name
  distinct(track_name_cleaned, .keep_all = TRUE) %>% 
  select(-track_name_cleaned) %>% 
  left_join(sp_albums_df, by = "album_id") %>% 
  arrange(album_release_date)

track_uris <- unique(sp_tracks_df_clean$track_uri)
### create playlist
plist_name <- str_glue("{input_artist_name} - The Complete Collection")

new_plist <- create_playlist(user_id = "12139910710",
                             name = plist_name,
                             public = TRUE,
                             collaborative = FALSE,
                             description = "Made with programming using R by github.com/sillasgonzaga")

### add tracks to playlist

# quebrar as musicas em grupos de cem
tracks_to_add <- tibble(track_uris) %>% 
  mutate(batch = seq_along(track_uris) %/% 100,
         position = row_number() - 1) %>% 
  split(.$batch)

for (i in seq_along(tracks_to_add)){
  add_tracks_to_playlist(
    playlist_id = new_plist$id,
    uris = tracks_to_add[[i]]$track_uris,
    position = min(tracks_to_add[[i]]$position)
  )
}

cat(new_plist$external_urls$spotify)