get_playlist_tracks() limit is set pretty low

[blechturm]

According to the documentation the limit parameter can range from 1 to 100.

This is a bit low since many playlists have several hundred entries. Is there a way to increase this or is this a limitation of the spotify API?

Thanks!

EDIT: I just realized how the pagination system of the spotify API works, see: https://developer.spotify.com/documentation/web-api/

Its pretty straight forward to parse through the whole playlist by increasing the offset parameter as long as there non-null list returns. The issue can be closed :-). But maybe adding an example in the documentation would be helpful for future users.

Btw.: thanks for building this package! Awesome work!

[antaldaniel]

Documentation pull request or examples are more than welcome.

[blechturm]

I can try... Would be my first pull request tho 😅

Is a tidyverse and purrr centric example okay or should it be base-R?

[antaldaniel]

Maybe you can write the example here and we'll see if it will be an illustration or an example or something else. Not the entire tidyverse is imported into spotifyr.

[blechturm]

How about this:

library(tidyverse)
library(purrr)
library(spotifyr)

# The get_playlist_tracks() function can only fetch 100 tracks each time it is called. 
# This is a limitation of the Spotify API. 

# We use the popular Lofi Hiphop Playlist as example. It has currently 300 entries.

playlist_id <- "0vvXsWCC9xrXsKd4FyS8kM"

# As we can see only the first 100 tracks are fetched

nrow(get_playlist_tracks(playlist_id))

# To get all the other tracks we need to increase the offset parameter
# of the function, while leaving the limit parameter at its default (100).

# Note that the playlist entries are identified with indices using zero based numbering. 
# This simply means that the first index starts at 0 instead on 1 as we are accustomed in R. 

# The first one hundred tracks of the playlist correspond to an index ranging from {0,1,2, ... , 99}. 
# To get the next one hundred tracks in the playlist we need the indices ranging from {100, 101, 102, ... , 199}. 

# This is achieved by setting the offset parameter 100 to fetch the next hundred tracks 
# then setting the offset parameter to 200 and so on.

# We create an arbitrarily long vector of offset values starting with 0

offset_vector <- c(0,seq(from=100, to = 10000, by  = 100))

# This vector is then used to create a dataframe we will use to iterate.

fetched_playlist <- data.frame(offset  = offset_vector) %>%
  
# We now iterate through all the values of the offset vector using the map 
# function of the purrr package and call the  get_playlist_tracks() function.
  
  mutate(results = map(offset, 
                        ~get_playlist_tracks(
                          playlist_id, 
                          limit=100,
                          offset = .x),
                        )) %>%
  
# To bind each fetched list element together we select the results column and unnest it

  select(results) %>%
  unnest(results)


# We now have a dataframe containing all the tracks in the playlist

glimpse(fetched_playlist)

[blechturm]

I could create a Rmarkdown file that uses this approach to find out which tempo or key dominates the playlist. This is btw. the use case I used your package for.

[eledroos]

Hmm tried running this code exactly and I got the following error:

Error in `list_unchop()`:
! Can't combine `x[[1]]` <data.frame> and `x[[6]]` <list>.

Did you encounter anything similar?

[blechturm]

``Hi @eledroos ,

Just tried the code again. You are right. Something in the purrr package might have changed.

This should work:

library(tidyverse)
library(purrr)
library(spotifyr)

# The get_playlist_tracks() function can only fetch 100 tracks each time it is called. 
# This is a limitation of the Spotify API. 

# We use the popular Lofi Hiphop Playlist as example. It has currently 300 entries.

playlist_id <- "0vvXsWCC9xrXsKd4FyS8kM"

# As we can see only the first 100 tracks are fetched

nrow(get_playlist_tracks(playlist_id))

# To get all the other tracks we need to increase the offset parameter
# of the function, while leaving the limit parameter at its default (100).

# Note that the playlist entries are identified with indices using zero based numbering. 
# This simply means that the first index starts at 0 instead on 1 as we are accustomed in R. 

# The first one hundred tracks of the playlist correspond to an index ranging from {0,1,2, ... , 99}. 
# To get the next one hundred tracks in the playlist we need the indices ranging from {100, 101, 102, ... , 199}. 

# This is achieved by setting the offset parameter 100 to fetch the next hundred tracks 
# then setting the offset parameter to 200 and so on.

# We create an arbitrarily long vector of offset values starting with 0

offset_vector <- c(0,seq(from=100, to = 10000, by  = 100))

# This vector is then used to create a dataframe we will use to iterate.

fetched_playlist <- tibble(offset  = offset_vector) %>%
  
  # We now iterate through all the values of the offset vector using the map 
  # function of the purrr package and call the  get_playlist_tracks() function.
  
  mutate(results = map(offset, 
                       ~get_playlist_tracks(
                         playlist_id, 
                         limit=100,
                         offset = .x),
  )) %>%
  
  # we first have to drop all empty lists 
  filter(lengths(results) > 0) %>%
  # To bind each fetched list element together we select the results column and unnest it
  select(results) %>%
  unnest(results)


# We now have a dataframe containing all the tracks in the playlist

glimpse(fetched_playlist)

[eledroos]

That worked thanks!

Out of curiosity, if I now wanted to get the features of fetched_playlist, would I repeat this procedure for get_track_audio_features()?

[blechturm]

to get the audio features you have to use the track id values from the unnested dataframe and plug them into get_track_audio_features()

purrr is great for this...