Skip to contents

Creating custom Sleeper API calls

The Sleeper API is pretty extensive. If there is something you’d like to access that’s beyond the current scope of ffscrapr, you can use the lower-level “sleeper_getendpoint” function to create a GET request and access the data, while still using the authentication and rate-limiting features I’ve already created.

Here is an example of how you can call one of the endpoints - in this case, let’s pull Sleeper’s trending players data!

We’ll start by opening up this page, https://docs.sleeper.com/#trending-players, which is the documentation page for this particular endpoint. From here, we can see that Sleeper’s documentation says the endpoint is:

https://api.sleeper.app/v1/players/<sport>/trending/<type>?lookback_hours=<hours>&limit=<int>

On first glance, you can see that it takes two parameters within the endpoint call itself (sport and type) and we can further adjust the query with HTTP parameters lookback_hours and limit. The sleeper_getendpoint function already has the https://api.sleeper.app/v1/ part encoded, so all we’ll need to do is pass in the remaining part of the URL as the endpoint, and pass the HTTP parameters in as arguments to the function (these are case sensitive!)

We can use the glue package to parameterise this, although you can also use base R’s paste function just as easily.


type <- "add"

query <- glue::glue('players/nfl/trending/{type}')

query
#> players/nfl/trending/add

response_trending <- sleeper_getendpoint(query,lookback_hours = 48, limit = 10)
#> Using request.R from "ffscrapr"

str(response_trending, max.level = 1)
#> List of 3
#>  $ content :List of 10
#>  $ query   : chr "https://api.sleeper.app/v1/players/nfl/trending/add?lookback_hours=48&limit=10"
#>  $ response:List of 9
#>   ..- attr(*, "class")= chr "response"
#>  - attr(*, "class")= chr "sleeper_api"

Along with the parsed content, the function also returns the query and the response that was sent by the server. These are helpful for debugging, but we can turn the content into a dataframe with some careful application of the tidyverse.


df_trending <- response_trending %>% 
  purrr::pluck("content") %>% 
  dplyr::bind_rows()

head(df_trending)
#> # A tibble: 6 × 2
#>   player_id  count
#>   <chr>      <int>
#> 1 6820      521768
#> 2 2161      374758
#> 3 7045      235794
#> 4 3164      205688
#> 5 GB        203645
#> 6 HOU       154698

This isn’t very helpful without knowing who these players are, so let’s pull the players endpoint in as well - this one has a convenient function!


players <- sleeper_players() %>% 
  select(player_id, player_name, pos, team, age)

trending <- df_trending %>% 
  left_join(players, by = "player_id")

trending
#> # A tibble: 10 × 6
#>    player_id  count player_name           pos   team    age
#>    <chr>      <int> <chr>                 <chr> <chr> <dbl>
#>  1 6820      521768 Clyde Edwards-Helaire RB    KC     25.1
#>  2 2161      374758 Jerick McKinnon       RB    KC     32.1
#>  3 7045      235794 Joshua Kelley         RB    LAC    26.5
#>  4 3164      205688 Ezekiel Elliott       RB    NE     28.9
#>  5 GB        203645 NA                    DEF   GB     NA  
#>  6 HOU       154698 NA                    DEF   HOU    NA  
#>  7 10235     140729 Roschon Johnson       RB    CHI    23.3
#>  8 9487      137447 Parker Washington     WR    JAX    22.2
#>  9 NO        132410 NA                    DEF   NO     NA  
#> 10 4089      108768 Gerald Everett        TE    LAC    29.9

There - this means something to us now! As of this writing (2020-11-10), Kalen Ballage was the most added player. Haven’t we been bitten by this before?