fitzRoy v1.8.0 on CRAN

I’m pleased to announce that fitzRoy v1.8.0 has just been accepted to CRAN!

This one is a little different to recent releases. There are no headline new functions — instead it’s a maintenance release focused on the boring stuff that makes everything else work: faster data loading, scraping that doesn’t get blocked, and much better behaviour when a data source is having a bad day.

You can read about all the changes below.

library(fitzRoy)

Faster player stats

fetch_player_stats_afltables() and fetch_player_stats_footywire() now read canonical Parquet files from the fitzroy_data release, rather than the legacy .rda files they used previously.

The data you get back is unchanged — same columns, same types, same rows. It just arrives faster, and the cached files are smaller.

fitzRoy::fetch_player_stats_afltables(2024) |> dim()

i Looking for data from 2024-01-01 to 2024-12-31

v Looking for data from 2024-01-01 to 2024-12-31 [16ms]

i Fetching cached data from <github.com/jimmyday12/fitzRoy_data>

v Fetching cached data from <github.com/jimmyday12/fitzRoy_data> [2s]

i No new data found! Returning cached data

v No new data found! Returning cached data [12ms]

i Tidying data

v Tidying data [33ms]

[1] 9936   81

This adds nanoparquet as a dependency, which is a deliberately lightweight Parquet reader with no further dependencies of its own.

Scraping that doesn’t get blocked

Several data sources have added basic bot protection that rejects requests carrying a generic default User-Agent — R’s own, or curl’s. That was showing up as intermittent, confusing connection failures in functions like fetch_player_details_footywire(), fetch_awards() and fetch_team_stats().

Rather than pretending to be a browser, fitzRoy now identifies itself honestly:

fitzRoy R package (https://github.com/jimmyday12/fitzRoy)

If you need to change it — say you’re running fitzRoy behind something that requires a particular string — you can:

options(fitzRoy.user_agent = "your string here")

Better errors when a site is down

fitzRoy scrapes live websites, so sometimes a source is simply unreachable. Previously that surfaced as whatever the underlying reader happened to throw, which wasn’t much help:

Error: cannot open the connection

fetch_results_afltables() now tells you what it was actually trying to do:

Error in `fetch_results_afltables()`:
! Could not read data from <https://afltables.com/afl/stats/biglists/bg3.txt>.
i The site may be unavailable, or you may not have an internet connection.
x cannot open the connection

This also brings the package in line with CRAN policy, which asks that packages using internet resources fail gracefully with an informative message.

Bug Fixes

Fixed an R CMD check ERROR on CRAN’s Fedora machine. A test file worked out the current season by calling fetch_results_afltables() at the top level of the file — and top-level test code runs before the skip_on_cran() guards inside the test blocks, so it hit the network unconditionally on check machines that have none (#290)
Fixed fetch_player_details_footywire(), fetch_awards() and fetch_team_stats() failing intermittently due to bot protection
Fixed get_aflw_detailed_data() emitting a deprecation warning for get_aflw_cookie() — a function it called internally, that callers never touched and couldn’t see in the public API (#291)
Corrected two roxygen examples that pointed at the wrong replacement function (#291)

Under the hood

The test suite has been reworked to cope with data sources being temporarily unreachable (#292).

Because fitzRoy tests against live sites, a brief outage at afltables.com or footywire.com used to fail the whole build — which meant a red CI run usually said more about the internet than about the code, and a genuine regression could hide in the noise. Tests that depend on a live source now skip cleanly when that source can’t be reached, and fail only when something is actually broken.

Not something you’ll ever see as a user, but it makes every future release easier to trust.

Enjoy!