NEWS

nascaR.data 3.0.1 (2026-02-16)

Removed disk cache. The on-disk cache from v3.0.0 silently served stale data across R sessions. load_series() now downloads fresh data from R2 once per session (in-memory cache only).

Bundled data removed. The cup_series, xfinity_series, and truck_series datasets are no longer shipped with the package. All data is now served from Cloudflare R2. Use load_series() to access data:
```
cup <- load_series("cup")
nxs <- load_series("nxs")
truck <- load_series("truck")
```
arrow is now a required dependency (moved from Suggests to Imports) since all data access goes through parquet files on R2.
"xfinity" series renamed to "nxs". The second-tier series has changed title sponsors four times: Busch (1984-2007), Nationwide (2008-2014), Xfinity (2015-2025), and O'Reilly Auto Parts (2026-present). The identifier "nxs" is NASCAR's own sponsor-neutral abbreviation, so it will never go stale. Replace load_series("xfinity") with load_series("nxs").
find_driver(), find_team(), find_manufacturer() removed. The get_*_info() functions already include fuzzy matching and return actual data. Use get_driver_info("bell") instead of find_driver("bell").
data("cup_series") no longer works. Replace all data() calls with load_series(). The old lazy-loaded dataset names are gone.

load_series(): Downloads series data from R2 and caches in memory. Subsequent calls within the same session are instant. Use refresh = TRUE to force re-download.
clear_cache(): New exported function to reset the in-memory cache.
R2-canonical pipeline: The weekly GitHub Actions scraper now reads existing data from R2, appends new races, and uploads back to R2. No local rda files are generated or committed.

Migrated from httr to httr2 for HTTP requests. The scraper now uses httr2::request() with built-in retry logic (req_retry()).
Consolidated web scraping with httr2, imap_dfr() indexing, explicit column type coercion, placeholder detection, and empty table guards.
Data validation framework for schema, integrity, and value checks.
Code quality enforced with styler and lintr (zero warnings).
All stop() calls in package code replaced with rlang::abort().

Deprecating the weekly branch! Weekly race results will be added to the main branch. CRAN-stable version is available via CRAN:

install.packages("nascaR.data")
remotes::install_github("kyleGrealis/nascaR.data") # please do not use "@weekly"

Interactive driver/team/manufacturer selection: When multiple matches are found, users can now select from a numbered list
Intelligent fuzzy matching: Dramatically improved search algorithm that handles typos, partial names, and word boundaries
- find_driver("kyle") -> returns Kyle Busch, Kyle Larson, Kyle Petty, etc.
- find_team("gibbs") -> finds Joe Gibbs Racing
- find_driver("earnhart") -> correctly finds Earnhardt family drivers
Flexible series input: All functions now accept both character strings AND data frames
- get_driver_info("kyle", "cup") check
- get_driver_info("kyle", "Cup Series") check
- get_driver_info("kyle", cup_series) check
Smart string matching: Handles variations like "cup", "Cup Series", "xfinity", "Xfinity Series" automatically

Consolidated codebase: Replaced three separate fuzzy matching files with one unified system
Priority-based matching: Exact matches > starts with > contains > word boundaries > fuzzy similarity
Non-interactive mode: Dashboard/script developers can set interactive = FALSE to get list returns
Removed dependency on problematic Levenshtein distance calculations
Eliminated interactive prompts that broke in non-interactive environments

Fixed fuzzy matching returning irrelevant results
Resolved cases where obvious matches weren't found due to strict string matching
Eliminated interactive readline prompts that failed in scripts and R Markdown

Added missing races. The Cup Series season finale was omitted for a number of years from 2002 to 2022. Thank you to Nick Triplett for the catching the mistake!
Seg Points has been removed. Instead, S1 & S2 variables correspond to the driver's finishing position during each segment.
Updated missing track information (length, surface type) for 32 Cup races with varying years, mostly pre-2000s.

dev branch will be updated with the most recent racing results every Monday at 10AM during the race season (February through November).
New get_*_info() functions provide summary statistics on a by-race, season, or career format.
Fuzzy matching has been included to search across the database in the respective series

Error handling for placeholder race. The "DriverAverages.com" site will sometimes have one row for the upcoming race. This has routinely caused issues for the weekly scraping functions accidentally "recognizing" that as having been a completed race. The oversight has been addressed by removing that (essentially) blank row, decreasing the index, and continuing to scrape for new race data.
Fuzzy finding is used within the functions returning the driver, team, or manufacturer information. There is room for improvement: "Chris Bell" would find "Christopher Bell" instead of "Chris Miller" as it does now. Improvements would include a stronger focus on the driver or owner's last name, but this will take some trial and error to really dial in.

BREAKING CHANGES:
- Web data scraped using R packages & removed existing Python scripts.
- Reduced number of datasets for Cup, Xfinity, & Truck series results by series only.
Integrated GitHub Actions for auto-updating all series results