Skip to content

Latest commit

 

History

History
218 lines (143 loc) · 12.4 KB

NEWS.md

File metadata and controls

218 lines (143 loc) · 12.4 KB

ffscrapr 1.4.3

The main goal of v1.4.3 is to patch some minor bugs.

Minor changes

  • ff_league() now has a column that returns the platform's season - this is most useful when running ff_league in batch. Thank you to @joeflan for the contribution! (Fixes #287)
  • Added helper script in dev/ for switching between test cache versions.
  • Added helper script in dev/ for rebuilding test cache.
  • ff_schedule() for Sleeper now should extend into playoff weeks. (Fixes #289)
  • ff_draft() for Sleeper now has an auction_amount column if it is an auction draft. (Fixes #291)
  • dp_cleannames() now can flip names originally presented in LastName, FirstName order into First Last, and also has a custom name database to convert common alternate names to a unified one.
  • nflfastr_stat_mapping and dp_name_mapping are namespaced within the package, so that they can be used internally and externally a little more robustly.
  • Rebuilt test cache in ffscrapr-tests.

ffscrapr 1.4.2

The main goal of v1.4.2 is to patch some minor bugs.

Minor changes

  • espn_rosters now has a week argument to return a specific week of data. (Fixes #276, thanks @scottfrechette!)
  • ff_scoringhistory now tries to return a platform-specific player ID, where available. (Fixes #275)
  • Roxygen configured to use markdown (with the roxygen2md package) - hopefully fixes the lifecycle badge issues?
  • sleeper_players() - gsis_id gets str_squish applied to it to resolve some stowaway whitespaces.
  • mfl_players() - add any_of handling for columns to handle column name changes.
  • sleeper_transactions now handles offseason transactions correctly

ffscrapr 1.4.1

The main goal of v1.4.1 is to patch some issues raised by CRAN checks and also correct some bugs in the new experimental ff_scoringhistory and ff_starter_positions functions released in v1.4.0.

New features

  • nflfastr_stat_mapping is a dataframe that maps nflfastr columns to fantasy scoring rules, and is now exported for end-user usage. It is primarily used inside of ff_scoringhistory().

Minor changes

  • Added release_questions to help remind me to maintain test infrastructure
  • mfl_starter_positions now correctly calculates offensive starters (first via "iop_starters" if defined and then otherwise by subtracting idp starters) as well as kdst_starters (Resolves #264)
  • Redirected nflfastr download functions to the new nflverse/ repository locations. (Resolves #268)
  • .ffscrapr_env relocated from being a child of the base environment to being a child of the empty environment (Resolves #269)
  • ff_scoringhistory refactored and adds better support for MFL's fumbles and firstdowns. (Resolves #265)

ffscrapr 1.4.0

The main goal of v1.4.0 is to add preliminary support for connecting ffscrapr to nflfastR weekly data, and to clean up bugs from v1.3.0. Huge thanks goes to Joe Sydlowski for his contributions on scoring history (and everything else DynastyProcess!)

New Features

  • nflfastr_weekly() imports weekly offensive statistics from nflfastR's data repository.
  • nflfastr_rosters() imports team rosters from nflfastR's roster repository.
  • ff_scoringhistory() connects your league's scoring settings to the nflfastr data (c/o the functions above), and allows you to reconstruct historical scoring for your league!
  • ff_starter_positions() describes the starter rules for each player/position, including min and max starters at each position accounting for flex spots. This should be useful for calculating things like value over replacement!

Breaking Changes

  • ff_scoring for ESPN loses the "override_pos" which only had a value of 16 for D/ST specific rules
  • ff_scoring rules for ESPN and Sleeper have been expanded from one line per rule to one line per position per rule

Minor Changes

  • Force mfl_playerscores to use season + league specific players call where possible (#239)
  • Reduce minimum rows for flea_rosters test to 200, which fixes an API check issue (#242)
  • Switching memoise backend to cachem package, fixes CRAN check issue where digest is no longer imported by memoise. Cachem also apparently more performant! (#244)
  • sleeper_transactions now correctly handles multiple dropped players in one transaction. (#246).
  • mfl_transactions - adds numeric string parsing to fix bug in auction bid amount (#)
  • mfl_draft now calculates age as of timestamp and also adds an overall column. (#259, #260)

ffscrapr 1.3.0

The main goal of ffscrapr 1.3.0 is to add support for the ESPN platform. It also includes several bug fixes, code quality improvements, and a major refactor of tests to reduce overall package size.

A huge thank-you goes to Tony ElHabr for his contributions to the package for the ESPN methods.

Breaking Changes

  • custom_players arguments are deprecated for MFL - it will now return them by default.

ESPN Details

ESPN is a tricky and undocumented API. Private leagues are accessible if you use the SWID/ESPN_S2 authentication arguments, which are a lot like API keys - see the ESPN authentication vignette.

Unsupported functions:

  • ff_draftpicks() - this does not apply to ESPN primarily because it does not support draft pick trades.
  • ff_userleagues() - ESPN does not support looking up user's leagues, even when authenticated
  • Username and password features - ESPN used to have a way to authenticate via username/password, but this has been recently made more difficult. It is an area that can be revisited if/when the Python package manages it, but at this time will only be accessible with the SWID/ESPN_S2 keys.

New Functions

  • dp_cleannames() is a utility function for cleaning player names that removes common suffixes, periods, and apostrophes.
  • espn_potentialpoints() calculates the optimal lineup for each week. This is

Minor patches

  • Converted GET requests to use httr::RETRY instead - this adds some robustness for server-side issues. As suggested by Maelle Salmon's blog post on not reinventing the wheel.
  • Added some type conversions and renaming for snake_case consistency to mfl_rosters and mfl_playerscores
  • Fixed bug in MFL's ff_playerscores() function so that it correctly pulls older names. (Resolves #196)
  • Refactored all tests to move test cache files to a separate/non-package location (https://github.com/dynastyprocess/ffscrapr-tests) - so that it is not included in CRAN's package sizing
  • Fixed bugs in MFL's ff_starters() function - bad default arg, bad players call. (Fixes #202)
  • Resolve MFL's playerscores to handle vectorized request (Fixes #206)
  • Resolve bugs related to .fn_choose_season for tests (Fixes #217, #219)
  • Resolved bug in MFL's ff_rosters() by adding a week parameter (Fixes #215)
  • Coerced ff_transactions() bid_amount into a numeric (Fixes #210)
  • Removed bye franchises from ff_starters() (Fixes #212)

ffscrapr 1.2.2

Minor patches to dp_import functions to address issues discovered by CRAN checks.

Also adds minor helper function, dp_cleannames(), which is a utility function for cleaning player names that removes common suffixes, periods, and apostrophes.

Messed up the export here, whoops. Fixing for next release.

Minor patches

  • Refactored dp_values() and dp_playerids() functions to use httr backend for compat with httptest, preventing CRAN errors.
  • Added inst-level redactor for httptest.

ffscrapr 1.2.1

Minor patches

  • Caching vignette outputs in tests/testthat to making vignette-rebuilding less internet reliant
  • Changing the league_id output of sleeper_userleagues to be a character column (because of cran no-longdouble support)

ffscrapr 1.2.0

The main goal of ffscrapr 1.2.0 is to add a full set of methods for Fleaflicker. This release also adds improved caching options, including writing to your filesystem for persistent caching (see the vignette!), and one hotfix for sleeper_getendpoint.

BREAKING CHANGES

  • sleeper_getendpoint() now behaves more like the other getendpoint functions - first argument is the endpoint and any further args are passed as query parameters.

Other tweaks to existing platforms/methods

  • Small copyedits to existing vignettes.
  • Added filesystem cache capabilities and a vignette to detailing how to use it.

Fleaflicker notes

All functions now have Fleaflicker methods! Here are notes about what isn't the same:

  • fleaflicker_players() requires a connection/leagueID by default - acts a little oddly on game days as of right now.
  • ff_playerscores() - Fleaflicker's API returns season level data easily, week-level is not readily available yet without some workarounds. Everything else seems to be okay.

ffscrapr 1.1.0

The main goal of ffscrapr 1.1.0 is to add a full set of methods for Sleeper. Also adds two new generics: ff_userleagues() and ff_starters().

New generic functions

Here is a list of new functions available at the top level (ie for all platforms)

  • ff_userleagues() returns a list of user leagues. This is deployed slightly differently for MFL and Sleeper - MFL requires authentication to access user's leagues, while Sleeper doesn't have authentication so you can look up any user you like.
  • ff_starters() returns a list of players started/not-started each week. MFL will return the actual score of each player each week and calculate whether they were optimal, while Sleeper just returns the player themselves.

Sleeper notes

Almost all functions now have Sleeper methods - implemented in what hopes to be relatively familiar manner to MFL. Outlining the specifics of what isn't the same:

  • sleeper_userleagues() is a wrapper on ff_userleagues() that makes it easier to look up user leagues without first creating a connection object.
  • ff_playerscores() is not available for Sleeper because Sleeper removed the player stats endpoint - it will generate a warning (rather than an error). Thinking about creating some functions to calculate scoring based on nflfastr.
  • sleeper_getendpoint() is a little more simple than MFL's equivalent - just pass a string url (minus api.sleeper.app/v1) or pass in chunks of code, the function will automatically paste them together with "/".
  • Added generic and method for ff_userleagues() - Sleeper league IDs are more annoying than MFL to handle, so the more intuitive way is to look up the user's league_ids by username first. MFL does have a parallel feature even if used for different purposes.
  • Added two vignettes, showing "Getting Started" as well as one for custom API calls

MFL changes

  • Added method for ff_userleagues()
  • Added handling for offensive_points and defensive_points in ff_standings() (#69, nice.)
  • Added ff_starters() as requested by #76 (thanks, Mike!)
  • Added an httr::handle_reset() call to fix login-caching bug.
  • Polished vignettes a little.

Other release tweaks in 1.1.0

  • Now uses {checkmate} for testing.
  • Silently swallowing up unused args in mfl_connect and sleeper_connect.
  • Uses describeIn instead of rdname for method documentation.
  • Wrap all documentation examples in donttest - ratelimiting AND running in under 5 seconds each is pretty challenging!

ffscrapr 1.0.0

This is the first (major) version of ffscrapr and it is intended to build out the full set of functions for the first API platform: MFL.

Future versions will add more platforms via methods mapped to the same functions.

Functions include:

  • ff_connect (and sibling mfl_connect) to establish connection parameters and ratelimiting
  • mfl_getendpoint as a low-level function for making GET requests from MFL
  • ff_draft gets draft results
  • ff_draftpicks gets current and future draft picks that have not yet been selected
  • ff_franchises gets franchise-level identifiers and divisions
  • ff_league gets league-level summaries of rules, players, and franchises
  • ff_playerscores gets playerweek-level scores
  • ff_rosters gets franchise-level rosters complete with naming
  • ff_schedule gets weekly fantasy schedules
  • ff_scoring gets scoring rules
  • ff_standings gets league-level season summaries
  • ff_transactions gets a list of all transactions and cleans them into a data frame.