README.md
In wehoop: Access Women's Basketball Play by Play Data

player_core parity fixtures

Golden-master oracle for espn_basketball_player_core(), captured from the Python implementation that currently produces the released player_core dataset. The R function is a port; these files are what it must reproduce.

| | | |---|---| | Oracle functions | sportsdataverse.wnba.helper_wnba_player_core, sportsdataverse.wbb.helper_wbb_player_core | | sdv-py version | 0.0.75 | | Captured | 2026-08-03 | | Payload source | wehoop-wnba-raw/wnba/player_core/json/{id}.json, wehoop-wbb-raw/wbb/player_core/json/{id}.json (committed raw trees) | | Capture command | per-league helper on each payload, pl.concat, write_csv |

Both sdv-py helpers are thin re-exports of one implementation — the core-v2 athlete resource is the same payload shape across nba/wnba/mbb/wbb, so the projection is league-agnostic. That is why one R function serves both leagues here, and why hoopR::espn_basketball_player_core() is byte-identical to this one. A change to either must land in both in the same session, verified.

wnba_1002.json, wnba_1007.json, wbb_13905.json, wbb_10000.json — ESPN core-v2 athlete payloads, copied unmodified from the raw trees (git may normalise line endings; the parsed JSON is identical). Never hand-written.
expected_player_core.csv — the oracle. 4 rows x 35 columns.

Both leagues are represented deliberately: a divergence that only shows on college payloads must fail here, not in production.

| athlete | league | covers | |---|---|---| | 1002 Jessica Breland | wnba | college and draft — the fully-populated pro path | | 1007 Ta'Shia Phillips | wnba | no college, no draft, and the literal "NA" position abbreviation (see below) | | 13905 Natasha Cloud | wbb | college + draft on the college side | | 10000 Rebecca Tobin | wbb | no college node yet a resolved birth_country — college payloads carry birthCountry at the TOP LEVEL rather than nested under birthPlace. The men's fixtures cannot reach this branch. |

`"NA"` is a value, not a missing marker

ESPN's "Not Available" position has abbreviation: "NA" — the two characters N,A. readr's default na = c("", "NA") turns that into a missing value even under col_character(), which silently corrupts the oracle and makes a correct port look wrong. The test therefore reads with na = "" only. Empty cells, which is what genuine nulls serialise to, still read as NA.

35 columns: athlete_id integer (the join key into player_box / player_season_stats, never a float-origin "123.0"); age, position_id, college_id, current_team_id, experience_years, status_id, draft_year, draft_round, draft_selection integer; height, weight double; active logical; the rest character.

current_team_id is the athlete's team today, not their team in any past season, and height/weight/jersey are a current snapshot — ESPN overwrites them in place. The season a released row is filed under is participation (who appeared that year, from player_box), not the vintage of the bio.

College and team ids come from parsing the $ref URL (/colleges/{id}, /teams/{id}) — the port must never fetch the ref.

Re-capture only when the Python oracle changes deliberately, and re-copy the payloads in the same pass — a refreshed oracle against stale payloads is worse than no oracle.