Lineup Lounge

Cricket games. The home screen lists the modes; Lineup, Ballpark and Who Am I? each have a daily round (with its own stats and streak) and a practice mode, and Showdown is free play.

• Lineup: put five players, teams or records in order. Five attempts; each attempt marks every row as in the right spot (green), one place off (amber) or further away. Practice is unlimited and filtered by format (IPL, Test, ODI, T20I, World Cup, T20 World Cup).
• Ballpark: five estimation questions a day, out of 500, across all of cricket: careers, record books, crowds, dates and size-it-up sums ("how many cricket balls would fill an Olympic pool?"). One guess each, scored by how many times off it is: 100 within 5%, 70 at 2× off, 52 at 3×, 0 at 10×, the same whether too high or too low. The reveal shows a log number line.
• Showdown: stat cards against the computer. Pick a format (ODI, T20I, Test or IPL); 30 cards from that format's deck (64 to 100 players) are dealt 15 each. Each round both sides play a card; you and the computer take turns naming a stat, and the higher number scores 10 points (5 each on a tie). Most points after 15 rounds wins. A game in progress is saved per format.
• Who Am I?: a stat card with the name and team hidden and the photo blurred; name the player in five guesses, picked from that format's deck. Each wrong guess shows what it shares with the answer (team, role, era, batting hand) and unlocks a hint in the player's own voice: how they play, their debut and a career highlight, the team-mates they kept (one from a different team), then their nickname, birthplace and team. A close guess unlocks two, and the photo clears a little with every guess that gets closer. The daily card's format rotates through ODI, T20I, Test and IPL; practice lets you pick the format.

npm install
npm run dev     # http://localhost:5173
npm test        # scoring logic + validation of every puzzle in the dataset
npm run build

Data

src/data/cricket.json, src/data/ballpark.json, src/data/cards.json and src/data/whoami.json are generated; don't edit them by hand. Sources:

Source	Used for	Licence
Cricsheet ball-by-ball data + register	IPL puzzles (complete, 2008 onward), player names and countries	ODC-By 1.0
Wikipedia record lists	International career/match records, World Cup timelines	CC BY-SA 4.0
Wikidata	Full names for the few players Cricsheet only lists by initials	CC0
Wikipedia (non-free files)	IPL franchise logos in public/logos/ipl/	Trademarks of their owners; used only to identify the teams
Wikipedia player infoboxes	Test, ODI and T20I career stats on the Showdown cards; roles, batting hands and career years for Who Am I?	CC BY-SA 4.0
Player photos in public/players/	Showdown and Who Am I? cards	Each file's source and credit is in scripts/data/player_photos.json

Cricsheet is missing some international matches (withheld or not yet covered), so international puzzles come from Wikipedia's maintained record tables, not from Cricsheet aggregates.

Refreshing

Raw downloads live in data-raw/ (git-ignored, about 80 MB).

# 1. Cricsheet (men's matches + register)
cd data-raw/cricsheet
for f in tests odis t20s ipl; do curl -sfLO https://cricsheet.org/downloads/${f}_json.zip && unzip -qo ${f}_json.zip -d $f; done
curl -sfLO https://cricsheet.org/register/people.csv && curl -sfLO https://cricsheet.org/register/names.csv
cd ../..

# 2. Wikipedia pages: re-fetch data-raw/wiki/*.json with the parse API
#    (action=parse&prop=text|revid) and revision_dates.json with prop=revisions.

# 3. Build
python3 scripts/data/parse_cricsheet.py   # per-match / per-player-match records
python3 scripts/data/gen_ipl.py           # ranked IPL lists
python3 scripts/data/gen_wiki.py          # Wikipedia record puzzles + tournament lists
python3 scripts/data/build.py             # pick 5 items per list, shuffle, validate -> src/data/cricket.json
python3 scripts/data/gen_ballpark.py      # Cricsheet totals + worked sums + scripts/data/ballpark_manual.json -> src/data/ballpark.json
python3 scripts/data/gen_cards.py         # Showdown decks -> src/data/cards.json (add --refresh to re-fetch Wikipedia)
python3 scripts/data/gen_whoami.py        # Who Am I? bios and daily order -> src/data/whoami.json (after gen_cards.py)
npm test

build.py refuses to write a dataset with tied values, duplicate labels, a puzzle that starts already solved, or a player name still in initials form. For the last one, add the full name to scripts/data/name_overrides.json (keyed by Cricsheet identifier) after checking it.

gen_ballpark.py takes careers, records and crowds from scripts/data/ballpark_manual.json (each checked against its linked page or Statsguru before it goes in; if a fact can't be sourced, the question is reworded around one that can), works out the size-it-up sums and date spans from sourced figures (the Laws' pitch, ball and stumps; the Equator; Everest) so the arithmetic can't drift, and computes totals from Cricsheet: the IPL, and the 2011 World Cup, the most recent one Cricsheet covers in full. It deals the questions into days of five with at most two of a family, no shared topic, no question that gives another's answer away and at most one IPL question.

gen_cards.py builds one Showdown deck per format from the players listed in scripts/data/cards_players.json. Test, ODI and T20I figures are read from the career table in each player's Wikipedia infobox (cached in data-raw/wiki/players/), because Cricsheet's international coverage starts in the 2000s; the IPL deck is computed from Cricsheet. Infoboxes of current players lag a few matches behind, so each deck carries an "as of" date. Every stat is higher-wins. The script fails if a player has no column for the format or two cards have identical stats. A card gets a photo when public/players/<name-in-lower-case-with-hyphens>.jpg exists (the script lists the players without one; they show initials on the team colour), so re-run it after adding photos.

gen_whoami.py reads cards.json and the same cached infoboxes and writes src/data/whoami.json: each player's role (batter, bowler, all-rounder or wicketkeeper), batting hand, and the years of their first and last match in each format they have a card for (the infobox's debut and last-match rows; Cricsheet for the IPL). whoami_hints.py writes four hints per card from the same infobox (bowling style, birthplace, nickname, shirt number, debut and farewell opponents) and from Cricsheet line-ups (team-mates, World Cup and IPL finals, best innings and bowling figures, IPL awards). Since Cricsheet has every IPL match but only some internationals, an international match count is quoted only when it has the player's whole career in that format, and a best innings only when it equals the highest score on the card; older players get team-mates from overlapping careers in the same side. The four infoboxes with no Role row are filled in from scripts/data/whoami_overrides.json. It also sets the order each deck's cards come up as the daily, arranged so nobody is the answer twice within 14 days. It never writes cards.json.