How-to
Build and judge solo progress
Hold solo progress as a new value, place and move tiles, and check before drawing or finishing.
Keep immutable progress
TilePlay stores the original bag, returned tiles, draw position, hand and grid. placeFromHand, moveOnTable, liftToHand, swapWithHand, draw and trade return progress values. Keep the return value; do not mutate the tiles map to imitate those operations.
import {loadTileWordsFromModule} from "@johnmorrisdotca/kumimoji/words";
import {deal,placeFromHand,judgeWithWords,encodeGrid} from "@johnmorrisdotca/kumimoji";
const words = await loadTileWordsFromModule("english");
let play = deal("catdog",3);
for (let col=0;col<3;col++) play = placeFromHand(play,0,`0,${col}`);
console.log(encodeGrid(play.tiles),judgeWithWords(play.tiles,words).sound);cat trueAsk before drawing or finishing
mayDraw and isFinished take a native GridVerdict. Every run must be an accepted word and the grid connected; rules can also read diagonals. Use judgeWithWords after changes. A full hand, an invalid grid or an empty bag can change which action is offered. Wild tiles need assigned readings before their runs can be judged.
Keep coordinates and encoding separate
squareAt names integer row/column coordinates, including negatives. encodeGrid trims the occupied bounds and writes a canonical relative shape; it does not preserve how far the table was panned. decodeGrid returns null for an invalid encoded grid. Views covers rotation and zoom without changing rules.