Build a Sudoku Game and Learn State Management in React

Sitian Liu
5 min readMar 6, 2018


Sudoku is one of my favorite logic-based number-placement puzzles. Due to the lack of quality Sudoku apps on the web and app stores, I decided to build my own version. It is packed with user-friendly and beginner-friendly features that make the game much more enjoyable. You can play it here. In this post, I will talk about designing, managing and working with the game’s state for the main UIs. You will read about how the right state design helps to keep the code clean and pleasant to work with. I hope it will help you reason, structure and manage an application’s state.

Game Rules

Fill a 9×9 grid with digits so that each column, each row, and each of the nine 3×3 squares that compose the grid contains all of the digits from 1 to 9.

peers: cells belonging to the same column, row or square

The Game

Game generation

Sudoku is a constraint satisfaction problem. Solving and generating Sudoku puzzles has been an explored topic(see notes). This post focuses on the state management of the game UI, but here is roughly the game generation code if you are interested.

Represent the Puzzle in 2D Array

The puzzle is represented as a 9x9 two-dimensional array. Each element of the array represents a cell’s current value and its notes:

with each cell’s object contains its current value as well as its notes.

The Overall Game State

For the game to work, it needs to be able to:

  • determine the game’s progress: know whether the game is completed and how many numbers are left for each group of peers, cells belonging to the same column, row or square.
  • track possible conflicts within each peer group: know the number of times a number is used per peer group(a complete group is one with each number used exactly once)

You can identify conflicts by comparing a cell’s numeric value with values of its 24 peers(8 per column, row or square) and determine the game’s progress by scanning the entire 9x9 puzzle. However, this approach is inefficient, verbose in code and hard to understand. To keep the implementation performant, clean and pleasant to work with, the game state will track the number of times each number used on a peer group basis. On initialization it will generate the puzzle and create count objects for all the peer groups:

The Cell Component

The cell component lights up with different colors signaling useful information about its current value, relationship to the selected cell, conflicts, and notes based on the game’s current state:

Computing the Cell Component’s Inputs

Cell component needs { value, isPeer, isSelected, sameValue, prefilled, notes, conflict } to render its correct presentation to the user. { value, notes, prefilled } are directly accessed from the cell from the puzzle’s 2D array. However, determining the rest of inputs will need inferences from the overall game state particularly in relation to the selected cell. Keeping which cell is selected in the game state is rather straightforward:

component methods for getting selected cell and setting the selected coordinates

With the getSelectedCell() method in the Game component, I can now infer the values of { isSelected, sameValue, isPeer, conflict} by comparing the current cell with the selected:

The Number Control Component

Each number control component allows a user to assign the corresponding numerical value to the selected cell. The presentation itself also visualizes the assignment progress of the number.

The NumberControl component needs:

  • a callback that would update the selected cell’s value on user clicks
  • a computed value of the number’s progress

fillNumber(number), a Game component’s function to update the selected cell’s value is written to create handlers of user clicks on the NumberControl components:

The number’s progress is computed by the following two steps:

  1. Per column, row and square groups, get the number of groups already have the number filled
  2. Take the minimum of the three values as the progress for the number

Now the Game component has what it needs to render the NumberControl components for the numbers, 1 to 9:

The Tool Components

In order to undo or redo an update to the game, it must keep track of its update history. An array object is used to keep a linear history of user’s updates to the game. The code is also extended to keep track of the history of board updates:

initialize the game state with a history array and historyOffset

With a history array and its current historyOffset in hand, redo/undo updates are just decrement/increment updates to historyOffset of the array:

Erasing a cell is as simple as updating its current value to null. Hinting at an unfilled cell is also trivial since the correct value can be obtained from the solution in the state:

For erase and hint, I created erasedSelected and filledSelectedWithSolution respecitively

Now I have everything I need to render the action elements and pass them the correct update handlers:

Now a user has a proper Sudoku game to play with!

Wrapping Things Up

This is my thought process of implementing and designing the state for the game’s main components, the parts worth talking about. The Sudoku application benefited tremendously from having a good state design. The toughest thing to get right is state management. Planning the management of the state is crucial to your success in building maintainable applications. Make sure to spend plenty of time designing the application’s state at a high level before diving into the implementation details.

Related Materials



Sitian Liu

Chasing great designs, delightful user experiences and happy developer experiences. Find me at