generating TTT game tree in lua - algorithm

Question

I am attempting to write a tic-tac-toe game in lua, and plan on using the minimax algorithm to decide non-human moves. The first step in this involves generating a tree of all possible board states from a single input state. I am trying to recursively do this, but cannot seem to figure out how. (I think) I understand conceptually how this should be done, but am having trouble implementing it in lua.
I am trying to structure my tree in the following manner. Each node is a list with two fields.
{ config = {}, children = {} }
Config is a list of integers (0,1,2) that represent empty, X, and O and defines a TTT board state. Children is a list nodes which are all possible board states one move away from the current node.
Here is my function that I currently have to build the game tree
function tree_builder(board, player)
supertemp = {}
for i in ipairs(board.config) do
--iterate through the current board state.
--for each empty location create a new node
--representing a possible board state
if board.config[i] == 0 then
temp = {config = {}, children = {}}
for j in ipairs(board.config) do
temp.config[j] = board.config[j]
end
temp.config[i] = player
temp.children = tree_builder(temp, opposite(player))
supertemp[i] = temp
end
end
return supertemp
end
The function is called in the following manner:
start_board = {config = {1,0,0,0}, children = {} }
start_board.children = tree_builder(start_board, 1)
When I comment out the recursive element of the function (the line "temp.children = builder(temp, opposite(player))") and only generate the first level of children. the output is correct. When called via code that is conceptually identical to (I am using love2D so formatting is different):
for i in pairs(start_board.children) do
print (start_board.children[i].config)
The three children are:
1,1,0,0
1,0,1,0
1,0,0,1
However, once I add the recursive element, the following is output for the same three children
1,1,2,1
1,1,2,1
1,1,2,1
I have been searching online for help and most of what I have found is conceptual in nature or involves implementation in different languages. I believe I have implemented the recursive element wrongly, but cannot wrap my head around the reasons why.

Answer 1

Don't understand what opposite(player) means in temp.children = tree_builder(temp, opposite(player)).
Notice that a recursion need an end condition.
This is my solution under your structure:
local COL = 3
local ROW = 3
local function printBoard( b )
local output = ""
local i = 1
for _,v in ipairs(b.config) do
output = output .. v .. ( (i % COL == 0) and '\n' or ',' )
i = i + 1
end
print( output )
end
local function shallowCopy( t )
local t2 = {}
for k,v in pairs(t) do
t2[k] = v
end
return t2
end
local MAX_STEP = COL * ROW
local ING = 0
local P1 = 1
local P2 = 2
local TIE = 3
local STATUS = { [P1] = "P1 Win", [P2] = "P2 Win", [TIE] = "Tied" }
local start_board = { config = {P1,0,0,0,0,0,0,0,0}, children = {} }
local function checkIfOver( board, step )
local config = board.config
local over = false
--check rows
for i=0,ROW-1 do
over = true
for j=1,COL do
if 0 == config[i*COL+1] or config[i*COL+j] ~= config[i*COL+1] then
over = false
end
end
if over then
return config[i*COL+1]
end
end
--check cols
for i=1,COL do
over = true
for j=0,ROW-1 do
if 0 == config[i] or config[i] ~= config[i+COL*j] then
over = false
end
end
if over then
return config[i]
end
end
--check diagonals
if config[1] ~= 0 and config[1] == config[5] and config[5] == config[9] then
return config[1]
end
if config[3] ~=0 and config[3] == config[5] and config[5] == config[7] then
return config[3]
end
if step >= MAX_STEP then
return TIE
else
return ING
end
end
local function treeBuilder( board, step )
--check the game is over
local over = checkIfOver( board, step )
if over ~= ING then
printBoard( board )
print( STATUS[over], '\n---------\n' )
return
end
local child
local childCfg
for i,v in ipairs(board.config) do
if 0 == v then
child = { config = {}, children = {} }
childCfg = shallowCopy( board.config )
childCfg[i] = (step % 2 == 0) and P1 or P2
child.config = childCfg
table.insert( board.children, child )
treeBuilder( child, step + 1 )
end
end
end
treeBuilder( start_board, 1 )