psuedo.txt

function findTok(subVec, tok){
    for i from 0 to length(subVec)
        if subVec[i].token == tok  
            return i

    return -1
}

function genMatrix(string input, M, tokens){
    last = ""
    
    while !input.end() //populate matrix
        tok = ""

        do
            input >> tok
        while tok == "" AND !input.end()

        if tok == "" //trailing whitespaces at the end of the file will slip through the do-while
            break

        tokInd = findTok(tokens, tok) //find index of the current token

        if tokInd == -1
            newWord = {length(M), 0, tok}
            tokens.push(newWord)
            tokInd = length(M)
            M.push([])
            
        parentVec = findTok(tokens, last)

        if parentVec != -1
            ind = inVec(M[parentVec], tok)

            if ind != -1
                M[parentVec][ind].prob++
            else
                M[parentVec].push({1.0, tok})

            tokens[parentVec].occurrences++
        
        last = tok

    for i from 0 to length(M)   //calculate transition probabilites
        for j from 0 to length(M[i])
            M[i][j].prob = M[i][j].prob / tokens[i].occurrences
}

function genText(M, tokens, size){
    first = rand(length(M)) //pick a random first word from matrix
    choice = 0
    curWord = M[first].token

    for i from 0 to size
        print(curWord)
        curIndex = findTok(tokens, curWord)
        probSum = 0.0
        choice = rand(100)

        if length(M[curIndex]) == 0 //if there are no transitions available
            for j from 0 to length(tokens)
                probsum += 100 * (tokens[j].occurrences / length(tokens))

                if choice < probsum
                    curWord = tokens[j].token
                    break
        else
            for j from 0 to length(M[curIndex])
                probsum += M[curIndex][j].prob * 100

                if choice < probsum
                    curWord = M[curIndex][j].token
                    break
    
    print(curWord)
}