Python - How to make a crossword puzzle faster

Question

I have the following function, which is part of a crossword puzzle solver:

def CrosswordPossibleWords(p_words, p_cw_words):
    """For each word found in the crossword, find the possible words and keep track of the one with the minimum possible words.

    Keyword arguments:
    p_words    -- The dictionary words.
    p_cw_words -- The crossword word attributes.
    """
    l_min = 999999999
    l_min_index = -1
    l_index = 0
    l_choices = []
    for l_cw_word in p_cw_words:
        if l_cw_word[2] >= l_min_length and '-' in l_cw_word[4]:
            pattern = re.compile('^' + l_cw_word[4].replace('.', '%').replace('-', '.').upper() + '$', re.UNICODE)
            l_choice = []
            for l_word in [w for w in p_words if len(w) == len(l_cw_word[4])]:
                if re.match(pattern, l_word):
                    l_choice.append(l_word)
            l_choices.append(l_choice)
            if len(l_choice) < l_min:
                l_min_index = l_index
                l_min = len(l_choice)
        else:
            l_choices.append([])
        l_index = l_index + 1
    return (l_choices, l_min_index)

Crossword puzzles look like:

[row, col, length, direction, word]

I have '.'in a word if I cannot solve this word and '-'if I do not know this letter.

How can I make this code faster? It currently takes about 2.5 seconds. Thought to use numpy strings; since numpy seems to be 10 times faster, but I don't know anything about numpy and don’t know if I can use all current string functions with it.

Any ideas?

Answer 1

You can break your vocabulary into word lengths before calling this function, so it does not need to do this again with every call.

Answer 2

, , , - , dicts:

def CrosswordPossibleWords(p_words, p_cw_words):
    """For each word found in the crossword, find the possible words and keep track of the one with the minimum possible words.

    Keyword arguments:
    p_words    -- The dictionary words.
    p_cw_words -- The crossword word attributes.
    """
    l_min = 999999999
    l_min_index = -1
    l_index = 0
    l_choices = {}    # using dict instead of list
    for l_cw_word in p_cw_words:
        if l_cw_word[2] >= l_min_length and '-' in l_cw_word[4]:
            pattern = re.compile('^' + l_cw_word[4].replace('.', '%').replace('-', '.').upper() + '$', re.UNICODE)
                l_choice = {}  # using dict instead of list

            for l_word in [w for w in p_words if len(w) == len(l_cw_word[4])]:
                if re.match(pattern, l_word):

                    l_choice[l_word]=None

            l_choices[l_choice]=None

            if len(l_choice) < l_min:  ##
                l_min_index = l_index  ## Get rid of this.
                l_min = len(l_choice)  ##
        else:
            l_choices.append([])    # why append empty list?
        l_index = l_index + 1
        l_choices=list(l_choices.keys())   # ...you probably need the list again...
    return (l_choices, l_min_index)