contractions

# Export smallest unique contractions of a list of words for Espanso

from collections import defaultdict

def shortest_unique_component(word, words_count):
    if words_count is None:
        return None

    for i in range(len(word)):
        contraction = word[:i+1]
        if words_count[contraction] == 1:
            return contraction

def process_words(file_name):
    words_count = defaultdict(int)
    with open(file_name, 'r') as file:
        for line in file:
            words = line.strip().split()
            for word in words:
                for i in range(len(word)):
                    contraction = word[:i+1]
                    words_count[contraction] += 1
    return words_count if words_count else None

def main():
    file_name = input("Enter the file name containing the list of words: ")
    words_count = process_words(file_name)

    matches = []
    with open(file_name, 'r') as file:
        for line in file:
            words = line.strip().split()
            for word in words:
                contraction = shortest_unique_component(word, words_count)
                if contraction is not None and contraction != word:
                    matches.append((contraction, word))

    with open("contractions.yml", "w") as output_file:
        output_file.write("matches:\n")
        for contraction, word in matches:
            output_file.write(f" - trigger: {contraction}\n")
            output_file.write(f"   replace: {word}\n")

if __name__ == "__main__":
    main()