pinyin-heteronym-combination.py

import re
from pypinyin import pinyin, Style


def _recursive(a_list, b_list=None):
    # a: [x] or x
    if not isinstance(a_list, (list, tuple)):
        a_list = [a_list]

    # b: [x] or None
    b_list = b_list or []

    # b: [x, y] or [[x, y], [a, b, c]]
    if b_list and isinstance(b_list[0], (list, tuple)):
        if len(b_list) > 1:
            b_list = get_all_cp(*b_list)
        else:
            # avoid tail blank list collapse all output
            b_list = b_list[0]

    # b: [x] or []
    if not b_list:
        for a in a_list:
            yield a
        return

    # a: [x] or []
    if not a_list:
        for b in b_list:
            yield b
        return

    # a: [x]
    # b: [y]
    for a in a_list:
        for b in b_list:
            if isinstance(b, tuple):
                yield (a, *b)
            else:
                yield a, b


def get_all_cp(*args):
    if not args:
        return []
    return list(_recursive(args[0], args[1:]))


def to_pinyin_list(text):
    if not text:
        return [text]
    zh = re.compile(r'[\u4e00 -\u9fa5]+')
    if zh.match(text):
        results = pinyin(text, heteronym=True, style=Style.NORMAL, strict=False)
        print('=>', results)
        return list(get_all_cp(*results))
    return [text]


print(to_pinyin_list('单白重'))

first = "单"
last = "白重"
first_pinyin_list = [''.join(x) for x in to_pinyin_list(first)]
last_pinyin_list = [''.join(x) for x in to_pinyin_list(last)]
first_last_pinyin_list = get_all_cp(first_pinyin_list, last_pinyin_list)
print(first_last_pinyin_list)

Output:

=> [['dan', 'chan', 'shan'], ['bai', 'bo'], ['zhong', 'chong', 'tong']]
[('dan', 'bai', 'zhong'), ('dan', 'bai', 'chong'), ('dan', 'bai', 'tong'), ('dan', 'bo', 'zhong'), ('dan', 'bo', 'chong'), ('dan', 'bo', 'tong'), ('chan', 'bai', 'zhong'), ('chan', 'bai', 'chong'), ('chan', 'bai', 'tong'), ('chan', 'bo', 'zhong'), ('chan', 'bo', 'chong'), ('chan', 'bo', 'tong'), ('shan', 'bai', 'zhong'), ('shan', 'bai', 'chong'), ('shan', 'bai', 'tong'), ('shan', 'bo', 'zhong'), ('shan', 'bo', 'chong'), ('shan', 'bo', 'tong')]


=> [['dan', 'chan', 'shan']]
=> [['bai', 'bo'], ['zhong', 'chong', 'tong']]
[('dan', 'baizhong'), ('dan', 'baichong'), ('dan', 'baitong'), ('dan', 'bozhong'), ('dan', 'bochong'), ('dan', 'botong'), ('chan', 'baizhong'), ('chan', 'baichong'), ('chan', 'baitong'), ('chan', 'bozhong'), ('chan', 'bochong'), ('chan', 'botong'), ('shan', 'baizhong'), ('shan', 'baichong'), ('shan', 'baitong'), ('shan', 'bozhong'), ('shan', 'bochong'), ('shan', 'botong')]