使用Python轉換word文檔到JSON

需求是有一堆這樣的word文檔，要轉換成試題，供web界面使用。

#!/usr/bin/env python3
import docx
import re
import json

file = docx.Document("./2018《廉潔自律准則》知識競賽試題及答案.docx")
list = []
val = None
for para in file.paragraphs:
    v = para.text.split()
    for line in v:
        items = re.compile("\.").split(line)
        if (re.match('\d', line)):
            if len(items) > 1:
                val = {'no': items[0]}
                q = ''.join([str(x) for x in items[1:]])
                key = re.search('(?<=(\(|（))\s*[A-D]*', q)
                if key != None:
                    val['k'] = key.group(0).lstrip()

                val['q'] = re.sub('(?<=(\(|（))\s*[A-D]*\s*', ' ', q)
                list.append(val)
        if (re.match('A', line)):
            if len(items) > 1:
                val['a'] = ''.join([str(x) for x in items[1:]])
        if (re.match('B', line)):
            if len(items) > 1:
                val['b'] = ''.join([str(x) for x in items[1:]])
        if (re.match('C', line)):
            if len(items) > 1:
                val['c'] = ''.join([str(x) for x in items[1:]])
        if (re.match('D', line)):
            if len(items) > 1:
                val['d'] = ''.join([str(x) for x in items[1:]])

with open('data.json', 'w') as outfile:
    json.dump(list, outfile, ensure_ascii=False)

轉換過程並不完美，因為word文檔並非標准，大約有90%左右的沒有問題，還有部分是有問題的。