weibo.com-crawler/Processor/CommentProcessor.py

def process_comment(res_data):
    result = {}

    def add_comment(root, father_user_id, user_id, comment):
        if father_user_id in root.keys():
            root[father_user_id]['comments'][user_id] = comment
            return root
        for key in root.keys():
            comments = root[key]['comments']
            new_comments = add_comment(
                comments, father_user_id, user_id, comment)
            if new_comments:
                root[key]['comments'] = new_comments
                return root
        return False

    for data in res_data:
        comment_id = data['id']
        host_user_id = data['user']['screen_name']
        if '//' in data['text_raw']:
            texts = data['text_raw'].split('//')
            father_id = ''
            for text in texts[::-1]:
                text = text.split(':')
                if len(text) == 1:
                    content = text[0] if text[0] == '' else '快转微博'
                    if father_id == '':
                        result[host_user_id] = {
                            'comment_id': comment_id, 'content': content, 'comments': {}
                        }
                    else:
                        result = add_comment(result, father_id, host_user_id, {
                                             'content': content,
                                             'comment_id': comment_id,
                                             'comments': {}
                                             })
                else:
                    user_id = text[0][1:]
                    content = text[1] if text[1] != '' else '快转微博'
                    if father_id == '':
                        if user_id not in result.keys():
                            result[user_id] = {
                                'content': content,
                                'comments': {}
                            }
                        father_id = user_id
                    else:
                        result = add_comment(result, father_id, user_id, {
                                             'content': content,
                                             'comments': {}
                                             })
                        father_id = user_id
        else:
            content = data['text_raw']
            result[host_user_id] = {
                'comment_id': comment_id, 'content': content, 'comments': {}
            }
    return result