def fix_json(name): data = iotools.load_raw(name) lines = data.split("\n") if lines[-1] == "": del lines[-1] data = "[%s]" % ",".join(lines) iotools.save_raw(data, name) data = iotools.load_json(name) iotools.save_json(data, name)
#! /usr/bin/env python -u # coding=utf-8 import re from utils import iotools __author__ = 'xl' data = iotools.load_raw("../data/raw/glossary.txt") ret = {} for block in re.findall(r"^::: (.+?)$(.*?)(?=\n:::)$", data, re.MULTILINE | re.DOTALL): ret[block[0].strip()]=block[1].strip() print len(ret)