Python 处理Excel数据

# coding: utf-8
import codecs
import json
import re
from collections import OrderedDict, defaultdict


import pandas as pd


FILE_PATH = u'处理的数据.xlsx'




def clean_name(name):
    name = re.sub(u'[\d|\s|省|市|区]', '', name)
    return name




df = pd.read_excel(FILE_PATH)




new_df = df.iloc[3:-2, 2:]


new_df = new_df.fillna('')


exporter = map(clean_name, df.iloc[3:-2, 0])   # 供方
importer = map(clean_name, df.iloc[1, 2:-3])    # 需方


data = defaultdict(list)
for idx, importer_name in enumerate(importer):
    for idx_2 in range(len(exporter)):
        exporter_name = exporter[idx_2]
        amount = new_df.iloc[idx_2, idx]
        data[importer_name].append(OrderedDict(
            [('name', exporter_name), ('import', amount)]))  # 进口
        data[exporter_name].append(OrderedDict(
            [('name', importer_name), ('export', amount)]))  # 出口




with codecs.open('data1.json', 'wb', 'utf-8') as F:

    json.dump(data.items(), F, ensure_ascii=False, indent=4)