def write_excel(lists, filename):
wb = Workbook()
ws = wb.active
ws.title = "职位信息"
ws.cell(row=1, column=1).value = '发布时间'
ws.cell(row=1, column=2).value = '工作时间'
ws.cell(row=1, column=3).value = '职位名称'
ws.cell(row=1, column=4).value = '职位ID'
ws.cell(row=1, column=5).value = '公司ID'
ws.cell(row=1, column=6).value = '职位类型'
ws.cell(row=1, column=7).value = '公司名称'
ws.cell(row=1, column=8).value = '所在城市'
ws.cell(row=1, column=9).value = '文化程度'
ws.cell(row=1, column=10).value = '所属行业'
ws.cell(row=1, column=11).value = '融资阶段'
ws.cell(row=1, column=12).value = '职位薪酬'
ws.cell(row=1, column=13).value = '公司规模'
ws.cell(row=1, column=14).value = '平均薪资'
rownum = 2
for each_item in lists:
info_list = each_item.get('joblist')
for each_job_info_obj in info_list:
ws.cell(row=rownum,
column=1).value = each_job_info_obj['formatCreateTime']
ws.cell(row=rownum, column=2).value = each_job_info_obj['workYear']
ws.cell(row=rownum,
column=3).value = each_job_info_obj['positionName']
ws.cell(row=rownum,
column=4).value = each_job_info_obj['positionId']
ws.cell(row=rownum,
column=5).value = each_job_info_obj['companyId']
ws.cell(row=rownum,
column=6).value = each_job_info_obj['positionType']
ws.cell(row=rownum,
column=7).value = each_job_info_obj['companyName']
ws.cell(row=rownum, column=8).value = each_job_info_obj['city']
ws.cell(row=rownum,
column=9).value = each_job_info_obj['education']
ws.cell(row=rownum,
column=10).value = each_job_info_obj['industryField']
ws.cell(row=rownum,
column=11).value = each_job_info_obj['financeStage']
ws.cell(row=rownum, column=12).value = each_job_info_obj['salary']
ws.cell(row=rownum,
column=13).value = each_job_info_obj['companySize']
ws.cell(row=rownum, column=14).value = toolkit.normalize(
each_job_info_obj['salary'])
rownum += 1
wb.save('d:/' + filename + '.xlsx')
logging.info('Excel生成成功!')
def write_excel(lists, filename):
wb = Workbook()
ws = wb.active
ws.title = "职位信息"
ws.cell(row=1, column=1).value = '发布时间'
ws.cell(row=1, column=2).value = '工作时间'
ws.cell(row=1, column=3).value = '职位名称'
ws.cell(row=1, column=4).value = '职位ID'
ws.cell(row=1, column=5).value = '公司ID'
ws.cell(row=1, column=6).value = '职位类型'
ws.cell(row=1, column=7).value = '公司名称'
ws.cell(row=1, column=8).value = '所在城市'
ws.cell(row=1, column=9).value = '文化程度'
ws.cell(row=1, column=10).value = '所属行业'
ws.cell(row=1, column=11).value = '融资阶段'
ws.cell(row=1, column=12).value = '职位薪酬'
ws.cell(row=1, column=13).value = '公司规模'
ws.cell(row=1, column=14).value = '平均薪资'
rownum = 2
for each_item in lists:
info_list = each_item.get('joblist')
for each_job_info_obj in info_list:
ws.cell(row=rownum, column=1).value = each_job_info_obj['formatCreateTime']
ws.cell(row=rownum, column=2).value = each_job_info_obj['workYear']
ws.cell(row=rownum, column=3).value = each_job_info_obj['positionName']
ws.cell(row=rownum, column=4).value = each_job_info_obj['positionId']
ws.cell(row=rownum, column=5).value = each_job_info_obj['companyId']
ws.cell(row=rownum, column=6).value = each_job_info_obj['positionType']
ws.cell(row=rownum, column=7).value = each_job_info_obj['companyName']
ws.cell(row=rownum, column=8).value = each_job_info_obj['city']
ws.cell(row=rownum, column=9).value = each_job_info_obj['education']
ws.cell(row=rownum, column=10).value = each_job_info_obj['industryField']
ws.cell(row=rownum, column=11).value = each_job_info_obj['financeStage']
ws.cell(row=rownum, column=12).value = each_job_info_obj['salary']
ws.cell(row=rownum, column=13).value = each_job_info_obj['companySize']
ws.cell(row=rownum, column=14).value = toolkit.normalize(each_job_info_obj['salary'])
rownum += 1
wb.save('d:/' + filename + '.xlsx')
print('Excel生成成功!')
def write_excel(output_path, lists, filename):
wb = Workbook()
ws = wb.active
ws.title = "职位信息"
ws.cell(row=1, column=1).value = '发布时间'
ws.cell(row=1, column=2).value = '工作时间'
ws.cell(row=1, column=3).value = '职位名称'
ws.cell(row=1, column=4).value = '公司名称'
ws.cell(row=1, column=5).value = '公司规模'
ws.cell(row=1, column=6).value = '所在城市'
ws.cell(row=1, column=7).value = '学历要求'
ws.cell(row=1, column=8).value = '工作经验'
ws.cell(row=1, column=9).value = '职位薪酬'
ws.cell(row=1, column=10).value = '平均薪资'
rownum = 2
for each_item in lists:
info_list = each_item.get('joblist')
for each_job_info_obj in info_list:
ws.cell(row=rownum,
column=1).value = each_job_info_obj['formatCreateTime']
ws.cell(row=rownum, column=2).value = each_job_info_obj['workYear']
ws.cell(row=rownum,
column=3).value = each_job_info_obj['positionName']
ws.cell(row=rownum,
column=4).value = each_job_info_obj['companyFullName']
ws.cell(row=rownum,
column=5).value = each_job_info_obj['companySize']
ws.cell(row=rownum, column=6).value = each_job_info_obj['city']
ws.cell(row=rownum,
column=7).value = each_job_info_obj['education']
ws.cell(row=rownum, column=8).value = each_job_info_obj['workYear']
ws.cell(row=rownum, column=9).value = each_job_info_obj['salary']
ws.cell(row=rownum, column=10).value = toolkit.normalize(
each_job_info_obj['salary'])
rownum += 1
wb.save(output_path + filename + '.xlsx')
print('Excel生成成功!')
def write_csv(output_path, lists, filename):
f = open(output_path + filename + '.csv', 'wt', encoding='utf-8')
f.write('发布时间,工作时间,职位名称,公司名称,公司规模,所在城市,学历要求,工作经验,单位薪酬,平均薪资')
f.write('\n')
for each_item in lists:
info_list = each_item.get('joblist')
for each_job_info_obj in info_list:
f.write(each_job_info_obj['formatCreateTime'] + ',' +
each_job_info_obj['workYear'] + ',' +
each_job_info_obj['positionName'] + ',' +
str(each_job_info_obj['positionId']) + ',' +
str(each_job_info_obj['companyId']) + ',' +
each_job_info_obj['positionAdvantage'] + ',' +
each_job_info_obj['companyFullName'] + ',' +
each_job_info_obj['city'] + ',' +
each_job_info_obj['education'] + ',' +
each_job_info_obj['industryField'] + ',' +
each_job_info_obj['financeStage'] + ',' +
each_job_info_obj['salary'] + ',' +
str(each_job_info_obj['companySize']) + ',' +
str(toolkit.normalize(each_job_info_obj['salary'])))
f.write('\n')
f.flush()
f.close()