python爬虫及语言处理练习
家里长辈想要一款定制的月份工作表制作软件，试着稍微编一编，目前完成度如下（至八月份）。
进度节点清单：
目前进度：
目前时间显示：
时间表生成：
周末时间分析：
政府政策爬取及处理：
政府政策节日分析：
政府政策调休分析：
人员信息录入：
特殊节假日录入：
人员信息及特殊节假日查询：
目前代码总览：
def now(): #当前时间
import time
todayyear = time.localtime()[0] #目前年份
todaymonth = time.localtime()[1] #目前月份
todayday = time.localtime()[2] #目前日
return "目前时间:\n"+str(todayyear)+"年"+str(todaymonth)+"月"+str(todayday)+"日"
def allyearcalendar(): #全年日历显示
import time
import calendar
todayyear = time.localtime()[0] # 目前年份
return calendar.calendar(todayyear)
def mcalendar(year, month): #全月日历显示
import time
import calendar
return calendar.month(year, month)
def weekend(todayyear, todaymonth): #周末分析
weekend1 = []
import calendar
a = calendar.monthcalendar(todayyear, todaymonth)
for i in range(0,len(a)):
for j in range(5,7):
if a[i][j] !=0:
weekend1.append(a[i][j])
for i1 in range(0, len(weekend1)):
day = str(todaymonth)+"月"+str(weekend1[i1])+"日"
if i1 == 0:
print("在"+day,sep=",",end =",")
if i1 != len(weekend1)-1 and i1 != 0:
print(day, sep=",", end=",")
if i1 == len(weekend1)-1:
print(day+"为周末", sep=",", end="。\n")
return weekend1
def holiday(per, last, expectyear): #国务院节假日通知爬取
import requests
import re
from fake_useragent import UserAgent #反爬虫头部信息模拟
ua = UserAgent()
headers = {'User-Agent': ua.ie}
web = 'https://baike.baidu.com/item/国务院办公厅关于' + expectyear + '年部分节假日安排的通知'
result = requests.get(url=web,headers=headers)
reg2 = r'(.*)'
reg_ques = re.compile(reg2)
a1 = reg_ques.findall(result.text)
iper= 0
ilast = 0
for i in range(0, len(a1)):
if per in a1[i]:
iper = i
if last in a1[i]:
ilast = i
del a1[ilast:len(a1)]
del a1[0:iper+1]
return a1
def holidaydesign1(holiday, ChineseHolidaylist, todayyear ):#国务院节假日信息处理（假日）
special = []
class specialtime(object): # 特殊时间点结构化(节日)
def __init__(self, Name, Year, Month, Day, Num):
self.Name = name
self.Year = year
self.Month = month
self.Day = day
self.Num = num
def check(self):
print("假期名称为: %s,假期开始年为: %s," % (self.Name,self.Year),end='')
print(" , 假期开始月为: %s ,假期开始日为: %s ,假期放多少天：%s天。" % (self.Month, self.Day, self.Num))
import jieba
name = " "
year = " "
month = " "
day = " "
num = " "
Truenum = []
Truenum2 = []
Trueinfo = []
ChineseNum = ['一', '二', '三', '四', '五', '六', '七', '八', '九', '十']
for i1 in range(0, len(holiday)):
for i2 in range(0, len(ChineseNum)):
if ChineseNum[i2] in holiday[i1]:
Truenum.append(i1)
[Truenum2.append(i) for i in Truenum if not i in Truenum2] #查重
for i in range(0,len(Truenum2)):#节日处理
Trueinfo.append(holiday[Truenum2[i]])
content = Trueinfo[i]
seg_list = jieba.lcut(content, cut_all=False) #自然语言处理
judge = 0
judge2 = 0
todayyearsave = todayyear
for i1 in range(0, len(seg_list)):
for i2 in range(0, len(ChineseHolidaylist)):
if ChineseHolidaylist[i2] in seg_list[i1] and judge == 0:
name = seg_list[i1]
save1 = seg_list[i1]
judge = 1
if '：' in seg_list[i1] and judge == 1:
save2 = i1
judge = 3
if '年' in seg_list[i1] and judge2 == 0:
todayyearsave = seg_list[i1-1]
judge2 = 1
if '月' in seg_list[i1] and judge == 3:
save22 = seg_list[i1-1]
judge = 4
if '日' in seg_list[i1] and judge == 4:
save23 = seg_list[i1-1]
judge = 5
if '放假' in seg_list[i1] and judge == 5:
save3 = i1
judge = 6
if '共' in seg_list[i1] and judge == 6:
save4 = seg_list[i1+1]
special.append(save1)
name = save1
year = todayyearsave
month = save22
day = save23
num = save4
special[len(special) - 1] = specialtime(name, year, month, day, num)
break
if '周末' in seg_list[i1] and judge == 6:
special.append(save1)
name = save1
year = todayyearsave
month = save22
day = save23
num = "1（不加周末）"
special[len(special) - 1] = specialtime(name, year, month, day, num)
break
return special
def holidaydesign2(holiday, ChineseHolidaylist, todayyear): # 国务院节假日信息处理（调休）
special = []
class worktime(object): # 特殊时间点结构化(节日)
def __init__(self, Year, Month, Day):
self.Year = year
self.Month = month
self.Day = day
def check(self):
print("调休开始年为: %s" % (self.Year), end='')
print(" , 调休开始月为: %s ,调休开始日为: %s" % (self.Month, self.Day))
import jieba
year = " "
month = " "
day = " "
Truenum = []
Truenum2 = []
Trueinfo = []
ChineseNum = ['一', '二', '三', '四', '五', '六', '七', '八', '九', '十']
for i1 in range(0, len(holiday)):
for i2 in range(0, len(ChineseNum)):
if ChineseNum[i2] in holiday[i1]:
Truenum.append(i1)
[Truenum2.append(i) for i in Truenum if not i in Truenum2] # 查重
for i in range(0, len(Truenum2)): # 节日处理
Trueinfo.append(holiday[Truenum2[i]])
content = Trueinfo[i]
seg_list = jieba.lcut(content, cut_all=False) # 自然语言处理
judge = 0
judge2 = 0
todayyearsave = todayyear
for i1 in range(0, len(seg_list)):
if '年' in seg_list[i1] and judge2 == 0:
todayyearsave = seg_list[i1 - 1]
judge2 = 1
if '年' in seg_list[i1] and judge2 == 1:
todayyearsave = seg_list[i1-1]
if '。' in seg_list[i1] and judge == 0:
judge = 1
if '月' in seg_list[i1] and judge == 1:
save1 = seg_list[i1-1]
judge = 2
if '日' in seg_list[i1] and judge == 2:
save2 = seg_list[i1-1]
judge = 3
if judge == 3:
special.append(str(i))
year = todayyearsave
month = save1
day = save2
special[len(special)-1] = worktime(year, month, day)
judge = 1
judge2 = 0
if '上班' in seg_list[i1]:
break
return special
def extratimeadd1(names,years): #自行添加时间点(节日)
special = []
class specialtime(object): # 特殊时间点结构化(节日)
def __init__(self, Name, Year, Month, Day):
self.Name = name
self.Year = year
self.Month = month
self.Day = day
def check(self):
print("额外假期名称为: %s,额外假期年为: %s," % (self.Name, self.Year), end='')
print("额外假期月为: %s ,额外假期日为: %s " % (self.Month, self.Day))
name = names
year = str(years)
month = ''
day = ''
judge = 1
while judge == 1:
month = input('额外假期月为：')
day = input('额外假期日为：')
special.append(name+'1')
special[len(special)-1] = specialtime(name, year, month, day)
judge = int(input("输入1继续操作："))
return special
def numberinfo(): #人员信息录入
class number(object):
def __init__(self, Name, Info1, Info2, Info3, Info4, Info5):
self.Name = name
self.Info1 = info1
self.Info2 = info2
self.Info3 = info3
self.Info4 = info4
self.Info5 = info5
def check(self):
print("人员信息为: %s,信息1为: %s" % (self.Name, self.Info1), end='')
print(" , 信息2为: %s ,信息3为: %s ,信息4为:%s ,信息5为:%s" % (self.Info2, self.Info3, self.Info4, self.Info5))
name = ""
info1 = ""
info2 = ""
info3 = ""
info4 = ""
info5 = ""
special = []
print("开始人员信息录入")
judge = 1
while judge == 1:
name = input("输入人员姓名：")
info1 = input("输入人员信息1：")
info2 = input("输入人员信息2：")
info3 = input("输入人员信息3：")
info4 = input("输入人员信息4：")
info5 = input("输入人员信息5：")
special.append(len(special))
special[len(special)-1] = number(name, info1, info2 , info3 , info4 , info5)
special[len(special)-1].check()
judge = int(input("输入1继续操作："))
print('录入完成')
return special
todayyear = int(input("输入工作表年份:")) #期望年份
todaymonth = int(input("输入工作表月份:")) #期望月份
extraname = '调休'
print(mcalendar(todayyear,todaymonth))
ChineseHolidaylist = ['元旦', '春节', '清明节', '劳动节', '端午节', '中秋节', '国庆节', '妇女节', '青年节', '儿童节', '中国人民解放军建军纪念日']
Per = '假日安排公布'
Last = '节假日期间，各地区、各部门要妥善安排好值班和安全、保卫等工作，遇有重大突发事件，要按规定及时报告并妥善处置，确保人民群众祥和平安度过节日假期'
print(now())
Holiday = holiday(Per, Last, str(todayyear))
Weekend = weekend(todayyear, todaymonth)
Holidayinfo = holidaydesign1(Holiday, ChineseHolidaylist, todayyear)
Workinfo = holidaydesign2(Holiday, ChineseHolidaylist, todayyear)
print(Weekend)
Numberinfo = numberinfo()
Extratimeadd1 = extratimeadd1(extraname, todayyear)
for i in range(0, len(Holiday)):
print(Holiday[i])
for i in range(0, len(Holidayinfo)):
Holidayinfo[i].check()
for i in range(0, len(Workinfo)):
Workinfo[i].check()
for i in range(0, len(Numberinfo)):
Numberinfo[i].check()
for i in range(0, len(Extratimeadd1)):
Extratimeadd1[i].check()小结：
大致做到节点4了，长辈也不着急用，再加上大连疫情复发，做的便比较慢。
其实已经半个月没有再写了(´-ω-`)
编程思路就两步：
能够获取制作工作表需要用到的数据的代码，以及通过用户操作能处理相关数据生成符合要求的工作表的代码。