def getText():
txt = open(r"C:\Users\CYG\Desktop\keyword_text.txt", "r",
encoding='utf-8').read()
# print(txt)
txt = txt.lower()
pattern = re.compile(
r'\t|\n|\.|-|:|;|\)|\(|\?|\s[0-9]{1,2}\s|"') # 定义正则表达式匹配模式（空格等）
txt = re.sub(pattern, ' ', txt)
# print(string_data)
text = txt.split()
res = collections.Counter(text)
word_counts_top = res.most_common(1500)
# print(word_counts_top)
mysql = Mysql(dbname='736')
for sort_name, num in word_counts_top:
if len(sort_name) >= 3:
try:
print(sort_name, num)
insert_data = []
# pattern = re.compile("[^a-zA-Z0-9]")
# sort_name = pattern.sub('',sort_name)
filter_list = [
'con', 'product', 'products', 'solutions', 'solution',
'blog', 'css', 'js', 'images', 'infoimages', 'fonts',
'static', 'ajax'
]
if sort_name not in filter_list and len(
sort_name) >= 3: #注意一些特殊语言的掺杂其中 所以加了个长度判定
insert_data.append({'sort_name': sort_name})
insert_data.append({'num': num})
mysql.table('sort_name').insert(insert_data)
except BaseException as e:
print(e)
def main():
# 创建页码队列
redisqueue = RedisQueue('keyword_url')
qsize = redisqueue.qsize()
mysql = Mysql()
for i in crawl_thread_list:
download_thread = CrawlThread(i, redisqueue, qsize, time.time(), mysql)
download_thread.start()
print(qsize)
time.sleep(20)
print('采集结束')
pass
def getTextMysql():
txt = open(r"D:\pydata\work\scrapy_data\www.ogniskanadziei.pl.txt",
"r",
encoding='utf-8').read()
# print(txt)
txt = txt.lower()
pattern = re.compile(
r'\t|\n|\.|-|:|;|\)|\(|\?|\s[0-9]{1,2}\s|"') # 定义正则表达式匹配模式（空格等）
txt = re.sub(pattern, ' ', txt)
# print(string_data)
text = txt.split()
text_all = collections.Counter(text)
# print(text_all)
word_counts_top = text_all.most_common()
# print(word_counts_top)
mysql = Mysql(dbname='scrapy')
for sort_name, num in word_counts_top:
if len(sort_name) >= 0:
try:
# print(sort_name,num)
insert_data = []
# pattern = re.compile("[^a-zA-Z0-9]")
# sort_name = pattern.sub('',sort_name)
res = mysql.table('sort_name').where([{
'sort_name': ['=', sort_name]
}]).find()
if res:
insert_data.append({'num': int(res[2]) + int(num)})
mysql.table('sort_name').where([{
'sort_name': ['=', sort_name]
}]).update(insert_data)
else:
insert_data.append({'sort_name': sort_name})
insert_data.append({'num': num})
mysql.table('sort_name').insert(insert_data)
except BaseException as e:
print(e)
#物料词和非物料词扇形图
def bar3(self):
#（数量+id+物料名字）
temp_x = []
y = []
num_tupple = self.get_material_bie()
for i in num_tupple:
temp_x.append(i[2])
y.append(i[0])
x = tuple(temp_x)
print(x, y)
plt.rcParams['font.sans-serif'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = False
x = range(len(temp_x))
p1 = plt.bar(x, y)
plt.xticks(x, temp_x)
plt.show()
pass
if __name__ == '__main__':
mysql = Mysql(dbname='industro')
data = Data(mysql)
# data.bar3()
# data.pie_show()
if data.bie_or_bar == 1:
data.pie_show()
elif data.bie_or_bar == 2:
data.pie_show2()
else:
data.bar_show()
print('结束执行')
from mysql_class import Mysql
mysql = Mysql('localhost', 'root', 'password', 'ex01')
# param=str(input("情输入一个名字:"))
sql = 'insert into students(name) values(%s);'
# sql='update students set name=%s where id=1'
# sql1='delete from students where id=1'
# mysql.excute_one(sql,param)
# mysql.excute_one(sql1)
# sql='select name from students where id<6'
# result=mysql.get_one(sql)
result = mysql.excute_many(sql, ['xiaohua', 'xiama'])
print(result)
# 将数据传入哈希对象中, hashlib.sha1(password.encode('utf-8'))等价于m=hashlib.sha1()和
# m.update(password.encode('utf-8'))这两句;当内容太长的时候,可以使用update多次更新内容,等效更新一次长数据
# m.update(password.encode('utf-8'))
# 获取二进制
# new_password = m.digest()
# 获取16进制
new_password1 = m.hexdigest()
# print(new_password)
# print(new_password1)
# 连接数据库
mydb = Mysql('localhost', 'root', 'password', 'ex01')
# 查询用户名和密码
sql = 'select password from userinfo where name=%s'
result = mydb.get_many(sql, (name))
print(result)
if len(result) == 0:
print("用户名错误")
else:
if result[0][0] == new_password1:
print("登陆成功")
else:
print("密码错误")
from mysql_class import Mysql
import hashlib
# 用户输入
user = input("请输入名字:")
password = input("请输入密码:")
m = hashlib.sha1()
m.update(password.encode('utf-8'))
new_password = m.hexdigest()
param=[user,new_password]
mydb = Mysql('localhost', 'root', 'password', 'ex01')
sql = 'select name from userinfo'
names = mydb.get_many(sql)
# print(names)
for name in names:
if name[0] == user:
print(type(name[0]))
print(user)
print("该用户已经存在")
break
else:
sql1 = 'insert into userinfo(name,password) values(%s,%s)'
mydb.excute_one(sql1, param)
print("注册成功")
while True:
# 用户输入
name = input("请输入用户名:")
password = input("请输入密码:")
m = hashlib.sha1(password.encode('utf-8'))
# 将数据传入哈希对象中, hashlib.sha1(password.encode('utf-8'))等价于m=hashlib.sha1()和
# m.update(password.encode('utf-8'))这两句;当内容太长的时候,可以使用update多次更新内容,等效更新一次长数据
# m.update(password.encode('utf-8'))
# 获取二进制
# new_password = m.digest()
# 获取16进制
new_password1 = m.hexdigest()
# print(new_password)
# print(new_password1)
# 连接数据库
mydb = Mysql('localhost', 'root', 'password', 'ex01')
# 插入数据，插入sha1的密码,方式一：格式化的方式输入
# sql = 'insert into students(name,password) values("%s","%s")' % (name, new_password1)
# 方式二：利用excute_one()中的param输入
sql = 'insert into students(name,password) values(%s,%s)'
print(sql)
mydb.excute_one(sql,(name,new_password1))