def get_delete():
username = request.forms.get('username')
passwd = request.forms.get('passwd')
if check_login(username , passwd):
sql.main(9 , **{'name':username})
return '<p><b>User has been deleted from database!</b></p>'
else:
return '<p><b>Password wrong or User wrong , delete user Failed!</p>'
def blog_delete(md5url):
'''
删除博文，但是不删除对于博文的评价，评价是很重要的信息数据，不能随便的删除
'''
import json
response.set_header('Access-Control-Allow-Origin','*')
sql.main(10 , **{'md5url' : md5url})
return json.dumps({'delete': True} , ensure_ascii = False , skipkeys = True)
def blog_comment():
'''
这里的comment的数据制定了我们对博文的评价成程度
1,2,3,4,5几个等级
'''
import json
response.set_header('Access-Control-Allow-Origin','*')
data = eval(request.body.readlines()[0].decode('utf8'))
sql.main(7 , **{'md5url' : data['md5url'] , 'grade' : data['grade']})
return json.dumps({'done': True} , ensure_ascii = False , skipkeys = True)
def crawl_sample(filename, ans, flag):
soup = BeautifulSoup(ans, 'lxml')
# 提取正文，富文本化
main = soup.article
feature = {}
if main == None:
main = soup.find(id='article_content')
if main == None:
print('目标URL不是博文页面')
return
else:
# 此时进入csdn的另外的主题里面
feature['md5url'] = filename
feature['size'] = len(main.get_text())
feature['number_like'] = int(soup.find(id='btnDigg').dd.string)
feature['number_reader'] = int(''.join(
filter(str.isdigit,
soup.find(class_='link_view').string)))
feature['number_code'] = len(main.find_all(name='pre'))
feature['number_photo'] = len(main.find_all(name='img'))
feature['number_link'] = len(main.find_all(name='a'))
else:
feature['md5url'] = filename
feature['size'] = len(main.get_text())
feature['number_like'] = int(
soup.find(class_='left_fixed').find(class_='txt').string)
feature['number_reader'] = int(
soup.find(class_='btn-noborder').find(class_='txt').string)
# feature['number_comment'] = int(soup.find(class_ = 'load_comment').span.string)
# 之后的 grade and analyse_grade 都是用户的添加和我们的自然语言文本分析的结果
feature['number_code'] = len(main.find_all(name='pre'))
feature['number_photo'] = len(main.find_all(name='img'))
# feaure 存储进入数据库
print('样本入数据库结束')
noise = main.find_all(class_='article_bar clearfix')[0]
noise.decompose() # 删除标签
feature['number_link'] = len(main.find_all(name='a'))
if flag != 0:
sql.main(6, **feature) # 只有不是测试网而言的网页才一会进入aifeature表中存储对应的网页样本信息
else:
return feature # 信息返还给我们的后端的对应路由
k = main.prettify().replace('"', '\'')
# 为了高亮代码块，我们需要将csdn的css样式表插入进来，head头需要包含
ans = "<html><head><meta charset='utf8'></head><body>" + k + '</body></html>'
pages = {'md5url': filename, 'content': ans}
sql.main(4, **pages)
print('文本入数据库结束')
feature['blog_name'] = soup.title.string
return feature
def loaddataset(self, filename):
# 加载数据集
save = sql.main(5)
self.datamat = [] # 数据集
self.classlabels = [] # 标志集
for line in save:
self.datamat.append(
[line[1], line[2], line[3], line[5], line[6], line[7], 1.0])
self.classlabels.append(line[4])
self.datamat = np.mat(self.datamat)
self.nSampNum, self.nSampDim = np.shape(self.datamat)
self.nSampDim -= 1 # 偏置值不进行计算
def load_data_for_grade(inx = None):
'''
该函数从数据库中抽取所有的样本空间数据，将信息规整病返还给监督学习组件去学习
'''
save = []
if inx == None:
save = sql.main(5) # 获取全部的样本信息
else:
inx.append(0)
save = [inx]
print('save' , save)
# data
data = []
for i in save:
# 数值型变量离散化
p = []
# size 离散化
if 0 <= i[1] < 1000 : p.append(1)
elif 1000 <= i[1] < 2000 : p.append(2)
elif 2000 <= i[1] < 3000 : p.append(3)
elif 3000 <= i[1] < 5000 : p.append(4)
else : p.append(5)
# number_reader
if 0 <= i[2] < 200 : p.append(1)
elif 200 <= i[2] < 400 : p.append(2)
elif 400 <= i[2] < 600 : p.append(3)
elif 600 <= i[2] < 800 : p.append(4)
elif 800 <= i[2] < 1700 : p.append(5)
else : p.append(6)
# number_like
if i[3] == 0 : p.append(1)
elif 1 <= i[3] <= 3 : p.append(2)
else : p.append(3)
# number_code
if i[5] == 0 : p.append(1)
elif 1 <= i[5] <= 3 : p.append(2)
else : p.append(3)
# number_photo
if i[6] == 0 : p.append(1)
elif 1 <= i[6] <= 2 : p.append(2)
elif 3 <= i[6] <= 4 : p.append(3)
else : p.append(4)
# number_link
if i[7] == 0 : p.append(1)
elif 1 <= i[7] <= 3 : p.append(2)
else : p.append(3)
# grade --++
p.append(int(i[4]))
data.append(p)
label = ['content size' , 'number_reader' , 'number_like' , 'number_code' , 'number_photo' , 'number_link']
return data , label
def main(ans):
k = 20 # 默认只有10个
data = sql.main(5)
test = []
label = []
for i in data:
p = []
p.append(i[1])
p.extend(i[3:])
test.append(p)
label.append(i[2])
ansp = [
ans['size'], ans['number_like'], ans['grade'], ans['number_code'],
ans['number_photo'], ans['number_link']
]
return classify(np.array(test), ansp, label, k)
def history():
import json
response.set_header('Access-Control-Allow-Origin','*')
res = sql.main(5)
new = []
for i in res:
p = {}
p['md5url'] = i[0]
p['size'] = i[1]
p['number_reader'] = i[2]
p['number_like'] = i[3]
p['grade'] = i[4]
p['number_code'] = i[5]
p['number_photo'] = i[6]
p['number_link'] = i[7]
new.append(p)
return json.dumps(new , ensure_ascii = False , skipkeys = True)
def load_data_for_reader(inx = None):
'''
该函数一样使用决策树的代码，只不过改变我们的决策的目标，决策目标更换成对应的阅读量
'''
save = []
if inx == None : save = sql.main(5)
else:
save = [inx]
print(save)
data = []
for i in save:
p = []
# size 离散化
if 0 <= i[1] < 1000 : p.append(1)
elif 1000 <= i[1] < 5000 : p.append(2)
else : p.append(3)
# number_like
if i[3] == 0 : p.append(1)
elif 1 <= i[3] < 5 : p.append(2)
else : p.append(3)
# grade
if 0 <= i[4] < 60 : p.append(1)
elif 60 <= i[4] < 80 : p.append(2)
else : p.append(3)
# number_code
if i[5] == 0 : p.append(1)
elif 1 <= i[5] <= 3 : p.append(2)
else : p.append(3)
# number_photo
if i[6] == 0 : p.append(1)
elif 1 <= i[6] <= 3 : p.append(2)
else : p.append(3)
# number_link
if i[7] == 0 : p.append(1)
elif 1 <= i[7] <= 3 : p.append(2)
else : p.append(3)
# number_reader
if 0 <= i[2] < 500 : p.append(1)
elif 500 <= i[2] < 1000 : p.append(2)
else : p.append(3)
data.append(p)
label = ['content size' , 'number_reader' , 'number_like' , 'number_code' , 'number_photo' , 'number_link']
return data , label
def check_login(username , passwd):
print(username , passwd)
save = sql.main(1 , **{'name' : username})
if save[1] == passwd : return True
else : return False
def blog_open(md5url):
'''
该模块返回一个新的富文本博文给用户
'''
response.set_header('Access-Control-Allow-Origin','*')
return sql.main(3 , **{'md5url' :md5url})[0][1].decode('utf8')
def get_signup():
username = request.forms.get('username')
passwd = request.forms.get('passwd')
sql.main(2 , **{'name' : username , 'passwd' : passwd})
return '<p><b>Create login successfully!</b></p>'
