def get_images(client, data):
print('[GET] getting images...')
images = []
reply = b''
for imgtag in data.find_all('img'):
src = imgtag.get('src')
src = src.replace(" ", "_")
src = src.replace("%20", "_")
request = get_from_src(src, client.addr, client)
client.sock.send(request)
headers = head.get_head(client)
reply = get.get_data(headers, client)
src = src.split("/")[-1]
if (src.startswith("/") == False):
src = "/{}".format(str(src))
url = "{}{}".format(str(client.addr), str(src))
imgtag['src'] = "./images{}".format(src)
print('Request: ', str(request))
images.append((src, reply))
return images
def main(country, yeari, yearf):
data = get_data()
cldata = clean.clean_data(data)
filfix = filter_fixer(cldata, country, yeari, yearf)
pop_plot_r, gr_plot_r = analyse_this(filfix)
nomb, altnomb, capi, regi, langu, cpopul = enrich_that(filfix)
rpdf = make_PDF(nomb, altnomb, capi, regi, langu, cpopul, pop_plot_r,
gr_plot_r)
return rpdf
def task(q, bloom, batch_size):
pid = os.getpid()
print(f"开启任务线程-{pid}:")
urls = [q.get() for i in range(batch_size) if not q.empty()]
urls = [url for url in urls if urlparse(url).netloc not in bloom]
print(urls)
r = get_data(urls)
# update visited site
list(map(bloom.add, [urlparse(url).netloc for url in urls]))
all_friend_links = list(chain(*[site.friends for site in r]))
return pid, all_friend_links
def send_msg(self):
data = ''
# === Handling GET ===
"""
Go to file: [[get.py]].
"""
if self.command == "GET":
get.setup(self)
headers = head.get_head(self)
data, content_type = get.get_data(headers, self)
if content_type == "html":
parser_.set_html_file(self, data)
else:
parser_.set_object(self, data)
# === Handling PUT ===
"""
Go to file: [[put.py]].
"""
if self.command == "PUT":
put.put_data(self)
headers = head.get_head(self)
# === Handling POST ===
"""
Go to file: [[post.py]].
"""
if self.command == "POST":
post.post_data(self)
headers = head.get_head(self)
# === Handling HEAD ===
"""
Go to file: [[head.py]].
"""
if self.command == "HEAD":
head.head_data(self)
return data
Created on Thu Aug 13 15:26:08 2020
@author: COM
"""
#%%
# import sys
# sys.path.append("C:\\pylib")
import cx_Oracle
import os
import get
import recommend_hybrid
import recommend_svd
review = get.get_data()
def recommend(user_i, df):
try:
result = recommend_hybrid.do(user_i, df)
if len(result) < 100:
result2 = recommend_svd.do(user_i, df)
for place in result2["place"]:
if place not in result["place"]:
row = {
"place":
place,
"rating":
float(result2.loc[result2["place"] == place, "rating"])
}