· 9 years ago · Dec 22, 2016, 01:50 AM
1from bs4 import BeautifulSoup
2import requests
3import smtplib
4from email.mime.multipart import MIMEMultipart
5from email.mime.text import MIMEText
6import urllib2
7import time
8from lxml import etree
9keyword1 = "marine"
10keyword2 = "boost"
11keyword3 = "red"
12print("starting")
13r = requests.get('http://kithnyc.com/sitemap_products_1.xml?from=60594372&to=9545825095')
14soup = BeautifulSoup(r.text, 'lxml')
15links = soup.find_all('loc')
16for link in links: #This is where script should scrape until link is found
17 if keyword1 in link.text and keyword2 in link.text and keyword3 in link.text:
18 print(link.text)
19 break
20jake = link.text
21print(jake)
22from_adr='@gmail.com'
23to_adr='@gmail.com'
24msg = MIMEMultipart('alternative')
25msg['Subject'] = "Emailing a link"
26msg['From'] = from_adr
27msg['To'] = to_adr
28html = """
29<html>
30<head></head>
31<body>
32<p>Link:</p>
33<a href="http://www.somewhere.com/whatever.foo">Link Text</a>
34</body>
35</html>
36"""
37print("found")
38part1=MIMEText(str(jake), 'html')
39part2=MIMEText(jake, 'text')
40msg.attach(part1)
41msg.attach(part2)
42s = smtplib.SMTP('smtp.gmail.com', 587)
43s.starttls()
44s.login("@gmail.com", "pass")
45s.sendmail(from_adr, to_adr, msg.as_string())
46s.quit
47url = str(jake) #second script
48response = urllib2.urlopen(url)
49htmlparser = etree.HTMLParser()
50tree = etree.parse(response, htmlparser)
51nodeList = tree.xpath('//script[@type="text/javascript"][contains(.,"Kith")][contains(.,"variants")]')
52blue = len(nodeList)
53if blue != 0:
54 for node in nodeList:
55 print etree.tostring(node)
56 server = smtplib.SMTP('smtp.gmail.com', 587)
57 server.starttls()
58 server.login("@gmail.com", "pass")
59 variants = etree.tostring(node)
60 msg2 = str(variants)
61 server.sendmail("@gmail.com", "@gmail.com", msg2)
62 server.quit()