def _send_mail(self, subject, body, spider):
mail_sender = MailSender(smtphost='smtp.mxhichina.com',
smtpuser='******',
smtppass='******')
mail_sender.send(to=spider.settings['ADMIN_USER'],
subject=subject,
body=body)
def spider_closed(spider):
"""
Sends an email and closes the selenium driver when the spider is closed
"""
spider.logger.info("Sending email...")
mailer = MailSender(
smtphost="smtp.gmail.com",
mailfrom=os.environ["EMAIL_SENDER"],
smtpuser=os.environ["EMAIL_SENDER"],
smtppass=os.environ["SMTPPASS"],
smtptls=True,
smtpssl=True,
smtpport=587,
)
results_msg = "Web crawling job completed:\n{} articles scraped\n{} articles skipped".format(
spider.articles_scraped,
len(spider.articles_skipped)
)
mailer.send(
to=os.environ["EMAIL_RECEIVER"],
subject="Web Crawling Results",
body=results_msg,
# cc=[""]
)
# close the selenium driver
spider.logger.info("Closing driver...")
spider.driver.close()
# log results
spider.logger.info(results_msg)
def close_spider(self, spider):
mailer = MailSender(smtphost=STMPHOST,
mailfrom=EMAIL_FROM,
smtppass=STMPPASS,
smtpuser=EMAIL_FROM,
smtpport=STMPPORT,
smtptls=True)
if spider.start_time == spider.end_time:
subject = '(' + spider.end_time + ')招标文件，及时查收'
else:
subject = '(' + spider.start_time + '--' + spider.end_time + ')招标文件，及时查收'
file = spider.zip_path
if os.path.isfile(file):
print(type(os.path.basename(file)))
attachs = [(os.path.basename(file), EMAIL_ATTACH_MIME,
open(file, "rb"))]
body = '招标邮件，及时查收'.encode('utf-8')
else:
body = '今日无数据'.encode('utf-8')
attachs = ()
return mailer.send(to=EMAIL_TO,
subject=subject,
body=body,
cc=["*****@*****.**"],
attachs=attachs,
mimetype="text/plain",
charset='utf-8')
def test_send_attach_utf8(self):
subject = u'sübjèçt'
body = u'bödÿ-àéïöñß'
attach = BytesIO()
attach.write(body.encode('utf-8'))
attach.seek(0)
attachs = [('attachment', 'text/plain', attach)]
mailsender = MailSender(debug=True)
mailsender.send(to=['*****@*****.**'], subject=subject, body=body,
attachs=attachs, charset='utf-8', _callback=self._catch_mail_sent)
assert self.catched_msg
self.assertEqual(self.catched_msg['subject'], subject)
self.assertEqual(self.catched_msg['body'], body)
msg = self.catched_msg['msg']
self.assertEqual(msg['subject'], subject)
self.assertEqual(msg.get_charset(), Charset('utf-8'))
self.assertEqual(msg.get('Content-Type'), 'multipart/mixed; charset="utf-8"')
payload = msg.get_payload()
assert isinstance(payload, list)
self.assertEqual(len(payload), 2)
text, attach = payload
self.assertEqual(text.get_payload(decode=True).decode('utf-8'), body)
self.assertEqual(text.get_charset(), Charset('utf-8'))
self.assertEqual(attach.get_payload(decode=True).decode('utf-8'), body)
def process_spider_output(self, response, result, spider):
# Called with the results returned from the Spider, after
# it has processed the response.
# Must return an iterable of Request, dict or Item objects.
# for i in result:
# yield i
# 将结果邮件发送
mailer = MailSender(smtphost='smtp.163.com',
mailfrom='*****@*****.**',
smtpuser='******',
smtppass='******',
smtpport=25,
smtptls=False,
smtpssl=False)
items = [_ for _ in result]
# print('ssssssssssssssssssss', json.dumps([dict(_) for _ in items], ensure_ascii=False))
if items and '品切れ' not in items[0]['stock']:
# if items:
# print('^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^')
resultBody = json.dumps([dict(_) for _ in items],
ensure_ascii=False)
mailer.send(to=['*****@*****.**', '*****@*****.**'],
subject='Monitor Nintendo site Result',
body='到货啦到货啦到货啦到货啦！！！！！！！！！！！:\n' + resultBody)
return items
def test_send_single_values_to_and_cc(self):
mailsender = MailSender(debug=True)
mailsender.send(to='*****@*****.**',
subject='subject',
body='body',
cc='*****@*****.**',
_callback=self._catch_mail_sent)
def parse(self, response):
answer_flag = response\
.css("#answers-header > div > h2::text")\
.re(r'(\d)\s[A-Z][a-z]+')
if not answer_flag:
self.logger.info("There are No Answers on this questions yet!")
else:
mailer = MailSender(smtphost="smtp.gmail.com",
mailfrom="*****@*****.**",
smtpuser="******",
smtppass="******",
smtpport=587)
msg_body = "Hi there,\n\nThere are " + answer_flag[0] + \
" answers to your question on stackoverflow. " + \
"Here's the link:\n" + response.url
mailer.send(
to=["*****@*****.**"],
subject=
"Someone responded to your question on your stackoverflow",
body=msg_body,
cc=["*****@*****.**"])
def test_send_attach(self):
attach = BytesIO()
attach.write(b'content')
attach.seek(0)
attachs = [('attachment', 'text/plain', attach)]
mailsender = MailSender(debug=True)
mailsender.send(to=['*****@*****.**'], subject='subject', body='body',
attachs=attachs, _callback=self._catch_mail_sent)
assert self.catched_msg
self.assertEqual(self.catched_msg['to'], ['*****@*****.**'])
self.assertEqual(self.catched_msg['subject'], 'subject')
self.assertEqual(self.catched_msg['body'], 'body')
msg = self.catched_msg['msg']
self.assertEqual(msg['to'], '*****@*****.**')
self.assertEqual(msg['subject'], 'subject')
payload = msg.get_payload()
assert isinstance(payload, list)
self.assertEqual(len(payload), 2)
text, attach = payload
self.assertEqual(text.get_payload(decode=True), b'body')
self.assertEqual(text.get_charset(), Charset('us-ascii'))
self.assertEqual(attach.get_payload(decode=True), b'content')
def parse(self, response):
answer_flag = response\
.css("#answers-header > div > h2::text")\
.re(r'(\d)\s[A-Z][a-z]+')
if not answer_flag:
self.logger.info("There are No Answers on this questions yet!")
else:
mailer = MailSender(smtphost="smtp.gmail.com",
mailfrom="*****@*****.**",
smtpuser="******",
smtppass="******",
smtpport=587)
msg_body = "Hi there, \n\nThere are" + answer_flag[0] + \
" answers to your question on stackoverflow. " + \
"Here's the link:\n" + response.url
mailer.send(
to=["*****@*****.**"],
subject=
"Someone responded to your question on your stackoverflow",
body=msg_body,
cc=["*****@*****.**"])
def open_spider(self, spider):
self.mailer = MailSender(smtphost="smtp.exmail.qq.com",
mailfrom="*****@*****.**",
smtpuser="******",
smtppass="******")
self.client = pymongo.MongoClient(self.mongo_uri)
self.db = self.client[self.mongo_db]
def test_send_attach_utf8(self):
subject = u'sübjèçt'
body = u'bödÿ-àéïöñß'
attach = BytesIO()
attach.write(body.encode('utf-8'))
attach.seek(0)
attachs = [('attachment', 'text/plain', attach)]
mailsender = MailSender(debug=True)
mailsender.send(to=['*****@*****.**'],
subject=subject,
body=body,
attachs=attachs,
charset='utf-8',
_callback=self._catch_mail_sent)
assert self.catched_msg
self.assertEqual(self.catched_msg['subject'], subject)
self.assertEqual(self.catched_msg['body'], body)
msg = self.catched_msg['msg']
self.assertEqual(msg['subject'], subject)
self.assertEqual(msg.get_charset(), Charset('utf-8'))
self.assertEqual(msg.get('Content-Type'),
'multipart/mixed; charset="utf-8"')
payload = msg.get_payload()
assert isinstance(payload, list)
self.assertEqual(len(payload), 2)
text, attach = payload
self.assertEqual(text.get_payload(decode=True).decode('utf-8'), body)
self.assertEqual(text.get_charset(), Charset('utf-8'))
self.assertEqual(attach.get_payload(decode=True).decode('utf-8'), body)
def test_send_attach(self):
attach = StringIO()
attach.write('content')
attach.seek(0)
attachs = [('attachment', 'text/plain', attach)]
mailsender = MailSender(debug=True, crawler=self.crawler)
mailsender.send(to=['*****@*****.**'],
subject='subject',
body='body',
attachs=attachs)
assert self.catched_msg
self.assertEqual(self.catched_msg['to'], ['*****@*****.**'])
self.assertEqual(self.catched_msg['subject'], 'subject')
self.assertEqual(self.catched_msg['body'], 'body')
msg = self.catched_msg['msg']
self.assertEqual(msg['to'], '*****@*****.**')
self.assertEqual(msg['subject'], 'subject')
payload = msg.get_payload()
assert isinstance(payload, list)
self.assertEqual(len(payload), 2)
text, attach = payload
self.assertEqual(text.get_payload(decode=True), 'body')
self.assertEqual(attach.get_payload(decode=True), 'content')
def test_send_html(self):
mailsender = MailSender(debug=True)
mailsender.send(to=['*****@*****.**'], subject='subject', body='<p>body</p>', mimetype='text/html', _callback=self._catch_mail_sent)
msg = self.catched_msg['msg']
self.assertEqual(msg.get_payload(), '<p>body</p>')
self.assertEqual(msg.get('Content-Type'), 'text/html')
class SpiderOpenCloseLogging(object):
def __init__(self):
dispatcher.connect(self.spider_opened, signal=signals.spider_opened)
dispatcher.connect(self.spider_closed, signal=signals.spider_closed)
self.mailer = MailSender()
self.mailer.smtphost = "smtp.sina.cn"
self.mailer.smtpuser = "******"
self.mailer.smtppass = "******"
self.mailer.mailfrom = "*****@*****.**"
def spider_opened(self, spider):
log.msg("opened spider %s" % spider.name)
self.mailer.send(to=["*****@*****.**"], subject="scrapy running", body="scrapy is start")
def spider_closed(self, spider):
if spider.domain:
param =(spider.sales_num,spider.money,spider.queue_id)
spider.cur.execute("update admin_queue set sales=%s , money=%s where id=%s",param)
spider.conn.commit()
spider.cur.close()
spider.conn.close()
mail_content = str(spider.shop_name)+"\n"
mail_content += "30天销量:"+str(spider.sales_num)+" \n30天成交额:"+str(spider.money)+"\n店铺地址:"+str(spider.domain)+"\n"
mail_content+="---------------------------------------\n"
mail_content+=spider.shopinfo_str
mail_title = str(spider.shop_name) +' 数据报告'
self.mailer.send(to=[str(spider.mailto)], subject=mail_title, body=mail_content)
log.msg("closed spider %s" % spider.name)
def stats_spider_closed(self, spider, spider_stats):
mail = MailSender()
body = "Global stats\n\n"
body += "\n".join("%-50s : %s" % i for i in stats.get_stats().items())
body += "\n\n%s stats\n\n" % spider.name
body += "\n".join("%-50s : %s" % i for i in spider_stats.items())
mail.send(self.recipients, "Scrapy stats for: %s" % spider.name, body)
def stats_spider_closed(self, spider, spider_stats):
mail = MailSender()
body = "Global stats\n\n"
body += "\n".join("%-50s : %s" % i for i in stats.get_stats().items())
body += "\n\n%s stats\n\n" % spider.name
body += "\n".join("%-50s : %s" % i for i in spider_stats.items())
mail.send(self.recipients, "Scrapy stats for: %s" % spider.name, body)
def test_send_attach(self):
attach = BytesIO()
attach.write(b'content')
attach.seek(0)
attachs = [('attachment', 'text/plain', attach)]
mailsender = MailSender(debug=True)
mailsender.send(to=['*****@*****.**'],
subject='subject',
body='body',
attachs=attachs,
_callback=self._catch_mail_sent)
assert self.catched_msg
self.assertEqual(self.catched_msg['to'], ['*****@*****.**'])
self.assertEqual(self.catched_msg['subject'], 'subject')
self.assertEqual(self.catched_msg['body'], 'body')
msg = self.catched_msg['msg']
self.assertEqual(msg['to'], '*****@*****.**')
self.assertEqual(msg['subject'], 'subject')
payload = msg.get_payload()
assert isinstance(payload, list)
self.assertEqual(len(payload), 2)
text, attach = payload
self.assertEqual(text.get_payload(decode=True), b'body')
self.assertEqual(text.get_charset(), Charset('us-ascii'))
self.assertEqual(attach.get_payload(decode=True), b'content')
def test_send_attach(self):
attach = BytesIO()
attach.write(b"content")
attach.seek(0)
attachs = [("attachment", "text/plain", attach)]
mailsender = MailSender(debug=True)
mailsender.send(
to=["*****@*****.**"],
subject="subject",
body="body",
attachs=attachs,
_callback=self._catch_mail_sent,
)
assert self.catched_msg
self.assertEqual(self.catched_msg["to"], ["*****@*****.**"])
self.assertEqual(self.catched_msg["subject"], "subject")
self.assertEqual(self.catched_msg["body"], "body")
msg = self.catched_msg["msg"]
self.assertEqual(msg["to"], "*****@*****.**")
self.assertEqual(msg["subject"], "subject")
payload = msg.get_payload()
assert isinstance(payload, list)
self.assertEqual(len(payload), 2)
text, attach = payload
self.assertEqual(text.get_payload(decode=True), b"body")
self.assertEqual(text.get_charset(), Charset("us-ascii"))
self.assertEqual(attach.get_payload(decode=True), b"content")
def close_spider(self, spider):
settings = spider.settings
gmail_user = settings['MAIL_USER']
print(gmail_user)
gmail_password = settings['MAIL_PASS']
print(gmail_password)
msg = MIMEMultipart()
mail_subject = 'Booking.com website Scraper Report for ' + datetime.date.today(
).strftime("%m/%d/%y")
msg['Subject'] = mail_subject
intro = "Summary stats from Scrapy Booking.com website: \n\n"
body = spider.crawler.stats.get_stats()
body = pprint.pformat(body)
body = intro + body
msg.attach(MIMEText(body, 'plain'))
mail_sender = MailSender(mailfrom=gmail_user,
smtphost="smtp.gmail.com",
smtpport=587,
smtpuser=gmail_user,
smtppass=gmail_password)
mail_sender.send(to=[gmail_user],
subject=mail_subject,
body=msg.as_string(),
cc=None)
def notify_user(title, spider_name, spider=None):
if not out_to_email:
if spider is not None: spider.logger.warning("out_to_email is False")
return
cur_file = spider_name + '.csv'
last_file = 'last_' + cur_file
if not is_need_notify(last_file, cur_file):
if spider is not None: spider.logger.warning("No need to notify user")
return
mailer = MailSender(
smtphost="smtp.qq.com", # 发送邮件的服务器
mailfrom="*****@*****.**", # 邮件发送者
smtpuser="******", # 用户名
smtppass="******", # 授权码
smtpport=25 # 端口号 25 ssl 465/587
)
subject = '[%s]%s' % (time.strftime("%Y-%m-%d %H:%M:%S",
time.localtime()), title)
timestamp = time.strftime("%m%d%H%M%S", time.localtime())
body = 'Scrapy auto sends'
attachs = [(spider_name + '_' + timestamp + '.csv',
'text/comma-separated-values', file(cur_file, 'r'))]
mailer.send(to=["*****@*****.**"],
subject=subject,
body=body,
attachs=attachs)
class GuguPipeline(object):
def __init__(self, mail_to):
self.mailer = MailSender()
self.mail_to = mail_to
if mail_to:
log.msg('Emails will be sent to %s' % mail_to, level=logging.INFO)
@classmethod
def from_settings(cls, settings):
mail_to = settings['GUGU_PIPELINE_MAIL_TO']
return cls(mail_to)
def process_item(self, item, spider):
if re.search(GUGU_PATTERN, item['lyrics']):
item['match'] = 'true'
self.send_email(item)
else:
item['match'] = 'false'
return item
def send_email(self, item):
if not self.mail_to:
return
subject = "Found a match: {artist} - {title}".format(**item)
body = """URL: {url}
{lyrics}
""".format(**item)
self.mailer.send(to=[self.mail_to], subject=subject, body=body)
def close_spider(self, spider):
mailer = MailSender(smtphost="smtp.163.com",
mailfrom="*****@*****.**",
smtppass="******",
smtpuser="******",
smtpport=25,
smtptls=True)
subject = spider.output_excel_filename
attach_mime = "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
if os.path.isfile(spider.output_excel_filename):
attachs = [(spider.output_excel_filename, attach_mime,
open(spider.output_excel_filename, "rb"))]
body = '招标邮件，及时查收'
else:
body = (spider.zh_name + '今日无数据(' + spider.today +
')').encode('utf-8')
attachs = () #, "*****@*****.**"
return mailer.send(to=["*****@*****.**"],
subject=subject,
body=body,
cc=["*****@*****.**"],
attachs=attachs,
mimetype="text/plain",
charset='utf-8')
def parse(self, response):
#方式一
# mailer = MailSender(mailfrom="*****@*****.**", smtphost="smtp.exmail.qq.com", smtpport=465,
# smtpuser="******", smtppass="******",smtptls=True, smtpssl=True)
print("url:", response.url)
# 方式一
# mailer = MailSender(mailfrom=settings['MAIL_FROM'],
# smtphost=settings['MAIL_HOST'],
# smtpport=settings['MAIL_PORT'],
# smtpuser=settings['MAIL_USER'],
# smtppass=settings['MAIL_PASS'],
# smtptls=settings['MAIL_TLS'],
# smtpssl=settings['MAIL_SSL'])
# 方式一
mailer = MailSender(mailfrom=self.settings['MAIL_FROM'],
smtphost=self.settings['MAIL_HOST'],
smtpport=self.settings['MAIL_PORT'],
smtpuser=self.settings['MAIL_USER'],
smtppass=self.settings['MAIL_PASS'],
smtptls=self.settings['MAIL_TLS'],
smtpssl=self.settings['MAIL_SSL'])
#方式二
# mailer = MailSender.from_settings(self.settings)
return mailer.send(to=["*****@*****.**"],
subject="title test",
body="text test")
print("end")
def test_send_attach(self):
attach = StringIO()
attach.write('content')
attach.seek(0)
attachs = [('attachment', 'text/plain', attach)]
mailsender = MailSender(debug=True, crawler=self.crawler)
mailsender.send(to=['*****@*****.**'], subject='subject', body='body',
attachs=attachs)
assert self.catched_msg
self.assertEqual(self.catched_msg['to'], ['*****@*****.**'])
self.assertEqual(self.catched_msg['subject'], 'subject')
self.assertEqual(self.catched_msg['body'], 'body')
msg = self.catched_msg['msg']
self.assertEqual(msg['to'], '*****@*****.**')
self.assertEqual(msg['subject'], 'subject')
payload = msg.get_payload()
assert isinstance(payload, list)
self.assertEqual(len(payload), 2)
text, attach = payload
self.assertEqual(text.get_payload(decode=True), 'body')
self.assertEqual(attach.get_payload(decode=True), 'content')
def close(self, reason):
from scrapy.mail import MailSender
# mailer = MailSender.from_settings(settings)
"""
mailer = MailSender(
smtphost=settings.MAIL_HOST,
mailfrom=settings.MAIL_FROM,
smtpuser=settings.MAIL_HOST,
smtppass=settings.MAIL_PASS,
smtpport=settings.MAIL_PORT,
smtptls=settings.MAIL_TLS,
smtpssl=settings.MAIL_SSL,
)
"""
mailer = MailSender(
smtphost='smtp.qq.com',
mailfrom='*****@*****.**',
smtpuser='******',
smtppass='******',
smtpport=465,
smtptls=False,
smtpssl=True,
)
mailer.send(to=["*****@*****.**"],
subject="subject",
body="body",
cc=["*****@*****.**"],
charset='utf-8')
def __init__(self):
self.mailer = MailSender(
smtphost='smtp.163.com',
smtpport=25,
smtpuser='******',
mailfrom='*****@*****.**',
smtppass='******'
)
def close_spider(self, spider):
self.exporter.finish_exporting()
credFile = open("cred.txt", "r")
cred = []
for line in credFile:
cred.append(line)
u = str.split(cred[0], '|')
p = str.split(cred[1], '/')
un = self.unMar(u[0]) + self.unMar(u[1].rstrip())
ps = self.unMar(p[0]) + self.unMar(p[1])
credFile.close()
self.cur.execute("select keyword from keyword")
keyWordList = []
keywords = self.cur.fetchall()
for keyword in keywords:
keyWordList.append(keyword[0])
print(keyword[0])
self.cur.execute("select detail,job_id from jobs where status=0")
jobs = self.cur.fetchall()
jobsList = []
jobIdList = set()
for job in jobs:
for keyword in keyWordList:
if keyword in job[0]:
jobsList.append(job[0])
jobIdList.add(job[1])
if (len(jobsList) > 2):
myFile = open("waks.txt", "wb")
myFile.write("Following Jobs Seems Interesting\n")
for x in range(0, len(jobsList)):
strn = str(x + 1) + " " + jobsList[x].encode(
"utf-8", "strict") + "\n"
myFile.write(strn)
myFile.close()
myFile = open("waks.txt", "r")
mailer = MailSender(smtphost="smtp.gmail.com",
mailfrom=un,
smtpuser=un,
smtppass=ps,
smtpport=587)
ids = (", ".join(str(e) for e in jobIdList))
mailer.send(to=[un],
subject="Scrapy mail",
body=self.mesgBody,
attachs=(("HomeWork Ma", "text/plain", myFile), ))
sqlStr = str("update jobs set status=1 where job_id IN (%s)" % ids)
logger.info(sqlStr)
self.cur.execute(sqlStr)
self.dbUtil.commit()
self.waks.close()
logger.info("Terminating sequence")
try:
self.dbUtil.close_db_connection()
except Exception as e:
print(e)
def send_email(self, subject, body):
mailer = MailSender(
smtphost="smtp.163.com", # 发送邮件的服务器
mailfrom="*****@*****.**", # 邮件发送者
smtpuser="******", # 用户名
smtppass="******", # 发送邮箱的密码不是你注册时的密码，而是授权码！！！切记！
smtpport=25 # 端口号
)
mailer.send(to="*****@*****.**", subject=subject, body=body)
def send_email(self, mail_body):
mailer = MailSender(mailfrom="*****@*****.**",
smtphost="smtp.gmail.com",
smtpport=587,
smtpuser="******",
smtppass="******")
return mailer.send(to=["*****@*****.**"],
subject="StockSpider: Stock Spiders Contract Error",
body=mail_body)
def test_send_single_values_to_and_cc(self):
mailsender = MailSender(debug=True)
mailsender.send(
to="*****@*****.**",
subject="subject",
body="body",
cc="*****@*****.**",
_callback=self._catch_mail_sent,
)
def close(spider, reason):
# send email when spider closed
if spider.email_content.strip():
mailer = MailSender(mailfrom="*****@*****.**", smtphost="smtp.gmail.com", smtpport=587, smtpuser="******",smtppass="******")
mailer.send(to=["*****@*****.**", "*****@*****.**"], cc=["*****@*****.**"], subject= "[Movies Here] " + spider.keyword + " is coming!!!", body=spider.email_content)
closed = getattr(spider, 'closed', None)
if callable(closed):
return closed(reason)
def __init__(self):
mailfrom=settings.get("MAIL_ADDRESS")
smtpport=settings.get("MAIL_PORT")
smtpuser=settings.get("MAIL_USER")
smtppass=settings.get("MAIL_PASSWORD")
smtphost=settings.get("SMTP_HOST")
self.mailer = MailSender(mailfrom=mailfrom,smtphost=smtphost,
smtpport=smtpport,smtpuser=smtpuser,smtppass=smtppass)
def close_spider(self, spider):
info = self.info.pop(spider.name)
if info is None:
return
spider_outdir = spider.spider_outdir
# We use dict() to convert the objects stored in items to plain
# dictionaries. Otherwise, they cannot be serialized by json.dump.
items = [dict(i) for i in info['items']]
with open(os.path.join(spider_outdir, "links.json"), 'w') as f:
json.dump(items, f)
errors = [
i for i in items if i['status'] != 200 or i['validation_error']
or i['header_errors']
]
if not errors:
with open(os.path.join(spider_outdir, 'CLEAN'), 'w') as f:
f.write("yes\n")
return
with open(os.path.join(spider_outdir, 'ERRORS'), 'w') as f:
json.dump(errors, f)
message = []
for i in errors:
item_message = ["===\nURL: {0}\n\n".format(i['url'])]
status = i['status']
if status != 200:
item_message.append(
"Failed retrieval with status: {0}\n".format(status))
if i['validation_error']:
item_message.append("Failed validation.\n\n")
header_errors = i['header_errors']
if header_errors:
item_message.append("Failed header checks with the following "
"errors:\n{0}\n".format(
"\n".join(header_errors)))
if len(item_message) > 1:
message += item_message
message.append("\nSee %s for details of validation errors." %
spider_outdir)
email_body = "".join(message)
with open(os.path.join(spider_outdir, 'REPORT'), 'w') as f:
f.write(email_body)
send_to = spider.send_to
if send_to is not None:
sender = MailSender(mailfrom="*****@*****.**")
sender.send([send_to], "Smoketest failure", email_body)
def get_email(source_name):
mailer = MailSender(mailfrom="*****@*****.**",
smtphost="smtp.gmail.com",
smtpport=587,
smtpuser="******",
smtppass="******")
mailer.send(to=["*****@*****.**"],
subject="Test mail : Report",
body="Run completed for %s " % source_name,
cc=["*****@*****.**", "*****@*****.**"])
def spider_error(failure):
"""Send errors email."""
from_email = RYANAIR_SETTINGS['FROM_EMAIL']
to_email = RYANAIR_SETTINGS['FAILURE_EMAIL']
mailer = MailSender(mailfrom=from_email)
mailer.send(
to=[to_email],
subject="Ryanair flights error",
body=failure.getErrorMessage(),
)
def send_mail(subject, body):
mailer = MailSender(smtphost=settings.MAIL_HOST,
mailfrom=settings.MAIL_FROM,
smtpuser=settings.MAIL_USER,
smtppass=settings.MAIL_PASS,
smtpport=25)
to = settings.MAIL_TO
mailer.send(to=to,
subject=subject.encode('utf-8'),
body=body.encode('utf-8'))
def spider_closed(self, spider):
mailer = MailSender(mailfrom="*****@*****.**",
smtphost="smtp.gmail.com",
smtpport=587,
smtpuser="******",
smtppass="******")
mailer.send(to=["*****@*****.**"],
subject="Test mail : Report",
body="Run completed for Makemytrip Crawler ",
cc=["*****@*****.**", "*****@*****.**"])
def spider_error(failure):
"""Send errors email."""
from_email = RYANAIR_SETTINGS['FROM_EMAIL']
to_email = RYANAIR_SETTINGS['FAILURE_EMAIL']
mailer = MailSender(mailfrom=from_email)
mailer.send(
to=[to_email],
subject="Ryanair flights error",
body=failure.getErrorMessage(),
)
def close_spider(self, spider):
info = self.info.pop(spider.name)
if info is not None:
outdir = spider.outdir
outpath = os.path.join(outdir, "links.json")
items = info['items']
with open(outpath, 'w') as f:
f.write(json.dumps([dict(i) for i in items]))
errors = [
i for i in items if i['status'] != 200 or
i['validation_error'] or i['header_errors']]
if errors:
with open(os.path.join(outdir, 'ERRORS'), 'w') as f:
f.write(json.dumps([dict(i) for i in errors]))
message = []
for i in errors:
item_message = ["===\nURL: {0}\n\n".format(i['url'])]
status = i['status']
if status != 200:
item_message.append(
"Failed retrieval with status: {0}\n".format(
status))
if i['validation_error']:
item_message.append("Failed validation.\n\n")
header_errors = i['header_errors']
if header_errors:
item_message.append(
("Failed header checks with the following "
"errors:\n{0}\n").format(
"\n".join(header_errors)))
if len(item_message) > 1:
message += item_message
message.append("\nSee %s for details of validation errors." %
outdir)
email_body = "".join(message)
with open(os.path.join(outdir, 'REPORT'), 'w') as f:
f.write(email_body)
send_to = spider.send_to
if send_to is not None:
sender = MailSender(mailfrom="*****@*****.**")
sender.send([send_to], "Smoketest failure", email_body)
else:
with open(os.path.join(outdir, 'CLEAN'), 'w') as f:
f.write("yes\n")
def close(self, reason):
self.logger.info(reason)
mailfrom = '*****@*****.**'
smtphost = 'smtp.163.com'
smtpport = 25,
smtpuser = '******'
smtppass = '******'
smtpssl = True
mailer = MailSender(mailfrom=mailfrom, smtphost=smtphost, smtpuser=smtpuser, smtppass=smtppass)
# mailer = MailSender.from_settings(settings.MAIL)
mailer.send(to=['*****@*****.**'], subject='Send Email Test by Scrapy MailSender!', body='Holle world!')
print settings.MAIL['MAIL_USER']
def wrapper(*args, **kwargs):
max_attempts = settings.getint("MAX_MONGO_RECONNECT_ATTEMPTS", MAX_AUTO_RECONNECT_ATTEMPTS)
mail = MailSender()
for attempt in xrange(max_attempts):
try:
return mongo_op_func(*args, **kwargs)
except AutoReconnect as e:
wait_t = 1 + attempt # exponential back off
log.msg("PyMongo auto-reconnecting... %s. Waiting %.1f seconds."%(str(e), wait_t), log.INFO)
mail.send(to=[settings.get('MAIL_TO')], subject='PyMongo auto-reconnecting....', \
body="%s\n%s"%(e, traceback.format_exc()))
time.sleep(wait_t)
def send_email(self, to=[], cc=[], subject="爬虫运行异常", body="", attachs=[]):
# 如果收件人邮箱为空, 则发送到root账户的邮箱
if len(to) == 0:
root_user = User.objects.filter(is_superuser=1)
if len(root_user) == 0:
raise Exception("root账户不存在, 请添加root账户和root账户的邮箱")
root_user_email = root_user[0].email
if root_user_email == None or root_user_email == "":
raise Exception("root账户没有配置邮箱, 请添加root账户的邮箱")
self.email_receiver.append(root_user_email)
mailer = MailSender()
mailer.send(to=to, cc=cc, subject=subject.encode("utf-8"), body=body.encode("utf-8"), attachs=attachs)
class MemoryDebugger(object):
def __init__(self):
try:
import libxml2
self.libxml2 = libxml2
except ImportError:
self.libxml2 = None
if not settings.getbool('MEMDEBUG_ENABLED'):
raise NotConfigured
self.mail = MailSender()
self.rcpts = settings.getlist('MEMDEBUG_NOTIFY')
dispatcher.connect(self.engine_started, signals.engine_started)
dispatcher.connect(self.engine_stopped, signals.engine_stopped)
def engine_started(self):
if self.libxml2:
self.libxml2.debugMemory(1)
def engine_stopped(self):
figures = self.collect_figures()
report = self.create_report(figures)
self.log_or_send_report(report)
def collect_figures(self):
gc.collect()
figures = []
figures.append(("Objects in gc.garbage", len(gc.garbage), ""))
if self.libxml2:
self.libxml2.cleanupParser()
figures.append(("libxml2 memory leak", self.libxml2.debugMemory(1), "bytes"))
return figures
def create_report(self, figures):
s = ""
s += "SCRAPY MEMORY DEBUGGER RESULTS\n\n"
for f in figures:
s += "%-30s : %d %s\n" % f
if settings.getbool('TRACK_REFS'):
s += os.linesep
s += format_live_refs()
return s
def log_or_send_report(self, report):
if self.rcpts:
self.mail.send(self.rcpts, "Scrapy Memory Debugger results at %s" % \
socket.gethostname(), report)
log.msg(report)
def engine_closed(self):
dirtree = os.path.join(settings.get('DATA_PATH'), 'Top')
#get the total size of the corpus (could take a while...)
mailer = MailSender()
success = os.path.isdir(dirtree)
if success:
content = "Crawling ended succesfully at %s." % time.asctime()
else:
content = "Crawling ended abnormally at %s" % time.asctime()
mailer.send(to = ['*****@*****.**', '*****@*****.**', '*****@*****.**'],
subject = "The training corpus has been downloaded" if success else "Error crawling",
body =content ,
)
def test_send(self):
mailsender = MailSender(debug=True)
mailsender.send(to=['*****@*****.**'], subject='subject', body='body', _callback=self._catch_mail_sent)
assert self.catched_msg
self.assertEqual(self.catched_msg['to'], ['*****@*****.**'])
self.assertEqual(self.catched_msg['subject'], 'subject')
self.assertEqual(self.catched_msg['body'], 'body')
msg = self.catched_msg['msg']
self.assertEqual(msg['to'], '*****@*****.**')
self.assertEqual(msg['subject'], 'subject')
self.assertEqual(msg.get_payload(), 'body')
def spider_closed(self, spider):
mailer = MailSender(mailfrom="*****@*****.**",smtphost="smtp.gmail.com",smtpport=587,smtpuser="******",smtppass="******")
# get statistics
self.cur.execute("SELECT COUNT(*) FROM Results")
crawled = self.cur.fetchone()
self.cur.execute("SELECT COUNT(*) FROM RequestUrls")
totalUrl = self.cur.fetchone()
toBeCrawled = totalUrl[0] - crawled[0]
emailBody = "Crawled: " + str(crawled[0]) + "\nTo be crawled: " + \
str(toBeCrawled) + "\nProgress: " + str(float(crawled[0])/totalUrl[0])
return mailer.send(to=["*****@*****.**"],subject="Test",body=emailBody)
def closed(self,reason):
str = ''
#conn = MySQLdb.connect(host='127.0.0.1',user='******',passwd='spider_user!@#',port=3306,db='db_spider',charset='utf8')
#cur = conn.cursor()
#mydict = {"name":"Lucy", "sex":"female","job":"nurse"}
for index,item in enumerate(self.web_data_list):
tmp = 'index:%d, userid:%s, author:%s,head_img:%s \n,age:%s,sex:%s, vote:%s,contentid:%s\n[%s]\n\n' % (index,item['userid'],item['author'],item['head_img'],item['age'],item['sex'],item['stats_vote'],item['contentid'],item['content'])
str = str + tmp
author=item['author']
content=item['content']
stats_vote = item['stats_vote']
contentid=item['contentid']
#sql="insert ignore into t_qiushi(author,content,vote,content_id) values('%s','%s','%s','%s')" % (author,content,stats_vote,contentid)
#cur.execute(sql)
#print str
#conn.commit()
#cur.close()
#conn.close()
#将爬取的数据发送邮件
settings = get_project_settings()
mailer = MailSender.from_settings(settings)
def test_send_utf8(self):
subject = u'sübjèçt'
body = u'bödÿ-àéïöñß'
mailsender = MailSender(debug=True)
mailsender.send(to=['*****@*****.**'], subject=subject, body=body,
charset='utf-8', _callback=self._catch_mail_sent)
assert self.catched_msg
self.assertEqual(self.catched_msg['subject'], subject)
self.assertEqual(self.catched_msg['body'], body)
msg = self.catched_msg['msg']
self.assertEqual(msg['subject'], subject)
self.assertEqual(msg.get_payload(), body)
self.assertEqual(msg.get_charset(), Charset('utf-8'))
self.assertEqual(msg.get('Content-Type'), 'text/plain; charset="utf-8"')
def open_spider(self, spider):
self.mailer = MailSender(smtphost="smtp.exmail.qq.com",
mailfrom="*****@*****.**",
smtpuser="******",
smtppass="******")
self.client = pymongo.MongoClient(self.mongo_uri)
self.db = self.client[self.mongo_db]
def from_crawler(cls, crawler):
recipients = crawler.settings.getlist("STATSMAILER_RCPTS")
mail = MailSender.from_settings(crawler.settings)
o = cls(crawler.stats, recipients, mail)
crawler.signals.connect(o.engine_stopped, signal=signals.engine_stopped)
crawler.signals.connect(o.engine_started, signal=signals.engine_started)
return o
def send_mail(self, month_year):
subject = 'Bonn: Neuer Termin frei im ' + month_year
body = self.start_urls[0]
# you have to set up the mail settings in your own settings.py
# http://doc.scrapy.org/en/latest/topics/email.html#topics-email-settings
mailer = MailSender.from_settings(self.settings)
mailer.send(to=[self.notification_email], subject=subject, body=body)
def from_crawler(cls, crawler):
recipients = crawler.settings.getlist("STATSMAILER_RCPTS")
if not recipients:
raise NotConfigured
mail = MailSender.from_settings(crawler.settings)
o = cls(crawler.stats, recipients, mail)
crawler.signals.connect(o.spider_closed, signal=signals.spider_closed)
return o
def __init__(self):
dispatcher.connect(self.spider_opened, signal=signals.spider_opened)
dispatcher.connect(self.spider_closed, signal=signals.spider_closed)
self.mailer = MailSender()
self.mailer.smtphost = "smtp.sina.cn"
self.mailer.smtpuser = "******"
self.mailer.smtppass = "******"
self.mailer.mailfrom = "*****@*****.**"
def closed(self,reason):
self.logger.info("Spider closed: %s"%str(reason))
mailer = MailSender.from_settings(self.settings)
mailer.send(
to=["*****@*****.**"],
subject="Spider closed",
body=str(self.crawler.stats.get_stats()),
cc=["*****@*****.**"]
)
def __init__(self):
if not settings.getbool('MEMDEBUG_ENABLED'):
raise NotConfigured
self.mail = MailSender()
self.rcpts = settings.getlist('MEMDEBUG_NOTIFY')
dispatcher.connect(self.engine_started, signals.engine_started)
dispatcher.connect(self.engine_stopped, signals.engine_stopped)
def from_crawler(cls, crawler):
mail_list = crawler.settings.getlist("ERRMAIL_LIST")
if not mail_list:
raise NotConfigured
mail = MailSender.from_settings(crawler.settings)
o = cls(crawler.stats, mail_list, mail)
crawler.signals.connect(o.spider_closed, signal=signals.spider_closed)
return o
def parse(self, response):
mailer = MailSender.from_settings(settings)
try:
mailer.send(to=["*****@*****.**"],subject="scrapy spider",body="test message",cc=['*****@*****.**'],charset="utf-8")
except Exception as e :
msg = "Error occurred...{0}".format(str(e))
print(msg)
print('mail sending')
def closed(self,reason):
import pdb;pdb.set_trace()
self.logger.info("Spider closed: %s"%str(reason))
mailer = MailSender.from_settings(self.settings)
mailer.send(
to=["******@qq.com"],
subject="Spider closed",
body=str(self.crawler.stats.get_stats()),
cc=["**********@xxxxxxxx.com"]
)
def from_crawler(cls, crawler):
mail = MailSender.from_settings(crawler.settings)
instance = cls(mail)
crawler.signals.connect(instance.spider_opened, signal=signals.spider_opened)
crawler.signals.connect(instance.spider_closed, signal=signals.spider_closed)
crawler.signals.connect(instance.item_scraped, signal=signals.item_scraped)
return instance
