def fetch_url(url, opener=None, timeout=60.0, chunk_size=16384):
if opener is None:
opener = urllib2.build_opener()
try:
output = StringIO()
fileobj = yield idiokit.thread(opener.open, url, timeout=timeout)
try:
while True:
data = yield idiokit.thread(fileobj.read, chunk_size)
if not data:
break
output.write(data)
finally:
fileobj.close()
info = fileobj.info()
info = email.parser.Parser().parsestr(str(info), headersonly=True)
output.seek(0)
idiokit.stop(info, output)
except urllib2.HTTPError as he:
raise HTTPError(he.code, he.msg, he.hdrs, he.fp)
except urllib2.URLError as error:
if _is_timeout(error.reason):
raise FetchUrlTimeout("fetching URL timed out")
raise FetchUrlFailed(str(error))
except socket.error as error:
if _is_timeout(error):
raise FetchUrlTimeout("fetching URL timed out")
raise FetchUrlFailed(str(error))
except httplib.HTTPException as error:
raise FetchUrlFailed(str(error))
def _login(self, server, user, password):
yield idiokit.thread(server.ehlo)
if server.has_extn("starttls"):
yield idiokit.thread(server.starttls)
yield idiokit.thread(server.ehlo)
if user is not None and password is not None and server.has_extn("auth"):
yield idiokit.thread(server.login, user, password)
def feed(self, query, feed_all):
collab = wiki.GraphingWiki(
self.collab_url,
ssl_verify_cert=not self.collab_ignore_cert,
ssl_ca_certs=self.collab_extra_ca_certs)
yield idiokit.thread(collab.authenticate, self.collab_user, self.collab_password)
yield idiokit.sleep(5)
token = None
current = dict()
while True:
try:
result = yield idiokit.thread(collab.request, "IncGetMeta", query, token)
except wiki.WikiFailure as fail:
self.log.error("IncGetMeta failed: {0!r}".format(fail))
else:
incremental, token, (removed, updates) = result
removed = set(removed)
if not incremental:
removed.update(current)
current.clear()
for page, keys in updates.iteritems():
event = current.setdefault(page, events.Event())
event.add("id:open", self.page_id(page))
event.add("gwikipagename", page)
event.add("collab url", self.collab_url + urllib.quote(page.encode("utf8")))
removed.discard(page)
for key, (discarded, added) in keys.iteritems():
for value in map(normalize, discarded):
event.discard(key, value)
for value in map(normalize, added):
event.add(key, value)
if not feed_all:
yield idiokit.send(event)
for page in removed:
current.pop(page, None)
event = events.Event()
event.add("id:close", self.page_id(page))
event.add("gwikipagename", page)
event.add("collab url", self.collab_url + page)
yield idiokit.send(event)
if feed_all:
for page in current:
yield idiokit.send(current[page])
yield idiokit.sleep(self.poll_interval)
def run_mailbox(self, min_delay=5.0, max_delay=60.0):
mailbox = None
try:
while True:
item = yield idiokit.next()
while True:
delay = min(min_delay, max_delay)
while mailbox is None:
try:
mailbox = yield idiokit.thread(self.connect)
except (imaplib.IMAP4.abort, socket.error) as error:
self.log.error(
"Failed IMAP connection ({0})".format(
utils.format_exception(error)))
else:
break
self.log.info(
"Retrying connection in {0:.2f} seconds".format(
delay))
yield idiokit.sleep(delay)
delay = min(2 * delay, max_delay)
event, name, args, keys = item
if event.result().unsafe_is_set():
break
try:
method = getattr(mailbox, name)
result = yield idiokit.thread(method, *args, **keys)
except (imaplib.IMAP4.abort, socket.error) as error:
yield idiokit.thread(self.disconnect, mailbox)
self.log.error("Lost IMAP connection ({0})".format(
utils.format_exception(error)))
mailbox = None
event.fail(LostConnection, "", None)
break
except imaplib.IMAP4.error as error:
event.fail(type(error), error, None)
break
else:
event.succeed(result)
break
finally:
if mailbox is not None:
yield idiokit.thread(self.disconnect, mailbox)
def build_mail(self,
events,
to=[],
cc=[],
bcc=[],
template="",
template_values={},
**keys):
"""
Return a mail object produced based on collected events and session parameters.
The "events" parameter is None when we just want to test building a mail.
"""
if events is None:
events = []
csv = templates.CSVFormatter()
template_keys = {
"csv": csv,
"attach_csv": templates.AttachUnicode(csv),
"attach_and_embed_csv": templates.AttachAndEmbedUnicode(csv),
"attach_zip": templates.AttachZip(csv),
"to": templates.Const(format_addresses(to)),
"cc": templates.Const(format_addresses(cc)),
"bcc": templates.Const(format_addresses(bcc))
}
for key, value in dict(template_values).iteritems():
template_keys[key] = templates.Event(value)
mail_template = MailTemplate(template, **template_keys)
msg = yield idiokit.thread(mail_template.format, events)
idiokit.stop(msg)
def _connect(self, host, port, retry_interval=60.0):
server = None
while server is None:
self.log.info(u"Connecting to SMTP server {0!r} port {1}".format(
host, port))
try:
server = yield idiokit.thread(
smtplib.SMTP,
host,
port,
timeout=self.smtp_connection_timeout)
except (socket.error, smtplib.SMTPException) as exc:
self.log.error(u"Failed connecting to SMTP server: {0}".format(
utils.format_exception(exc)))
else:
self.log.info(u"Connected to the SMTP server")
break
self.log.info(
u"Retrying SMTP connection in {0:.2f} seconds".format(
retry_interval))
yield idiokit.sleep(retry_interval)
idiokit.stop(server)
def _compress(self, queue):
while True:
compress_path = yield queue.wait()
try:
path = yield idiokit.thread(_compress, compress_path)
self.log.info("Compressed archive {0!r}".format(path))
except ValueError:
self.log.error("Invalid path {0!r}".format(compress_path))
def run_mailbox(self, min_delay=5.0, max_delay=60.0):
mailbox = None
try:
while True:
item = yield idiokit.next()
while True:
delay = min(min_delay, max_delay)
while mailbox is None:
try:
mailbox = yield idiokit.thread(self.connect)
except (imaplib.IMAP4.abort, socket.error) as error:
self.log.error("Failed IMAP connection ({0})".format(utils.format_exception(error)))
else:
break
self.log.info("Retrying connection in {0:.2f} seconds".format(delay))
yield idiokit.sleep(delay)
delay = min(2 * delay, max_delay)
event, name, args, keys = item
if event.result().unsafe_is_set():
break
try:
method = getattr(mailbox, name)
result = yield idiokit.thread(method, *args, **keys)
except (imaplib.IMAP4.abort, socket.error) as error:
yield idiokit.thread(self.disconnect, mailbox)
self.log.error("Lost IMAP connection ({0})".format(utils.format_exception(error)))
mailbox = None
except imaplib.IMAP4.error as error:
event.fail(type(error), error, None)
break
else:
event.succeed(result)
break
finally:
if mailbox is not None:
yield idiokit.thread(self.disconnect, mailbox)
def _connect(self, host, port, retry_interval=60.0):
server = None
while server is None:
self.log.info(u"Connecting to SMTP server {0!r} port {1}".format(host, port))
try:
server = yield idiokit.thread(smtplib.SMTP, host, port, timeout=self.smtp_connection_timeout)
except (socket.error, smtplib.SMTPException) as exc:
self.log.error(u"Failed connecting to SMTP server: {0}".format(utils.format_exception(exc)))
else:
self.log.info(u"Connected to the SMTP server")
break
self.log.info(u"Retrying SMTP connection in {0:.2f} seconds".format(retry_interval))
yield idiokit.sleep(retry_interval)
idiokit.stop(server)
def _manage_cache(self, query):
token = None
wikikeys = set()
for valueset in self.keys.values():
wikikeys.update(valueset)
while True:
try:
result = yield idiokit.thread(self.collab.request,
"IncGetMeta", query, token)
except (socket.error, wiki.WikiFailure) as exc:
self.log.error("IncGetMeta failed: {0}".format(exc))
else:
incremental, token, (removed, updates) = result
removed = set(removed)
if not incremental:
removed.update(self.cache.keys())
self.cache.clear()
for page, keys in updates.iteritems():
event = self.cache.setdefault(page, events.Event())
event.add("gwikipagename", page)
removed.discard(page)
for key, (discarded, added) in keys.iteritems():
for value in discarded:
event.discard(key, value)
for value in added:
if key in wikikeys:
event.add(key, value)
for page in removed:
self.cache.pop(page, None)
if removed or updates:
self.log.info(
"Updated {0} pages and removed {1} pages ({2} pages in cache)"
.format(len(updates), len(removed),
len(self.cache.keys())))
yield idiokit.sleep(self.poll_interval)
def _manage_cache(self, query):
token = None
wikikeys = set()
for valueset in self.keys.values():
wikikeys.update(valueset)
while True:
try:
result = yield idiokit.thread(self.collab.request, "IncGetMeta", query, token)
except (socket.error, wiki.WikiFailure) as exc:
self.log.error("IncGetMeta failed: {0}".format(exc))
else:
incremental, token, (removed, updates) = result
removed = set(removed)
if not incremental:
removed.update(self.cache.keys())
self.cache.clear()
for page, keys in updates.iteritems():
event = self.cache.setdefault(page, events.Event())
event.add("gwikipagename", page)
removed.discard(page)
for key, (discarded, added) in keys.iteritems():
for value in discarded:
event.discard(key, value)
for value in added:
if key in wikikeys:
event.add(key, value)
for page in removed:
self.cache.pop(page, None)
if removed or updates:
self.log.info("Updated {0} pages and removed {1} pages ({2} pages in cache)".format(
len(updates), len(removed), len(self.cache.keys())))
yield idiokit.sleep(self.poll_interval)
def build_mail(self, events, to=[], cc=[], bcc=[], template="", template_values={}, **keys):
"""
Return a mail object produced based on collected events and session parameters.
The "events" parameter is None when we just want to test building a mail.
"""
if events is None:
events = []
csv = templates.CSVFormatter()
template_keys = {
"csv": csv,
"attach_csv": templates.AttachUnicode(csv),
"attach_and_embed_csv": templates.AttachAndEmbedUnicode(csv),
"attach_zip": templates.AttachZip(csv),
"to": templates.Const(format_addresses(to)),
"cc": templates.Const(format_addresses(cc)),
"bcc": templates.Const(format_addresses(bcc)),
}
for key, value in dict(template_values).iteritems():
template_keys[key] = templates.Event(value)
mail_template = MailTemplate(template, **template_keys)
msg = yield idiokit.thread(mail_template.format, events)
idiokit.stop(msg)
def report(self, eventlist, retries=None, to=[], cc=[], bcc=[], **keys):
if retries is None:
retries = self.max_retries
msg = yield self.build_mail(eventlist, to=to, cc=cc, bcc=bcc, **keys)
prep_recipient_header(msg, "to", to)
prep_recipient_header(msg, "cc", cc)
prep_recipient_header(msg, "bcc", bcc)
# FIXME: Use encoding after getaddresses
from_addr = getaddresses([self.mail_sender])[0]
if "from" not in msg:
msg["from"] = formataddr(from_addr)
subject = decode_subject(msg.get("subject", ""))
header_recipients = clean_recipients(msg.get_all("to", []) + msg.get_all("cc", []) + msg.get_all("bcc", []))
if self.mail_receiver_override is not None:
actual_recipients = clean_recipients(self.mail_receiver_override)
recipient_string = u"{actual_recipients} (overridden from {header_recipients})".format(
actual_recipients=format_recipients(actual_recipients),
header_recipients=format_recipients(header_recipients),
)
else:
actual_recipients = header_recipients
recipient_string = unicode(format_recipients(actual_recipients))
# No need to keep both the mail object and mail data in memory.
msg_data = msg.as_string()
del msg
event = events.Event(
{
"type": "mail",
"subject": subject,
"to": to,
"cc": cc,
"bcc": bcc,
"sender": from_addr[1],
"recipients": actual_recipients,
"event count": unicode(len(eventlist)),
}
)
sent = False
if not actual_recipients:
self.log.info(
u'Skipped message "{subject}": {recipients}'.format(subject=subject, recipients=recipient_string),
event=event.union(status="skipped (no recipients)"),
)
elif not eventlist:
self.log.info(
u'Skipped message "{subject}" to {recipients}: no events'.format(
subject=subject, recipients=recipient_string
),
event=event.union(status="skipped (no events)"),
)
else:
server = yield self._connect(self.smtp_host, self.smtp_port)
try:
yield self._login(server, self.smtp_auth_user, self.smtp_auth_password)
self.log.info(
u'Sending message "{subject}" to {recipients}'.format(subject=subject, recipients=recipient_string)
)
try:
yield idiokit.thread(server.sendmail, from_addr[1], actual_recipients, msg_data)
except smtplib.SMTPDataError as data_error:
self.log.error(
u"Could not send the message to {recipients}: {error}. Dropping message from queue".format(
recipients=recipient_string, error=utils.format_exception(data_error)
)
)
except smtplib.SMTPRecipientsRefused as refused:
for recipient, reason in refused.recipients.iteritems():
self.log.error(
u"Could not the send message to {recipients}: {error}. Dropping message from queue".format(
recipients=recipient_string, error=utils.format_exception(reason)
)
)
except (socket.error, smtplib.SMTPException) as exc:
self.log.error(
u"Could not send the message to {recipients}: {error}".format(
recipients=recipient_string, error=utils.format_exception(exc)
)
)
if retries >= 1:
self.log.info(u"Retrying sending in 60 seconds")
self.requeue(60.0, retries=retries - 1)
else:
self.log.error(u"Failed all retries, dropping the mail from the queue")
else:
sent = True
self.log.info(
u'Sent message "{subject}" to {recipients}'.format(
subject=subject, recipients=recipient_string
),
event=event.union(status="sent"),
)
finally:
yield idiokit.thread(server.quit)
idiokit.stop(sent)
def fetch_url(url,
opener=None,
timeout=60.0,
chunk_size=16384,
cookies=None,
auth=None,
cert=None,
verify=True,
proxies=None):
if opener is not None:
raise TypeError("'opener' argument is no longer supported")
handlers = [
_CustomHTTPSHandler(cert=cert, verify=verify),
urllib2.ProxyHandler(proxies)
]
if cookies is not None:
handlers.append(urllib2.HTTPCookieProcessor(cookies))
if auth is not None:
username, password = auth
passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
if isinstance(url, urllib2.Request):
uri = url.get_full_url()
passmgr.add_password(None, uri, username, password)
else:
passmgr.add_password(None, url, username, password)
handlers.append(urllib2.HTTPBasicAuthHandler(passmgr))
opener = urllib2.build_opener(*handlers)
try:
output = StringIO()
fileobj = yield idiokit.thread(opener.open, url, timeout=timeout)
try:
while True:
data = yield idiokit.thread(fileobj.read, chunk_size)
if not data:
break
output.write(data)
finally:
fileobj.close()
info = fileobj.info()
info = email.parser.Parser().parsestr(str(info), headersonly=True)
output.seek(0)
idiokit.stop(info, output)
except urllib2.HTTPError as he:
raise HTTPError(he.code, he.msg, he.hdrs, he.fp)
except urllib2.URLError as error:
if _is_timeout(error.reason):
raise FetchUrlTimeout("fetching URL timed out")
raise FetchUrlFailed(str(error))
except socket.error as error:
if _is_timeout(error):
raise FetchUrlTimeout("fetching URL timed out")
raise FetchUrlFailed(str(error))
except CertificateError as error:
raise SSLCertificateError(str(error))
except httplib.HTTPException as error:
raise FetchUrlFailed(str(error))
def report(self, eventlist, retries=None, to=[], cc=[], bcc=[], **keys):
if retries is None:
retries = self.max_retries
msg = yield self.build_mail(eventlist, to=to, cc=cc, bcc=bcc, **keys)
prep_recipient_header(msg, "to", to)
prep_recipient_header(msg, "cc", cc)
prep_recipient_header(msg, "bcc", bcc)
# FIXME: Use encoding after getaddresses
from_addr = getaddresses([self.mail_sender])[0]
if "from" not in msg:
msg["from"] = formataddr(from_addr)
subject = decode_subject(msg.get("subject", ""))
header_recipients = clean_recipients(msg.get_all("to", []) + msg.get_all("cc", []) + msg.get_all("bcc", []))
if self.mail_receiver_override is not None:
actual_recipients = clean_recipients(self.mail_receiver_override)
recipient_string = u"{actual_recipients} (overridden from {header_recipients})".format(
actual_recipients=format_recipients(actual_recipients),
header_recipients=format_recipients(header_recipients)
)
else:
actual_recipients = header_recipients
recipient_string = unicode(format_recipients(actual_recipients))
# No need to keep both the mail object and mail data in memory.
msg_data = msg.as_string()
del msg
event = events.Event({
"type": "mail",
"subject": subject,
"to": to,
"cc": cc,
"bcc": bcc,
"sender": from_addr[1],
"recipients": actual_recipients,
"event count": unicode(len(eventlist))
})
sent = False
if not actual_recipients:
self.log.info(
u"Skipped message \"{subject}\": {recipients}".format(
subject=subject,
recipients=recipient_string
),
event=event.union(status="skipped (no recipients)")
)
elif not eventlist:
self.log.info(
u"Skipped message \"{subject}\" to {recipients}: no events".format(
subject=subject,
recipients=recipient_string
),
event=event.union(status="skipped (no events)")
)
else:
server = yield self._connect(self.smtp_host, self.smtp_port)
try:
yield self._login(server, self.smtp_auth_user, self.smtp_auth_password)
self.log.info(u"Sending message \"{subject}\" to {recipients}".format(
subject=subject,
recipients=recipient_string
))
try:
yield idiokit.thread(server.sendmail, from_addr[1], actual_recipients, msg_data)
except smtplib.SMTPDataError as data_error:
self.log.error(u"Could not send the message to {recipients}: {error}. Dropping message from queue".format(
recipients=recipient_string,
error=utils.format_exception(data_error)
))
except smtplib.SMTPRecipientsRefused as refused:
for recipient, reason in refused.recipients.iteritems():
self.log.error(u"Could not the send message to {recipients}: {error}. Dropping message from queue".format(
recipients=recipient_string,
error=utils.format_exception(reason)
))
except (socket.error, smtplib.SMTPException) as exc:
self.log.error(u"Could not send the message to {recipients}: {error}".format(
recipients=recipient_string,
error=utils.format_exception(exc)
))
if retries >= 1:
self.log.info(u"Retrying sending in 60 seconds")
self.requeue(60.0, retries=retries - 1)
else:
self.log.error(u"Failed all retries, dropping the mail from the queue")
else:
sent = True
self.log.info(
u"Sent message \"{subject}\" to {recipients}".format(
subject=subject,
recipients=recipient_string
),
event=event.union(status="sent")
)
finally:
yield idiokit.thread(server.quit)
idiokit.stop(sent)
def fetch_url(
url,
opener=None,
timeout=60.0,
chunk_size=16384,
cookies=None,
auth=None,
cert=None,
verify=True,
proxies=None
):
if opener is not None:
raise TypeError("'opener' argument is no longer supported")
handlers = [
_CustomHTTPSHandler(cert=cert, verify=verify),
urllib2.ProxyHandler(proxies)
]
if cookies is not None:
handlers.append(urllib2.HTTPCookieProcessor(cookies))
if auth is not None:
username, password = auth
passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
if isinstance(url, urllib2.Request):
uri = url.get_full_url()
passmgr.add_password(None, uri, username, password)
else:
passmgr.add_password(None, url, username, password)
handlers.append(urllib2.HTTPBasicAuthHandler(passmgr))
opener = urllib2.build_opener(*handlers)
try:
output = StringIO()
fileobj = yield idiokit.thread(opener.open, url, timeout=timeout)
try:
while True:
data = yield idiokit.thread(fileobj.read, chunk_size)
if not data:
break
output.write(data)
finally:
fileobj.close()
info = fileobj.info()
info = email.parser.Parser().parsestr(str(info), headersonly=True)
output.seek(0)
idiokit.stop(info, output)
except urllib2.HTTPError as he:
raise HTTPError(he.code, he.msg, he.hdrs, he.fp)
except urllib2.URLError as error:
if _is_timeout(error.reason):
raise FetchUrlTimeout("fetching URL timed out")
raise FetchUrlFailed(str(error))
except socket.error as error:
if _is_timeout(error):
raise FetchUrlTimeout("fetching URL timed out")
raise FetchUrlFailed(str(error))
except httplib.HTTPException as error:
raise FetchUrlFailed(str(error))
