def getdate(msg):
if not 'date' in msg:
return None
datehdr = msg['date'].strip()
try:
return Utils.mktime_tz(Utils.parsedate_tz(datehdr))
except:
return None
def p_http(t):
from email import Utils
import time
tt = Utils.parsedate_tz(t)
ts = time.mktime(tt[:9])
utc = 0 # (datetime.utcnow() - datetime.now()).seconds
if tt[9] is not None:
utc += tt[9]
return datetime.fromtimestamp(ts - utc)
def getdate(msg):
if not 'date' in msg:
return None
datehdr = msg['date'].strip()
try:
return Utils.mktime_tz(Utils.parsedate_tz(datehdr))
except:
return None
def getdate(msg):
"""Returns the date/time from msg in seconds-since-epoch, if possible.
Otherwise, returns None."""
if not 'date' in msg:
# No Date header present.
return None
datehdr = msg['date'].strip()
try:
return Utils.mktime_tz(Utils.parsedate_tz(datehdr))
except:
# Some sort of error occured, likely because of an invalid date.
return None
def getdate(msg):
"""Returns the date/time from msg in seconds-since-epoch, if possible.
Otherwise, returns None."""
if not 'date' in msg:
# No Date header present.
return None
datehdr = msg['date'].strip()
try:
return Utils.mktime_tz(Utils.parsedate_tz(datehdr))
except:
# Some sort of error occured, likely because of an invalid date.
return None
def __getMailMessage(self, addNewLine=False):
view = self.view
m = Mail.MailMessage(itsView=view)
m.fromAddress = Mail.EmailAddress.getEmailAddress(view, "*****@*****.**")
toOne = Mail.EmailAddress.getEmailAddress(view, "*****@*****.**")
toTwo = Mail.EmailAddress.getEmailAddress(view, "*****@*****.**", "John Johnson")
m.toAddress = []
m.toAddress.append(toOne)
m.toAddress.append(toTwo)
ccOne = Mail.EmailAddress.getEmailAddress(view, "*****@*****.**")
m.ccAddress = []
m.ccAddress.append(ccOne)
m.subject = "test mail"
if addNewLine:
m.subject += "\n"
m.headers['Content-Length'] = "75"
m.headers['Content-Type'] = "text/plain; charset=utf-8; format=flowed"
m.headers['Content-Transfer-Encoding'] = "7bit"
m.headers['Mime-Version'] = "1.0"
m.headers['Received'] = "from [192.168.101.37] (w002.z065106067.sjc-ca.dsl.cnc.net [65.106.67.2]) by kahuna.test.com (8.12.8/8.12.8) with ESMTP id i7GKWWpo017020; Mon, 16 Aug 2004 13:32:32 -0700"
m.headers['References'] = "<*****@*****.**> <*****@*****.**> <*****@*****.**> <*****@*****.**>"
dateString = "Mon, 9 Aug 2004 13:55:15 -0700"
m.dateSent = datetime.fromtimestamp(emailUtils.mktime_tz(emailUtils.parsedate_tz(dateString)), view.tzinfo.getInstance("Etc/GMT-7"))
m.dateSentString = dateString
m.itsItem.body = u"This is the body"
if not IGNORE_ATTACHMENTS:
m.rfc2822Message = utils.dataToBinary(m, "rfc2822Message", self.__mail)
self.__mailMessage = m
return self.__mailMessage
def RFC2822DateToDatetime(dateString, tz=None):
"""
Parses an RFC 2822 date value (e.g. in the Date: header) and
returns a C{datetime.datetime} value. Uses C{email.utils.parsedate_tz},
so that a variety of date formats (not all required by the RFC) can
be parsed.
@param dateString: Input date value
@type dateString: C{basestring}
@param tz: The timezone to use for the return value. If C{None}, and the
input header specified a timezone offset, the returned
C{datetime} will be a UTC value. If C{None}, and no timezone
offset was specified, the returned value will be as parsed.
If a non-C{None} C{tz} is passed in, the returned datetime will
have a C{tzinfo} of C{tz}.
@type tz: C{datetime.tzinfo}
@rtype: C{datetime.datetime}
"""
parsed = Utils.parsedate_tz(dateString)
if parsed is None:
return None
result = datetime(*parsed[:6])
if parsed[9] is not None:
# i.e. parsed[9] is an offset. Let's convert to UTC
result -= timedelta(seconds=parsed[9])
if tz is not None:
result = result.replace(tzinfo=PyICU.ICUtzinfo.getInstance("UTC"))
result = result.astimezone(tz)
else: # (shrug) no tz
result = result.replace(tzinfo=tz)
return result
def RFC2822DateToDatetime(dateString, tz=None):
"""
Parses an RFC 2822 date value (e.g. in the Date: header) and
returns a C{datetime.datetime} value. Uses C{email.utils.parsedate_tz},
so that a variety of date formats (not all required by the RFC) can
be parsed.
@param dateString: Input date value
@type dateString: C{basestring}
@param tz: The timezone to use for the return value. If C{None}, and the
input header specified a timezone offset, the returned
C{datetime} will be a UTC value. If C{None}, and no timezone
offset was specified, the returned value will be as parsed.
If a non-C{None} C{tz} is passed in, the returned datetime will
have a C{tzinfo} of C{tz}.
@type tz: C{datetime.tzinfo}
@rtype: C{datetime.datetime}
"""
parsed = Utils.parsedate_tz(dateString)
if parsed is None:
return None
result = datetime(*parsed[:6])
if parsed[9] is not None:
# i.e. parsed[9] is an offset. Let's convert to UTC
result -= timedelta(seconds=parsed[9])
if tz is not None:
result = result.replace(tzinfo=PyICU.ICUtzinfo.getInstance("UTC"))
result = result.astimezone(tz)
else: # (shrug) no tz
result = result.replace(tzinfo=tz)
return result
def init_headers(self):
from email import Header, Errors, Utils
self.headers = []
self.title = None
for h in self.msg.keys():
for s in (self.msg.get_all(h) or []):
if not s: continue
v = decode_header(s)
self.headers.append((h, v))
if h.lower() == 'date':
try:
self.mtime = int(Utils.mktime_tz(Utils.parsedate_tz(v)))
except (TypeError, ValueError, OverflowError):
pass
elif h.lower() == 'subject':
self.title = v
if not self.title:
fname = self.msg.get_filename()
if fname:
self.title = decode_header(fname)
return
def update_list (mlist, **kw):
archive = mlist.data.get ('list_archive')
mboxes = []
if archive != None:
links = LinkExtractor (archive).get_links ()
for link in links:
if link.endswith ('.txt.gz'):
mboxes.append (urlparse.urljoin (archive, link))
mlist.data.setdefault ('archives', {})
for url in mboxes:
parsed = urlparse.urlparse (url)
con = httplib.HTTPConnection (parsed[1])
headers = {'Accept-encoding': 'gzip'}
if kw.get('timestamps', True):
dats = mlist.data['archives'].get (url, {})
mod = dats.get ('modified')
if mod != None:
headers['If-Modified-Since'] = mod
etag = dats.get ('etag')
if etag != None:
headers['Etag'] =etag
con.request ('GET', parsed[2], None, headers)
res = con.getresponse ()
if res.status != 200:
continue
urlfile = url
urlslash = urlfile.rfind ('/')
if urlslash >= 0:
urlfile = urlfile[urlslash+1:]
pulse.utils.log ('Processing list archive %s %s' % (mlist.ident, urlfile))
urlbase = url[:-7] + '/'
tmp = pulse.utils.tmpfile()
fd = open (tmp, 'w')
fd.write (gzip.GzipFile (fileobj=StringIO.StringIO (res.read ())).read ())
fd.close ()
mbox = mailbox.PortableUnixMailbox (open(tmp, 'rb'))
i = -1
for msg in mbox:
i += 1
msgfrom = msgdate = msgsubject = msgid = msgparent = None
for hdr in msg.headers:
lower = hdr.lower()
if lower.startswith ('from:'):
msgfrom = hdr[5:]
elif lower.startswith ('date:'):
msgdate = hdr[5:]
elif lower.startswith ('subject:'):
msgsubject = hdr[8:].strip()
elif lower.startswith ('message-id:'):
msgid = hdr[11:]
elif lower.startswith ('in-reply-to:'):
msgparent = hdr[12:]
if msgid == None:
continue
msgid = pulse.utils.utf8dec (emailutils.parseaddr (msgid)[1])
ident = mlist.ident + u'/' + msgid
post = pulse.db.ForumPost.get_or_create (ident, u'ListPost')
postdata = {'forum': mlist, 'name': msgsubject}
if msgparent != None:
msgparent = pulse.utils.utf8dec (emailutils.parseaddr (msgparent)[1])
pident = mlist.ident + '/' + msgparent
parent = pulse.db.ForumPost.get_or_create (pident, u'ListPost')
parent.forum = mlist
postdata['parent_ident'] = parent.ident
msgfrom = emailutils.parseaddr (msgfrom)
personident = u'/person/' + pulse.utils.utf8dec (msgfrom[1])
person = pulse.db.Entity.get_or_create (personident, u'Person')
person.extend (name=msgfrom[0])
postdata['author_ident'] = person.ident
if person.ident != personident:
postdata['alias_ident'] = personident
pulse.db.Queue.push (u'people', person.ident)
msgdate = emailutils.parsedate_tz (msgdate)
try:
dt = datetime.datetime (*msgdate[:6])
if msgdate[-1] != None:
dt = dt + datetime.timedelta (seconds=msgdate[-1])
except:
dt = None
postdata['datetime'] = dt
postdata['weeknum'] = pulse.utils.weeknum (dt)
postdata['web'] = urlbase + 'msg%05i.html' % i
post.update (postdata)
try:
os.remove (tmp)
except:
pass
mlist.data['archives'].setdefault (url, {})
mlist.data['archives'][url]['modified'] = res.getheader ('Last-Modified')
mlist.data['archives'][url]['etag'] = res.getheader ('Etag')
update_graphs (mlist, 100, **kw)
def get_message_date(msg):
try:
return int(Utils.mktime_tz(Utils.parsedate_tz(msg['date'])))
except:
return 0
self.method = "If-modified-since"
self.ratio = 0
return
else:
self.method = "Error %d" % err.code
return
except urllib2.URLError:
self.method = "Error URL"
return
except socket.timeout:
self.method = "Error timeout"
return
if "last-modified" in f.info():
date = Utils.mktime_tz(
Utils.parsedate_tz(f.info()["last-modified"]))
if date == self.date:
self.method = "Last-modified"
self.ratio = 0
return
else:
date = time.time()
if "content-length" in f.info():
length = f.info()["content-length"]
if length == self.length:
self.method = "Content-length"
self.ratio = 0
return
else:
self.length = length
print '***Header in massage: '
for header, value in msg.items():
print header + ':'
print " " + value
if msg.is_multipart():
print 'this program can not handler MIME multipart message!'
sys.exit(-1)
#特定头
print '-'*60
if 'subject' in msg:
print 'Subject: ', msg['subject']
print '-'*60
print 'Message Body: \n'
print msg.get_payload()
#解析日期
print '-'*60
if 'date' not in msg:
exit(-1)
datehdr = msg['date'].strip()
print datehdr
dateval = Utils.mktime_tz(Utils.parsedate_tz(datehdr))
print dateval
print 'Message was sent on', time.strftime("%A, %B %d %Y at %I:%M %p", time.localtime(dateval))
def get_message_date(msg):
try:
return int(Utils.mktime_tz(Utils.parsedate_tz(msg["date"])))
except:
return 0
