59
def _parse_message(message):
60
"""Get a date and dict of an RFC822 message."""
61
parsed = email.parser.Parser().parsestr(message)
64
for key in ('from', 'to', 'subject', 'message-id'):
65
value = parsed.get(key, None)
67
message_dict[key] = value
69
date = parsed.get('date')
72
pdate = parsedate_tz(date)
73
date = datetime.datetime(
75
tzinfo=dateutil.tz.tzoffset('', pdate[9]))
78
message_dict['date'] = date.isoformat() if date is not None else None
80
return date, message_dict
83
def _utc_datetime(dt):
84
return dt.astimezone(dateutil.tz.tzutc())
87
def _utc_timestamp(dt):
88
return time.mktime(_utc_datetime(dt).timetuple()) - time.timezone
91
def _utc_timeuuid(dt, lowest_val=True):
92
return convert_time_to_uuid(_utc_timestamp(dt), lowest_val)
95
def _cmp_timeuuid(a, b):
97
return cmp(a.time, b.time)
101
def _bound_timeuuid(a, b, max=False):
102
if a == '' or _cmp_timeuuid(b, a) == (1 if max else -1):
107
def _make_bounds(memo, range_start, range_finish, backward):
110
memo = uuid.UUID(memo)
115
if range_start is not None:
116
start = _bound_timeuuid(
117
start, _utc_timeuuid(range_start), max=True)
118
if range_finish is not None:
119
finish = _bound_timeuuid(
120
finish, _utc_timeuuid(range_finish, lowest_val=False))
121
return memo, start, finish
54
124
class CassandraConnection(object):
56
126
def __init__(self, keyspace, host):
69
139
def add_message(self, archive_uuid, message):
70
140
message_uuid = uuid.uuid4()
71
parsed = email.parser.Parser().parsestr(message)
72
date = parsed.get('date')
75
pdate = parsedate_tz(date)
76
date = datetime.datetime(
78
tzinfo=dateutil.tz.tzoffset('', pdate[9]))
83
{'date_created': datetime.datetime.utcnow().isoformat() + 'Z',
85
'from': parsed.get('From'),
86
'to': parsed.get('To'),
87
'subject': parsed.get('Subject'),
88
'date': date.isoformat() if date is not None else None,
89
'message_id': parsed.get('Message-ID'),
141
message_date, message_dict = _parse_message(message)
142
message_dict['content'] = message
143
message_dict['date_created'] = (
144
datetime.datetime.utcnow().isoformat() + 'Z')
145
self.messages.insert(message_uuid, message_dict)
91
146
self.archive_messages.insert(
93
{date.astimezone(dateutil.tz.tzutc()): message_uuid})
148
{_utc_timestamp(message_date): message_uuid})
95
'Imported %s into %s' % (parsed.get('Message-ID'), archive_uuid))
150
'Imported %s into %s'
151
% (message_dict.get('message-id', None), archive_uuid))
96
152
return message_uuid
98
154
def _format_message(self, message):
100
'date': message['date'],
101
'from': message['from'],
102
'subject': message['subject'],
156
'date': message.get('date'),
157
'from': message.get('from'),
158
'subject': message.get('subject'),
159
'message-id': message.get('message-id'),
105
def get_messages(self, archive_uuid, order, count, start):
162
def _trim(self, sequence, end):
163
"""Return the sequence with one of the ends trimmed.
165
:param end: if true, remove the last element. otherwise remove
173
def get_messages(self, archive_uuid, order, count, memo, backward=False,
174
start_date=None, finish_date=None):
106
175
if order in ("date", "-date"):
107
176
reversed = order[0] == '-'
109
178
raise AssertionError("Unsupported order.")
180
memo, start, finish = _make_bounds(
181
memo, start_date, finish_date, backward)
183
# Get up to n+1 messages from the memo: the last item of the
184
# previous batch (because that's where the memo starts) + this
110
186
pairs = self.archive_messages.get(
111
archive_uuid, column_count=count + 1,
112
column_start=start, column_reversed=reversed).items()
187
archive_uuid, column_count=count + 1, column_start=start,
188
column_finish=finish, column_reversed=reversed).items()
190
if len(pairs) and memo and pairs[0][0] <= memo:
191
# The memo (from the previous batch) was included in the result.
193
pairs = self._trim(pairs, False ^ backward)
194
elif len(pairs) > count:
195
# There was no memo in the result, so the n+1th element is
196
# unnecessary. Kill it.
197
pairs = self._trim(pairs, True ^ backward)
200
return (None, [], None)
202
assert 0 < len(pairs) <= count
204
# We've narrowed down the message references. Fetch the messages.
113
205
ids = [v for k, v in pairs]
114
206
messages = self.messages.multiget(
115
ids, columns=['date', 'from', 'subject'])
116
actual_count = len(pairs)
117
if len(pairs) > count:
118
assert len(pairs) == count + 1
120
next_memo = str(pairs[count][0])
207
ids, columns=['date', 'from', 'subject', 'message-id'])
124
[self._format_message(messages[id]) for id in ids[:actual_count]],
211
[self._format_message(messages[id]) for id in ids],