aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/vk/tracking
diff options
context:
space:
mode:
authorEgor Tensin <Egor.Tensin@gmail.com>2016-06-18 23:56:48 +0300
committerEgor Tensin <Egor.Tensin@gmail.com>2016-06-18 23:56:48 +0300
commit48b19f2c9fec8a7e2f26ddd058c794d4f5a33894 (patch)
tree23f397fdad7a38ea2f7d3890593fddedd0572867 /vk/tracking
parentREADME update (diff)
downloadvk-scripts-48b19f2c9fec8a7e2f26ddd058c794d4f5a33894.tar.gz
vk-scripts-48b19f2c9fec8a7e2f26ddd058c794d4f5a33894.zip
vk.utils.tracking -> vk.tracking
Diffstat (limited to 'vk/tracking')
-rw-r--r--vk/tracking/__init__.py8
-rw-r--r--vk/tracking/db/__init__.py7
-rw-r--r--vk/tracking/db/backend/__init__.py5
-rw-r--r--vk/tracking/db/backend/csv.py63
-rw-r--r--vk/tracking/db/backend/log.py75
-rw-r--r--vk/tracking/db/backend/null.py37
-rw-r--r--vk/tracking/db/format.py35
-rw-r--r--vk/tracking/db/record.py98
-rw-r--r--vk/tracking/db/timestamp.py34
-rw-r--r--vk/tracking/online_streaks.py140
-rw-r--r--vk/tracking/status_tracker.py99
11 files changed, 601 insertions, 0 deletions
diff --git a/vk/tracking/__init__.py b/vk/tracking/__init__.py
new file mode 100644
index 0000000..0f9c422
--- /dev/null
+++ b/vk/tracking/__init__.py
@@ -0,0 +1,8 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from .online_streaks import OnlineStreakEnumerator
+from .status_tracker import StatusTracker
+
+__all__ = 'online_streaks', 'status_tracker',
diff --git a/vk/tracking/db/__init__.py b/vk/tracking/db/__init__.py
new file mode 100644
index 0000000..9e6b74a
--- /dev/null
+++ b/vk/tracking/db/__init__.py
@@ -0,0 +1,7 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from .format import Format
+
+__all__ = 'format'
diff --git a/vk/tracking/db/backend/__init__.py b/vk/tracking/db/backend/__init__.py
new file mode 100644
index 0000000..4b3c278
--- /dev/null
+++ b/vk/tracking/db/backend/__init__.py
@@ -0,0 +1,5 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+__all__ = 'csv', 'log', 'null'
diff --git a/vk/tracking/db/backend/csv.py b/vk/tracking/db/backend/csv.py
new file mode 100644
index 0000000..10a504f
--- /dev/null
+++ b/vk/tracking/db/backend/csv.py
@@ -0,0 +1,63 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from collections.abc import Iterable
+import csv
+
+from ..record import Record
+from ..timestamp import Timestamp
+
+class Writer:
+ def __init__(self, fd):
+ self._fd = fd
+ self._writer = csv.writer(fd, lineterminator='\n')
+
+ def __enter__(self):
+ return self
+
+ def __exit__(self, *args):
+ pass
+
+ def on_initial_status(self, user):
+ self._write_record(user)
+ self._fd.flush()
+
+ def on_status_update(self, user):
+ self._write_record(user)
+ self._fd.flush()
+
+ def on_connection_error(self, e):
+ pass
+
+ def _write_record(self, user):
+ if not self:
+ return
+ self._write_row(self._record_to_row(Record.from_user(user)))
+
+ def _write_row(self, row):
+ self._writer.writerow(row)
+
+ @staticmethod
+ def _record_to_row(record):
+ return [str(record.get_timestamp())] + [str(record[field]) for field in record]
+
+class Reader(Iterable):
+ def __init__(self, fd):
+ self._reader = csv.reader(fd)
+
+ def __enter__(self):
+ return self
+
+ def __exit__(self, *args):
+ pass
+
+ def __iter__(self):
+ return map(Reader._record_from_row, self._reader)
+
+ @staticmethod
+ def _record_from_row(row):
+ record = Record(Timestamp.from_string(row[0]))
+ for i in range(len(Record.FIELDS)):
+ record[Record.FIELDS[i]] = row[i + 1]
+ return record
diff --git a/vk/tracking/db/backend/log.py b/vk/tracking/db/backend/log.py
new file mode 100644
index 0000000..625257b
--- /dev/null
+++ b/vk/tracking/db/backend/log.py
@@ -0,0 +1,75 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+import logging
+
+class Writer:
+ def __init__(self, fd):
+ self._logger = logging.getLogger(__file__)
+ self._logger.setLevel(logging.INFO)
+ handler = logging.StreamHandler(fd)
+ handler.setFormatter(logging.Formatter(
+ fmt='[%(asctime)s] %(message)s',
+ datefmt='%Y-%m-%d %H:%M:%S'))
+ self._logger.addHandler(handler)
+
+ def __enter__(self):
+ return self
+
+ def __exit__(self, *args):
+ pass
+
+ def info(self, msg):
+ self._logger.info(msg)
+
+ def exception(self, e):
+ self._logger.exception(e)
+
+ def on_initial_status(self, user):
+ if user.is_online():
+ self.info(self._format_user_is_online(user))
+ else:
+ self.info(self._format_user_is_offline(user))
+ self.info(self._format_user_last_seen(user))
+
+ def on_status_update(self, user):
+ if user.is_online():
+ self.info(self._format_user_went_online(user))
+ else:
+ self.info(self._format_user_went_offline(user))
+ self.info(self._format_user_last_seen(user))
+
+ def on_connection_error(self, e):
+ #self.exception(e)
+ pass
+
+ @staticmethod
+ def _format_user(user):
+ if user.has_last_name():
+ return '{} {}'.format(user.get_first_name(), user.get_last_name())
+ else:
+ return '{}'.format(user.get_first_name())
+
+ @staticmethod
+ def _format_user_is_online(user):
+ return '{} is ONLINE.'.format(Writer._format_user(user))
+
+ @staticmethod
+ def _format_user_is_offline(user):
+ return '{} is OFFLINE.'.format(Writer._format_user(user))
+
+ @staticmethod
+ def _format_user_last_seen(user):
+ return '{} was last seen at {} using {}.'.format(
+ Writer._format_user(user),
+ user.get_last_seen_time_local(),
+ user.get_last_seen_platform().get_description_for_sentence())
+
+ @staticmethod
+ def _format_user_went_online(user):
+ return '{} went ONLINE.'.format(Writer._format_user(user))
+
+ @staticmethod
+ def _format_user_went_offline(user):
+ return '{} went OFFLINE.'.format(Writer._format_user(user))
diff --git a/vk/tracking/db/backend/null.py b/vk/tracking/db/backend/null.py
new file mode 100644
index 0000000..139a9f0
--- /dev/null
+++ b/vk/tracking/db/backend/null.py
@@ -0,0 +1,37 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from collections.abc import Iterable
+
+class Writer:
+ def __init__(self, fd):
+ pass
+
+ def __enter__(self):
+ return self
+
+ def __exit__(self, *args):
+ pass
+
+ def on_initial_status(self, user):
+ pass
+
+ def on_status_update(self, user):
+ pass
+
+ def on_connection_error(self, e):
+ pass
+
+class Reader(Iterable):
+ def __init__(self, fd):
+ pass
+
+ def __enter__(self):
+ return self
+
+ def __exit__(self, *args):
+ pass
+
+ def __iter__(self):
+ pass
diff --git a/vk/tracking/db/format.py b/vk/tracking/db/format.py
new file mode 100644
index 0000000..4856094
--- /dev/null
+++ b/vk/tracking/db/format.py
@@ -0,0 +1,35 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from enum import Enum
+
+from .backend import *
+
+class Format(Enum):
+ CSV = 'csv'
+ LOG = 'log'
+ NULL = 'null'
+
+ def create_writer(self, fd):
+ if self is Format.CSV:
+ return csv.Writer(fd)
+ elif self is Format.LOG:
+ return log.Writer(fd)
+ elif self is Format.NULL:
+ return null.Writer(fd)
+ else:
+ raise NotImplementedError('unsupported database format: ' + str(self))
+
+ def create_reader(self, fd):
+ if self is Format.CSV:
+ return csv.Reader(fd)
+ elif self is Format.LOG:
+ raise NotImplementedError()
+ elif self is Format.NULL:
+ return null.Reader(fd)
+ else:
+ raise NotImplementedError('unsupported database format: ' + str(self))
+
+ def __str__(self):
+ return self.value
diff --git a/vk/tracking/db/record.py b/vk/tracking/db/record.py
new file mode 100644
index 0000000..93be97c
--- /dev/null
+++ b/vk/tracking/db/record.py
@@ -0,0 +1,98 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from collections import OrderedDict
+from collections.abc import MutableMapping
+from datetime import datetime
+
+from .timestamp import Timestamp
+from vk.user import LastSeen, LastSeenField, User, UserField
+
+class Record(MutableMapping):
+ FIELDS = (
+ UserField.UID,
+ UserField.FIRST_NAME,
+ UserField.LAST_NAME,
+ UserField.SCREEN_NAME,
+ UserField.ONLINE,
+ LastSeenField.TIME,
+ LastSeenField.PLATFORM,
+ )
+
+ def __init__(self, timestamp=None, fields=None):
+ if timestamp is None:
+ timestamp = Timestamp()
+ if fields is None:
+ fields = OrderedDict()
+ self._timestamp = timestamp
+ self._fields = fields
+
+ def __getitem__(self, field):
+ if field is LastSeenField.TIME:
+ return Timestamp(self._fields[field])
+ return self._fields[field]
+
+ def __setitem__(self, field, value):
+ if field is LastSeenField.TIME:
+ if isinstance(value, str):
+ value = Timestamp.from_string(value).dt
+ elif isinstance(value, Timestamp):
+ value = value.dt
+ elif isinstance(value, datetime):
+ pass
+ else:
+ raise TypeError()
+ if isinstance(field, LastSeenField):
+ self._fields[field] = LastSeen.parse(field, value)
+ elif isinstance(field, UserField):
+ self._fields[field] = User.parse(field, value)
+ else:
+ raise TypeError()
+
+ def __delitem__(self, field):
+ del self._fields[field]
+
+ def __iter__(self):
+ return iter(self._fields)
+
+ def __len__(self):
+ return len(self._fields)
+
+ def get_timestamp(self):
+ return self._timestamp
+
+ @staticmethod
+ def from_user(user):
+ record = Record()
+ for field in Record.FIELDS:
+ if isinstance(field, UserField):
+ record[field] = user[field]
+ elif isinstance(field, LastSeenField):
+ record[field] = user.get_last_seen()[field]
+ else:
+ assert False
+ return record
+
+ def _update_last_seen_field(self, last_seen, field):
+ if field is LastSeenField.TIME:
+ last_seen[field] = self[field].dt
+ else:
+ last_seen[field] = self[field]
+
+ def _update_user_field(self, user, field):
+ user[field] = self[field]
+
+ def to_user(self):
+ user = User()
+ last_seen = LastSeen()
+ for field in self:
+ if isinstance(field, LastSeenField):
+ self._update_last_seen_field(last_seen, field)
+ elif isinstance(field, UserField):
+ self._update_user_field(user, field)
+ else:
+ assert False
+ if len(last_seen):
+ user.set_last_seen(last_seen)
+ return user
diff --git a/vk/tracking/db/timestamp.py b/vk/tracking/db/timestamp.py
new file mode 100644
index 0000000..35cf5b3
--- /dev/null
+++ b/vk/tracking/db/timestamp.py
@@ -0,0 +1,34 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from datetime import datetime, timezone
+
+class Timestamp:
+ @staticmethod
+ def _new():
+ return datetime.utcnow()
+
+ @staticmethod
+ def _is_timezone_aware(dt):
+ return dt.tzinfo is not None and dt.tzinfo.utcoffset(dt) is not None
+
+ @staticmethod
+ def _lose_timezone(dt):
+ if Timestamp._is_timezone_aware(dt):
+ return dt.astimezone(timezone.utc).replace(tzinfo=None)
+ return dt
+
+ def __init__(self, dt=None):
+ if dt is None:
+ dt = self._new()
+ dt = dt.replace(microsecond=0)
+ dt = self._lose_timezone(dt)
+ self.dt = dt
+
+ @staticmethod
+ def from_string(s):
+ return Timestamp(datetime.strptime(s, '%Y-%m-%dT%H:%M:%SZ'))
+
+ def __str__(self):
+ return self.dt.isoformat() + 'Z'
diff --git a/vk/tracking/online_streaks.py b/vk/tracking/online_streaks.py
new file mode 100644
index 0000000..5c9aa48
--- /dev/null
+++ b/vk/tracking/online_streaks.py
@@ -0,0 +1,140 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from collections import OrderedDict
+from collections.abc import MutableMapping
+from datetime import timedelta
+from enum import Enum
+
+from vk.user import User
+
+class Weekday(Enum):
+ MONDAY = 0
+ TUESDAY = 1
+ WEDNESDAY = 2
+ THURSDAY = 3
+ FRIDAY = 4
+ SATURDAY = 5
+ SUNDAY = 6
+
+ def __str__(self):
+ return self.name[0] + self.name[1:].lower()
+
+class OnlineStreakEnumerator(MutableMapping):
+ def __init__(self, date_from=None, date_to=None):
+ self._records = {}
+ self._date_from = date_from
+ self._date_to = date_to
+
+ def __getitem__(self, user):
+ return self._records[user]
+
+ def __setitem__(self, user, record):
+ self._records[user] = record
+
+ def __delitem__(self, user):
+ del self._records[user]
+
+ def __iter__(self):
+ return iter(self._records)
+
+ def __len__(self):
+ return len(self._records)
+
+ def _cut_period(self, streak):
+ user, time_from, time_to = streak
+ #print(user.get_first_name(), time_from, self._date_from)
+ if self._date_from is not None:
+ if time_to < self._date_from:
+ #print(1)
+ return None
+ if time_from < self._date_from:
+ #print(2)
+ time_from = self._date_from
+ if self._date_to is not None:
+ if time_from > self._date_to:
+ #print(3)
+ return None
+ if time_to > self._date_to:
+ #print(4)
+ time_to = self._date_to
+ return user, time_from, time_to
+
+ def enum(self, db_reader):
+ for record in db_reader:
+ streak = self._insert_record(record)
+ if streak is not None:
+ streak = self._cut_period(streak)
+ if streak is not None:
+ yield streak
+
+ def group_by_user(self, db_reader):
+ by_user = {}
+ for user, time_from, time_to in self.enum(db_reader):
+ if user not in by_user:
+ by_user[user] = timedelta()
+ by_user[user] += time_to - time_from
+ return by_user
+
+ def group_by_date(self, db_reader):
+ by_date = OrderedDict()
+ for _, time_from, time_to in self.enum(db_reader):
+ for date, duration in self._enum_dates_and_durations(time_from, time_to):
+ if date not in by_date:
+ by_date[date] = timedelta()
+ by_date[date] += duration
+ return by_date
+
+ def group_by_weekday(self, db_reader):
+ by_weekday = OrderedDict()
+ for weekday in Weekday:
+ by_weekday[weekday] = timedelta()
+ for _, time_from, time_to in self.enum(db_reader):
+ for date, duration in self._enum_dates_and_durations(time_from, time_to):
+ by_weekday[Weekday(date.weekday())] += duration
+ return by_weekday
+
+ def group_by_hour(self, db_reader):
+ by_hour = OrderedDict()
+ for i in range(24):
+ by_hour[i] = timedelta()
+ for _, time_from, time_to in self.enum(db_reader):
+ for hour, duration in self._enum_hours_and_durations(time_from, time_to):
+ by_hour[hour] += duration
+ return by_hour
+
+ @staticmethod
+ def _enum_dates_and_durations(time_from, time_to):
+ while time_from.date() != time_to.date():
+ next_day = time_from.date() + timedelta(days=1)
+ yield time_from.date(), next_day - time_from
+ time_from = next_day
+ yield time_to.date(), time_to - time_from
+
+ @staticmethod
+ def _enum_hours_and_durations(time_from, time_to):
+ while time_from.date() != time_to.date() or time_from.hour != time_to.hour:
+ next_hour = time_from.replace(minute=0, second=0) + timedelta(hours=1)
+ yield time_from.hour, next_hour - time_from
+ time_from = next_hour
+ yield time_to.hour, time_to - time_from
+
+ def _insert_record(self, record):
+ return self._insert_user(record.to_user())
+
+ def _known_user(self, user):
+ return user.get_uid() in self._records
+
+ def _unknown_user(self, user):
+ return not self._known_user(user)
+
+ def _insert_user(self, user):
+ if user not in self or self[user].is_offline():
+ self[user] = user
+ return None
+ if user.is_online():
+ return None
+ streak = user, self[user].get_last_seen_time(), user.get_last_seen_time()
+ self[user] = user
+ return streak
diff --git a/vk/tracking/status_tracker.py b/vk/tracking/status_tracker.py
new file mode 100644
index 0000000..3d1f032
--- /dev/null
+++ b/vk/tracking/status_tracker.py
@@ -0,0 +1,99 @@
+# Copyright 2016 Egor Tensin <Egor.Tensin@gmail.com>
+# This file is licensed under the terms of the MIT License.
+# See LICENSE.txt for details.
+
+from collections import Callable
+import time
+
+import vk.error
+from vk.user import UserField
+
+class StatusTracker:
+ DEFAULT_TIMEOUT = 5
+
+ def __init__(self, api, timeout=DEFAULT_TIMEOUT):
+ self._api = api
+ self._timeout = timeout
+ self._on_initial_status = []
+ self._on_status_update = []
+ self._on_connection_error = []
+
+ def _wait_after_connection_error(self):
+ time.sleep(self._timeout)
+
+ def add_database_writer(self, writer):
+ self.add_initial_status_handler(lambda user: writer.on_initial_status(user))
+ self.add_status_update_handler(lambda user: writer.on_status_update(user))
+ self.add_connection_error_handler(lambda e: writer.on_connection_error(e))
+
+ def add_initial_status_handler(self, fn):
+ self._assert_is_callback(fn)
+ self._on_initial_status.append(fn)
+
+ def add_status_update_handler(self, fn):
+ self._assert_is_callback(fn)
+ self._on_status_update.append(fn)
+
+ def add_connection_error_handler(self, fn):
+ self._assert_is_callback(fn)
+ self._on_connection_error.append(fn)
+
+ @staticmethod
+ def _assert_is_callback(fn):
+ if not isinstance(fn, Callable):
+ raise TypeError()
+
+ _USER_FIELDS = UserField.SCREEN_NAME, UserField.ONLINE, UserField.LAST_SEEN
+
+ def _query_status(self, uids):
+ return {user.get_uid(): user for user in self._api.users_get(uids, self._USER_FIELDS)}
+
+ def _notify_status(self, user):
+ for fn in self._on_initial_status:
+ fn(user)
+
+ def _notify_status_update(self, user):
+ for fn in self._on_status_update:
+ fn(user)
+
+ def _notify_connection_error(self, e):
+ for fn in self._on_connection_error:
+ fn(e)
+
+ def _query_initial_status(self, uids):
+ while True:
+ try:
+ return self._query_status(uids)
+ except vk.error.ConnectionError as e:
+ self._notify_connection_error(e)
+ self._wait_after_connection_error()
+
+ def _query_status_updates(self, uids):
+ while True:
+ self._wait_after_connection_error()
+ try:
+ return self._query_status(uids)
+ except vk.error.ConnectionError as e:
+ self._notify_connection_error(e)
+
+ @staticmethod
+ def _filter_status_updates(old_users, new_users):
+ for uid, user in new_users.items():
+ if old_users[uid].is_online() != user.is_online():
+ old_users[uid] = user
+ yield user
+
+ def _do_loop(self, uids):
+ users = self._query_initial_status(uids)
+ for user in users.values():
+ self._notify_status(user)
+ while True:
+ updated_users = self._query_status_updates(uids)
+ for user in self._filter_status_updates(users, updated_users):
+ self._notify_status_update(user)
+
+ def loop(self, uids):
+ try:
+ self._do_loop(uids)
+ except KeyboardInterrupt:
+ pass