| 1 |
from __future__ import annotations |
| 2 |
|
| 3 |
from processview.utils.singleton import singleton |
| 4 |
from processview.core.dataset import Dataset |
| 5 |
from processview.core.dataset import DatasetIdentifier |
| 6 |
from collections import OrderedDict |
| 7 |
from silx.utils.enum import Enum as _Enum |
| 8 |
import threading |
| 9 |
from typing import Optional, Type, Union |
| 10 |
from datetime import datetime |
| 11 |
import weakref |
| 12 |
import logging |
| 13 |
|
| 14 |
_logger = logging.getLogger(__name__) |
| 15 |
|
| 16 |
|
| 17 |
class DatasetState(_Enum): |
| 18 |
"""possible dataset status relatif to a Process""" |
| 19 |
|
| 20 |
ON_GOING = "on going" |
| 21 |
SUCCEED = "succeed" |
| 22 |
FAILED = "failed" |
| 23 |
PENDING = "pending" |
| 24 |
SKIPPED = "skipped" |
| 25 |
WAIT_USER_VALIDATION = "waiting user validation" |
| 26 |
CANCELLED = "cancelled" |
| 27 |
|
| 28 |
|
| 29 |
@singleton |
| 30 |
class ProcessManager: |
| 31 |
""" |
| 32 |
Manager to register and observe `SuperviseProcess`. |
| 33 |
""" |
| 34 |
|
| 35 |
def __init__(self): |
| 36 |
self._processes = {} |
| 37 |
self._dataset_process_states: dict[int, dict[str, tuple]] = {} |
| 38 |
"""key is processID, Value is a tuple of |
| 39 |
(dataset state, details)""" |
| 40 |
self._processID = 0 |
| 41 |
self._updateCallback = set() |
| 42 |
"""list of callback to trigger when an update is generated""" |
| 43 |
self._newProcessCallback = set() |
| 44 |
"""list of callback to trigger when a process is added""" |
| 45 |
self._processRemovedCallback = set() |
| 46 |
"""list of callback to trigger when a process is removed""" |
| 47 |
self._process_renamed_callback = set() |
| 48 |
"""list of callback to trigger when a process is renamed""" |
| 49 |
self.lock = threading.Lock() |
| 50 |
|
| 51 |
def met(self, process, dataset) -> bool: |
| 52 |
"""check if a dataset has already met a dataset""" |
| 53 |
if isinstance(dataset, DatasetIdentifier): |
| 54 |
dataset_id = dataset |
| 55 |
elif dataset is None: |
| 56 |
return False |
| 57 |
else: |
| 58 |
dataset_id = dataset.get_identifier() |
| 59 |
if process.process_id in self._dataset_process_states: |
| 60 |
return dataset_id in self._dataset_process_states[process.process_id] |
| 61 |
|
| 62 |
def process_renamed(self, process): |
| 63 |
""" |
| 64 |
the process name only matter for the GUI side |
| 65 |
""" |
| 66 |
for callback in self._process_renamed_callback: |
| 67 |
callback(process) |
| 68 |
|
| 69 |
def register(self, process) -> int: |
| 70 |
""" |
| 71 |
Register a process to the manager |
| 72 |
|
| 73 |
:param SuperviseProcess process: |
| 74 |
""" |
| 75 |
from processview.core.superviseprocess import ( |
| 76 |
SuperviseProcess, |
| 77 |
) # avoid cyclic import |
| 78 |
|
| 79 |
if not isinstance(process, SuperviseProcess): |
| 80 |
raise TypeError( |
| 81 |
f"{process} is expected to be an isntance of {SuperviseProcess}" |
| 82 |
) |
| 83 |
process_id = self._processID |
| 84 |
self._processID += 1 |
| 85 |
self._processes[process_id] = weakref.ref(process) |
| 86 |
process._set_process_id(process_id) |
| 87 |
|
| 88 |
with self.lock: |
| 89 |
for callback in self._newProcessCallback: |
| 90 |
callback() |
| 91 |
return process_id |
| 92 |
|
| 93 |
def _find_dataset_id(self, data) -> Dataset: |
| 94 |
if isinstance(data, Dataset): |
| 95 |
return data.get_identifier() |
| 96 |
elif isinstance(data, DatasetIdentifier): |
| 97 |
return data |
| 98 |
else: |
| 99 |
raise TypeError( |
| 100 |
"dataset should be an instance of Dataset or " |
| 101 |
"DatasetIdentifier. Get {} instead".format(type(data)) |
| 102 |
) |
| 103 |
|
| 104 |
def _find_process(self, process): |
| 105 |
from processview.core.superviseprocess import ( |
| 106 |
SuperviseProcess, |
| 107 |
) # avoid cyclic import |
| 108 |
|
| 109 |
if isinstance(process, SuperviseProcess): |
| 110 |
return process |
| 111 |
elif isinstance(process, str): |
| 112 |
for p in self.get_processes(): |
| 113 |
if p.name == process: |
| 114 |
return p |
| 115 |
else: |
| 116 |
raise TypeError( |
| 117 |
"process should be an instance of SuperviseProcess or" |
| 118 |
" str. Get {} instead".format(type(process)) |
| 119 |
) |
| 120 |
|
| 121 |
def unregister(self, process): |
| 122 |
""" |
| 123 |
Unregister a process to the manager |
| 124 |
|
| 125 |
:param BaseProcess process: |
| 126 |
""" |
| 127 |
with self.lock: |
| 128 |
if process.process_id in self._processes and process.is_master_process: |
| 129 |
del self._processes[process.process_id] |
| 130 |
|
| 131 |
for callback in self._processRemovedCallback: |
| 132 |
callback() |
| 133 |
|
| 134 |
def get_processes(self) -> tuple: |
| 135 |
""" |
| 136 |
|
| 137 |
:return: tuple of processes currently registered |
| 138 |
""" |
| 139 |
processes = [] |
| 140 |
for _, p in self._processes.items(): |
| 141 |
if p() is not None: |
| 142 |
processes.append(p()) |
| 143 |
return tuple(processes) |
| 144 |
|
| 145 |
def get_process(self, process_id: int): |
| 146 |
if process_id in self._processes: |
| 147 |
return self._processes[process_id]() |
| 148 |
|
| 149 |
def get_datasets(self) -> tuple: |
| 150 |
""" |
| 151 |
|
| 152 |
:return: tuple of datasets |
| 153 |
""" |
| 154 |
res = set() |
| 155 |
for _, dataset_states in self._dataset_process_states.items(): |
| 156 |
[res.add(dataset) for dataset in dataset_states] |
| 157 |
return tuple(res) |
| 158 |
|
| 159 |
def remove_dataset(self, dataset: Dataset): |
| 160 |
""" |
| 161 |
Remove a dataset from the list of registered dataset |
| 162 |
""" |
| 163 |
dataset_id = self._find_dataset_id(dataset) |
| 164 |
for process_id in self._dataset_process_states: |
| 165 |
self._dataset_process_states[process_id].pop(dataset_id, None) |
| 166 |
|
| 167 |
def notify_dataset_state( |
| 168 |
self, |
| 169 |
dataset: Dataset | DatasetIdentifier, |
| 170 |
process, |
| 171 |
state: DatasetState, |
| 172 |
details=None, |
| 173 |
) -> None: |
| 174 |
""" |
| 175 |
Update dataset state |
| 176 |
|
| 177 |
:param Dataset dataset: dataset which state is updated |
| 178 |
:param BaseProcess process: Process concern by the new state |
| 179 |
:param DatasetState state: current State |
| 180 |
:param str info: details about the error or success |
| 181 |
:return: |
| 182 |
""" |
| 183 |
from processview.core.superviseprocess import ( |
| 184 |
SuperviseProcess, |
| 185 |
) # avoid cyclic import |
| 186 |
|
| 187 |
if not isinstance(process, SuperviseProcess): |
| 188 |
_logger.warning( |
| 189 |
f"Unable to update dataset {dataset} state. Process does not inherit from SuperviseProcess" |
| 190 |
) |
| 191 |
if process.process_id not in self._processes: |
| 192 |
self.register(process) |
| 193 |
if process.process_id not in self._dataset_process_states: |
| 194 |
self._dataset_process_states[process.process_id] = OrderedDict() |
| 195 |
if details is None: |
| 196 |
details = "" |
| 197 |
if isinstance(dataset, Dataset): |
| 198 |
dataset_id = dataset.get_identifier() |
| 199 |
else: |
| 200 |
dataset_id = dataset |
| 201 |
if not isinstance(dataset_id, DatasetIdentifier): |
| 202 |
raise TypeError( |
| 203 |
f"dataset should be an instance of {Dataset} or {DatasetIdentifier}. Got {type(dataset)}" |
| 204 |
) |
| 205 |
# as we store the identifier and not a string anymore we might need to remove another instanciation |
| 206 |
process_states = self._dataset_process_states[process.process_id] |
| 207 |
to_remove = set() |
| 208 |
for id in process_states: |
| 209 |
if id == dataset_id: |
| 210 |
to_remove.add(id) |
| 211 |
for id in to_remove: |
| 212 |
del process_states[id] |
| 213 |
self._dataset_process_states[process.process_id] = process_states |
| 214 |
self._dataset_process_states[process.process_id][dataset_id] = ( |
| 215 |
state, |
| 216 |
datetime.now(), |
| 217 |
details, |
| 218 |
) |
| 219 |
self.updated() |
| 220 |
|
| 221 |
def get_dataset_state(self, dataset_id, process) -> Union[None, DatasetState]: |
| 222 |
""" |
| 223 |
|
| 224 |
:param Dataset dataset_id: |
| 225 |
:param BaseProcess process: |
| 226 |
:return: DatasetState relative to provided process if know |
| 227 |
:rtype: Union[None, DatasetState] |
| 228 |
""" |
| 229 |
dataset_id = self._find_dataset_id(dataset_id) |
| 230 |
assert isinstance(dataset_id, DatasetIdentifier) |
| 231 |
process = self._find_process(process) |
| 232 |
if process is None: |
| 233 |
_logger.warning("process {} is no more supervised".format(process)) |
| 234 |
return |
| 235 |
if dataset_id is None: |
| 236 |
_logger.warning("dataset {} is no more supervised".format(dataset_id)) |
| 237 |
elif process.process_id in self._dataset_process_states: |
| 238 |
if dataset_id in self._dataset_process_states[process.process_id]: |
| 239 |
return self._dataset_process_states[process.process_id][dataset_id][0] |
| 240 |
return None |
| 241 |
|
| 242 |
def get_dataset_details(self, dataset_id, process) -> Union[None, str]: |
| 243 |
""" |
| 244 |
|
| 245 |
:param Dataset dataset_id: |
| 246 |
:param BaseProcess process: |
| 247 |
:return: DatasetState relative to provided process if know |
| 248 |
""" |
| 249 |
dataset_id = self._find_dataset_id(dataset_id) |
| 250 |
process = self._find_process(process) |
| 251 |
if process is None: |
| 252 |
_logger.warning("process {} is no more supervised".format(process)) |
| 253 |
return |
| 254 |
if dataset_id is None: |
| 255 |
_logger.warning("dataset {} is no more supervised".format(dataset_id)) |
| 256 |
if process.process_id in self._dataset_process_states: |
| 257 |
if dataset_id in self._dataset_process_states[process.process_id]: |
| 258 |
return self._dataset_process_states[process.process_id][dataset_id][2] |
| 259 |
return None |
| 260 |
|
| 261 |
def get_dataset_time_stamp(self, dataset_id, process) -> Union[None, str]: |
| 262 |
""" |
| 263 |
|
| 264 |
:param Dataset dataset_id: |
| 265 |
:param BaseProcess process: |
| 266 |
:return: DatasetState relative to provided process if know |
| 267 |
""" |
| 268 |
dataset_id = self._find_dataset_id(dataset_id) |
| 269 |
process = self._find_process(process) |
| 270 |
if process is None: |
| 271 |
_logger.warning("process {} is no more supervised".format(process)) |
| 272 |
return |
| 273 |
if dataset_id is None: |
| 274 |
_logger.warning("dataset {} is no more supervised".format(dataset_id)) |
| 275 |
if process.process_id in self._dataset_process_states: |
| 276 |
if dataset_id in self._dataset_process_states[process.process_id]: |
| 277 |
return self._dataset_process_states[process.process_id][dataset_id][1] |
| 278 |
return None |
| 279 |
|
| 280 |
def get_dataset_stream( |
| 281 |
self, dataset, time_stamp=False |
| 282 |
) -> tuple[int, DatasetState, list]: |
| 283 |
""" |
| 284 |
|
| 285 |
:param Dataset dataset: dataset the stream is focus on |
| 286 |
:param bool time_stamp: if True then return timestamp in the list of |
| 287 |
elements |
| 288 |
:return: stream of (process ID, DatasetState) for a given dataset |
| 289 |
""" |
| 290 |
stream = [] |
| 291 |
for process_id, dataset_states in self._dataset_process_states.items(): |
| 292 |
dst_id = dataset.get_identifier() |
| 293 |
if dst_id in dataset_states: |
| 294 |
state, _timestamp, _ = dataset_states[dst_id] |
| 295 |
stream.append((process_id, state, _timestamp)) |
| 296 |
# order the stream |
| 297 |
stream = sorted(stream, key=lambda elmt: elmt[2]) |
| 298 |
if not time_stamp: |
| 299 |
stream = [s[:-1] for s in stream] |
| 300 |
return tuple(stream) |
| 301 |
|
| 302 |
def get_process_history(self, process, time_stamp=False) -> tuple: |
| 303 |
""" |
| 304 |
Return the know history of the process. |
| 305 |
|
| 306 |
:param BaseProcess process: |
| 307 |
:param bool time_stamp: if True then return timestamp in the list of |
| 308 |
elements |
| 309 |
:return: tuple of (DatasetIdentifier, state, [timestamp]) |
| 310 |
""" |
| 311 |
history = [] |
| 312 |
if process.process_id in self._dataset_process_states: |
| 313 |
dataset_states = self._dataset_process_states[process.process_id] |
| 314 |
for dataset_id, info in dataset_states.items(): |
| 315 |
state, _timestamp, details = info |
| 316 |
history.append((dataset_id, state, _timestamp)) |
| 317 |
|
| 318 |
history = sorted(history, key=lambda elmt: elmt[2]) |
| 319 |
if not time_stamp: |
| 320 |
history = [s[:-1] for s in history] |
| 321 |
return tuple(history) |
| 322 |
|
| 323 |
def updated(self): |
| 324 |
"""Function 'open' for monckey patch""" |
| 325 |
with self.lock: |
| 326 |
for callback in self._updateCallback: |
| 327 |
callback() |
| 328 |
|
| 329 |
def add_update_callback(self, callback) -> None: |
| 330 |
""" |
| 331 |
|
| 332 |
:param callback: add a callback to be trigger when dataset state change |
| 333 |
""" |
| 334 |
with self.lock: |
| 335 |
self._updateCallback.add(callback) |
| 336 |
|
| 337 |
def remove_update_callback(self, callback) -> None: |
| 338 |
""" |
| 339 |
|
| 340 |
:param callback: remove a callback from the stack of callback to be |
| 341 |
call when dataset state change |
| 342 |
""" |
| 343 |
with self.lock: |
| 344 |
if callback in self._updateCallback: |
| 345 |
self._updateCallback.remove(callback) |
| 346 |
|
| 347 |
def add_new_process_callback(self, callback): |
| 348 |
self._newProcessCallback.add(callback) |
| 349 |
|
| 350 |
def remove_new_process_callback(self, callback): |
| 351 |
with self.lock: |
| 352 |
if callback in self._newProcessCallback: |
| 353 |
self._newProcessCallback.remove(callback) |
| 354 |
|
| 355 |
def add_process_removed_callback(self, callback): |
| 356 |
self._processRemovedCallback.add(callback) |
| 357 |
|
| 358 |
def remove_process_removed_callback(self, callback): |
| 359 |
with self.lock: |
| 360 |
if callback in self._processRemovedCallback: |
| 361 |
self._processRemovedCallback.remove(callback) |
| 362 |
|
| 363 |
def add_process_name_changed_callback(self, callback): |
| 364 |
self._process_renamed_callback.add(callback) |
| 365 |
|
| 366 |
def clear(self): |
| 367 |
""" |
| 368 |
clear registered processes and dataset states |
| 369 |
|
| 370 |
:return: |
| 371 |
""" |
| 372 |
self._processes = {} |
| 373 |
self._dataset_process_states = {} |