data.py 45.5 KB
Newer Older
Markus Scheidgen's avatar
Markus Scheidgen committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
# Copyright 2018 Markus Scheidgen
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an"AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

15
'''
Markus Scheidgen's avatar
Markus Scheidgen committed
16
17
18
19
20
21
This module comprises a set of persistent document classes that hold all user related
data. These are information about users, their uploads and datasets, the associated
calculations, and files


.. autoclass:: Calc
Markus Scheidgen's avatar
Markus Scheidgen committed
22

Markus Scheidgen's avatar
Markus Scheidgen committed
23
.. autoclass:: Upload
Markus Scheidgen's avatar
Markus Scheidgen committed
24

25
'''
Markus Scheidgen's avatar
Markus Scheidgen committed
26

27
from typing import cast, List, Any, Tuple, Iterator, Dict, cast, Iterable
28
from mongoengine import StringField, DateTimeField, DictField, BooleanField, IntField
Markus Scheidgen's avatar
Markus Scheidgen committed
29
import logging
30
from structlog import wrap_logger
31
from contextlib import contextmanager
32
import os.path
33
34
from datetime import datetime
from pymongo import UpdateOne
35
import hashlib
36
from structlog.processors import StackInfoRenderer, format_exc_info, TimeStamper
Markus Scheidgen's avatar
Markus Scheidgen committed
37

38
from nomad import utils, config, infrastructure, search, datamodel
39
from nomad.files import PathObject, UploadFiles, ExtractError, ArchiveBasedStagingUploadFiles, PublicUploadFiles, StagingUploadFiles
40
from nomad.processing.base import Proc, process, task, PENDING, SUCCESS, FAILURE
41
from nomad.parsing import parser_dict, match_parser, Backend
Markus Scheidgen's avatar
Markus Scheidgen committed
42
43
44
from nomad.normalizing import normalizers


45
46
47
48
49
50
51
def _pack_log_event(logger, method_name, event_dict):
    try:
        log_data = dict(event_dict)
        log_data.update(**{
            key: value
            for key, value in getattr(logger, '_context', {}).items()
            if key not in ['service', 'release', 'upload_id', 'calc_id', 'mainfile', 'process_status']})
52
        log_data.update(logger=logger.name)
53
54
55
56
57
58
59
60
61
62
63

        return log_data
    except Exception:
        # raising an exception would cause an indefinite loop
        return event_dict


_log_processors = [
    StackInfoRenderer(),
    _pack_log_event,
    format_exc_info,
64
    TimeStamper(fmt="%Y-%m-%d %H:%M.%S", utc=False)]
65
66


Markus Scheidgen's avatar
Markus Scheidgen committed
67
class Calc(Proc):
68
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
69
70
71
72
73
    Instances of this class represent calculations. This class manages the elastic
    search index entry, files, and archive for the respective calculation.

    It also contains the calculations processing and its state.

74
75
    The attribute list, does not include the various metadata properties generated
    while parsing, including ``code_name``, ``code_version``, etc.
Markus Scheidgen's avatar
Markus Scheidgen committed
76
77

    Attributes:
78
        calc_id: the calc_id of this calc
Markus Scheidgen's avatar
Markus Scheidgen committed
79
80
81
        parser: the name of the parser used to process this calc
        upload_id: the id of the upload used to create this calculation
        mainfile: the mainfile (including path in upload) that was used to create this calc
82

83
84
        metadata: the metadata record wit calc and user metadata, see :class:`datamodel.EntryMetadata`
    '''
85
    calc_id = StringField(primary_key=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
86
87
88
89
    upload_id = StringField()
    mainfile = StringField()
    parser = StringField()

90
91
    metadata = DictField()

Markus Scheidgen's avatar
Markus Scheidgen committed
92
    meta: Any = {
Markus Scheidgen's avatar
Markus Scheidgen committed
93
        'indexes': [
94
95
96
97
            'upload_id',
            ('upload_id', 'mainfile'),
            ('upload_id', 'parser'),
            ('upload_id', 'tasks_status'),
98
            ('upload_id', 'process_status'),
99
100
101
            ('upload_id', 'metadata.nomad_version'),
            'metadata.published',
            'metadata.datasets'
Markus Scheidgen's avatar
Markus Scheidgen committed
102
            'metadata.pid'
Markus Scheidgen's avatar
Markus Scheidgen committed
103
104
105
106
107
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
108
        self._parser_backend: Backend = None
109
110
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
111
        self._calc_proc_logs: List[Any] = None
Markus Scheidgen's avatar
Markus Scheidgen committed
112

113
        self._entry_metadata = None
114

Markus Scheidgen's avatar
Markus Scheidgen committed
115
116
    @classmethod
    def get(cls, id):
117
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
118

Markus Scheidgen's avatar
Markus Scheidgen committed
119
    @property
120
121
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
122

123
124
125
126
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
127
            self._upload.worker_hostname = self.worker_hostname
128
129
        return self._upload

130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
    def apply_entry_metadata(self, entry_metadata: datamodel.EntryMetadata):
        self.metadata = entry_metadata.m_to_dict(
            include_defaults=True,
            categories=[datamodel.MongoMetadata])  # TODO use embedded doc?

    def create_metadata(self) -> datamodel.EntryMetadata:
        '''
        Returns a :class:`nomad.datamodel.EntryMetadata` with values from this
        processing object, not necessarely the user metadata nor the metadata from
        the archive.
        '''
        entry_metadata = datamodel.EntryMetadata()
        entry_metadata.domain = parser_dict[self.parser].domain
        entry_metadata.upload_id = self.upload_id
        entry_metadata.calc_id = self.calc_id
        entry_metadata.mainfile = self.mainfile
        entry_metadata.nomad_version = config.version
        entry_metadata.nomad_commit = config.commit
        entry_metadata.uploader = self.upload.user_id
        entry_metadata.upload_time = self.upload.upload_time
        entry_metadata.upload_name = self.upload.name

        return entry_metadata

    def entry_metadata(self, upload_files: UploadFiles) -> datamodel.EntryMetadata:
        '''
        Returns a complete set of :class:`nomad.datamodel.EntryMetadata` including
        the user metadata and metadata from the archive.

        Arguments:
            upload_files:
                The :class:`nomad.files.UploadFiles` instance to read the archive from.
            cache:
                A boolean that indicates if the archive file should be left unclosed,
                e.g. if this method is called for many entries of the same upload.
        '''
        archive = upload_files.read_archive(self.calc_id)
167
168
169
170
171
172
173
174
175
176
177
178
        try:
            entry_metadata = datamodel.EntryMetadata.m_from_dict(
                archive[self.calc_id][datamodel.EntryArchive.section_metadata.name].to_dict())

        except KeyError:
            # Due hard processing failures, it might be possible that an entry might not
            # have an archive
            if self._entry_metadata is not None:
                entry_metadata = self._entry_metadata

            else:
                entry_metadata = self.create_metadata()
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194

        entry_metadata.m_update_from_dict(self.metadata)

        return entry_metadata

    def user_metadata(self) -> datamodel.EntryMetadata:
        '''
        Returns a :class:`nomad.datamodel.EntryMetadata` with values from this
        processing object and the user metadata, not necessarely the metadata from
        the archive.
        '''
        entry_metadata = self.create_metadata()
        entry_metadata.m_update_from_dict(self.metadata)

        return entry_metadata

195
196
197
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
198
199
            self._upload_files = ArchiveBasedStagingUploadFiles(
                self.upload_id, is_authorized=lambda: True, upload_path=self.upload.upload_path)
200
201
        return self._upload_files

Markus Scheidgen's avatar
Markus Scheidgen committed
202
    def get_logger(self, **kwargs):
203
        '''
204
205
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
206
        '''
207
208
209
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile, calc_id=self.calc_id, **kwargs)
210

211
212
        if self._calc_proc_logs is None:
            self._calc_proc_logs = []
213

214
215
        def save_to_calc_log(logger, method_name, event_dict):
            try:
216
217
                # sanitize the event_dict, because all kinds of values might have been added
                dump_dict = {key: str(value) for key, value in event_dict.items()}
218
219
                dump_dict.update(level=method_name.upper())
                self._calc_proc_logs.append(dump_dict)
220

221
222
223
            except Exception:
                # Exceptions here will cause indefinite loop
                pass
224

225
            return event_dict
226

227
        return wrap_logger(logger, processors=_log_processors + [save_to_calc_log])
228

229
230
    @process
    def re_process_calc(self):
231
        '''
232
233
234
        Processes a calculation again. This means there is already metadata and
        instead of creating it initially, we are just updating the existing
        records.
235
        '''
236
        parser = match_parser(self.upload_files.raw_file_object(self.mainfile).os_path, strict=False)
237
        logger = self.get_logger()
238
239
240

        if parser is None and not config.reprocess_unmatched:
            self.errors = ['no parser matches during re-process, will not re-process this calc']
241

242
243
244
245
246
247
248
            try:
                upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
                with upload_files.read_archive(self.calc_id) as archive:
                    self.upload_files.write_archive(self.calc_id, archive[self.calc_id].to_dict())

            except Exception as e:
                logger.error('could not copy archive for non matching, non reprocessed entry', exc_info=e)
249
                raise e
250

251
252
253
254
255
            # mock the steps of actual processing
            self._continue_with('parsing')
            self._continue_with('normalizing')
            self._continue_with('archiving')
            self._complete()
256
257
            return

258
        if parser is None:
259
260
            self.get_logger().error('no parser matches during re-process, use the old parser')
            self.errors = ['no matching parser found during re-processing']
261
        if self.parser != parser.name:
262
263
264
            self.parser = parser.name
            logger.info(
                'different parser matches during re-process, use new parser',
265
                parser=parser.name)
266

267
        try:
268
269
270
271
272
273
            self._entry_metadata = self.user_metadata()
            self._entry_metadata.calc_hash = self.upload_files.calc_hash(self.mainfile)
            self._entry_metadata.nomad_version = config.version
            self._entry_metadata.nomad_commit = config.commit
            self._entry_metadata.last_processing = datetime.utcnow()
            self._entry_metadata.files = self.upload_files.calc_files(self.mainfile)
274
275
276
277
278
279

            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
            # close loghandler that was not closed due to failures
280
281
282
283
284
285
            try:
                if self._parser_backend and self._parser_backend.resource:
                    self._parser_backend.resource.unload()
            except Exception as e:
                logger.error('could unload processing results', exc_info=e)

Markus Scheidgen's avatar
Markus Scheidgen committed
286
    @process
287
    def process_calc(self):
288
        '''
289
290
        Processes a new calculation that has no prior records in the mongo, elastic,
        or filesystem storage. It will create an initial set of (user) metadata.
291
        '''
292
        logger = self.get_logger()
293
        if self.upload is None:
294
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
295
296

        try:
297
298
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
299
300
301
302
303
304
            self._entry_metadata = self.create_metadata()
            self._entry_metadata.calc_hash = self.upload_files.calc_hash(self.mainfile)
            self._entry_metadata.last_processing = datetime.utcnow()
            self._entry_metadata.files = self.upload_files.calc_files(self.mainfile)

            if len(self._entry_metadata.files) >= config.auxfile_cutoff:
305
306
307
308
                self.warning(
                    'This calc has many aux files in its directory. '
                    'Have you placed many calculations in the same directory?')

Markus Scheidgen's avatar
Markus Scheidgen committed
309
310
311
312
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
313
            # close loghandler that was not closed due to failures
314
315
316
317
318
319
            try:
                if self._parser_backend and self._parser_backend.resource:
                    self._parser_backend.resource.unload()
            except Exception as e:
                logger.error('could unload processing results', exc_info=e)

320
321
322
323
    def fail(self, *errors, log_level=logging.ERROR, **kwargs):
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
324
325
326
            if self.parser is not None:
                parser = parser_dict[self.parser]
                if hasattr(parser, 'code_name'):
327
                    self._entry_metadata.code_name = parser.code_name
328

329
330
            self._entry_metadata.processed = False
            self.apply_entry_metadata(self._entry_metadata)
331
332
333
334
335
            if self._parser_backend and self._parser_backend.resource:
                backend = self._parser_backend
            else:
                backend = None
            self._entry_metadata.apply_domain_metadata(backend)
336
            self._entry_metadata.a_elastic.index()
337
        except Exception as e:
338
339
            self.get_logger().error(
                'could not index after processing failure', exc_info=e)
340

341
        try:
342
            self.write_archive(None)
343
        except Exception as e:
344
345
            self.get_logger().error(
                'could not write archive after processing failure', exc_info=e)
346

347
348
        super().fail(*errors, log_level=log_level, **kwargs)

349
350
351
352
    def on_process_complete(self, process_name):
        # the save might be necessary to correctly read the join condition from the db
        self.save()
        # in case of error, the process_name might be unknown
353
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
354
355
            self.upload.reload()
            self.upload.check_join()
Markus Scheidgen's avatar
Markus Scheidgen committed
356
357
358

    @task
    def parsing(self):
359
        ''' The *task* that encapsulates all parsing related actions. '''
360
        context = dict(parser=self.parser, step=self.parser)
361
        logger = self.get_logger(**context)
362
        parser = parser_dict[self.parser]
363
        self._entry_metadata.parser_name = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
364

365
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
366
367
368
369
            try:
                self._parser_backend = parser.run(
                    self.upload_files.raw_file_object(self.mainfile).os_path, logger=logger)
            except Exception as e:
370
                self.fail('parser failed with exception', exc_info=e, error=str(e), **context)
371
                return
372
            except SystemExit:
373
                self.fail('parser raised system exit', error='system exit', **context)
374
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
375

Markus Scheidgen's avatar
Markus Scheidgen committed
376
377
        if self._parser_backend.status[0] != 'ParseSuccess':
            error = self._parser_backend.status[1]
378
            self.fail('parser failed', error=error, **context)
379
380
381
382
383
384
385
386

    @contextmanager
    def use_parser_backend(self, processor_name):
        self._parser_backend.reset_status()
        yield self._parser_backend

        if self._parser_backend.status[0] == 'ParseSuccess':
            warnings = getattr(self._parser_backend, '_warnings', [])
387

388
            if len(warnings) > 0:
389
390
391
392
                self.get_logger().warn(
                    'processor completed successful with warnings',
                    processor=processor_name, warnings=[str(warning) for warning in warnings])

393
            else:
394
395
396
397
                self.get_logger().info(
                    'processor completed successful',
                    processor=processor_name)

398
        else:
399
            errors = self._parser_backend.status[1]
400
401
402
            self.get_logger().error(
                'processor completed with failure',
                processor=processor_name, errors=str(errors))
Markus Scheidgen's avatar
Markus Scheidgen committed
403
404
405

    @task
    def normalizing(self):
406
        ''' The *task* that encapsulates all normalizing related actions. '''
407
408
409
410
411

        # allow normalizer to access and add data to the entry metadata
        self._parser_backend.entry_archive.m_add_sub_section(
            datamodel.EntryArchive.section_metadata, self._entry_metadata)

Markus Scheidgen's avatar
Markus Scheidgen committed
412
        for normalizer in normalizers:
413
            if normalizer.domain != parser_dict[self.parser].domain:
414
415
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
416
            normalizer_name = normalizer.__name__
417
            context = dict(normalizer=normalizer_name, step=normalizer_name)
418
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
419
420

            with utils.timer(
421
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
422
                with self.use_parser_backend(normalizer_name) as backend:
423
424
425
                    try:
                        normalizer(backend).normalize(logger=logger)
                    except Exception as e:
426
                        self._parser_backend.finishedParsingSession('ParseFailure', [str(e)])
427
428
429
                        logger.error(
                            'normalizer failed with exception', exc_info=e,
                            error=str(e), **context)
430
431
432
433
434
435
436
437
438
                        break
                    else:
                        if self._parser_backend.status[0] != 'ParseSuccess':
                            error = self._parser_backend.status[1]
                            self.fail('normalizer failed', error=error, **context)
                            break
                        else:
                            logger.debug(
                                'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
439
440
441

    @task
    def archiving(self):
442
        ''' The *task* that encapsulates all archival related actions. '''
443
444
        logger = self.get_logger()

445
446
        self._entry_metadata.apply_domain_metadata(self._parser_backend)
        self._entry_metadata.processed = True
447

448
449
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
450
            self.apply_entry_metadata(self._entry_metadata)
451
452

        # index in search
453
        with utils.timer(logger, 'indexed', step='index'):
454
            self._entry_metadata.a_elastic.index()
455

456
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
457
        with utils.timer(
458
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
459
                input_size=self.mainfile_file.size) as log_data:
460

461
            archive_size = self.write_archive(self._parser_backend)
462
            log_data.update(archive_size=archive_size)
Markus Scheidgen's avatar
Markus Scheidgen committed
463

464
465
466
467
468
469
470
471
472
    def write_archive(self, backend: Backend):
        if self._calc_proc_logs is None:
            self._calc_proc_logs = []

        if backend is not None:
            entry_archive = backend.entry_archive.m_copy()
        else:
            entry_archive = datamodel.EntryArchive()

473
474
475
        if entry_archive.section_metadata is None:
            entry_archive.m_add_sub_section(datamodel.EntryArchive.section_metadata, self._entry_metadata)

476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
        entry_archive.processing_logs = self._calc_proc_logs

        try:
            return self.upload_files.write_archive(self.calc_id, entry_archive.m_to_dict())
        except Exception as e:
            if backend is None:
                raise e

            # most likely failed due to domain data, try to write metadata and processing logs
            entry_archive = datamodel.EntryArchive()
            entry_archive.m_add_sub_section(datamodel.EntryArchive.section_metadata, self._entry_metadata)
            entry_archive.processing_logs = self._calc_proc_logs
            self.upload_files.write_archive(self.calc_id, entry_archive.m_to_dict())
            raise e

491
    def __str__(self):
492
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
493

494

495
class Upload(Proc):
496
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
497
498
499
500
501
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
502
503
        upload_path: the path were the uploaded files was stored
        temporary: True if the uploaded file should be removed after extraction
Markus Scheidgen's avatar
Markus Scheidgen committed
504
505
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
506
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
507
508
        published: Boolean that indicates the publish status
        publish_time: Date when the upload was initially published
509
        last_update: Date of the last publishing/re-processing
Markus Scheidgen's avatar
Markus Scheidgen committed
510
        joined: Boolean indicates if the running processing has joined (:func:`check_join`)
511
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
512
513
514
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
515
516
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
517
    embargo_length = IntField(default=36)
Markus Scheidgen's avatar
Markus Scheidgen committed
518
519
520

    name = StringField(default=None)
    upload_time = DateTimeField()
521
    user_id = StringField(required=True)
522
523
    published = BooleanField(default=False)
    publish_time = DateTimeField()
524
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
525

526
527
    joined = BooleanField(default=False)

Markus Scheidgen's avatar
Markus Scheidgen committed
528
529
    meta: Any = {
        'indexes': [
530
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
Markus Scheidgen's avatar
Markus Scheidgen committed
531
532
533
534
535
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
536
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
537

538
539
    @property
    def metadata(self) -> dict:
540
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
541
542
543
544
545
        Getter, setter for user metadata. Metadata is pickled to and from the public
        bucket to allow sharing among all processes. Usually uploads do not have (much)
        user defined metadata, but users provide all metadata per upload as part of
        the publish process. This will change, when we introduce editing functionality
        and metadata will be provided through different means.
546
        '''
547
548
549
550
551
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
552
553
554

    @metadata.setter
    def metadata(self, data: dict) -> None:
555
556
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
557

Markus Scheidgen's avatar
Markus Scheidgen committed
558
    @classmethod
559
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
560
        return cls.get_by_id(id, 'upload_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
561
562

    @classmethod
563
    def user_uploads(cls, user: datamodel.User, **kwargs) -> List['Upload']:
564
        ''' Returns all uploads for the given user. Kwargs are passed to mongo query. '''
565
        return cls.objects(user_id=str(user.user_id), **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
566

567
568
    @property
    def uploader(self):
569
        return datamodel.User.get(self.user_id)
570

Markus Scheidgen's avatar
Markus Scheidgen committed
571
572
    def get_logger(self, **kwargs):
        logger = super().get_logger()
573
574
        user = self.uploader
        user_name = '%s %s' % (user.first_name, user.last_name)
575
        # We are not using 'user_id' because logstash (?) will filter these entries ?!
576
577
        logger = logger.bind(
            upload_id=self.upload_id, upload_name=self.name, user_name=user_name,
578
            user=user.user_id, **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
579
580
581
582
        return logger

    @classmethod
    def create(cls, **kwargs) -> 'Upload':
583
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
584
585
586
        Creates a new upload for the given user, a user given name is optional.
        It will populate the record with a signed url and pending :class:`UploadProc`.
        The upload will be already saved to the database.
587
588

        Arguments:
589
            user: The user that created the upload.
590
        '''
591
592
        # use kwargs to keep compatibility with super method
        user: datamodel.User = kwargs['user']
593
        del(kwargs['user'])
594

595
596
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
597
        kwargs.update(user_id=user.user_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
598
        self = super().create(**kwargs)
599

Markus Scheidgen's avatar
Markus Scheidgen committed
600
        self._continue_with('uploading')
601

Markus Scheidgen's avatar
Markus Scheidgen committed
602
603
        return self

604
    def delete(self):
605
        ''' Deletes this upload process state entry and its calcs. '''
606
607
608
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

609
    def delete_upload_local(self):
610
        '''
611
        Deletes the upload, including its processing state and
612
        staging files. Local version without celery processing.
613
        '''
614
615
616
        logger = self.get_logger()

        with utils.lnr(logger, 'staged upload delete failed'):
617
            with utils.timer(
618
                    logger, 'upload deleted from index', step='index',
619
                    upload_size=self.upload_files.size):
620
                search.delete_upload(self.upload_id)
621

622
            with utils.timer(
623
                    logger, 'staged upload deleted', step='files',
624
625
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
626
627

            self.delete()
628

629
    @process
630
    def delete_upload(self):
631
        '''
632
633
        Deletes of the upload, including its processing state and
        staging files. This starts the celery process of deleting the upload.
634
        '''
635
        self.delete_upload_local()
636

637
        return True  # do not save the process status on the delete upload
638

639
    @process
640
    def publish_upload(self):
641
        '''
642
643
        Moves the upload out of staging to the public area. It will
        pack the staging upload files in to public upload files.
644
        '''
645
646
        assert self.processed_calcs > 0

647
        logger = self.get_logger()
648
        logger.info('started to publish')
649

650
        with utils.lnr(logger, 'publish failed'):
651
            with self.entries_metadata(self.metadata) as calcs:
652

653
                with utils.timer(
654
                        logger, 'upload metadata updated', step='metadata',
655
                        upload_size=self.upload_files.size):
656

657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
                    def create_update(calc):
                        calc.published = True
                        calc.with_embargo = calc.with_embargo if calc.with_embargo is not None else False
                        return UpdateOne(
                            {'_id': calc.calc_id},
                            {'$set': {'metadata': calc.m_to_dict(
                                include_defaults=True, categories=[datamodel.MongoMetadata])}})

                    Calc._get_collection().bulk_write([create_update(calc) for calc in calcs])

                if isinstance(self.upload_files, StagingUploadFiles):
                    with utils.timer(
                            logger, 'staged upload files packed', step='pack',
                            upload_size=self.upload_files.size):
                        self.upload_files.pack(calcs)
672

673
                with utils.timer(
674
                        logger, 'index updated', step='index',
675
                        upload_size=self.upload_files.size):
676
677
678
679
680
681
682
683
684
685
686
687
                    search.publish(calcs)

                if isinstance(self.upload_files, StagingUploadFiles):
                    with utils.timer(
                            logger, 'staged upload deleted', step='delete staged',
                            upload_size=self.upload_files.size):
                        self.upload_files.delete()
                        self.published = True
                        self.publish_time = datetime.utcnow()
                        self.last_update = datetime.utcnow()
                        self.save()
                else:
688
                    self.last_update = datetime.utcnow()
689
                    self.save()
690

691
692
    @process
    def re_process_upload(self):
693
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
694
695
696
        A *process* that performs the re-processing of a earlier processed
        upload.

697
698
699
700
701
702
        Runs the distributed process of fully reparsing/renormalizing an existing and
        already published upload. Will renew the archive part of the upload and update
        mongo and elastic search entries.

        TODO this implementation does not do any re-matching. This will be more complex
        due to handling of new or missing matches.
703
        '''
704
705
706
        logger = self.get_logger()
        logger.info('started to re-process')

707
708
709
710
711
        # mock the steps of actual processing
        self._continue_with('uploading')

        # extract the published raw files into a staging upload files instance
        self._continue_with('extracting')
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726

        if self.published:
            try:
                staging_upload_files = StagingUploadFiles(self.upload_id)
                # public files exist and there is a staging directory, it is probably old
                # and we delete it first
                staging_upload_files.delete()
                logger.warn('deleted old staging files')

            except KeyError as e:
                logger.info('reprocessing published files')
        else:
            logger.info('reprocessing staging files')

        staging_upload_files = self.upload_files.to_staging_upload_files(create=True)
727
728

        self._continue_with('parse_all')
729
        try:
730
            # check if a calc is already/still processing
731
732
733
734
735
736
737
738
            processing = Calc.objects(
                upload_id=self.upload_id,
                **Calc.process_running_mongoengine_query()).count()

            if processing > 0:
                logger.warn(
                    'processes are still/already running on calc, they will be resetted',
                    count=processing)
739
740

            # reset all calcs
741
            Calc._get_collection().update_many(
742
743
744
                dict(upload_id=self.upload_id),
                {'$set': Calc.reset_pymongo_update(worker_hostname=self.worker_hostname)})

745
            # process call calcs
746
747
748
            Calc.process_all(Calc.re_process_calc, dict(upload_id=self.upload_id), exclude=['metadata'])

            logger.info('completed to trigger re-process of all calcs')
749
750
        except Exception as e:
            # try to remove the staging copy in failure case
751
752
            logger.error('failed to trigger re-process of all calcs', exc_info=e)

753
754
755
            if self.published:
                if staging_upload_files is not None and staging_upload_files.exists():
                    staging_upload_files.delete()
756
757

            raise e
758
759
760
761

        # the packing and removing of the staging upload files, will be trigged by
        # the 'cleanup' task after processing all calcs

762
763
    @process
    def re_pack(self):
764
        ''' A *process* that repacks the raw and archive data based on the current embargo data. '''
765
766
767
768
769
770
771
772
        assert self.published

        # mock the steps of actual processing
        self._continue_with('uploading')
        self._continue_with('extracting')
        self._continue_with('parse_all')
        self._continue_with('cleanup')

773
        self.upload_files.re_pack(self.user_metadata())
774
775
776
        self.joined = True
        self._complete()

Markus Scheidgen's avatar
Markus Scheidgen committed
777
    @process
778
    def process_upload(self):
779
        ''' A *process* that performs the initial upload processing. '''
Markus Scheidgen's avatar
Markus Scheidgen committed
780
781
782
783
784
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
785
        ''' A no-op *task* as a stand-in for receiving upload data. '''
Markus Scheidgen's avatar
Markus Scheidgen committed
786
787
        pass

788
    @property
789
790
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
791
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
792
793
794

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
795
                self.upload_id, is_authorized=lambda: True, **kwargs)
796

797
        return self._upload_files
798

799
800
801
802
803
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
804
805
    @task
    def extracting(self):
806
        '''
807
808
        The *task* performed before the actual parsing/normalizing: extracting
        the uploaded files.
809
        '''
810
811
812
813
814
        # extract the uploaded file
        self._upload_files = ArchiveBasedStagingUploadFiles(
            upload_id=self.upload_id, is_authorized=lambda: True, create=True,
            upload_path=self.upload_path)

Markus Scheidgen's avatar
Markus Scheidgen committed
815
816
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
817
818
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
819
820
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
821
822
823
824

            if self.temporary:
                os.remove(self.upload_path)
                self.upload_path = None
825

826
        except KeyError:
827
            self.fail('processing requested for non existing upload', log_level=logging.ERROR)
828
829
830
            return
        except ExtractError:
            self.fail('bad .zip/.tar file', log_level=logging.INFO)
Markus Scheidgen's avatar
Markus Scheidgen committed
831
832
            return

833
    def _preprocess_files(self, path):
834
        '''
835
836
        Some files need preprocessing. Currently we need to add a stripped POTCAR version
        and always restrict/embargo the original.
837
        '''
838
        if os.path.basename(path).startswith('POTCAR'):
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
            # create checksum
            hash = hashlib.sha224()
            with open(self.staging_upload_files.raw_file_object(path).os_path, 'rb') as orig_f:
                for line in orig_f.readlines():
                    hash.update(line)

            checksum = hash.hexdigest()

            # created stripped POTCAR
            stripped_path = path + '.stripped'
            with open(self.staging_upload_files.raw_file_object(stripped_path).os_path, 'wt') as stripped_f:
                stripped_f.write('Stripped POTCAR file. Checksum of original file (sha224): %s\n' % checksum)
            os.system(
                '''
                    awk < %s >> %s '
                    BEGIN { dump=1 }
                    /End of Dataset/ { dump=1 }
                    dump==1 { print }
                    /END of PSCTR/ { dump=0 }'
                ''' % (
                    self.staging_upload_files.raw_file_object(path).os_path,
                    self.staging_upload_files.raw_file_object(stripped_path).os_path))

862
    def match_mainfiles(self) -> Iterator[Tuple[str, object]]:
863
        '''
864
865
866
867
868
        Generator function that matches all files in the upload to all parsers to
        determine the upload's mainfiles.

        Returns:
            Tuples of mainfile, filename, and parsers
869
        '''
870
        directories_with_match: Dict[str, str] = dict()
871
872
        upload_files = self.staging_upload_files
        for filename in upload_files.raw_file_manifest():
873
            self._preprocess_files(filename)
874
            try:
875
                parser = match_parser(upload_files.raw_file_object(filename).os_path)
876
                if parser is not None:
877
878
                    directory = os.path.dirname(filename)
                    if directory in directories_with_match:
879
880
881
                        # TODO this might give us the chance to store directory based relationship
                        # between calcs for the future?
                        pass
882
883
                    else:
                        directories_with_match[directory] = filename
884
885

                    yield filename, parser
886
887
888
889
            except Exception as e:
                self.get_logger().error(
                    'exception while matching pot. mainfile',
                    mainfile=filename, exc_info=e)
890

Markus Scheidgen's avatar
Markus Scheidgen committed
891
892
    @task
    def parse_all(self):
893
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
894
        The *task* used to identify mainfile/parser combinations among the upload's files, creates
895
        respective :class:`Calc` instances, and triggers their processing.
896
        '''
897
898
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
899
900
        with utils.timer(
                logger, 'upload extracted', step='matching',
901
902
                upload_size=self.upload_files.size):
            for filename, parser in self.match_mainfiles():
903
                calc = Calc.create(
904
                    calc_id=self.upload_files.calc_id(filename),
905
                    mainfile=filename, parser=parser.name,
906
                    worker_hostname=self.worker_hostname,
907
908
                    upload_id=self.upload_id)

909
                calc.process_calc()
Markus Scheidgen's avatar
Markus Scheidgen committed
910

911
    def on_process_complete(self, process_name):
912
        if process_name == 'process_upload' or process_name == 're_process_upload':
913
914
915
            self.check_join()

    def check_join(self):
916
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
917
918
919
920
921
922
        Performs an evaluation of the join condition and triggers the :func:`cleanup`
        task if necessary. The join condition allows to run the ``cleanup`` after
        all calculations have been processed. The upload processing stops after all
        calculation processings have been triggered (:func:`parse_all` or
        :func:`re_process_upload`). The cleanup task is then run within the last
        calculation process (the one that triggered the join by calling this method).
923
        '''
924
925
926
927
        total_calcs = self.total_calcs
        processed_calcs = self.processed_calcs

        self.get_logger().debug('check join', processed_calcs=processed_calcs, total_calcs=total_calcs)
928
929
        # check if process is not running anymore, i.e. not still spawining new processes to join
        # check the join condition, i.e. all calcs have been processed
930
        if not self.process_running and processed_calcs >= total_calcs:
931
            # this can easily be called multiple times, e.g. upload finished after all calcs finished
932
933
934
            modified_upload = self._get_collection().find_one_and_update(
                {'_id': self.upload_id, 'joined': {'$ne': True}},
                {'$set': {'joined': True}})
935
936
937
938
939
940
            if modified_upload is not None:
                self.get_logger().debug('join')
                self.cleanup()
            else:
                # the join was already done due to a prior call
                pass
941

942
943
944
    def reset(self):
        self.joined = False
        super().reset()
Markus Scheidgen's avatar
Markus Scheidgen committed
945

946
947
948
949
950
951
    @classmethod
    def reset_pymongo_update(cls, worker_hostname: str = None):
        update = super().reset_pymongo_update()
        update.update(joined=False)
        return update

952
    def _cleanup_after_processing(self):
953
954
955
956
957
958
        # send email about process finish
        user = self.uploader
        name = '%s %s' % (user.first_name, user.last_name)
        message = '\n'.join([
            'Dear %s,' % name,
            '',