data.py 49.9 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
'''
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
from nomad.normalizing import normalizers
43
44
45
from nomad.datamodel import EntryArchive
from nomad.archive import query_archive
import phonopyparser
Markus Scheidgen's avatar
Markus Scheidgen committed
46
47


48
49
50
51
52
53
54
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']})
55
        log_data.update(logger=logger.name)
56
57
58
59
60
61
62
63
64
65
66

        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,
67
    TimeStamper(fmt="%Y-%m-%d %H:%M.%S", utc=False)]
68
69


Markus Scheidgen's avatar
Markus Scheidgen committed
70
class Calc(Proc):
71
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
72
73
74
75
76
    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.

77
78
    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
79
80

    Attributes:
81
        calc_id: the calc_id of this calc
Markus Scheidgen's avatar
Markus Scheidgen committed
82
83
84
        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
85

86
87
        metadata: the metadata record wit calc and user metadata, see :class:`datamodel.EntryMetadata`
    '''
88
    calc_id = StringField(primary_key=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
89
90
91
92
    upload_id = StringField()
    mainfile = StringField()
    parser = StringField()

93
94
    metadata = DictField()

Markus Scheidgen's avatar
Markus Scheidgen committed
95
    meta: Any = {
Markus Scheidgen's avatar
Markus Scheidgen committed
96
        'indexes': [
97
            'upload_id',
98
            'parser',
99
100
101
            ('upload_id', 'mainfile'),
            ('upload_id', 'parser'),
            ('upload_id', 'tasks_status'),
102
            ('upload_id', 'process_status'),
103
            ('upload_id', 'metadata.nomad_version'),
104
105
            'metadata.processed',
            'metadata.last_processing',
106
            'metadata.published',
107
            'metadata.datasets',
Markus Scheidgen's avatar
Markus Scheidgen committed
108
            'metadata.pid'
Markus Scheidgen's avatar
Markus Scheidgen committed
109
110
111
112
113
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
114
        self._parser_backend: Backend = None
115
116
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
117
        self._calc_proc_logs: List[Any] = None
118

119
        self._entry_metadata = None
120

Markus Scheidgen's avatar
Markus Scheidgen committed
121
122
    @classmethod
    def get(cls, id):
123
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
124

Markus Scheidgen's avatar
Markus Scheidgen committed
125
    @property
126
127
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
128

129
130
131
132
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
133
            self._upload.worker_hostname = self.worker_hostname
134
135
        return self._upload

136
137
138
139
140
141
142
143
144
145
146
147
    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()
148
149
        if self.parser is not None:
            entry_metadata.domain = parser_dict[self.parser].domain
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
        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)
174
175
176
177
178
179
180
181
182
183
184
185
        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()
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201

        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

202
203
204
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
205
206
            self._upload_files = ArchiveBasedStagingUploadFiles(
                self.upload_id, is_authorized=lambda: True, upload_path=self.upload.upload_path)
207
208
        return self._upload_files

Markus Scheidgen's avatar
Markus Scheidgen committed
209
    def get_logger(self, **kwargs):
210
        '''
211
212
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
213
        '''
214
215
216
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile, calc_id=self.calc_id, **kwargs)
217

218
219
        if self._calc_proc_logs is None:
            self._calc_proc_logs = []
220

221
222
        def save_to_calc_log(logger, method_name, event_dict):
            try:
223
224
                # 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()}
225
226
                dump_dict.update(level=method_name.upper())
                self._calc_proc_logs.append(dump_dict)
227

228
229
230
            except Exception:
                # Exceptions here will cause indefinite loop
                pass
231

232
            return event_dict
233

234
        return wrap_logger(logger, processors=_log_processors + [save_to_calc_log])
235

236
    @process
237
    def re_process_calc(self):
238
        '''
239
240
241
        Processes a calculation again. This means there is already metadata and
        instead of creating it initially, we are just updating the existing
        records.
242
        '''
243
        parser = match_parser(self.upload_files.raw_file_object(self.mainfile).os_path, strict=False)
244
        logger = self.get_logger()
245
246
247

        if parser is None and not config.reprocess_unmatched:
            self.errors = ['no parser matches during re-process, will not re-process this calc']
248

249
250
251
252
253
254
255
            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)
256
                raise e
257

258
259
260
261
262
            # mock the steps of actual processing
            self._continue_with('parsing')
            self._continue_with('normalizing')
            self._continue_with('archiving')
            self._complete()
263
264
            return

265
        if parser is None:
266
267
268
269
            self.get_logger().warn('no parser matches during re-process, use the old parser')
            self.warnings = ['no matching parser found during re-processing']

        elif self.parser != parser.name:
270
271
272
273
274
275
276
277
278
            if parser_dict[self.parser].name == parser.name:
                # parser was just renamed
                self.parser = parser.name

            else:
                self.parser = parser.name
                logger.info(
                    'different parser matches during re-process, use new parser',
                    parser=parser.name)
279

280
        try:
281
282
283
284
285
286
            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)
287
288
289
290
291
292
293

            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
            # close loghandler that was not closed due to failures
            try:
294
295
                if self._parser_backend and self._parser_backend.resource:
                    self._parser_backend.resource.unload()
296
            except Exception as e:
297
                logger.error('could unload processing results', exc_info=e)
298

299
    @process
300
    def process_calc(self):
301
        '''
302
303
        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.
304
        '''
305
        logger = self.get_logger()
306
        if self.upload is None:
307
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
308
309

        try:
310
311
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
312
313
314
315
316
317
            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:
318
319
320
321
                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
322
323
324
325
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
326
            # close loghandler that was not closed due to failures
327
            try:
328
329
                if self._parser_backend and self._parser_backend.resource:
                    self._parser_backend.resource.unload()
330
            except Exception as e:
331
                logger.error('could unload processing results', exc_info=e)
332

333
    def on_fail(self):
334
335
336
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
337
            self._entry_metadata.processed = False
338

339
            self.apply_entry_metadata(self._entry_metadata)
340
341
342
343
344
            if self._parser_backend and self._parser_backend.resource:
                backend = self._parser_backend
            else:
                backend = None
            self._entry_metadata.apply_domain_metadata(backend)
345

346
            self._entry_metadata.a_elastic.index()
347
        except Exception as e:
348
349
            self.get_logger().error(
                'could not index after processing failure', exc_info=e)
350

351
        try:
352
            self.write_archive(None)
353
        except Exception as e:
354
355
            self.get_logger().error(
                'could not write archive after processing failure', exc_info=e)
356

357
358
359
360
361
362
363
364
    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
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
            self.upload.reload()
            self.upload.check_join()

Markus Scheidgen's avatar
Markus Scheidgen committed
365
366
    @task
    def parsing(self):
367
        ''' The *task* that encapsulates all parsing related actions. '''
368
        context = dict(parser=self.parser, step=self.parser)
369
        logger = self.get_logger(**context)
370
        parser = parser_dict[self.parser]
371
        self._entry_metadata.parser_name = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
372

373
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
374
375
376
377
            try:
                self._parser_backend = parser.run(
                    self.upload_files.raw_file_object(self.mainfile).os_path, logger=logger)
            except Exception as e:
378
                self.fail('parser failed with exception', exc_info=e, error=str(e), **context)
379
                return
380
            except SystemExit:
381
                self.fail('parser raised system exit', error='system exit', **context)
382
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
383

Markus Scheidgen's avatar
Markus Scheidgen committed
384
385
        if self._parser_backend.status[0] != 'ParseSuccess':
            error = self._parser_backend.status[1]
386
            self.fail('parser failed', error=error, **context)
387

388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
    def process_phonon(self):
        """Function that is run for phonon calculation before cleanup.
        This task is run by the celery process that is calling the join for the
        upload.

        This function re-opens the Archive for this calculation to add method
        information from another referenced archive. Updates the method
        information in section_encyclopedia as well as the DFT domain metadata.
        """
        try:
            # Re-create a backend
            context = dict(parser=self.parser, step=self.parser)
            logger = self.get_logger(**context)
            metainfo = phonopyparser.metainfo.m_env
            backend = Backend(metainfo, logger=logger, domain="dft")

            # Open the archive of the phonon calculation.
            upload_files = StagingUploadFiles(self.upload_id, is_authorized=lambda: True)
            with upload_files.read_archive(self.calc_id) as archive:
                arch = query_archive(archive, {self.calc_id: self.calc_id})[self.calc_id]
                phonon_archive = EntryArchive.m_from_dict(arch)
409
410
411
412
413
414

            # Save Archive contents, metadata and logs from the old entry
            backend.entry_archive = phonon_archive
            self._parser_backend = backend
            self._entry_metadata = backend.entry_archive.section_metadata
            self._calc_proc_logs = phonon_archive.processing_logs
415
416
417
418
419

            # Read in the first referenced calculation. The reference is given as
            # an absolute path which needs to be converted into a path that is
            # relative to upload root.
            scc = backend.entry_archive.section_run[0].section_single_configuration_calculation[0]
420
            relative_ref = scc.section_calculation_to_calculation_refs[0].calculation_to_calculation_external_url
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
            ref_id = upload_files.calc_id(relative_ref)
            with upload_files.read_archive(ref_id) as archive:
                arch = query_archive(archive, {ref_id: ref_id})[ref_id]
                ref_archive = EntryArchive.m_from_dict(arch)

            # Get encyclopedia method information directly from the referenced calculation.
            ref_enc_method = ref_archive.section_encyclopedia.method
            backend.entry_archive.section_encyclopedia.method = ref_enc_method

            # Overwrite old entry with new data. The metadata is updated with
            # new timestamp and method details taken from the referenced
            # archive.
            self._entry_metadata.last_processing = datetime.utcnow()
            self._entry_metadata.dft.xc_functional = ref_archive.section_metadata.dft.xc_functional
            self._entry_metadata.dft.basis_set = ref_archive.section_metadata.dft.basis_set
            self._entry_metadata.dft.update_group_hash()

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

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

            # persist the archive
            with utils.timer(
                    logger, 'archived', step='archive',
                    input_size=self.mainfile_file.size) as log_data:

                archive_size = self.write_archive(self._parser_backend)
                log_data.update(archive_size=archive_size)

        except Exception as e:
            logger.error("Could not retrieve method information for phonon calculation.", exception=e)

457
458
459
460
461
462
463
    @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', [])
464

465
            if len(warnings) > 0:
466
467
468
469
                self.get_logger().warn(
                    'processor completed successful with warnings',
                    processor=processor_name, warnings=[str(warning) for warning in warnings])

470
            else:
471
472
473
474
                self.get_logger().info(
                    'processor completed successful',
                    processor=processor_name)

475
        else:
476
            errors = self._parser_backend.status[1]
477
478
479
            self.get_logger().error(
                'processor completed with failure',
                processor=processor_name, errors=str(errors))
Markus Scheidgen's avatar
Markus Scheidgen committed
480
481
482

    @task
    def normalizing(self):
483
        ''' The *task* that encapsulates all normalizing related actions. '''
484
485
486
487
488

        # 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
489
        for normalizer in normalizers:
490
            if normalizer.domain != parser_dict[self.parser].domain:
491
492
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
493
            normalizer_name = normalizer.__name__
494
            context = dict(normalizer=normalizer_name, step=normalizer_name)
495
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
496
497

            with utils.timer(
498
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
499
                with self.use_parser_backend(normalizer_name) as backend:
500
501
502
                    try:
                        normalizer(backend).normalize(logger=logger)
                    except Exception as e:
503
                        self._parser_backend.finishedParsingSession('ParseFailure', [str(e)])
504
505
506
                        logger.error(
                            'normalizer failed with exception', exc_info=e,
                            error=str(e), **context)
507
508
509
510
511
512
513
514
515
                        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
516
517
518

    @task
    def archiving(self):
519
        ''' The *task* that encapsulates all archival related actions. '''
520
521
        logger = self.get_logger()

522
523
        self._entry_metadata.apply_domain_metadata(self._parser_backend)
        self._entry_metadata.processed = True
524

525
526
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
527
            self.apply_entry_metadata(self._entry_metadata)
528
529

        # index in search
530
        with utils.timer(logger, 'indexed', step='index'):
531
            self._entry_metadata.a_elastic.index()
532

533
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
534
        with utils.timer(
535
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
536
                input_size=self.mainfile_file.size) as log_data:
537

538
            archive_size = self.write_archive(self._parser_backend)
539
            log_data.update(archive_size=archive_size)
Markus Scheidgen's avatar
Markus Scheidgen committed
540

541
    def write_archive(self, backend: Backend):
542
543
544
545
546
547
        def filter_processing_logs(logs):
            if len(logs) > 100:
                return [
                    log for log in logs
                    if log.get('level') != 'DEBUG']
            return logs
Markus Scheidgen's avatar
Markus Scheidgen committed
548

549
550
551
552
553
554
555
556
        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()

557
558
559
        if entry_archive.section_metadata is None:
            entry_archive.m_add_sub_section(datamodel.EntryArchive.section_metadata, self._entry_metadata)

560
        entry_archive.processing_logs = filter_processing_logs(self._calc_proc_logs)
561
562
563
564
565
566
567
568
569
570

        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)
571
            entry_archive.processing_logs = filter_processing_logs(self._calc_proc_logs)
572
573
            self.upload_files.write_archive(self.calc_id, entry_archive.m_to_dict())
            raise e
Markus Scheidgen's avatar
Markus Scheidgen committed
574

575
    def __str__(self):
576
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
577

578

579
class Upload(Proc):
580
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
581
582
583
584
585
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
586
587
        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
588
589
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
590
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
591
592
        published: Boolean that indicates the publish status
        publish_time: Date when the upload was initially published
593
        last_update: Date of the last publishing/re-processing
594
        joined: Boolean indicates if the running processing has joined (:func:`check_join`)
595
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
596
597
598
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
599
600
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
601
    embargo_length = IntField(default=36)
Markus Scheidgen's avatar
Markus Scheidgen committed
602
603
604

    name = StringField(default=None)
    upload_time = DateTimeField()
605
    user_id = StringField(required=True)
606
607
    published = BooleanField(default=False)
    publish_time = DateTimeField()
608
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
609

610
611
    joined = BooleanField(default=False)

Markus Scheidgen's avatar
Markus Scheidgen committed
612
613
    meta: Any = {
        'indexes': [
614
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
615
        ]
Markus Scheidgen's avatar
Markus Scheidgen committed
616
617
618
619
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
620
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
621

622
623
    @property
    def metadata(self) -> dict:
624
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
625
626
627
628
629
        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.
630
        '''
631
632
633
634
635
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
636
637
638

    @metadata.setter
    def metadata(self, data: dict) -> None:
639
640
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
641

Markus Scheidgen's avatar
Markus Scheidgen committed
642
    @classmethod
643
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
644
        return cls.get_by_id(id, 'upload_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
645
646

    @classmethod
647
    def user_uploads(cls, user: datamodel.User, **kwargs) -> List['Upload']:
648
        ''' Returns all uploads for the given user. Kwargs are passed to mongo query. '''
649
        return cls.objects(user_id=str(user.user_id), **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
650

651
652
    @property
    def uploader(self):
653
        return datamodel.User.get(self.user_id)
654

Markus Scheidgen's avatar
Markus Scheidgen committed
655
656
    def get_logger(self, **kwargs):
        logger = super().get_logger()
657
658
        user = self.uploader
        user_name = '%s %s' % (user.first_name, user.last_name)
659
        # We are not using 'user_id' because logstash (?) will filter these entries ?!
660
661
        logger = logger.bind(
            upload_id=self.upload_id, upload_name=self.name, user_name=user_name,
662
            user=user.user_id, **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
663
664
665
666
        return logger

    @classmethod
    def create(cls, **kwargs) -> 'Upload':
667
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
668
669
670
        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.
671
672

        Arguments:
673
            user: The user that created the upload.
674
        '''
675
676
        # use kwargs to keep compatibility with super method
        user: datamodel.User = kwargs['user']
677
        del(kwargs['user'])
678

679
680
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
681
        kwargs.update(user_id=user.user_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
682
        self = super().create(**kwargs)
683

Markus Scheidgen's avatar
Markus Scheidgen committed
684
        self._continue_with('uploading')
685

Markus Scheidgen's avatar
Markus Scheidgen committed
686
687
        return self

688
    def delete(self):
689
        ''' Deletes this upload process state entry and its calcs. '''
690
691
692
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

693
    def delete_upload_local(self):
694
        '''
695
        Deletes the upload, including its processing state and
696
        staging files. Local version without celery processing.
697
        '''
698
699
        logger = self.get_logger()

700
        with utils.lnr(logger, 'upload delete failed'):
701
            with utils.timer(
702
                    logger, 'upload deleted from index', step='index',
703
                    upload_size=self.upload_files.size):
704
                search.delete_upload(self.upload_id)
705

706
            with utils.timer(
707
                    logger, 'upload deleted', step='files',
708
709
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
710
711

            self.delete()
712

713
    @process
714
    def delete_upload(self):
715
        '''
716
717
        Deletes of the upload, including its processing state and
        staging files. This starts the celery process of deleting the upload.
718
        '''
719
        self.delete_upload_local()
720

721
        return True  # do not save the process status on the delete upload
722

723
    @process
724
    def publish_upload(self):
725
        '''
726
727
        Moves the upload out of staging to the public area. It will
        pack the staging upload files in to public upload files.
728
        '''
729
730
        assert self.processed_calcs > 0

731
        logger = self.get_logger()
732
        logger.info('started to publish')
733

734
        with utils.lnr(logger, 'publish failed'):
735
            with self.entries_metadata(self.metadata) as calcs:
736

737
                with utils.timer(
738
                        logger, 'upload metadata updated', step='metadata',
739
                        upload_size=self.upload_files.size):
740

741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
                    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)
756

757
                with utils.timer(
758
                        logger, 'index updated', step='index',
759
                        upload_size=self.upload_files.size):
760
761
762
763
764
765
766
767
768
769
770
771
                    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:
772
                    self.last_update = datetime.utcnow()
773
                    self.save()
774

775
776
    @process
    def re_process_upload(self):
777
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
778
779
780
        A *process* that performs the re-processing of a earlier processed
        upload.

781
782
783
784
785
786
        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.
787
        '''
788
789
790
        logger = self.get_logger()
        logger.info('started to re-process')

791
792
793
794
795
        # 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')
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810

        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)
811
812

        self._continue_with('parse_all')
813
        try:
814
            # check if a calc is already/still processing
815
816
817
818
819
820
821
822
            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)
823

824
            # reset all calcs
825
            Calc._get_collection().update_many(
826
827
828
                dict(upload_id=self.upload_id),
                {'$set': Calc.reset_pymongo_update(worker_hostname=self.worker_hostname)})

829
830
            # process call calcs
            Calc.process_all(Calc.re_process_calc, dict(upload_id=self.upload_id), exclude=['metadata'])
831
832

            logger.info('completed to trigger re-process of all calcs')
833
834
        except Exception as e:
            # try to remove the staging copy in failure case
835
836
            logger.error('failed to trigger re-process of all calcs', exc_info=e)

837
838
839
            if self.published:
                if staging_upload_files is not None and staging_upload_files.exists():
                    staging_upload_files.delete()
840
841

            raise e
842
843
844
845

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

846
847
    @process
    def re_pack(self):
848
        ''' A *process* that repacks the raw and archive data based on the current embargo data. '''
849
850
851
852
853
854
855
856
        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')

857
        self.upload_files.re_pack(self.user_metadata())
858
        self.joined = True
859
860
        self._complete()

Markus Scheidgen's avatar
Markus Scheidgen committed
861
    @process
862
    def process_upload(self):
863
        ''' A *process* that performs the initial upload processing. '''
Markus Scheidgen's avatar
Markus Scheidgen committed
864
865
866
867
868
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
869
        ''' A no-op *task* as a stand-in for receiving upload data. '''
Markus Scheidgen's avatar
Markus Scheidgen committed
870
871
        pass

872
    @property
873
874
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
875
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
876
877
878

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
879
                self.upload_id, is_authorized=lambda: True, **kwargs)
880

881
        return self._upload_files
882

883
884
885
886
887
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
888
889
    @task
    def extracting(self):
890
        '''
891
892
        The *task* performed before the actual parsing/normalizing: extracting
        the uploaded files.
893
        '''
894
895
896
897
898
        # 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
899
900
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
901
902
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
903
904
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
905
906
907
908

            if self.temporary:
                os.remove(self.upload_path)
                self.upload_path = None
909

910
        except KeyError:
911
            self.fail('processing requested for non existing upload', log_level=logging.ERROR)
912
913
914
            return
        except ExtractError:
            self.fail('bad .zip/.tar file', log_level=logging.INFO)
Markus Scheidgen's avatar
Markus Scheidgen committed
915
916
            return

917
    def _preprocess_files(self, path):
918
        '''
919
920
        Some files need preprocessing. Currently we need to add a stripped POTCAR version
        and always restrict/embargo the original.
921
        '''
922
        if os.path.basename(path).startswith('POTCAR'):
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
            # 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))

946
947
948
949
    def match_mainfiles(self) -> Iterator[Tuple[str, object]]:
        '''
        Generator function that matches all files in the upload to all parsers to
        determine the upload's mainfiles.
950
951

        Returns:
952
953
            Tuples of mainfile, filename, and parsers
        '''
954
        directories_with_match: Dict[str, str] = dict()
955
        upload_files = self.staging_upload_files
956
957
        for filename in upload_files.raw_file_manifest():
            self._preprocess_files(filename)
958
            try:
959
                parser = match_parser(upload_files.raw_file_object(filename).os_path)
960
                if parser is not None:
961
                    directory = os.path.dirname(filename)
962
                    if directory in directories_with_match:
963
964
965
                        # TODO this might give us the chance to store directory based relationship
                        # between calcs for the future?
                        pass
966
                    else:
967
968
969
                        directories_with_match[directory] = filename

                    yield filename, parser
970
971
972
            except Exception as e:
                self.get_logger().error(
                    'exception while matching pot. mainfile',
973
                    mainfile=filename, exc_info=e)