data.py 44.9 KB
Newer Older
Markus Scheidgen's avatar
Markus Scheidgen committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# 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.

"""
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

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

Markus Scheidgen's avatar
Markus Scheidgen committed
27
from typing import cast, List, Any, ContextManager, Tuple, Generator, Dict, cast
28
from mongoengine import StringField, DateTimeField, DictField, BooleanField
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
Markus Scheidgen's avatar
Markus Scheidgen committed
36

37
from nomad import utils, coe_repo, config, infrastructure, search, datamodel
38
from nomad.files import PathObject, UploadFiles, ExtractError, ArchiveBasedStagingUploadFiles, PublicUploadFiles, StagingUploadFiles
39
from nomad.processing.base import Proc, process, task, PENDING, SUCCESS, FAILURE
40
from nomad.parsing import parser_dict, match_parser, LocalBackend
Markus Scheidgen's avatar
Markus Scheidgen committed
41
from nomad.normalizing import normalizers
42
from nomad.datamodel import UploadWithMetadata, Domain
Markus Scheidgen's avatar
Markus Scheidgen committed
43
44


Markus Scheidgen's avatar
Markus Scheidgen committed
45
class Calc(Proc):
Markus Scheidgen's avatar
Markus Scheidgen committed
46
47
48
49
50
51
52
53
54
55
    """
    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.

    The attribute list, does not include the various repository properties generated
    while parsing, including ``program_name``, ``program_version``, etc.

    Attributes:
56
        calc_id: the calc_id of this calc
Markus Scheidgen's avatar
Markus Scheidgen committed
57
58
59
        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
60

61
        metadata: the metadata record wit calc and user metadata, see :class:`datamodel.CalcWithMetadata`
Markus Scheidgen's avatar
Markus Scheidgen committed
62
    """
63
    calc_id = StringField(primary_key=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
64
65
66
67
    upload_id = StringField()
    mainfile = StringField()
    parser = StringField()

68
69
    metadata = DictField()

Markus Scheidgen's avatar
Markus Scheidgen committed
70
    meta: Any = {
Markus Scheidgen's avatar
Markus Scheidgen committed
71
        'indexes': [
72
73
74
75
            'upload_id',
            ('upload_id', 'mainfile'),
            ('upload_id', 'parser'),
            ('upload_id', 'tasks_status'),
76
77
            ('upload_id', 'process_status'),
            ('upload_id', 'metadata.nomad_version')
Markus Scheidgen's avatar
Markus Scheidgen committed
78
79
80
81
82
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
83
        self._parser_backend: LocalBackend = None
84
85
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
86
        self._calc_proc_logwriter = None
87
        self._calc_proc_logwriter_ctx: ContextManager = None
Markus Scheidgen's avatar
Markus Scheidgen committed
88

89
90
91
92
93
94
95
96
97
98
    @classmethod
    def from_calc_with_metadata(cls, calc_with_metadata):
        calc = Calc.create(
            calc_id=calc_with_metadata.calc_id,
            upload_id=calc_with_metadata.upload_id,
            mainfile=calc_with_metadata.mainfile,
            metadata=calc_with_metadata.to_dict())

        return calc

Markus Scheidgen's avatar
Markus Scheidgen committed
99
100
    @classmethod
    def get(cls, id):
101
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
102

Markus Scheidgen's avatar
Markus Scheidgen committed
103
    @property
104
105
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
106

107
108
109
110
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
111
            self._upload.worker_hostname = self.worker_hostname
112
113
        return self._upload

114
115
116
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
117
118
            self._upload_files = ArchiveBasedStagingUploadFiles(
                self.upload_id, is_authorized=lambda: True, upload_path=self.upload.upload_path)
119
120
        return self._upload_files

Markus Scheidgen's avatar
Markus Scheidgen committed
121
    def get_logger(self, **kwargs):
122
123
124
125
        """
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
        """
126
127
128
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile, calc_id=self.calc_id, **kwargs)
129

130
        if self._calc_proc_logwriter_ctx is None:
131
132
133
134
135
136
137
138
139
140
141
142
143
144
            try:
                self._calc_proc_logwriter_ctx = self.upload_files.archive_log_file(self.calc_id, 'wt')
                self._calc_proc_logwriter = self._calc_proc_logwriter_ctx.__enter__()  # pylint: disable=E1101
            except KeyError:
                # cannot open log file
                pass

        if self._calc_proc_logwriter_ctx is None:
            return logger
        else:
            def save_to_calc_log(logger, method_name, event_dict):
                if self._calc_proc_logwriter is not None:
                    program = event_dict.get('normalizer', 'parser')
                    event = event_dict.get('event', '')
145
146
147
148

                    entry = '[%s] %s, %s: %s' % (method_name, datetime.utcnow().isoformat(), program, event)
                    if len(entry) > 140:
                        self._calc_proc_logwriter.write(entry[:140])
149
150
151
152
153
154
155
156
                        self._calc_proc_logwriter.write('...')
                    else:
                        self._calc_proc_logwriter.write(entry)
                    self._calc_proc_logwriter.write('\n')

                return event_dict

            return wrap_logger(logger, processors=[save_to_calc_log])
157

158
159
160
161
162
163
164
165
166
167
    @process
    def re_process_calc(self):
        """
        Processes a calculation again. This means there is already metadata and
        instead of creating it initially, we are just updating the existing
        records.
        """
        logger = self.get_logger()

        try:
168
169
170
171
172
173
174
            calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
            calc_with_metadata.upload_id = self.upload_id
            calc_with_metadata.calc_id = self.calc_id
            calc_with_metadata.calc_hash = self.upload_files.calc_hash(self.mainfile)
            calc_with_metadata.mainfile = self.mainfile
            calc_with_metadata.nomad_version = config.version
            calc_with_metadata.nomad_commit = config.commit
175
            calc_with_metadata.last_processing = datetime.utcnow()
176
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
177
            self.metadata = calc_with_metadata.to_dict()
178
179
180
181
182
183
184
185
186
187
188
189
190

            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
            # close loghandler that was not closed due to failures
            try:
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

Markus Scheidgen's avatar
Markus Scheidgen committed
191
    @process
192
    def process_calc(self):
193
194
195
196
        """
        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.
        """
197
        logger = self.get_logger()
198
        if self.upload is None:
199
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
200
201

        try:
202
203
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
204
            calc_with_metadata = datamodel.CalcWithMetadata(
205
206
207
208
209
210
                upload_id=self.upload_id,
                calc_id=self.calc_id,
                calc_hash=self.upload_files.calc_hash(self.mainfile),
                mainfile=self.mainfile)
            calc_with_metadata.published = False
            calc_with_metadata.uploader = self.upload.uploader.to_popo()
211
            calc_with_metadata.upload_time = self.upload.upload_time
212
            calc_with_metadata.nomad_version = config.version
213
            calc_with_metadata.nomad_commit = config.commit
214
            calc_with_metadata.last_processing = datetime.utcnow()
215
216
217
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
            self.metadata = calc_with_metadata.to_dict()

218
219
220
221
222
            if len(calc_with_metadata.files) >= config.auxfile_cutoff:
                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
223
224
225
226
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
227
            # close loghandler that was not closed due to failures
228
            try:
229
230
231
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
232
233
234
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

235
236
237
238
    def fail(self, *errors, log_level=logging.ERROR, **kwargs):
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
239
            calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
240
241
242
243
244
245
246
247
            calc_with_metadata.formula = config.services.not_processed_value
            calc_with_metadata.basis_set = config.services.not_processed_value
            calc_with_metadata.xc_functional = config.services.not_processed_value
            calc_with_metadata.system = config.services.not_processed_value
            calc_with_metadata.crystal_system = config.services.not_processed_value
            calc_with_metadata.spacegroup = config.services.not_processed_value
            calc_with_metadata.spacegroup_symbol = config.services.not_processed_value
            calc_with_metadata.code_version = config.services.not_processed_value
248
249
250
251
252
253
254

            calc_with_metadata.code_name = config.services.not_processed_value
            if self.parser is not None:
                parser = parser_dict[self.parser]
                if hasattr(parser, 'code_name'):
                    calc_with_metadata.code_name = parser.code_name

255
256
257
258
259
260
261
262
            calc_with_metadata.processed = False
            self.metadata = calc_with_metadata.to_dict()
            search.Entry.from_calc_with_metadata(calc_with_metadata).save()
        except Exception as e:
            self.get_logger().error('could not index after processing failure', exc_info=e)

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

263
264
265
266
    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
267
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
268
269
            self.upload.reload()
            self.upload.check_join()
Markus Scheidgen's avatar
Markus Scheidgen committed
270
271
272

    @task
    def parsing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
273
        """ The *task* that encapsulates all parsing related actions. """
274
        context = dict(parser=self.parser, step=self.parser)
275
        logger = self.get_logger(**context)
276
        parser = parser_dict[self.parser]
277
        self.metadata['parser_name'] = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
278

279
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
280
281
282
283
284
285
286
            try:
                self._parser_backend = parser.run(
                    self.upload_files.raw_file_object(self.mainfile).os_path, logger=logger)
            except Exception as e:
                self.fail(
                    'parser failed with exception', level=logging.ERROR,
                    exc_info=e, error=str(e), **context)
287
                return
288
289
290
291
292
            except SystemExit:
                self.fail(
                    'parser raised system exit', level=logging.ERROR,
                    error='system exit', **context)
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
293

294
295
        # add the non code specific calc metadata to the backend
        # all other quantities have been determined by parsers/normalizers
296
        self._parser_backend.openNonOverlappingSection('section_entry_info')
297
        self._parser_backend.addValue('upload_id', self.upload_id)
298
        self._parser_backend.addValue('calc_id', self.calc_id)
299
        self._parser_backend.addValue('calc_hash', self.metadata['calc_hash'])
300
        self._parser_backend.addValue('mainfile', self.mainfile)
301
        self._parser_backend.addValue('parser_name', self.parser)
302
303
304
305
306
307
308
309
310
311
312
313
        filepaths = self.metadata['files']
        self._parser_backend.addValue('number_of_files', len(filepaths))
        self._parser_backend.addValue('filepaths', filepaths)
        uploader = self.upload.uploader
        self._parser_backend.addValue(
            'entry_uploader_name', '%s, %s' % (uploader.first_name, uploader.last_name))
        self._parser_backend.addValue(
            'entry_uploader_id', str(uploader.user_id))
        self._parser_backend.addValue('entry_upload_time', int(self.upload.upload_time.timestamp()))
        self._parser_backend.closeNonOverlappingSection('section_entry_info')

        self.add_processor_info(self.parser)
314

Markus Scheidgen's avatar
Markus Scheidgen committed
315
        if self._parser_backend.status[0] != 'ParseSuccess':
316
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
317
            error = self._parser_backend.status[1]
318
            self.fail(error, level=logging.INFO, **context)
319
320
321
322
323
324
325
326

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

    def add_processor_info(self, processor_name: str) -> None:
327
        self._parser_backend.openContext('/section_entry_info/0')
328
329
330
331
332
333
334
        self._parser_backend.openNonOverlappingSection('section_archive_processing_info')
        self._parser_backend.addValue('archive_processor_name', processor_name)

        if self._parser_backend.status[0] == 'ParseSuccess':
            warnings = getattr(self._parser_backend, '_warnings', [])
            if len(warnings) > 0:
                self._parser_backend.addValue('archive_processor_status', 'WithWarnings')
335
                self._parser_backend.addValue('number_of_archive_processor_warnings', len(warnings))
336
337
338
339
                self._parser_backend.addArrayValues('archive_processor_warnings', [str(warning) for warning in warnings])
            else:
                self._parser_backend.addValue('archive_processor_status', 'Success')
        else:
340
            errors = self._parser_backend.status[1]
341
342
343
            self._parser_backend.addValue('archive_processor_error', str(errors))

        self._parser_backend.closeNonOverlappingSection('section_archive_processing_info')
344
        self._parser_backend.closeContext('/section_entry_info/0')
Markus Scheidgen's avatar
Markus Scheidgen committed
345
346
347

    @task
    def normalizing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
348
        """ The *task* that encapsulates all normalizing related actions. """
Markus Scheidgen's avatar
Markus Scheidgen committed
349
        for normalizer in normalizers:
350
351
352
            if normalizer.domain != config.domain:
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
353
            normalizer_name = normalizer.__name__
354
            context = dict(normalizer=normalizer_name, step=normalizer_name)
355
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
356
357

            with utils.timer(
358
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
359
                with self.use_parser_backend(normalizer_name) as backend:
360
361
362
363
364
365
366
                    try:
                        normalizer(backend).normalize(logger=logger)
                    except Exception as e:
                        self.fail(
                            'normalizer failed with exception', level=logging.ERROR,
                            exc_info=e, error=str(e), **context)
                        self._parser_backend.status = ['ParseFailure', str(e)]
Markus Scheidgen's avatar
Markus Scheidgen committed
367

368
369
            failed = self._parser_backend.status[0] != 'ParseSuccess'
            if failed:
370
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
371
                error = self._parser_backend.status[1]
372
                self.fail(error, level=logging.WARNING, error=error, **context)
373
374
375
376
                break
            else:
                logger.debug(
                    'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
377
378
379

    @task
    def archiving(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
380
        """ The *task* that encapsulates all archival related actions. """
381
382
        logger = self.get_logger()

383
        calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
384
        calc_with_metadata.apply_domain_metadata(self._parser_backend)
385
        calc_with_metadata.processed = True
386

387
388
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
389
            self.metadata = calc_with_metadata.to_dict()
390
391

        # index in search
392
        with utils.timer(logger, 'indexed', step='index'):
393
            search.Entry.from_calc_with_metadata(calc_with_metadata).save()
394

395
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
396
        with utils.timer(
397
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
398
                input_size=self.mainfile_file.size) as log_data:
399
            with self.upload_files.archive_file(self.calc_id, 'wt') as out:
400
                self._parser_backend.write_json(out, pretty=True, root_sections=Domain.instance.root_sections)
401

402
            log_data.update(archive_size=self.upload_files.archive_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
403
404
405
406

        # close loghandler
        if self._calc_proc_logwriter is not None:
            with utils.timer(
407
                    logger, 'archived log', step='logs',
Markus Scheidgen's avatar
Markus Scheidgen committed
408
                    input_size=self.mainfile_file.size) as log_data:
409
                self._calc_proc_logwriter_ctx.__exit__(None, None, None)  # pylint: disable=E1101
410
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
411

412
                log_data.update(log_size=self.upload_files.archive_log_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
413

414
    def __str__(self):
415
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
416

417

418
class Upload(Proc):
Markus Scheidgen's avatar
Markus Scheidgen committed
419
420
421
422
423
424
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
425
426
        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
427
428
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
429
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
430
431
432
433
        published: Boolean that indicates the publish status
        publish_time: Date when the upload was initially published
        last_update: Date of the last (re-)publishing
        joined: Boolean indicates if the running processing has joined (:func:`check_join`)
Markus Scheidgen's avatar
Markus Scheidgen committed
434
435
436
437
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
438
439
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
Markus Scheidgen's avatar
Markus Scheidgen committed
440
441
442

    name = StringField(default=None)
    upload_time = DateTimeField()
443
    user_id = StringField(required=True)
444
445
    published = BooleanField(default=False)
    publish_time = DateTimeField()
446
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
447

448
449
    joined = BooleanField(default=False)

Markus Scheidgen's avatar
Markus Scheidgen committed
450
451
    meta: Any = {
        'indexes': [
452
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
Markus Scheidgen's avatar
Markus Scheidgen committed
453
454
455
456
457
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
458
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
459

460
461
    @property
    def metadata(self) -> dict:
Markus Scheidgen's avatar
Markus Scheidgen committed
462
463
464
465
466
467
468
        """
        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.
        """
469
470
471
472
473
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
474
475
476

    @metadata.setter
    def metadata(self, data: dict) -> None:
477
478
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
479

Markus Scheidgen's avatar
Markus Scheidgen committed
480
    @classmethod
481
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
482
        upload = cls.get_by_id(id, 'upload_id')
483
        if upload is not None:
484
485
486
            return upload

        raise KeyError()
Markus Scheidgen's avatar
Markus Scheidgen committed
487
488

    @classmethod
489
490
491
    def user_uploads(cls, user: coe_repo.User, **kwargs) -> List['Upload']:
        """ Returns all uploads for the given user. Kwargs are passed to mongo query. """
        return cls.objects(user_id=str(user.user_id), **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
492

493
494
495
496
    @property
    def uploader(self):
        return coe_repo.User.from_user_id(self.user_id)

Markus Scheidgen's avatar
Markus Scheidgen committed
497
498
    def get_logger(self, **kwargs):
        logger = super().get_logger()
499
500
501
502
503
        user = self.uploader
        user_name = '%s %s' % (user.first_name, user.last_name)
        logger = logger.bind(
            upload_id=self.upload_id, upload_name=self.name, user_name=user_name,
            user_id=user.user_id, **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
504
505
506
507
508
509
510
511
        return logger

    @classmethod
    def create(cls, **kwargs) -> 'Upload':
        """
        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.
512
513

        Arguments:
514
            user (coe_repo.User): The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
515
        """
516
        user: coe_repo.User = kwargs['user']
517
518
519
        del(kwargs['user'])
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
520
        kwargs.update(user_id=str(user.user_id))
Markus Scheidgen's avatar
Markus Scheidgen committed
521
        self = super().create(**kwargs)
522

Markus Scheidgen's avatar
Markus Scheidgen committed
523
        self._continue_with('uploading')
524

Markus Scheidgen's avatar
Markus Scheidgen committed
525
526
        return self

527
528
529
530
531
    def delete(self):
        """ Deletes this upload process state entry and its calcs. """
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

532
    def delete_upload_local(self, with_coe_repo: bool = False):
533
534
        """
        Deletes of the upload, including its processing state and
535
        staging files. Local version without celery processing.
536
537
538
539
        """
        logger = self.get_logger()

        with utils.lnr(logger, 'staged upload delete failed'):
540

541
542
543
544
545
            if with_coe_repo and self.published:
                with utils.timer(
                        logger, 'upload deleted from repo db', step='repo',
                        upload_size=self.upload_files.size):
                    coe_repo.Upload.delete(self.upload_id)
546

547
            with utils.timer(
548
                    logger, 'upload deleted from index', step='index',
549
                    upload_size=self.upload_files.size):
550
                search.delete_upload(self.upload_id)
551

552
            with utils.timer(
553
                    logger, 'staged upload deleted', step='files',
554
555
556
557
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
                self.delete()

558
559
560
561
562
563
564
565
    @process
    def delete_upload(self, with_coe_repo: bool = False):
        """
        Deletes of the upload, including its processing state and
        staging files. This starts the celery process of deleting the upload.
        """
        self.delete_upload_local(with_coe_repo=with_coe_repo)

566
        return True  # do not save the process status on the delete upload
567

568
    @process
569
    def publish_upload(self):
570
571
572
573
574
        """
        Moves the upload out of staging to add it to the coe repository. It will
        pack the staging upload files in to public upload files, add entries to the
        coe repository db and remove this instance and its calculation from the
        processing state db.
575
576
577

        If the upload is already published (i.e. re-publish), it will update user metadata from
        repository db, publish to repository db if not exists, update the search index.
578
        """
579
580
        assert self.processed_calcs > 0

581
        logger = self.get_logger()
582
        logger.info('started to publish')
583

584
585
        with utils.lnr(logger, '(re-)publish failed'):
            upload_with_metadata = self.to_upload_with_metadata(self.metadata)
586
            calcs = upload_with_metadata.calcs
587

588
            if config.repository_db.publish_enabled:
589
590
591
592
593
594
595
                if config.repository_db.mode == 'coe' and isinstance(self.upload_files, StagingUploadFiles):
                    with utils.timer(
                            logger, 'coe extracted raw-file copy created', step='repo',
                            upload_size=self.upload_files.size):

                        self.upload_files.create_extracted_copy()

596
597
598
599
600
601
602
                coe_upload = coe_repo.Upload.from_upload_id(upload_with_metadata.upload_id)
                if coe_upload is None:
                    with utils.timer(
                            logger, 'upload added to repository', step='repo',
                            upload_size=self.upload_files.size):
                        coe_upload = coe_repo.Upload.publish(upload_with_metadata)

603
                with utils.timer(
604
                        logger, 'upload PIDs read from repository', step='repo',
605
                        upload_size=self.upload_files.size):
606
607
                    for calc, coe_calc in zip(calcs, coe_upload.calcs):
                        calc.pid = coe_calc.coe_calc_id
608

609
            with utils.timer(
610
                    logger, 'upload metadata updated', step='metadata',
611
                    upload_size=self.upload_files.size):
612
613

                def create_update(calc):
614
                    calc.published = True
615
616
617
618
619
                    return UpdateOne(
                        {'_id': calc.calc_id},
                        {'$set': {'metadata': calc.to_dict()}})

                Calc._get_collection().bulk_write([create_update(calc) for calc in calcs])
620

621
622
623
624
625
            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(upload_with_metadata)
626
627
628
629

            with utils.timer(
                    logger, 'index updated', step='index',
                    upload_size=self.upload_files.size):
630
                search.publish(calcs)
631

632
633
634
635
636
637
            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
638
639
                    self.publish_time = datetime.utcnow()
                    self.last_update = datetime.utcnow()
640
641
                    self.save()
            else:
642
                self.last_update = datetime.utcnow()
643
                self.save()
644

645
646
647
    @process
    def re_process_upload(self):
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
648
649
650
        A *process* that performs the re-processing of a earlier processed
        upload.

651
652
653
654
655
656
657
658
659
        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.
        """
        assert self.published

660
661
662
        logger = self.get_logger()
        logger.info('started to re-process')

663
664
665
666
667
668
        # 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')
        public_upload_files = cast(PublicUploadFiles, self.upload_files)
669
        staging_upload_files = public_upload_files.to_staging_upload_files(create=True)
670
671

        self._continue_with('parse_all')
672
        try:
673
674
675
676
            # we use a copy of the mongo queryset; reasons are cursor timeouts and
            # changing results on modifying the calc entries
            calcs = list(Calc.objects(upload_id=self.upload_id))
            for calc in calcs:
677
678
679
680
681
682
683
684
                if calc.process_running:
                    if calc.current_process == 're_process_calc':
                        logger.warn('re_process_calc is already running', calc_id=calc.calc_id)
                    else:
                        logger.warn('a process is already running on calc', calc_id=calc.calc_id)

                    continue

685
                calc.reset(worker_hostname=self.worker_hostname)
686

687
                parser = match_parser(calc.mainfile, staging_upload_files, strict=False)
688
                if parser is None:
689
690
                    logger.error(
                        'no parser matches during re-process, use the old parser',
691
                        calc_id=calc.calc_id)
692
693
694
695
                elif calc.parser != parser.name:
                    calc.parser = parser.name
                    logger.info(
                        'different parser matches during re-process, use new parser',
696
                        calc_id=calc.calc_id, parser=parser.name)
697
698
699
                calc.re_process_calc()
        except Exception as e:
            # try to remove the staging copy in failure case
700
            if staging_upload_files is not None and staging_upload_files.exists():
701
702
703
                staging_upload_files.delete()

            raise e
704
705
706
707

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

Markus Scheidgen's avatar
Markus Scheidgen committed
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
    @process
    def re_pack(self):
        """ A *process* that repacks the raw and archive data based on the current embargo data. """
        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')

        self.upload_files.re_pack(self.to_upload_with_metadata())
        self.joined = True
        self._complete()

Markus Scheidgen's avatar
Markus Scheidgen committed
723
    @process
724
    def process_upload(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
725
        """ A *process* that performs the initial upload processing. """
Markus Scheidgen's avatar
Markus Scheidgen committed
726
727
728
729
730
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
731
        """ A no-op *task* as a stand-in for receiving upload data. """
Markus Scheidgen's avatar
Markus Scheidgen committed
732
733
        pass

734
    @property
735
736
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
737
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
738
739
740

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
741
                self.upload_id, is_authorized=lambda: True, **kwargs)
742

743
        return self._upload_files
744

745
746
747
748
749
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
750
751
    @task
    def extracting(self):
752
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
753
        The *task* performed before the actual parsing/normalizing. Extracting and bagging
754
755
756
        the uploaded files, computing all keys, create an *upload* entry in the NOMAD-coe
        repository db, etc.
        """
757
758
759
760
761
        # 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
762
763
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
764
765
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
766
767
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
768
769
770
771

            if self.temporary:
                os.remove(self.upload_path)
                self.upload_path = None
772

773
        except KeyError:
774
            self.fail('processing requested for non existing upload', log_level=logging.ERROR)
775
776
777
            return
        except ExtractError:
            self.fail('bad .zip/.tar file', log_level=logging.INFO)
Markus Scheidgen's avatar
Markus Scheidgen committed
778
779
            return

780
781
782
783
784
    def _preprocess_files(self, path):
        """
        Some files need preprocessing. Currently we need to add a stripped POTCAR version
        and always restrict/embargo the original.
        """
785
        if os.path.basename(path).startswith('POTCAR'):
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
            # 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))

809
    def match_mainfiles(self) -> Generator[Tuple[str, object], None, None]:
810
811
812
813
814
815
816
        """
        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
        """
817
        directories_with_match: Dict[str, str] = dict()
818
819
        upload_files = self.staging_upload_files
        for filename in upload_files.raw_file_manifest():
820
            self._preprocess_files(filename)
821
            try:
822
                parser = match_parser(filename, upload_files)
823
                if parser is not None:
824
825
                    directory = os.path.dirname(filename)
                    if directory in directories_with_match:
826
827
828
                        # TODO this might give us the chance to store directory based relationship
                        # between calcs for the future?
                        pass
829
830
                    else:
                        directories_with_match[directory] = filename
831
832

                    yield filename, parser
833
834
835
836
            except Exception as e:
                self.get_logger().error(
                    'exception while matching pot. mainfile',
                    mainfile=filename, exc_info=e)
837

Markus Scheidgen's avatar
Markus Scheidgen committed
838
839
    @task
    def parse_all(self):
840
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
841
        The *task* used to identify mainfile/parser combinations among the upload's files, creates
842
843
        respective :class:`Calc` instances, and triggers their processing.
        """
844
845
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
846
847
        with utils.timer(
                logger, 'upload extracted', step='matching',
848
849
                upload_size=self.upload_files.size):
            for filename, parser in self.match_mainfiles():
850
                calc = Calc.create(
851
                    calc_id=self.upload_files.calc_id(filename),
852
                    mainfile=filename, parser=parser.name,
853
                    worker_hostname=self.worker_hostname,
854
855
                    upload_id=self.upload_id)

856
                calc.process_calc()
Markus Scheidgen's avatar
Markus Scheidgen committed
857

858
    def on_process_complete(self, process_name):
859
        if process_name == 'process_upload' or process_name == 're_process_upload':
860
861
862
            self.check_join()

    def check_join(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
863
864
865
866
867
868
869
870
        """
        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).
        """
871
872
873
874
        total_calcs = self.total_calcs
        processed_calcs = self.processed_calcs

        self.get_logger().debug('check join', processed_calcs=processed_calcs, total_calcs=total_calcs)
875
876
        # 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
877
        if not self.process_running and processed_calcs >= total_calcs:
878
            # this can easily be called multiple times, e.g. upload finished after all calcs finished
879
880
881
            modified_upload = self._get_collection().find_one_and_update(
                {'_id': self.upload_id, 'joined': {'$ne': True}},
                {'$set': {'joined': True}})
882
883
884
885
886
887
            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
888

889
890
891
    def reset(self):
        self.joined = False
        super().reset()
Markus Scheidgen's avatar
Markus Scheidgen committed
892

893
894
895
896
897
    @property
    def gui_url(self):
        base = config.api_url()[:-3]
        if base.endswith('/'):
            base = base[:-1]
898
        return '%s/gui/uploads/' % base
899

900
    def _cleanup_after_processing(self):
901
902
903
904
905
906
        # send email about process finish
        user = self.uploader
        name = '%s %s' % (user.first_name, user.last_name)
        message = '\n'.join([
            'Dear %s,' % name,
            '',
907
            'your data %suploaded at %s has completed processing.' % (
908
                '"%s" ' % self.name if self.name else '', self.upload_time.isoformat()),  # pylint: disable=no-member
909
910
911
912
913
            'You can review your data on your upload page: %s' % self.gui_url,
            '',
            'If you encouter any issues with your upload, please let us know and replay to this email.',
            '',
            'The nomad team'
914
        ])
915
916
917
918
919
920
921
        try:
            infrastructure.send_mail(
                name=name, email=user.email, message=message, subject='Processing completed')
        except Exception as e:
            # probably due to email configuration problems
            # don't fail or present this error to clients
            self.logger.error('could not send after processing email', exc_info=e)
Markus Scheidgen's avatar
Markus Scheidgen committed
922

923
924
925
926
927
928
929
930
931
932
    def _cleanup_after_re_processing(self):
        logger = self.get_logger()
        logger.info('started to repack re-processed upload')

        staging_upload_files = self.upload_files.to_staging_upload_files()

        with utils.timer(
                logger, 'reprocessed staged upload packed', step='delete staged',
                upload_size=self.upload_files.size):

933
            staging_upload_files.pack(self.to_upload_with_metadata(), skip_raw=True)
934
935
936
937
938
939

        with utils.timer(
                logger, 'reprocessed staged upload deleted', step='delete staged',
                upload_size=self.upload_files.size):

            staging_upload_files.delete()
940
            self.last_update = datetime.utcnow()
941
942
943
944
            self.save()

    @task
    def cleanup(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
945
946
947
948
        """
        The *task* that "cleans" the processing, i.e. removed obsolete files and performs
        pending archival operations. Depends on the type of processing.
        """
949
950
951
952
953
954
955
        search.refresh()

        if self.current_process == 're_process_upload':
            self._cleanup_after_re_processing()
        else:
            self._cleanup_after_processing()

956
    def get_calc(self, calc_id) -> Calc:
Markus Scheidgen's avatar
Markus Scheidgen committed
957
        """ Returns the upload calc with the given id or ``None``. """
958
959
        return Calc.objects(upload_id=self.upload_id, calc_id=calc_id).first()

Markus Scheidgen's avatar
Markus Scheidgen committed
960
    @property
961
    def processed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
962
963
964
965
        """
        The number of successfully or not successfully processed calculations. I.e.
        calculations that have finished processing.
        """
966
        return Calc.objects(upload_id=self.upload_id, tasks_status__in=[SUCCESS, FAILURE]).count()
967
968
969

    @property
    def total_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
970
        """ The number of all calculations. """
971
972
973
974
        return Calc.objects(upload_id=self.upload_id).count()

    @property
    def failed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
975
        """ The number of calculations with failed processing. """
976
        return Calc.objects(upload_id=self.upload_id, tasks_status=FAILURE).count()
977

978
    @property
Markus Scheidgen's avatar
Markus Scheidgen committed
979
980
    def pending_calcs(self) -> int:
        """ The number of calculations with pending processing. """
981
        return Calc.objects(upload_id=self.upload_id, tasks_status=PENDING).count()
982

983
    def all_calcs(self, start, end, order_by=None):
Markus Scheidgen's avatar
Markus Scheidgen committed
984
985
986
987
988
989
990
991
        """
        Returns all calculations, paginated and ordered.

        Arguments:
            start: the start index of the requested page
            end: the end index of the requested page
            order_by: the property to order by
        """
992
993
        query = Calc.objects(upload_id=self.upload_id)[start:end]
        return query.order_by(order_by) if order_by is not None else query
994

995
996
    @property
    def outdated_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
997
        """ All successfully processed and outdated calculations. """
998
999
1000
1001
        return Calc