data.py 43 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, 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
    """
    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.

52
53
    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
54
55

    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
            ('upload_id', 'process_status'),
77
78
79
            ('upload_id', 'metadata.nomad_version'),
            'metadata.published',
            'metadata.datasets'
Markus Scheidgen's avatar
Markus Scheidgen committed
80
            'metadata.pid'
Markus Scheidgen's avatar
Markus Scheidgen committed
81
82
83
84
85
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
86
        self._parser_backend: LocalBackend = None
87
88
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
89
        self._calc_proc_logwriter = None
90
        self._calc_proc_logwriter_ctx: ContextManager = None
Markus Scheidgen's avatar
Markus Scheidgen committed
91

92
93
94
95
96
97
98
99
100
101
    @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
102
103
    @classmethod
    def get(cls, id):
104
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
105

Markus Scheidgen's avatar
Markus Scheidgen committed
106
    @property
107
108
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
109

110
111
112
113
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
114
            self._upload.worker_hostname = self.worker_hostname
115
116
        return self._upload

117
118
119
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
120
121
            self._upload_files = ArchiveBasedStagingUploadFiles(
                self.upload_id, is_authorized=lambda: True, upload_path=self.upload.upload_path)
122
123
        return self._upload_files

Markus Scheidgen's avatar
Markus Scheidgen committed
124
    def get_logger(self, **kwargs):
125
126
127
128
        """
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
        """
129
130
131
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile, calc_id=self.calc_id, **kwargs)
132

133
        if self._calc_proc_logwriter_ctx is None:
134
135
136
137
138
139
140
141
142
143
144
145
146
147
            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', '')
148
149
150
151

                    entry = '[%s] %s, %s: %s' % (method_name, datetime.utcnow().isoformat(), program, event)
                    if len(entry) > 140:
                        self._calc_proc_logwriter.write(entry[:140])
152
153
154
155
156
157
158
159
                        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])
160

161
162
163
164
165
166
167
168
169
170
    @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:
171
172
173
174
175
176
177
            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
178
            calc_with_metadata.last_processing = datetime.utcnow()
179
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
180
            self.metadata = calc_with_metadata.to_dict()
181
182
183
184
185
186
187
188
189
190
191
192
193

            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
194
    @process
195
    def process_calc(self):
196
197
198
199
        """
        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.
        """
200
        logger = self.get_logger()
201
        if self.upload is None:
202
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
203
204

        try:
205
206
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
207
            calc_with_metadata = datamodel.CalcWithMetadata(
208
209
210
211
212
                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
213
            calc_with_metadata.uploader = self.upload.user_id
214
            calc_with_metadata.upload_time = self.upload.upload_time
215
            calc_with_metadata.nomad_version = config.version
216
            calc_with_metadata.nomad_commit = config.commit
217
            calc_with_metadata.last_processing = datetime.utcnow()
218
219
220
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
            self.metadata = calc_with_metadata.to_dict()

221
222
223
224
225
            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
226
227
228
229
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
230
            # close loghandler that was not closed due to failures
231
            try:
232
233
234
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
235
236
237
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

238
239
240
241
    def fail(self, *errors, log_level=logging.ERROR, **kwargs):
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
242
            calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
243
244
245
246
247
248
249
250
            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
251
252
253
254
255
256
257

            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

258
259
260
261
262
263
264
265
            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)

266
267
268
269
    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
270
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
271
272
            self.upload.reload()
            self.upload.check_join()
Markus Scheidgen's avatar
Markus Scheidgen committed
273
274
275

    @task
    def parsing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
276
        """ The *task* that encapsulates all parsing related actions. """
277
        context = dict(parser=self.parser, step=self.parser)
278
        logger = self.get_logger(**context)
279
        parser = parser_dict[self.parser]
280
        self.metadata['parser_name'] = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
281

282
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
283
284
285
286
287
288
289
            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)
290
                return
291
292
293
294
295
            except SystemExit:
                self.fail(
                    'parser raised system exit', level=logging.ERROR,
                    error='system exit', **context)
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
296

297
298
        # add the non code specific calc metadata to the backend
        # all other quantities have been determined by parsers/normalizers
299
        self._parser_backend.openNonOverlappingSection('section_entry_info')
300
        self._parser_backend.addValue('upload_id', self.upload_id)
301
        self._parser_backend.addValue('calc_id', self.calc_id)
302
        self._parser_backend.addValue('calc_hash', self.metadata['calc_hash'])
303
        self._parser_backend.addValue('mainfile', self.mainfile)
304
        self._parser_backend.addValue('parser_name', self.parser)
305
306
307
308
309
310
311
312
313
314
315
316
        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)
317

Markus Scheidgen's avatar
Markus Scheidgen committed
318
        if self._parser_backend.status[0] != 'ParseSuccess':
319
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
320
            error = self._parser_backend.status[1]
321
            self.fail(error, level=logging.INFO, **context)
322
323
324
325
326
327
328
329

    @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:
330
        self._parser_backend.openContext('/section_entry_info/0')
331
332
333
334
335
336
337
        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')
338
                self._parser_backend.addValue('number_of_archive_processor_warnings', len(warnings))
339
340
341
342
                self._parser_backend.addArrayValues('archive_processor_warnings', [str(warning) for warning in warnings])
            else:
                self._parser_backend.addValue('archive_processor_status', 'Success')
        else:
343
            errors = self._parser_backend.status[1]
344
345
346
            self._parser_backend.addValue('archive_processor_error', str(errors))

        self._parser_backend.closeNonOverlappingSection('section_archive_processing_info')
347
        self._parser_backend.closeContext('/section_entry_info/0')
Markus Scheidgen's avatar
Markus Scheidgen committed
348
349
350

    @task
    def normalizing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
351
        """ The *task* that encapsulates all normalizing related actions. """
Markus Scheidgen's avatar
Markus Scheidgen committed
352
        for normalizer in normalizers:
353
354
355
            if normalizer.domain != config.domain:
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
356
            normalizer_name = normalizer.__name__
357
            context = dict(normalizer=normalizer_name, step=normalizer_name)
358
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
359
360

            with utils.timer(
361
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
362
                with self.use_parser_backend(normalizer_name) as backend:
363
364
365
366
367
368
369
                    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
370

371
372
            failed = self._parser_backend.status[0] != 'ParseSuccess'
            if failed:
373
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
374
                error = self._parser_backend.status[1]
375
                self.fail(error, level=logging.WARNING, error=error, **context)
376
377
378
379
                break
            else:
                logger.debug(
                    'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
380
381
382

    @task
    def archiving(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
383
        """ The *task* that encapsulates all archival related actions. """
384
385
        logger = self.get_logger()

386
        calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
387
        calc_with_metadata.apply_domain_metadata(self._parser_backend)
388
        calc_with_metadata.processed = True
389

390
391
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
392
            self.metadata = calc_with_metadata.to_dict()
393
394

        # index in search
395
        with utils.timer(logger, 'indexed', step='index'):
396
            search.Entry.from_calc_with_metadata(calc_with_metadata).save()
397

398
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
399
        with utils.timer(
400
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
401
                input_size=self.mainfile_file.size) as log_data:
402
            with self.upload_files.archive_file(self.calc_id, 'wt') as out:
403
                self._parser_backend.write_json(out, pretty=True, root_sections=Domain.instance.root_sections)
404

405
            log_data.update(archive_size=self.upload_files.archive_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
406
407
408
409

        # close loghandler
        if self._calc_proc_logwriter is not None:
            with utils.timer(
410
                    logger, 'archived log', step='logs',
Markus Scheidgen's avatar
Markus Scheidgen committed
411
                    input_size=self.mainfile_file.size) as log_data:
412
                self._calc_proc_logwriter_ctx.__exit__(None, None, None)  # pylint: disable=E1101
413
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
414

415
                log_data.update(log_size=self.upload_files.archive_log_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
416

417
    def __str__(self):
418
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
419

420

421
class Upload(Proc):
Markus Scheidgen's avatar
Markus Scheidgen committed
422
423
424
425
426
427
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
428
429
        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
430
431
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
432
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
433
434
        published: Boolean that indicates the publish status
        publish_time: Date when the upload was initially published
435
        last_update: Date of the last publishing/re-processing
Markus Scheidgen's avatar
Markus Scheidgen committed
436
        joined: Boolean indicates if the running processing has joined (:func:`check_join`)
Markus Scheidgen's avatar
Markus Scheidgen committed
437
438
439
440
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
441
442
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
Markus Scheidgen's avatar
Markus Scheidgen committed
443
444
445

    name = StringField(default=None)
    upload_time = DateTimeField()
446
    user_id = StringField(required=True)
447
448
    published = BooleanField(default=False)
    publish_time = DateTimeField()
449
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
450

451
452
    joined = BooleanField(default=False)

Markus Scheidgen's avatar
Markus Scheidgen committed
453
454
    meta: Any = {
        'indexes': [
455
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
Markus Scheidgen's avatar
Markus Scheidgen committed
456
457
458
459
460
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
461
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
462

463
464
    @property
    def metadata(self) -> dict:
Markus Scheidgen's avatar
Markus Scheidgen committed
465
466
467
468
469
470
471
        """
        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.
        """
472
473
474
475
476
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
477
478
479

    @metadata.setter
    def metadata(self, data: dict) -> None:
480
481
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
482

Markus Scheidgen's avatar
Markus Scheidgen committed
483
    @classmethod
484
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
485
        return cls.get_by_id(id, 'upload_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
486
487

    @classmethod
488
    def user_uploads(cls, user: datamodel.User, **kwargs) -> List['Upload']:
489
490
        """ 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
491

492
493
    @property
    def uploader(self):
494
        return datamodel.User.get(self.user_id)
495

Markus Scheidgen's avatar
Markus Scheidgen committed
496
497
    def get_logger(self, **kwargs):
        logger = super().get_logger()
498
499
500
501
502
        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
503
504
505
506
507
508
509
510
        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.
511
512

        Arguments:
513
            user: The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
514
        """
515
516
        # use kwargs to keep compatibility with super method
        user: datamodel.User = kwargs['user']
517
        del(kwargs['user'])
518

519
520
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
521
        kwargs.update(user_id=user.user_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
522
        self = super().create(**kwargs)
523

Markus Scheidgen's avatar
Markus Scheidgen committed
524
        self._continue_with('uploading')
525

Markus Scheidgen's avatar
Markus Scheidgen committed
526
527
        return self

528
529
530
531
532
    def delete(self):
        """ Deletes this upload process state entry and its calcs. """
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

533
    def delete_upload_local(self):
534
        """
535
        Deletes the upload, including its processing state and
536
        staging files. Local version without celery processing.
537
538
539
540
        """
        logger = self.get_logger()

        with utils.lnr(logger, 'staged upload delete failed'):
541
            with utils.timer(
542
                    logger, 'upload deleted from index', step='index',
543
                    upload_size=self.upload_files.size):
544
                search.delete_upload(self.upload_id)
545

546
            with utils.timer(
547
                    logger, 'staged upload deleted', step='files',
548
549
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
550
551

            self.delete()
552

553
    @process
554
    def delete_upload(self):
555
556
557
558
        """
        Deletes of the upload, including its processing state and
        staging files. This starts the celery process of deleting the upload.
        """
559
        self.delete_upload_local()
560

561
        return True  # do not save the process status on the delete upload
562

563
    @process
564
    def publish_upload(self):
565
        """
566
567
        Moves the upload out of staging to the public area. It will
        pack the staging upload files in to public upload files.
568
        """
569
570
        assert self.processed_calcs > 0

571
        logger = self.get_logger()
572
        logger.info('started to publish')
573

574
        with utils.lnr(logger, 'publish failed'):
575
            upload_with_metadata = self.to_upload_with_metadata(self.metadata)
576
            calcs = upload_with_metadata.calcs
577

578
            with utils.timer(
579
                    logger, 'upload metadata updated', step='metadata',
580
                    upload_size=self.upload_files.size):
581
582

                def create_update(calc):
583
                    calc.published = True
584
                    calc.with_embargo = calc.with_embargo if calc.with_embargo is not None else False
585
586
587
588
589
                    return UpdateOne(
                        {'_id': calc.calc_id},
                        {'$set': {'metadata': calc.to_dict()}})

                Calc._get_collection().bulk_write([create_update(calc) for calc in calcs])
590

591
592
593
594
595
            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)
596
597
598
599

            with utils.timer(
                    logger, 'index updated', step='index',
                    upload_size=self.upload_files.size):
600
                search.publish(calcs)
601

602
603
604
605
606
607
            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
608
609
                    self.publish_time = datetime.utcnow()
                    self.last_update = datetime.utcnow()
610
611
                    self.save()
            else:
612
                self.last_update = datetime.utcnow()
613
                self.save()
614

615
616
617
    @process
    def re_process_upload(self):
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
618
619
620
        A *process* that performs the re-processing of a earlier processed
        upload.

621
622
623
624
625
626
627
628
629
        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

630
631
632
        logger = self.get_logger()
        logger.info('started to re-process')

633
634
635
636
637
638
        # 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)
639
        staging_upload_files = public_upload_files.to_staging_upload_files(create=True)
640
641

        self._continue_with('parse_all')
642
        try:
643
644
645
646
            # 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:
647
648
649
650
651
652
653
654
                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

655
                calc.reset(worker_hostname=self.worker_hostname)
656

657
                parser = match_parser(calc.mainfile, staging_upload_files, strict=False)
658
                if parser is None:
659
660
                    logger.error(
                        'no parser matches during re-process, use the old parser',
661
                        calc_id=calc.calc_id)
662
663
664
665
                elif calc.parser != parser.name:
                    calc.parser = parser.name
                    logger.info(
                        'different parser matches during re-process, use new parser',
666
                        calc_id=calc.calc_id, parser=parser.name)
667
668
669
                calc.re_process_calc()
        except Exception as e:
            # try to remove the staging copy in failure case
670
            if staging_upload_files is not None and staging_upload_files.exists():
671
672
673
                staging_upload_files.delete()

            raise e
674
675
676
677

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

678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
    @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
693
    @process
694
    def process_upload(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
695
        """ A *process* that performs the initial upload processing. """
Markus Scheidgen's avatar
Markus Scheidgen committed
696
697
698
699
700
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
701
        """ A no-op *task* as a stand-in for receiving upload data. """
Markus Scheidgen's avatar
Markus Scheidgen committed
702
703
        pass

704
    @property
705
706
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
707
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
708
709
710

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
711
                self.upload_id, is_authorized=lambda: True, **kwargs)
712

713
        return self._upload_files
714

715
716
717
718
719
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
720
721
    @task
    def extracting(self):
722
        """
723
724
        The *task* performed before the actual parsing/normalizing: extracting
        the uploaded files.
725
        """
726
727
728
729
730
        # 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
731
732
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
733
734
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
735
736
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
737
738
739
740

            if self.temporary:
                os.remove(self.upload_path)
                self.upload_path = None
741

742
        except KeyError:
743
            self.fail('processing requested for non existing upload', log_level=logging.ERROR)
744
745
746
            return
        except ExtractError:
            self.fail('bad .zip/.tar file', log_level=logging.INFO)
Markus Scheidgen's avatar
Markus Scheidgen committed
747
748
            return

749
750
751
752
753
    def _preprocess_files(self, path):
        """
        Some files need preprocessing. Currently we need to add a stripped POTCAR version
        and always restrict/embargo the original.
        """
754
        if os.path.basename(path).startswith('POTCAR'):
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
            # 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))

778
    def match_mainfiles(self) -> Generator[Tuple[str, object], None, None]:
779
780
781
782
783
784
785
        """
        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
        """
786
        directories_with_match: Dict[str, str] = dict()
787
788
        upload_files = self.staging_upload_files
        for filename in upload_files.raw_file_manifest():
789
            self._preprocess_files(filename)
790
            try:
791
                parser = match_parser(filename, upload_files)
792
                if parser is not None:
793
794
                    directory = os.path.dirname(filename)
                    if directory in directories_with_match:
795
796
797
                        # TODO this might give us the chance to store directory based relationship
                        # between calcs for the future?
                        pass
798
799
                    else:
                        directories_with_match[directory] = filename
800
801

                    yield filename, parser
802
803
804
805
            except Exception as e:
                self.get_logger().error(
                    'exception while matching pot. mainfile',
                    mainfile=filename, exc_info=e)
806

Markus Scheidgen's avatar
Markus Scheidgen committed
807
808
    @task
    def parse_all(self):
809
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
810
        The *task* used to identify mainfile/parser combinations among the upload's files, creates
811
812
        respective :class:`Calc` instances, and triggers their processing.
        """
813
814
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
815
816
        with utils.timer(
                logger, 'upload extracted', step='matching',
817
818
                upload_size=self.upload_files.size):
            for filename, parser in self.match_mainfiles():
819
                calc = Calc.create(
820
                    calc_id=self.upload_files.calc_id(filename),
821
                    mainfile=filename, parser=parser.name,
822
                    worker_hostname=self.worker_hostname,
823
824
                    upload_id=self.upload_id)

825
                calc.process_calc()
Markus Scheidgen's avatar
Markus Scheidgen committed
826

827
    def on_process_complete(self, process_name):
828
        if process_name == 'process_upload' or process_name == 're_process_upload':
829
830
831
            self.check_join()

    def check_join(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
832
833
834
835
836
837
838
839
        """
        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).
        """
840
841
842
843
        total_calcs = self.total_calcs
        processed_calcs = self.processed_calcs

        self.get_logger().debug('check join', processed_calcs=processed_calcs, total_calcs=total_calcs)
844
845
        # 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
846
        if not self.process_running and processed_calcs >= total_calcs:
847
            # this can easily be called multiple times, e.g. upload finished after all calcs finished
848
849
850
            modified_upload = self._get_collection().find_one_and_update(
                {'_id': self.upload_id, 'joined': {'$ne': True}},
                {'$set': {'joined': True}})
851
852
853
854
855
856
            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
857

858
859
860
    def reset(self):
        self.joined = False
        super().reset()
Markus Scheidgen's avatar
Markus Scheidgen committed
861

862
    def _cleanup_after_processing(self):
863
864
865
866
867
868
        # send email about process finish
        user = self.uploader
        name = '%s %s' % (user.first_name, user.last_name)
        message = '\n'.join([
            'Dear %s,' % name,
            '',
869
            'your data %suploaded at %s has completed processing.' % (
870
                '"%s" ' % self.name if self.name else '', self.upload_time.isoformat()),  # pylint: disable=no-member
871
            'You can review your data on your upload page: %s' % config.gui_url(),
872
873
874
875
            '',
            'If you encouter any issues with your upload, please let us know and replay to this email.',
            '',
            'The nomad team'
876
        ])
877
878
879
880
881
882
883
        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
884

885
886
887
888
889
890
891
892
893
894
    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):

895
            staging_upload_files.pack(self.to_upload_with_metadata(), skip_raw=True)
896
897
898
899
900
901

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

            staging_upload_files.delete()
902
            self.last_update = datetime.utcnow()
903
904
905
906
            self.save()

    @task
    def cleanup(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
907
908
909
910
        """
        The *task* that "cleans" the processing, i.e. removed obsolete files and performs
        pending archival operations. Depends on the type of processing.
        """
911
912
913
914
915
916
917
        search.refresh()

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

918
    def get_calc(self, calc_id) -> Calc:
Markus Scheidgen's avatar
Markus Scheidgen committed
919
        """ Returns the upload calc with the given id or ``None``. """
920
921
        return Calc.objects(upload_id=self.upload_id, calc_id=calc_id).first()

Markus Scheidgen's avatar
Markus Scheidgen committed
922
    @property
923
    def processed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
924
925
926
927
        """
        The number of successfully or not successfully processed calculations. I.e.
        calculations that have finished processing.
        """
928
        return Calc.objects(upload_id=self.upload_id, tasks_status__in=[SUCCESS, FAILURE]).count()
929
930
931

    @property
    def total_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
932
        """ The number of all calculations. """
933
934
935
936
        return Calc.objects(upload_id=self.upload_id).count()

    @property
    def failed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
937
        """ The number of calculations with failed processing. """
938
        return Calc.objects(upload_id=self.upload_id, tasks_status=FAILURE).count()
939

940
    @property
Markus Scheidgen's avatar
Markus Scheidgen committed
941
942
    def pending_calcs(self) -> int:
        """ The number of calculations with pending processing. """
943
        return Calc.objects(upload_id=self.upload_id, tasks_status=PENDING).count()
944

945
    def all_calcs(self, start, end, order_by=None):
Markus Scheidgen's avatar
Markus Scheidgen committed
946
947
948
949
950
951
952
953
        """
        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
        """
954
955
        query = Calc.objects(upload_id=self.upload_id)[start:end]
        return query.order_by(order_by) if order_by is not None else query
956

957
958
    @property
    def outdated_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
959
        """ All successfully processed and outdated calculations. """
960
961
962
963
        return Calc.objects(
            upload_id=self.upload_id, tasks_status=SUCCESS,
            metadata__nomad_version__ne=config.version)

964
965
    @property
    def calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
966
        """ All successfully processed calculations. """
Markus Scheidgen's avatar
Markus Scheidgen committed
967
        return Calc.objects(upload_id=self.upload_id, tasks_status=SUCCESS)