data.py 37.8 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
22
23
24
25
26
# 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
    :members:
.. autoclass:: Upload
    :members:
"""

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
76
            'upload_id',
            ('upload_id', 'mainfile'),
            ('upload_id', 'parser'),
            ('upload_id', 'tasks_status'),
            ('upload_id', 'process_status')
Markus Scheidgen's avatar
Markus Scheidgen committed
77
78
79
80
81
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
82
        self._parser_backend: LocalBackend = None
83
84
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
85
        self._calc_proc_logwriter = None
86
        self._calc_proc_logwriter_ctx: ContextManager = None
Markus Scheidgen's avatar
Markus Scheidgen committed
87
88
89

    @classmethod
    def get(cls, id):
90
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
91

Markus Scheidgen's avatar
Markus Scheidgen committed
92
    @property
93
94
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
95

96
97
98
99
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
100
            self._upload.worker_hostname = self.worker_hostname
101
102
        return self._upload

103
104
105
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
106
107
            self._upload_files = ArchiveBasedStagingUploadFiles(
                self.upload_id, is_authorized=lambda: True, upload_path=self.upload.upload_path)
108
109
        return self._upload_files

Markus Scheidgen's avatar
Markus Scheidgen committed
110
    def get_logger(self, **kwargs):
111
112
113
114
        """
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
        """
115
116
117
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile, calc_id=self.calc_id, **kwargs)
118

119
        if self._calc_proc_logwriter_ctx is None:
120
121
122
123
124
125
126
127
128
129
130
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', '')
                    entry = '[%s] %s: %s' % (method_name, program, event)
                    if len(entry) > 120:
                        self._calc_proc_logwriter.write(entry[:120])
                        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])
145

146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
    @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:
            self.metadata['nomad_version'] = config.version
            self.metadata['nomad_commit'] = config.commit
            self.metadata['last_processing'] = datetime.now()

            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
172
    @process
173
    def process_calc(self):
174
175
176
177
        """
        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.
        """
178
        logger = self.get_logger()
179
        if self.upload is None:
180
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
181
182

        try:
183
184
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
185
            calc_with_metadata = datamodel.CalcWithMetadata(
186
187
188
189
190
191
                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()
192
            calc_with_metadata.upload_time = self.upload.upload_time
193
            calc_with_metadata.nomad_version = config.version
194
            calc_with_metadata.nomad_commit = config.commit
195
196
197
198
            calc_with_metadata.last_processing = datetime.now()
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
            self.metadata = calc_with_metadata.to_dict()

199
200
201
202
203
            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
204
205
206
207
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
208
            # close loghandler that was not closed due to failures
209
            try:
210
211
212
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
213
214
215
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

216
217
218
219
    def fail(self, *errors, log_level=logging.ERROR, **kwargs):
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
220
            calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
221
222
223
224
225
226
227
228
            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
229
230
231
232
233
234
235

            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

236
237
238
239
240
241
242
243
            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)

244
245
246
247
    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
248
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
249
250
            self.upload.reload()
            self.upload.check_join()
Markus Scheidgen's avatar
Markus Scheidgen committed
251
252
253

    @task
    def parsing(self):
254
        context = dict(parser=self.parser, step=self.parser)
255
        logger = self.get_logger(**context)
256
        parser = parser_dict[self.parser]
257
        self.metadata['parser_name'] = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
258

259
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
260
261
262
263
264
265
266
            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)
267
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
268

269
270
        # add the non code specific calc metadata to the backend
        # all other quantities have been determined by parsers/normalizers
271
        self._parser_backend.openNonOverlappingSection('section_entry_info')
272
        self._parser_backend.addValue('upload_id', self.upload_id)
273
        self._parser_backend.addValue('calc_id', self.calc_id)
274
        self._parser_backend.addValue('calc_hash', self.metadata['calc_hash'])
275
        self._parser_backend.addValue('mainfile', self.mainfile)
276
        self._parser_backend.addValue('parser_name', self.parser)
277
278
279
280
281
282
283
284
285
286
287
288
        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)
289

Markus Scheidgen's avatar
Markus Scheidgen committed
290
        if self._parser_backend.status[0] != 'ParseSuccess':
291
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
292
            error = self._parser_backend.status[1]
293
            self.fail(error, level=logging.INFO, **context)
294
295
296
297
298
299
300
301

    @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:
302
        self._parser_backend.openContext('/section_entry_info/0')
303
304
305
306
307
308
309
        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')
310
                self._parser_backend.addValue('number_of_archive_processor_warnings', len(warnings))
311
312
313
314
                self._parser_backend.addArrayValues('archive_processor_warnings', [str(warning) for warning in warnings])
            else:
                self._parser_backend.addValue('archive_processor_status', 'Success')
        else:
315
            errors = self._parser_backend.status[1]
316
317
318
            self._parser_backend.addValue('archive_processor_error', str(errors))

        self._parser_backend.closeNonOverlappingSection('section_archive_processing_info')
319
        self._parser_backend.closeContext('/section_entry_info/0')
Markus Scheidgen's avatar
Markus Scheidgen committed
320
321
322
323

    @task
    def normalizing(self):
        for normalizer in normalizers:
324
325
326
            if normalizer.domain != config.domain:
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
327
            normalizer_name = normalizer.__name__
328
            context = dict(normalizer=normalizer_name, step=normalizer_name)
329
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
330
331

            with utils.timer(
332
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
333
                with self.use_parser_backend(normalizer_name) as backend:
334
335
336
337
338
339
340
                    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
341

342
343
            failed = self._parser_backend.status[0] != 'ParseSuccess'
            if failed:
344
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
345
                error = self._parser_backend.status[1]
346
                self.fail(error, level=logging.WARNING, error=error, **context)
347
348
349
350
                break
            else:
                logger.debug(
                    'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
351
352
353

    @task
    def archiving(self):
354
355
        logger = self.get_logger()

356
        calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
357
        calc_with_metadata.apply_domain_metadata(self._parser_backend)
358
        calc_with_metadata.processed = True
359

360
361
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
362
            self.metadata = calc_with_metadata.to_dict()
363
364

        # index in search
365
        with utils.timer(logger, 'indexed', step='index'):
366
            search.Entry.from_calc_with_metadata(calc_with_metadata).save()
367

368
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
369
        with utils.timer(
370
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
371
                input_size=self.mainfile_file.size) as log_data:
372
            with self.upload_files.archive_file(self.calc_id, 'wt') as out:
373
                self._parser_backend.write_json(out, pretty=True, root_sections=Domain.instance.root_sections)
374

375
            log_data.update(archive_size=self.upload_files.archive_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
376
377
378
379

        # close loghandler
        if self._calc_proc_logwriter is not None:
            with utils.timer(
380
                    logger, 'archived log', step='logs',
Markus Scheidgen's avatar
Markus Scheidgen committed
381
                    input_size=self.mainfile_file.size) as log_data:
382
                self._calc_proc_logwriter_ctx.__exit__(None, None, None)  # pylint: disable=E1101
383
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
384

385
                log_data.update(log_size=self.upload_files.archive_log_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
386

387
    def __str__(self):
388
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
389

390

391
class Upload(Proc):
Markus Scheidgen's avatar
Markus Scheidgen committed
392
393
394
395
396
397
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
398
399
        upload_path: the path were the uploaded files was stored
        temporary: True if the uploaded file should be removed after extraction
400
        metadata: optional user provided additional meta data
Markus Scheidgen's avatar
Markus Scheidgen committed
401
402
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
403
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
404
405
406
407
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
408
409
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
Markus Scheidgen's avatar
Markus Scheidgen committed
410
411
412

    name = StringField(default=None)
    upload_time = DateTimeField()
413
    user_id = StringField(required=True)
414
415
    published = BooleanField(default=False)
    publish_time = DateTimeField()
416
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
417
418
419

    meta: Any = {
        'indexes': [
420
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
Markus Scheidgen's avatar
Markus Scheidgen committed
421
422
423
424
425
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
426
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
427

428
429
    @property
    def metadata(self) -> dict:
430
431
432
433
434
435
        # TODO user_metadata needs to be stored in the public bucket, since staging data might not be shared
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
436
437
438

    @metadata.setter
    def metadata(self, data: dict) -> None:
439
440
441
        # TODO user_metadata needs to be stored in the public bucket, since staging data might not be shared
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
442

Markus Scheidgen's avatar
Markus Scheidgen committed
443
    @classmethod
444
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
445
        upload = cls.get_by_id(id, 'upload_id')
446
        if upload is not None:
447
448
449
            return upload

        raise KeyError()
Markus Scheidgen's avatar
Markus Scheidgen committed
450
451

    @classmethod
452
453
454
    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
455

456
457
458
459
    @property
    def uploader(self):
        return coe_repo.User.from_user_id(self.user_id)

Markus Scheidgen's avatar
Markus Scheidgen committed
460
461
    def get_logger(self, **kwargs):
        logger = super().get_logger()
462
        logger = logger.bind(upload_id=self.upload_id, upload_name=self.name, **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
463
464
465
466
467
468
469
470
        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.
471
472

        Arguments:
473
            user (coe_repo.User): The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
474
        """
475
        user: coe_repo.User = kwargs['user']
476
477
478
        del(kwargs['user'])
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
479
        kwargs.update(user_id=str(user.user_id))
Markus Scheidgen's avatar
Markus Scheidgen committed
480
        self = super().create(**kwargs)
481

Markus Scheidgen's avatar
Markus Scheidgen committed
482
        self._continue_with('uploading')
483

Markus Scheidgen's avatar
Markus Scheidgen committed
484
485
        return self

486
487
488
489
490
    def delete(self):
        """ Deletes this upload process state entry and its calcs. """
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

491
    def delete_upload_local(self, with_coe_repo: bool = False):
492
493
        """
        Deletes of the upload, including its processing state and
494
        staging files. Local version without celery processing.
495
496
497
498
        """
        logger = self.get_logger()

        with utils.lnr(logger, 'staged upload delete failed'):
499

500
501
502
503
504
            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)
505

506
            with utils.timer(
507
                    logger, 'upload deleted from index', step='index',
508
                    upload_size=self.upload_files.size):
509
                search.delete_upload(self.upload_id)
510

511
            with utils.timer(
512
                    logger, 'staged upload deleted', step='files',
513
514
515
516
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
                self.delete()

517
518
519
520
521
522
523
524
    @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)

525
        return True  # do not save the process status on the delete upload
526

527
    @process
528
    def publish_upload(self):
529
530
531
532
533
        """
        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.
534
535
536

        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.
537
        """
538
539
        assert self.processed_calcs > 0

540
        logger = self.get_logger()
541
        logger.info('started to publish')
542

543
544
        with utils.lnr(logger, '(re-)publish failed'):
            upload_with_metadata = self.to_upload_with_metadata(self.metadata)
545
            calcs = upload_with_metadata.calcs
546

547
            if config.repository_db.publish_enabled:
548
549
550
551
552
553
554
                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()

555
556
557
558
559
560
561
                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)

562
                with utils.timer(
563
                        logger, 'upload PIDs read from repository', step='repo',
564
                        upload_size=self.upload_files.size):
565
566
                    for calc, coe_calc in zip(calcs, coe_upload.calcs):
                        calc.pid = coe_calc.coe_calc_id
567

568
            with utils.timer(
569
                    logger, 'upload metadata updated', step='metadata',
570
                    upload_size=self.upload_files.size):
571
572

                def create_update(calc):
573
                    calc.published = True
574
575
576
577
578
                    return UpdateOne(
                        {'_id': calc.calc_id},
                        {'$set': {'metadata': calc.to_dict()}})

                Calc._get_collection().bulk_write([create_update(calc) for calc in calcs])
579

580
581
582
583
584
            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)
585
586
587
588

            with utils.timer(
                    logger, 'index updated', step='index',
                    upload_size=self.upload_files.size):
589
                search.publish(calcs)
590

591
592
593
594
595
596
597
598
599
600
601
            if isinstance(self.upload_files, StagingUploadFiles):
                with utils.timer(
                        logger, 'staged upload deleted', step='delete staged',
                        upload_size=self.upload_files.size):
                    self.upload_files.delete()
                    self.published = True
                    self.publish_time = datetime.now()
                    self.last_update = datetime.now()
                    self.save()
            else:
                self.last_update = datetime.now()
602
                self.save()
603

604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
    @process
    def re_process_upload(self):
        """
        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

        self.reset()
        # 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)
        public_upload_files.to_staging_upload_files(create=True)

        self._continue_with('parse_all')
        for calc in self.calcs:
            calc.reset()
            calc.re_process_calc()

        # 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
633
    @process
634
    def process_upload(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
635
636
637
638
639
640
641
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
        pass

642
    @property
643
644
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
645
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
646
647
648

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
649
                self.upload_id, is_authorized=lambda: True, **kwargs)
650

651
        return self._upload_files
652

653
654
655
656
657
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
658
659
    @task
    def extracting(self):
660
661
662
663
664
        """
        Task performed before the actual parsing/normalizing. Extracting and bagging
        the uploaded files, computing all keys, create an *upload* entry in the NOMAD-coe
        repository db, etc.
        """
665
666
667
668
669
        # 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
670
671
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
672
673
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
674
675
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
676
677
678
679

            if self.temporary:
                os.remove(self.upload_path)
                self.upload_path = None
680

681
        except KeyError:
682
            self.fail('processing requested for non existing upload', log_level=logging.ERROR)
683
684
685
            return
        except ExtractError:
            self.fail('bad .zip/.tar file', log_level=logging.INFO)
Markus Scheidgen's avatar
Markus Scheidgen committed
686
687
            return

688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
    def _preprocess_files(self, path):
        """
        Some files need preprocessing. Currently we need to add a stripped POTCAR version
        and always restrict/embargo the original.
        """
        if os.path.basename(path) == 'POTCAR':
            # 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))

717
    def match_mainfiles(self) -> Generator[Tuple[str, object], None, None]:
718
719
720
721
722
723
724
        """
        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
        """
725
        directories_with_match: Dict[str, str] = dict()
726
727
        upload_files = self.staging_upload_files
        for filename in upload_files.raw_file_manifest():
728
            self._preprocess_files(filename)
729
            try:
730
                parser = match_parser(filename, upload_files)
731
                if parser is not None:
732
733
                    directory = os.path.dirname(filename)
                    if directory in directories_with_match:
734
735
736
                        # TODO this might give us the chance to store directory based relationship
                        # between calcs for the future?
                        pass
737
738
                    else:
                        directories_with_match[directory] = filename
739
740

                    yield filename, parser
741
742
743
744
            except Exception as e:
                self.get_logger().error(
                    'exception while matching pot. mainfile',
                    mainfile=filename, exc_info=e)
745

Markus Scheidgen's avatar
Markus Scheidgen committed
746
747
    @task
    def parse_all(self):
748
        """
749
        Identified mainfile/parser combinations among the upload's files, creates
750
751
        respective :class:`Calc` instances, and triggers their processing.
        """
752
753
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
754
755
        with utils.timer(
                logger, 'upload extracted', step='matching',
756
757
                upload_size=self.upload_files.size):
            for filename, parser in self.match_mainfiles():
758
                calc = Calc.create(
759
                    calc_id=self.upload_files.calc_id(filename),
760
                    mainfile=filename, parser=parser.name,
761
                    worker_hostname=self.worker_hostname,
762
763
                    upload_id=self.upload_id)

764
                calc.process_calc()
Markus Scheidgen's avatar
Markus Scheidgen committed
765

766
    def on_process_complete(self, process_name):
767
        if process_name == 'process_upload' or process_name == 're_process_upload':
768
769
770
771
772
773
774
775
776
777
            self.check_join()

    def check_join(self):
        total_calcs = self.total_calcs
        processed_calcs = self.processed_calcs

        self.get_logger().debug('check join', processed_calcs=processed_calcs, total_calcs=total_calcs)
        if not self.process_running and processed_calcs >= total_calcs:
            self.get_logger().debug('join')
            self.join()
778
779
780

    def join(self):
        self.cleanup()
Markus Scheidgen's avatar
Markus Scheidgen committed
781

782
783
784
785
786
787
788
    @property
    def gui_url(self):
        base = config.api_url()[:-3]
        if base.endswith('/'):
            base = base[:-1]
        return '%s/uploads/' % base

789
    def _cleanup_after_processing(self):
790
791
792
793
794
795
        # send email about process finish
        user = self.uploader
        name = '%s %s' % (user.first_name, user.last_name)
        message = '\n'.join([
            'Dear %s,' % name,
            '',
796
            'your data %suploaded at %s has completed processing.' % (
797
                '"%s" ' % self.name if self.name else '', self.upload_time.isoformat()),  # pylint: disable=no-member
798
799
800
801
802
            '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'
803
        ])
804
805
806
807
808
809
810
        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
811

812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
    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):

            staging_upload_files.pack(self.to_upload_with_metadata())

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

            staging_upload_files.delete()
            self.last_update = datetime.now()
            self.save()

    @task
    def cleanup(self):
        search.refresh()

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

841
842
843
    def get_calc(self, calc_id) -> Calc:
        return Calc.objects(upload_id=self.upload_id, calc_id=calc_id).first()

Markus Scheidgen's avatar
Markus Scheidgen committed
844
    @property
845
    def processed_calcs(self):
846
        return Calc.objects(upload_id=self.upload_id, tasks_status__in=[SUCCESS, FAILURE]).count()
847
848
849
850
851
852
853

    @property
    def total_calcs(self):
        return Calc.objects(upload_id=self.upload_id).count()

    @property
    def failed_calcs(self):
854
        return Calc.objects(upload_id=self.upload_id, tasks_status=FAILURE).count()
855

856
857
    @property
    def pending_calcs(self):
858
        return Calc.objects(upload_id=self.upload_id, tasks_status=PENDING).count()
859

860
861
862
    def all_calcs(self, start, end, order_by=None):
        query = Calc.objects(upload_id=self.upload_id)[start:end]
        return query.order_by(order_by) if order_by is not None else query
863
864
865

    @property
    def calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
866
        return Calc.objects(upload_id=self.upload_id, tasks_status=SUCCESS)
Markus Scheidgen's avatar
Markus Scheidgen committed
867

868
    def to_upload_with_metadata(self, user_metadata: dict = None) -> UploadWithMetadata:
869
        # prepare user metadata per upload and per calc
870
871
872
        if user_metadata is not None:
            calc_metadatas: Dict[str, Any] = dict()
            upload_metadata: Dict[str, Any] = dict()
873

874
            upload_metadata.update(user_metadata)
875
876
            if 'calculations' in upload_metadata:
                del(upload_metadata['calculations'])
Markus Scheidgen's avatar
Markus Scheidgen committed
877

878
            for calc in user_metadata.get('calculations', []):  # pylint: disable=no-member
879
                calc_metadatas[calc['mainfile']] = calc
Markus Scheidgen's avatar
Markus Scheidgen committed
880

881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
            user_upload_time = upload_metadata.get('_upload_time', None)

            def get_metadata(calc: Calc):
                """
                Assemble metadata from calc's processed calc metadata and the uploads
                user metadata.
                """
                calc_data = calc.metadata
                calc_with_metadata = datamodel.CalcWithMetadata(**calc_data)
                calc_metadata = dict(upload_metadata)
                calc_metadata.update(calc_metadatas.get(calc.mainfile, {}))
                calc_with_metadata.apply_user_metadata(calc_metadata)

                return calc_with_metadata
        else:
            user_upload_time = None

            def get_metadata(calc: Calc):
                return datamodel.CalcWithMetadata(**calc.metadata)

Markus Scheidgen's avatar
Markus Scheidgen committed
901
902
903
904
905
        result = UploadWithMetadata(
            upload_id=self.upload_id,
            uploader=utils.POPO(id=int(self.user_id)),
            upload_time=self.upload_time if user_upload_time is None else user_upload_time)

906
        result.calcs = [get_metadata(calc) for calc in Calc.objects(upload_id=self.upload_id)]
907

Markus Scheidgen's avatar
Markus Scheidgen committed
908
        return result
909

910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
    def compress_and_set_metadata(self, metadata: Dict[str, Any]) -> None:
        """
        Stores the given user metadata in the upload document. This is the metadata
        adhering to the API model (``UploadMetaData``). Most quantities can be stored
        for the upload and for each calculation. This method will try to move same values
        from the calculation to the upload to "compress" the data.
        """
        compressed = {
            key: value for key, value in metadata.items() if key != 'calculations'}
        calculations: List[Dict[str, Any]] = []
        compressed['calculations'] = calculations

        for calc in metadata.get('calculations', []):
            compressed_calc: Dict[str, Any] = {}
            calculations.append(compressed_calc)
            for key, value in calc.items():
                if key in ['_pid', 'mainfile']:
                    # these quantities are explicitly calc specific and have to stay with
                    # the calc
                    compressed_calc[key] = value
                else:
                    if key not in compressed:
                        compressed[key] = value
                    elif compressed[key].__repr__ != value.__repr__:
                        compressed_calc[key] = value
                    else:
                        compressed[key] = value

        self.metadata = compressed

940
    def __str__(self):
941
        return 'upload %s upload_id%s' % (super().__str__(), self.upload_id)