data.py 44.1 KB
Newer Older
Markus Scheidgen's avatar
Markus Scheidgen committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# Copyright 2018 Markus Scheidgen
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an"AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""
This module comprises a set of persistent document classes that hold all user related
data. These are information about users, their uploads and datasets, the associated
calculations, and files


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

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

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

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

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


Markus Scheidgen's avatar
Markus Scheidgen committed
45
class Calc(Proc):
Markus Scheidgen's avatar
Markus Scheidgen committed
46
47
48
49
50
51
52
53
54
55
    """
    Instances of this class represent calculations. This class manages the elastic
    search index entry, files, and archive for the respective calculation.

    It also contains the calculations processing and its state.

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

    Attributes:
56
        calc_id: the calc_id of this calc
Markus Scheidgen's avatar
Markus Scheidgen committed
57
58
59
        parser: the name of the parser used to process this calc
        upload_id: the id of the upload used to create this calculation
        mainfile: the mainfile (including path in upload) that was used to create this calc
60

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

68
69
    metadata = DictField()

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

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
83
        self._parser_backend: LocalBackend = None
84
85
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
86
        self._calc_proc_logwriter = None
87
        self._calc_proc_logwriter_ctx: ContextManager = None
Markus Scheidgen's avatar
Markus Scheidgen committed
88
89
90

    @classmethod
    def get(cls, id):
91
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
92

Markus Scheidgen's avatar
Markus Scheidgen committed
93
    @property
94
95
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
96

97
98
99
100
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
101
            self._upload.worker_hostname = self.worker_hostname
102
103
        return self._upload

104
105
106
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
107
108
            self._upload_files = ArchiveBasedStagingUploadFiles(
                self.upload_id, is_authorized=lambda: True, upload_path=self.upload.upload_path)
109
110
        return self._upload_files

Markus Scheidgen's avatar
Markus Scheidgen committed
111
    def get_logger(self, **kwargs):
112
113
114
115
        """
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
        """
116
117
118
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile, calc_id=self.calc_id, **kwargs)
119

120
        if self._calc_proc_logwriter_ctx is None:
121
122
123
124
125
126
127
128
129
130
131
132
133
134
            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', '')
135
136
137
138

                    entry = '[%s] %s, %s: %s' % (method_name, datetime.utcnow().isoformat(), program, event)
                    if len(entry) > 140:
                        self._calc_proc_logwriter.write(entry[:140])
139
140
141
142
143
144
145
146
                        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])
147

148
149
150
151
152
153
154
155
156
157
    @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:
158
159
160
161
162
163
164
            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
165
            calc_with_metadata.last_processing = datetime.utcnow()
166
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
167
            self.metadata = calc_with_metadata.to_dict()
168
169
170
171
172
173
174
175
176
177
178
179
180

            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
181
    @process
182
    def process_calc(self):
183
184
185
186
        """
        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.
        """
187
        logger = self.get_logger()
188
        if self.upload is None:
189
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
190
191

        try:
192
193
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
194
            calc_with_metadata = datamodel.CalcWithMetadata(
195
196
197
198
199
200
                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()
201
            calc_with_metadata.upload_time = self.upload.upload_time
202
            calc_with_metadata.nomad_version = config.version
203
            calc_with_metadata.nomad_commit = config.commit
204
            calc_with_metadata.last_processing = datetime.utcnow()
205
206
207
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
            self.metadata = calc_with_metadata.to_dict()

208
209
210
211
212
            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
213
214
215
216
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
217
            # close loghandler that was not closed due to failures
218
            try:
219
220
221
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
222
223
224
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

225
226
227
228
    def fail(self, *errors, log_level=logging.ERROR, **kwargs):
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
229
            calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
230
231
232
233
234
235
236
237
            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
238
239
240
241
242
243
244

            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

245
246
247
248
249
250
251
252
            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)

253
254
255
256
    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
257
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
258
259
            self.upload.reload()
            self.upload.check_join()
Markus Scheidgen's avatar
Markus Scheidgen committed
260
261
262

    @task
    def parsing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
263
        """ The *task* that encapsulates all parsing related actions. """
264
        context = dict(parser=self.parser, step=self.parser)
265
        logger = self.get_logger(**context)
266
        parser = parser_dict[self.parser]
267
        self.metadata['parser_name'] = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
268

269
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
270
271
272
273
274
275
276
            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)
277
                return
278
279
280
281
282
            except SystemExit:
                self.fail(
                    'parser raised system exit', level=logging.ERROR,
                    error='system exit', **context)
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
283

284
285
        # add the non code specific calc metadata to the backend
        # all other quantities have been determined by parsers/normalizers
286
        self._parser_backend.openNonOverlappingSection('section_entry_info')
287
        self._parser_backend.addValue('upload_id', self.upload_id)
288
        self._parser_backend.addValue('calc_id', self.calc_id)
289
        self._parser_backend.addValue('calc_hash', self.metadata['calc_hash'])
290
        self._parser_backend.addValue('mainfile', self.mainfile)
291
        self._parser_backend.addValue('parser_name', self.parser)
292
293
294
295
296
297
298
299
300
301
302
303
        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)
304

Markus Scheidgen's avatar
Markus Scheidgen committed
305
        if self._parser_backend.status[0] != 'ParseSuccess':
306
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
307
            error = self._parser_backend.status[1]
308
            self.fail(error, level=logging.INFO, **context)
309
310
311
312
313
314
315
316

    @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:
317
        self._parser_backend.openContext('/section_entry_info/0')
318
319
320
321
322
323
324
        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')
325
                self._parser_backend.addValue('number_of_archive_processor_warnings', len(warnings))
326
327
328
329
                self._parser_backend.addArrayValues('archive_processor_warnings', [str(warning) for warning in warnings])
            else:
                self._parser_backend.addValue('archive_processor_status', 'Success')
        else:
330
            errors = self._parser_backend.status[1]
331
332
333
            self._parser_backend.addValue('archive_processor_error', str(errors))

        self._parser_backend.closeNonOverlappingSection('section_archive_processing_info')
334
        self._parser_backend.closeContext('/section_entry_info/0')
Markus Scheidgen's avatar
Markus Scheidgen committed
335
336
337

    @task
    def normalizing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
338
        """ The *task* that encapsulates all normalizing related actions. """
Markus Scheidgen's avatar
Markus Scheidgen committed
339
        for normalizer in normalizers:
340
341
342
            if normalizer.domain != config.domain:
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
343
            normalizer_name = normalizer.__name__
344
            context = dict(normalizer=normalizer_name, step=normalizer_name)
345
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
346
347

            with utils.timer(
348
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
349
                with self.use_parser_backend(normalizer_name) as backend:
350
351
352
353
354
355
356
                    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
357

358
359
            failed = self._parser_backend.status[0] != 'ParseSuccess'
            if failed:
360
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
361
                error = self._parser_backend.status[1]
362
                self.fail(error, level=logging.WARNING, error=error, **context)
363
364
365
366
                break
            else:
                logger.debug(
                    'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
367
368
369

    @task
    def archiving(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
370
        """ The *task* that encapsulates all archival related actions. """
371
372
        logger = self.get_logger()

373
        calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
374
        calc_with_metadata.apply_domain_metadata(self._parser_backend)
375
        calc_with_metadata.processed = True
376

377
378
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
379
            self.metadata = calc_with_metadata.to_dict()
380
381

        # index in search
382
        with utils.timer(logger, 'indexed', step='index'):
383
            search.Entry.from_calc_with_metadata(calc_with_metadata).save()
384

385
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
386
        with utils.timer(
387
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
388
                input_size=self.mainfile_file.size) as log_data:
389
            with self.upload_files.archive_file(self.calc_id, 'wt') as out:
390
                self._parser_backend.write_json(out, pretty=True, root_sections=Domain.instance.root_sections)
391

392
            log_data.update(archive_size=self.upload_files.archive_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
393
394
395
396

        # close loghandler
        if self._calc_proc_logwriter is not None:
            with utils.timer(
397
                    logger, 'archived log', step='logs',
Markus Scheidgen's avatar
Markus Scheidgen committed
398
                    input_size=self.mainfile_file.size) as log_data:
399
                self._calc_proc_logwriter_ctx.__exit__(None, None, None)  # pylint: disable=E1101
400
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
401

402
                log_data.update(log_size=self.upload_files.archive_log_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
403

404
    def __str__(self):
405
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
406

407

408
class Upload(Proc):
Markus Scheidgen's avatar
Markus Scheidgen committed
409
410
411
412
413
414
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
415
416
        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
417
418
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
419
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
420
421
422
423
        published: Boolean that indicates the publish status
        publish_time: Date when the upload was initially published
        last_update: Date of the last (re-)publishing
        joined: Boolean indicates if the running processing has joined (:func:`check_join`)
Markus Scheidgen's avatar
Markus Scheidgen committed
424
425
426
427
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
428
429
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
Markus Scheidgen's avatar
Markus Scheidgen committed
430
431
432

    name = StringField(default=None)
    upload_time = DateTimeField()
433
    user_id = StringField(required=True)
434
435
    published = BooleanField(default=False)
    publish_time = DateTimeField()
436
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
437

438
439
    joined = BooleanField(default=False)

Markus Scheidgen's avatar
Markus Scheidgen committed
440
441
    meta: Any = {
        'indexes': [
442
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
Markus Scheidgen's avatar
Markus Scheidgen committed
443
444
445
446
447
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
448
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
449

450
451
    @property
    def metadata(self) -> dict:
Markus Scheidgen's avatar
Markus Scheidgen committed
452
453
454
455
456
457
458
        """
        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.
        """
459
460
461
462
463
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
464
465
466

    @metadata.setter
    def metadata(self, data: dict) -> None:
467
468
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
469

Markus Scheidgen's avatar
Markus Scheidgen committed
470
    @classmethod
471
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
472
        upload = cls.get_by_id(id, 'upload_id')
473
        if upload is not None:
474
475
476
            return upload

        raise KeyError()
Markus Scheidgen's avatar
Markus Scheidgen committed
477
478

    @classmethod
479
480
481
    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
482

483
484
485
486
    @property
    def uploader(self):
        return coe_repo.User.from_user_id(self.user_id)

Markus Scheidgen's avatar
Markus Scheidgen committed
487
488
    def get_logger(self, **kwargs):
        logger = super().get_logger()
489
490
491
492
493
        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
494
495
496
497
498
499
500
501
        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.
502
503

        Arguments:
504
            user (coe_repo.User): The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
505
        """
506
        user: coe_repo.User = kwargs['user']
507
508
509
        del(kwargs['user'])
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
510
        kwargs.update(user_id=str(user.user_id))
Markus Scheidgen's avatar
Markus Scheidgen committed
511
        self = super().create(**kwargs)
512

Markus Scheidgen's avatar
Markus Scheidgen committed
513
        self._continue_with('uploading')
514

Markus Scheidgen's avatar
Markus Scheidgen committed
515
516
        return self

517
518
519
520
521
    def delete(self):
        """ Deletes this upload process state entry and its calcs. """
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

522
    def delete_upload_local(self, with_coe_repo: bool = False):
523
524
        """
        Deletes of the upload, including its processing state and
525
        staging files. Local version without celery processing.
526
527
528
529
        """
        logger = self.get_logger()

        with utils.lnr(logger, 'staged upload delete failed'):
530

531
532
533
534
535
            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)
536

537
            with utils.timer(
538
                    logger, 'upload deleted from index', step='index',
539
                    upload_size=self.upload_files.size):
540
                search.delete_upload(self.upload_id)
541

542
            with utils.timer(
543
                    logger, 'staged upload deleted', step='files',
544
545
546
547
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
                self.delete()

548
549
550
551
552
553
554
555
    @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)

556
        return True  # do not save the process status on the delete upload
557

558
    @process
559
    def publish_upload(self):
560
561
562
563
564
        """
        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.
565
566
567

        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.
568
        """
569
570
        assert self.processed_calcs > 0

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

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

578
            if config.repository_db.publish_enabled:
579
580
581
582
583
584
585
                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()

586
587
588
589
590
591
592
                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)

593
                with utils.timer(
594
                        logger, 'upload PIDs read from repository', step='repo',
595
                        upload_size=self.upload_files.size):
596
597
                    for calc, coe_calc in zip(calcs, coe_upload.calcs):
                        calc.pid = coe_calc.coe_calc_id
598

599
            with utils.timer(
600
                    logger, 'upload metadata updated', step='metadata',
601
                    upload_size=self.upload_files.size):
602
603

                def create_update(calc):
604
                    calc.published = True
605
606
607
608
609
                    return UpdateOne(
                        {'_id': calc.calc_id},
                        {'$set': {'metadata': calc.to_dict()}})

                Calc._get_collection().bulk_write([create_update(calc) for calc in calcs])
610

611
612
613
614
615
            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)
616
617
618
619

            with utils.timer(
                    logger, 'index updated', step='index',
                    upload_size=self.upload_files.size):
620
                search.publish(calcs)
621

622
623
624
625
626
627
            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
628
629
                    self.publish_time = datetime.utcnow()
                    self.last_update = datetime.utcnow()
630
631
                    self.save()
            else:
632
                self.last_update = datetime.utcnow()
633
                self.save()
634

635
636
637
    @process
    def re_process_upload(self):
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
638
639
640
        A *process* that performs the re-processing of a earlier processed
        upload.

641
642
643
644
645
646
647
648
649
        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

650
651
652
        logger = self.get_logger()
        logger.info('started to re-process')

653
654
655
656
657
658
        # 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)
659
        staging_upload_files = public_upload_files.to_staging_upload_files(create=True)
660
661

        self._continue_with('parse_all')
662
        try:
663
664
665
666
            # 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:
667
668
669
670
671
672
673
674
                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

675
                calc.reset(worker_hostname=self.worker_hostname)
676

677
                parser = match_parser(calc.mainfile, staging_upload_files, strict=False)
678
                if parser is None:
679
680
                    logger.error(
                        'no parser matches during re-process, use the old parser',
681
                        calc_id=calc.calc_id)
682
683
684
685
                elif calc.parser != parser.name:
                    calc.parser = parser.name
                    logger.info(
                        'different parser matches during re-process, use new parser',
686
                        calc_id=calc.calc_id, parser=parser.name)
687
688
689
                calc.re_process_calc()
        except Exception as e:
            # try to remove the staging copy in failure case
690
            if staging_upload_files is not None and staging_upload_files.exists():
691
692
693
                staging_upload_files.delete()

            raise e
694
695
696
697

        # 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
698
    @process
699
    def process_upload(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
700
        """ A *process* that performs the initial upload processing. """
Markus Scheidgen's avatar
Markus Scheidgen committed
701
702
703
704
705
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
706
        """ A no-op *task* as a stand-in for receiving upload data. """
Markus Scheidgen's avatar
Markus Scheidgen committed
707
708
        pass

709
    @property
710
711
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
712
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
713
714
715

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
716
                self.upload_id, is_authorized=lambda: True, **kwargs)
717

718
        return self._upload_files
719

720
721
722
723
724
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
725
726
    @task
    def extracting(self):
727
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
728
        The *task* performed before the actual parsing/normalizing. Extracting and bagging
729
730
731
        the uploaded files, computing all keys, create an *upload* entry in the NOMAD-coe
        repository db, etc.
        """
732
733
734
735
736
        # 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
737
738
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
739
740
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
741
742
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
743
744
745
746

            if self.temporary:
                os.remove(self.upload_path)
                self.upload_path = None
747

748
        except KeyError:
749
            self.fail('processing requested for non existing upload', log_level=logging.ERROR)
750
751
752
            return
        except ExtractError:
            self.fail('bad .zip/.tar file', log_level=logging.INFO)
Markus Scheidgen's avatar
Markus Scheidgen committed
753
754
            return

755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
    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))

784
    def match_mainfiles(self) -> Generator[Tuple[str, object], None, None]:
785
786
787
788
789
790
791
        """
        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
        """
792
        directories_with_match: Dict[str, str] = dict()
793
794
        upload_files = self.staging_upload_files
        for filename in upload_files.raw_file_manifest():
795
            self._preprocess_files(filename)
796
            try:
797
                parser = match_parser(filename, upload_files)
798
                if parser is not None:
799
800
                    directory = os.path.dirname(filename)
                    if directory in directories_with_match:
801
802
803
                        # TODO this might give us the chance to store directory based relationship
                        # between calcs for the future?
                        pass
804
805
                    else:
                        directories_with_match[directory] = filename
806
807

                    yield filename, parser
808
809
810
811
            except Exception as e:
                self.get_logger().error(
                    'exception while matching pot. mainfile',
                    mainfile=filename, exc_info=e)
812

Markus Scheidgen's avatar
Markus Scheidgen committed
813
814
    @task
    def parse_all(self):
815
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
816
        The *task* used to identify mainfile/parser combinations among the upload's files, creates
817
818
        respective :class:`Calc` instances, and triggers their processing.
        """
819
820
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
821
822
        with utils.timer(
                logger, 'upload extracted', step='matching',
823
824
                upload_size=self.upload_files.size):
            for filename, parser in self.match_mainfiles():
825
                calc = Calc.create(
826
                    calc_id=self.upload_files.calc_id(filename),
827
                    mainfile=filename, parser=parser.name,
828
                    worker_hostname=self.worker_hostname,
829
830
                    upload_id=self.upload_id)

831
                calc.process_calc()
Markus Scheidgen's avatar
Markus Scheidgen committed
832

833
    def on_process_complete(self, process_name):
834
        if process_name == 'process_upload' or process_name == 're_process_upload':
835
836
837
            self.check_join()

    def check_join(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
838
839
840
841
842
843
844
845
        """
        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).
        """
846
847
848
849
        total_calcs = self.total_calcs
        processed_calcs = self.processed_calcs

        self.get_logger().debug('check join', processed_calcs=processed_calcs, total_calcs=total_calcs)
850
851
        # 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
852
        if not self.process_running and processed_calcs >= total_calcs:
853
            # this can easily be called multiple times, e.g. upload finished after all calcs finished
854
855
856
            modified_upload = self._get_collection().find_one_and_update(
                {'_id': self.upload_id, 'joined': {'$ne': True}},
                {'$set': {'joined': True}})
857
858
859
860
861
862
            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
863

864
865
866
    def reset(self):
        self.joined = False
        super().reset()
Markus Scheidgen's avatar
Markus Scheidgen committed
867

868
869
870
871
872
873
874
    @property
    def gui_url(self):
        base = config.api_url()[:-3]
        if base.endswith('/'):
            base = base[:-1]
        return '%s/uploads/' % base

875
    def _cleanup_after_processing(self):
876
877
878
879
880
881
        # send email about process finish
        user = self.uploader
        name = '%s %s' % (user.first_name, user.last_name)
        message = '\n'.join([
            'Dear %s,' % name,
            '',
882
            'your data %suploaded at %s has completed processing.' % (
883
                '"%s" ' % self.name if self.name else '', self.upload_time.isoformat()),  # pylint: disable=no-member
884
885
886
887
888
            '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'
889
        ])
890
891
892
893
894
895
896
        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
897

898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
    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()
915
            self.last_update = datetime.utcnow()
916
917
918
919
            self.save()

    @task
    def cleanup(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
920
921
922
923
        """
        The *task* that "cleans" the processing, i.e. removed obsolete files and performs
        pending archival operations. Depends on the type of processing.
        """
924
925
926
927
928
929
930
        search.refresh()

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

931
    def get_calc(self, calc_id) -> Calc:
Markus Scheidgen's avatar
Markus Scheidgen committed
932
        """ Returns the upload calc with the given id or ``None``. """
933
934
        return Calc.objects(upload_id=self.upload_id, calc_id=calc_id).first()

Markus Scheidgen's avatar
Markus Scheidgen committed
935
    @property
936
    def processed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
937
938
939
940
        """
        The number of successfully or not successfully processed calculations. I.e.
        calculations that have finished processing.
        """
941
        return Calc.objects(upload_id=self.upload_id, tasks_status__in=[SUCCESS, FAILURE]).count()
942
943
944

    @property
    def total_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
945
        """ The number of all calculations. """
946
947
948
949
        return Calc.objects(upload_id=self.upload_id).count()

    @property
    def failed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
950
        """ The number of calculations with failed processing. """
951
        return Calc.objects(upload_id=self.upload_id, tasks_status=FAILURE).count()
952

953
    @property
Markus Scheidgen's avatar
Markus Scheidgen committed
954
955
    def pending_calcs(self) -> int:
        """ The number of calculations with pending processing. """
956
        return Calc.objects(upload_id=self.upload_id, tasks_status=PENDING).count()
957

958
    def all_calcs(self, start, end, order_by=None):
Markus Scheidgen's avatar
Markus Scheidgen committed
959
960
961
962
963
964
965
966
        """
        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
        """
967
968
        query = Calc.objects(upload_id=self.upload_id)[start:end]
        return query.order_by(order_by) if order_by is not None else query
969

970
971
    @property
    def outdated_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
972
        """ All successfully processed and outdated calculations. """
973
974
975
976
        return Calc.objects(
            upload_id=self.upload_id, tasks_status=SUCCESS,
            metadata__nomad_version__ne=config.version)

977
978
    @property
    def calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
979
        """ All successfully processed calculations. """
Markus Scheidgen's avatar
Markus Scheidgen committed
980
        return Calc.objects(upload_id=self.upload_id, tasks_status=SUCCESS)
Markus Scheidgen's avatar
Markus Scheidgen committed
981

982
    def to_upload_with_metadata(self, user_metadata: dict = None) -> UploadWithMetadata:
Markus Scheidgen's avatar
Markus Scheidgen committed
983
984
985
986
987
988
989
990
991
        """
        This is the :py:mod:`nomad.datamodel` transformation method to transform
        processing uploads into datamodel uploads. It will also implicitely transform
        all calculations of this upload.

        Arguments:
            user_metadata: A dict of user metadata that is applied to the resulting
                datamodel data and the respective calculations.
        """
992
        # prepare user metadata per upload and per calc
993
994
995
        if user_metadata is not None:
            calc_metadatas: Dict[str, Any] = dict()
            upload_metadata: Dict[str, Any] = dict()
996