data.py 43.3 KB
Newer Older
Markus Scheidgen's avatar
Markus Scheidgen committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# Copyright 2018 Markus Scheidgen
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an"AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

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


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

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

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

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

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


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

    It also contains the calculations processing and its state.

52
53
    The attribute list, does not include the various metadata properties generated
    while parsing, including ``code_name``, ``code_version``, etc.
Markus Scheidgen's avatar
Markus Scheidgen committed
54
55

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

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

68
69
    metadata = DictField()

Markus Scheidgen's avatar
Markus Scheidgen committed
70
    meta: Any = {
Markus Scheidgen's avatar
Markus Scheidgen committed
71
        'indexes': [
72
73
74
75
            'upload_id',
            ('upload_id', 'mainfile'),
            ('upload_id', 'parser'),
            ('upload_id', 'tasks_status'),
76
            ('upload_id', 'process_status'),
77
78
79
            ('upload_id', 'metadata.nomad_version'),
            'metadata.published',
            'metadata.datasets'
Markus Scheidgen's avatar
Markus Scheidgen committed
80
            'metadata.pid'
Markus Scheidgen's avatar
Markus Scheidgen committed
81
82
83
84
85
        ]
    }

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

92
93
94
95
96
97
98
99
100
101
    @classmethod
    def from_calc_with_metadata(cls, calc_with_metadata):
        calc = Calc.create(
            calc_id=calc_with_metadata.calc_id,
            upload_id=calc_with_metadata.upload_id,
            mainfile=calc_with_metadata.mainfile,
            metadata=calc_with_metadata.to_dict())

        return calc

Markus Scheidgen's avatar
Markus Scheidgen committed
102
103
    @classmethod
    def get(cls, id):
104
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
105

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

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

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

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

133
        if self._calc_proc_logwriter_ctx is None:
134
135
136
137
138
139
140
141
142
143
144
145
146
147
            try:
                self._calc_proc_logwriter_ctx = self.upload_files.archive_log_file(self.calc_id, 'wt')
                self._calc_proc_logwriter = self._calc_proc_logwriter_ctx.__enter__()  # pylint: disable=E1101
            except KeyError:
                # cannot open log file
                pass

        if self._calc_proc_logwriter_ctx is None:
            return logger
        else:
            def save_to_calc_log(logger, method_name, event_dict):
                if self._calc_proc_logwriter is not None:
                    program = event_dict.get('normalizer', 'parser')
                    event = event_dict.get('event', '')
148
149
150
151

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

                return event_dict

            return wrap_logger(logger, processors=[save_to_calc_log])
160

161
162
163
164
165
166
167
168
169
170
    @process
    def re_process_calc(self):
        """
        Processes a calculation again. This means there is already metadata and
        instead of creating it initially, we are just updating the existing
        records.
        """
        logger = self.get_logger()

        try:
171
172
173
174
175
176
177
            calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
            calc_with_metadata.upload_id = self.upload_id
            calc_with_metadata.calc_id = self.calc_id
            calc_with_metadata.calc_hash = self.upload_files.calc_hash(self.mainfile)
            calc_with_metadata.mainfile = self.mainfile
            calc_with_metadata.nomad_version = config.version
            calc_with_metadata.nomad_commit = config.commit
178
            calc_with_metadata.last_processing = datetime.utcnow()
179
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
180
            self.metadata = calc_with_metadata.to_dict()
181
182
183
184
185
186
187
188
189
190
191
192
193

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

Markus Scheidgen's avatar
Markus Scheidgen committed
194
    @process
195
    def process_calc(self):
196
197
198
199
        """
        Processes a new calculation that has no prior records in the mongo, elastic,
        or filesystem storage. It will create an initial set of (user) metadata.
        """
200
        logger = self.get_logger()
201
        if self.upload is None:
202
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
203
204

        try:
205
206
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
207
            calc_with_metadata = datamodel.CalcWithMetadata(
208
209
210
211
212
                upload_id=self.upload_id,
                calc_id=self.calc_id,
                calc_hash=self.upload_files.calc_hash(self.mainfile),
                mainfile=self.mainfile)
            calc_with_metadata.published = False
213
            calc_with_metadata.uploader = self.upload.user_id
214
            calc_with_metadata.upload_time = self.upload.upload_time
215
            calc_with_metadata.upload_name = self.upload.name
216
            calc_with_metadata.nomad_version = config.version
217
            calc_with_metadata.nomad_commit = config.commit
218
            calc_with_metadata.last_processing = datetime.utcnow()
219
220
221
            calc_with_metadata.files = self.upload_files.calc_files(self.mainfile)
            self.metadata = calc_with_metadata.to_dict()

222
223
224
225
226
            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
227
228
229
230
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
231
            # close loghandler that was not closed due to failures
232
            try:
233
234
235
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
236
237
238
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

239
240
241
242
    def fail(self, *errors, log_level=logging.ERROR, **kwargs):
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
243
            calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
244
245
246
247
248
249
250
251
            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
252
253
254
255
256
257
258

            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

259
260
261
262
263
264
265
266
            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)

267
268
269
270
    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
271
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
272
273
            self.upload.reload()
            self.upload.check_join()
Markus Scheidgen's avatar
Markus Scheidgen committed
274
275
276

    @task
    def parsing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
277
        """ The *task* that encapsulates all parsing related actions. """
278
        context = dict(parser=self.parser, step=self.parser)
279
        logger = self.get_logger(**context)
280
        parser = parser_dict[self.parser]
281
        self.metadata['parser_name'] = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
282

283
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
284
285
286
287
288
289
290
            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)
291
                return
292
293
294
295
296
            except SystemExit:
                self.fail(
                    'parser raised system exit', level=logging.ERROR,
                    error='system exit', **context)
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
297

298
299
        # add the non code specific calc metadata to the backend
        # all other quantities have been determined by parsers/normalizers
300
        self._parser_backend.openNonOverlappingSection('section_entry_info')
301
        self._parser_backend.addValue('upload_id', self.upload_id)
302
        self._parser_backend.addValue('calc_id', self.calc_id)
303
        self._parser_backend.addValue('calc_hash', self.metadata['calc_hash'])
304
        self._parser_backend.addValue('mainfile', self.mainfile)
305
        self._parser_backend.addValue('parser_name', self.parser)
306
307
308
309
310
311
312
313
314
315
316
317
        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)
318

Markus Scheidgen's avatar
Markus Scheidgen committed
319
        if self._parser_backend.status[0] != 'ParseSuccess':
320
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
321
            error = self._parser_backend.status[1]
322
            self.fail(error, level=logging.INFO, **context)
323
324
325
326
327
328
329
330

    @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:
331
        self._parser_backend.openContext('/section_entry_info/0')
332
333
334
335
336
337
338
        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')
339
                self._parser_backend.addValue('number_of_archive_processor_warnings', len(warnings))
340
341
342
343
                self._parser_backend.addArrayValues('archive_processor_warnings', [str(warning) for warning in warnings])
            else:
                self._parser_backend.addValue('archive_processor_status', 'Success')
        else:
344
            errors = self._parser_backend.status[1]
345
346
347
            self._parser_backend.addValue('archive_processor_error', str(errors))

        self._parser_backend.closeNonOverlappingSection('section_archive_processing_info')
348
        self._parser_backend.closeContext('/section_entry_info/0')
Markus Scheidgen's avatar
Markus Scheidgen committed
349
350
351

    @task
    def normalizing(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
352
        """ The *task* that encapsulates all normalizing related actions. """
Markus Scheidgen's avatar
Markus Scheidgen committed
353
        for normalizer in normalizers:
354
355
356
            if normalizer.domain != config.domain:
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
357
            normalizer_name = normalizer.__name__
358
            context = dict(normalizer=normalizer_name, step=normalizer_name)
359
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
360
361

            with utils.timer(
362
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
363
                with self.use_parser_backend(normalizer_name) as backend:
364
365
366
367
368
369
370
                    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
371

372
373
            failed = self._parser_backend.status[0] != 'ParseSuccess'
            if failed:
374
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
375
                error = self._parser_backend.status[1]
376
                self.fail(error, level=logging.WARNING, error=error, **context)
377
378
379
380
                break
            else:
                logger.debug(
                    'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
381
382
383

    @task
    def archiving(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
384
        """ The *task* that encapsulates all archival related actions. """
385
386
        logger = self.get_logger()

387
        calc_with_metadata = datamodel.CalcWithMetadata(**self.metadata)
388
        calc_with_metadata.apply_domain_metadata(self._parser_backend)
389
        calc_with_metadata.processed = True
390

391
392
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
393
            self.metadata = calc_with_metadata.to_dict()
394
395

        # index in search
396
        with utils.timer(logger, 'indexed', step='index'):
397
            search.Entry.from_calc_with_metadata(calc_with_metadata).save()
398

399
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
400
        with utils.timer(
401
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
402
                input_size=self.mainfile_file.size) as log_data:
403
            with self.upload_files.archive_file(self.calc_id, 'wt') as out:
404
                self._parser_backend.write_json(out, pretty=True, root_sections=Domain.instance.root_sections)
405

406
            log_data.update(archive_size=self.upload_files.archive_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
407
408
409
410

        # close loghandler
        if self._calc_proc_logwriter is not None:
            with utils.timer(
411
                    logger, 'archived log', step='logs',
Markus Scheidgen's avatar
Markus Scheidgen committed
412
                    input_size=self.mainfile_file.size) as log_data:
413
                self._calc_proc_logwriter_ctx.__exit__(None, None, None)  # pylint: disable=E1101
414
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
415

416
                log_data.update(log_size=self.upload_files.archive_log_file_object(self.calc_id).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
417

418
    def __str__(self):
419
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
420

421

422
class Upload(Proc):
Markus Scheidgen's avatar
Markus Scheidgen committed
423
424
425
426
427
428
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
429
430
        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
431
432
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
433
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
434
435
        published: Boolean that indicates the publish status
        publish_time: Date when the upload was initially published
436
        last_update: Date of the last publishing/re-processing
Markus Scheidgen's avatar
Markus Scheidgen committed
437
        joined: Boolean indicates if the running processing has joined (:func:`check_join`)
Markus Scheidgen's avatar
Markus Scheidgen committed
438
439
440
441
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
442
443
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
Markus Scheidgen's avatar
Markus Scheidgen committed
444
445
446

    name = StringField(default=None)
    upload_time = DateTimeField()
447
    user_id = StringField(required=True)
448
449
    published = BooleanField(default=False)
    publish_time = DateTimeField()
450
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
451

452
453
    joined = BooleanField(default=False)

Markus Scheidgen's avatar
Markus Scheidgen committed
454
455
    meta: Any = {
        'indexes': [
456
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
Markus Scheidgen's avatar
Markus Scheidgen committed
457
458
459
460
461
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
462
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
463

464
465
    @property
    def metadata(self) -> dict:
Markus Scheidgen's avatar
Markus Scheidgen committed
466
467
468
469
470
471
472
        """
        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.
        """
473
474
475
476
477
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
478
479
480

    @metadata.setter
    def metadata(self, data: dict) -> None:
481
482
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
483

Markus Scheidgen's avatar
Markus Scheidgen committed
484
    @classmethod
485
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
486
        return cls.get_by_id(id, 'upload_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
487
488

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

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

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

    @classmethod
    def create(cls, **kwargs) -> 'Upload':
        """
        Creates a new upload for the given user, a user given name is optional.
        It will populate the record with a signed url and pending :class:`UploadProc`.
        The upload will be already saved to the database.
512
513

        Arguments:
514
            user: The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
515
        """
516
517
        # use kwargs to keep compatibility with super method
        user: datamodel.User = kwargs['user']
518
        del(kwargs['user'])
519

520
521
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
522
        kwargs.update(user_id=user.user_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
523
        self = super().create(**kwargs)
524

Markus Scheidgen's avatar
Markus Scheidgen committed
525
        self._continue_with('uploading')
526

Markus Scheidgen's avatar
Markus Scheidgen committed
527
528
        return self

529
530
531
532
533
    def delete(self):
        """ Deletes this upload process state entry and its calcs. """
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

534
    def delete_upload_local(self):
535
        """
536
        Deletes the upload, including its processing state and
537
        staging files. Local version without celery processing.
538
539
540
541
        """
        logger = self.get_logger()

        with utils.lnr(logger, 'staged upload delete failed'):
542
            with utils.timer(
543
                    logger, 'upload deleted from index', step='index',
544
                    upload_size=self.upload_files.size):
545
                search.delete_upload(self.upload_id)
546

547
            with utils.timer(
548
                    logger, 'staged upload deleted', step='files',
549
550
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
551
552

            self.delete()
553

554
    @process
555
    def delete_upload(self):
556
557
558
559
        """
        Deletes of the upload, including its processing state and
        staging files. This starts the celery process of deleting the upload.
        """
560
        self.delete_upload_local()
561

562
        return True  # do not save the process status on the delete upload
563

564
    @process
565
    def publish_upload(self):
566
        """
567
568
        Moves the upload out of staging to the public area. It will
        pack the staging upload files in to public upload files.
569
        """
570
571
        assert self.processed_calcs > 0

572
        logger = self.get_logger()
573
        logger.info('started to publish')
574

575
        with utils.lnr(logger, 'publish failed'):
576
            upload_with_metadata = self.to_upload_with_metadata(self.metadata)
577
            calcs = upload_with_metadata.calcs
578

579
            with utils.timer(
580
                    logger, 'upload metadata updated', step='metadata',
581
                    upload_size=self.upload_files.size):
582
583

                def create_update(calc):
584
                    calc.published = True
585
                    calc.with_embargo = calc.with_embargo if calc.with_embargo is not None else False
586
587
588
589
590
                    return UpdateOne(
                        {'_id': calc.calc_id},
                        {'$set': {'metadata': calc.to_dict()}})

                Calc._get_collection().bulk_write([create_update(calc) for calc in calcs])
591

592
593
594
595
596
            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)
597
598
599
600

            with utils.timer(
                    logger, 'index updated', step='index',
                    upload_size=self.upload_files.size):
601
                search.publish(calcs)
602

603
604
605
606
607
608
            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
609
610
                    self.publish_time = datetime.utcnow()
                    self.last_update = datetime.utcnow()
611
612
                    self.save()
            else:
613
                self.last_update = datetime.utcnow()
614
                self.save()
615

616
617
618
    @process
    def re_process_upload(self):
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
619
620
621
        A *process* that performs the re-processing of a earlier processed
        upload.

622
623
624
625
626
627
628
629
630
        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

631
632
633
        logger = self.get_logger()
        logger.info('started to re-process')

634
635
636
637
638
639
        # 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)
640
        staging_upload_files = public_upload_files.to_staging_upload_files(create=True)
641
642

        self._continue_with('parse_all')
643
        try:
644
645
646
647
            # 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:
648
649
650
651
652
653
654
655
                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

656
                calc.reset(worker_hostname=self.worker_hostname)
657

658
                parser = match_parser(calc.mainfile, staging_upload_files, strict=False)
659
                if parser is None:
660
661
                    logger.error(
                        'no parser matches during re-process, use the old parser',
662
                        calc_id=calc.calc_id)
663
664
665
666
                elif calc.parser != parser.name:
                    calc.parser = parser.name
                    logger.info(
                        'different parser matches during re-process, use new parser',
667
                        calc_id=calc.calc_id, parser=parser.name)
668
669
670
                calc.re_process_calc()
        except Exception as e:
            # try to remove the staging copy in failure case
671
            if staging_upload_files is not None and staging_upload_files.exists():
672
673
674
                staging_upload_files.delete()

            raise e
675
676
677
678

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

679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
    @process
    def re_pack(self):
        """ A *process* that repacks the raw and archive data based on the current embargo data. """
        assert self.published

        # mock the steps of actual processing
        self._continue_with('uploading')
        self._continue_with('extracting')
        self._continue_with('parse_all')
        self._continue_with('cleanup')

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

Markus Scheidgen's avatar
Markus Scheidgen committed
694
    @process
695
    def process_upload(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
696
        """ A *process* that performs the initial upload processing. """
Markus Scheidgen's avatar
Markus Scheidgen committed
697
698
699
700
701
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
702
        """ A no-op *task* as a stand-in for receiving upload data. """
Markus Scheidgen's avatar
Markus Scheidgen committed
703
704
        pass

705
    @property
706
707
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
708
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
709
710
711

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
712
                self.upload_id, is_authorized=lambda: True, **kwargs)
713

714
        return self._upload_files
715

716
717
718
719
720
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
721
722
    @task
    def extracting(self):
723
        """
724
725
        The *task* performed before the actual parsing/normalizing: extracting
        the uploaded files.
726
        """
727
728
729
730
731
        # 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
732
733
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
734
735
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
736
737
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
738
739
740
741

            if self.temporary:
                os.remove(self.upload_path)
                self.upload_path = None
742

743
        except KeyError:
744
            self.fail('processing requested for non existing upload', log_level=logging.ERROR)
745
746
747
            return
        except ExtractError:
            self.fail('bad .zip/.tar file', log_level=logging.INFO)
Markus Scheidgen's avatar
Markus Scheidgen committed
748
749
            return

750
751
752
753
754
    def _preprocess_files(self, path):
        """
        Some files need preprocessing. Currently we need to add a stripped POTCAR version
        and always restrict/embargo the original.
        """
755
        if os.path.basename(path).startswith('POTCAR'):
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
            # 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))

779
    def match_mainfiles(self) -> Generator[Tuple[str, object], None, None]:
780
781
782
783
784
785
786
        """
        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
        """
787
        directories_with_match: Dict[str, str] = dict()
788
789
        upload_files = self.staging_upload_files
        for filename in upload_files.raw_file_manifest():
790
            self._preprocess_files(filename)
791
            try:
792
                parser = match_parser(filename, upload_files)
793
                if parser is not None:
794
795
                    directory = os.path.dirname(filename)
                    if directory in directories_with_match:
796
797
798
                        # TODO this might give us the chance to store directory based relationship
                        # between calcs for the future?
                        pass
799
800
                    else:
                        directories_with_match[directory] = filename
801
802

                    yield filename, parser
803
804
805
806
            except Exception as e:
                self.get_logger().error(
                    'exception while matching pot. mainfile',
                    mainfile=filename, exc_info=e)
807

Markus Scheidgen's avatar
Markus Scheidgen committed
808
809
    @task
    def parse_all(self):
810
        """
Markus Scheidgen's avatar
Markus Scheidgen committed
811
        The *task* used to identify mainfile/parser combinations among the upload's files, creates
812
813
        respective :class:`Calc` instances, and triggers their processing.
        """
814
815
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
816
817
        with utils.timer(
                logger, 'upload extracted', step='matching',
818
819
                upload_size=self.upload_files.size):
            for filename, parser in self.match_mainfiles():
820
                calc = Calc.create(
821
                    calc_id=self.upload_files.calc_id(filename),
822
                    mainfile=filename, parser=parser.name,
823
                    worker_hostname=self.worker_hostname,
824
825
                    upload_id=self.upload_id)

826
                calc.process_calc()
Markus Scheidgen's avatar
Markus Scheidgen committed
827

828
    def on_process_complete(self, process_name):
829
        if process_name == 'process_upload' or process_name == 're_process_upload':
830
831
832
            self.check_join()

    def check_join(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
833
834
835
836
837
838
839
840
        """
        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).
        """
841
842
843
844
        total_calcs = self.total_calcs
        processed_calcs = self.processed_calcs

        self.get_logger().debug('check join', processed_calcs=processed_calcs, total_calcs=total_calcs)
845
846
        # 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
847
        if not self.process_running and processed_calcs >= total_calcs:
848
            # this can easily be called multiple times, e.g. upload finished after all calcs finished
849
850
851
            modified_upload = self._get_collection().find_one_and_update(
                {'_id': self.upload_id, 'joined': {'$ne': True}},
                {'$set': {'joined': True}})
852
853
854
855
856
857
            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
858

859
860
861
    def reset(self):
        self.joined = False
        super().reset()
Markus Scheidgen's avatar
Markus Scheidgen committed
862

863
    def _cleanup_after_processing(self):
864
865
866
867
868
869
        # send email about process finish
        user = self.uploader
        name = '%s %s' % (user.first_name, user.last_name)
        message = '\n'.join([
            'Dear %s,' % name,
            '',
870
            'your data %suploaded at %s has completed processing.' % (
871
                '"%s" ' % self.name if self.name else '', self.upload_time.isoformat()),  # pylint: disable=no-member
872
            'You can review your data on your upload page: %s' % config.gui_url(),
873
874
875
876
            '',
            'If you encouter any issues with your upload, please let us know and replay to this email.',
            '',
            'The nomad team'
877
        ])
878
879
880
881
882
883
884
        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
885

886
887
888
889
890
891
892
893
894
895
    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):

896
            staging_upload_files.pack(self.to_upload_with_metadata(), skip_raw=True)
897
898
899
900
901
902

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

            staging_upload_files.delete()
903
            self.last_update = datetime.utcnow()
904
905
906
907
            self.save()

    @task
    def cleanup(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
908
909
910
911
        """
        The *task* that "cleans" the processing, i.e. removed obsolete files and performs
        pending archival operations. Depends on the type of processing.
        """
912
913
914
915
916
917
918
        search.refresh()

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

919
    def get_calc(self, calc_id) -> Calc:
Markus Scheidgen's avatar
Markus Scheidgen committed
920
        """ Returns the upload calc with the given id or ``None``. """
921
922
        return Calc.objects(upload_id=self.upload_id, calc_id=calc_id).first()

Markus Scheidgen's avatar
Markus Scheidgen committed
923
    @property
924
    def processed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
925
926
927
928
        """
        The number of successfully or not successfully processed calculations. I.e.
        calculations that have finished processing.
        """
929
        return Calc.objects(upload_id=self.upload_id, tasks_status__in=[SUCCESS, FAILURE]).count()
930
931
932

    @property
    def total_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
933
        """ The number of all calculations. """
934
935
936
937
        return Calc.objects(upload_id=self.upload_id).count()

    @property
    def failed_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
938
        """ The number of calculations with failed processing. """
939
        return Calc.objects(upload_id=self.upload_id, tasks_status=FAILURE).count()
940

941
    @property
Markus Scheidgen's avatar
Markus Scheidgen committed
942
943
    def pending_calcs(self) -> int:
        """ The number of calculations with pending processing. """
944
        return Calc.objects(upload_id=self.upload_id, tasks_status=PENDING).count()
945

946
    def all_calcs(self, start, end, order_by=None):
Markus Scheidgen's avatar
Markus Scheidgen committed
947
948
949
950
951
952
953
954
        """
        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
        """
955
956
        query = Calc.objects(upload_id=self.upload_id)[start:end]
        return query.order_by(order_by) if order_by is not None else query
957

958
959
    @property
    def outdated_calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
960
        """ All successfully processed and outdated calculations. """
961
962
963
964
        return Calc.objects(
            upload_id=self.upload_id, tasks_status=SUCCESS,
            metadata__nomad_version__ne=config.version)

965
966
    @property
    def calcs(self):
Markus Scheidgen's avatar
Markus Scheidgen committed
967
        """ All successfully processed calculations. """
Markus Scheidgen's avatar