data.py 19.7 KB
Newer Older
Markus Scheidgen's avatar
Markus Scheidgen committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
# Copyright 2018 Markus Scheidgen
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an"AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

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


.. autoclass:: Calc
    :members:
.. autoclass:: Upload
    :members:
"""

27
from typing import List, Any, ContextManager, Tuple, Generator
28
from elasticsearch.exceptions import NotFoundError
Markus Scheidgen's avatar
Markus Scheidgen committed
29
from mongoengine import StringField, BooleanField, DateTimeField, DictField, IntField
Markus Scheidgen's avatar
Markus Scheidgen committed
30
import logging
31
from structlog import wrap_logger
32
from contextlib import contextmanager
Markus Scheidgen's avatar
Markus Scheidgen committed
33

34
from nomad import utils, coe_repo, datamodel
35
from nomad.uploads import PathObject, ArchiveBasedStagingUploadFiles
Markus Scheidgen's avatar
Markus Scheidgen committed
36
from nomad.repo import RepoCalc, RepoUpload
37
from nomad.processing.base import Proc, Chord, process, task, PENDING, SUCCESS, FAILURE
Markus Scheidgen's avatar
Markus Scheidgen committed
38
from nomad.parsing import parsers, parser_dict
Markus Scheidgen's avatar
Markus Scheidgen committed
39
from nomad.normalizing import normalizers
Markus Scheidgen's avatar
Markus Scheidgen committed
40
from nomad.utils import lnr
Markus Scheidgen's avatar
Markus Scheidgen committed
41
42
43
44
45


class NotAllowedDuringProcessing(Exception): pass


46
class Calc(Proc, datamodel.Calc):
Markus Scheidgen's avatar
Markus Scheidgen committed
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
    """
    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:
        archive_id: the hash based archive id of the calc
        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
    """
    archive_id = StringField(primary_key=True)
    upload_id = StringField()
    mainfile = StringField()
    parser = StringField()

    meta: Any = {
        'indices': [
69
            'upload_id', 'mainfile', 'code', 'parser', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
70
71
72
73
74
75
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._parser_backend = None
76
77
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
78
        self._calc_proc_logwriter = None
79
        self._calc_proc_logwriter_ctx: ContextManager = None
Markus Scheidgen's avatar
Markus Scheidgen committed
80
81
82
83
84

    @classmethod
    def get(cls, id):
        return cls.get_by_id(id, 'archive_id')

Markus Scheidgen's avatar
Markus Scheidgen committed
85
    @property
86
87
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
88

89
90
91
92
    @property
    def calc_hash(self) -> str:
        return utils.archive.calc_hash(self.archive_id)

93
94
95
96
97
98
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
        return self._upload

99
100
101
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
102
            self._upload_files = ArchiveBasedStagingUploadFiles(self.upload_id, is_authorized=lambda: True, local_path=self.upload.local_path)
103
104
105
106
107
108
        return self._upload_files

    @property
    def upload_hash(self):
        return utils.archive.upload_hash(self.archive_id)

Markus Scheidgen's avatar
Markus Scheidgen committed
109
110
111
112
    def delete(self):
        """
        Delete this calculation and all associated data. This includes all files,
        the archive, and this search index entry.
113
        TODO is this needed? Or do we always delete hole uploads in bulk.
Markus Scheidgen's avatar
Markus Scheidgen committed
114
        """
115
116
        # delete all files
        self.upload_files.delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
117
118

        # delete the search index entry
119
120
121
122
123
124
        try:
            elastic_entry = RepoCalc.get(self.archive_id)
            if elastic_entry is not None:
                elastic_entry.delete()
        except NotFoundError:
            pass
Markus Scheidgen's avatar
Markus Scheidgen committed
125
126
127
128
129
130
131
132

        # delete this mongo document
        super().delete()

    def get_logger(self, **kwargs):
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile,
133
            upload_hash=self.upload_hash, calc_hash=self.calc_hash,
Markus Scheidgen's avatar
Markus Scheidgen committed
134
            archive_id=self.archive_id, **kwargs)
135

Markus Scheidgen's avatar
Markus Scheidgen committed
136
137
        return logger

138
139
140
141
142
143
144
145
    def get_calc_logger(self, **kwargs):
        """
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
        """
        logger = self.get_logger(**kwargs)

        if self._calc_proc_logwriter is None:
146
            self._calc_proc_logwriter_ctx = self.upload_files.archive_log_file(self.calc_hash, 'wt')
147
            self._calc_proc_logwriter = self._calc_proc_logwriter_ctx.__enter__()  # pylint: disable=E1101
148

149
        def save_to_calc_log(logger, method_name, event_dict):
150
151
152
153
154
155
156
157
158
159
160
            program = event_dict.get('normalizer', 'parser')
            event = event_dict.get('event', '')
            entry = '[%s] %s: %s' % (method_name, program, event)
            if len(entry) > 120:
                self._calc_proc_logwriter.write(entry[:120])
                self._calc_proc_logwriter.write('...')
            else:
                self._calc_proc_logwriter.write(entry)
            self._calc_proc_logwriter.write('\n')
            return event_dict

161
        return wrap_logger(logger, processors=[save_to_calc_log])
162

Markus Scheidgen's avatar
Markus Scheidgen committed
163
164
    @process
    def process(self):
165
        logger = self.get_logger()
166
        if self.upload is None:
167
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
168
169
170
171
172
173

        try:
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
174
            # close loghandler that was not closed due to failures
175
            try:
176
177
178
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
179
180
181
182
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

            # inform parent proc about completion
183
            self.upload.completed_child()
Markus Scheidgen's avatar
Markus Scheidgen committed
184
185
186

    @task
    def parsing(self):
187
188
        context = dict(parser=self.parser, step=self.parser)
        logger = self.get_calc_logger(**context)
189
        parser = parser_dict[self.parser]
Markus Scheidgen's avatar
Markus Scheidgen committed
190

191
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
192
193
            self._parser_backend = parser.run(
                self.upload_files.raw_file_object(self.mainfile).os_path, logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
194

195
196
197
198
199
200
        self._parser_backend.openNonOverlappingSection('section_calculation_info')
        self._parser_backend.addValue('upload_id', self.upload_id)
        self._parser_backend.addValue('archive_id', self.archive_id)
        self._parser_backend.addValue('main_file', self.mainfile)
        self._parser_backend.addValue('parser_name', self.parser)

Markus Scheidgen's avatar
Markus Scheidgen committed
201
        if self._parser_backend.status[0] != 'ParseSuccess':
202
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
203
            error = self._parser_backend.status[1]
204
            self._parser_backend.addValue('parse_status', 'ParseFailure')
205
            self.fail(error, level=logging.DEBUG, **context)
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
        else:
            self._parser_backend.addValue('parse_status', 'ParseSuccess')

        self._parser_backend.closeNonOverlappingSection('section_calculation_info')

        self.add_processor_info(self.parser)

    @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:
        self._parser_backend.openContext('/section_calculation_info/0')
        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')
                self._parser_backend.addValue('archive_processor_warning_number', len(warnings))
                self._parser_backend.addArrayValues('archive_processor_warnings', [str(warning) for warning in warnings])
            else:
                self._parser_backend.addValue('archive_processor_status', 'Success')
        else:
            errors = self._parser_backend.status[1]
            self._parser_backend.addValue('archive_processor_error', str(errors))

        self._parser_backend.closeNonOverlappingSection('section_archive_processing_info')
        self._parser_backend.closeContext('/section_calculation_info/0')
Markus Scheidgen's avatar
Markus Scheidgen committed
238
239
240
241
242

    @task
    def normalizing(self):
        for normalizer in normalizers:
            normalizer_name = normalizer.__name__
243
244
            context = dict(normalizer=normalizer_name, step=normalizer_name)
            logger = self.get_calc_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
245
246

            with utils.timer(
247
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
248
249
                with self.use_parser_backend(normalizer_name) as backend:
                    normalizer(backend).normalize(logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
250

251
252
            failed = self._parser_backend.status[0] != 'ParseSuccess'
            if failed:
253
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
254
                error = self._parser_backend.status[1]
255
                self.fail(error, level=logging.WARNING, **context)
256
257
258
259
                break
            else:
                logger.debug(
                    'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
260
261
262

    @task
    def archiving(self):
263
264
265
266
267
        logger = self.get_logger()

        upload_hash, calc_hash = self.archive_id.split('/')
        additional = dict(
            mainfile=self.mainfile,
268
            upload_time=self.upload.upload_time,
269
270
            staging=True,
            restricted=False,
271
            user_id=self.upload.user_id,
272
            aux_files=list(self.upload_files.calc_files(self.mainfile, with_mainfile=False)))
273
274

        with utils.timer(logger, 'indexed', step='index'):
275
            # persist to elastic search
276
            repo_calc = RepoCalc.create_from_backend(
277
278
279
280
                self._parser_backend,
                additional=additional,
                upload_hash=upload_hash,
                calc_hash=calc_hash,
281
282
                upload_id=self.upload_id)
            repo_calc.persist()
283

Markus Scheidgen's avatar
Markus Scheidgen committed
284
285
286
287
        with utils.timer(
                logger, 'archived', step='archive',
                input_size=self.mainfile_file.size) as log_data:

288
            # persist the archive
289
            with self.upload_files.archive_file(self.calc_hash, 'wt') as out:
290
291
                self._parser_backend.write_json(out, pretty=True)

292
            log_data.update(archive_size=self.upload_files.archive_file_object(self.calc_hash).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
293
294
295
296
297
298

        # close loghandler
        if self._calc_proc_logwriter is not None:
            with utils.timer(
                    logger, 'archived log', step='archive_log',
                    input_size=self.mainfile_file.size) as log_data:
299
                self._calc_proc_logwriter_ctx.__exit__(None, None, None)  # pylint: disable=E1101
300
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
301

302
                log_data.update(log_size=self.upload_files.archive_log_file_object(self.calc_hash).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
303

Markus Scheidgen's avatar
Markus Scheidgen committed
304

305
class Upload(Chord, datamodel.Upload):
Markus Scheidgen's avatar
Markus Scheidgen committed
306
307
308
309
310
311
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
312
        local_path: optional local path, e.g. for files that are already somewhere on the server
Markus Scheidgen's avatar
Markus Scheidgen committed
313
314
315
316
317
318
        additional_metadata: optional user provided additional meta data
        upload_id: the upload id generated by the database
        in_staging: true if the upload is still in staging and can be edited by the uploader
        is_private: true if the upload and its derivitaves are only visible to the uploader
        upload_time: the timestamp when the system realised the upload
        upload_hash: the hash of the uploaded file
319
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
320
321
322
323
324
325
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)

    name = StringField(default=None)
326
    local_path = StringField(default=None)
Markus Scheidgen's avatar
Markus Scheidgen committed
327
328
329
330
331
332
333
334
    additional_metadata = DictField(default=None)

    in_staging = BooleanField(default=True)
    is_private = BooleanField(default=False)

    upload_time = DateTimeField()
    upload_hash = StringField(default=None)

335
    user_id = StringField(required=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
336

337
338
    coe_repo_upload_id = IntField(default=None)

339
340
    _initiated_parsers = IntField(default=-1)

Markus Scheidgen's avatar
Markus Scheidgen committed
341
342
    meta: Any = {
        'indexes': [
343
            'upload_hash', 'user_id', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
344
345
346
347
348
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
349
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
350
351
352
353
354
355

    @classmethod
    def get(cls, id):
        return cls.get_by_id(id, 'upload_id')

    @classmethod
356
    def user_uploads(cls, user: coe_repo.User) -> List['Upload']:
Markus Scheidgen's avatar
Markus Scheidgen committed
357
        """ Returns all uploads for the given user. Currently returns all uploads. """
358
        return cls.objects(user_id=str(user.user_id), in_staging=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
359

360
361
362
363
    @property
    def uploader(self):
        return coe_repo.User.from_user_id(self.user_id)

Markus Scheidgen's avatar
Markus Scheidgen committed
364
365
366
367
368
369
370
371
    def get_logger(self, **kwargs):
        logger = super().get_logger()
        logger = logger.bind(upload_id=self.upload_id, **kwargs)
        return logger

    def delete(self):
        logger = self.get_logger(task='delete')

372
        if not (self.completed or self.current_task == 'uploading'):
Markus Scheidgen's avatar
Markus Scheidgen committed
373
374
            raise NotAllowedDuringProcessing()

375
376
        with lnr(logger, 'delete all files of upload'):
            self.upload_files.delete()
377

378
        with lnr(logger, 'deleting calcs db entries'):
379
            # delete repo entries
Markus Scheidgen's avatar
Markus Scheidgen committed
380
            self.to(RepoUpload).delete()
381
382
383

            # delete calc processings
            Calc.objects(upload_id=self.upload_id).delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
384

385
        with lnr(logger, 'deleting upload db entry'):
Markus Scheidgen's avatar
Markus Scheidgen committed
386
387
388
389
390
391
392
393
            super().delete()

    @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.
394
395

        Arguments:
396
            user (coe_repo.User): The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
397
        """
398
        user: coe_repo.User = kwargs['user']
399
400
401
        del(kwargs['user'])
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
402
        kwargs.update(user_id=str(user.user_id))
Markus Scheidgen's avatar
Markus Scheidgen committed
403
        self = super().create(**kwargs)
404

Markus Scheidgen's avatar
Markus Scheidgen committed
405
        self._continue_with('uploading')
406

Markus Scheidgen's avatar
Markus Scheidgen committed
407
408
        return self

409
    def unstage(self, meta_data):
410
        self.get_logger().info('unstage')
411
412
413
414

        if not (self.completed or self.current_task == 'uploading'):
            raise NotAllowedDuringProcessing()

Markus Scheidgen's avatar
Markus Scheidgen committed
415
        self.in_staging = False
Markus Scheidgen's avatar
Markus Scheidgen committed
416
        self.to(RepoUpload).unstage()
417
        coe_repo.Upload.add(self, meta_data)
Markus Scheidgen's avatar
Markus Scheidgen committed
418
419
        self.save()

Markus Scheidgen's avatar
Markus Scheidgen committed
420
421
422
423
424
425
426
427
428
    @process
    def process(self):
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
        pass

429
    @property
430
431
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
432
            self._upload_files = ArchiveBasedStagingUploadFiles(self.upload_id, is_authorized=lambda: True, local_path=self.local_path)
433
        return self._upload_files
434

Markus Scheidgen's avatar
Markus Scheidgen committed
435
436
    @task
    def extracting(self):
437
438
439
440
441
442
        """
        Task performed before the actual parsing/normalizing. Extracting and bagging
        the uploaded files, computing all keys, create an *upload* entry in the NOMAD-coe
        repository db, etc.
        """
        # extract the uploaded file, this will also create a bagit bag.
Markus Scheidgen's avatar
Markus Scheidgen committed
443
444
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
445
446
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
447
448
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
Markus Scheidgen's avatar
Markus Scheidgen committed
449
        except KeyError as e:
450
            self.fail('process request for non existing upload', level=logging.ERROR)
Markus Scheidgen's avatar
Markus Scheidgen committed
451
452
            return

453
        # create and save a hash for the upload
Markus Scheidgen's avatar
Markus Scheidgen committed
454
        try:
455
            self.upload_hash = self.upload_id  # TODO self.upload_file.upload_hash()
456
        except Exception as e:
Markus Scheidgen's avatar
Markus Scheidgen committed
457
458
459
            self.fail('could not create upload hash', e)
            return

460
        # check if the file was already uploaded and processed before
Markus Scheidgen's avatar
Markus Scheidgen committed
461
        if self.to(RepoUpload).exists():
Markus Scheidgen's avatar
Markus Scheidgen committed
462
463
464
            self.fail('The same file was already uploaded and processed.', level=logging.INFO)
            return

465
    def match_mainfiles(self) -> Generator[Tuple[str, object], None, None]:
466
467
468
469
470
471
472
        """
        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
        """
473
        for filename in self.upload_files.raw_file_manifest():
474
475
            for parser in parsers:
                try:
476
                    with self.upload_files.raw_file(filename) as mainfile_f:
477
                        if parser.is_mainfile(filename, lambda fn: mainfile_f):
478
                            yield filename, parser
479
                except Exception as e:
480
                    self.get_logger().error(
481
482
483
                        'exception while matching pot. mainfile',
                        mainfile=filename, exc_info=e)

Markus Scheidgen's avatar
Markus Scheidgen committed
484
485
    @task
    def parse_all(self):
486
        """
487
        Identified mainfile/parser combinations among the upload's files, creates
488
489
        respective :class:`Calc` instances, and triggers their processing.
        """
490
491
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
492
        # TODO: deal with multiple possible parser specs
Markus Scheidgen's avatar
Markus Scheidgen committed
493
494
        with utils.timer(
                logger, 'upload extracted', step='matching',
495
                upload_size=self.upload_files.size):
496
            total_calcs = 0
497
            for filename, parser in self.match_mainfiles():
498
499
500
501
502
503
504
                calc = Calc.create(
                    archive_id='%s/%s' % (self.upload_hash, utils.hash(filename)),
                    mainfile=filename, parser=parser.name,
                    upload_id=self.upload_id)

                calc.process()
                total_calcs += 1
Markus Scheidgen's avatar
Markus Scheidgen committed
505

506
507
508
509
510
        # have to save the total_calcs information for chord management
        self.spwaned_childred(total_calcs)

    def join(self):
        self.cleanup()
Markus Scheidgen's avatar
Markus Scheidgen committed
511
512
513

    @task
    def cleanup(self):
514
515
516
517
518
519
520
        # TODO issue #83
        with utils.timer(
                self.get_logger(), 'pack staging upload', step='cleaning',
                upload_size=self.upload_files.size):
            pass
            # self.upload_files.pack()
            # self.upload_files.delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
521
522

    @property
523
524
525
526
527
528
529
530
531
532
533
    def processed_calcs(self):
        return Calc.objects(upload_id=self.upload_id, status__in=[SUCCESS, FAILURE]).count()

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

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

534
535
536
537
    @property
    def pending_calcs(self):
        return Calc.objects(upload_id=self.upload_id, status=PENDING).count()

538
539
    def all_calcs(self, start, end, order_by='mainfile'):
        return Calc.objects(upload_id=self.upload_id)[start:end].order_by(order_by)
540
541
542
543

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