data.py 18.5 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
Markus Scheidgen's avatar
Markus Scheidgen committed
28
from mongoengine import StringField, BooleanField, DateTimeField, DictField, IntField
Markus Scheidgen's avatar
Markus Scheidgen committed
29
import logging
30
from structlog import wrap_logger
31
from contextlib import contextmanager
Markus Scheidgen's avatar
Markus Scheidgen committed
32

33
from nomad import utils, coe_repo, datamodel
34
from nomad.uploads import PathObject, ArchiveBasedStagingUploadFiles
Markus Scheidgen's avatar
Markus Scheidgen committed
35
from nomad.repo import RepoCalc, RepoUpload
36
from nomad.processing.base import Proc, Chord, process, task, PENDING, SUCCESS, FAILURE
Markus Scheidgen's avatar
Markus Scheidgen committed
37
from nomad.parsing import parsers, parser_dict
Markus Scheidgen's avatar
Markus Scheidgen committed
38
39
40
41
42
43
from nomad.normalizing import normalizers


class NotAllowedDuringProcessing(Exception): pass


44
class Calc(Proc, datamodel.Calc):
Markus Scheidgen's avatar
Markus Scheidgen committed
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
    """
    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': [
67
            'upload_id', 'mainfile', 'code', 'parser', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
68
69
70
71
72
73
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._parser_backend = None
74
75
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
76
        self._calc_proc_logwriter = None
77
        self._calc_proc_logwriter_ctx: ContextManager = None
Markus Scheidgen's avatar
Markus Scheidgen committed
78
79
80
81
82

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

Markus Scheidgen's avatar
Markus Scheidgen committed
83
    @property
84
85
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
86

87
88
89
90
    @property
    def calc_hash(self) -> str:
        return utils.archive.calc_hash(self.archive_id)

91
92
93
94
95
96
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
        return self._upload

97
98
99
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
100
            self._upload_files = ArchiveBasedStagingUploadFiles(self.upload_id, is_authorized=lambda: True, local_path=self.upload.local_path)
101
102
103
104
105
106
        return self._upload_files

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

Markus Scheidgen's avatar
Markus Scheidgen committed
107
108
109
110
    def get_logger(self, **kwargs):
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile,
111
            upload_hash=self.upload_hash, calc_hash=self.calc_hash,
Markus Scheidgen's avatar
Markus Scheidgen committed
112
            archive_id=self.archive_id, **kwargs)
113

Markus Scheidgen's avatar
Markus Scheidgen committed
114
115
        return logger

116
117
118
119
120
121
122
123
    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:
124
            self._calc_proc_logwriter_ctx = self.upload_files.archive_log_file(self.calc_hash, 'wt')
125
            self._calc_proc_logwriter = self._calc_proc_logwriter_ctx.__enter__()  # pylint: disable=E1101
126

127
        def save_to_calc_log(logger, method_name, event_dict):
128
129
130
131
132
133
134
135
136
137
138
            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

139
        return wrap_logger(logger, processors=[save_to_calc_log])
140

Markus Scheidgen's avatar
Markus Scheidgen committed
141
142
    @process
    def process(self):
143
        logger = self.get_logger()
144
        if self.upload is None:
145
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
146
147
148
149
150
151

        try:
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
152
            # close loghandler that was not closed due to failures
153
            try:
154
155
156
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
157
158
159
160
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

            # inform parent proc about completion
161
            self.upload.completed_child()
Markus Scheidgen's avatar
Markus Scheidgen committed
162
163
164

    @task
    def parsing(self):
165
166
        context = dict(parser=self.parser, step=self.parser)
        logger = self.get_calc_logger(**context)
167
        parser = parser_dict[self.parser]
Markus Scheidgen's avatar
Markus Scheidgen committed
168

169
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
170
171
            self._parser_backend = parser.run(
                self.upload_files.raw_file_object(self.mainfile).os_path, logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
172

173
174
175
176
177
178
        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
179
        if self._parser_backend.status[0] != 'ParseSuccess':
180
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
181
            error = self._parser_backend.status[1]
182
            self._parser_backend.addValue('parse_status', 'ParseFailure')
183
            self.fail(error, level=logging.DEBUG, **context)
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
        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
216
217
218
219
220

    @task
    def normalizing(self):
        for normalizer in normalizers:
            normalizer_name = normalizer.__name__
221
222
            context = dict(normalizer=normalizer_name, step=normalizer_name)
            logger = self.get_calc_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
223
224

            with utils.timer(
225
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
226
227
                with self.use_parser_backend(normalizer_name) as backend:
                    normalizer(backend).normalize(logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
228

229
230
            failed = self._parser_backend.status[0] != 'ParseSuccess'
            if failed:
231
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
232
                error = self._parser_backend.status[1]
233
                self.fail(error, level=logging.WARNING, **context)
234
235
236
237
                break
            else:
                logger.debug(
                    'completed normalizer successfully', normalizer=normalizer_name)
Markus Scheidgen's avatar
Markus Scheidgen committed
238
239
240

    @task
    def archiving(self):
241
242
243
244
245
        logger = self.get_logger()

        upload_hash, calc_hash = self.archive_id.split('/')
        additional = dict(
            mainfile=self.mainfile,
246
            upload_time=self.upload.upload_time,
247
248
            staging=True,
            restricted=False,
249
            user_id=self.upload.user_id,
250
            aux_files=list(self.upload_files.calc_files(self.mainfile, with_mainfile=False)))
251
252

        with utils.timer(logger, 'indexed', step='index'):
253
            # persist to elastic search
254
            repo_calc = RepoCalc.create_from_backend(
255
256
257
258
                self._parser_backend,
                additional=additional,
                upload_hash=upload_hash,
                calc_hash=calc_hash,
259
260
                upload_id=self.upload_id)
            repo_calc.persist()
261

Markus Scheidgen's avatar
Markus Scheidgen committed
262
263
264
265
        with utils.timer(
                logger, 'archived', step='archive',
                input_size=self.mainfile_file.size) as log_data:

266
            # persist the archive
267
            with self.upload_files.archive_file(self.calc_hash, 'wt') as out:
268
269
                self._parser_backend.write_json(out, pretty=True)

270
            log_data.update(archive_size=self.upload_files.archive_file_object(self.calc_hash).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
271
272
273
274
275
276

        # 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:
277
                self._calc_proc_logwriter_ctx.__exit__(None, None, None)  # pylint: disable=E1101
278
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
279

280
                log_data.update(log_size=self.upload_files.archive_log_file_object(self.calc_hash).size)
Markus Scheidgen's avatar
Markus Scheidgen committed
281

Markus Scheidgen's avatar
Markus Scheidgen committed
282

283
class Upload(Chord, datamodel.Upload):
Markus Scheidgen's avatar
Markus Scheidgen committed
284
285
286
287
288
289
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
290
        local_path: optional local path, e.g. for files that are already somewhere on the server
Markus Scheidgen's avatar
Markus Scheidgen committed
291
292
293
294
295
        additional_metadata: optional user provided additional meta data
        upload_id: the upload id generated by the database
        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
296
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
297
298
299
300
301
302
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)

    name = StringField(default=None)
303
    local_path = StringField(default=None)
Markus Scheidgen's avatar
Markus Scheidgen committed
304
305
306
307
308
309
310
    additional_metadata = DictField(default=None)

    is_private = BooleanField(default=False)

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

311
    user_id = StringField(required=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
312

313
314
    coe_repo_upload_id = IntField(default=None)

315
316
    _initiated_parsers = IntField(default=-1)

Markus Scheidgen's avatar
Markus Scheidgen committed
317
318
    meta: Any = {
        'indexes': [
319
            'upload_hash', 'user_id', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
320
321
322
323
324
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
325
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
326
327
328
329
330
331

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

    @classmethod
332
    def user_uploads(cls, user: coe_repo.User) -> List['Upload']:
Markus Scheidgen's avatar
Markus Scheidgen committed
333
        """ Returns all uploads for the given user. Currently returns all uploads. """
334
        return cls.objects(user_id=str(user.user_id))
Markus Scheidgen's avatar
Markus Scheidgen committed
335

336
337
338
339
    @property
    def uploader(self):
        return coe_repo.User.from_user_id(self.user_id)

Markus Scheidgen's avatar
Markus Scheidgen committed
340
341
342
343
344
345
    def get_logger(self, **kwargs):
        logger = super().get_logger()
        logger = logger.bind(upload_id=self.upload_id, **kwargs)
        return logger

    def delete(self):
346
        if not (self.completed or self.current_task == 'uploading'):
Markus Scheidgen's avatar
Markus Scheidgen committed
347
            raise NotAllowedDuringProcessing()
348
349
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
350
351
352
353
354
355
356

    @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.
357
358

        Arguments:
359
            user (coe_repo.User): The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
360
        """
361
        user: coe_repo.User = kwargs['user']
362
363
364
        del(kwargs['user'])
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
365
        kwargs.update(user_id=str(user.user_id))
Markus Scheidgen's avatar
Markus Scheidgen committed
366
        self = super().create(**kwargs)
367

Markus Scheidgen's avatar
Markus Scheidgen committed
368
        self._continue_with('uploading')
369

Markus Scheidgen's avatar
Markus Scheidgen committed
370
371
        return self

372
    def unstage(self, meta_data):
373
        self.get_logger().info('unstage')
374
375
376
377

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

378
379
        self.delete()

Markus Scheidgen's avatar
Markus Scheidgen committed
380
        self.to(RepoUpload).unstage()
381
        coe_repo.Upload.add(self, meta_data)
Markus Scheidgen's avatar
Markus Scheidgen committed
382
383
        self.save()

384
385
386
        self.upload_files.pack()
        self.upload_files.delete()

Markus Scheidgen's avatar
Markus Scheidgen committed
387
388
389
390
391
392
393
394
395
    @process
    def process(self):
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
        pass

396
    @property
397
398
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
399
            self._upload_files = ArchiveBasedStagingUploadFiles(self.upload_id, is_authorized=lambda: True, local_path=self.local_path)
400
        return self._upload_files
401

Markus Scheidgen's avatar
Markus Scheidgen committed
402
403
    @task
    def extracting(self):
404
405
406
407
408
409
        """
        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
410
411
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
412
413
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
414
415
                    upload_size=self.upload_files.size):
                self.upload_files.extract()
Markus Scheidgen's avatar
Markus Scheidgen committed
416
        except KeyError as e:
417
            self.fail('process request for non existing upload', level=logging.ERROR)
Markus Scheidgen's avatar
Markus Scheidgen committed
418
419
            return

420
        # create and save a hash for the upload
Markus Scheidgen's avatar
Markus Scheidgen committed
421
        try:
422
            self.upload_hash = self.upload_id  # TODO self.upload_file.upload_hash()
423
        except Exception as e:
Markus Scheidgen's avatar
Markus Scheidgen committed
424
425
426
            self.fail('could not create upload hash', e)
            return

427
        # check if the file was already uploaded and processed before
Markus Scheidgen's avatar
Markus Scheidgen committed
428
        if self.to(RepoUpload).exists():
Markus Scheidgen's avatar
Markus Scheidgen committed
429
430
431
            self.fail('The same file was already uploaded and processed.', level=logging.INFO)
            return

432
    def match_mainfiles(self) -> Generator[Tuple[str, object], None, None]:
433
434
435
436
437
438
439
        """
        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
        """
440
        for filename in self.upload_files.raw_file_manifest():
441
442
            for parser in parsers:
                try:
443
                    with self.upload_files.raw_file(filename) as mainfile_f:
444
                        if parser.is_mainfile(filename, lambda fn: mainfile_f):
445
                            yield filename, parser
446
                except Exception as e:
447
                    self.get_logger().error(
448
449
450
                        'exception while matching pot. mainfile',
                        mainfile=filename, exc_info=e)

Markus Scheidgen's avatar
Markus Scheidgen committed
451
452
    @task
    def parse_all(self):
453
        """
454
        Identified mainfile/parser combinations among the upload's files, creates
455
456
        respective :class:`Calc` instances, and triggers their processing.
        """
457
458
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
459
        # TODO: deal with multiple possible parser specs
Markus Scheidgen's avatar
Markus Scheidgen committed
460
461
        with utils.timer(
                logger, 'upload extracted', step='matching',
462
                upload_size=self.upload_files.size):
463
            total_calcs = 0
464
            for filename, parser in self.match_mainfiles():
465
466
467
468
469
470
471
                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
472

473
474
475
476
477
        # 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
478
479
480

    @task
    def cleanup(self):
481
482
483
484
485
        # TODO issue #83
        with utils.timer(
                self.get_logger(), 'pack staging upload', step='cleaning',
                upload_size=self.upload_files.size):
            pass
Markus Scheidgen's avatar
Markus Scheidgen committed
486
487

    @property
488
489
490
491
492
493
494
495
496
497
498
    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()

499
500
501
502
    @property
    def pending_calcs(self):
        return Calc.objects(upload_id=self.upload_id, status=PENDING).count()

503
504
    def all_calcs(self, start, end, order_by='mainfile'):
        return Calc.objects(upload_id=self.upload_id)[start:end].order_by(order_by)
505
506
507
508

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