data.py 51.9 KB
Newer Older
Markus Scheidgen's avatar
Markus Scheidgen committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
# 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.

15
'''
Markus Scheidgen's avatar
Markus Scheidgen committed
16
17
18
19
20
21
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

25
'''
26
27

from typing import cast, List, Any, Tuple, Iterator, Dict, cast, Iterable
28
from mongoengine import StringField, DateTimeField, DictField, BooleanField, IntField
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
36
from structlog.processors import StackInfoRenderer, format_exc_info, TimeStamper
Alvin Noe Ladines's avatar
Alvin Noe Ladines committed
37
38
39
import yaml
import json
from cachetools import cached, LRUCache
Markus Scheidgen's avatar
Markus Scheidgen committed
40

41
from nomad import utils, config, infrastructure, search, datamodel
Markus Scheidgen's avatar
Markus Scheidgen committed
42
43
44
from nomad.files import (
    PathObject, UploadFiles, ExtractError, ArchiveBasedStagingUploadFiles,
    PublicUploadFiles, StagingUploadFiles)
45
from nomad.processing.base import Proc, process, task, PENDING, SUCCESS, FAILURE
Markus Scheidgen's avatar
Markus Scheidgen committed
46
from nomad.parsing.parsers import parser_dict, match_parser
Markus Scheidgen's avatar
Markus Scheidgen committed
47
from nomad.normalizing import normalizers
Alvin Noe Ladines's avatar
Alvin Noe Ladines committed
48
from nomad.datamodel import EntryArchive, EditableUserMetadata
49
from nomad.archive import query_archive
Markus Scheidgen's avatar
Markus Scheidgen committed
50
from nomad.datamodel.encyclopedia import EncyclopediaMetadata
Markus Scheidgen's avatar
Markus Scheidgen committed
51
52


53
54
55
56
section_metadata = datamodel.EntryArchive.section_metadata.name
section_workflow = datamodel.EntryArchive.section_workflow.name


Alvin Noe Ladines's avatar
Alvin Noe Ladines committed
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
_editable_metadata = {
    quantity.name: quantity for quantity in EditableUserMetadata.m_def.definitions}


@cached(cache=LRUCache(maxsize=100))
def metadata_cached(path):
    for ext in config.aux_metadata_exts:
        full_path = '%s.%s' % (path, ext)
        if os.path.isfile(full_path):
            with open(full_path) as f:
                if full_path.endswith('json'):
                    return json.load(f)
                elif full_path.endswith('yaml'):
                    return yaml.load(f, Loader=getattr(yaml, 'FullLoader'))
                else:
                    return {}
    return {}


76
77
78
79
80
81
82
def _pack_log_event(logger, method_name, event_dict):
    try:
        log_data = dict(event_dict)
        log_data.update(**{
            key: value
            for key, value in getattr(logger, '_context', {}).items()
            if key not in ['service', 'release', 'upload_id', 'calc_id', 'mainfile', 'process_status']})
83
        log_data.update(logger=logger.name)
84
85
86
87
88
89
90
91
92
93
94

        return log_data
    except Exception:
        # raising an exception would cause an indefinite loop
        return event_dict


_log_processors = [
    StackInfoRenderer(),
    _pack_log_event,
    format_exc_info,
95
    TimeStamper(fmt="%Y-%m-%d %H:%M.%S", utc=False)]
96
97


Markus Scheidgen's avatar
Markus Scheidgen committed
98
class Calc(Proc):
99
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
100
101
102
103
104
    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.

105
106
    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
107
108

    Attributes:
109
        calc_id: the calc_id of this calc
Markus Scheidgen's avatar
Markus Scheidgen committed
110
111
112
        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
113

114
115
        metadata: the metadata record wit calc and user metadata, see :class:`datamodel.EntryMetadata`
    '''
116
    calc_id = StringField(primary_key=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
117
118
119
120
    upload_id = StringField()
    mainfile = StringField()
    parser = StringField()

121
122
    metadata = DictField()

Markus Scheidgen's avatar
Markus Scheidgen committed
123
    meta: Any = {
Markus Scheidgen's avatar
Markus Scheidgen committed
124
        'indexes': [
125
            'upload_id',
126
            'parser',
127
128
129
            ('upload_id', 'mainfile'),
            ('upload_id', 'parser'),
            ('upload_id', 'tasks_status'),
130
            ('upload_id', 'process_status'),
131
            ('upload_id', 'metadata.nomad_version'),
132
133
            'metadata.processed',
            'metadata.last_processing',
134
            'metadata.published',
135
            'metadata.datasets',
Markus Scheidgen's avatar
Markus Scheidgen committed
136
            'metadata.pid'
Markus Scheidgen's avatar
Markus Scheidgen committed
137
138
139
140
141
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
142
        self._parser_results: EntryArchive = None
143
144
        self._upload: Upload = None
        self._upload_files: ArchiveBasedStagingUploadFiles = None
145
        self._calc_proc_logs: List[Any] = None
146

147
        self._entry_metadata = None
148

Markus Scheidgen's avatar
Markus Scheidgen committed
149
150
    @classmethod
    def get(cls, id):
151
        return cls.get_by_id(id, 'calc_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
152

Markus Scheidgen's avatar
Markus Scheidgen committed
153
    @property
154
155
    def mainfile_file(self) -> PathObject:
        return self.upload_files.raw_file_object(self.mainfile)
Markus Scheidgen's avatar
Markus Scheidgen committed
156

157
158
159
160
    @property
    def upload(self) -> 'Upload':
        if not self._upload:
            self._upload = Upload.get(self.upload_id)
161
            self._upload.worker_hostname = self.worker_hostname
162
163
        return self._upload

164
165
166
167
168
169
170
171
172
173
174
175
    def apply_entry_metadata(self, entry_metadata: datamodel.EntryMetadata):
        self.metadata = entry_metadata.m_to_dict(
            include_defaults=True,
            categories=[datamodel.MongoMetadata])  # TODO use embedded doc?

    def create_metadata(self) -> datamodel.EntryMetadata:
        '''
        Returns a :class:`nomad.datamodel.EntryMetadata` with values from this
        processing object, not necessarely the user metadata nor the metadata from
        the archive.
        '''
        entry_metadata = datamodel.EntryMetadata()
176
        if self.parser is not None:
177
178
179
            parser = parser_dict[self.parser]
            if parser.domain:
                entry_metadata.domain = parser_dict[self.parser].domain
180
181
182
        entry_metadata.upload_id = self.upload_id
        entry_metadata.calc_id = self.calc_id
        entry_metadata.mainfile = self.mainfile
183
184
        entry_metadata.nomad_version = config.meta.version
        entry_metadata.nomad_commit = config.meta.commit
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
        entry_metadata.uploader = self.upload.user_id
        entry_metadata.upload_time = self.upload.upload_time
        entry_metadata.upload_name = self.upload.name

        return entry_metadata

    def entry_metadata(self, upload_files: UploadFiles) -> datamodel.EntryMetadata:
        '''
        Returns a complete set of :class:`nomad.datamodel.EntryMetadata` including
        the user metadata and metadata from the archive.

        Arguments:
            upload_files:
                The :class:`nomad.files.UploadFiles` instance to read the archive from.
            cache:
                A boolean that indicates if the archive file should be left unclosed,
                e.g. if this method is called for many entries of the same upload.
        '''
        archive = upload_files.read_archive(self.calc_id)
204
        try:
205
206
207
208
209
210
211
212
213
            # instead of loading the whole archive, it should be enough to load the
            # parts that are referenced by section_metadata/EntryMetadata
            # TODO somehow it should determine which root setions too load from the metainfo
            # or configuration
            calc_archive = archive[self.calc_id]
            entry_archive_dict = {section_metadata: calc_archive[section_metadata].to_dict()}
            if section_workflow in calc_archive:
                entry_archive_dict[section_workflow] = calc_archive[section_workflow].to_dict()
            entry_metadata = datamodel.EntryArchive.m_from_dict(entry_archive_dict)[section_metadata]
214
215
216
217
218
219
220
221
222

        except KeyError:
            # Due hard processing failures, it might be possible that an entry might not
            # have an archive
            if self._entry_metadata is not None:
                entry_metadata = self._entry_metadata

            else:
                entry_metadata = self.create_metadata()
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238

        entry_metadata.m_update_from_dict(self.metadata)

        return entry_metadata

    def user_metadata(self) -> datamodel.EntryMetadata:
        '''
        Returns a :class:`nomad.datamodel.EntryMetadata` with values from this
        processing object and the user metadata, not necessarely the metadata from
        the archive.
        '''
        entry_metadata = self.create_metadata()
        entry_metadata.m_update_from_dict(self.metadata)

        return entry_metadata

239
240
241
    @property
    def upload_files(self) -> ArchiveBasedStagingUploadFiles:
        if not self._upload_files:
242
243
            self._upload_files = ArchiveBasedStagingUploadFiles(
                self.upload_id, is_authorized=lambda: True, upload_path=self.upload.upload_path)
244
245
        return self._upload_files

Markus Scheidgen's avatar
Markus Scheidgen committed
246
    def get_logger(self, **kwargs):
247
        '''
248
249
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
250
        '''
251
252
253
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile, calc_id=self.calc_id, **kwargs)
254

255
256
        if self._calc_proc_logs is None:
            self._calc_proc_logs = []
257

258
259
        def save_to_calc_log(logger, method_name, event_dict):
            try:
260
261
                # sanitize the event_dict, because all kinds of values might have been added
                dump_dict = {key: str(value) for key, value in event_dict.items()}
262
263
                dump_dict.update(level=method_name.upper())
                self._calc_proc_logs.append(dump_dict)
264

265
266
267
            except Exception:
                # Exceptions here will cause indefinite loop
                pass
268

269
            return event_dict
270

271
        return wrap_logger(logger, processors=_log_processors + [save_to_calc_log])
272

273
    @process
274
    def re_process_calc(self):
275
        '''
276
277
278
        Processes a calculation again. This means there is already metadata and
        instead of creating it initially, we are just updating the existing
        records.
279
        '''
280
        parser = match_parser(self.upload_files.raw_file_object(self.mainfile).os_path, strict=False)
281
        logger = self.get_logger()
282
283
284

        if parser is None and not config.reprocess_unmatched:
            self.errors = ['no parser matches during re-process, will not re-process this calc']
285

286
287
288
289
290
291
292
            try:
                upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
                with upload_files.read_archive(self.calc_id) as archive:
                    self.upload_files.write_archive(self.calc_id, archive[self.calc_id].to_dict())

            except Exception as e:
                logger.error('could not copy archive for non matching, non reprocessed entry', exc_info=e)
293
                raise e
294

295
296
297
298
299
            # mock the steps of actual processing
            self._continue_with('parsing')
            self._continue_with('normalizing')
            self._continue_with('archiving')
            self._complete()
300
301
            return

302
        if parser is None:
303
304
305
306
            self.get_logger().warn('no parser matches during re-process, use the old parser')
            self.warnings = ['no matching parser found during re-processing']

        elif self.parser != parser.name:
307
308
309
310
311
312
313
314
315
            if parser_dict[self.parser].name == parser.name:
                # parser was just renamed
                self.parser = parser.name

            else:
                self.parser = parser.name
                logger.info(
                    'different parser matches during re-process, use new parser',
                    parser=parser.name)
316

317
        try:
318
319
            self._entry_metadata = self.user_metadata()
            self._entry_metadata.calc_hash = self.upload_files.calc_hash(self.mainfile)
320
321
            self._entry_metadata.nomad_version = config.meta.version
            self._entry_metadata.nomad_commit = config.meta.commit
322
323
            self._entry_metadata.last_processing = datetime.utcnow()
            self._entry_metadata.files = self.upload_files.calc_files(self.mainfile)
324
325
326
327
328
329
330

            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
            # close loghandler that was not closed due to failures
            try:
Markus Scheidgen's avatar
Markus Scheidgen committed
331
                if self._parser_results and self._parser_results.m_resource:
332
                    self._parser_results.section_metadata = None
Markus Scheidgen's avatar
Markus Scheidgen committed
333
                    self._parser_results.m_resource.unload()
334
            except Exception as e:
335
                logger.error('could not unload processing results', exc_info=e)
336

337
338
339
340
341
342
    def _setup_fallback_metadata(self):
        self._entry_metadata = self.create_metadata()
        self._entry_metadata.calc_hash = self.upload_files.calc_hash(self.mainfile)
        self._entry_metadata.last_processing = datetime.utcnow()
        self._entry_metadata.files = self.upload_files.calc_files(self.mainfile)

343
    @process
344
    def process_calc(self):
345
        '''
346
347
        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.
348
        '''
349
        logger = self.get_logger()
350
        if self.upload is None:
351
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
352
353

        try:
354
355
            # save preliminary minimum calc metadata in case processing fails
            # successful processing will replace it with the actual metadata
356
            self._setup_fallback_metadata()
357
358

            if len(self._entry_metadata.files) >= config.auxfile_cutoff:
359
360
361
362
                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
363
364
365
366
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
367
            # close loghandler that was not closed due to failures
368
            try:
Markus Scheidgen's avatar
Markus Scheidgen committed
369
                if self._parser_results and self._parser_results.m_resource:
370
                    self._parser_results.section_metadata = None
Markus Scheidgen's avatar
Markus Scheidgen committed
371
                    self._parser_results.m_resource.unload()
372
            except Exception as e:
373
                logger.error('could unload processing results', exc_info=e)
374

375
    def on_fail(self):
376
377
378
        # in case of failure, index a minimum set of metadata and mark
        # processing failure
        try:
379
380
381
            if self._entry_metadata is None:
                self._setup_fallback_metadata()

382
            self._entry_metadata.processed = False
383

384
            self.apply_entry_metadata(self._entry_metadata)
Markus Scheidgen's avatar
Markus Scheidgen committed
385
386
            self._entry_metadata.apply_domain_metadata(self._parser_results)
            if self._parser_results and self._parser_results.m_resource:
387
                self._parser_results.section_metadata = None
Markus Scheidgen's avatar
Markus Scheidgen committed
388
                self._parser_results.m_resource.unload()
389

390
            self._entry_metadata.a_elastic.index()
391
        except Exception as e:
392
393
            self.get_logger().error(
                'could not index after processing failure', exc_info=e)
394

395
        try:
396
            self.write_archive(None)
397
        except Exception as e:
398
399
            self.get_logger().error(
                'could not write archive after processing failure', exc_info=e)
400

401
402
403
404
405
406
407
408
    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
        if process_name == 'process_calc' or process_name == 're_process_calc' or process_name is None:
            self.upload.reload()
            self.upload.check_join()

Markus Scheidgen's avatar
Markus Scheidgen committed
409
410
    @task
    def parsing(self):
411
        ''' The *task* that encapsulates all parsing related actions. '''
412
        context = dict(parser=self.parser, step=self.parser)
413
        logger = self.get_logger(**context)
414
        parser = parser_dict[self.parser]
415
        self._entry_metadata.parser_name = self.parser
Markus Scheidgen's avatar
Markus Scheidgen committed
416

417
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
418
            try:
419
420
421
                self._parser_results = EntryArchive()
                # allow parsers to read/write metadata
                self._parser_results.m_add_sub_section(EntryArchive.section_metadata, self._entry_metadata)
Markus Scheidgen's avatar
Markus Scheidgen committed
422
423
424
                parser.parse(
                    self.upload_files.raw_file_object(self.mainfile).os_path,
                    self._parser_results, logger=logger)
425

426
            except Exception as e:
427
                self.fail('parser failed with exception', exc_info=e, error=str(e), **context)
428
                return
429
            except SystemExit:
430
                self.fail('parser raised system exit', error='system exit', **context)
431
                return
Markus Scheidgen's avatar
Markus Scheidgen committed
432

433
434
435
436
437
438
439
440
441
442
    def process_phonon(self):
        """Function that is run for phonon calculation before cleanup.
        This task is run by the celery process that is calling the join for the
        upload.

        This function re-opens the Archive for this calculation to add method
        information from another referenced archive. Updates the method
        information in section_encyclopedia as well as the DFT domain metadata.
        """
        try:
443
            logger = self.get_logger(parser=self.parser, step=self.parser)
444
445
446
447
448
449

            # Open the archive of the phonon calculation.
            upload_files = StagingUploadFiles(self.upload_id, is_authorized=lambda: True)
            with upload_files.read_archive(self.calc_id) as archive:
                arch = query_archive(archive, {self.calc_id: self.calc_id})[self.calc_id]
                phonon_archive = EntryArchive.m_from_dict(arch)
Markus Scheidgen's avatar
Markus Scheidgen committed
450
            self._entry_metadata = phonon_archive.section_metadata
451
            self._calc_proc_logs = phonon_archive.processing_logs
452

Markus Scheidgen's avatar
Markus Scheidgen committed
453
454
            # Re-create the parse results
            self._parser_results = phonon_archive
Markus Scheidgen's avatar
Markus Scheidgen committed
455

456
457
458
            # Read in the first referenced calculation. The reference is given as
            # an absolute path which needs to be converted into a path that is
            # relative to upload root.
Markus Scheidgen's avatar
Markus Scheidgen committed
459
            scc = self._parser_results.section_run[0].section_single_configuration_calculation[0]
460
            relative_ref = scc.section_calculation_to_calculation_refs[0].calculation_to_calculation_external_url
461
462
463
464
465
466
            ref_id = upload_files.calc_id(relative_ref)
            with upload_files.read_archive(ref_id) as archive:
                arch = query_archive(archive, {ref_id: ref_id})[ref_id]
                ref_archive = EntryArchive.m_from_dict(arch)

            # Get encyclopedia method information directly from the referenced calculation.
467
            ref_enc_method = ref_archive.section_metadata.encyclopedia.method
468
            if ref_enc_method is None or len(ref_enc_method) == 0 or ref_enc_method.functional_type is None:
469
                raise ValueError("No method information available in referenced calculation.")
Markus Scheidgen's avatar
Markus Scheidgen committed
470
            self._parser_results.section_metadata.encyclopedia.method = ref_enc_method
471
472
473
474
475
476
477
478

            # Overwrite old entry with new data. The metadata is updated with
            # new timestamp and method details taken from the referenced
            # archive.
            self._entry_metadata.last_processing = datetime.utcnow()
            self._entry_metadata.dft.xc_functional = ref_archive.section_metadata.dft.xc_functional
            self._entry_metadata.dft.basis_set = ref_archive.section_metadata.dft.basis_set
            self._entry_metadata.dft.update_group_hash()
479
            self._entry_metadata.encyclopedia.status = EncyclopediaMetadata.status.type.success
480
        except Exception as e:
481
            logger.error("Could not retrieve method information for phonon calculation.", exc_info=e)
Markus Scheidgen's avatar
Markus Scheidgen committed
482
483
484
485
            if self._entry_metadata is None:
                self._setup_fallback_metadata()
                self._entry_metadata.processed = False

486
487
488
489
490
491
            try:
                if self._entry_metadata.encyclopedia is None:
                    self._entry_metadata.encyclopedia = EncyclopediaMetadata()
                self._entry_metadata.encyclopedia.status = EncyclopediaMetadata.status.type.failure
            except Exception as e:
                logger.error("Could set encyclopedia status.", exc_info=e)
492

493
        finally:
494
495
496
497
498
499
500
501
502
503
504
505
506
            # persist the calc metadata
            with utils.timer(logger, 'saved calc metadata', step='metadata'):
                self.apply_entry_metadata(self._entry_metadata)

            # index in search
            with utils.timer(logger, 'indexed', step='index'):
                self._entry_metadata.a_elastic.index()

            # persist the archive
            with utils.timer(
                    logger, 'archived', step='archive',
                    input_size=self.mainfile_file.size) as log_data:

Markus Scheidgen's avatar
Markus Scheidgen committed
507
                archive_size = self.write_archive(self._parser_results)
508
509
                log_data.update(archive_size=archive_size)

Markus Scheidgen's avatar
Markus Scheidgen committed
510
511
    @task
    def normalizing(self):
512
        ''' The *task* that encapsulates all normalizing related actions. '''
513
514

        # allow normalizer to access and add data to the entry metadata
515
516
517
        if self._parser_results.section_metadata is None:
            self._parser_results.m_add_sub_section(
                datamodel.EntryArchive.section_metadata, self._entry_metadata)
518

Markus Scheidgen's avatar
Markus Scheidgen committed
519
        for normalizer in normalizers:
520
            if normalizer.domain != parser_dict[self.parser].domain:
521
522
                continue

Markus Scheidgen's avatar
Markus Scheidgen committed
523
            normalizer_name = normalizer.__name__
524
            context = dict(normalizer=normalizer_name, step=normalizer_name)
525
            logger = self.get_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
526

Markus Scheidgen's avatar
Markus Scheidgen committed
527
528
529
530
531
532
            with utils.timer(logger, 'normalizer executed', input_size=self.mainfile_file.size):
                try:
                    normalizer(self._parser_results).normalize(logger=logger)
                    logger.info('processor completed successfull', **context)
                except Exception as e:
                    self.fail('normalizer failed with exception', exc_info=e, error=str(e), **context)
Markus Scheidgen's avatar
Markus Scheidgen committed
533

Alvin Noe Ladines's avatar
Alvin Noe Ladines committed
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
    def _read_metadata_from_file(self, logger):
        # metadata file name defined in nomad.config nomad_metadata.yaml/json
        # which can be placed in the directory containing the mainfile or somewhere up
        # highest priority is directory with mainfile

        metadata_file = config.aux_metadata_file
        metadata_dir = os.path.dirname(self.mainfile_file.os_path)

        metadata = {}
        metadata_path = None
        while metadata_dir:
            metadata_part = metadata_cached(os.path.join(metadata_dir, metadata_file))

            for key, val in metadata_part.items():
                metadata.setdefault(key, val)

            metadata_dir = os.path.dirname(metadata_dir)
            if metadata_path is not None:
                break

        for key, val in metadata.items():
            definition = _editable_metadata.get(key, None)
            if not definition:
                logger.warn('Cannot set metadata %s' % key)
                continue
            self._entry_metadata.m_set(definition, val)

Markus Scheidgen's avatar
Markus Scheidgen committed
561
562
    @task
    def archiving(self):
563
        ''' The *task* that encapsulates all archival related actions. '''
564
565
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
566
        self._entry_metadata.apply_domain_metadata(self._parser_results)
567
        self._entry_metadata.processed = True
568

Alvin Noe Ladines's avatar
Alvin Noe Ladines committed
569
570
571
        # read metadata from file
        self._read_metadata_from_file(logger)

572
573
        # persist the calc metadata
        with utils.timer(logger, 'saved calc metadata', step='metadata'):
574
            self.apply_entry_metadata(self._entry_metadata)
575
576

        # index in search
577
        with utils.timer(logger, 'indexed', step='index'):
578
            self._entry_metadata.a_elastic.index()
579

580
        # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
581
        with utils.timer(
582
                logger, 'archived', step='archive',
Markus Scheidgen's avatar
Markus Scheidgen committed
583
                input_size=self.mainfile_file.size) as log_data:
584

Markus Scheidgen's avatar
Markus Scheidgen committed
585
            archive_size = self.write_archive(self._parser_results)
586
            log_data.update(archive_size=archive_size)
Markus Scheidgen's avatar
Markus Scheidgen committed
587

Markus Scheidgen's avatar
Markus Scheidgen committed
588
    def write_archive(self, archive: EntryArchive):
589
590
591
592
593
594
        def filter_processing_logs(logs):
            if len(logs) > 100:
                return [
                    log for log in logs
                    if log.get('level') != 'DEBUG']
            return logs
Markus Scheidgen's avatar
Markus Scheidgen committed
595

596
597
598
        if self._calc_proc_logs is None:
            self._calc_proc_logs = []

Markus Scheidgen's avatar
Markus Scheidgen committed
599
600
        if archive is not None:
            archive = archive.m_copy()
601
        else:
Markus Scheidgen's avatar
Markus Scheidgen committed
602
            archive = datamodel.EntryArchive()
603

Markus Scheidgen's avatar
Markus Scheidgen committed
604
605
        if archive.section_metadata is None:
            archive.m_add_sub_section(datamodel.EntryArchive.section_metadata, self._entry_metadata)
606

Markus Scheidgen's avatar
Markus Scheidgen committed
607
        archive.processing_logs = filter_processing_logs(self._calc_proc_logs)
608
609

        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
610
            return self.upload_files.write_archive(self.calc_id, archive.m_to_dict())
611
612
        except Exception as e:
            # most likely failed due to domain data, try to write metadata and processing logs
Markus Scheidgen's avatar
Markus Scheidgen committed
613
614
615
616
            archive = datamodel.EntryArchive()
            archive.m_add_sub_section(datamodel.EntryArchive.section_metadata, self._entry_metadata)
            archive.processing_logs = filter_processing_logs(self._calc_proc_logs)
            self.upload_files.write_archive(self.calc_id, archive.m_to_dict())
617
            raise e
Markus Scheidgen's avatar
Markus Scheidgen committed
618

619
    def __str__(self):
620
        return 'calc %s calc_id=%s upload_id%s' % (super().__str__(), self.calc_id, self.upload_id)
621

622

623
class Upload(Proc):
624
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
625
626
627
628
629
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
630
631
        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
632
633
        upload_id: the upload id generated by the database
        upload_time: the timestamp when the system realised the upload
634
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
635
636
        published: Boolean that indicates the publish status
        publish_time: Date when the upload was initially published
637
        last_update: Date of the last publishing/re-processing
638
        joined: Boolean indicates if the running processing has joined (:func:`check_join`)
639
    '''
Markus Scheidgen's avatar
Markus Scheidgen committed
640
641
642
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)
643
644
    upload_path = StringField(default=None)
    temporary = BooleanField(default=False)
645
    embargo_length = IntField(default=36)
Markus Scheidgen's avatar
Markus Scheidgen committed
646
647
648

    name = StringField(default=None)
    upload_time = DateTimeField()
649
    user_id = StringField(required=True)
650
651
    published = BooleanField(default=False)
    publish_time = DateTimeField()
652
    last_update = DateTimeField()
Markus Scheidgen's avatar
Markus Scheidgen committed
653

654
655
    joined = BooleanField(default=False)

Markus Scheidgen's avatar
Markus Scheidgen committed
656
657
    meta: Any = {
        'indexes': [
658
            'user_id', 'tasks_status', 'process_status', 'published', 'upload_time'
659
        ]
Markus Scheidgen's avatar
Markus Scheidgen committed
660
661
662
663
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
664
        self._upload_files: ArchiveBasedStagingUploadFiles = None
Markus Scheidgen's avatar
Markus Scheidgen committed
665

666
667
    @property
    def metadata(self) -> dict:
668
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
669
670
671
672
673
        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.
674
        '''
675
676
677
678
679
        try:
            upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True)
        except KeyError:
            return None
        return upload_files.user_metadata
680
681
682

    @metadata.setter
    def metadata(self, data: dict) -> None:
683
684
        upload_files = PublicUploadFiles(self.upload_id, is_authorized=lambda: True, create=True)
        upload_files.user_metadata = data
685

Markus Scheidgen's avatar
Markus Scheidgen committed
686
    @classmethod
687
    def get(cls, id: str, include_published: bool = True) -> 'Upload':
688
        return cls.get_by_id(id, 'upload_id')
Markus Scheidgen's avatar
Markus Scheidgen committed
689
690

    @classmethod
691
    def user_uploads(cls, user: datamodel.User, **kwargs) -> List['Upload']:
692
        ''' Returns all uploads for the given user. Kwargs are passed to mongo query. '''
693
        return cls.objects(user_id=str(user.user_id), **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
694

695
696
    @property
    def uploader(self):
697
        return datamodel.User.get(self.user_id)
698

Markus Scheidgen's avatar
Markus Scheidgen committed
699
700
    def get_logger(self, **kwargs):
        logger = super().get_logger()
701
702
        user = self.uploader
        user_name = '%s %s' % (user.first_name, user.last_name)
703
        # We are not using 'user_id' because logstash (?) will filter these entries ?!
704
705
        logger = logger.bind(
            upload_id=self.upload_id, upload_name=self.name, user_name=user_name,
706
            user=user.user_id, **kwargs)
Markus Scheidgen's avatar
Markus Scheidgen committed
707
708
709
710
        return logger

    @classmethod
    def create(cls, **kwargs) -> 'Upload':
711
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
712
713
714
        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.
715
716

        Arguments:
717
            user: The user that created the upload.
718
        '''
719
720
        # use kwargs to keep compatibility with super method
        user: datamodel.User = kwargs['user']
721
        del(kwargs['user'])
722

723
724
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
725
        kwargs.update(user_id=user.user_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
726
        self = super().create(**kwargs)
727

Markus Scheidgen's avatar
Markus Scheidgen committed
728
        self._continue_with('uploading')
729

Markus Scheidgen's avatar
Markus Scheidgen committed
730
731
        return self

732
    def delete(self):
733
        ''' Deletes this upload process state entry and its calcs. '''
734
735
736
        Calc.objects(upload_id=self.upload_id).delete()
        super().delete()

737
    def delete_upload_local(self):
738
        '''
739
        Deletes the upload, including its processing state and
740
        staging files. Local version without celery processing.
741
        '''
742
743
        logger = self.get_logger()

744
        with utils.lnr(logger, 'upload delete failed'):
745
            with utils.timer(
746
                    logger, 'upload deleted from index', step='index',
747
                    upload_size=self.upload_files.size):
748
                search.delete_upload(self.upload_id)
749

750
            with utils.timer(
751
                    logger, 'upload deleted', step='files',
752
753
                    upload_size=self.upload_files.size):
                self.upload_files.delete()
754
755

            self.delete()
756

757
    @process
758
    def delete_upload(self):
759
        '''
760
761
        Deletes of the upload, including its processing state and
        staging files. This starts the celery process of deleting the upload.
762
        '''
763
        self.delete_upload_local()
764

765
        return True  # do not save the process status on the delete upload
766

767
    @process
768
    def publish_upload(self):
769
        '''
770
771
        Moves the upload out of staging to the public area. It will
        pack the staging upload files in to public upload files.
772
        '''
773
774
        assert self.processed_calcs > 0

775
        logger = self.get_logger()
776
        logger.info('started to publish')
777

778
        with utils.lnr(logger, 'publish failed'):
779
            with self.entries_metadata(self.metadata) as calcs:
780

781
                with utils.timer(
782
                        logger, 'upload metadata updated', step='metadata',
783
                        upload_size=self.upload_files.size):
784

785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
                    def create_update(calc):
                        calc.published = True
                        calc.with_embargo = calc.with_embargo if calc.with_embargo is not None else False
                        return UpdateOne(
                            {'_id': calc.calc_id},
                            {'$set': {'metadata': calc.m_to_dict(
                                include_defaults=True, categories=[datamodel.MongoMetadata])}})

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

                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(calcs)
800

801
                with utils.timer(
802
                        logger, 'index updated', step='index',
803
                        upload_size=self.upload_files.size):
804
805
806
807
808
809
810
811
812
813
814
815
                    search.publish(calcs)

                if isinstance(self.upload_files, StagingUploadFiles):
                    with utils.timer(
                            logger, 'staged upload deleted', step='delete staged',
                            upload_size=self.upload_files.size):
                        self.upload_files.delete()
                        self.published = True
                        self.publish_time = datetime.utcnow()
                        self.last_update = datetime.utcnow()
                        self.save()
                else:
816
                    self.last_update = datetime.utcnow()
817
                    self.save()
818

819
820
    @process
    def re_process_upload(self):
821
        '''
Markus Scheidgen's avatar
Markus Scheidgen committed
822
823
824
        A *process* that performs the re-processing of a earlier processed
        upload.

825
826
827
828
829
830
        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.
831
        '''
832
833
834
        logger = self.get_logger()
        logger.info('started to re-process')

835
836
837
838
839
        # 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')
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854

        if self.published:
            try:
                staging_upload_files = StagingUploadFiles(self.upload_id)
                # public files exist and there is a staging directory, it is probably old
                # and we delete it first
                staging_upload_files.delete()
                logger.warn('deleted old staging files')

            except KeyError as e:
                logger.info('reprocessing published files')
        else:
            logger.info('reprocessing staging files')

        staging_upload_files = self.upload_files.to_staging_upload_files(create=True)
855
856

        self._continue_with('parse_all')
857
        try:
858
            # check if a calc is already/still processing
859
860
861
862
863
864
865
866
            processing = Calc.objects(
                upload_id=self.upload_id,
                **Calc.process_running_mongoengine_query()).count()

            if processing > 0:
                logger.warn(
                    'processes are still/already running on calc, they will be resetted',
                    count=processing)
867

868
            # reset all calcs
869
            Calc._get_collection().update_many(
870
871
872
                dict(upload_id=self.upload_id),
                {'$set': Calc.reset_pymongo_update(worker_hostname=self.worker_hostname)})

873
874
            # process call calcs
            Calc.process_all(Calc.re_process_calc, dict(upload_id=self.upload_id), exclude=['metadata'])
875
876

            logger.info('completed to trigger re-process of all calcs')
877
878
        except Exception as e:
            # try to remove the staging copy in failure case
879
880
            logger.error('failed to trigger re-process of all calcs', exc_info=e)

881
882
883
            if self.published:
                if staging_upload_files is not None and staging_upload_files.exists():
                    staging_upload_files.delete()
884
885

            raise e
886
887
888
889

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

890
891
    @process
    def re_pack(self):
892
        ''' A *process* that repacks the raw and archive data based on the current embargo data. '''
893
894
895
896
897
898
899
900
        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')

901
        self.upload_files.re_pack(self.user_metadata())
902
        self.joined = True
903
904
        self._complete()

Markus Scheidgen's avatar
Markus Scheidgen committed
905
    @process
906
    def process_upload(self):
907
        ''' A *process* that performs the initial upload processing. '''
Markus Scheidgen's avatar
Markus Scheidgen committed
908
909
910
911
912
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
913
        ''' A no-op *task* as a stand-in for receiving upload data. '''
Markus Scheidgen's avatar
Markus Scheidgen committed
914
915
        pass

916
    @property
917
918
    def upload_files(self) -> UploadFiles:
        upload_files_class = ArchiveBasedStagingUploadFiles if not self.published else PublicUploadFiles
919
        kwargs = dict(upload_path=self.upload_path) if not self.published else {}
920
921
922

        if not self._upload_files or not isinstance(self._upload_files, upload_files_class):
            self._upload_files = upload_files_class(
923
                self.upload_id, is_authorized=lambda: True, **kwargs)
924

925
        return self._upload_files
926

927
928
929
930
931
    @property
    def staging_upload_files(self) -> ArchiveBasedStagingUploadFiles:
        assert not self.published
        return cast(ArchiveBasedStagingUploadFiles, self.upload_files)

Markus Scheidgen's avatar
Markus Scheidgen committed
932
933
    @task
    def extracting(self):
934
        '''
Markus Scheidgen's avatar