data.py 17.8 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
27
28
# 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:
"""

from typing import List, Any
from datetime import datetime
29
from elasticsearch.exceptions import NotFoundError
Markus Scheidgen's avatar
Markus Scheidgen committed
30
from mongoengine import StringField, BooleanField, DateTimeField, DictField, IntField
Markus Scheidgen's avatar
Markus Scheidgen committed
31
import logging
32
import base64
Markus Scheidgen's avatar
Markus Scheidgen committed
33
import time
34
from structlog import wrap_logger
Markus Scheidgen's avatar
Markus Scheidgen committed
35

36
from nomad import config, utils
37
from nomad.files import UploadFile, ArchiveFile, ArchiveLogFile
Markus Scheidgen's avatar
Markus Scheidgen committed
38
from nomad.repo import RepoCalc
Markus Scheidgen's avatar
Markus Scheidgen committed
39
from nomad.user import User
40
from nomad.processing.base import Proc, Chord, process, task, PENDING, SUCCESS, FAILURE, RUNNING
Markus Scheidgen's avatar
Markus Scheidgen committed
41
from nomad.parsing import parsers, parser_dict
Markus Scheidgen's avatar
Markus Scheidgen committed
42
from nomad.normalizing import normalizers
Markus Scheidgen's avatar
Markus Scheidgen committed
43
from nomad.utils import lnr
Markus Scheidgen's avatar
Markus Scheidgen committed
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73


class NotAllowedDuringProcessing(Exception): pass


class Calc(Proc):
    """
    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
        mainfile_tmp_path: path to the mainfile extracted for processing
    """
    archive_id = StringField(primary_key=True)
    upload_id = StringField()
    mainfile = StringField()
    parser = StringField()
    mainfile_tmp_path = StringField()

    meta: Any = {
        'indices': [
74
            'upload_id', 'mainfile', 'code', 'parser', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
75
76
77
78
79
80
81
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._parser_backend = None
        self._upload = None
82
        self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
83
84
85
86
87
88
89
90
91

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

    def delete(self):
        """
        Delete this calculation and all associated data. This includes all files,
        the archive, and this search index entry.
92
        TODO is this needed? Or do we always delete hole uploads in bulk.
Markus Scheidgen's avatar
Markus Scheidgen committed
93
94
95
        """
        # delete the archive
        if self.archive_id is not None:
96
            ArchiveFile(self.archive_id).delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
97
98

        # delete the search index entry
99
100
101
102
103
104
        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
105
106
107
108
109
110
111
112
113
114

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

    def get_logger(self, **kwargs):
        upload_hash, calc_hash = self.archive_id.split('/')
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile,
            upload_hash=upload_hash, calc_hash=calc_hash, **kwargs)
115

Markus Scheidgen's avatar
Markus Scheidgen committed
116
117
        return logger

118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
    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:
            self._calc_proc_logwriter = ArchiveLogFile(self.archive_id).open('wt')

        def save_to_cacl_log(logger, method_name, event_dict):
            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

        return wrap_logger(logger, processors=[save_to_cacl_log])

Markus Scheidgen's avatar
Markus Scheidgen committed
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
    @property
    def json_dict(self):
        """ A json serializable dictionary representation. """
        data = {
            'archive_id': self.archive_id,
            'mainfile': self.mainfile,
            'upload_id': self.upload_id,
            'parser': self.parser
        }
        data.update(super().json_dict)
        return {key: value for key, value in data.items() if value is not None}

    @process
    def process(self):
        self._upload = Upload.get(self.upload_id)
157
        logger = self.get_logger()
Markus Scheidgen's avatar
Markus Scheidgen committed
158
        if self._upload is None:
159
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
160
161
162
163
164
165

        try:
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
166
            # close loghandler that was not closed due to failures
167
            try:
168
169
170
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
171
172
173
174
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

            # inform parent proc about completion
175
            self._upload.completed_child()
Markus Scheidgen's avatar
Markus Scheidgen committed
176
177
178

    @task
    def parsing(self):
179
        logger = self.get_calc_logger(parser=self.parser)
180
        parser = parser_dict[self.parser]
181
        with utils.timer(logger, 'parser executed', step=self.parser):
182
            self._parser_backend = parser.run(self.mainfile_tmp_path, logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
183
        if self._parser_backend.status[0] != 'ParseSuccess':
184
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
185
186
187
188
189
190
191
            error = self._parser_backend.status[1]
            self.fail(error, level=logging.DEBUG)

    @task
    def normalizing(self):
        for normalizer in normalizers:
            normalizer_name = normalizer.__name__
192
            logger = self.get_calc_logger(normalizer=normalizer_name)
193
            with utils.timer(logger, 'normalizer executed', step=normalizer_name):
194
                normalizer(self._parser_backend).normalize(logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
195
            if self._parser_backend.status[0] != 'ParseSuccess':
196
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
197
198
199
                error = self._parser_backend.status[1]
                self.fail(error, normalizer=normalizer_name, level=logging.WARNING)
                return
200
            logger.debug(
Markus Scheidgen's avatar
Markus Scheidgen committed
201
202
203
204
                'completed normalizer successfully', normalizer=normalizer_name)

    @task
    def archiving(self):
205
206
207
208
209
210
211
212
213
214
215
        logger = self.get_logger()

        upload_hash, calc_hash = self.archive_id.split('/')
        additional = dict(
            mainfile=self.mainfile,
            upload_time=self._upload.upload_time,
            staging=True,
            restricted=False,
            user_id=self._upload.user_id)

        with utils.timer(logger, 'indexed', step='index'):
216
217
218
219
220
221
222
223
            # persist to elastic search
            RepoCalc.create_from_backend(
                self._parser_backend,
                additional=additional,
                upload_hash=upload_hash,
                calc_hash=calc_hash,
                upload_id=self.upload_id)

224
        with utils.timer(logger, 'archived', step='archive'):
225
226
227
228
            # persist the archive
            with ArchiveFile(self.archive_id).write_archive_json() as out:
                self._parser_backend.write_json(out, pretty=True)

229
        with utils.timer(logger, 'archived log', step='archive_log'):
230
231
232
233
            # close loghandler
            if self._calc_proc_logwriter is not None:
                self._calc_proc_logwriter.close()
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
234

Markus Scheidgen's avatar
Markus Scheidgen committed
235

236
class Upload(Chord):
Markus Scheidgen's avatar
Markus Scheidgen committed
237
238
239
240
241
242
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
243
        local_path: optional local path, e.g. for files that are already somewhere on the server
Markus Scheidgen's avatar
Markus Scheidgen committed
244
245
246
247
248
249
        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
250
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
251
252
253
254
255
256
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)

    name = StringField(default=None)
257
    local_path = StringField(default=None)
Markus Scheidgen's avatar
Markus Scheidgen committed
258
259
260
261
262
263
264
265
    additional_metadata = DictField(default=None)

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

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

266
    user_id = StringField(required=True)
267
268
    upload_url = StringField(default=None)
    upload_command = StringField(default=None)
Markus Scheidgen's avatar
Markus Scheidgen committed
269

270
271
    _initiated_parsers = IntField(default=-1)

Markus Scheidgen's avatar
Markus Scheidgen committed
272
273
    meta: Any = {
        'indexes': [
274
            'upload_hash', 'user_id', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
275
276
277
278
279
280
281
282
283
284
285
286
287
288
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
        self._upload = None

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

    @classmethod
    def user_uploads(cls, user: User) -> List['Upload']:
        """ Returns all uploads for the given user. Currently returns all uploads. """
Markus Scheidgen's avatar
Markus Scheidgen committed
289
        return cls.objects(user_id=user.email, in_staging=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
290
291
292
293
294
295
296
297
298
299
300
301
302
303

    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')

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

        with lnr(logger, 'delete upload file'):
            try:
304
                UploadFile(self.upload_id, local_path=self.local_path).delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
305
306
307
308
309
310
311
312
313
            except KeyError:
                if self.current_task == 'uploading':
                    logger.debug(
                        'Upload exist, but file does not exist. '
                        'It was probably aborted and deleted.')
                else:
                    logger.debug('Upload exist, but uploaded file does not exist.')

        with lnr(logger, 'deleting calcs'):
314
            # delete archive files
315
            ArchiveFile.delete_archives(upload_hash=self.upload_hash)
316
317

            # delete repo entries
Markus Scheidgen's avatar
Markus Scheidgen committed
318
            RepoCalc.delete_upload(upload_id=self.upload_id)
319
320
321

            # delete calc processings
            Calc.objects(upload_id=self.upload_id).delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
322
323
324
325

        with lnr(logger, 'deleting upload'):
            super().delete()

326
327
    @classmethod
    def _external_objects_url(cls, url):
328
        """ Replaces the given internal object storage url with an URL that allows
329
330
            external access.
        """
331
        return 'http://%s:%s%s%s' % (config.services.api_host, config.services.api_port, config.services.api_base_path, url)
332

Markus Scheidgen's avatar
Markus Scheidgen committed
333
334
335
336
337
338
    @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.
339
340
341

        Arguments:
            user (User): The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
342
        """
343
344
345
346
347
        user: User = kwargs['user']
        del(kwargs['user'])
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
        kwargs.update(user_id=user.email)
Markus Scheidgen's avatar
Markus Scheidgen committed
348
        self = super().create(**kwargs)
349
350
351
352

        basic_auth_token = base64.b64encode(b'%s:' % user.generate_auth_token()).decode('utf-8')

        self.upload_url = cls._external_objects_url('/uploads/%s/file' % self.upload_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
353
        self.upload_command = 'curl -H "Authorization: Basic %s" "%s" --upload-file local_file' % (
354
355
            basic_auth_token, self.upload_url)

Markus Scheidgen's avatar
Markus Scheidgen committed
356
        self._continue_with('uploading')
357

Markus Scheidgen's avatar
Markus Scheidgen committed
358
359
360
361
362
363
364
365
366
        return self

    @property
    def is_stale(self) -> bool:
        if self.current_task == 'uploading' and self.upload_time is None:
            return (datetime.now() - self.create_time).days > 1
        else:
            return False

Markus Scheidgen's avatar
Markus Scheidgen committed
367
    def unstage(self):
368
        self.get_logger().info('unstage')
Markus Scheidgen's avatar
Markus Scheidgen committed
369
        self.in_staging = False
370
        RepoCalc.unstage(upload_id=self.upload_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
371
372
        self.save()

Markus Scheidgen's avatar
Markus Scheidgen committed
373
374
375
376
377
    @property
    def json_dict(self) -> dict:
        """ A json serializable dictionary representation. """
        data = {
            'name': self.name,
378
            'local_path': self.local_path,
Markus Scheidgen's avatar
Markus Scheidgen committed
379
380
            'additional_metadata': self.additional_metadata,
            'upload_id': self.upload_id,
381
            'upload_url': self.upload_url,
382
            'upload_command': self.upload_command,
Markus Scheidgen's avatar
Markus Scheidgen committed
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
            'upload_time': self.upload_time.isoformat() if self.upload_time is not None else None,
            'is_stale': self.is_stale,
        }
        data.update(super().json_dict)
        return {key: value for key, value in data.items() if value is not None}

    @process
    def process(self):
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
        pass

    @task
    def extracting(self):
        logger = self.get_logger()
        try:
402
            with utils.timer(logger, 'upload extracted', step='extracting'):
403
404
                self._upload = UploadFile(self.upload_id, local_path=self.local_path)
                self._upload.extract()
Markus Scheidgen's avatar
Markus Scheidgen committed
405
406
407
408
409
410
        except KeyError as e:
            self.fail('process request for non existing upload', level=logging.INFO)
            return

        try:
            self.upload_hash = self._upload.hash()
411
        except Exception as e:
Markus Scheidgen's avatar
Markus Scheidgen committed
412
413
414
            self.fail('could not create upload hash', e)
            return

Markus Scheidgen's avatar
Markus Scheidgen committed
415
        if RepoCalc.upload_exists(self.upload_hash):
Markus Scheidgen's avatar
Markus Scheidgen committed
416
417
418
419
420
            self.fail('The same file was already uploaded and processed.', level=logging.INFO)
            return

    @task
    def parse_all(self):
421
422
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
423
        # TODO: deal with multiple possible parser specs
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
        with utils.timer(logger, 'upload extracted', step='matching'):
            total_calcs = 0
            for filename in self._upload.filelist:
                for parser in parsers:
                    try:
                        if parser.is_mainfile(filename, lambda fn: self._upload.open_file(fn)):
                            tmp_mainfile = self._upload.get_path(filename)
                            calc = Calc.create(
                                archive_id='%s/%s' % (self.upload_hash, utils.hash(filename)),
                                mainfile=filename, parser=parser.name,
                                mainfile_tmp_path=tmp_mainfile,
                                upload_id=self.upload_id)

                            calc.process()
                            total_calcs += 1
                    except Exception as e:
                        self.warning(
                            'exception while matching pot. mainfile',
                            mainfile=filename, exc_info=e)
Markus Scheidgen's avatar
Markus Scheidgen committed
443

444
445
446
447
448
        # 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
449
450
451
452

    @task
    def cleanup(self):
        try:
453
            with utils.timer(self.get_logger(), 'processing cleaned up', step='cleaning'):
454
                upload = UploadFile(self.upload_id, local_path=self.local_path)
Markus Scheidgen's avatar
Markus Scheidgen committed
455
        except KeyError as e:
Markus Scheidgen's avatar
Markus Scheidgen committed
456
            self.fail('Upload does not exist', exc_info=e)
Markus Scheidgen's avatar
Markus Scheidgen committed
457
458
            return

459
        upload.remove_extract()
Markus Scheidgen's avatar
Markus Scheidgen committed
460
461
462
        self.get_logger().debug('closed upload')

    @property
463
464
465
466
467
468
469
470
471
472
473
    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()

474
475
476
477
    @property
    def pending_calcs(self):
        return Calc.objects(upload_id=self.upload_id, status=PENDING).count()

478
479
    def all_calcs(self, start, end, order_by='mainfile'):
        return Calc.objects(upload_id=self.upload_id)[start:end].order_by(order_by)
480

Markus Scheidgen's avatar
Markus Scheidgen committed
481
    @staticmethod
482
483
484
485
486
    def repair_all():
        """
        Utitlity function that will look for suspiciously looking conditions in
        all uncompleted downloads. It ain't a perfect world.
        """
487
488
489
        # TODO this was added as a quick fix to #37.
        # Even though it might be strictly necessary, there should be a tested backup
        # solution for it Chords to not work properly due to failed to fail processings
490
491
492
493
494
495
496
497
498
499
        uploads = Upload.objects(status__in=[PENDING, RUNNING])
        for upload in uploads:
            completed = upload.processed_calcs
            total = upload.total
            pending = upload.pending_calcs

            if completed + pending == total:
                time.sleep(2)
                if pending == upload.pending_calcs:
                    Calc.objects(upload_id=upload.upload_id, status=PENDING).delete()