validation.tasks.ValidateSubmission.mark_model() - Code Metrics - Inspection of "2069987b2b129050867b63cad1489b442b195120" - cnr-ibba/IMAGE-InjectTool - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Branch — master (206998)

by Paolo

created 2019-06-27 14:41 UTC

validation.tasks.ValidateSubmission.mark_model() C

↳ Parent: validation.tasks

Complexity

Conditions

Size

Total Lines	56
Code Lines	32

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	32
dl	0
loc	56
rs	5.9999
c	0
b	0
f	0
cc	10
nop	4

How to fix Long Method Complexity

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Fri Oct  5 11:22:33 2018

@author: Paolo Cozzi <[email protected]>

Useful staff to deal with validation process

"""

import json
import traceback

from collections import Counter
from celery.utils.log import get_task_logger

from django.conf import settings
from django.core.mail import send_mass_mail

from common.constants import (
    READY, ERROR, LOADED, NEED_REVISION, COMPLETED, STATUSES, KNOWN_STATUSES)
from validation.helpers import construct_validation_message
from image.celery import app as celery_app, MyTask
from image_app.helpers import get_admin_emails
from image_app.models import Submission, Sample, Animal
from submissions.helpers import send_message
from validation.models import ValidationSummary

from .models import ValidationResult as ValidationResultModel
from .helpers import MetaDataValidation, OntologyCacheError, RulesetError

# Get an instance of a logger
logger = get_task_logger(__name__)

# get a dictionary from status name (ie {0: 'Waiting'})
key2status = dict([x.value for x in STATUSES])


# A class to deal with validation errors
class ValidationError(Exception):
    pass


class ValidateSubmission(object):
    """
    An helper class for submission task, useful to pass parameters like
    submission data between tasks"""

    # define my class attributes
    def __init__(self, submission_obj, ruleset):
        # track submission object
        self.submission_obj = submission_obj

        # track ruleset
        self.ruleset = ruleset

        # collect all unique messages for samples and animals
        self.messages_animals = Counter()
        self.messages_samples = Counter()

        # track global statuses for animals and samples
        # Don't set keys: if you take a key which doesn't exists, you will
        # get 0 instead of key errors. This is how Counter differ from a
        # default dictionary object
        self.statuses_animals = Counter()
        self.statuses_samples = Counter()

    def check_valid_statuses(self):
        """Check if validation return with an unsupported status message"""

        # test for keys in model_statuses
        for key in self.statuses_animals.keys():
            if key not in KNOWN_STATUSES:
                logger.error("Unsupported status '%s' from validation" % key)
                return False

        for key in self.statuses_samples.keys():
            if key not in KNOWN_STATUSES:
                logger.error("Unsupported status '%s' from validation" % key)
                return False

        # if I arrive here, all validation statuses are handled
        return True

    def __has_key_in_rules(self, key):
        """Generic function to test errors in validation rules"""

        if (self.statuses_animals[key] > 0 or
                self.statuses_samples[key] > 0):
            return True

        else:
            return False

    def has_errors_in_rules(self):
        "Return True if there is any errors in validation rules"""

        return self.__has_key_in_rules('Error')

    def has_warnings_in_rules(self):
        "Return True if there is any warnings in validation rules"""

        return self.__has_key_in_rules('Warning')

    def has_errors_in_json(self):
        "Return True if there is any error in JSON"""

        return self.__has_key_in_rules('JSON')

    def validate_model(self, model):
        logger.debug("Validating %s" % (model))

        # thsi could be animal or sample
        if isinstance(model, Sample):
            model_statuses = self.statuses_samples

        elif isinstance(model, Animal):
            model_statuses = self.statuses_animals

        # get data in biosample format
        data = model.to_biosample()

        # input is a list object
        usi_result = self.ruleset.check_usi_structure([data])

        # if I have errors here, JSON isn't valid: this is not an error
        # on user's data but on InjectTool itself
        if len(usi_result) > 0:
            # update counter for JSON
            model_statuses.update({'JSON': len(usi_result)})

            model_statuses.update(['Issues', 'Known'])

            # update model results
            self.mark_model(model, usi_result, NEED_REVISION)

            # It make no sense continue validation since JSON is wrong
            return

        # no check_duplicates: it checks against alias (that is a pk)
        # HINT: improve check_duplicates or implement database constraints

        # check against image metadata
        ruleset_result = self.ruleset.validate(data)

        # update status and track data in a overall variable
        self.update_statuses(model_statuses, model, ruleset_result)

    # inspired from validation.deal_with_validation_results
    def update_statuses(self, model_statuses, model, result):
        # get overall status (ie Pass, Error)
        overall = result.get_overall_status()

        # set model as valid even if has some warnings
        if overall in ["Pass", "Warning"]:
            self.mark_model(model, result, READY)

        else:
            model_statuses['Issues'] += 1
            self.mark_model(model, result, NEED_REVISION)

        # update a collections.Counter objects by key
        model_statuses.update({overall})
        model_statuses['Known'] += 1

    def mark_model(self, model, result, status):
        """Set status to a model and instantiate a ValidationResult obj"""

        if isinstance(result, list):
            messages = result
            comparable_messages = result
            overall_status = "Wrong JSON structure"

        else:
            messages = result.get_messages()

            # get comparable messages for batch update
            comparable_messages = list()
            for result_set in result.result_set:
                comparable_messages.append(result_set.get_comparable_str())
            overall_status = result.get_overall_status()

        # Save all messages for validation summary
        if isinstance(model, Sample):
            for message in comparable_messages:
                # messages_samples iss a counter object
                self.messages_samples.update({message})

        # is as an animal object
        elif isinstance(model, Animal):
            for message in comparable_messages:
                self.messages_animals.update({message})

        # get a validation result model or create a new one
        if hasattr(model.name, 'validationresult'):
            validationresult = model.name.validationresult

        else:
            validationresult = ValidationResultModel()
            model.name.validationresult = validationresult

        # setting valdiationtool results and save
        validationresult.messages = messages
        validationresult.status = overall_status
        validationresult.save()

        # ok, don't update Name statuses for submitted objects which
        # already are in biosamples and pass validation
        if model.name.status == COMPLETED and status == READY:
            logger.debug(
                "Ignoring %s: status was '%s' and validation is OK" % (
                    model, key2status[model.name.status]))

        else:
            logger.debug(
                "Marking %s with '%s' status (%s)" % (
                    model, key2status[status], messages))

            # update model status and save
            model.name.status = status
            model.name.save()

    def create_validation_summary(self):
        """
        This function will create ValidationSummary object that will be used
        on validation_summary view
        """

        for model_type in ['animal', 'sample']:
            summary_obj, created = ValidationSummary.objects.get_or_create(
                submission=self.submission_obj, type=model_type)

            if created:
                logger.debug(
                    "Created %s validationSummary for %s" % (
                        model_type, self.submission_obj))

            # reset all_count
            summary_obj.reset_all_count()

            if model_type == 'animal':
                messages = self.messages_animals
                model_statuses = self.statuses_animals

            # Im cycling with animal and sample type
            else:
                messages = self.messages_samples
                model_statuses = self.statuses_samples

            summary_obj.submission = self.submission_obj

            # they are counter object, so no Keyerror and returns 0
            summary_obj.pass_count = model_statuses['Pass']
            summary_obj.warning_count = model_statuses['Warning']
            summary_obj.error_count = model_statuses['Error']
            summary_obj.json_count = model_statuses['JSON']
            summary_obj.issues_count = model_statuses['Issues']
            summary_obj.validation_known_count = model_statuses['Known']

            validation_messages = list()

            for message, count in messages.items():
                validation_messages.append({
                    'message': message,
                    'count': count
                })

            summary_obj.messages = validation_messages
            summary_obj.type = model_type
            summary_obj.save()

        logger.debug(
            "Results for submission %s: animals - %s, samples - %s" % (
                self.submission_obj,
                dict(self.statuses_animals),
                dict(self.statuses_samples))
        )


class ValidateTask(MyTask):
    name = "Validate Submission"
    description = """Validate submission data against IMAGE rules"""

    # http://docs.celeryproject.org/en/latest/userguide/tasks.html#instantiation
    # A task is not instantiated for every request, but is registered in
    # the task registry as a global instance. This means that the __init__
    # constructor will only be called once per process, and that the
    # task class is semantically closer to an Actor. if you have a task and
    # you route every request to the same process, then it will keep state
    # between requests. This can also be useful to cache resources, For
    # example, a base Task class that caches a database connection

    # TODO: extract a generic send_message for all modules which need it
    def send_message(self, submission_obj):
        """
        Update submission.status and submission message using django
        channels

        Args:
            submission_obj (image_app.models.Submission): an UID submission
            object
        """

        send_message(
            submission_obj,
            validation_message=construct_validation_message(submission_obj))

    def __generic_error_report(
            self, submission_obj, status, message, notify_admins=False):
        """
        Generic report for updating submission objects and send email after
        an exception is called

        Args:
            submission_obj (image_app.models.Submission): an UID submission
            object
            status (int): a :py:class:`common.constants.STATUSES` object
            message (str): a text object
            notify_admins (bool): send mail to the admins or not
        """

        # mark submission with its status
        submission_obj.status = status
        submission_obj.message = message
        submission_obj.save()

        self.send_message(submission_obj)

        # get exception info
        einfo = traceback.format_exc()

        # send a mail to the user with the stacktrace (einfo)
        email_subject = "Error in IMAGE Validation: %s" % (message)
        email_message = (
            "Something goes wrong with validation. Please report "
            "this to InjectTool team\n\n %s" % str(einfo))

        submission_obj.owner.email_user(
            email_subject,
            email_message,
        )

        # TODO: should this be a common.helpers method?
        if notify_admins:
            # submit mail to admins
            datatuple = (
                email_subject,
                email_message,
                settings.DEFAULT_FROM_EMAIL,
                get_admin_emails())

            send_mass_mail((datatuple, ))

    # Ovverride default on failure method
    # This is not a failed validation for a wrong value, this is an
    # error in task that mean an error in coding
    def on_failure(self, exc, task_id, args, kwargs, einfo):
        logger.error('{0!r} failed: {1!r}'.format(task_id, exc))

        # define message
        message = "Unknown error in validation - %s" % str(exc)

        # get submissio object
        submission_obj = Submission.objects.get(pk=args[0])

        # call generic report which update submission and send email
        self.__generic_error_report(
            submission_obj, ERROR, message, notify_admins=True)

        # returns None: this task will have the ERROR status

    # TODO: define a method to inform user for error in validation (Task run
    # with success but errors in data)

    def temporary_error_report(self, exc, submission_obj):
        """
        Deal with known issues in validation task. Notify the user using
        email and set status as READY in order to recall this task

        Args:
            exc (Exception): an py:exc`Exception` object
            submission_obj (image_app.models.Submission): an UID submission
            object

        Return
            str: "success" since this task is correctly managed
        """

        logger.error("Error in validation: %s" % exc)

        message = "Errors in EBI API endpoints. Please try again later"
        logger.error(message)

        # call generic report which update submission and send email
        self.__generic_error_report(submission_obj, LOADED, message)

        return "success"

    def ruleset_error_report(self, exc, submission_obj):
        """
        Deal with ruleset issue in validation task. Notify the user using
        email and set status as ERROR, since he can't do anything without
        admin intervention

        Args:
            exc (Exception): an py:exc`Exception` object
            submission_obj (image_app.models.Submission): an UID submission
            object

        Return
            str: "success" since this task is correctly managed
        """

        logger.error("Error ruleset: %s" % exc)

        message = (
            "Error in IMAGE-metadata ruleset. Please inform InjectTool team")
        logger.error(message)

        # call generic report which update submission and send email
        self.__generic_error_report(
            submission_obj, ERROR, message, notify_admins=True)

        return "success"

    def run(self, submission_id):
        """a function to perform validation steps"""

        logger.info("Validate Submission started")

        # get submissio object
        submission_obj = Submission.objects.get(pk=submission_id)

        # read rules when task starts. Model issues when starting
        # OntologyCache at start
        try:
            self.ruleset = MetaDataValidation()

        except OntologyCacheError as exc:
            return self.temporary_error_report(exc, submission_obj)

        except RulesetError as exc:
            return self.ruleset_error_report(exc, submission_obj)

        # get a submission data helper instance
        validate_submission = ValidateSubmission(submission_obj, self.ruleset)

        try:
            for animal in Animal.objects.filter(
                    name__submission=submission_obj).order_by('id'):
                validate_submission.validate_model(animal)

            for sample in Sample.objects.filter(
                    name__submission=submission_obj).order_by('id'):
                validate_submission.validate_model(sample)

        # TODO: errors in validation should raise custom exception
        except json.decoder.JSONDecodeError as exc:
            return self.temporary_error_report(exc, submission_obj)

        except Exception as exc:
            raise self.retry(exc=exc)

        # if error messages changes in IMAGE-ValidationTool, all this
        # stuff isn't valid and I throw an exception

        if not validate_submission.check_valid_statuses():
            message = (
                "Unsupported validation status for submission %s" % (
                    submission_obj))

            # debug: print error in log
            logger.error(message)

            # create validation summary
            validate_submission.create_validation_summary()

            # mark submission with ERROR (this is not related to user data)
            # calling the appropriate method passing ERROR as status
            self.submission_fail(submission_obj, message, status=ERROR)

            # raise an exception since is an InjectTool issue
            raise ValidationError(message)

        # If I have any error in JSON is a problem of injectool
        if validate_submission.has_errors_in_json():
            # create validation summary
            validate_submission.create_validation_summary()

            # mark submission with NEED_REVISION
            self.submission_fail(submission_obj, "Wrong JSON structure")

            # debug
            logger.warning(
                "Wrong JSON structure for submission %s" % (submission_obj))

        # set a proper value for status (READY or NEED_REVISION)
        # If I will found any error or warning, I will
        # return a message and I will set NEED_REVISION
        elif validate_submission.has_errors_in_rules():
            # create validation summary
            validate_submission.create_validation_summary()

            message = (
                "Error in metadata. Need revisions before submit")

            # mark submission with NEED_REVISION
            self.submission_fail(submission_obj, message)

            logger.warning(
                "Error in metadata for submission %s" % (submission_obj))

        # WOW: I can submit those data
        elif validate_submission.has_warnings_in_rules():
            # create validation summary
            validate_submission.create_validation_summary()

            message = "Submission validated with some warnings"

            # mark submission with READY status
            self.submission_ready(submission_obj, message)

            logger.info(
                "Submission %s validated with some warning" % (submission_obj))

        else:
            # create validation summary
            validate_submission.create_validation_summary()

            message = "Submission validated with success"

            # mark submission with READY status
            self.submission_ready(submission_obj, message)

            logger.info(
                "Submission %s validated with success" % (submission_obj))

        logger.info("Validate Submission completed")

        return "success"

    def __mark_submission(self, submission_obj, message, status):
        """Mark submission with status and message"""

        submission_obj.status = status
        submission_obj.message = message
        submission_obj.save()

        self.send_message(submission_obj)

    def submission_fail(self, submission_obj, message, status=NEED_REVISION):
        """Mark a submission with NEED_REVISION status"""

        # ovverride message
        message = ("Validation got errors: %s" % (message))
        self.__mark_submission(submission_obj, message, status)

    def submission_ready(self, submission_obj, message):
        """Mark a submission with READY status"""

        self.__mark_submission(submission_obj, message, READY)


# register explicitly tasks
# https://github.com/celery/celery/issues/3744#issuecomment-271366923
celery_app.tasks.register(ValidateTask)


1			#!/usr/bin/env python3
2			# -- coding: utf-8 --
3			"""
4			Created on Fri Oct 5 11:22:33 2018
5
6			@author: Paolo Cozzi <[email protected]>
7
8			Useful staff to deal with validation process
9
10			"""
11
12			import json
13			import traceback
14
15			from collections import Counter
16			from celery.utils.log import get_task_logger
17
18			from django.conf import settings
19			from django.core.mail import send_mass_mail
20
21			from common.constants import (
22			READY, ERROR, LOADED, NEED_REVISION, COMPLETED, STATUSES, KNOWN_STATUSES)
23			from validation.helpers import construct_validation_message
24			from image.celery import app as celery_app, MyTask
25			from image_app.helpers import get_admin_emails
26			from image_app.models import Submission, Sample, Animal
27			from submissions.helpers import send_message
28			from validation.models import ValidationSummary
29
30			from .models import ValidationResult as ValidationResultModel
31			from .helpers import MetaDataValidation, OntologyCacheError, RulesetError
32
33			# Get an instance of a logger
34			logger = get_task_logger(__name__)
35
36			# get a dictionary from status name (ie {0: 'Waiting'})
37			key2status = dict([x.value for x in STATUSES])
38
39
40			# A class to deal with validation errors
41			class ValidationError(Exception):
42			pass
43
44
45			class ValidateSubmission(object):
46			"""
47			An helper class for submission task, useful to pass parameters like
48			submission data between tasks"""
49
50			# define my class attributes
51			def __init__(self, submission_obj, ruleset):
52			# track submission object
53			self.submission_obj = submission_obj
54
55			# track ruleset
56			self.ruleset = ruleset
57
58			# collect all unique messages for samples and animals
59			self.messages_animals = Counter()
60			self.messages_samples = Counter()
61
62			# track global statuses for animals and samples
63			# Don't set keys: if you take a key which doesn't exists, you will
64			# get 0 instead of key errors. This is how Counter differ from a
65			# default dictionary object
66			self.statuses_animals = Counter()
67			self.statuses_samples = Counter()
68
69			def check_valid_statuses(self):
70			"""Check if validation return with an unsupported status message"""
71
72			# test for keys in model_statuses
73			for key in self.statuses_animals.keys():
74			if key not in KNOWN_STATUSES:
75			logger.error("Unsupported status '%s' from validation" % key)
76			return False
77
78			for key in self.statuses_samples.keys():
79			if key not in KNOWN_STATUSES:
80			logger.error("Unsupported status '%s' from validation" % key)
81			return False
82
83			# if I arrive here, all validation statuses are handled
84			return True
85
86			def __has_key_in_rules(self, key):
87			"""Generic function to test errors in validation rules"""
88
89			if (self.statuses_animals[key] > 0 or
90			self.statuses_samples[key] > 0):
91			return True
92
93			else:
94			return False
95
96			def has_errors_in_rules(self):
97			"Return True if there is any errors in validation rules"""
98
99			return self.__has_key_in_rules('Error')
100
101			def has_warnings_in_rules(self):
102			"Return True if there is any warnings in validation rules"""
103
104			return self.__has_key_in_rules('Warning')
105
106			def has_errors_in_json(self):
107			"Return True if there is any error in JSON"""
108
109			return self.__has_key_in_rules('JSON')
110
111			def validate_model(self, model):
112			logger.debug("Validating %s" % (model))
113
114			# thsi could be animal or sample
115			if isinstance(model, Sample):
116			model_statuses = self.statuses_samples
117
118			elif isinstance(model, Animal):
119			model_statuses = self.statuses_animals
120
121			# get data in biosample format
122			data = model.to_biosample()
123
124			# input is a list object
125			usi_result = self.ruleset.check_usi_structure([data])
126
127			# if I have errors here, JSON isn't valid: this is not an error
128			# on user's data but on InjectTool itself
129			if len(usi_result) > 0:
130			# update counter for JSON
131			model_statuses.update({'JSON': len(usi_result)})
			0 ignored issues – show introduced 2019-06-21 15:24 UTC by Report Bug Copy Issue Report The variable `model_statuses` does not seem to be defined for all execution paths. Loading history...
132			model_statuses.update(['Issues', 'Known'])
133
134			# update model results
135			self.mark_model(model, usi_result, NEED_REVISION)
136
137			# It make no sense continue validation since JSON is wrong
138			return
139
140			# no check_duplicates: it checks against alias (that is a pk)
141			# HINT: improve check_duplicates or implement database constraints
142
143			# check against image metadata
144			ruleset_result = self.ruleset.validate(data)
145
146			# update status and track data in a overall variable
147			self.update_statuses(model_statuses, model, ruleset_result)
148
149			# inspired from validation.deal_with_validation_results
150			def update_statuses(self, model_statuses, model, result):
151			# get overall status (ie Pass, Error)
152			overall = result.get_overall_status()
153
154			# set model as valid even if has some warnings
155			if overall in ["Pass", "Warning"]:
156			self.mark_model(model, result, READY)
157
158			else:
159			model_statuses['Issues'] += 1
160			self.mark_model(model, result, NEED_REVISION)
161
162			# update a collections.Counter objects by key
163			model_statuses.update({overall})
164			model_statuses['Known'] += 1
165
166			def mark_model(self, model, result, status):
167			"""Set status to a model and instantiate a ValidationResult obj"""
168
169			if isinstance(result, list):
170			messages = result
171			comparable_messages = result
172			overall_status = "Wrong JSON structure"
173
174			else:
175			messages = result.get_messages()
176
177			# get comparable messages for batch update
178			comparable_messages = list()
179			for result_set in result.result_set:
180			comparable_messages.append(result_set.get_comparable_str())
181			overall_status = result.get_overall_status()
182
183			# Save all messages for validation summary
184			if isinstance(model, Sample):
185			for message in comparable_messages:
186			# messages_samples iss a counter object
187			self.messages_samples.update({message})
188
189			# is as an animal object
190			elif isinstance(model, Animal):
191			for message in comparable_messages:
192			self.messages_animals.update({message})
193
194			# get a validation result model or create a new one
195			if hasattr(model.name, 'validationresult'):
196			validationresult = model.name.validationresult
197
198			else:
199			validationresult = ValidationResultModel()
200			model.name.validationresult = validationresult
201
202			# setting valdiationtool results and save
203			validationresult.messages = messages
204			validationresult.status = overall_status
205			validationresult.save()
206
207			# ok, don't update Name statuses for submitted objects which
208			# already are in biosamples and pass validation
209			if model.name.status == COMPLETED and status == READY:
210			logger.debug(
211			"Ignoring %s: status was '%s' and validation is OK" % (
212			model, key2status[model.name.status]))
213
214			else:
215			logger.debug(
216			"Marking %s with '%s' status (%s)" % (
217			model, key2status[status], messages))
218
219			# update model status and save
220			model.name.status = status
221			model.name.save()
222
223			def create_validation_summary(self):
224			"""
225			This function will create ValidationSummary object that will be used
226			on validation_summary view
227			"""
228
229			for model_type in ['animal', 'sample']:
230			summary_obj, created = ValidationSummary.objects.get_or_create(
231			submission=self.submission_obj, type=model_type)
232
233			if created:
234			logger.debug(
235			"Created %s validationSummary for %s" % (
236			model_type, self.submission_obj))
237
238			# reset all_count
239			summary_obj.reset_all_count()
240
241			if model_type == 'animal':
242			messages = self.messages_animals
243			model_statuses = self.statuses_animals
244
245			# Im cycling with animal and sample type
246			else:
247			messages = self.messages_samples
248			model_statuses = self.statuses_samples
249
250			summary_obj.submission = self.submission_obj
251
252			# they are counter object, so no Keyerror and returns 0
253			summary_obj.pass_count = model_statuses['Pass']
254			summary_obj.warning_count = model_statuses['Warning']
255			summary_obj.error_count = model_statuses['Error']
256			summary_obj.json_count = model_statuses['JSON']
257			summary_obj.issues_count = model_statuses['Issues']
258			summary_obj.validation_known_count = model_statuses['Known']
259
260			validation_messages = list()
261
262			for message, count in messages.items():
263			validation_messages.append({
264			'message': message,
265			'count': count
266			})
267
268			summary_obj.messages = validation_messages
269			summary_obj.type = model_type
270			summary_obj.save()
271
272			logger.debug(
273			"Results for submission %s: animals - %s, samples - %s" % (
274			self.submission_obj,
275			dict(self.statuses_animals),
276			dict(self.statuses_samples))
277			)
278
279
280			class ValidateTask(MyTask):
281			name = "Validate Submission"
282			description = """Validate submission data against IMAGE rules"""
283
284			# http://docs.celeryproject.org/en/latest/userguide/tasks.html#instantiation
285			# A task is not instantiated for every request, but is registered in
286			# the task registry as a global instance. This means that the __init__
287			# constructor will only be called once per process, and that the
288			# task class is semantically closer to an Actor. if you have a task and
289			# you route every request to the same process, then it will keep state
290			# between requests. This can also be useful to cache resources, For
291			# example, a base Task class that caches a database connection
292
293			# TODO: extract a generic send_message for all modules which need it
294			def send_message(self, submission_obj):
295			"""
296			Update submission.status and submission message using django
297			channels
298
299			Args:
300			submission_obj (image_app.models.Submission): an UID submission
301			object
302			"""
303
304			send_message(
305			submission_obj,
306			validation_message=construct_validation_message(submission_obj))
307
308			def __generic_error_report(
309			self, submission_obj, status, message, notify_admins=False):
310			"""
311			Generic report for updating submission objects and send email after
312			an exception is called
313
314			Args:
315			submission_obj (image_app.models.Submission): an UID submission
316			object
317			status (int): a :py:class:`common.constants.STATUSES` object
318			message (str): a text object
319			notify_admins (bool): send mail to the admins or not
320			"""
321
322			# mark submission with its status
323			submission_obj.status = status
324			submission_obj.message = message
325			submission_obj.save()
326
327			self.send_message(submission_obj)
328
329			# get exception info
330			einfo = traceback.format_exc()
331
332			# send a mail to the user with the stacktrace (einfo)
333			email_subject = "Error in IMAGE Validation: %s" % (message)
334			email_message = (
335			"Something goes wrong with validation. Please report "
336			"this to InjectTool team\n\n %s" % str(einfo))
337
338			submission_obj.owner.email_user(
339			email_subject,
340			email_message,
341			)
342
343			# TODO: should this be a common.helpers method?
344			if notify_admins:
345			# submit mail to admins
346			datatuple = (
347			email_subject,
348			email_message,
349			settings.DEFAULT_FROM_EMAIL,
350			get_admin_emails())
351
352			send_mass_mail((datatuple, ))
353
354			# Ovverride default on failure method
355			# This is not a failed validation for a wrong value, this is an
356			# error in task that mean an error in coding
357			def on_failure(self, exc, task_id, args, kwargs, einfo):
358			logger.error('{0!r} failed: {1!r}'.format(task_id, exc))
359
360			# define message
361			message = "Unknown error in validation - %s" % str(exc)
362
363			# get submissio object
364			submission_obj = Submission.objects.get(pk=args[0])
365
366			# call generic report which update submission and send email
367			self.__generic_error_report(
368			submission_obj, ERROR, message, notify_admins=True)
369
370			# returns None: this task will have the ERROR status
371
372			# TODO: define a method to inform user for error in validation (Task run
373			# with success but errors in data)
374
375			def temporary_error_report(self, exc, submission_obj):
376			"""
377			Deal with known issues in validation task. Notify the user using
378			email and set status as READY in order to recall this task
379
380			Args:
381			exc (Exception): an py:exc`Exception` object
382			submission_obj (image_app.models.Submission): an UID submission
383			object
384
385			Return
386			str: "success" since this task is correctly managed
387			"""
388
389			logger.error("Error in validation: %s" % exc)
390
391			message = "Errors in EBI API endpoints. Please try again later"
392			logger.error(message)
393
394			# call generic report which update submission and send email
395			self.__generic_error_report(submission_obj, LOADED, message)
396
397			return "success"
398
399			def ruleset_error_report(self, exc, submission_obj):
400			"""
401			Deal with ruleset issue in validation task. Notify the user using
402			email and set status as ERROR, since he can't do anything without
403			admin intervention
404
405			Args:
406			exc (Exception): an py:exc`Exception` object
407			submission_obj (image_app.models.Submission): an UID submission
408			object
409
410			Return
411			str: "success" since this task is correctly managed
412			"""
413
414			logger.error("Error ruleset: %s" % exc)
415
416			message = (
417			"Error in IMAGE-metadata ruleset. Please inform InjectTool team")
418			logger.error(message)
419
420			# call generic report which update submission and send email
421			self.__generic_error_report(
422			submission_obj, ERROR, message, notify_admins=True)
423
424			return "success"
425
426			def run(self, submission_id):
427			"""a function to perform validation steps"""
428
429			logger.info("Validate Submission started")
430
431			# get submissio object
432			submission_obj = Submission.objects.get(pk=submission_id)
433
434			# read rules when task starts. Model issues when starting
435			# OntologyCache at start
436			try:
437			self.ruleset = MetaDataValidation()
438
439			except OntologyCacheError as exc:
440			return self.temporary_error_report(exc, submission_obj)
441
442			except RulesetError as exc:
443			return self.ruleset_error_report(exc, submission_obj)
444
445			# get a submission data helper instance
446			validate_submission = ValidateSubmission(submission_obj, self.ruleset)
447
448			try:
449			for animal in Animal.objects.filter(
450			name__submission=submission_obj).order_by('id'):
451			validate_submission.validate_model(animal)
452
453			for sample in Sample.objects.filter(
454			name__submission=submission_obj).order_by('id'):
455			validate_submission.validate_model(sample)
456
457			# TODO: errors in validation should raise custom exception
458			except json.decoder.JSONDecodeError as exc:
459			return self.temporary_error_report(exc, submission_obj)
460
461			except Exception as exc:
462			raise self.retry(exc=exc)
463
464			# if error messages changes in IMAGE-ValidationTool, all this
465			# stuff isn't valid and I throw an exception
466
467			if not validate_submission.check_valid_statuses():
468			message = (
469			"Unsupported validation status for submission %s" % (
470			submission_obj))
471
472			# debug: print error in log
473			logger.error(message)
474
475			# create validation summary
476			validate_submission.create_validation_summary()
477
478			# mark submission with ERROR (this is not related to user data)
479			# calling the appropriate method passing ERROR as status
480			self.submission_fail(submission_obj, message, status=ERROR)
481
482			# raise an exception since is an InjectTool issue
483			raise ValidationError(message)
484
485			# If I have any error in JSON is a problem of injectool
486			if validate_submission.has_errors_in_json():
487			# create validation summary
488			validate_submission.create_validation_summary()
489
490			# mark submission with NEED_REVISION
491			self.submission_fail(submission_obj, "Wrong JSON structure")
492
493			# debug
494			logger.warning(
495			"Wrong JSON structure for submission %s" % (submission_obj))
496
497			# set a proper value for status (READY or NEED_REVISION)
498			# If I will found any error or warning, I will
499			# return a message and I will set NEED_REVISION
500			elif validate_submission.has_errors_in_rules():
501			# create validation summary
502			validate_submission.create_validation_summary()
503
504			message = (
505			"Error in metadata. Need revisions before submit")
506
507			# mark submission with NEED_REVISION
508			self.submission_fail(submission_obj, message)
509
510			logger.warning(
511			"Error in metadata for submission %s" % (submission_obj))
512
513			# WOW: I can submit those data
514			elif validate_submission.has_warnings_in_rules():
515			# create validation summary
516			validate_submission.create_validation_summary()
517
518			message = "Submission validated with some warnings"
519
520			# mark submission with READY status
521			self.submission_ready(submission_obj, message)
522
523			logger.info(
524			"Submission %s validated with some warning" % (submission_obj))
525
526			else:
527			# create validation summary
528			validate_submission.create_validation_summary()
529
530			message = "Submission validated with success"
531
532			# mark submission with READY status
533			self.submission_ready(submission_obj, message)
534
535			logger.info(
536			"Submission %s validated with success" % (submission_obj))
537
538			logger.info("Validate Submission completed")
539
540			return "success"
541
542			def __mark_submission(self, submission_obj, message, status):
543			"""Mark submission with status and message"""
544
545			submission_obj.status = status
546			submission_obj.message = message
547			submission_obj.save()
548
549			self.send_message(submission_obj)
550
551			def submission_fail(self, submission_obj, message, status=NEED_REVISION):
552			"""Mark a submission with NEED_REVISION status"""
553
554			# ovverride message
555			message = ("Validation got errors: %s" % (message))
556			self.__mark_submission(submission_obj, message, status)
557
558			def submission_ready(self, submission_obj, message):
559			"""Mark a submission with READY status"""
560
561			self.__mark_submission(submission_obj, message, READY)
562
563
564			# register explicitly tasks
565			# https://github.com/celery/celery/issues/3744#issuecomment-271366923
566			celery_app.tasks.register(ValidateTask)
567

cnr-ibba / IMAGE-InjectTool

Branch — master (206998)

validation.tasks.ValidateSubmission.mark_model() C

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Complexity

Duplication Side-by-Side

Filter issues like