Add mvp for plagiarism, needs testing on paid account

submissions/exceptions.py

@@ -12,3 +12,7 @@ class CycleUpdateDeadlineError(BaseCustomException):
 
 class InvalidReportVettingValue(BaseCustomException):
     pass
+
+
+class ArxivPDFNotFound(Exception):
+    pass

submissions/forms.py

 from django import forms
+from django.conf import settings
 from django.contrib.auth.models import Group
 from django.db import transaction
 from django.utils import timezone
@@ -10,17 +11,18 @@ from .constants import ASSIGNMENT_BOOL, ASSIGNMENT_REFUSAL_REASONS, STATUS_RESUB
                        STATUS_REJECTED, STATUS_REJECTED_VISIBLE, STATUS_RESUBMISSION_INCOMING,\
                        STATUS_DRAFT, STATUS_UNVETTED, REPORT_ACTION_ACCEPT, REPORT_ACTION_REFUSE,\
                        STATUS_VETTED
-from .exceptions import InvalidReportVettingValue
-from .models import Submission, RefereeInvitation, Report, EICRecommendation, EditorialAssignment
+from . import exceptions, helpers
+from .models import Submission, RefereeInvitation, Report, EICRecommendation, EditorialAssignment,\
+                    iThenticateReport
 
 from scipost.constants import SCIPOST_SUBJECT_AREAS
 from scipost.services import ArxivCaller
 from scipost.models import Contributor
+import strings
 
 from crispy_forms.helper import FormHelper
 from crispy_forms.layout import Layout, Div, Field, HTML, Submit
-
-import strings
+import iThenticate
 
 
 class SubmissionSearchForm(forms.Form):
@@ -554,7 +556,7 @@ class VetReportForm(forms.Form):
             # The report is rejected
             report.status = self.cleaned_data['refusal_reason']
         else:
-            raise InvalidReportVettingValue(self.cleaned_data['action_option'])
+            raise exceptions.InvalidReportVettingValue(self.cleaned_data['action_option'])
         report.save()
         return report
 
@@ -643,3 +645,114 @@ class SubmissionCycleChoiceForm(forms.ModelForm):
         other_submission = self.instance.other_versions.first()
         if other_submission:
             self.fields['referees_reinvite'].queryset = other_submission.referee_invitations.all()
+
+
+class iThenticateReportForm(forms.ModelForm):
+    class Meta:
+        model = iThenticateReport
+        fields = []
+
+    def __init__(self, submission, *args, **kwargs):
+        self.submission = submission
+        super().__init__(*args, **kwargs)
+
+        if kwargs.get('files', {}).get('file'):
+            # Add file field if file data is coming in!
+            self.fields['file'] = forms.FileField()
+
+    def clean(self):
+        cleaned_data = super().clean()
+        doc_id = self.instance.doc_id
+        if not doc_id and not self.fields.get('file'):
+            try:
+                cleaned_data['document'] = helpers.retrieve_pdf_from_arxiv(
+                                        self.submission.arxiv_identifier_w_vn_nr)
+            except exceptions.ArxivPDFNotFound:
+                self.add_error(None, ('The pdf could not be found at arXiv.'
+                                      ' Please upload the pdf manually.'))
+                self.fields['file'] = forms.FileField()
+        elif not doc_id and cleaned_data.get('file'):
+            cleaned_data['document'] = cleaned_data['file']
+        elif doc_id:
+            self.document_id = doc_id
+
+        # Login client to append login-check to form
+        self.client = self.get_client()
+
+        # Document (id) is found
+        if cleaned_data.get('document'):
+            self.document = cleaned_data['document']
+            self.response = self.call_ithenticate()
+        elif hasattr(self, 'document_id'):
+            self.response = self.call_ithenticate()
+
+        if self.response:
+            return cleaned_data
+        # Don't return anything as someone submitted invalid data for the form at this point!
+        return None
+
+    def save(self, *args, **kwargs):
+        if self.instance:
+            report = self.instance
+        else:
+            report = iThenticateReport.objects.get_or_create(doc_id=self.response['data']['id'])
+        report.submission = self.submission
+        report.uploaded_time = data['uploaded_time']
+        report.processed_time = data['processed_time']
+        report.percent_match = data['percent_match']
+        report.save()
+        return report
+
+    def call_ithenticate(self):
+        if hasattr(self, 'document_id'):
+            # Update iThenticate status
+            return self.update_status()
+        elif hasattr(self, 'document'):
+            # Upload iThenticate document first time
+            return self.upload_document()
+
+    def get_client(self):
+        client = iThenticate.API.Client(settings.ITHENTICATE_USERNAME,
+                                        settings.ITHENTICATE_PASSWORD)
+        if client.login():
+            return client
+        self.add_error(None, "Failed to login to iThenticate.")
+        return None
+
+    def update_status(self):
+        client = self.client
+        response = client.documents.get(self.document_id)
+        if response['status'] == 200:
+            return response['data']
+        self.add_error(None, "Updating failed. iThenticate didn't return valid data [1]")
+        self.add_error(None, client.messages[0])
+        return None
+
+    def upload_document(self):
+        client = self.client
+
+        # Get first folder available
+        # TODO: Fix this ugly piece of crap
+        folders = client.folders.all()
+        if folders['status'] == 200:
+            folder_id = folders['data'][0]['id']
+        else:
+            self.add_error(None, "Uploading failed. iThenticate didn't return valid data [2]")
+            self.add_error(None, client.messages[0])
+
+        # Finally, upload the file
+        author = self.submission.authors.first()
+        response = client.documents.add(
+            self.document.read(),
+            folder_id,
+            author.user.first_name,
+            author.user.last_name,
+            self.submission.title,
+        )
+
+        if response['status'] == 200:
+            return response['data']
+
+        self.add_error(None, "Updating failed. iThenticate didn't return valid data [3]")
+        self.add_error(None, client.messages[0])
+        return None

submissions/helpers.py

+import requests
+
+from .exceptions import ArxivPDFNotFound
+
+
+def retrieve_pdf_from_arxiv(arxiv_id):
+    """
+    Try to download the pdf as bytes object from arXiv for a certain arXiv Identifier.
+    Raise ArxivPDFNotFound instead.
+
+    :arxiv_id: Arxiv Identifier with or without (takes latest version instead) version number
+    """
+    path_to_pdf = 'https://arxiv.org/pdf/{arxiv_id}.pdf'.format(arxiv_id=arxiv_id)
+    response = requests.get(path_to_pdf)
+    if response.status_code != 200:
+        raise ArxivPDFNotFound('No pdf found on arXiv.')
+    return response.content

submissions/mixins.py

+from django.urls import NoReverseMatch
 from django.contrib.auth.mixins import PermissionRequiredMixin
+from django.views.generic.list import ListView
 
 from .models import Submission
 
@@ -20,7 +22,17 @@ class FriendlyPermissionMixin(PermissionRequiredMixin):
         return super().dispatch(request, *args, **kwargs)
 
 
-class SubmissionAdminViewMixin(FriendlyPermissionMixin):
+class SubmissionFormViewMixin:
+    def get_form_kwargs(self):
+        """
+        Ideally all ModelForms on Submission-related objects have a required argument `submission`.
+        """
+        kwargs = super().get_form_kwargs()
+        kwargs['submission'] = self._original_submission
+        return kwargs
+
+
+class SubmissionAdminViewMixin(FriendlyPermissionMixin, SubmissionFormViewMixin):
     """
     This mixin will provide all basic methods and checks required for Submission
     administrational actions regarding Submissions.
@@ -38,7 +50,46 @@ class SubmissionAdminViewMixin(FriendlyPermissionMixin):
         return not self.editorial_page
 
     def get_queryset(self):
+        """
+        Return either of two sets of Submissions, with an author filter for the current user.
+
+        This method is used in all Class-Based-Views. However, if one overwrites either one of the
+         - get_object()
+         - get_queryset()
+        methods, please don't forget to call super().method_name() to not remove this filter!
+        """
         qs = super().get_queryset()
         if self.pool:
             return qs.get_pool(self.request.user)
         return qs.filter_editorial_page(self.request.user)
+
+    def get_object(self):
+        """
+        Save the original Submission instance for performance reasons to the view,
+        which may be used in get_context_data().
+        """
+        obj = super().get_object()
+        self.submission = obj
+        return obj
+
+    def get_context_data(self, *args, **kwargs):
+        """
+        If the main object in a DetailView is not a Submission instance, it will be lost.
+        Here, explicitly save the Submission instance to the context data.
+        """
+        ctx = super().get_context_data(*args, **kwargs)
+
+        if not ctx.get('submission') and not isinstance(self, ListView):
+            # Call parent get_object() to explicitly save the submission which is related
+            # to the view's main object.
+            ctx['submission'] = self._original_submission
+        return ctx
+
+    @property
+    def _original_submission(self):
+        if hasattr(self, 'submission'):
+            return self.submission
+        obj = super().get_object()
+        if isinstance(obj, Submission):
+            return obj
+        return None

submissions/models.py

@@ -506,6 +506,11 @@ class iThenticateReport(TimeStampedModel):
     doc_id = models.IntegerField(primary_key=True)
     percent_match = models.IntegerField(null=True, blank=True)
 
+    def get_absolute_url(self):
+        return reverse('submissions:plagiarism', kwargs={
+                        'arxiv_identifier_w_vn_nr':
+                        self.to_submission.arxiv_identifier_w_vn_nr})
+
     @property
     def score(self):
         return self.percent_match

submissions/templates/submissions/admin/plagiarism_report.html

 {% extends 'scipost/_personal_page_base.html' %}
 
+{% load bootstrap %}
+
 {% block pagetitle %}: plagiarism report ({{ submission.arxiv_identifier_w_vn_nr }}){% endblock pagetitle %}
 
 {% block breadcrumb_items %}
@@ -18,8 +20,9 @@
         No Plagiarism Report found.
     {% endif %}
 
-    <form method="post" class="mt-3">
+    <form method="post" class="mt-3" enctype="multipart/form-data">
         {% csrf_token %}
+        {{form|bootstrap}}
         <input type="submit" class="btn btn-primary" value="{% if submission.plagiarism_report %}Update report status{% else %}Submit submission for plagiarism check{% endif %}">
     </form>
 

submissions/views.py

 import datetime
 import feedparser
 
-from django.conf import settings
 from django.contrib import messages
 from django.contrib.auth.decorators import login_required, permission_required
 from django.contrib.auth.models import Group
@@ -13,12 +12,11 @@ from django.template import Template, Context
 from django.utils import timezone
 from django.utils.decorators import method_decorator
 from django.views.generic.detail import DetailView
-from django.views.generic.edit import CreateView
+from django.views.generic.edit import CreateView, UpdateView
 from django.views.generic.list import ListView
 
 from guardian.decorators import permission_required_or_403
 from guardian.shortcuts import assign_perm, get_objects_for_user
-import iThenticate
 
 from .constants import SUBMISSION_STATUS_VOTING_DEPRECATED, STATUS_VETTED, STATUS_EIC_ASSIGNED,\
                        SUBMISSION_STATUS_PUBLICLY_INVISIBLE, SUBMISSION_STATUS, ED_COMM_CHOICES,\
@@ -31,7 +29,8 @@ from .forms import SubmissionIdentifierForm, RequestSubmissionForm, SubmissionSe
                    SetRefereeingDeadlineForm, RefereeSelectForm, RefereeRecruitmentForm,\
                    ConsiderRefereeInvitationForm, EditorialCommunicationForm,\
                    EICRecommendationForm, ReportForm, VetReportForm, VotingEligibilityForm,\
-                   SubmissionCycleChoiceForm, ReportPDFForm, SubmissionReportsForm
+                   SubmissionCycleChoiceForm, ReportPDFForm, SubmissionReportsForm,\
+                   iThenticateReportForm
 from .utils import SubmissionUtils
 
 from scipost.forms import ModifyPersonalMessageForm, RemarkForm
@@ -90,6 +89,7 @@ class RequestSubmission(CreateView):
             messages.warning(self.request, *error_messages)
         return super().form_invalid(form)
 
+
 @login_required
 @permission_required('scipost.can_submit_manuscript', raise_exception=True)
 def prefill_using_arxiv_identifier(request):
@@ -1472,23 +1472,20 @@ class EditorialSummaryView(SubmissionAdminViewMixin, ListView):
                                      .get(arxiv_identifier_w_vn_nr=arxiv_id))
         except (AssertionError, Submission.DoesNotExist):
             context['submission'] = None
-            context['latest_events'] = SubmissionEvent.objects.for_eic()#.last_hours()
+            context['latest_events'] = SubmissionEvent.objects.for_eic().last_hours()
         return context
 
 
-class PlagiarismView(SubmissionAdminViewMixin, DetailView):
+class PlagiarismView(SubmissionAdminViewMixin, UpdateView):
     permission_required = 'scipost.can_do_plagiarism_checks'
     template_name = 'submissions/admin/plagiarism_report.html'
     editorial_page = True
+    success_url = reverse_lazy('submissions:plagiarism')
+    form_class = iThenticateReportForm
 
-    def post(self, request, *args, **kwargs):
-        client = iThenticate.API.Client(settings.ITHENTICATE_USERNAME,
-                                        settings.ITHENTICATE_PASSWORD)
-        submission = self.get_object()
-        if submission.plagiarism_report:
-            # Plagiarism Report needs an update
-            client.documents.get()
-        else:
-            # Plagiarism Report needs to be uploaded still
-            client.folders.all()
-        raise NotImplementedError
+    def get_object(self):
+        submission = super().get_object()
+        return submission.plagiarism_report
+
+    # def post(self, request, *args, **kwargs):
+    #     raise NotImplementedError