diff --git a/scipost_django/submissions/admin.py b/scipost_django/submissions/admin.py
index 5895f5bfc5cd85dd3637ef48f202d0fea9bd6284..70358f049989e07b578852e97c639319ce81f2ff 100644
--- a/scipost_django/submissions/admin.py
+++ b/scipost_django/submissions/admin.py
@@ -112,18 +112,15 @@ class SubmissionAdmin(GuardedModelAdmin):
fieldsets = (
(
None,
- {
- "fields": ("preprint", "title", "abstract"),
- },
- ),
- (
- "Versioning",
{
"fields": (
- "thread_hash",
- "is_current",
- "is_resubmission_of",
- "list_of_changes",
+ "preprint",
+ "title",
+ "abstract",
+ "author_list",
+ "submission_date",
+ "submitted_to",
+ "status",
),
},
),
@@ -148,20 +145,44 @@ class SubmissionAdmin(GuardedModelAdmin):
"classes": ("collapse",),
"fields": (
"submitted_by",
- "author_list",
"authors",
"authors_claims",
"authors_false_claims",
),
},
),
+ (
+ "Versioning",
+ {
+ "classes": ("collapse",),
+ "fields": (
+ "thread_hash",
+ "is_current",
+ "is_resubmission_of",
+ "list_of_changes",
+ ),
+ },
+ ),
+ (
+ "Plagiarism",
+ {
+ "classes": ("collapse",),
+ "fields": ("internal_plagiarism_matches", "plagiarism_report"),
+ }
+ ),
+ (
+ "Conflicts of interest",
+ {
+ "classes": ("collapse",),
+ "fields": ("needs_conflicts_update",),
+ }
+ ),
(
"Refereeing",
{
"classes": ("collapse",),
"fields": (
"editor_in_charge",
- "status",
("visible_public", "visible_pool"),
"refereeing_cycle",
("open_for_commenting", "open_for_reporting"),
@@ -170,9 +191,7 @@ class SubmissionAdmin(GuardedModelAdmin):
"referees_flagged",
"referees_suggested",
"remarks_for_editors",
- "submitted_to",
"pdf_refereeing_pack",
- "plagiarism_report",
"fellows",
),
},
@@ -181,7 +200,7 @@ class SubmissionAdmin(GuardedModelAdmin):
"Meta",
{
"classes": ("collapse",),
- "fields": ("metadata", "submission_date", "needs_conflicts_update"),
+ "fields": ("metadata",),
},
),
)
diff --git a/scipost_django/submissions/api/serializers/submission.py b/scipost_django/submissions/api/serializers/submission.py
index ccbfc8028d870198383cf917639178b31bac346f..a22771e2d1ab3bc395e498808c28374eb1b3c8ef 100644
--- a/scipost_django/submissions/api/serializers/submission.py
+++ b/scipost_django/submissions/api/serializers/submission.py
@@ -84,6 +84,7 @@ class SubmissionPublicSerializer(serializers.ModelSerializer):
class SubmissionPublicSearchSerializer(serializers.ModelSerializer):
identifier = serializers.CharField(source="preprint.identifier_w_vn_nr")
submission_date = serializers.CharField(source="submission_date_ymd")
+ original_submission_date = serializers.CharField(source="original_submission_date_ymd")
url = serializers.URLField(source="get_absolute_url")
class Meta:
diff --git a/scipost_django/submissions/migrations/0114_submission_internal_plagiarism_matches.py b/scipost_django/submissions/migrations/0114_submission_internal_plagiarism_matches.py
new file mode 100644
index 0000000000000000000000000000000000000000..f87bdeaf80510204c88eff284c5e7a3fb7b68bfa
--- /dev/null
+++ b/scipost_django/submissions/migrations/0114_submission_internal_plagiarism_matches.py
@@ -0,0 +1,18 @@
+# Generated by Django 3.2.14 on 2022-11-09 07:21
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+ dependencies = [
+ ('submissions', '0113_alter_refereeinvitation_options'),
+ ]
+
+ operations = [
+ migrations.AddField(
+ model_name='submission',
+ name='internal_plagiarism_matches',
+ field=models.JSONField(blank=True, default=dict, null=True),
+ ),
+ ]
diff --git a/scipost_django/submissions/models/submission.py b/scipost_django/submissions/models/submission.py
index 131f0c9be671762b08eb64aebf810dc71760d37b..ad344779de188e9bff4f2635cadd35a7b89b6743 100644
--- a/scipost_django/submissions/models/submission.py
+++ b/scipost_django/submissions/models/submission.py
@@ -167,6 +167,11 @@ class Submission(models.Model):
blank=True,
related_name="to_submission",
)
+ internal_plagiarism_matches = models.JSONField(
+ default=dict,
+ blank=True,
+ null=True,
+ )
pdf_refereeing_pack = models.FileField(
upload_to="UPLOADS/REFEREE/%Y/%m/", max_length=200, blank=True
@@ -341,6 +346,11 @@ class Submission(models.Model):
"""Return the submission date in YYYY-MM-DD format."""
return self.submission_date.date()
+ @property
+ def original_submission_date_ymd(self):
+ """Return the submission date in YYYY-MM-DD format."""
+ return self.original_submission_date.date()
+
@property
def original_submission_date(self):
"""Return the submission_date of the first Submission in the thread."""
diff --git a/scipost_django/submissions/tasks.py b/scipost_django/submissions/tasks.py
index 35fc0c98cd9b1f5df8bfad223157922e82980551..be9f6c40db43c931108124ed2656e757742d11f2 100644
--- a/scipost_django/submissions/tasks.py
+++ b/scipost_django/submissions/tasks.py
@@ -2,16 +2,22 @@ __copyright__ = "Copyright © Stichting SciPost (SciPost Foundation)"
__license__ = "AGPL v3"
from datetime import timedelta
+from difflib import SequenceMatcher
+
from django.utils import timezone
from SciPost_v1.celery import app
from .models import Submission, EditorialAssignment, RefereeInvitation, Report
+from journals.models import Publication
+
@app.task(bind=True)
def send_editorial_assignment_invitations(self):
- """Send invitation email to 'next EditorialAssignment' in queue."""
+ """
+ Send next queued editorial assignment invitation emails.
+ """
qs = Submission.objects.unassigned().has_editor_invitations_to_be_sent().distinct()
submission_ids = qs.values_list("id", flat=True)
submissions_count = len(submission_ids)
@@ -49,3 +55,75 @@ def submit_submission_document_for_plagiarism(self):
for submission in submission_to_update:
report = submission.plagiarism_report
# do it...
+
+
+@app.task(bind=True)
+def check_for_internal_plagiarism_submission_matches(self, ratio_threshold=0.7):
+ """
+ Check incoming Submissions for internal plagiarism with preexisting Submissions.
+ """
+
+ submissions_to_check = Submission.objects.exclude(
+ internal_plagiarism_matches__has_key="submission_matches"
+ )
+
+ # Check in small batches to not overwhelm the server
+ for sub_to_check in submissions_to_check.all()[:5]:
+ submission_matches = []
+ # check all Submissions which predate, and are not in the thread of the sub
+ for sub in Submission.objects.exclude(
+ thread_hash=sub_to_check.thread_hash
+ ).filter(submission_date__lt=sub_to_check.submission_date):
+ sub_title_sm = SequenceMatcher(None, sub_to_check.title, sub.title)
+ ratio_title = sub_title_sm.ratio()
+ ratio_authors = SequenceMatcher(
+ None, sub_to_check.author_list, sub.author_list
+ ).ratio()
+ ratio_abstract = SequenceMatcher(
+ None, sub_to_check.abstract, sub.abstract
+ ).ratio()
+ if ratio_title > ratio_threshold or ratio_abstract > ratio_threshold:
+ submission_matches.append(
+ {
+ "identifier_w_vn_nr": sub.preprint.identifier_w_vn_nr,
+ "ratio_title": ratio_title,
+ "ratio_authors": ratio_authors,
+ "ratio_abstract": ratio_abstract,
+ }
+ )
+ sub_to_check.internal_plagiarism_matches["submission_matches"] = submission_matches
+ sub_to_check.save()
+
+
+@app.task(bind=True)
+def check_for_internal_plagiarism_publication_matches(self, ratio_threshold=0.7):
+ """
+ Check incoming Submissions for internal plagiarism with existing Publications.
+ """
+
+ submissions_to_check = Submission.objects.exclude(
+ internal_plagiarism_matches__has_key="publication_matches"
+ )
+
+ # Check in small batches to not overwhelm the server
+ for sub_to_check in submissions_to_check.all()[:5]:
+ publication_matches = []
+ for pub in Publication.objects.filter(publication_date__lt=sub_to_check.submission_date):
+ ratio_title = SequenceMatcher(None, sub_to_check.title, pub.title).ratio()
+ ratio_authors = SequenceMatcher(
+ None, sub_to_check.author_list, pub.author_list
+ ).ratio()
+ ratio_abstract = SequenceMatcher(
+ None, sub_to_check.abstract, pub.abstract
+ ).ratio()
+ if ratio_title > ratio_threshold or ratio_abstract > ratio_threshold:
+ publication_matches.append(
+ {
+ "doi_label": pub.doi_label,
+ "ratio_title": ratio_title,
+ "ratio_authors": ratio_authors,
+ "ratio_abstract": ratio_abstract,
+ }
+ )
+ sub_to_check.internal_plagiarism_matches["publication_matches"] = publication_matches
+ sub_to_check.save()
diff --git a/scipost_django/submissions/templates/submissions/admin/plagiarism_internal_check.html b/scipost_django/submissions/templates/submissions/admin/plagiarism_internal_check.html
index b73c9ef8176bd9cb29bd3e762fadbf154d0ea806..39e0f0ea5ea93df95eb6ddc78028d6e70565cafd 100644
--- a/scipost_django/submissions/templates/submissions/admin/plagiarism_internal_check.html
+++ b/scipost_django/submissions/templates/submissions/admin/plagiarism_internal_check.html
@@ -15,6 +15,9 @@
<h3 class="mb-4">by {{ submission.author_list }}</h3>
<h1 class="highlight">Submission matches</h1>
+ {% if not "submission_matches" in submission.internal_plagiarism_matches %}
+ <h2 class="text-danger border border-danger m-2 p-2">This automated internal plagiarism check has not finished running yet; please come back later!</h2>
+ {% endif %}
<table class="table">
<thead>
<tr>
@@ -28,7 +31,9 @@
{% for match in submission_matches %}
<tr>
<td>
- {{ match.submission.preprint.identifier_w_vn_nr }}<br>
+ {{ match.submission.preprint.identifier_w_vn_nr }}
+  <small class="text-muted">Thread hash: {{ match.submission.thread_hash }}</small>
+ <br>
<a href="{{ match.submission.get_absolute_url }}" target="_blank">{{ match.submission.title }}</a><br>
{{ match.submission.author_list }}
</td>
@@ -43,6 +48,9 @@
</table>
<h1 class="highlight">Publication matches</h1>
+ {% if not "publication_matches" in submission.internal_plagiarism_matches %}
+ <h2 class="text-danger border border-danger m-2 p-2">This automated internal plagiarism check has not finished running yet; please come back later!</h2>
+ {% endif %}
<table class="table">
<thead>
<tr>
diff --git a/scipost_django/submissions/templates/submissions/admin/plagiarism_report.html b/scipost_django/submissions/templates/submissions/admin/plagiarism_report.html
index d8db6c77fe6fe1c3748283d863a9e0b9d89bd689..18ec45c65a1406abb5a68c3fea907553a681caf1 100644
--- a/scipost_django/submissions/templates/submissions/admin/plagiarism_report.html
+++ b/scipost_django/submissions/templates/submissions/admin/plagiarism_report.html
@@ -10,7 +10,7 @@
{% endblock %}
{% block content %}
- <h1>Plagiarism Report for <a href="{{ submission.get_absolute_url }}">{{ submission.preprint.identifier_w_vn_nr }}</a></h1>
+ <h1>iThenticate Plagiarism Report for <a href="{{ submission.get_absolute_url }}">{{ submission.preprint.identifier_w_vn_nr }}</a></h1>
<h2>{{ submission.title }}</h2>
<h3 class="mb-4">by {{ submission.author_list }}</h3>
{% if submission.plagiarism_report %}
diff --git a/scipost_django/submissions/templates/submissions/pool/_submission_info_table.html b/scipost_django/submissions/templates/submissions/pool/_submission_info_table.html
index e5f8016c002155f88783554df995d8c2e475a9c4..c96ab3ff7982b0b33a91223833ab6819708dd6ce 100644
--- a/scipost_django/submissions/templates/submissions/pool/_submission_info_table.html
+++ b/scipost_django/submissions/templates/submissions/pool/_submission_info_table.html
@@ -99,15 +99,28 @@
{% if perms.scipost.can_do_plagiarism_checks %}
<tr>
- <td>Plagiarism score</td>
+ <td>iThenticate plagiarism score</td>
<td>
{% if submission.plagiarism_report %}
{{ submission.plagiarism_report.score }}%
+  <a href="{% url 'submissions:plagiarism_report' submission.preprint.identifier_w_vn_nr %}" class="ms-2 btn btn-default" target="_blank">View report pdf</a>
{% else %}
<a href="{% url 'submissions:plagiarism' submission.preprint.identifier_w_vn_nr %}">Run plagiarism check</a>
{% endif %}
</td>
</tr>
+ <tr>
+ <td>Internal plagiarism matches</td>
+ <td>
+ {% if submission.internal_plagiarism_matches.submission_matches %}
+ Sub: {{ submission.internal_plagiarism_matches.submission_matches|length }} 
+ {% endif %}
+ {% if submission.internal_plagiarism_matches.publication_matches %}
+ Pub: {{ submission.internal_plagiarism_matches.publication_matches|length }} 
+ {% endif %}
+ <a href="{% url 'submissions:plagiarism_internal' submission.preprint.identifier_w_vn_nr %}" target="_blank">View details</a>
+ </td>
+ </tr>
{% endif %}
{% if perms.scipost.can_manage_college_composition %}
diff --git a/scipost_django/submissions/views.py b/scipost_django/submissions/views.py
index 503c82d33d1a231f8a0cd6863085448085ec9222..482c4bb101a3c94d923265cc18e4418e84c8f1db 100644
--- a/scipost_django/submissions/views.py
+++ b/scipost_django/submissions/views.py
@@ -2977,50 +2977,33 @@ class PlagiarismInternalView(SubmissionAdminViewMixin, DetailView):
def get_context_data(self, *args, **kwargs):
context = super().get_context_data(*args, **kwargs)
submission = self.get_object()
- ratio_threshold = 0.7
-
- # Submissions:
- submission_matches = []
- for sub in Submission.objects.exclude(pk=submission.id):
- sub_title_sm = SequenceMatcher(None, submission.title, sub.title)
- ratio_title = sub_title_sm.ratio()
- ratio_authors = SequenceMatcher(
- None, submission.author_list, sub.author_list
- ).ratio()
- ratio_abstract = SequenceMatcher(
- None, submission.abstract, sub.abstract
- ).ratio()
- if ratio_title > ratio_threshold or ratio_abstract > ratio_threshold:
- submission_matches.append(
+
+ context["submission_matches"] = []
+ if "submission_matches" in submission.internal_plagiarism_matches:
+ for sub_match in submission.internal_plagiarism_matches["submission_matches"]:
+ context["submission_matches"].append(
{
- "submission": sub,
- "ratio_title": ratio_title,
- "ratio_authors": ratio_authors,
- "ratio_abstract": ratio_abstract,
+ "submission": Submission.objects.get(
+ preprint__identifier_w_vn_nr=sub_match["identifier_w_vn_nr"],
+ ),
+ "ratio_title": sub_match["ratio_title"],
+ "ratio_authors": sub_match["ratio_authors"],
+ "ratio_abstract": sub_match["ratio_abstract"],
}
)
- context["submission_matches"] = submission_matches
-
- # Publications:
- publication_matches = []
- for pub in Publication.objects.all():
- ratio_title = SequenceMatcher(None, submission.title, pub.title).ratio()
- ratio_authors = SequenceMatcher(
- None, submission.author_list, pub.author_list
- ).ratio()
- ratio_abstract = SequenceMatcher(
- None, submission.abstract, pub.abstract
- ).ratio()
- if ratio_title > ratio_threshold or ratio_abstract > ratio_threshold:
- publication_matches.append(
+
+ context["publication_matches"] = []
+ if "publication_matches" in submission.internal_plagiarism_matches:
+ for pub_match in submission.internal_plagiarism_matches["publication_matches"]:
+ context["publication_matches"].append(
{
- "publication": pub,
- "ratio_title": ratio_title,
- "ratio_authors": ratio_authors,
- "ratio_abstract": ratio_abstract,
+ "publication": Publication.objects.get(doi_label=pub_match["doi_label"]),
+ "ratio_title": pub_match["ratio_title"],
+ "ratio_authors": pub_match["ratio_authors"],
+ "ratio_abstract": pub_match["ratio_abstract"],
}
)
- context["publication_matches"] = publication_matches
+
return context