From 8bec49b7571bf1beb7310e03b15edae5f47b6ef1 Mon Sep 17 00:00:00 2001
From: "J.-S. Caux" <J.S.Caux@uva.nl>
Date: Fri, 30 Apr 2021 05:51:22 +0200
Subject: [PATCH] Basic working version of ChemRxiv integration
---
preprints/models.py | 10 +-
scipost/services.py | 66 +++++++++++
submissions/constants.py | 20 +++-
submissions/forms.py | 103 ++++++++++++++++--
.../submit_choose_preprint_server.html | 12 ++
submissions/urls.py | 5 +
submissions/views.py | 48 ++++++--
7 files changed, 241 insertions(+), 23 deletions(-)
diff --git a/preprints/models.py b/preprints/models.py
index 846566100..eb1d8fc41 100644
--- a/preprints/models.py
+++ b/preprints/models.py
@@ -47,7 +47,10 @@ class Preprint(models.Model):
"""Return the absolute URL of the pdf for the meta tag for Google Scholar."""
if self._file: # means this is a SciPost-hosted preprint
return "https://scipost.org%s" % self.get_absolute_url()
- return self.get_absolute_url().replace("/abs/", "/pdf/")
+ elif self.is_arXiv:
+ return self.get_absolute_url().replace("/abs/", "/pdf/")
+ elif self.is_ChemRxiv:
+ return self.get_absolute_url()
@property
def is_SciPost(self):
@@ -58,3 +61,8 @@ class Preprint(models.Model):
def is_arXiv(self):
"""Return True if this preprint is hosted on arXiv."""
return 'arxiv.org' in self.url
+
+ @property
+ def is_ChemRxiv(self):
+ """Return True if this preprint is hosted on ChemRxiv."""
+ return 'chemrxiv' in self.url
diff --git a/scipost/services.py b/scipost/services.py
index 63278cdff..3308bf2e3 100644
--- a/scipost/services.py
+++ b/scipost/services.py
@@ -11,6 +11,7 @@ import logging
arxiv_logger = logging.getLogger('scipost.services.arxiv')
doi_logger = logging.getLogger('scipost.services.doi')
+figshare_logger = logging.getLogger('scipost.services.figshare')
class DOICaller:
@@ -159,3 +160,68 @@ class ArxivCaller:
if len(data.get('entries', [])) > 0:
return 'title' in data['entries'][0]
return False
+
+
+class FigshareCaller:
+ """
+ Figshare caller to get data from api.figshare.com.
+ """
+
+ query_base_url = 'https://api.figshare.com/v2/articles/%s/versions/%s'
+
+ def __init__(self, identifier_w_vn_nr):
+ self.identifier_w_vn_nr = identifier_w_vn_nr
+ self.identifier = identifier_w_vn_nr.split('.')[0]
+ self.version = identifier_w_vn_nr.split('.v')[1]
+ figshare_logger.info(
+ 'New figshare API call for identifier %s.v%s' % (self.identifier, self.version))
+ self._call_figshare()
+ if self.is_valid:
+ self._format_data()
+
+ def _call_figshare(self):
+ url = self.query_base_url % (self.identifier, self.version)
+ request = requests.get(url)
+ response_content = request.json()
+ figshare_logger.info('GET [{identifier_w_vn_nr} [request] | {url}'.format(
+ identifier_w_vn_nr=self.identifier_w_vn_nr,
+ url=url,
+ ))
+ if self._result_present(response_content):
+ self.is_valid = True
+ self._figshare_data = response_content
+ self.metadata = response_content
+ else:
+ self.is_valid = False
+
+ figshare_logger.info('GET [{identifier}] [response {valid}] | {response}'.format(
+ identifier=self.identifier,
+ valid='VALID' if self.is_valid else 'INVALID',
+ response=response_content,
+ ))
+
+ def _format_data(self):
+ title = self._figshare_data['title']
+ author_list = [author['full_name'] for author in self._figshare_data.get('authors', [])]
+ # author_list is given as a comma separated list of names on the relevant models (Commentary, Submission)
+ author_list = ", ".join(author_list)
+ chemrxiv_identifier_w_vn_nr = self._figshare_data['doi'].partition('chemrxiv.')[2]
+ abstract = self._figshare_data['description']
+ pub_date = self._figshare_data['published_date']
+ self.data = {
+ 'title': title,
+ 'author_list': author_list,
+ 'chemrxiv_identifier_w_vn_nr': chemrxiv_identifier_w_vn_nr,
+ 'pub_abstract': abstract,
+ 'abstract': abstract, # Duplicate for Commentary/Submission cross-compatibility
+ 'pub_date': pub_date,
+ }
+ figshare_logger.info('GET [{identifier}] [formatted data] | {data}'.format(
+ identifier=self.identifier,
+ data=self.data,
+ ))
+
+ def _result_present(self, data):
+ if data['id'] == int(self.identifier):
+ return True
+ return False
diff --git a/submissions/constants.py b/submissions/constants.py
index c1578141f..bacf8cd21 100644
--- a/submissions/constants.py
+++ b/submissions/constants.py
@@ -266,11 +266,19 @@ PLAGIARISM_STATUSES = (
)
# Define regexes
-arxiv_regex_wo_vn = '[0-9]{4,}.[0-9]{4,}'
-arxiv_regex_w_vn = '[0-9]{4,}.[0-9]{4,}v[0-9]{1,2}'
scipost_regex_wo_vn = 'scipost_[0-9]{4,}_[0-9]{4,}'
scipost_regex_w_vn = 'scipost_[0-9]{4,}_[0-9]{4,}v[0-9]{1,2}'
-SUBMISSIONS_WO_VN_REGEX = '(?P<identifier_wo_vn_nr>(%s|%s))' % (
- arxiv_regex_wo_vn, scipost_regex_wo_vn)
-SUBMISSIONS_COMPLETE_REGEX = '(?P<identifier_w_vn_nr>(%s|%s))' % (
- arxiv_regex_w_vn, scipost_regex_w_vn)
+arxiv_regex_wo_vn = '[0-9]{4,}.[0-9]{4,}'
+arxiv_regex_w_vn = '[0-9]{4,}.[0-9]{4,}v[0-9]{1,2}'
+chemrxiv_regex_wo_vn = 'chemrxiv_[0-9]+'
+chemrxiv_regex_w_vn = 'chemrxiv_[0-9]+.v[0-9]{1,2}'
+SUBMISSIONS_WO_VN_REGEX = '(?P<identifier_wo_vn_nr>(%s|%s|%s))' % (
+ scipost_regex_wo_vn,
+ arxiv_regex_wo_vn,
+ chemrxiv_regex_wo_vn
+)
+SUBMISSIONS_COMPLETE_REGEX = '(?P<identifier_w_vn_nr>(%s|%s|%s))' % (
+ scipost_regex_w_vn,
+ arxiv_regex_w_vn,
+ chemrxiv_regex_w_vn
+)
diff --git a/submissions/forms.py b/submissions/forms.py
index 1c603d1d0..996cd0d05 100644
--- a/submissions/forms.py
+++ b/submissions/forms.py
@@ -45,13 +45,15 @@ from ontology.models import AcademicField, Specialty
from preprints.helpers import get_new_scipost_identifier
from preprints.models import Preprint
from profiles.models import Profile
-from scipost.services import ArxivCaller
+from scipost.services import ArxivCaller, FigshareCaller
from scipost.models import Contributor, Remark
import strings
import iThenticate
ARXIV_IDENTIFIER_PATTERN_NEW = r'^[0-9]{4,}\.[0-9]{4,5}v[0-9]{1,2}$'
+CHEMRXIV_DOI_PATTERN = r'https://doi.org/10.26434/chemrxiv.[0-9]+.v[0-9]{1,2}$'
+CHEMRXIV_IDENTIFIER_PATTERN = r'^[0-9]+\.v[0-9]{1,2}$'
class SubmissionSearchForm(forms.Form):
@@ -163,12 +165,35 @@ def check_arxiv_identifier_w_vn_nr(identifier):
if published_id:
error_message = ('This paper has been published under DOI %(published_id)s. '
- 'It cannot be submitted again.'),
+ 'It cannot be submitted again.')
raise forms.ValidationError(error_message, code='published',
params={'published_id': published_id})
return arxiv_data, metadata, identifier
+def check_chemrxiv_identifier_w_vn_nr(chemrxiv_identifier_w_vn_nr):
+ caller = FigshareCaller(chemrxiv_identifier_w_vn_nr)
+ if caller.is_valid:
+ chemrxiv_data = caller.data
+ metadata = caller.metadata
+ else:
+ error_message = 'A preprint associated to this identifier does not exist.'
+ raise forms.ValidationError(error_message)
+
+ # Check if this article has already been published (according to Figshare)
+ published_id = None
+ if 'resource_doi' in metadata:
+ published_id = metadata['resource_doi']
+
+ if published_id:
+ error_message = ('This paper has been published under DOI %(published_id)s. '
+ 'It cannot be submitted again.')
+ raise forms.ValidationError(error_message, code='published',
+ params={'published_id': published_id})
+ identifier = 'chemrxiv_' + chemrxiv_identifier_w_vn_nr
+ return chemrxiv_data, metadata, identifier
+
+
class SubmissionPrefillForm(forms.Form):
"""
Base class for all SubmissionPrefillForms (one per integrated preprint server).
@@ -297,6 +322,57 @@ class ArXivPrefillForm(SubmissionPrefillForm):
return form_data
+class ChemRxivPrefillForm(SubmissionPrefillForm):
+ """
+ Provide initial data for SubmissionForm (ChemRxiv preprint server route).
+
+ This adds the `chemrxiv_identifier_w_vn_nr` kwarg to those
+ from the `SubmissionPrefillForm` base class.
+ """
+ chemrxiv_identifier_w_vn_nr = forms.RegexField(
+ label='',
+ regex=CHEMRXIV_IDENTIFIER_PATTERN, strip=True,
+ error_messages={'invalid': 'Invalid ChemRxiv identifier'},
+ widget=forms.TextInput()
+ )
+
+ def __init__(self, *args, **kwargs):
+ self.preprint_server = 'ChemRxiv'
+ self.chemrxiv_data = {}
+ self.metadata = {}
+ self.identifier = None
+ super().__init__(*args, **kwargs)
+
+ def clean_chemrxiv_identifier_w_vn_nr(self):
+ """
+ Do basic prechecks based on the ChemRxiv identifier.
+ """
+ self.chemrxiv_data, self.metadata, self.identifier = \
+ check_chemrxiv_identifier_w_vn_nr(self.cleaned_data['chemrxiv_identifier_w_vn_nr'])
+ check_identifier_is_unused(self.identifier)
+ return self.cleaned_data['chemrxiv_identifier_w_vn_nr']
+
+ def get_prefill_data(self):
+ """
+ Return dictionary to prefill `SubmissionForm`.
+ """
+ form_data = super().get_prefill_data()
+ form_data.update(self.chemrxiv_data)
+ form_data['identifier_w_vn_nr'] = self.identifier
+ form_data['chemrxiv_doi'] = ('https://doi.org/10.26434/chemrxiv.'
+ + self.cleaned_data['chemrxiv_identifier_w_vn_nr'])
+
+ if self.is_resubmission():
+ form_data.update({
+ 'approaches': self.latest_submission.approaches,
+ 'referees_flagged': self.latest_submission.referees_flagged,
+ 'referees_suggested': self.latest_submission.referees_suggested,
+ 'acad_field': self.latest_submission.acad_field.id,
+ 'specialties': [s.id for s in self.latest_submission.specialties.all()]
+ })
+ return form_data
+
+
##################
#
# Submission form
@@ -328,6 +404,7 @@ class SubmissionForm(forms.ModelForm):
help_text=('Please submit the processed .pdf (not the source files; '
'these will only be required at the post-acceptance proofs stage)'))
arxiv_link = forms.URLField(label='arXiv link (including version nr)')
+ chemrxiv_doi = forms.URLField(label='ChemRxiv DOI (including version nr)')
class Meta:
model = Submission
@@ -355,6 +432,8 @@ class SubmissionForm(forms.ModelForm):
'thread_hash': forms.HiddenInput(),
'arxiv_link': forms.TextInput(
attrs={'placeholder': 'Full URL, ex.: https://arxiv.org/abs/1234.56789v1'}),
+ 'chemrxiv_doi': forms.TextInput(
+ attrs={'placeholder': 'Full URL, ex.: https://doi.org/10.26434/chemrxiv.1234567.v1'}),
'code_repository_url': forms.TextInput(
attrs={'placeholder': 'If applicable; please give the full URL'}),
'data_repository_url': forms.TextInput(
@@ -379,11 +458,11 @@ class SubmissionForm(forms.ModelForm):
def __init__(self, *args, **kwargs):
self.requested_by = kwargs.pop('requested_by')
self.preprint_server = kwargs.pop('preprint_server')
- data = data = args[0] if len(args) > 1 else kwargs.get('data', {})
+ data = args[0] if len(args) > 1 else kwargs.get('data', {})
self.thread_hash = kwargs['initial'].get('thread_hash', None) or data.get('thread_hash')
self.is_resubmission_of = kwargs['initial'].get(
'is_resubmission_of', None) or data.get('is_resubmission_of')
- self.arxiv_data = {}
+ self.preprint_data = {}
self.metadata = {} # container for possible external server-provided metadata
super().__init__(*args, **kwargs)
@@ -398,6 +477,8 @@ class SubmissionForm(forms.ModelForm):
if not self.preprint_server == 'arXiv':
# No arXiv-specific data required.
del self.fields['arxiv_link']
+ if not self.preprint_server == 'ChemRxiv':
+ del self.fields['chemrxiv_doi']
if not self.is_resubmission():
del self.fields['is_resubmission_of']
@@ -495,7 +576,10 @@ class SubmissionForm(forms.ModelForm):
check_identifier_is_unused(identifier)
if self.preprint_server == 'arXiv':
- self.arxiv_data, self.metadata, identifier = check_arxiv_identifier_w_vn_nr(identifier)
+ self.preprint_data, self.metadata, identifier = check_arxiv_identifier_w_vn_nr(identifier)
+ elif self.preprint_server == 'ChemRxiv':
+ self.preprint_data, self.metadata, identifier = check_chemrxiv_identifier_w_vn_nr(
+ identifier.lstrip('chemrxiv_'))
return identifier
@transaction.atomic
@@ -507,12 +591,17 @@ class SubmissionForm(forms.ModelForm):
submission.submitted_by = self.requested_by.contributor
# Save identifiers
+ url = None
+ if self.preprint_server == 'arXiv':
+ url = self.cleaned_data.get('arxiv_link', '')
+ elif self.preprint_server == 'ChemRxiv':
+ url = self.cleaned_data.get('chemrxiv_doi', '')
preprint, __ = Preprint.objects.get_or_create(
identifier_w_vn_nr=self.cleaned_data['identifier_w_vn_nr'],
- url=self.cleaned_data.get('arxiv_link', ''),
+ url=url,
_file=self.cleaned_data.get('preprint_file', None), )
- # Save metadata directly from ArXiv call without possible user interception
+ # Save metadata directly from preprint server call without possible user interception
submission.metadata = self.metadata
submission.preprint = preprint
diff --git a/submissions/templates/submissions/submit_choose_preprint_server.html b/submissions/templates/submissions/submit_choose_preprint_server.html
index 8d8726261..7d525641c 100644
--- a/submissions/templates/submissions/submit_choose_preprint_server.html
+++ b/submissions/templates/submissions/submit_choose_preprint_server.html
@@ -70,6 +70,18 @@
{% include 'bi/arrow-right.html' %}
<input type="submit" class="btn btn-primary text-white" value="Query arXiv"/>
</form>
+ {% elif server.name == 'ChemRxiv' %}
+ <h3>Please provide the ChemRxiv identifier for your Submission</h3>
+ <p><em>without the url prefix but with version number, e.g.
+ <br> #######.v#</em></p>
+ <form action="{% url 'submissions:submit_manuscript_chemrxiv' journal_doi_label=journal.doi_label %}" method="get">
+ {{ chemrxiv_prefill_form }}
+ {% if thread_hash %}
+ <input type="hidden" name="thread_hash" value="{{ thread_hash }}"/>
+ {% endif %}
+ {% include 'bi/arrow-right.html' %}
+ <input type="submit" class="btn btn-primary text-white" value="Query ChemRxiv"/>
+ </form>
{% endif %}
</div>
</div>
diff --git a/submissions/urls.py b/submissions/urls.py
index e13a35eb2..e9ddab7f8 100644
--- a/submissions/urls.py
+++ b/submissions/urls.py
@@ -153,6 +153,11 @@ urlpatterns = [
views.RequestSubmissionUsingArXivView.as_view(),
name='submit_manuscript_arxiv'
),
+ path( # Submit using arXiv (thread_hash as GET param if resubmission)
+ 'submit_manuscript/<journal_doi_label:journal_doi_label>/chemrxiv',
+ views.RequestSubmissionUsingChemRxivView.as_view(),
+ name='submit_manuscript_chemrxiv'
+ ),
url(
r'^withdraw_manuscript/{regex}/$'.format(regex=SUBMISSIONS_COMPLETE_REGEX),
diff --git a/submissions/views.py b/submissions/views.py
index a754173b8..db2364fa9 100644
--- a/submissions/views.py
+++ b/submissions/views.py
@@ -40,7 +40,7 @@ from .models import (
EditorialAssignment, RefereeInvitation, Report, SubmissionEvent)
from .mixins import SubmissionMixin, SubmissionAdminViewMixin
from .forms import (
- ArXivPrefillForm, SciPostPrefillForm,
+ SciPostPrefillForm, ArXivPrefillForm, ChemRxivPrefillForm,
SubmissionForm, SubmissionSearchForm, RecommendationVoteForm,
ConsiderAssignmentForm, InviteEditorialAssignmentForm, EditorialAssignmentForm, VetReportForm,
SetRefereeingDeadlineForm, RefereeSearchForm,
@@ -163,15 +163,24 @@ def submit_choose_preprint_server(request, journal_doi_label):
thread_hash = request.GET.get('thread_hash') or None
# Each integrated preprint server has a prefill form:
scipost_prefill_form = SciPostPrefillForm(
- requested_by=request.user, journal_doi_label=journal_doi_label, thread_hash=thread_hash)
+ requested_by=request.user,
+ journal_doi_label=journal_doi_label,
+ thread_hash=thread_hash)
arxiv_prefill_form = ArXivPrefillForm(
- requested_by=request.user, journal_doi_label=journal_doi_label, thread_hash=thread_hash)
+ requested_by=request.user,
+ journal_doi_label=journal_doi_label,
+ thread_hash=thread_hash)
+ chemrxiv_prefill_form = ChemRxivPrefillForm(
+ requested_by=request.user,
+ journal_doi_label=journal_doi_label,
+ thread_hash=thread_hash)
context = {
'journal': journal,
'thread_hash': thread_hash,
'preprint_servers': preprint_servers,
'scipost_prefill_form': scipost_prefill_form,
'arxiv_prefill_form': arxiv_prefill_form,
+ 'chemrxiv_prefill_form': chemrxiv_prefill_form,
}
return render(request, 'submissions/submit_choose_preprint_server.html', context)
@@ -191,7 +200,7 @@ class RequestSubmissionView(LoginRequiredMixin, PermissionRequiredMixin, CreateV
def get(self, request, journal_doi_label):
"""
- Redirect to `submit_choose_preprint_server` if arXiv identifier is not known.
+ Redirect to `submit_choose_preprint_server` if preprint identifier is not known.
"""
if self.prefill_form.is_valid():
if self.prefill_form.is_resubmission():
@@ -201,13 +210,13 @@ class RequestSubmissionView(LoginRequiredMixin, PermissionRequiredMixin, CreateV
'Please check everything carefully!')
messages.success(request, resubmessage, fail_silently=True)
else:
- if self.prefill_form.preprint_server == 'arXiv':
+ if self.prefill_form.preprint_server in ['arXiv', 'ChemRxiv']:
readymessage = ('We have pre-filled the form where possible. '
'Please check everything carefully!')
else:
readymessage = 'Your submission form is now ready to be filled in.'
messages.success(request, readymessage, fail_silently=True)
- # Gather data from ArXiv API if prefill form is valid
+ # Gather data from preprint server API if prefill form is valid
self.initial_data = self.prefill_form.get_prefill_data()
return super().get(request)
else:
@@ -300,6 +309,25 @@ class RequestSubmissionUsingArXivView(RequestSubmissionView):
return kwargs
+class RequestSubmissionUsingChemRxivView(RequestSubmissionView):
+ """Formview to submit a new Submission using ChemRxiv."""
+
+ def get(self, request, journal_doi_label):
+ """
+ Redirect to `submit_choose_preprint_server` if ChemRxiv identifier is not known.
+ """
+ self.prefill_form = ChemRxivPrefillForm(
+ request.GET or None,
+ requested_by=self.request.user,
+ journal_doi_label=journal_doi_label,
+ thread_hash=request.GET.get('thread_hash'))
+ return super().get(request, journal_doi_label)
+
+ def get_form_kwargs(self):
+ """Form requires extra kwargs."""
+ kwargs = super().get_form_kwargs()
+ kwargs['preprint_server'] = 'ChemRxiv'
+ return kwargs
@login_required
@@ -417,6 +445,7 @@ def submission_detail_wo_vn_nr(request, identifier_wo_vn_nr):
def submission_detail(request, identifier_w_vn_nr):
"""Public detail page of Submission."""
+
submission = get_object_or_404(Submission, preprint__identifier_w_vn_nr=identifier_w_vn_nr)
context = {
'can_read_editorial_information': False,
@@ -431,9 +460,9 @@ def submission_detail(request, identifier_w_vn_nr):
if not request.user.is_authenticated:
raise Http404
elif not request.user.has_perm(
- 'scipost.can_assign_submissions') and not submission.fellows.filter(
- contributor__user=request.user).exists():
- raise Http404
+ 'scipost.can_assign_submissions') and not submission.fellows.filter(
+ contributor__user=request.user).exists():
+ raise Http404
if is_author:
context['proofs_decision_form'] = ProofsDecisionForm()
@@ -483,6 +512,7 @@ def submission_detail(request, identifier_w_vn_nr):
'is_author': is_author,
'is_author_unchecked': is_author_unchecked,
})
+ print(context)
return render(request, 'submissions/submission_detail.html', context)
--
GitLab