From 1d3919651eba0cc66cb0f3be311a5e208d92a118 Mon Sep 17 00:00:00 2001
From: George Katsikas <giorgakis.katsikas@gmail.com>
Date: Thu, 21 Dec 2023 17:12:31 +0200
Subject: [PATCH] add chemrxiv preprint caller and logger
---
scipost_django/SciPost_v1/settings/base.py | 12 +
.../SciPost_v1/settings/production_do1.py | 3 +
scipost_django/scipost/services.py | 67 +++++
scipost_django/submissions/forms/__init__.py | 75 +++--
.../submit_choose_preprint_server.html | 260 +++++++++++-------
5 files changed, 278 insertions(+), 139 deletions(-)
diff --git a/scipost_django/SciPost_v1/settings/base.py b/scipost_django/SciPost_v1/settings/base.py
index 52855255f..5865cea0b 100644
--- a/scipost_django/SciPost_v1/settings/base.py
+++ b/scipost_django/SciPost_v1/settings/base.py
@@ -489,6 +489,12 @@ LOGGING = {
"filename": "/path/to/logs/arxiv.log",
"formatter": "verbose",
},
+ "scipost_file_chemrxiv": {
+ "level": "INFO",
+ "class": "logging.FileHandler",
+ "filename": "/path/to/logs/chemrxiv.log",
+ "formatter": "verbose",
+ },
"scipost_file_doi": {
"level": "INFO",
"class": "logging.FileHandler",
@@ -521,6 +527,12 @@ LOGGING = {
"propagate": True,
"formatter": "simple",
},
+ "scipost.services.chemrxiv": {
+ "handlers": ["scipost_file_chemrxiv"],
+ "level": "INFO",
+ "propagate": True,
+ "formatter": "simple",
+ },
"scipost.services.doi": {
"handlers": ["scipost_file_doi"],
"level": "INFO",
diff --git a/scipost_django/SciPost_v1/settings/production_do1.py b/scipost_django/SciPost_v1/settings/production_do1.py
index 03dbe9f22..043e790d7 100644
--- a/scipost_django/SciPost_v1/settings/production_do1.py
+++ b/scipost_django/SciPost_v1/settings/production_do1.py
@@ -63,6 +63,9 @@ ITHENTICATE_PASSWORD = get_secret("ITHENTICATE_PASSWORD")
LOGGING["handlers"]["scipost_file_arxiv"][
"filename"
] = "/home/scipost/SciPost_logs/arxiv.log"
+LOGGING["handlers"]["scipost_file_chemrxiv"][
+ "filename"
+] = "/home/scipost/SciPost_logs/chemrxiv.log"
LOGGING["handlers"]["scipost_file_doi"][
"filename"
] = "/home/scipost/SciPost_logs/doi.log"
diff --git a/scipost_django/scipost/services.py b/scipost_django/scipost/services.py
index 519940b11..090276d9b 100644
--- a/scipost_django/scipost/services.py
+++ b/scipost_django/scipost/services.py
@@ -3,6 +3,8 @@ __license__ = "AGPL v3"
# Module for making external api calls as needed in the submissions cycle
+import re
+from django.db.models import query
import feedparser
import requests
import datetime
@@ -13,8 +15,10 @@ from common.utils import remove_extra_spacing
from submissions.constants import FIGSHARE_PREPRINT_SERVERS
from submissions.models import PreprintServer
+from submissions.regexes import CHEMRXIV_DOI_PATTERN
arxiv_logger = logging.getLogger("scipost.services.arxiv")
+chemrxiv_logger = logging.getLogger("scipost.services.chemrxiv")
doi_logger = logging.getLogger("scipost.services.doi")
figshare_logger = logging.getLogger("scipost.services.figshare")
osfpreprints_logger = logging.getLogger("scipost.services.osfpreprints")
@@ -188,6 +192,69 @@ class ArxivCaller:
return False
+class ChemRxivCaller:
+ """ChemRxiv Caller will help retrieve Submission data from ChemRxiv API."""
+
+ query_base_url = "https://chemrxiv.org/engage/chemrxiv/public-api/v1/items/"
+ doi_pattern = r"(https://doi.org/)?" + f"({CHEMRXIV_DOI_PATTERN})"
+ url_id_pattern = (
+ r"(https://chemrxiv.org/engage/chemrxiv/article-details/)?([a-z0-9]{24})"
+ )
+ valid_patterns = f"(({doi_pattern})|({url_id_pattern}))"
+
+ def __init__(self, identifier):
+ self.identifier = identifier
+ chemrxiv_logger.info("New ChemRxiv call for identifier %s" % identifier)
+
+ self.metadata = self._call_chemrxiv()
+ self.data = self._format_data(self.metadata) if self.is_valid else None
+
+ def _call_chemrxiv(self):
+ if m := re.match(self.doi_pattern, self.identifier):
+ doi = m.group(2)
+ url = self.query_base_url + "doi/" + doi
+ elif m := re.match(self.url_id_pattern, self.identifier):
+ item_id = m.group(2)
+ url = self.query_base_url + item_id
+ else:
+ self.is_valid = False
+ return
+
+ request = requests.get(url)
+ try:
+ response_content = request.json()
+ except:
+ self.is_valid = False
+ return
+
+ chemrxiv_logger.info(
+ "GET [{chemrxiv}] [request] | {url}".format(
+ chemrxiv=self.identifier, url=url
+ )
+ )
+
+ self.url = url
+ self.is_valid = True
+ return response_content
+
+ @staticmethod
+ def _format_data(data):
+ formatted_data = {
+ "title": data.get("title", ""),
+ "author_list": ", ".join(
+ [a["firstName"] + " " + a["lastName"] for a in data.get("authors", [])]
+ ),
+ "abstract": data.get("abstract", ""),
+ "pub_date": datetime.datetime.fromisoformat(pub_date)
+ if (pub_date := data.get("publishedDate"))
+ else None,
+ "identifier_w_vn_nr": data.get("doi", ""),
+ "preprint_link": "https://doi.org/" + data.get("doi", ""),
+ }
+
+ return formatted_data
+
+
class FigshareCaller:
"""
Figshare caller to get data from api.figshare.com.
diff --git a/scipost_django/submissions/forms/__init__.py b/scipost_django/submissions/forms/__init__.py
index 8bb3534ca..af0949cd3 100644
--- a/scipost_django/submissions/forms/__init__.py
+++ b/scipost_django/submissions/forms/__init__.py
@@ -80,7 +80,13 @@ from preprints.helpers import get_new_scipost_identifier
from preprints.models import Preprint
from proceedings.models import Proceedings
from profiles.models import Profile
-from scipost.services import DOICaller, ArxivCaller, FigshareCaller, OSFPreprintsCaller
+from scipost.services import (
+ ChemRxivCaller,
+ DOICaller,
+ ArxivCaller,
+ FigshareCaller,
+ OSFPreprintsCaller,
+)
from scipost.models import Contributor, Remark
from series.models import Collection
import strings
@@ -721,45 +727,18 @@ def check_arxiv_identifier_w_vn_nr(identifier):
def check_chemrxiv_doi(doi):
"""
Call Crossref to get ChemRxiv preprint data.
+ `doi` is the DOI of the preprint, but can also be a link to the submission, or its database ID.
"""
- caller = DOICaller(doi)
- if caller.is_valid:
- data = caller.data
- metadata = caller.data["crossref_data"]
- else:
- error_message = "A preprint associated to this DOI does not exist."
- raise forms.ValidationError(error_message)
-
- # Check if the type of this resource is indeed a preprint
- if "subtype" in metadata:
- if metadata["subtype"] != "preprint":
- error_message = (
- "This does not seem to be a preprint: the type "
- "returned by Crossref on behalf of "
- "%(preprint_server) is %(subtype). "
- "Please contact techsupport."
- )
- raise forms.ValidationError(
- error_message,
- code="wrong_subtype",
- params={
- "preprint_server": preprint_server.name,
- "subtype": metadata["subtype"],
- },
- )
- else:
- raise forms.ValidationError(
- "Crossref failed to return a subtype. Please contact techsupport.",
- code="wrong_subtype",
+ caller = ChemRxivCaller(doi)
+ if not caller.is_valid or not (data := caller.data):
+ error_message = (
+ "The preprint could not be found. Please check the provided identifier."
)
+ raise forms.ValidationError(error_message)
# Explicitly add ChemRxiv as the preprint server:
data["preprint_server"] = PreprintServer.objects.get(name="ChemRxiv")
- data["preprint_link"] = "https://doi.org/%s" % doi
- # Build the identifier by stripping the DOI prefix:
- identifier = doi
- data["identifier_w_vn_nr"] = identifier
- return data, metadata, identifier
+ return data, caller.metadata, data.get("identifier_w_vn_nr")
def check_figshare_identifier_w_vn_nr(preprint_server, figshare_identifier_w_vn_nr):
@@ -1054,14 +1033,14 @@ class ChemRxivPrefillForm(SubmissionPrefillForm):
chemrxiv_doi = forms.RegexField(
label="",
- regex=CHEMRXIV_DOI_PATTERN,
+ regex=ChemRxivCaller.valid_patterns,
strip=True,
error_messages={"invalid": "Invalid ChemRxiv DOI"},
widget=forms.TextInput(),
)
def __init__(self, *args, **kwargs):
- self.crossref_data = {}
+ self.data = {}
self.metadata = {}
super().__init__(*args, **kwargs)
@@ -1070,7 +1049,7 @@ class ChemRxivPrefillForm(SubmissionPrefillForm):
identifier = self.cleaned_data.get("chemrxiv_doi", None).partition("/")[2]
check_identifier_is_unused(identifier)
- self.crossref_data, self.metadata, identifier = check_chemrxiv_doi(
+ self.data, self.metadata, identifier = check_chemrxiv_doi(
self.cleaned_data["chemrxiv_doi"]
)
return identifier
@@ -1080,7 +1059,21 @@ class ChemRxivPrefillForm(SubmissionPrefillForm):
Return dictionary to prefill `SubmissionForm`.
"""
form_data = super().get_prefill_data()
- form_data.update(self.crossref_data)
+ form_data.update(self.data)
+
+ # check metadata for specialties
+ category_titles = (
+ [c["name"].title() for c in self.metadata["categories"]]
+ if self.metadata
+ else []
+ )
+ form_data["specialties"] = Specialty.objects.filter(name__in=category_titles)
+
+ # check keywords for topics
+ keyword_titles = (
+ [k.title() for k in self.metadata["keywords"]] if self.metadata else []
+ )
+ form_data["topics"] = Topic.objects.filter(name__in=keyword_titles)
if self.is_resubmission():
form_data.update(
@@ -1619,10 +1612,10 @@ class SubmissionForm(forms.ModelForm):
def clean_title(self):
return remove_extra_spacing(self.cleaned_data["title"])
-
+
def clean_abstract(self):
return remove_extra_spacing(self.cleaned_data["abstract"])
-
+
@transaction.atomic
def save(self):
"""
diff --git a/scipost_django/submissions/templates/submissions/submit_choose_preprint_server.html b/scipost_django/submissions/templates/submissions/submit_choose_preprint_server.html
index 23cd5b349..ac9ef0248 100644
--- a/scipost_django/submissions/templates/submissions/submit_choose_preprint_server.html
+++ b/scipost_django/submissions/templates/submissions/submit_choose_preprint_server.html
@@ -4,7 +4,9 @@
{% load bootstrap %}
{% load journals_extras %}
-{% block pagetitle %}: submit manuscript{% endblock pagetitle %}
+{% block pagetitle %}
+ : submit manuscript
+{% endblock pagetitle %}
{% block breadcrumb_items %}
{{ block.super }}
@@ -14,7 +16,16 @@
{% block content %}
<div class="row">
<div class="col-12">
- <h1 class="highlight">{% if thread_hash %}Resubmit a{% else %}Submit a new{% endif %} manuscript to {{ journal }}</h1>
+ <h1 class="highlight">
+
+ {% if thread_hash %}
+ Resubmit a
+ {% else %}
+ Submit a new
+ {% endif %}
+
+ manuscript to {{ journal }}
+ </h1>
</div>
</div>
@@ -22,106 +33,159 @@
<div class="row">
<div class="col-md-6">
- <h2 class=highlight>Step 3</h2>
- {% if preprint_server_list|length > 1 %}
- <h3>Which preprint server do you wish to use for your {% if thread_hash %}re{% endif %}submission?</h3>
- <p>SciPost's own preprint server is always available.
- We however strongly encourage you to use your field's customary
- preprint server, to ensure maximal dissemination of your preprint.</p>
- <p>If you use an external preprint server, make sure your submission
- is publicly available there before proceeding here (our machines will query theirs
- in order to automatically prefill our submission form for you).</p>
- {% else %}
- <h3>Please proceed with the SciPost preprint server (others are not available)</h3>
- {% endif %}
- </div>
- <div class="col-md-6">
- {% include 'submissions/submit_steps.html' with step=3 thread_hash=thread_hash %}
+ <h2 class=highlight>Step 3</h2>
+
+ {% if preprint_server_list|length > 1 %}
+ <h3>
+ Which preprint server do you wish to use for your
+
+ {% if thread_hash %}re{% endif %}
+
+ submission?
+ </h3>
+ <p>
+ SciPost's own preprint server is always available.
+ We however strongly encourage you to use your field's customary
+ preprint server, to ensure maximal dissemination of your preprint.
+ </p>
+ <p>
+ If you use an external preprint server, make sure your submission
+ is publicly available there before proceeding here (our machines will query theirs
+ in order to automatically prefill our submission form for you).
+ </p>
+ {% else %}
+ <h3>Please proceed with the SciPost preprint server (others are not available)</h3>
+ {% endif %}
+
</div>
+ <div class="col-md-6">{% include 'submissions/submit_steps.html' with step=3 thread_hash=thread_hash %}</div>
</div>
- <br>
+ <br />
<div class="container">
<div class="row row-cols-1 row-cols-lg-2 row-cols-xl-3">
- {% for preprint_server in preprint_server_list %}
- <div class="col col-sm-12 col-lg-6 mb-2">
- <div class="card mb-4">
- <div class="card-header bg-dark text-white">
- <h3 class="p-2 m-0"><em>{% if thread_hash %}Resubmit{% else %}Submit{% endif %} via</em> <strong class="text-warning">{{ preprint_server.server }}</strong></h3>
- </div>
- <div class="card-body text-center">
- {% if preprint_server.server.name == 'SciPost' %}
- <form action="{% url 'submissions:submit_manuscript_scipost' journal_doi_label=journal.doi_label %}" method="get">
- {{ preprint_server.prefill_form }}
- {% if thread_hash %}
- <input type="hidden" name="thread_hash" value="{{ thread_hash }}"/>
- {% endif %}
- <input type="submit" class="btn btn-primary text-white" value="Go to the SciPost submission form"/>
- </form>
- {% elif preprint_server.server.name == 'arXiv' %}
- <h3>Please provide the arXiv identifier for your Submission</h3>
- <p><em>without the website prefix but with version number, e.g.
- <br> ####.####(#)v#(#)</em></p>
- <form action="{% url 'submissions:submit_manuscript_arxiv' journal_doi_label=journal.doi_label %}" method="get">
- {{ preprint_server.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 arXiv"/>
- </form>
- {% elif preprint_server.server.name == 'ChemRxiv' %}
- <h3>Please provide the ChemRxiv DOI for your Submission</h3>
- <p class="mb-0">Pattern: either
- <ul>
- <li>old style: <em>10.#####/chemrxiv.#######(.v#)</em></li>
- <li>new style: <em>10.#####/chemrxiv-*****(-v#)</em></li>
- </ul>
- </p>
- <form action="{% url 'submissions:submit_manuscript_chemrxiv' journal_doi_label=journal.doi_label %}" method="get">
- {{ preprint_server.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>
- {% elif preprint_server.server.served_by and preprint_server.server.served_by.name == 'OSFPreprints' %}
- <h3>Please provide the {{ preprint_server.server.name }} identifier for your Submission</h3>
- <p><em>just a few lowercase alphanumeric characters, e.g. #####</em></p>
- <p><em>Hint: on the submission's {{ preprint_server.server.name }} page, look for the last few characters in the DOI to see the identifier.</em></p>
- <p><em>Example: if the DOI is <strong>10.31235/osf.io/xxzbx</strong>,
- then you need <strong>xxzbx</strong></em></p>
- <form action="{% url 'submissions:submit_manuscript_osfpreprints' journal_doi_label=journal.doi_label %}" method="get">
- {{ preprint_server.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 {{ preprint_server.server.name }}"/>
- </form>
- {% elif preprint_server.server.served_by and preprint_server.server.served_by.name == 'Figshare' %}
- <h3>Please provide the {{ preprint_server.server.name }} identifier for your Submission</h3>
- <p><em>without the url prefix but with version number, e.g.
- <br> ########.v#</em></p>
- <p><em>Hint: on the submission's {{ preprint_server.server.name }} page, click on the "Cite" button to see the identifier.</em></p>
- <form action="{% url 'submissions:submit_manuscript_figshare' journal_doi_label=journal.doi_label %}" method="get">
- {{ preprint_server.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 {{ preprint_server.server.name }}"/>
- </form>
- {% endif %}
- </div>
- </div>
- </div>
- {% endfor %}
+
+ {% for preprint_server in preprint_server_list %}
+ <div class="col col-sm-12 col-lg-6 mb-2">
+ <div class="card mb-4">
+ <div class="card-header bg-dark text-white">
+ <h3 class="p-2 m-0">
+ <em>
+
+ {% if thread_hash %}
+ Resubmit
+ {% else %}
+ Submit
+ {% endif %}
+
+ via</em> <strong class="text-warning">{{ preprint_server.server }}</strong>
+ </h3>
+ </div>
+ <div class="card-body">
+
+ {% if preprint_server.server.name == 'SciPost' %}
+ <form action="{% url 'submissions:submit_manuscript_scipost' journal_doi_label=journal.doi_label %}"
+ method="get">
+ {{ preprint_server.prefill_form }}
+
+ {% if thread_hash %}<input type="hidden" name="thread_hash" value="{{ thread_hash }}" />{% endif %}
+
+ <input type="submit"
+ class="btn btn-primary text-white"
+ value="Go to the SciPost submission form" />
+ </form>
+ {% elif preprint_server.server.name == 'arXiv' %}
+ <h3>Please provide the arXiv identifier for your Submission</h3>
+ <p>
+ <em>without the website prefix but with version number, e.g.
+ <br />
+ ####.####(#)v#(#)</em>
+ </p>
+ <form action="{% url 'submissions:submit_manuscript_arxiv' journal_doi_label=journal.doi_label %}"
+ method="get">
+ {{ preprint_server.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 arXiv" />
+ </form>
+ {% elif preprint_server.server.name == 'ChemRxiv' %}
+ <h3>Please provide the ChemRxiv DOI/URL for your Submission</h3>
+ <p class="mb-0">
+ Patterns:
+ <ul>
+ <li>10.#####/chemrxiv-*****(-v#)</li>
+ <li>https://chemrxiv.org/****/article-details/****</li>
+ </ul>
+ </p>
+ <form action="{% url 'submissions:submit_manuscript_chemrxiv' journal_doi_label=journal.doi_label %}"
+ method="get">
+ {{ preprint_server.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>
+ {% elif preprint_server.server.served_by and preprint_server.server.served_by.name == 'OSFPreprints' %}
+ <h3>Please provide the {{ preprint_server.server.name }} identifier for your Submission</h3>
+ <p>
+ <em>just a few lowercase alphanumeric characters, e.g. #####</em>
+ </p>
+ <p>
+ <em>Hint: on the submission's {{ preprint_server.server.name }} page, look for the last few characters in the DOI to see the identifier.</em>
+ </p>
+ <p>
+ <em>Example: if the DOI is <strong>10.31235/osf.io/xxzbx</strong>,
+ then you need <strong>xxzbx</strong></em>
+ </p>
+ <form action="{% url 'submissions:submit_manuscript_osfpreprints' journal_doi_label=journal.doi_label %}"
+ method="get">
+ {{ preprint_server.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 {{ preprint_server.server.name }}" />
+ </form>
+ {% elif preprint_server.server.served_by and preprint_server.server.served_by.name == 'Figshare' %}
+ <h3>Please provide the {{ preprint_server.server.name }} identifier for your Submission</h3>
+ <p>
+ <em>without the url prefix but with version number, e.g.
+ <br />
+ ########.v#</em>
+ </p>
+ <p>
+ <em>Hint: on the submission's {{ preprint_server.server.name }} page, click on the "Cite" button to see the identifier.</em>
+ </p>
+ <form action="{% url 'submissions:submit_manuscript_figshare' journal_doi_label=journal.doi_label %}"
+ method="get">
+ {{ preprint_server.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 {{ preprint_server.server.name }}" />
+ </form>
+ {% endif %}
+
+ </div>
+ </div>
+ </div>
+ {% endfor %}
+
+ </div>
</div>
- </div>
- {% else %}
- <h3>You are currently not allowed to submit a manuscript.</h3>
- {% endif %}
-{% endblock content %}
+ {% else %}
+ <h3>You are currently not allowed to submit a manuscript.</h3>
+ {% endif %}
+
+ {% endblock content %}
--
GitLab