summaryrefslogtreecommitdiffstats
path: root/python
diff options
context:
space:
mode:
Diffstat (limited to 'python')
-rw-r--r--python/fatcat_tools/importers/common.py12
-rw-r--r--python/fatcat_tools/importers/datacite.py12
-rw-r--r--python/fatcat_tools/importers/ingest.py7
-rw-r--r--python/fatcat_tools/importers/pubmed.py1
-rw-r--r--python/fatcat_tools/workers/elasticsearch.py15
-rw-r--r--python/fatcat_web/routes.py4
-rw-r--r--python/fatcat_web/templates/container_search.html89
-rw-r--r--python/fatcat_web/templates/container_view.html2
-rw-r--r--python/fatcat_web/templates/entity_macros.html78
-rw-r--r--python/fatcat_web/templates/release_search.html59
-rw-r--r--python/fatcat_web/templates/release_view.html12
-rw-r--r--python/tests/files/datacite/datacite_doc_31.json53
-rw-r--r--python/tests/files/datacite/datacite_doc_32.json53
-rw-r--r--python/tests/files/datacite/datacite_result_31.json24
-rw-r--r--python/tests/files/datacite/datacite_result_32.json23
-rw-r--r--python/tests/import_datacite.py2
16 files changed, 336 insertions, 110 deletions
diff --git a/python/fatcat_tools/importers/common.py b/python/fatcat_tools/importers/common.py
index 99c330a6..eafc6546 100644
--- a/python/fatcat_tools/importers/common.py
+++ b/python/fatcat_tools/importers/common.py
@@ -458,7 +458,8 @@ class EntityImporter:
creator_id = rv.ident
except ApiException as ae:
# If anything other than a 404 (not found), something is wrong
- assert ae.status == 404
+ if ae.status != 404:
+ raise ae
self._orcid_id_map[orcid] = creator_id # might be None
return creator_id
@@ -479,7 +480,8 @@ class EntityImporter:
release_id = rv.ident
except ApiException as ae:
# If anything other than a 404 (not found), something is wrong
- assert ae.status == 404
+ if ae.status != 404:
+ raise ae
self._doi_id_map[doi] = release_id # might be None
return release_id
@@ -495,7 +497,8 @@ class EntityImporter:
release_id = rv.ident
except ApiException as ae:
# If anything other than a 404 (not found), something is wrong
- assert ae.status == 404
+ if ae.status != 404:
+ raise ae
self._pmid_id_map[pmid] = release_id # might be None
return release_id
@@ -512,7 +515,8 @@ class EntityImporter:
container_id = rv.ident
except ApiException as ae:
# If anything other than a 404 (not found), something is wrong
- assert ae.status == 404
+ if ae.status != 404:
+ raise ae
self._issnl_id_map[issnl] = container_id # might be None
return container_id
diff --git a/python/fatcat_tools/importers/datacite.py b/python/fatcat_tools/importers/datacite.py
index 81f00876..d998f266 100644
--- a/python/fatcat_tools/importers/datacite.py
+++ b/python/fatcat_tools/importers/datacite.py
@@ -496,10 +496,12 @@ class DataciteImporter(EntityImporter):
if not desc.get('descriptionType') == 'Abstract':
continue
- # Description maybe a string or list.
+ # Description maybe a string, int or list.
text = desc.get('description', '')
if not text:
continue
+ if isinstance(text, int):
+ text = '{}'.format(text)
if isinstance(text, list):
try:
text = "\n".join(text)
@@ -758,6 +760,14 @@ class DataciteImporter(EntityImporter):
given_name = clean(given_name)
if surname:
surname = clean(surname)
+
+ # Perform a final assertion that name does not reduce to zero
+ # (e.g. whitespace only name).
+ if name:
+ name = name.strip()
+ if not name:
+ continue
+
if raw_affiliation == '':
continue
diff --git a/python/fatcat_tools/importers/ingest.py b/python/fatcat_tools/importers/ingest.py
index 4772bfaa..6cf1604b 100644
--- a/python/fatcat_tools/importers/ingest.py
+++ b/python/fatcat_tools/importers/ingest.py
@@ -19,6 +19,7 @@ class IngestFileResultImporter(EntityImporter):
editgroup_description=eg_desc,
editgroup_extra=eg_extra,
**kwargs)
+ self.use_glutton_match = False
self.default_link_rel = kwargs.get("default_link_rel", "web")
assert self.default_link_rel
self.require_grobid = require_grobid
@@ -107,9 +108,10 @@ class IngestFileResultImporter(EntityImporter):
elif err.status == 400:
self.counts['warn-extid-invalid'] += 1
continue
+ raise err
release_ident = release.ident
break
- if not release_ident and row.get('grobid'):
+ if self.use_glutton_match and not release_ident and row.get('grobid'):
# try biblio-glutton extracted hit
if row['grobid'].get('fatcat_release'):
release_ident = row['grobid']['fatcat_release'].split('_')[-1]
@@ -197,8 +199,7 @@ class IngestFileResultImporter(EntityImporter):
if not existing:
return True
- # the following checks all assume there is an existing item
-
+ # NOTE: the following checks all assume there is an existing item
if (fe.release_ids[0] in existing.release_ids) and existing.urls:
# TODO: could still, in theory update with the new URL?
self.counts['exists'] += 1
diff --git a/python/fatcat_tools/importers/pubmed.py b/python/fatcat_tools/importers/pubmed.py
index abcb21d9..3d3e3a8c 100644
--- a/python/fatcat_tools/importers/pubmed.py
+++ b/python/fatcat_tools/importers/pubmed.py
@@ -782,6 +782,7 @@ class PubmedImporter(EntityImporter):
# NOTE: API behavior might change in the future?
if "release_edit_editgroup_id_ident_id_key" in err.body:
self.counts['skip-update-conflict'] += 1
+ return False
else:
raise err
finally:
diff --git a/python/fatcat_tools/workers/elasticsearch.py b/python/fatcat_tools/workers/elasticsearch.py
index 525f372b..e58b3da1 100644
--- a/python/fatcat_tools/workers/elasticsearch.py
+++ b/python/fatcat_tools/workers/elasticsearch.py
@@ -19,7 +19,7 @@ class ElasticsearchReleaseWorker(FatcatWorker):
def __init__(self, kafka_hosts, consume_topic, poll_interval=10.0, offset=None,
elasticsearch_backend="http://localhost:9200", elasticsearch_index="fatcat",
- batch_size=200):
+ batch_size=200, api_host="https://api.fatcat.wiki/v0"):
super().__init__(kafka_hosts=kafka_hosts,
consume_topic=consume_topic)
self.consumer_group = "elasticsearch-updates3"
@@ -30,9 +30,11 @@ class ElasticsearchReleaseWorker(FatcatWorker):
self.entity_type = ReleaseEntity
self.elasticsearch_document_name = "release"
self.transform_func = release_to_elasticsearch
+ self.api_host = api_host
def run(self):
ac = ApiClient()
+ api = public_api(self.api_host)
def fail_fast(err, partitions):
if err is not None:
@@ -103,13 +105,20 @@ class ElasticsearchReleaseWorker(FatcatWorker):
if entity_dict.get('name') and not entity_dict.get('title'):
continue
entity = entity_from_json(json_str, self.entity_type, api_client=ac)
+ if self.elasticsearch_document_name == "changelog":
+ key = entity.index
+ # might need to fetch from API
+ if not (entity.editgroup and entity.editgroup.editor):
+ entity = api.get_changelog_entry(entity.index)
+ else:
+ key = entity.ident
# TODO: handle deletions from index
bulk_actions.append(json.dumps({
- "index": { "_id": entity.ident, },
+ "index": { "_id": key, },
}))
bulk_actions.append(json.dumps(
self.transform_func(entity)))
- print("Upserting, eg, {} (of {} releases in elasticsearch)".format(entity.ident, len(batch)))
+ print("Upserting, eg, {} (of {} {} in elasticsearch)".format(key, len(batch), self.elasticsearch_document_name))
elasticsearch_endpoint = "{}/{}/{}/_bulk".format(
self.elasticsearch_backend,
self.elasticsearch_index,
diff --git a/python/fatcat_web/routes.py b/python/fatcat_web/routes.py
index 8583d255..58f4b7e0 100644
--- a/python/fatcat_web/routes.py
+++ b/python/fatcat_web/routes.py
@@ -707,6 +707,8 @@ def generic_search():
def release_search():
query = request.args.get('q')
+ if not query:
+ query = '*'
fulltext_only = bool(request.args.get('fulltext_only'))
issnl = request.args.get('container_issnl')
@@ -731,6 +733,8 @@ def release_search():
def container_search():
query = request.args.get('q')
+ if not query:
+ query = '*'
offset = request.args.get('offset', '0')
offset = max(0, int(offset)) if offset.isnumeric() else 0
diff --git a/python/fatcat_web/templates/container_search.html b/python/fatcat_web/templates/container_search.html
index 1a089643..1a804595 100644
--- a/python/fatcat_web/templates/container_search.html
+++ b/python/fatcat_web/templates/container_search.html
@@ -9,6 +9,7 @@
{% endif %}
{% endblock %}
+
{% block fullmain %}
<div class="ui vertical stripe segment" style="background-color: #EEE; padding-top: 4.5em;">
@@ -17,8 +18,7 @@
<form class="" role="search" action="/container/search" method="get">
<div class="ui form">
<div class="ui action input huge fluid">
- <input type="text" placeholder="Query..." name="q" value="{% if query %}{{ query }}{% endif %}" aria-label="search container metadata">
- <button class="ui button">Search</button>
+ <input type="text" placeholder="Query..." name="q" value="{% if query %}{{ query }}{% endif %}" aria-label="search container metadata"> <button class="ui button">Search</button>
</div>
<br>Can also lookup by <b><a href="/container/lookup">identifier</a></b> or search <b><a href="/release/search?q={{ query or "" }}">releases</a></b>.
</div>
@@ -30,49 +30,56 @@
<br>
{% if found %}
-{% if found.results %}
- {{ entity_macros.top_results(found) }}
+ {% if found.results %}
-{% for entity in found.results %}
-<div>
- <h4 style="margin-top: 1em; margin-bottom: 4px; font-size: 1.1em;">
- <a href="/container/{{ entity.ident }}" style="color: #2224c7;">{{ entity['name'] }}</a>
- {% if entity.is_oa %}<i class="icon unlock orange small"></i>{% endif %}
- </h4>
- {% if entity.publisher %}
- <h5 style="margin-top: 4px; margin-bottom: 4px; font-size: 1em;">{{ entity.publisher }}</h5>
- {% endif %}
- {% if entity.issnl %}
- <a href="https://portal.issn.org/resource/ISSN/{{entity.issnl }}" style="color: green;">issn:{{ entity.issnl }}</a>
- {% endif %}
- {% if entity.container_type %}
- &nbsp;{{ entity.container_type }}
- {% endif %}
-</div>
-{% endfor %}
-{% if found.results|length > 8 %}
- <div class="ui divider"></div>
- <div style="text-align: center">
- {{ entity_macros.bottom_results(found, endpoint='container_search')}}
- </div>
-{% endif %}
-{% else %}
+ {{ entity_macros.top_results(found) }}
-Raw query was: <i>{{ found.query.q }}</i>
+ {% for entity in found.results %}
+ <div>
+ <h4 style="margin-top: 1em; margin-bottom: 4px; font-size: 1.1em;">
+ <a href="/container/{{ entity.ident }}" style="color: #2224c7;">{{ entity['name'] }}</a>
+ {% if entity.is_oa %}<i class="icon unlock orange small"></i>{% endif %}
+ </h4>
+ {% if entity.publisher %}
+ <h5 style="margin-top: 4px; margin-bottom: 4px; font-size: 1em;">{{ entity.publisher }}</h5>
+ {% endif %}
+ {% if entity.issnl %}
+ <a href="https://portal.issn.org/resource/ISSN/{{entity.issnl }}" style="color: green;">issn:{{ entity.issnl }}</a>
+ {% endif %}
+ {% if entity.container_type %}
+ &nbsp;{{ entity.container_type }}
+ {% endif %}
+ </div>
+ {% endfor %}
+
+ {% if found.results|length > 8 %}
+ <div class="ui divider"></div>
+ <div style="text-align: center">
+ {{ entity_macros.bottom_results(found, endpoint='container_search') }}
+ </div>
+ {% endif %}
-<div class="ui centered stackable grid" style="padding-top: 15%;">
- <div class="row">
- <div class="four wide column">
- <img src="/static/paper_man_confused.gif" alt="confused paper man">
+ {% else %}
+
+ Raw query was: <i>{{ found.query.q }}</i>
+
+ <div class="ui centered stackable grid" style="padding-top: 15%;">
+ <div class="row">
+ <div class="four wide column">
+ <img src="/static/paper_man_confused.gif" alt="confused paper man">
+ </div>
+ <div class="six wide column">
+ <h2>No results found!</h2>
+ <p>You could try elsewhere:</p>
+ <ul>
+ <li>Search <a href="https://scholar.google.com/scholar?q={{ found.query.q | urlencode }}">Google Scholar</a></li>
+ </ul>
+ </div>
+ </div>
</div>
- <div class="six wide column">
- <h2>No results found!</h2>
- <p>You could try elsewhere:</p>
- <ul>
- <li>Search <a href="https://scholar.google.com/scholar?q={{ found.query.q | urlencode }}">Google Scholar</a></li>
- </ul>
-</div>
-{% endif %}
+
+ {% endif %}
+
{% endif %}
</div>
diff --git a/python/fatcat_web/templates/container_view.html b/python/fatcat_web/templates/container_view.html
index 8d7c4cae..cf444956 100644
--- a/python/fatcat_web/templates/container_view.html
+++ b/python/fatcat_web/templates/container_view.html
@@ -23,6 +23,7 @@
Published by {{ container.publisher }}
{% endif %}
+{% if container.state == "active" %}
<h3>Search Releases from this Container</h3>
<form class="" role="search" action="/release/search" method="get">
<div class="ui form">
@@ -33,6 +34,7 @@
</div>
</div>
</form>
+{% endif %}
{% if container._random_releases %}
<h3>Example Publications</h3>
diff --git a/python/fatcat_web/templates/entity_macros.html b/python/fatcat_web/templates/entity_macros.html
index 7c5436c6..c22eb106 100644
--- a/python/fatcat_web/templates/entity_macros.html
+++ b/python/fatcat_web/templates/entity_macros.html
@@ -54,10 +54,12 @@
</a>
</div>
+{% if entity.ident %}
<div class="two ui buttons bottom attached">
<a href="{% if editgroup %}/editgroup/{{ editgroup.editgroup_id }}{% endif %}/{{ entity_type }}/{{ entity.ident }}/edit" class="ui blue button">Edit Metadata</a>
<a href="/{{ entity_type }}/{{ entity.ident }}/history" class="ui button">View History</a>
</div>
+{% endif %}
{%- endmacro %}
@@ -156,17 +158,35 @@
[blank]
{% endif %}
</a>
+
+ {# release type suffix #}
+ {% if paper.release_type in ("article-journal", "paper-conference") %}
+ {# pass #}
+ {% elif paper.release_type in ("book", "chapter", "dataset") %}
+ <b style="text-transform: uppercase;">[{{ paper.release_type }}]</b>
+ {% elif not paper.release_type %}
+ <b style="text-transform: uppercase; color: black;">[unknown-media]</b>
+ {% else %}
+ <b style="text-transform: uppercase;">[{{ paper.release_type }}]</b>
+ {% endif %}
+
+ {# show original_title #}
+ {% if paper.original_title and paper.title != paper.original_title %}
+ <br>
+ <i style="font-weight: normal;">
+ {{ paper.original_title[:512] }} {% if paper.original_title|length > 512 %}...{% endif %}
+ </i>
+ {% endif %}
+
</h4>
+
{% if paper.best_pdf_url %}
<div style="float: right; padding: 4px;">
&nbsp;&nbsp;<a href="{{ paper.best_pdf_url }}" class="ui violet tag label"><i class="file icon"></i>fulltext</a>
</div>
{% endif %}
- {#
- <h5 style="margin-top: 4px; margin-bottom: 4px; font-size: 1em;">{{ ", ".join(paper.contrib_names[:12]) }}
- {% if paper.contrib_names|length > 12 %}<i>(+{{ paper.contrib_names|length - 12 }} others)</i>{% endif %}
- </h5>
- #}
+
+ {# ### AUTHOR ROW #}
{% if paper.contrib_names %}
<div style="margin-top: 0px; margin-bottom: 0px; font-size: 1em;">
<b>
@@ -175,25 +195,14 @@
</b>
</div>
{% endif %}
+
+
+ {# ### JOURNAL ROW #}
{% if paper.release_year %}
{{ paper.release_year }}
{% endif %}
- {% if paper.release_type %}
- {% if paper.release_type in ("article-journal", "paper-conference") %}
- <span class="ui black basic label small">{{ paper.release_type }}</span>
- {% elif paper.release_type in ("book") %}
- <span class="ui brown basic label small">{{ paper.release_type }}</span>
- {% else %}
- <span class="ui grey basic label small">{{ paper.release_type }}</span>
- {% endif %}
- {% endif %}
- {% if paper.withdrawn_status %}
- <span class="ui red label small">{{ paper.withdrawn_status }}</span>
- {% endif %}
- {% if paper.release_stage and paper.release_stage != "published" %}
- <span class="ui pink basic label small">{{ paper.release_stage }}</span>
- {% elif not paper.release_stage %}
- <span class="ui red basic label small">unknown</span>
+ {% if paper.release_year and paper.container_name %}
+ |
{% endif %}
{% if paper.container_name %}
{% if paper.container_id %}
@@ -203,8 +212,22 @@
{% endif %}
{% if paper.container_is_oa %}<i class="icon unlock orange small"></i>{% endif %}
{% endif %}
- {% if paper.doi or paper.pmid or paper.arxiv_id or paper.jstor_id %}
- <br>
+ {% if paper.withdrawn_status %}
+ <b style="color: red;"><code>[{{ paper.withdrawn_status }}]</code></b>
+ {% endif %}
+ {% if paper.release_stage == "accepted" %}
+ <b style="color: darkmagenta;"><code>[{{ paper.release_stage }} manuscript]</code></b>
+ {% elif paper.release_stage == "submitted" %}
+ <b style="color: magenta;"><code>[pre-print]</code></b>
+ {% elif paper.release_stage and paper.release_stage != "published" %}
+ <b style="color: magenta;"><code>[{{ paper.release_stage }}]</code></b>
+ {% elif not paper.release_stage %}
+ <b style="color: red;"><code>[unpublished?]</code></b>
+ {% endif %}
+
+ {# ### IDENTIFIERS #}
+ {% if paper.doi or paper.pmid or paper.arxiv_id or paper.jstor_id or paper.pmcid %}
+ <br>
{% endif %}
{% if paper.doi %}
<a href="https://doi.org/{{paper.doi }}" style="color: green;">doi:{{ paper.doi }}</a> &nbsp;
@@ -212,13 +235,18 @@
{% if paper.pmid %}
<a href="https://www.ncbi.nlm.nih.gov/pubmed/{{paper.pmid }}" style="color: green;">pmid:{{ paper.pmid }}</a> &nbsp;
{% endif %}
+ {% if paper.pmcid %}
+ <a href="https://pubmed.ncbi.nlm.nih.gov/{{paper.pmcid }}/" style="color: green;">pmcid:{{ paper.pmcid }}</a> &nbsp;
+ {% endif %}
{% if paper.arxiv_id %}
<a href="https://arxiv.org/abs/{{paper.arxiv_id }}" style="color: green;">arXiv:{{ paper.arxiv_id }}</a> &nbsp;
{% endif %}
- {% if False %} {# XXX: elastic release work grouping searches #}
+
+{# WIP: elastic release work grouping searches
<br>
<a href="/work/{{ paper.work_id }}"><i class="sitemap icon"></i> and 5 other versions of the same work!</a>
- {% endif %}
+#}
+
</div>
{% endmacro %}
diff --git a/python/fatcat_web/templates/release_search.html b/python/fatcat_web/templates/release_search.html
index 47f4330a..70c46c0a 100644
--- a/python/fatcat_web/templates/release_search.html
+++ b/python/fatcat_web/templates/release_search.html
@@ -35,37 +35,44 @@
<br>
{% if found %}
-{% if found.results %}
+ {% if found.results %}
+
{{ entity_macros.top_results(found) }}
- {% for paper in found.results %}
- {{ entity_macros.release_search_result_row(paper) }}
-{% endfor %}
-{% if found.results|length > 8 %}
- <div class="ui divider"></div>
- <div style="text-align: center">
- {{ entity_macros.bottom_results(found)}}
- </div>
-{% endif %}
-{% else %}
+ {% for paper in found.results %}
+ {{ entity_macros.release_search_result_row(paper) }}
+ {% endfor %}
+
+ {% if found.results|length > 8 %}
+ <div class="ui divider"></div>
+ <div style="text-align: center">
+ {{ entity_macros.bottom_results(found, endpoint='release_search') }}
+ </div>
+ {% endif %}
+
+ {% else %}
-Raw query was: <i>{{ found.query.q }}</i>
+ Raw query was: <i>{{ found.query.q }}</i>
-<div class="ui centered stackable grid" style="padding-top: 15%;">
- <div class="row">
- <div class="four wide column">
- <img src="/static/paper_man_confused.gif" alt="confused paper man">
+ <div class="ui centered stackable grid" style="padding-top: 15%;">
+ <div class="row">
+ <div class="four wide column">
+ <img src="/static/paper_man_confused.gif" alt="confused paper man">
+ </div>
+ <div class="six wide column">
+ <h2>No results found!</h2>
+ <p>You could try elsewhere:</p>
+ <ul>
+ <li>Search <a href="https://dissem.in/search?q={{ found.query.q | urlencode }}">dissem.in</a></li>
+ <li>Search <a href="https://www.base-search.net/Search/Results?lookfor={{ found.query.q | urlencode }}">BASE</a></li>
+ <li>Search <a href="https://scholar.google.com/scholar?q={{ found.query.q | urlencode }}">Google Scholar</a></li>
+ </ul>
+ </div>
+ </div>
</div>
- <div class="six wide column">
- <h2>No results found!</h2>
- <p>You could try elsewhere:</p>
- <ul>
- <li>Search <a href="https://dissem.in/search?q={{ found.query.q | urlencode }}">dissem.in</a></li>
- <li>Search <a href="https://www.base-search.net/Search/Results?lookfor={{ found.query.q | urlencode }}">BASE</a></li>
- <li>Search <a href="https://scholar.google.com/scholar?q={{ found.query.q | urlencode }}">Google Scholar</a></li>
- </ul>
-</div>
-{% endif %}
+
+ {% endif %}
+
{% endif %}
</div>
diff --git a/python/fatcat_web/templates/release_view.html b/python/fatcat_web/templates/release_view.html
index d7c4e76e..60e4624e 100644
--- a/python/fatcat_web/templates/release_view.html
+++ b/python/fatcat_web/templates/release_view.html
@@ -3,9 +3,9 @@
{% import "entity_macros.html" as entity_macros %}
{% extends "entity_base.html" %}
-{# HTML metadata embeddings #}
-{% if release and release.status == "active" %}
- {% block extra_head %}
+{% block extra_head %}
+ {% if release and release.state == "active" %}
+ {# HTML metadata embeddings #}
<link rel="canonical" href="https://{{ config.FATCAT_DOMAIN }}/release/{{ entity.ident }}">
<meta name="twitter:card" content="summary">
@@ -70,8 +70,8 @@
<meta name="citation_pdf_url" content="{{ url.url }}">
{% endif %}
{% endfor %}{% endfor %}
- {% endblock %}
-{% endif %}
+ {% endif %}
+{% endblock %}
{% block main_extra_attr %}itemscope itemtype="http://schema.org/ScholarlyArticle" itemid="#release"{% endblock %}
@@ -388,7 +388,7 @@ accessible version.
<br>grouping other versions (eg, pre-print) and variants of this release
</div>
-{% if release._can_citeproc %}
+{% if release.state == "active" and release._can_citeproc %}
<div class="ui segment attached accordion">
<div class="title" style="padding: 0px;">
<i class="dropdown icon"></i><b>Cite This Release</b>
diff --git a/python/tests/files/datacite/datacite_doc_31.json b/python/tests/files/datacite/datacite_doc_31.json
new file mode 100644
index 00000000..83af3e4d
--- /dev/null
+++ b/python/tests/files/datacite/datacite_doc_31.json
@@ -0,0 +1,53 @@
+{
+ "id": "10.17912/micropub.biology.000143",
+ "type": "dois",
+ "attributes": {
+ "doi": "10.17912/micropub.biology.000143",
+ "identifiers": null,
+ "creators": [
+ {
+ "raw_name": " ",
+ "givenName": "",
+ "familyName": "",
+ "affiliation": [],
+ "role": "author"
+ }
+ ],
+ "titles": [
+ {
+ "title": "Sample"
+ }
+ ],
+ "publisher": "microPublication Biology",
+ "publicationYear": 2019,
+ "types": {
+ "resourceTypeGeneral": "DataPaper"
+ },
+ "relatedIdentifiers": [],
+ "sizes": [],
+ "formats": [],
+ "version": null,
+ "rightsList": [],
+ "descriptions": [
+ {
+ "description": "Biological liquid-liquid phase separation",
+ "descriptionType": "Abstract"
+ }
+ ],
+ "geoLocations": [],
+ "fundingReferences": [],
+ "url": "https://www.micropublication.org/journals/biology/micropub.biology.000143",
+ "created": "2019-08-19T14:43:08.000Z",
+ "registered": "2019-08-19T14:43:09.000Z",
+ "published": "2019",
+ "updated": "2019-11-09T12:32:02.000Z"
+ },
+ "relationships": {
+ "client": {
+ "data": {
+ "id": "caltech.micropub",
+ "type": "clients"
+ }
+ }
+ }
+}
diff --git a/python/tests/files/datacite/datacite_doc_32.json b/python/tests/files/datacite/datacite_doc_32.json
new file mode 100644
index 00000000..7ea7e873
--- /dev/null
+++ b/python/tests/files/datacite/datacite_doc_32.json
@@ -0,0 +1,53 @@
+{
+ "id": "10.17912/micropub.biology.000143",
+ "type": "dois",
+ "attributes": {
+ "doi": "10.17912/micropub.biology.000143",
+ "identifiers": null,
+ "creators": [
+ {
+ "raw_name": " ",
+ "givenName": "",
+ "familyName": "",
+ "affiliation": [],
+ "role": "author"
+ }
+ ],
+ "titles": [
+ {
+ "title": "Sample"
+ }
+ ],
+ "publisher": "microPublication Biology",
+ "publicationYear": 2019,
+ "types": {
+ "resourceTypeGeneral": "DataPaper"
+ },
+ "relatedIdentifiers": [],
+ "sizes": [],
+ "formats": [],
+ "version": null,
+ "rightsList": [],
+ "descriptions": [
+ {
+ "description": 1234567890,
+ "descriptionType": "Abstract"
+ }
+ ],
+ "geoLocations": [],
+ "fundingReferences": [],
+ "url": "https://www.micropublication.org/journals/biology/micropub.biology.000143",
+ "created": "2019-08-19T14:43:08.000Z",
+ "registered": "2019-08-19T14:43:09.000Z",
+ "published": "2019",
+ "updated": "2019-11-09T12:32:02.000Z"
+ },
+ "relationships": {
+ "client": {
+ "data": {
+ "id": "caltech.micropub",
+ "type": "clients"
+ }
+ }
+ }
+}
diff --git a/python/tests/files/datacite/datacite_result_31.json b/python/tests/files/datacite/datacite_result_31.json
new file mode 100644
index 00000000..193104b0
--- /dev/null
+++ b/python/tests/files/datacite/datacite_result_31.json
@@ -0,0 +1,24 @@
+{
+ "abstracts": [
+ {
+ "content": "Biological liquid-liquid phase separation",
+ "lang": "fr",
+ "mimetype": "text/plain"
+ }
+ ],
+ "contribs": [],
+ "ext_ids": {
+ "doi": "10.17912/micropub.biology.000143"
+ },
+ "extra": {
+ "datacite": {
+ "resourceTypeGeneral": "DataPaper"
+ },
+ "container_name": "microPublication Biology"
+ },
+ "refs": [],
+ "release_stage": "published",
+ "release_year": 2019,
+ "publisher": "microPublication Biology",
+ "title": "Sample"
+}
diff --git a/python/tests/files/datacite/datacite_result_32.json b/python/tests/files/datacite/datacite_result_32.json
new file mode 100644
index 00000000..1a84a043
--- /dev/null
+++ b/python/tests/files/datacite/datacite_result_32.json
@@ -0,0 +1,23 @@
+{
+ "abstracts": [
+ {
+ "content": "1234567890",
+ "mimetype": "text/plain"
+ }
+ ],
+ "contribs": [],
+ "ext_ids": {
+ "doi": "10.17912/micropub.biology.000143"
+ },
+ "extra": {
+ "datacite": {
+ "resourceTypeGeneral": "DataPaper"
+ },
+ "container_name": "microPublication Biology"
+ },
+ "refs": [],
+ "release_stage": "published",
+ "release_year": 2019,
+ "publisher": "microPublication Biology",
+ "title": "Sample"
+}
diff --git a/python/tests/import_datacite.py b/python/tests/import_datacite.py
index 15650375..c9210ea4 100644
--- a/python/tests/import_datacite.py
+++ b/python/tests/import_datacite.py
@@ -287,7 +287,7 @@ def test_datacite_conversions(datacite_importer):
for now.
"""
datacite_importer.debug = True
- for i in range(31):
+ for i in range(33):
src = 'tests/files/datacite/datacite_doc_{0:02d}.json'.format(i)
dst = 'tests/files/datacite/datacite_result_{0:02d}.json'.format(i)
with open(src, 'r') as f:
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-23 15:14:02 +0000