From d5b24df069fc96d396afbb302633a077e5dbfb39 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Thu, 15 Apr 2021 18:34:11 -0700
Subject: first iteration of basic citation inbound/outbound views

---
 python/fatcat_web/ref_routes.py | 50 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
 create mode 100644 python/fatcat_web/ref_routes.py

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
new file mode 100644
index 00000000..a49813c4
--- /dev/null
+++ b/python/fatcat_web/ref_routes.py
@@ -0,0 +1,50 @@
+"""
+Flask endpoints for reference (citation) endpoints. Eg, listing references
+"inbound" and "outbound" from a specific release or work.
+"""
+
+from typing import Optional
+
+from flask import render_template, abort, redirect, request
+from fatcat_openapi_client import *
+from fatcat_openapi_client.rest import ApiException
+
+from fatcat_tools.references import enrich_inbound_refs_fatcat, enrich_outbound_refs_fatcat, get_inbound_refs, get_outbound_refs
+from fatcat_web import app, api, auth_api
+from fatcat_web.forms import *
+from fatcat_web.entity_helpers import *
+
+
+@app.route('/release/<string(length=26):ident>/refs/inbound', methods=['GET'])
+def release_view_refs_inbound(ident):
+
+    # lookup release ident, ensure it exists
+    try:
+        release = api.get_release(ident)
+    except ApiException as ae:
+        abort(ae.status)
+
+    offset = request.args.get('offset', '0')
+    offset = max(0, int(offset)) if offset.isnumeric() else 0
+
+    hits = get_inbound_refs(release_ident=ident, es_client=app.es_client, offset=offset, limit=30)
+    enriched_refs = enrich_inbound_refs_fatcat(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
+
+    return render_template('release_view_fuzzy_refs.html', direction="inbound", entity=release, hits=hits, enriched_refs=enriched_refs), 200
+
+@app.route('/release/<string(length=26):ident>/refs/outbound', methods=['GET'])
+def release_view_refs_outbound(ident):
+
+    # lookup release ident, ensure it exists
+    try:
+        release = api.get_release(ident)
+    except ApiException as ae:
+        abort(ae.status)
+
+    offset = request.args.get('offset', '0')
+    offset = max(0, int(offset)) if offset.isnumeric() else 0
+
+    hits = get_outbound_refs(release_ident=ident, es_client=app.es_client, offset=offset, limit=30)
+    enriched_refs = enrich_outbound_refs_fatcat(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
+
+    return render_template('release_view_fuzzy_refs.html', direction="outbound", entity=release, hits=hits, enriched_refs=enriched_refs), 200
-- 
cgit v1.2.3


From 314aba35d06eb80be0c5ffc068774bb9bca38e76 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Thu, 15 Apr 2021 23:31:07 -0700
Subject: web: initial implementation of fuzzy citation parsing and matching
 tool

---
 python/fatcat_web/forms.py                       | 41 +++++++++++
 python/fatcat_web/ref_routes.py                  | 46 +++++++++++++
 python/fatcat_web/templates/reference_match.html | 86 ++++++++++++++++++++++++
 3 files changed, 173 insertions(+)
 create mode 100644 python/fatcat_web/templates/reference_match.html

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_web/forms.py b/python/fatcat_web/forms.py
index 1c9fb199..19176a59 100644
--- a/python/fatcat_web/forms.py
+++ b/python/fatcat_web/forms.py
@@ -482,3 +482,44 @@ class EntityTomlForm(EntityEditForm):
 
         etf.toml.data = entity_to_toml(entity, pop_fields=pop_fields)
         return etf
+
+class ReferenceMatchForm(FlaskForm):
+
+    submit_type = SelectField('submit_type',
+        [validators.DataRequired()],
+        choices=['parse', 'match'])
+
+    raw_citation = TextAreaField("Citation String", render_kw={'rows':'3'})
+
+    title = StringField("Title")
+    journal = StringField("Journal or Conference")
+    first_author = StringField("First Author")
+    #year = IntegerField('Year Released',
+    #    [validators.Optional(True), valid_year])
+    year = StringField("Year Released")
+    volume = StringField("Volume")
+    issue = StringField("Issue")
+    pages = StringField("Pages")
+
+    @staticmethod
+    def from_grobid_parse(parse_dict, raw_citation):
+        """
+        Initializes form from GROBID extraction
+        """
+        rmf = ReferenceMatchForm()
+        rmf.raw_citation.data = raw_citation
+
+        direct_fields = ['title', 'journal', 'volume', 'issue', 'pages']
+        for k in direct_fields:
+            if parse_dict.get(k):
+                a = getattr(rmf, k)
+                a.data = parse_dict[k]
+
+        date = parse_dict.get('date')
+        if date and len(date) >= 4 and date[0:4].isdigit():
+            rmf.year.data = int(date[0:4])
+
+        if parse_dict.get('authors'):
+            rmf.first_author.data = parse_dict['authors'][0].get('name')
+
+        return rmf
diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index a49813c4..dc39299f 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -8,8 +8,11 @@ from typing import Optional
 from flask import render_template, abort, redirect, request
 from fatcat_openapi_client import *
 from fatcat_openapi_client.rest import ApiException
+from fuzzycat.grobid_unstructured import grobid_api_process_citation, transform_grobid_ref_xml, grobid_ref_to_release
+from fuzzycat.simple import close_fuzzy_biblio_matches, close_fuzzy_release_matches
 
 from fatcat_tools.references import enrich_inbound_refs_fatcat, enrich_outbound_refs_fatcat, get_inbound_refs, get_outbound_refs
+from fatcat_tools.transforms.access import release_access_options
 from fatcat_web import app, api, auth_api
 from fatcat_web.forms import *
 from fatcat_web.entity_helpers import *
@@ -48,3 +51,46 @@ def release_view_refs_outbound(ident):
     enriched_refs = enrich_outbound_refs_fatcat(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
 
     return render_template('release_view_fuzzy_refs.html', direction="outbound", entity=release, hits=hits, enriched_refs=enriched_refs), 200
+
+@app.route('/reference/match', methods=['GET', 'POST'])
+def reference_match():
+
+    form = ReferenceMatchForm()
+    grobid_status = None
+    grobid_dict = None
+
+    if form.is_submitted():
+        if form.validate_on_submit():
+            if form.submit_type.data == 'parse':
+                resp_xml = grobid_api_process_citation(form.raw_citation.data)
+                if not resp_xml:
+                    grobid_status = "failed"
+                    return render_template('reference_match.html', form=form, grobid_status=grobid_status), 400
+                grobid_dict = transform_grobid_ref_xml(resp_xml)
+                if not grobid_dict:
+                    grobid_status = "empty"
+                    return render_template('reference_match.html', form=form, grobid_status=grobid_status), 200
+                #print(grobid_dict)
+                release_stub = grobid_ref_to_release(grobid_dict)
+                # remove empty values from GROBID parsed dict
+                grobid_dict = {k: v for k, v in grobid_dict.items() if v is not None}
+                form = ReferenceMatchForm.from_grobid_parse(grobid_dict, form.raw_citation.data)
+                grobid_status = "success"
+                matches = close_fuzzy_release_matches(es_client=app.es_client, release=release_stub, match_limit=10) or []
+            elif form.submit_type.data == 'match':
+                matches = close_fuzzy_biblio_matches(es_client=app.es_client, biblio=form.data, match_limit=10) or []
+            else:
+                raise NotImplementedError()
+
+            for m in matches:
+                # expand releases more completely
+                m.release = api.get_release(m.release.ident, expand="container,files,filesets,webcaptures", hide="abstract,refs")
+                # hack in access options
+                m.access_options = release_access_options(m.release)
+
+            return render_template('reference_match.html', form=form, grobid_dict=grobid_dict, grobid_status=grobid_status, matches=matches), 200
+
+        elif form.errors:
+            return render_template('reference_match.html', form=form), 400
+
+    return render_template('reference_match.html', form=form), 200
diff --git a/python/fatcat_web/templates/reference_match.html b/python/fatcat_web/templates/reference_match.html
new file mode 100644
index 00000000..042b0607
--- /dev/null
+++ b/python/fatcat_web/templates/reference_match.html
@@ -0,0 +1,86 @@
+{% extends "base.html" %}
+{% import "entity_macros.html" as entity_macros %}
+{% import "edit_macros.html" as edit_macros %}
+
+{% block body %}
+
+<h1>Reference Fuzzy Match Tool</h1>
+
+<form class="ui form" id="reference_match" method="POST" action="/reference/match">
+  <input type="hidden" name="csrf_token" value="{{ csrf_token() }}"/>
+
+  <h3>Parse Citation</h3>
+
+  <p>Enter a citation string here and we will try to parse it (using GROBID)
+  into a structured format, then match against the catalog.
+
+  {{ edit_macros.form_field_basic(form.raw_citation) }}
+
+  <button class="ui primary submit button right floated" type="submit" name="submit_type" value="parse">
+    Parse
+  </button>
+
+  <br clear="all">
+  {% if grobid_status == "success" and grobid_dict %}
+    <div class="ui positive message">
+      <div class="header">Parsed successfully! See match results below</div>
+      {{ entity_macros.extra_metadata(grobid_dict) }}
+    </div>
+  {% endif %}
+
+  <br>
+  <hr>
+  <h3>Fuzzy Match Metadata</h3>
+
+  <p>Enter whatever bibliographic metadata fields you know, and we will try to
+  match to catalog entries.
+
+  <p><b>NOTE:</b> if you already know a persistent identifier (like a DOI), you
+  should use the <a href="/release/lookup">lookup tool</a> instead.
+
+    {{ edit_macros.form_field_inline(form.title) }}
+    {{ edit_macros.form_field_inline(form.first_author) }}
+
+    <br>
+    <div class="ui equal width fields">
+      {{ edit_macros.form_field_basic(form.year) }}
+      {{ edit_macros.form_field_basic(form.journal) }}
+    </div>
+    <div class="ui equal width fields">
+      {{ edit_macros.form_field_basic(form.volume) }}
+      {{ edit_macros.form_field_basic(form.issue) }}
+      {{ edit_macros.form_field_basic(form.pages) }}
+    </div>
+
+  <button class="ui primary submit button right floated" type="submit" name="submit_type" value="match">
+    Match
+  </button>
+  <br clear="all">
+
+</form>
+
+{% if matches is defined %}
+  <br>
+  <hr>
+  <h3>Match Results</h3>
+  <table class="ui very basic celled table">
+  <tbody>
+  {% for match in matches %}
+    <tr><td class="collapsing center aligned">
+          <br><b>{{ match.status.name }}</b>
+          <br>{{ match.reason.name }}
+        <td class="">
+          {{ entity_macros.release_summary(match.release) }}
+        <td class="">
+          {% if match.access_options %}
+            <a href="{{ match.access_options[0].access_url}}" class="ui tiny green active button">{{ match.access_options[0].access_type.name }}</a>
+            {% endif %}
+  {% endfor %}
+  </tbody>
+  </table>
+  {% if not matches %}
+    <p><i>None!</i>
+  {% endif %}
+{% endif %}
+
+{% endblock %}
-- 
cgit v1.2.3


From 570074b514259bf6345c376faea8128f279bd0b4 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Thu, 15 Apr 2021 23:48:06 -0700
Subject: web: inbound/outbound refs as links (temporarily); change URL names

---
 python/fatcat_web/ref_routes.py                          | 4 ++--
 python/fatcat_web/templates/entity_base.html             | 4 ++++
 python/fatcat_web/templates/release_view_fuzzy_refs.html | 2 +-
 3 files changed, 7 insertions(+), 3 deletions(-)

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index dc39299f..bd8ae550 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -18,7 +18,7 @@ from fatcat_web.forms import *
 from fatcat_web.entity_helpers import *
 
 
-@app.route('/release/<string(length=26):ident>/refs/inbound', methods=['GET'])
+@app.route('/release/<string(length=26):ident>/inbound-refs', methods=['GET'])
 def release_view_refs_inbound(ident):
 
     # lookup release ident, ensure it exists
@@ -35,7 +35,7 @@ def release_view_refs_inbound(ident):
 
     return render_template('release_view_fuzzy_refs.html', direction="inbound", entity=release, hits=hits, enriched_refs=enriched_refs), 200
 
-@app.route('/release/<string(length=26):ident>/refs/outbound', methods=['GET'])
+@app.route('/release/<string(length=26):ident>/outbound-refs', methods=['GET'])
 def release_view_refs_outbound(ident):
 
     # lookup release ident, ensure it exists
diff --git a/python/fatcat_web/templates/entity_base.html b/python/fatcat_web/templates/entity_base.html
index 36280f5d..c23dbef2 100644
--- a/python/fatcat_web/templates/entity_base.html
+++ b/python/fatcat_web/templates/entity_base.html
@@ -86,6 +86,10 @@
             {% elif entity_type == "release" and entity.state != 'deleted' %}
               {{ entity_tab("contribs", "Authors", "/contribs", entity._authors|count ) }}
               {{ entity_tab("references", "References", "/references", entity.refs|count) }}
+              {% if  entity.state == 'active' %}
+                {{ entity_tab("inbound-refs", "Inbound", "/inbound-refs") }}
+                {{ entity_tab("outbound-refs", "Outbound", "/outbound-refs") }}
+              {% endif %}
             {% endif %}
             {{ entity_tab("metadata", "Metadata", "/metadata") }}
           </div>
diff --git a/python/fatcat_web/templates/release_view_fuzzy_refs.html b/python/fatcat_web/templates/release_view_fuzzy_refs.html
index bc1fa171..9ceb6060 100644
--- a/python/fatcat_web/templates/release_view_fuzzy_refs.html
+++ b/python/fatcat_web/templates/release_view_fuzzy_refs.html
@@ -1,5 +1,5 @@
 {% set release = entity %}
-{% set entity_view = "references" %}
+{% set entity_view = "{{ direction }}-refs" %}
 {% set entity_type = "release" %}
 {% import "entity_macros.html" as entity_macros %}
 {% extends "entity_base.html" %}
-- 
cgit v1.2.3


From 61ed521cc40c1ee76692e9c4054e89fa63320600 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Thu, 24 Jun 2021 18:45:08 -0700
Subject: improvements to fuzzy refs view

- fixes to release summary macro
- show tab counts correctly by re-using generic entity get helper
- table styling; 'prev' link
- openlibrary access links
- parse-and-match button for unmatched+unstructured refs
---
 python/fatcat_web/ref_routes.py                    | 12 +--
 python/fatcat_web/templates/entity_macros.html     | 17 ++--
 .../templates/release_view_fuzzy_refs.html         | 93 +++++++++++++++-------
 3 files changed, 75 insertions(+), 47 deletions(-)

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index bd8ae550..e08aaf15 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -21,11 +21,7 @@ from fatcat_web.entity_helpers import *
 @app.route('/release/<string(length=26):ident>/inbound-refs', methods=['GET'])
 def release_view_refs_inbound(ident):
 
-    # lookup release ident, ensure it exists
-    try:
-        release = api.get_release(ident)
-    except ApiException as ae:
-        abort(ae.status)
+    release = generic_get_entity("release", ident)
 
     offset = request.args.get('offset', '0')
     offset = max(0, int(offset)) if offset.isnumeric() else 0
@@ -38,11 +34,7 @@ def release_view_refs_inbound(ident):
 @app.route('/release/<string(length=26):ident>/outbound-refs', methods=['GET'])
 def release_view_refs_outbound(ident):
 
-    # lookup release ident, ensure it exists
-    try:
-        release = api.get_release(ident)
-    except ApiException as ae:
-        abort(ae.status)
+    release = generic_get_entity("release", ident)
 
     offset = request.args.get('offset', '0')
     offset = max(0, int(offset)) if offset.isnumeric() else 0
diff --git a/python/fatcat_web/templates/entity_macros.html b/python/fatcat_web/templates/entity_macros.html
index 94770afb..24d1b6d0 100644
--- a/python/fatcat_web/templates/entity_macros.html
+++ b/python/fatcat_web/templates/entity_macros.html
@@ -394,30 +394,29 @@ yellow
   {% if release.release_type not in ["article-journal", "conference-paper"] %}
     <b>[{{ release.release_type or "unknown-type" }}]</b>
   {% endif %}
-<br>
+  {% if release.contribs %}<br>{% endif %}
   {% for contrib in release.contribs[:5] %}
     {% if contrib.creator %}
-      {{ contrib.creator.display_name }}
+      <a href="/contib/{{ contrib.creator.ident }}" style="color: black;">{{ contrib.creator.display_name }}</a>
     {% else %}
       {{ contrib.raw_name }}
-    {% endif %}
-    {% if not loop.last %}, {% endif %}
+    {%- endif %}
+    {%- if not loop.last %}, {% endif %}
   {% endfor %}
-  {% if release.contribs | length > 5 %}(+ more) {%endif %}
-<br>
+  {% if release.contribs | length > 5 %} <i>(+ more)</i> {%endif %}
+  {% if release.release_year or release.container or (release.extra and release.extra.container_name) %}<br>{% endif %}
   {% if release.release_year %}{{ release.release_year }} &nbsp;{% endif %}
   {% if release.container %}
-    <i>{{ release.container.name }}</i>
+    <a href="/container/{{ release.container.ident }}" style="color: black;"><i>{{ release.container.name }}</i></a>
   {% elif release.extra and release.extra.container_name %}
     <i>{{ release.extra.container_name }}</i>
   {% endif %}
   {% if release.release_stage != "published" %}
     &nbsp;<span style="color: brown; font-weight: bold;">{{ release.release_stage or "unpublished" }}</span>
   {% endif %}
-</small>
 <br>
   {% if release.version %}
-    <span style="color:green">version:{{ release.release_year }}</span>&nbsp;
+    <span style="color:green">version:{{ release.version }}</span>&nbsp;
   {% endif %}
   {% if release.number %}
     <span style="color:green">number:{{ release.number }}</span>&nbsp;
diff --git a/python/fatcat_web/templates/release_view_fuzzy_refs.html b/python/fatcat_web/templates/release_view_fuzzy_refs.html
index ee39d15b..7b286fd3 100644
--- a/python/fatcat_web/templates/release_view_fuzzy_refs.html
+++ b/python/fatcat_web/templates/release_view_fuzzy_refs.html
@@ -4,53 +4,90 @@
 {% import "entity_macros.html" as entity_macros %}
 {% extends "entity_base.html" %}
 
+{% macro pagination_row(hits) %}
+
+  {% if hits.offset %}
+    <a href="?offset={{ hits.offset - hits.limit }}">&laquo; prev</a> &nbsp;
+  {% endif %}
+  Showing {{ hits.offset + 1 }} - {{ hits.offset + hits.count_returned }} of {{ hits.count_total}} references (in {{ hits.query_wall_time_ms }}ms)
+  {% if hits.count_total != hits.count_returned and hits.offset + hits.limit < hits.count_total %}
+    &nbsp;<a href="?offset={{ hits.offset + hits.limit }}">next &raquo;</a>
+  {% endif %}
+{% endmacro %}
+
 {% block entity_main %}
 
 {% if direction == "inbound" %}
-  <h3>Inbound Matched References</h3>
-  <i>Other releases citing this one</i>
+  <h3>Referenced By</h3>
+  <i>Citations to this release by other works.</i>
 {% elif direction == "outbound" %}
-  <h3>Outbound Matched References</h3>
-  <i>This release citing other releases</i>
-{% endif %}
-
-<p>Found {{ hits.count_total }} references in {{ hits.query_wall_time_ms }}ms.
-{% if hits.count_total != hits.count_returned %}
-  Showing {{ hits.offset + 1 }} - {{ hits.offset + hits.count_returned }}
-  {% if hits.offset + hits.limit < hits.count_total %}
-    &nbsp;<a href="?offset={{ hits.offset + hits.limit }}">next...</a>
-  {% endif %}
+  <h3>References</h3>
+  <i>NOTE: currently batch computed and may include additional references sources, or be missing recent changes, compared to entity reference list.</i>
 {% endif %}
 
-<table class="ui very basic celled table">
+{% if enriched_refs %}
+<table class="ui table">
+<thead>
+  <tr><th colspan="3">
+    {{ pagination_row(hits) }}
+</thead>
 <tbody>
-{% for ref in enriched_refs %}
-  {% set release = ref.release %}
-  <tr><td class="collapsing center aligned">
+{% for row in enriched_refs %}
+  {% set release = row.release %}
+  <tr><td class="collapsing left aligned top aligned">
+        {# TODO: ref_locator? #}
         {% if direction == "outbound" %}
-          {% if ref.ref.ref_key %}
-            <code>[{{ ref.ref.ref_key }}]</code>
+          {% if row.ref.ref_key %}
+            <code>[{{ row.ref.ref_key }}]</code><br>
           {% endif %}
         {% endif %}
-        <br><b>{{ ref.ref.match_status }}</b>
-        {% if ref.ref.match_provenance %}
-          <br>{{ ref.ref.match_provenance }}
+        <b>{{ row.ref.match_status }}</b><br>
+        {% if row.ref.match_provenance %}
+          via {{ row.ref.match_provenance }}
         {% endif %}
       <td class="">
         {% if release %}
-          {% entity_macros.release_summary(release) %}
-        {% elif ref.ref.target_unstructured %}
-          <code>{{ ref.ref.target_unstructured }}</code>
+          {{ entity_macros.release_summary(release) }}
+        {% elif row.ref.target_unstructured %}
+          <code>{{ row.ref.target_unstructured }}</code>
         {% else %}
           <i>blank</i>
         {% endif %}
-        <td class="">
-          {% if ref.access %}
-            <a href="{{ ref.access[0].access_url}}" class="ui tiny green active button">{{ ref.access[0].access_type.name }}</a>
-          {% endif %}
+      <td class="center aligned">
+        {% if row.access %}
+          {% for access in row.access %}
+            <a href="{{ access.access_url}}" class="ui tiny green active button">{{ access.access_type.name }}</a><br>
+          {% endfor %}
+        {% elif row.ref.target_unstructured %}
+          <form class="ui form" id="reference_match" method="POST" action="/reference/match">
+            <input type="hidden" name="csrf_token" value="{{ csrf_token() }}"/>
+            <input type="hidden" name="raw_citation" value="{{ row.ref.target_unstructured }}">
+            <button class="ui tiny primary submit button" type="submit" name="submit_type" value="parse">
+              parse
+            </button>
+          </form>
+        {% endif %}
+        
+        {# TODO: include these as access options instead #}
+        {% if row.ref.target_openlibrary_work %}
+          <a href="https://openlibrary.org/work/{{ row.ref.target_openlibrary_work }}" class="ui tiny green active button">openlibrary.org</a>
+        {% endif %}
+        {% if row.ref.target_url %}
+          <a href="{{ ref.target_url | safe }}" class="ui tiny green active button">web</a>
+          <br><a href="https://web.archive.org/web/*/{{ row.ref.target_url | safe }}" class="ui tiny green active button">wayback (?)</a>
+        {% endif %}
 {% endfor %}
 </tbody>
+{% if hits.count_total != hits.count_returned %}
+  <tfoot>
+    <tr><th colspan="3">
+      {{ pagination_row(hits) }}
+  </tfoot>
+{% endif %}
 </table>
+{% else %}
+  <br><br><p><b>None found</b>
+{% endif %}
 
 {% endblock %}
 
-- 
cgit v1.2.3


From 7489ef7a979574effa74f1f17cebb81eefb1b71a Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Fri, 23 Jul 2021 11:56:42 -0700
Subject: refs: refactor web paths; enrich refs as generic; remove old refs
 link

---
 proposals/2021-01-29_citation_api.md               | 102 ++++++++-------------
 python/fatcat_tools/references.py                  |  85 +++++++----------
 python/fatcat_web/ref_routes.py                    |  16 ++--
 python/fatcat_web/templates/entity_base.html       |   5 +-
 .../templates/release_view_fuzzy_refs.html         |  12 +--
 5 files changed, 91 insertions(+), 129 deletions(-)

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/proposals/2021-01-29_citation_api.md b/proposals/2021-01-29_citation_api.md
index 1e329d61..f8d9e676 100644
--- a/proposals/2021-01-29_citation_api.md
+++ b/proposals/2021-01-29_citation_api.md
@@ -41,13 +41,13 @@ into a columnar file format like Parquet to get storage efficiency advances,
 type/schema enforcement, and easier ingest and use for large-scale data
 analysis.
 
-TODO: more?
-
 
 ## Schemas
 
 First, a combined JSON/pydantic/elasticsearch object that represents a
-reference between two things:
+reference from one thing to another, where the "source" must be known, but the
+"target" may either be known ("matched") or ambiguous (eg, just a reference
+string):
 
     BiblioRef ("bibliographic reference")
         _key: Optional[str] elasticsearch doc key
@@ -60,8 +60,6 @@ reference between two things:
         source_work_ident: Optional[str]
         source_wikipedia_article: Optional[str]
             with lang prefix like "en:Superglue"
-        # skipped: source_openlibrary_work
-        # skipped: source_url_surt
         source_release_stage: Optional[str]
         source_year: Optional[int]
 
@@ -71,7 +69,9 @@ reference between two things:
         ref_key: Optional[str]
             eg, "Lee86", "BIB23"
         ref_locator: Optional[str]
-            eg, page number
+            eg, specific page number in the book being referenced, if
+            applicable. Not used for, eg, first page of paper in a
+            volume/issue.
 
         # target of reference (identifiers)
         target_release_ident: Optional[str]
@@ -82,15 +82,15 @@ reference between two things:
             would not be stored in elasticsearch, but would be auto-generated
             by all "get" methods from the SURT, so calling code does not need
             to do SURT transform
-        # skipped: target_wikipedia_article
 
         match_provenance: str
             crossref, pubmed, grobid, etc
+            TODO: "ref_provenance"
         match_status: Optional[str]
             strong, weak, etc
-            TODO: "match_strength"?
+            TODO: "match_strength"? "match_confidence"?
         match_reason: Optional[str]
-            "doi", "isbn", "fuzzy title, author", etc
+            "doi", "isbn", "title-fuzzy, author", etc
             maybe "fuzzy-title-author"?
 
         target_unstructured: string (only if no release_ident link/match)
@@ -116,33 +116,22 @@ jinja templated to display lists of references in the user interface.
         size_bytes: Optional[int]
         thumbnail_url: Optional[str]
 
-    CslBiblioRef
-        # an "enriched" version of BiblioRef with metadata about the source or
-        # target entity. would be "hydrated" via a lookup to, eg, the
-        # `fatcat_release` elasticsearch index (fast mget fetch with a single
-        # request), as opposed to fatcat API fetches
-        biblio_ref: BiblioRef
-        source_csl/target_csl: free-form CSL-JSON
-        source_access/target_access: List[AccessOption]
-
-    FatcatBiblioRef
+    EnrichedBiblioRef
         # enriched version of BiblioRef with complete ReleaseEntity object as
-        # fetched from the fatcat API. CSL-JSON metadata would be derived from
-        # the full release entity.
+        # fetched from entity catalogs, if available. For example, fatcat API.
         biblio_ref: BiblioRef
         source_release/target_release: Optional[ReleaseEntity]
             complete ReleaseEntity from API, with optional expand/hide fields
-        source_csl/target_csl: free-form CSL-JSON
-            CSL-JSON version of ReleaseEntity metadata
         source_access/target_access: List[AccessOption]
+        # TODO: target_openlibrary? source_wikipedia?
 
 
 ## Datastore
 
 Would store in Elasticsearch as a live database, at least to start.
 
-TODO: try generating ~1 million of these objects to estimate index size (at
-billions of docs).
+Example Elasticsearch index `fatcat_ref_v02_20210716` has 1.8 billion docs
+(references), and consumes 435 GBytes of disk.
 
 Might be reasonable to use PostgreSQL in the future, with more explicit control
 over indexes and tuning for latency. But Elasticsearch is pretty easy to
@@ -172,59 +161,46 @@ operate (eg, replicas).
     count_inbound_refs(...) -> int
         same parameters as get_inbound_refs(), but returns just a count
 
-    get_all_outbound_refs(...) -> List[BiblioRef]
-    get_all_inbound_refs(...) -> List[BiblioRef]
-        same as get_outbound_refs()/get_inbound_refs(), but does a scroll (return list or iterator?)
-        (optional; maybe not public)
+    # UNIMPLEMENTED
+    #get_all_outbound_refs(...) -> List[BiblioRef]
+    #get_all_inbound_refs(...) -> List[BiblioRef]
+    #    same as get_outbound_refs()/get_inbound_refs(), but does a scroll (return list or iterator?)
+    #    (optional; maybe not public)
 
-    # run elasticsearch mget query for all ref idents and include "enriched" refs when possible
-    # for outbound URL refs, would do wayback CDX fetches to find a direct wayback URL
-    # TODO: for openlibrary, would this query openlibrary.org API? or some fatcat-specific index?
-    enrich_inbound_refs(refs: List[BiblioRef]) -> List[CslBiblioRef]
-    enrich_outbound_refs(refs: List[BiblioRef]) -> List[CslBiblioRef]
-
-    # run fatcat API fetches for each ref and return "enriched" refs
-    enrich_inbound_refs_fatcat(refs: List[BiblioRef], hide, expand) -> List[FatcatBiblioRef]
-    enrich_outbound_refs_fatcat(refs: List[BiblioRef], hide, expand) -> List[FatcatBiblioRef]
+    # run catalog API fetches for each and return "enriched" refs
+    enrich_inbound_refs(refs: List[BiblioRef], hide, expand) -> List[EnrichedBiblioRef]
+    enrich_outbound_refs(refs: List[BiblioRef], hide, expand) -> List[EnrichedBiblioRef]
 
 ## HTTP API Endpoints
 
-Possible HTTP API endpoints... not even sure we would use these or expose them
-publicly?
-
-    citations-api.fatcat.wiki
-        /refs/inbound
-            &release_ident=
-            &work_ident=
-            &openlibrary_work=
-            &url=
-        /refs/outbound
-            &release_ident=
-            &work_ident=
-        /refs/csl/outbound
-        /refs/fatcat/outbound
-
-    api.fatcat.wiki/citations/v0
-        /inbound
-
-    fatcat.wiki/release/{release_ident}/refs/outbound.json
-    fatcat.wiki/work/{work_ident}/refs/outbound.json
-        &filter_type
-        &filter_stage
+Initial web endpoints, including unstable pseudo-APIs:
+
+    fatcat.wiki/release/{release_ident}/refs/in (and .json)
+    fatcat.wiki/release/{release_ident}/refs/out (and .json)
         &limit
         &offset
+        &sort (for inbound)
+        &filter_stage (for inbound)
 
-    fatcat.wiki/refs/openlibrary/{openlibrary_ident}/inbound.json
+    fatcat.wiki/openlibrary/{openlibrary_ident}/refs/in (and .json)
+        &limit
+        &offset
+        &sort
+        &filter_stage
 
-    fatcat.wiki/refs/url/inbound.json
+    fatcat.wiki/web/refs/in (and .json)
         &url=
+        &limit
+        &offset
+        &sort (newest, oldest)
+        &filter_stage
 
 ## Design Notes
 
 This proposed schema is relatively close to what the "normalize" SQL table
 would look like (many-to-many relationship).
 
-Especiall for "redistributing as bulk corpus", we might want to consider an
+Especially for "redistributing as bulk corpus", we might want to consider an
 alternative data model which is a single source entity containing a list of
 outbound references. Could even be a single source *work* for fatcat content,
 with many release under the entity. One advantage of this is that source
diff --git a/python/fatcat_tools/references.py b/python/fatcat_tools/references.py
index 1d8a0d0d..a0079efd 100644
--- a/python/fatcat_tools/references.py
+++ b/python/fatcat_tools/references.py
@@ -3,26 +3,13 @@ Helper routines for working with the fatcat citation graph, which is a separate
 index of reference links between works in the main catalog.
 
 See bulk citation and citation API proposals for design documentation.
-
-TODO:
-
-    surt_ify() helper (URL to SURT for queries)
-    CSL enrichment method (using only elasticsearch mget)
-    CSL enrichment for fatcat enrichment
-    access transform
-    microfilm access in access transform
-
-    all_outbound_refs(...) -> List[BiblioRef]
-    all_inbound_refs(...) -> List[BiblioRef]
-        same as get_outbound_refs()/get_inbound_refs(), but does a scroll (return list or iterator?)
-        (optional; maybe not public)
 """
 
 import sys
 import json
 import datetime
 import argparse
-from typing import Optional, List, Any, Dict
+from typing import Optional, List, Any, Dict, Union
 
 from pydantic import BaseModel
 import elasticsearch
@@ -45,8 +32,6 @@ class BiblioRef(BaseModel):
     source_work_ident: Optional[str]
     # with lang prefix like "en:Superglue"
     source_wikipedia_article: Optional[str]
-    # skipped: source_openlibrary_work
-    # skipped: source_url_surt
     source_release_stage: Optional[str]
     source_year: Optional[int]
 
@@ -65,7 +50,6 @@ class BiblioRef(BaseModel):
     target_url_surt: Optional[str]
     # would not be stored in elasticsearch, but would be auto-generated by all "get" methods from the SURT, so calling code does not need to do SURT transform
     target_url: Optional[str]
-    # skipped: target_wikipedia_article
 
     # crossref, pubmed, grobid, etc
     match_provenance: Optional[str]
@@ -92,31 +76,20 @@ class BiblioRef(BaseModel):
         # TODO: if target_openlibrary_work, add an access option?
         return self
 
-class CslBiblioRef(BaseModel):
-    # an "enriched" version of BiblioRef with metadata about the source or
-    # target entity. would be "hydrated" via a lookup to, eg, the
-    # `fatcat_release` elasticsearch index (fast mget fetch with a single
-    # request), as opposed to fatcat API fetches
-    ref: BiblioRef
-    csl: Optional[Dict[str, Any]]
-    access: List[AccessOption]
-
-    class Config:
-        arbitrary_types_allowed = True
 
-class FatcatBiblioRef(BaseModel):
+class EnrichedBiblioRef(BaseModel):
     # enriched version of BiblioRef with complete ReleaseEntity object as
     # fetched from the fatcat API. CSL-JSON metadata would be derived from
     # the full release entity.
     ref: BiblioRef
     release: Optional[ReleaseEntity]
     # TODO: openlibrary work?
-    #csl: Optional[Dict[str, Any]]
     access: List[AccessOption]
 
     class Config:
         arbitrary_types_allowed = True
 
+
 class RefHits(BaseModel):
     count_returned: int
     count_total: int
@@ -124,9 +97,13 @@ class RefHits(BaseModel):
     limit: int
     query_time_ms: int
     query_wall_time_ms: int
-    result_refs: List[BiblioRef]
+    result_refs: List[Union[BiblioRef,EnrichedBiblioRef]]
 
-def _execute_ref_query(search: Any, limit: int, offset: Optional[int] = None) -> List[BiblioRef]:
+
+def _execute_ref_query(search: Any, limit: int, offset: Optional[int] = None) -> RefHits:
+    """
+    Internal helper for querying elasticsearch refs index and transforming hits
+    """
 
     limit = min((int(limit or 15), 200))
     if not offset or offset < 0:
@@ -179,7 +156,7 @@ def get_outbound_refs(
     limit: int = 100,
     offset: Optional[int] = None,
     es_index: str = "fatcat_ref",
-) -> List[BiblioRef]:
+) -> RefHits:
 
     search = Search(using=es_client, index=es_index)
 
@@ -199,6 +176,7 @@ def get_outbound_refs(
     hits.result_refs = sorted(hits.result_refs, key=lambda r: r.ref_index or 0)
     return hits
 
+
 def get_inbound_refs(
     es_client: Any,
     release_ident: Optional[str] = None,
@@ -208,12 +186,11 @@ def get_inbound_refs(
     url: Optional[str] = None,
     consolidate_works: bool = True,
     filter_stage: List[str] = [],
-    filter_type: List[str] = [],
+    sort: Optional[str] = None,
     limit: int = 25,
     offset: Optional[int] = None,
     es_index: str = "fatcat_ref",
 ) -> List[BiblioRef]:
-    # TODO: filter_stage, filter_type
 
     if url and not url_surt:
         url = surt_ify(url)
@@ -239,10 +216,19 @@ def get_inbound_refs(
     else:
         raise ValueError("require a lookup key")
 
-    search = search.sort("-source_year")
+    if filter_stage:
+        search = search.filter("term", source_stage=filter_stage)
+
+    if sort == "newest":
+        search = search.sort("-source_year")
+    elif sort == "oldest":
+        search = search.sort("source_year")
+    else:
+        search = search.sort("-source_year")
 
     return _execute_ref_query(search, limit=limit, offset=offset)
 
+
 def count_inbound_refs(
     es_client: Any,
     release_ident: Optional[str] = None,
@@ -251,7 +237,6 @@ def count_inbound_refs(
     url_surt: Optional[str] = None,
     url: Optional[str] = None,
     filter_stage: List[str] = [],
-    filter_type: List[str] = [],
     es_index: str = "fatcat_ref",
 ) -> int:
     """
@@ -274,28 +259,26 @@ def count_inbound_refs(
     else:
         raise ValueError("require a lookup key")
 
+    if filter_stage:
+        search = search.filter("term", source_stage=filter_stage)
+
     return search.count()
 
-# run elasticsearch mget query for all ref idents and include "enriched" refs when possible
-# for outbound URL refs, would do wayback CDX fetches to find a direct wayback URL
-# TODO: for openlibrary, would this query openlibrary.org API? or some fatcat-specific index?
-#enrich_inbound_refs(refs: List[BiblioRef]) -> List[CslBiblioRef]
-#enrich_outbound_refs(refs: List[BiblioRef]) -> List[CslBiblioRef]
 
 # run fatcat API fetches for each ref and return "enriched" refs
-def enrich_inbound_refs_fatcat(refs: List[BiblioRef], fatcat_api_client: Any, hide: Optional[str] = "refs", expand: Optional[str] = "container,files,webcaptures,filesets") -> List[FatcatBiblioRef]:
+def enrich_inbound_refs(refs: List[BiblioRef], fatcat_api_client: Any, hide: Optional[str] = "refs", expand: Optional[str] = "container,files,webcaptures,filesets") -> List[EnrichedBiblioRef]:
     enriched = []
     for ref in refs:
         if ref.source_release_ident:
             release = fatcat_api_client.get_release(ref.source_release_ident, hide=hide, expand=expand)
-            enriched.append(FatcatBiblioRef(
+            enriched.append(EnrichedBiblioRef(
                 ref=ref,
                 #csl=None,
                 access=release_access_options(release),
                 release=release,
             ))
         else:
-            enriched.append(FatcatBiblioRef(
+            enriched.append(EnrichedBiblioRef(
                 ref=ref,
                 #csl=None,
                 access=[],
@@ -303,21 +286,20 @@ def enrich_inbound_refs_fatcat(refs: List[BiblioRef], fatcat_api_client: Any, hi
             ))
     return enriched
 
-def enrich_outbound_refs_fatcat(refs: List[BiblioRef], fatcat_api_client: Any, hide: Optional[str] = "refs", expand: Optional[str] = "container,files,webcaptures,filesets") -> List[FatcatBiblioRef]:
+
+def enrich_outbound_refs(refs: List[BiblioRef], fatcat_api_client: Any, hide: Optional[str] = "refs", expand: Optional[str] = "container,files,webcaptures,filesets") -> List[EnrichedBiblioRef]:
     enriched = []
     for ref in refs:
         if ref.target_release_ident:
             release = fatcat_api_client.get_release(ref.target_release_ident, hide=hide, expand=expand)
-            enriched.append(FatcatBiblioRef(
+            enriched.append(EnrichedBiblioRef(
                 ref=ref,
-                #csl=None,
                 access=release_access_options(release),
                 release=release,
             ))
         else:
-            enriched.append(FatcatBiblioRef(
+            enriched.append(EnrichedBiblioRef(
                 ref=ref,
-                #csl=None,
                 access=[],
                 release=None,
             ))
@@ -325,6 +307,9 @@ def enrich_outbound_refs_fatcat(refs: List[BiblioRef], fatcat_api_client: Any, h
 
 
 def run_ref_query(args) -> None:
+    """
+    CLI helper/debug tool (prints to stdout)
+    """
     release_ident = None
     work_ident = None
     if args.ident.startswith("release_"):
diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index e08aaf15..e24b4ac6 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -11,14 +11,14 @@ from fatcat_openapi_client.rest import ApiException
 from fuzzycat.grobid_unstructured import grobid_api_process_citation, transform_grobid_ref_xml, grobid_ref_to_release
 from fuzzycat.simple import close_fuzzy_biblio_matches, close_fuzzy_release_matches
 
-from fatcat_tools.references import enrich_inbound_refs_fatcat, enrich_outbound_refs_fatcat, get_inbound_refs, get_outbound_refs
+from fatcat_tools.references import enrich_inbound_refs, enrich_outbound_refs, get_inbound_refs, get_outbound_refs
 from fatcat_tools.transforms.access import release_access_options
 from fatcat_web import app, api, auth_api
 from fatcat_web.forms import *
 from fatcat_web.entity_helpers import *
 
 
-@app.route('/release/<string(length=26):ident>/inbound-refs', methods=['GET'])
+@app.route('/release/<string(length=26):ident>/refs/in', methods=['GET'])
 def release_view_refs_inbound(ident):
 
     release = generic_get_entity("release", ident)
@@ -27,11 +27,12 @@ def release_view_refs_inbound(ident):
     offset = max(0, int(offset)) if offset.isnumeric() else 0
 
     hits = get_inbound_refs(release_ident=ident, es_client=app.es_client, offset=offset, limit=30)
-    enriched_refs = enrich_inbound_refs_fatcat(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
+    enriched_refs = enrich_inbound_refs(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
 
-    return render_template('release_view_fuzzy_refs.html', direction="inbound", entity=release, hits=hits, enriched_refs=enriched_refs), 200
+    return render_template('release_view_fuzzy_refs.html', direction="in", entity=release, hits=hits, enriched_refs=enriched_refs), 200
 
-@app.route('/release/<string(length=26):ident>/outbound-refs', methods=['GET'])
+
+@app.route('/release/<string(length=26):ident>/refs/out', methods=['GET'])
 def release_view_refs_outbound(ident):
 
     release = generic_get_entity("release", ident)
@@ -40,9 +41,10 @@ def release_view_refs_outbound(ident):
     offset = max(0, int(offset)) if offset.isnumeric() else 0
 
     hits = get_outbound_refs(release_ident=ident, es_client=app.es_client, offset=offset, limit=30)
-    enriched_refs = enrich_outbound_refs_fatcat(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
+    enriched_refs = enrich_outbound_refs(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
+
+    return render_template('release_view_fuzzy_refs.html', direction="out", entity=release, hits=hits, enriched_refs=enriched_refs), 200
 
-    return render_template('release_view_fuzzy_refs.html', direction="outbound", entity=release, hits=hits, enriched_refs=enriched_refs), 200
 
 @app.route('/reference/match', methods=['GET', 'POST'])
 def reference_match():
diff --git a/python/fatcat_web/templates/entity_base.html b/python/fatcat_web/templates/entity_base.html
index c23dbef2..78a151a0 100644
--- a/python/fatcat_web/templates/entity_base.html
+++ b/python/fatcat_web/templates/entity_base.html
@@ -85,10 +85,9 @@
               {{ entity_tab("coverage", "Coverage", "/coverage") }}
             {% elif entity_type == "release" and entity.state != 'deleted' %}
               {{ entity_tab("contribs", "Authors", "/contribs", entity._authors|count ) }}
-              {{ entity_tab("references", "References", "/references", entity.refs|count) }}
               {% if  entity.state == 'active' %}
-                {{ entity_tab("inbound-refs", "Inbound", "/inbound-refs") }}
-                {{ entity_tab("outbound-refs", "Outbound", "/outbound-refs") }}
+                {{ entity_tab("refs-out", "References", "/refs/out") }}
+                {{ entity_tab("refs-in", "Cited By", "/refs/in") }}
               {% endif %}
             {% endif %}
             {{ entity_tab("metadata", "Metadata", "/metadata") }}
diff --git a/python/fatcat_web/templates/release_view_fuzzy_refs.html b/python/fatcat_web/templates/release_view_fuzzy_refs.html
index 7b286fd3..43860a31 100644
--- a/python/fatcat_web/templates/release_view_fuzzy_refs.html
+++ b/python/fatcat_web/templates/release_view_fuzzy_refs.html
@@ -1,5 +1,5 @@
 {% set release = entity %}
-{% set entity_view = "{{ direction }}-refs" %}
+{% set entity_view = "refs-" + direction %}
 {% set entity_type = "release" %}
 {% import "entity_macros.html" as entity_macros %}
 {% extends "entity_base.html" %}
@@ -17,10 +17,10 @@
 
 {% block entity_main %}
 
-{% if direction == "inbound" %}
-  <h3>Referenced By</h3>
-  <i>Citations to this release by other works.</i>
-{% elif direction == "outbound" %}
+{% if direction == "in" %}
+  <h3>Cited By</h3>
+  <i>References to this release by other works.</i>
+{% elif direction == "out" %}
   <h3>References</h3>
   <i>NOTE: currently batch computed and may include additional references sources, or be missing recent changes, compared to entity reference list.</i>
 {% endif %}
@@ -36,7 +36,7 @@
   {% set release = row.release %}
   <tr><td class="collapsing left aligned top aligned">
         {# TODO: ref_locator? #}
-        {% if direction == "outbound" %}
+        {% if direction == "out" %}
           {% if row.ref.ref_key %}
             <code>[{{ row.ref.ref_key }}]</code><br>
           {% endif %}
-- 
cgit v1.2.3


From d01ab339a0aa568b6ccd6c56beb611a2e7ac9686 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Fri, 23 Jul 2021 11:58:41 -0700
Subject: remove unused imports (lint)

---
 python/fatcat_tools/references.py        | 3 +--
 python/fatcat_tools/transforms/access.py | 2 +-
 python/fatcat_web/ref_routes.py          | 7 ++-----
 3 files changed, 4 insertions(+), 8 deletions(-)

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_tools/references.py b/python/fatcat_tools/references.py
index da398239..67a16602 100644
--- a/python/fatcat_tools/references.py
+++ b/python/fatcat_tools/references.py
@@ -6,14 +6,13 @@ See bulk citation and citation API proposals for design documentation.
 """
 
 import sys
-import json
 import datetime
 import argparse
 from typing import Optional, List, Any, Dict, Union
 
 from pydantic import BaseModel
 import elasticsearch
-from elasticsearch_dsl import Search, Q
+from elasticsearch_dsl import Search
 from fatcat_openapi_client import ReleaseEntity
 
 from fatcat_tools import public_api
diff --git a/python/fatcat_tools/transforms/access.py b/python/fatcat_tools/transforms/access.py
index 231cd2b3..add8ff3b 100644
--- a/python/fatcat_tools/transforms/access.py
+++ b/python/fatcat_tools/transforms/access.py
@@ -1,6 +1,6 @@
 
 from enum import Enum
-from typing import Optional, List, Any, Dict
+from typing import Optional, List
 
 from pydantic import BaseModel
 from fatcat_openapi_client import ReleaseEntity
diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index e24b4ac6..7c3ba5bd 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -3,17 +3,14 @@ Flask endpoints for reference (citation) endpoints. Eg, listing references
 "inbound" and "outbound" from a specific release or work.
 """
 
-from typing import Optional
-
-from flask import render_template, abort, redirect, request
+from flask import render_template, request
 from fatcat_openapi_client import *
-from fatcat_openapi_client.rest import ApiException
 from fuzzycat.grobid_unstructured import grobid_api_process_citation, transform_grobid_ref_xml, grobid_ref_to_release
 from fuzzycat.simple import close_fuzzy_biblio_matches, close_fuzzy_release_matches
 
 from fatcat_tools.references import enrich_inbound_refs, enrich_outbound_refs, get_inbound_refs, get_outbound_refs
 from fatcat_tools.transforms.access import release_access_options
-from fatcat_web import app, api, auth_api
+from fatcat_web import app, api
 from fatcat_web.forms import *
 from fatcat_web.entity_helpers import *
 
-- 
cgit v1.2.3


From 05665c448e846e4834fa59dfe3cd0f11ac789ac9 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Fri, 23 Jul 2021 17:51:22 -0700
Subject: refs: generalize web endpoints; JSON content negotiation; openlibrary
 inbound view; etc

---
 python/fatcat_tools/references.py                  | 77 ++++++++++++-----
 python/fatcat_tools/transforms/access.py           |  2 +
 python/fatcat_web/ref_routes.py                    | 99 +++++++++++++++++-----
 .../templates/openlibrary_view_fuzzy_refs.html     | 29 +++++++
 4 files changed, 166 insertions(+), 41 deletions(-)
 create mode 100644 python/fatcat_web/templates/openlibrary_view_fuzzy_refs.html

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_tools/references.py b/python/fatcat_tools/references.py
index 73f57e18..81b55f41 100644
--- a/python/fatcat_tools/references.py
+++ b/python/fatcat_tools/references.py
@@ -10,13 +10,14 @@ import datetime
 import argparse
 from typing import Optional, List, Any, Dict, Union
 
-from pydantic import BaseModel
+from pydantic import BaseModel, validator
 import elasticsearch
 from elasticsearch_dsl import Search
 from fatcat_openapi_client import ReleaseEntity
 
 from fatcat_tools import public_api
 from fatcat_tools.transforms.access import release_access_options, AccessOption
+from fatcat_tools.transforms.entities import entity_to_dict
 
 
 class BiblioRef(BaseModel):
@@ -97,8 +98,17 @@ class EnrichedBiblioRef(BaseModel):
     # TODO: openlibrary work?
     access: List[AccessOption]
 
+    @validator('release')
+    def check_release(cls, v):
+        if v is not None and not isinstance(v, ReleaseEntity):
+            raise ValueError("expected a ReleaseEntity")
+        return v
+
     class Config:
         arbitrary_types_allowed = True
+        json_encoders = {
+            ReleaseEntity: entity_to_dict,
+        }
 
 
 class RefHits(BaseModel):
@@ -110,6 +120,11 @@ class RefHits(BaseModel):
     query_wall_time_ms: int
     result_refs: List[Union[BiblioRef,EnrichedBiblioRef]]
 
+    class Config:
+        json_encoders = {
+            ReleaseEntity: entity_to_dict,
+        }
+
 
 def _execute_ref_query(search: Any, limit: int, offset: Optional[int] = None) -> RefHits:
     """
@@ -268,40 +283,58 @@ def count_inbound_refs(
 def enrich_inbound_refs(refs: List[BiblioRef], fatcat_api_client: Any, hide: Optional[str] = "refs", expand: Optional[str] = "container,files,webcaptures,filesets") -> List[EnrichedBiblioRef]:
     enriched = []
     for ref in refs:
+        release = None
+        access = []
         if ref.source_release_ident:
             release = fatcat_api_client.get_release(ref.source_release_ident, hide=hide, expand=expand)
-            enriched.append(EnrichedBiblioRef(
-                ref=ref,
-                #csl=None,
-                access=release_access_options(release),
-                release=release,
-            ))
-        else:
-            enriched.append(EnrichedBiblioRef(
-                ref=ref,
-                #csl=None,
-                access=[],
-                release=None,
+            access = release_access_options(release)
+        if ref.source_wikipedia_article:
+            wiki_lang = ref.source_wikipedia.split(':')[0]
+            wiki_article = ':'.join(ref.source_wikipedia.split(':')[1:])
+            access.append(AccessOption(
+                access_type="wikipedia",
+                access_url=f"https://{wiki_lang}.wikipedia.org/wiki/{wiki_article}",
+                mimetype=None,
+                size_bytes=None,
+                thumbnail_url=None
             ))
+        enriched.append(EnrichedBiblioRef(
+            ref=ref,
+            access=access,
+            release=release,
+        ))
     return enriched
 
 
 def enrich_outbound_refs(refs: List[BiblioRef], fatcat_api_client: Any, hide: Optional[str] = "refs", expand: Optional[str] = "container,files,webcaptures,filesets") -> List[EnrichedBiblioRef]:
     enriched = []
     for ref in refs:
+        release = None
+        access = []
         if ref.target_release_ident:
             release = fatcat_api_client.get_release(ref.target_release_ident, hide=hide, expand=expand)
-            enriched.append(EnrichedBiblioRef(
-                ref=ref,
-                access=release_access_options(release),
-                release=release,
+            access = release_access_options(release)
+        if ref.target_openlibrary_work:
+            access.append(AccessOption(
+                access_type="openlibrary",
+                access_url=f"https://openlibrary.org/works/{ref.target_openlibrary_work}",
+                mimetype=None,
+                size_bytes=None,
+                thumbnail_url=None
             ))
-        else:
-            enriched.append(EnrichedBiblioRef(
-                ref=ref,
-                access=[],
-                release=None,
+        if ref.target_url and '://web.archive.org/' in ref.target_url:
+            access.append(AccessOption(
+                access_type="wayback",
+                access_url=ref.target_url,
+                mimetype=None,
+                size_bytes=None,
+                thumbnail_url=None
             ))
+        enriched.append(EnrichedBiblioRef(
+            ref=ref,
+            access=access,
+            release=release,
+        ))
     return enriched
 
 
diff --git a/python/fatcat_tools/transforms/access.py b/python/fatcat_tools/transforms/access.py
index add8ff3b..5ed64c7c 100644
--- a/python/fatcat_tools/transforms/access.py
+++ b/python/fatcat_tools/transforms/access.py
@@ -13,6 +13,8 @@ class AccessType(str, Enum):
     ia_file = "ia_file"
     ia_microfilm = "ia_microfilm"
     repository = "repository"
+    openlibrary = "openlibrary"
+    wikipedia = "wikipedia"
 
 class AccessOption(BaseModel):
 
diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index 7c3ba5bd..72f115cf 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -3,44 +3,82 @@ Flask endpoints for reference (citation) endpoints. Eg, listing references
 "inbound" and "outbound" from a specific release or work.
 """
 
-from flask import render_template, request
+from flask import render_template, request, jsonify, Response
 from fatcat_openapi_client import *
 from fuzzycat.grobid_unstructured import grobid_api_process_citation, transform_grobid_ref_xml, grobid_ref_to_release
 from fuzzycat.simple import close_fuzzy_biblio_matches, close_fuzzy_release_matches
 
-from fatcat_tools.references import enrich_inbound_refs, enrich_outbound_refs, get_inbound_refs, get_outbound_refs
+from fatcat_tools.references import enrich_inbound_refs, enrich_outbound_refs, get_inbound_refs, get_outbound_refs, RefHits
 from fatcat_tools.transforms.access import release_access_options
 from fatcat_web import app, api
+from fatcat_web.cors import crossdomain
 from fatcat_web.forms import *
 from fatcat_web.entity_helpers import *
 
-
-@app.route('/release/<string(length=26):ident>/refs/in', methods=['GET'])
-def release_view_refs_inbound(ident):
-
-    release = generic_get_entity("release", ident)
-
+def _refs_web(direction, release_ident=None, work_ident=None, openlibrary_id=None) -> RefHits:
     offset = request.args.get('offset', '0')
     offset = max(0, int(offset)) if offset.isnumeric() else 0
+    limit = request.args.get('offset', '30')
+    limit = min(max(0, int(limit)), 100) if limit.isnumeric() else 30
+    if direction == "in":
+        hits = get_inbound_refs(
+            release_ident=release_ident,
+            work_ident=work_ident,
+            openlibrary_work=openlibrary_id,
+            es_client=app.es_client,
+            offset=offset,
+            limit=limit,
+        )
+        hits.result_refs = enrich_inbound_refs(
+            hits.result_refs,
+            fatcat_api_client=api,
+            expand="container,files,webcaptures",
+        )
+    elif direction == "out":
+        hits = get_outbound_refs(
+            release_ident=release_ident,
+            work_ident=work_ident,
+            es_client=app.es_client,
+            offset=offset,
+            limit=limit,
+        )
+        hits.result_refs = enrich_outbound_refs(
+            hits.result_refs,
+            fatcat_api_client=api,
+            expand="container,files,webcaptures",
+        )
+    else:
+        raise ValueError()
+    return hits
+
+
+@app.route('/release/<string(length=26):ident>/refs-in', methods=['GET'])
+def release_view_refs_inbound(ident):
+    if request.accept_mimetypes.best == "application/json":
+        return release_view_refs_inbound_json(ident)
 
-    hits = get_inbound_refs(release_ident=ident, es_client=app.es_client, offset=offset, limit=30)
-    enriched_refs = enrich_inbound_refs(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
-
-    return render_template('release_view_fuzzy_refs.html', direction="in", entity=release, hits=hits, enriched_refs=enriched_refs), 200
+    release = generic_get_entity("release", ident)
+    hits = _refs_web("in", release_ident=ident)
+    return render_template('release_view_fuzzy_refs.html', direction="in", entity=release, hits=hits), 200
 
 
-@app.route('/release/<string(length=26):ident>/refs/out', methods=['GET'])
+@app.route('/release/<string(length=26):ident>/refs-out', methods=['GET'])
 def release_view_refs_outbound(ident):
+    if request.accept_mimetypes.best == "application/json":
+        return release_view_refs_outbound_json(ident)
 
     release = generic_get_entity("release", ident)
+    hits = _refs_web("out", release_ident=ident)
+    return render_template('release_view_fuzzy_refs.html', direction="out", entity=release, hits=hits), 200
 
-    offset = request.args.get('offset', '0')
-    offset = max(0, int(offset)) if offset.isnumeric() else 0
-
-    hits = get_outbound_refs(release_ident=ident, es_client=app.es_client, offset=offset, limit=30)
-    enriched_refs = enrich_outbound_refs(hits.result_refs, fatcat_api_client=api, expand="container,files,webcaptures")
+@app.route('/openlibrary/OL<int:id_num>W/refs-in', methods=['GET'])
+def openlibrary_view_refs_inbound(id_num):
+    if request.accept_mimetypes.best == "application/json":
+        return openlibrary_view_refs_inbound(id_num)
 
-    return render_template('release_view_fuzzy_refs.html', direction="out", entity=release, hits=hits, enriched_refs=enriched_refs), 200
+    openlibrary_id = f"OL{id_num}W"
+    hits = _refs_web("in", openlibrary_id=openlibrary_id)
+    return render_template('openlibrary_view_fuzzy_refs.html', openlibrary_id=openlibrary_id, direction="in", hits=hits), 200
 
 
 @app.route('/reference/match', methods=['GET', 'POST'])
@@ -85,3 +123,26 @@ def reference_match():
             return render_template('reference_match.html', form=form), 400
 
     return render_template('reference_match.html', form=form), 200
+
+
+### Pseudo-APIs #############################################################
+
+@app.route('/release/<string(length=26):ident>/refs-out.json', methods=['GET', 'OPTIONS'])
+@crossdomain(origin='*',headers=['access-control-allow-origin','Content-Type'])
+def release_view_refs_outbound_json(ident):
+    hits = _refs_web("out", release_ident=ident)
+    return Response(hits.json(exclude_unset=True), mimetype="application/json")
+
+
+@app.route('/release/<string(length=26):ident>/refs-in.json', methods=['GET', 'OPTIONS'])
+@crossdomain(origin='*',headers=['access-control-allow-origin','Content-Type'])
+def release_view_refs_inbound_json(ident):
+    hits = _refs_web("in", release_ident=ident)
+    return Response(hits.json(exclude_unset=True), mimetype="application/json")
+
+@app.route('/openlibrary/OL<int:id_num>W/refs-in', methods=['GET', 'OPTIONS'])
+@crossdomain(origin='*',headers=['access-control-allow-origin','Content-Type'])
+def openlibrary_view_refs_inbound_json(ident):
+    openlibrary_id = f"OL{id_num}W"
+    hits = _refs_web("in", openlibrary_id=openlibrary_id)
+    return Response(hits.json(exclude_unset=True), mimetype="application/json")
diff --git a/python/fatcat_web/templates/openlibrary_view_fuzzy_refs.html b/python/fatcat_web/templates/openlibrary_view_fuzzy_refs.html
new file mode 100644
index 00000000..161a7b50
--- /dev/null
+++ b/python/fatcat_web/templates/openlibrary_view_fuzzy_refs.html
@@ -0,0 +1,29 @@
+{% extends "base.html" %}
+{% import "refs_macros.html" as refs_macros %}
+
+{% block title %}Open Library Refs{% endblock %}
+
+{% block fullbody %}
+<h1 class="ui header">
+  {% if hits.result_refs and hits.result_refs[0].ref.target_unstructured %}
+    <i>{{ hits.result_refs[0].ref.target_unstructured }}</i>
+  {% endif %}
+  <span class="sub header"><a href="https://openlibrary.org/works/{{ openlibrary_id }}"><code>https://openlibrary.org/works/{{ openlibrary_id }}</code></a></span>
+</h1>
+
+{% if direction == "in" %}
+  <h3>Cited By</h3>
+  <p>This page lists references to this book from other works (eg, journal articles).
+{% elif direction == "out" %}
+  <h3>References</h3>
+  <i>Refernces from this book to other entities.</i>
+{% endif %}
+
+{% if hits.result_refs %}
+  {{ refs_macros.refs_table(hits, direction) }}
+{% else %}
+  <br><br><p><b>None found</b>
+{% endif %}
+
+{% endblock %}
+
-- 
cgit v1.2.3


From d5ea71b2892b84040b07107643d939f1cfb74d67 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Tue, 27 Jul 2021 17:00:32 -0700
Subject: refs: fix offset/limit bug

---
 python/fatcat_web/ref_routes.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index 72f115cf..6f887c4d 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -18,7 +18,7 @@ from fatcat_web.entity_helpers import *
 def _refs_web(direction, release_ident=None, work_ident=None, openlibrary_id=None) -> RefHits:
     offset = request.args.get('offset', '0')
     offset = max(0, int(offset)) if offset.isnumeric() else 0
-    limit = request.args.get('offset', '30')
+    limit = request.args.get('limit', '30')
     limit = min(max(0, int(limit)), 100) if limit.isnumeric() else 30
     if direction == "in":
         hits = get_inbound_refs(
-- 
cgit v1.2.3


From cb3d10a06c0166c4eb8ec2d48852e4bfc236ef27 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Tue, 27 Jul 2021 17:01:29 -0700
Subject: refs: support for wikipedia outbound refs, and display in tables

---
 python/fatcat_tools/references.py                  |  4 ++--
 python/fatcat_web/ref_routes.py                    | 26 ++++++++++++++++++---
 python/fatcat_web/templates/refs_macros.html       | 20 +++++++++++++---
 .../templates/wikipedia_view_fuzzy_refs.html       | 27 ++++++++++++++++++++++
 4 files changed, 69 insertions(+), 8 deletions(-)
 create mode 100644 python/fatcat_web/templates/wikipedia_view_fuzzy_refs.html

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_tools/references.py b/python/fatcat_tools/references.py
index 81b55f41..508cf19d 100644
--- a/python/fatcat_tools/references.py
+++ b/python/fatcat_tools/references.py
@@ -289,8 +289,8 @@ def enrich_inbound_refs(refs: List[BiblioRef], fatcat_api_client: Any, hide: Opt
             release = fatcat_api_client.get_release(ref.source_release_ident, hide=hide, expand=expand)
             access = release_access_options(release)
         if ref.source_wikipedia_article:
-            wiki_lang = ref.source_wikipedia.split(':')[0]
-            wiki_article = ':'.join(ref.source_wikipedia.split(':')[1:])
+            wiki_lang = ref.source_wikipedia_article.split(':')[0]
+            wiki_article = ':'.join(ref.source_wikipedia_article.split(':')[1:]).replace(' ', '_')
             access.append(AccessOption(
                 access_type="wikipedia",
                 access_url=f"https://{wiki_lang}.wikipedia.org/wiki/{wiki_article}",
diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index 6f887c4d..88ac0744 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -15,7 +15,7 @@ from fatcat_web.cors import crossdomain
 from fatcat_web.forms import *
 from fatcat_web.entity_helpers import *
 
-def _refs_web(direction, release_ident=None, work_ident=None, openlibrary_id=None) -> RefHits:
+def _refs_web(direction, release_ident=None, work_ident=None, openlibrary_id=None, wikipedia_article=None) -> RefHits:
     offset = request.args.get('offset', '0')
     offset = max(0, int(offset)) if offset.isnumeric() else 0
     limit = request.args.get('limit', '30')
@@ -37,6 +37,7 @@ def _refs_web(direction, release_ident=None, work_ident=None, openlibrary_id=Non
     elif direction == "out":
         hits = get_outbound_refs(
             release_ident=release_ident,
+            wikipedia_article=wikipedia_article,
             work_ident=work_ident,
             es_client=app.es_client,
             offset=offset,
@@ -74,12 +75,23 @@ def release_view_refs_outbound(ident):
 @app.route('/openlibrary/OL<int:id_num>W/refs-in', methods=['GET'])
 def openlibrary_view_refs_inbound(id_num):
     if request.accept_mimetypes.best == "application/json":
-        return openlibrary_view_refs_inbound(id_num)
+        return openlibrary_view_refs_inbound_json(id_num)
 
     openlibrary_id = f"OL{id_num}W"
     hits = _refs_web("in", openlibrary_id=openlibrary_id)
     return render_template('openlibrary_view_fuzzy_refs.html', openlibrary_id=openlibrary_id, direction="in", hits=hits), 200
 
+@app.route('/wikipedia/<string(length=2):wiki_lang>:<string:wiki_article>/refs-out', methods=['GET'])
+def wikipedia_view_refs_outbound(wiki_lang: str, wiki_article: str):
+    if request.accept_mimetypes.best == "application/json":
+        return wikipedia_view_refs_outbound_json(wiki_lang, wiki_article)
+
+    wiki_url = f"https://{wiki_lang}.wikipedia.org/wiki/{wiki_article}"
+    wiki_article = wiki_article.replace('_', ' ')
+    wikipedia_article = wiki_lang + ":" + wiki_article
+    hits = _refs_web("out", wikipedia_article=wikipedia_article)
+    return render_template('wikipedia_view_fuzzy_refs.html', wiki_article=wiki_article, wiki_lang=wiki_lang, wiki_url=wiki_url, direction="out", hits=hits), 200
+
 
 @app.route('/reference/match', methods=['GET', 'POST'])
 def reference_match():
@@ -140,9 +152,17 @@ def release_view_refs_inbound_json(ident):
     hits = _refs_web("in", release_ident=ident)
     return Response(hits.json(exclude_unset=True), mimetype="application/json")
 
-@app.route('/openlibrary/OL<int:id_num>W/refs-in', methods=['GET', 'OPTIONS'])
+@app.route('/openlibrary/OL<int:id_num>W/refs-in.json', methods=['GET', 'OPTIONS'])
 @crossdomain(origin='*',headers=['access-control-allow-origin','Content-Type'])
 def openlibrary_view_refs_inbound_json(ident):
     openlibrary_id = f"OL{id_num}W"
     hits = _refs_web("in", openlibrary_id=openlibrary_id)
     return Response(hits.json(exclude_unset=True), mimetype="application/json")
+
+@app.route('/wikipedia/<string(length=2):wiki_lang>:<string:wiki_article>/refs-out.json', methods=['GET', 'OPTIONS'])
+@crossdomain(origin='*',headers=['access-control-allow-origin','Content-Type'])
+def wikipedia_view_refs_outbound_json(wiki_lang: str, wiki_article: str):
+    wiki_article = wiki_article.replace('_', ' ')
+    wikipedia_article = wiki_lang + ":" + wiki_article
+    hits = _refs_web("out", wikipedia_article=wikipedia_article)
+    return Response(hits.json(exclude_unset=True), mimetype="application/json")
diff --git a/python/fatcat_web/templates/refs_macros.html b/python/fatcat_web/templates/refs_macros.html
index 405aca73..ba4d18ad 100644
--- a/python/fatcat_web/templates/refs_macros.html
+++ b/python/fatcat_web/templates/refs_macros.html
@@ -35,12 +35,26 @@
       <td class="">
         {% if release %}
           {{ entity_macros.release_summary(release) }}
-        {% elif row.ref.target_unstructured %}
+        {% elif direction == "in" and row.ref.source_wikipedia_article %}
+          {% set wiki_lang = row.ref.source_wikipedia_article.split(':')[0] %}
+          {% set wiki_article = ':'.join(row.ref.source_wikipedia_article.split(':')[1:]) %}
+          <b>
+            <a href="https://{{ wiki_lang }}.wikipedia.org/wiki/{{ wiki_article.replace(' ', '_') }}">
+              {{ wiki_article }}
+            </a>
+            [wikipedia]
+          </b>
+          <br>
+          <span style="color:green;">lang:{{ wiki_lang }}</span>&nbsp;
+          <a href="/wikipedia/{{ wiki_lang }}:{{ wiki_article.replace(' ', '_') }}/refs-out" style="color:green;">[references]</a>&nbsp;
+        {% elif direction == "out" and row.ref.target_unstructured %}
           <code>{{ row.ref.target_unstructured }}</code>
           {% if row.ref.target_openlibrary_work %}
-            <br><a href="https://openlibrary.org/{{ row.ref.target_openlibrary_work }}" style="color:green;">openlibrary:{{ row.ref.target_openlibrary_work }}</a>&nbsp;
+            <br>
+            <a href="https://openlibrary.org/{{ row.ref.target_openlibrary_work }}" style="color:green;">openlibrary:{{ row.ref.target_openlibrary_work }}</a>&nbsp;
+            <a href="/openlibrary/{{ row.ref.target_openlibrary_work}}/refs-in" style="color:green;">[cited-by]</a>&nbsp;
           {% endif %}
-        {% elif row.ref.target_csl %}
+        {% elif direction == "in" and row.ref.target_csl %}
           {{ entity_macros.csl_summary(row.ref.target_csl) }}
         {% else %}
           <i>blank</i>
diff --git a/python/fatcat_web/templates/wikipedia_view_fuzzy_refs.html b/python/fatcat_web/templates/wikipedia_view_fuzzy_refs.html
new file mode 100644
index 00000000..5b53d692
--- /dev/null
+++ b/python/fatcat_web/templates/wikipedia_view_fuzzy_refs.html
@@ -0,0 +1,27 @@
+{% extends "base.html" %}
+{% import "refs_macros.html" as refs_macros %}
+
+{% block title %}Wikipedia Refs{% endblock %}
+
+{% block fullbody %}
+<h1 class="ui header">
+  [{{ wiki_lang }}] {{ wiki_article }}
+  <span class="sub header"><a href="{{ wiki_url }}"><code>{{ wiki_url }}</code></a></span>
+</h1>
+
+{% if direction == "in" %}
+  <h3>Cited By</h3>
+  <p>This page lists references to a wikipedia article, from other works (eg, journal articles).
+{% elif direction == "out" %}
+  <h3>References</h3>
+  <i>Refernces from wikipedia article to other entities.</i>
+{% endif %}
+
+{% if hits.result_refs %}
+  {{ refs_macros.refs_table(hits, direction) }}
+{% else %}
+  <br><br><p><b>None found</b>
+{% endif %}
+
+{% endblock %}
+
-- 
cgit v1.2.3


From f3481c02bd7a50d9073902dba07fe265eecb93db Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Tue, 27 Jul 2021 18:57:09 -0700
Subject: refs: lint fixes

---
 python/fatcat_tools/references.py | 1 +
 python/fatcat_web/ref_routes.py   | 4 ++--
 2 files changed, 3 insertions(+), 2 deletions(-)

(limited to 'python/fatcat_web/ref_routes.py')

diff --git a/python/fatcat_tools/references.py b/python/fatcat_tools/references.py
index 508cf19d..496a46e1 100644
--- a/python/fatcat_tools/references.py
+++ b/python/fatcat_tools/references.py
@@ -99,6 +99,7 @@ class EnrichedBiblioRef(BaseModel):
     access: List[AccessOption]
 
     @validator('release')
+    @classmethod
     def check_release(cls, v):
         if v is not None and not isinstance(v, ReleaseEntity):
             raise ValueError("expected a ReleaseEntity")
diff --git a/python/fatcat_web/ref_routes.py b/python/fatcat_web/ref_routes.py
index 88ac0744..d4219012 100644
--- a/python/fatcat_web/ref_routes.py
+++ b/python/fatcat_web/ref_routes.py
@@ -3,7 +3,7 @@ Flask endpoints for reference (citation) endpoints. Eg, listing references
 "inbound" and "outbound" from a specific release or work.
 """
 
-from flask import render_template, request, jsonify, Response
+from flask import render_template, request, Response
 from fatcat_openapi_client import *
 from fuzzycat.grobid_unstructured import grobid_api_process_citation, transform_grobid_ref_xml, grobid_ref_to_release
 from fuzzycat.simple import close_fuzzy_biblio_matches, close_fuzzy_release_matches
@@ -154,7 +154,7 @@ def release_view_refs_inbound_json(ident):
 
 @app.route('/openlibrary/OL<int:id_num>W/refs-in.json', methods=['GET', 'OPTIONS'])
 @crossdomain(origin='*',headers=['access-control-allow-origin','Content-Type'])
-def openlibrary_view_refs_inbound_json(ident):
+def openlibrary_view_refs_inbound_json(id_num):
     openlibrary_id = f"OL{id_num}W"
     hits = _refs_web("in", openlibrary_id=openlibrary_id)
     return Response(hits.json(exclude_unset=True), mimetype="application/json")
-- 
cgit v1.2.3