From 87f40d0f8d15f16020964773e75d35ea22da049f Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Thu, 2 Apr 2020 17:37:35 -0700
Subject: basic fulltext search highlighting

---
 fatcat_covid19/search.py                      | 32 +++++++++++++++++++++++--
 fatcat_covid19/templates/base.html            |  4 ++++
 fatcat_covid19/templates/entity_macros.html   | 34 +++++++++++++++++----------
 fatcat_covid19/templates/fulltext_search.html |  4 ++--
 4 files changed, 58 insertions(+), 16 deletions(-)

diff --git a/fatcat_covid19/search.py b/fatcat_covid19/search.py
index 2534374..08b6466 100644
--- a/fatcat_covid19/search.py
+++ b/fatcat_covid19/search.py
@@ -1,4 +1,5 @@
 
+import json
 import datetime
 import requests
 from flask import abort, flash
@@ -32,7 +33,16 @@ def do_search(index, request, limit=30, offset=0, deep_page_limit=2000):
         abort(resp.status_code)
 
     content = resp.json()
-    results = [h['_source'] for h in content['hits']['hits']]
+    #print(json.dumps(content, indent=2))
+    results = []
+    for h in content['hits']['hits']:
+        r = h['_source']
+        r['_highlights'] = []
+        highlights = h.get('highlight', {})
+        for k in highlights:
+            r['_highlights'] += highlights[k]
+        results.append(r)
+    print(json.dumps(results, indent=2))
     for h in results:
         # Handle surrogate strings that elasticsearch returns sometimes,
         # probably due to mangled data processing in some pipeline.
@@ -66,7 +76,25 @@ def do_fulltext_search(q, limit=30, offset=0):
                 "default_operator": "AND",
                 "analyze_wildcard": True,
                 "lenient": True,
-                "fields": ["everything"],
+                "fields": [
+                    "everything",
+                    "abstract",
+                    "fulltext.body",
+                    "fulltext.annex",
+                ],
+            },
+        },
+        "highlight" : {
+            "number_of_fragments" : 3,
+            "fragment_size" : 150,
+            "fields" : {
+                "abstract": { },
+                "fulltext.body": { },
+                "fulltext.annex": { },
+                #"everything": { "number_of_fragments" : 3 },
+                #"fulltext.abstract": { "number_of_fragments" : 3 },
+                #"fulltext.body":     { "number_of_fragments" : 3 },
+                #"fulltext.annex":    { "number_of_fragments" : 3 },
             },
         },
     }
diff --git a/fatcat_covid19/templates/base.html b/fatcat_covid19/templates/base.html
index 07632dc..3193619 100644
--- a/fatcat_covid19/templates/base.html
+++ b/fatcat_covid19/templates/base.html
@@ -17,6 +17,10 @@
     @media only screen and (max-width: 479px) {
       .mobile-hide{ display: none !important; }
     }
+    {# for fulltext search result highlighting #}
+    .search_highlights em {
+      background-color: #FFFFBB; {# yellow #}
+    }
   </style>
   {% block extra_head %}{% endblock %}
 </head>
diff --git a/fatcat_covid19/templates/entity_macros.html b/fatcat_covid19/templates/entity_macros.html
index 66f0eb2..cc3a8a6 100644
--- a/fatcat_covid19/templates/entity_macros.html
+++ b/fatcat_covid19/templates/entity_macros.html
@@ -1,16 +1,7 @@
 
 {% macro fulltext_search_result_row(paper) -%}
 <div class="row">
-<div class="four wide right aligned column" style="padding: 1em;">
-  {% if paper.fulltext.t_thumbnail_url %}
-    <a href="{{ config.COVID19_FULLTEXT_HOST }}{{ paper.fulltext.pdf_url.replace('//', '/') }}">
-      <img src="{{ config.COVID19_FULLTEXT_HOST }}{{ paper.fulltext.t_thumbnail_url.replace('//', '/') }}" style="border: 1px solid grey; max-height: 12em;">
-    </a>
-  {% else %}
-    No Fulltext
-  {% endif %}
-</div>
-<div class="twelve wide column">
+<div class="thirteen wide column">
   {# ### TITLE ROW #}
   <h4 style="margin-top: 1em; margin-bottom: 0px; font-size: 1.2em;">
 
@@ -41,14 +32,18 @@
       <b style="text-transform: uppercase;">[{{ paper.release_type }}]</b>
     {% endif %}
   </h4>
+
+  {#
   <div style="float: right; padding: 4px;">
   &nbsp;&nbsp;<a href="{{ config.ELASTICSEARCH_BACKEND }}/{{ config.ELASTICSEARCH_FULLTEXT_INDEX }}/release/{{ paper.fatcat_ident }}" class="ui grey tag label"><i class="file icon"></i>json</a>
   </div>
+
   {% 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 %}
+  #}
 
 
   {# ### AUTHOR ROW #}
@@ -87,8 +82,14 @@
   {% endif %}
 
   {# ### ABSTRACT / QUERY HIGHLIGHT #}
-  {% if paper.abstract %}
-    <div style="padding: 1em;">
+  {% if paper._highlights %}
+    <div style="padding-top: 1em; padding-left: 0.5em;" class="search_highlights">
+    {% for highlight in paper._highlights %}
+      {{ highlight|safe }} ...
+    {% endfor %}
+    </div>
+  {% elif paper.abstract %}
+    <div style="padding-top: 1em; padding-left: 0.5em;">
       {% if paper.abstract[0]|length > 500 %}
         {{ paper.abstract[0][:500] }}...
       {% else %}
@@ -115,6 +116,15 @@
   <a href="/work/{{ paper.work_id }}"><i class="sitemap icon"></i> and 5 other versions of the same work!</a>
   {% endif %}
 </div>
+<div class="three wide left aligned column" style="padding: 1em;">
+  {% if paper.fulltext.thumbnail_url %}
+    <a href="{{ config.COVID19_FULLTEXT_HOST }}{{ paper.fulltext.pdf_url.replace('//', '/') }}">
+      <img src="{{ config.COVID19_FULLTEXT_HOST }}{{ paper.fulltext.thumbnail_url.replace('//', '/') }}" style="border: 1px solid grey; max-height: 12em; max-width: 100%;">
+    </a>
+  {% else %}
+    {# No Fulltext #}
+  {% endif %}
+</div>
 </div>
 {% endmacro %}
 
diff --git a/fatcat_covid19/templates/fulltext_search.html b/fatcat_covid19/templates/fulltext_search.html
index f1f2c6b..d2a8e4c 100644
--- a/fatcat_covid19/templates/fulltext_search.html
+++ b/fatcat_covid19/templates/fulltext_search.html
@@ -38,8 +38,8 @@
     <div class="ui container text" style="padding: 1em;">
       {{ entity_macros.top_results(found) }}
     </div>
-    <div class="ui container">
-      <div class="ui centered grid">
+    <div class="ui container" style="max-width: 950px !important;">
+      <div class="ui centered stackable grid">
         {% for paper in found.results %}
           {{ entity_macros.fulltext_search_result_row(paper) }}
         {% endfor %}
-- 
cgit v1.2.3