From 6513a5606431b5ab4fbe6eefb70510ba47ab5e04 Mon Sep 17 00:00:00 2001
From: spongebob33 <marc.cenon33@gmail.com>
Date: Fri, 26 Mar 2021 12:22:49 +0100
Subject: [PATCH] add core.ac.uk engine

---
 manage                |  1 +
 searx/engines/core.py | 82 +++++++++++++++++++++++++++++++++++++++++++
 searx/settings.yml    |  7 ++++
 3 files changed, 90 insertions(+)
 create mode 100644 searx/engines/core.py

diff --git a/manage b/manage
index 7c89d8a7..fe02bf01 100755
--- a/manage
+++ b/manage
@@ -35,6 +35,7 @@ PYLINT_FILES=(
     searx/engines/google_scholar.py
     searx/engines/yahoo_news.py
     searx/engines/apkmirror.py
+    searx/engines/core.py
     searx_extra/update/update_external_bangs.py
 )
 
diff --git a/searx/engines/core.py b/searx/engines/core.py
new file mode 100644
index 00000000..ee2443b5
--- /dev/null
+++ b/searx/engines/core.py
@@ -0,0 +1,82 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+"""CORE (science)
+
+"""
+# pylint: disable=missing-function-docstring
+
+from json import loads
+from datetime import datetime
+from urllib.parse import urlencode
+
+from searx import logger
+from searx.exceptions import SearxEngineAPIException
+
+logger = logger.getChild('CORE engine')
+
+about = {
+    "website": 'https://core.ac.uk',
+    "wikidata_id": None,
+    "official_api_documentation": 'https://core.ac.uk/documentation/api/',
+    "use_official_api": True,
+    "require_api_key": True,
+    "results": 'JSON',
+}
+
+categories = ['science']
+paging = True
+nb_per_page = 10
+
+api_key = 'unset'
+
+logger = logger.getChild('CORE engine')
+
+base_url = 'https://core.ac.uk:443/api-v2/search/'
+search_string = '{query}?page={page}&pageSize={nb_per_page}&apiKey={apikey}'
+
+def request(query, params):
+
+    if api_key == 'unset':
+        raise SearxEngineAPIException('missing CORE API key')
+
+    search_path = search_string.format(
+        query = urlencode({'q': query}),
+        nb_per_page = nb_per_page,
+        page = params['pageno'],
+        apikey = api_key,
+    )
+    params['url'] = base_url + search_path
+
+    logger.debug("query_url --> %s", params['url'])
+    return params
+
+def response(resp):
+    results = []
+    json_data = loads(resp.text)
+
+    for result in json_data['data']:
+
+        source = result['_source']
+        time = source['publishedDate'] or source['depositedDate']
+        if time :
+            date = datetime.fromtimestamp(time / 1000)
+        else:
+            date = None
+
+        metadata = []
+        if source['publisher'] and len(source['publisher']) > 3:
+            metadata.append(source['publisher'])
+        if source['topics']:
+            metadata.append(source['topics'][0])
+        if source['doi']:
+            metadata.append(source['doi'])
+        metadata = ' / '.join(metadata)
+
+        results.append({
+            'url': source['urls'][0].replace('http://', 'https://', 1),
+            'title': source['title'],
+            'content': source['description'],
+            'publishedDate': date,
+            'metadata' : metadata,
+        })
+
+    return results
diff --git a/searx/settings.yml b/searx/settings.yml
index 8b4f9972..df106be5 100644
--- a/searx/settings.yml
+++ b/searx/settings.yml
@@ -281,6 +281,13 @@ engines:
     categories : images
     shortcut : cce
 
+# - name : core.ac.uk
+#   engine : core
+#   categories : science
+#   shortcut : cor
+#   # get your API key from: https://core.ac.uk/api-keys/register/
+#   api_key : 'unset'
+
   - name : crossref
     engine : json_engine
     paging : True