Browse Source

[feat] engine: implementation of bpb

Bnyro 1 year ago
parent
commit
bf75a8c2a0
4 changed files with 88 additions and 2 deletions
  1. 13 0
      docs/dev/engines/online/bpb.rst
  2. 68 0
      searx/engines/bpb.py
  3. 5 0
      searx/settings.yml
  4. 2 2
      utils/brand.env

+ 13 - 0
docs/dev/engines/online/bpb.rst

@@ -0,0 +1,13 @@
+.. _bpb engine:
+
+===
+Bpb
+===
+
+.. contents:: Contents
+   :depth: 2
+   :local:
+   :backlinks: entry
+
+.. automodule:: searx.engines.bpb
+  :members:

+ 68 - 0
searx/engines/bpb.py

@@ -0,0 +1,68 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+# lint: pylint
+"""BPB refers to ``Bundeszentrale für poltische Bildung``, which is a German
+governmental institution aiming to reduce misinformation by providing resources
+about politics and history.
+"""
+
+from datetime import datetime
+from urllib.parse import urlencode
+
+about = {
+    'website': "https://www.bpb.de",
+    'official_api_documentation': None,
+    'use_official_api': False,
+    'require_api_key': False,
+    'results': 'JSON',
+    'language': 'de',
+}
+
+paging = True
+categories = ['general']
+
+
+base_url = "https://www.bpb.de"
+
+
+def request(query, params):
+    args = {
+        'query[term]': query,
+        'page': params['pageno'] - 1,
+        'sort[direction]': 'descending',
+        'payload[nid]': 65350,
+    }
+    params['url'] = f"{base_url}/bpbapi/filter/search?{urlencode(args)}"
+    return params
+
+
+def response(resp):
+    results = []
+
+    json_resp = resp.json()
+
+    for result in json_resp['teaser']:
+        img_src = None
+        if result['teaser']['image']:
+            img_src = base_url + result['teaser']['image']['sources'][-1]['url']
+
+        metadata = result['extension']['overline']
+        authors = ', '.join(author['name'] for author in result['extension'].get('authors', []))
+        if authors:
+            metadata += f" | {authors}"
+
+        publishedDate = None
+        if result['extension'].get('publishingDate'):
+            publishedDate = datetime.utcfromtimestamp(result['extension']['publishingDate'])
+
+        results.append(
+            {
+                'url': base_url + result['teaser']['link']['url'],
+                'title': result['teaser']['title'],
+                'content': result['teaser']['text'],
+                'img_src': img_src,
+                'publishedDate': publishedDate,
+                'metadata': metadata,
+            }
+        )
+
+    return results

+ 5 - 0
searx/settings.yml

@@ -419,6 +419,11 @@ engines:
       require_api_key: false
       results: HTML
 
+  - name: bpb
+    engine: bpb
+    shortcut: bpb
+    disabled: true
+
   - name: btdigg
     engine: btdigg
     shortcut: bt

+ 2 - 2
utils/brand.env

@@ -1,5 +1,5 @@
 export SEARXNG_URL=''
 export SEARXNG_PORT='8888'
 export SEARXNG_BIND_ADDRESS='127.0.0.1'
-export GIT_URL='https://github.com/searxng/searxng'
-export GIT_BRANCH='master'
+export GIT_URL='https://github.com//Bnyro/searxng'
+export GIT_BRANCH='bpb'