Allow setting site alts using environment vars (#155)

* Add ability to configure site alts w/ env vars

Site alternatives (i.e. twitter.com -> nitter.net) can now be configured
using environment variables:

WHOOGLE_ALT_TW='nitter.net' # twitter alt
WHOOGLE_ALT_YT='invidio.us' # youtube alt
WHOOGLE_ALT_IG='bibliogram.art/u' # instagram alt

Updated testing to confirm results have been modified.

* Add site alt vars to docker settings and readme

Dockerfile

@@ -36,6 +36,13 @@ ENV HTTPS_ONLY=$use_https
 ARG whoogle_port=5000
 ENV EXPOSE_PORT=$whoogle_port
 
+ARG twitter_alt=''
+ENV WHOOGLE_ALT_TW=$twitter_alt
+ARG youtube_alt=''
+ENV WHOOGLE_ALT_YT=$youtube_alt
+ARG instagram_alt=''
+ENV WHOOGLE_ALT_YT=$instagram_alt
+
 COPY . .
 
 EXPOSE $EXPOSE_PORT

README.md

@@ -129,6 +129,12 @@ Description=Whoogle
 #Environment=WHOOGLE_PROXY_PASS=<proxy password>
 #Environment=WHOOGLE_PROXY_TYPE=<proxy type (http|proxy4|proxy5)
 #Environment=WHOOGLE_PROXY_LOC=<proxy host/ip>
+# Site alternative configurations, uncomment to enable
+# Note: If not set, the feature will still be available
+# with default values. 
+#Environment=WHOOGLE_ALT_TW=nitter.net
+#Environment=WHOOGLE_ALT_YT=invidious.snopyta.org
+#Environment=WHOOGLE_ALT_IG=bibliogram.art/u
 Type=simple
 User=root
 WorkingDirectory=<whoogle_directory>
@@ -230,6 +236,9 @@ There are a few optional environment variables available for customizing a Whoog
 | WHOOGLE_PROXY_LOC  | The location of the proxy server (host or ip).                 |
 | EXPOSE_PORT        | The port where Whoogle will be exposed.                        |
 | HTTPS_ONLY         | Enforce HTTPS. (See [here](https://github.com/benbusby/whoogle-search#https-enforcement)) |
+| WHOOGLE_ALT_TW     | The twitter.com alternative to use when site alternatives are enabled in the config. |
+| WHOOGLE_ALT_YT     | The youtube.com alternative to use when site alternatives are enabled in the config. |
+| WHOOGLE_ALT_IG     | The instagram.com alternative to use when site alternatives are enabled in the config. |
 
 ## Usage
 Same as most search engines, with the exception of filtering by time range.

app.json

@@ -44,6 +44,21 @@
       "description": "The location of the proxy server (host or ip). Leave empty to disable.",
       "value": "",
       "required": false
+    },
+    "WHOOGLE_ALT_TW": {
+      "description": "The site to use as a replacement for twitter.com when site alternatives are enabled in the config."
+      "value": "",
+      "required": false
+    },
+    "WHOOGLE_ALT_YT": {
+      "description": "The site to use as a replacement for youtube.com when site alternatives are enabled in the config."
+      "value": "",
+      "required": false
+    },
+    "WHOOGLE_ALT_IG": {
+      "description": "The site to use as a replacement for instagram.com when site alternatives are enabled in the config."
+      "value": "",
+      "required": false
     }
   }
 }

app/utils/filter_utils.py

@@ -1,4 +1,5 @@
 from bs4 import BeautifulSoup
+import os
 import urllib.parse as urlparse
 from urllib.parse import parse_qs
 
@@ -17,9 +18,9 @@ BLACKLIST = [
 ]
 
 SITE_ALTS = {
-    'twitter.com': 'nitter.net',
-    'youtube.com': 'invidious.snopyta.org',
-    'instagram.com': 'bibliogram.art/u'
+    'twitter.com': os.getenv('WHOOGLE_ALT_TW', 'nitter.net'),
+    'youtube.com': os.getenv('WHOOGLE_ALT_YT', 'invidious.snopyta.org'),
+    'instagram.com': os.getenv('WHOOGLE_ALT_IG', 'bibliogram.art/u')
 }
 
 

docker-compose.yml

@@ -13,6 +13,12 @@ services:
       #- WHOOGLE_PROXY_PASS=<proxy password>
       #- WHOOGLE_PROXY_TYPE=<proxy type (http|socks4|socks5)
       #- WHOOGLE_PROXY_LOC=<proxy host/ip>
+      # Site alternative configurations, uncomment to enable
+      # Note: If not set, the feature will still be available
+      # with default values. 
+      #- WHOOGLE_ALT_TW=nitter.net
+      #- WHOOGLE_ALT_YT=invidious.snopyta.org
+      #- WHOOGLE_ALT_IG=bibliogram.art/u
     ports:
       - 5000:5000
     restart: unless-stopped

test/conftest.py

@@ -1,6 +1,17 @@
 from app import app
+from app.models.config import Config
 from app.utils.session_utils import generate_user_keys
 import pytest
+import random
+
+demo_config = {
+    'near': random.choice(['Seattle', 'New York', 'San Francisco']),
+    'dark_mode': str(random.getrandbits(1)),
+    'nojs': str(random.getrandbits(1)),
+    'lang_interface': random.choice(Config.LANGUAGES)['value'],
+    'lang_search': random.choice(Config.LANGUAGES)['value'],
+    'ctry': random.choice(Config.COUNTRIES)['value']
+}
 
 
 @pytest.fixture

test/test_results.py

@@ -3,6 +3,8 @@ from app.filter import Filter
 from app.utils.session_utils import generate_user_keys
 from datetime import datetime
 from dateutil.parser import *
+import json
+import os
 
 
 def get_search_results(data):
@@ -43,6 +45,18 @@ def test_post_results(client):
     assert len(get_search_results(rv.data)) <= 15
 
 
+def test_site_alts(client):
+    rv = client.post('/search', data=dict(q='twitter official account'))
+    assert b'twitter.com/Twitter' in rv.data
+
+    client.post('/config', data=dict(alts=True))
+    assert json.loads(client.get('/config').data)['alts']
+
+    rv = client.post('/search', data=dict(q='twitter official account'))
+    assert b'twitter.com/Twitter' not in rv.data
+    assert b'nitter.net/Twitter' in rv.data
+
+
 def test_recent_results(client):
     times = {
         'past year': 365,

test/test_routes.py

@@ -1,15 +1,6 @@
-from app.models.config import Config
 import json
-import random
 
-demo_config = {
-    'near': random.choice(['Seattle', 'New York', 'San Francisco']),
-    'dark_mode': str(random.getrandbits(1)),
-    'nojs': str(random.getrandbits(1)),
-    'lang_interface': random.choice(Config.LANGUAGES)['value'],
-    'lang_search': random.choice(Config.LANGUAGES)['value'],
-    'ctry': random.choice(Config.COUNTRIES)['value']
-}
+from test.conftest import demo_config
 
 
 def test_main(client):
@@ -58,4 +49,4 @@ def test_config(client):
 def test_opensearch(client):
     rv = client.get('/opensearch.xml')
     assert rv._status_code == 200
-    assert 'Whoogle' in str(rv.data)
+    assert '<ShortName>Whoogle</ShortName>' in str(rv.data)