[enh] add hostname_replace plugin

* backport of https://github.com/searx/searx/pull/2724
* allow to remove result if the replacement is the boolean value false

2 files changed, 34 insertions, 0 deletions

diff --git a/searx/plugins/__init__.py b/searx/plugins/__init__.py
index 3a35f7025..1153c9ed1 100644
--- a/searx/plugins/__init__.py
+++ b/searx/plugins/__init__.py
@@ -31,6 +31,7 @@ from searx.plugins import (oa_doi_rewrite,
                            hash_plugin,
                            infinite_scroll,
                            self_info,
+                           hostname_replace,
                            search_on_category_select,
                            tracker_url_remover,
                            vim_hotkeys)
@@ -182,6 +183,7 @@ plugins.register(oa_doi_rewrite)
 plugins.register(hash_plugin)
 plugins.register(infinite_scroll)
 plugins.register(self_info)
+plugins.register(hostname_replace)
 plugins.register(search_on_category_select)
 plugins.register(tracker_url_remover)
 plugins.register(vim_hotkeys)
diff --git a/searx/plugins/hostname_replace.py b/searx/plugins/hostname_replace.py
new file mode 100644
index 000000000..778b84615
--- /dev/null
+++ b/searx/plugins/hostname_replace.py
@@ -0,0 +1,32 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+import re
+from urllib.parse import urlunparse
+from searx import settings
+from searx.plugins import logger
+from flask_babel import gettext
+
+name = gettext('Hostname replace')
+description = gettext('Rewrite result hostnames or remove results based on the hostname')
+default_on = False
+preference_section = 'general'
+
+plugin_id = 'hostname_replace'
+
+replacements = {re.compile(p): r for (p, r) in settings[plugin_id].items()} if plugin_id in settings else {}
+
+logger = logger.getChild(plugin_id)
+parsed = 'parsed_url'
+
+
+def on_result(request, search, result):
+    if parsed not in result:
+        return True
+    for (pattern, replacement) in replacements.items():
+        if pattern.search(result[parsed].netloc):
+            if not replacement:
+                return False
+            result[parsed] = result[parsed]._replace(netloc=pattern.sub(replacement, result[parsed].netloc))
+            result['url'] = urlunparse(result[parsed])
+
+    return True