[feat] tracker url plugin: use ClearURL tracking param list

4 files changed, 94 insertions, 24 deletions

diff --git a/.github/workflows/data-update.yml b/.github/workflows/data-update.yml
index 4deca2ffc..9d22bcc94 100644
--- a/.github/workflows/data-update.yml
+++ b/.github/workflows/data-update.yml
@@ -33,6 +33,7 @@ jobs:
           - update_engine_traits.py
           - update_wikidata_units.py
           - update_engine_descriptions.py
+          - update_tracker_patterns.py
 
     permissions:
       contents: write
diff --git a/searx/data/__init__.py b/searx/data/__init__.py
index 9be1cd67e..5a859f8cd 100644
--- a/searx/data/__init__.py
+++ b/searx/data/__init__.py
@@ -23,6 +23,7 @@ OSM_KEYS_TAGS: dict[str, typing.Any]
 ENGINE_DESCRIPTIONS: dict[str, typing.Any]
 ENGINE_TRAITS: dict[str, typing.Any]
 LOCALES: dict[str, typing.Any]
+TRACKER_PATTERNS: list[dict[str, typing.Any]]
 
 lazy_globals = {
     "CURRENCIES": CurrenciesDB(),
@@ -34,6 +35,7 @@ lazy_globals = {
     "ENGINE_DESCRIPTIONS": None,
     "ENGINE_TRAITS": None,
     "LOCALES": None,
+    "TRACKER_PATTERNS": None,
 }
 
 data_json_files = {
@@ -45,6 +47,7 @@ data_json_files = {
     "ENGINE_DESCRIPTIONS": "engine_descriptions.json",
     "ENGINE_TRAITS": "engine_traits.json",
     "LOCALES": "locales.json",
+    "TRACKER_PATTERNS": "tracker_patterns.json",
 }
 
 
diff --git a/searx/plugins/tracker_url_remover.py b/searx/plugins/tracker_url_remover.py
index d9c767a36..efc593775 100644
--- a/searx/plugins/tracker_url_remover.py
+++ b/searx/plugins/tracker_url_remover.py
@@ -1,34 +1,31 @@
 # SPDX-License-Identifier: AGPL-3.0-or-later
-# pylint: disable=missing-module-docstring
+# pylint: disable=missing-module-docstring, unused-argument
 
 from __future__ import annotations
 import typing
 
 import re
-from urllib.parse import urlunparse, parse_qsl, urlencode
+from urllib.parse import urlparse, urlunparse, parse_qsl, urlencode
 
 from flask_babel import gettext
 
-from searx.plugins import Plugin, PluginInfo
+from searx.data import TRACKER_PATTERNS
+
+from . import Plugin, PluginInfo
+from ._core import log
 
 if typing.TYPE_CHECKING:
     from searx.search import SearchWithPlugins
     from searx.extended_types import SXNG_Request
-    from searx.result_types import Result
+    from searx.result_types import Result, LegacyResult
     from searx.plugins import PluginCfg
 
-regexes = {
-    re.compile(r'utm_[^&]+'),
-    re.compile(r'(wkey|wemail)[^&]*'),
-    re.compile(r'(_hsenc|_hsmi|hsCtaTracking|__hssc|__hstc|__hsfp)[^&]*'),
-    re.compile(r'&$'),
-}
-
 
 class SXNGPlugin(Plugin):
-    """Remove trackers arguments from the returned URL"""
+    """Remove trackers arguments from the returned URL."""
 
     id = "tracker_url_remover"
+    log = log.getChild(id)
 
     def __init__(self, plg_cfg: "PluginCfg") -> None:
         super().__init__(plg_cfg)
@@ -39,20 +36,53 @@ class SXNGPlugin(Plugin):
             preference_section="privacy",
         )
 
-    def on_result(
-        self, request: "SXNG_Request", search: "SearchWithPlugins", result: Result
-    ) -> bool:  # pylint: disable=unused-argument
-        if not result.parsed_url:
+    def on_result(self, request: "SXNG_Request", search: "SearchWithPlugins", result: Result) -> bool:
+
+        result.filter_urls(self.filter_url_field)
+        return True
+
+    @classmethod
+    def filter_url_field(cls, result: "Result|LegacyResult", field_name: str, url_src: str) -> bool | str:
+        """Returns bool ``True`` to use URL unchanged (``False`` to ignore URL).
+        If URL should be modified, the returned string is the new URL to use."""
+
+        if not url_src:
+            cls.log.debug("missing a URL in field %s", field_name)
             return True
 
-        parsed_query: list[tuple[str, str]] = parse_qsl(result.parsed_url.query)
-        for name_value in list(parsed_query):
-            param_name = name_value[0]
-            for reg in regexes:
-                if reg.match(param_name):
-                    parsed_query.remove(name_value)
-                    result.parsed_url = result.parsed_url._replace(query=urlencode(parsed_query))
-                    result.url = urlunparse(result.parsed_url)
+        new_url = url_src
+        parsed_new_url = urlparse(url=new_url)
+
+        for rule in TRACKER_PATTERNS:
+
+            if not re.match(rule["urlPattern"], new_url):
+                # no match / ignore pattern
+                continue
+
+            in_exceptions = False
+            for exception in rule["exceptions"]:
+                if re.match(exception, new_url):
+                    in_exceptions = True
                     break
+            if in_exceptions:
+                # pattern is in the list of exceptions / ignore pattern
+                # hint: we can't break the outer pattern loop since we have
+                # overlapping urlPattern like ".*"
+                continue
+
+            # remove tracker arguments from the url-query part
+            query_args: list[tuple[str, str]] = list(parse_qsl(parsed_new_url.query))
+
+            for name, val in query_args.copy():
+                for reg in rule["trackerParams"]:
+                    if re.match(reg, name):
+                        cls.log.debug("%s remove tracker arg: %s='%s'", parsed_new_url.netloc, name, val)
+                        query_args.remove((name, val))
+
+            parsed_new_url = parsed_new_url._replace(query=urlencode(query_args))
+            new_url = urlunparse(parsed_new_url)
+
+        if new_url != url_src:
+            return new_url
 
         return True
diff --git a/searxng_extra/update/update_tracker_patterns.py b/searxng_extra/update/update_tracker_patterns.py
new file mode 100644
index 000000000..f8928d354
--- /dev/null
+++ b/searxng_extra/update/update_tracker_patterns.py
@@ -0,0 +1,36 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+"""Fetch trackers"""
+
+import json
+import httpx
+
+from searx.data import data_dir
+
+DATA_FILE = data_dir / "tracker_patterns.json"
+CLEAR_LIST_URL = "https://raw.githubusercontent.com/ClearURLs/Rules/refs/heads/master/data.min.json"
+
+
+def fetch_clear_url_filters():
+    resp = httpx.get(CLEAR_LIST_URL)
+    if resp.status_code != 200:
+        # pylint: disable=broad-exception-raised
+        raise Exception(f"Error fetching ClearURL filter lists, HTTP code {resp.status_code}")
+
+    providers = resp.json()["providers"]
+    rules = []
+    for rule in providers.values():
+        rules.append(
+            {
+                "urlPattern": rule["urlPattern"].replace("\\\\", "\\"),  # fix javascript regex syntax
+                "exceptions": [exc.replace("\\\\", "\\") for exc in rule["exceptions"]],
+                "trackerParams": rule["rules"],
+            }
+        )
+
+    return rules
+
+
+if __name__ == '__main__':
+    filter_list = fetch_clear_url_filters()
+    with DATA_FILE.open("w", encoding='utf-8') as f:
+        json.dump(filter_list, f, indent=4, sort_keys=True, ensure_ascii=False)