logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://hacktivis.me/git/searx.git
commit: fbb080f3588ad43bf896a569257f3a95e1181d7b
parent 67e11c42b973932c8f568d80a0f25bfd7fc150ab
Author: dalf <alex@al-f.net>
Date:   Sat, 22 Oct 2016 14:01:53 +0200

Change plugin API :
- pre_search(request, search)
- post_search(request, search)
- on_result(request, search, result)

with
- request is the Flask request
- search a searx.Search instance
- result a searx result as usual

Diffstat:

Msearx/plugins/doai_rewrite.py8++++----
Msearx/plugins/https_rewrite.py3+--
Msearx/plugins/self_info.py16++++++++--------
Msearx/plugins/tracker_url_remover.py10+++++-----
Msearx/search.py32++++++++++----------------------
Mtests/unit/test_plugins.py52++++++++++++++++++++++++++--------------------------
6 files changed, 54 insertions(+), 67 deletions(-)

diff --git a/searx/plugins/doai_rewrite.py b/searx/plugins/doai_rewrite.py @@ -20,12 +20,12 @@ def extract_doi(url): return None -def on_result(request, ctx): - doi = extract_doi(ctx['result']['parsed_url']) +def on_result(request, search, result): + doi = extract_doi(result['parsed_url']) if doi and len(doi) < 50: for suffix in ('/', '.pdf', '/full', '/meta', '/abstract'): if doi.endswith(suffix): doi = doi[:-len(suffix)] - ctx['result']['url'] = 'http://doai.io/' + doi - ctx['result']['parsed_url'] = urlparse(ctx['result']['url']) + result['url'] = 'http://doai.io/' + doi + result['parsed_url'] = urlparse(ctx['result']['url']) return True diff --git a/searx/plugins/https_rewrite.py b/searx/plugins/https_rewrite.py @@ -220,8 +220,7 @@ def https_url_rewrite(result): return result -def on_result(request, ctx): - result = ctx['result'] +def on_result(request, search, result): if result['parsed_url'].scheme == 'http': https_url_rewrite(result) return True diff --git a/searx/plugins/self_info.py b/searx/plugins/self_info.py @@ -28,19 +28,19 @@ p = re.compile('.*user[ -]agent.*', re.IGNORECASE) # attach callback to the post search hook # request: flask request object # ctx: the whole local context of the pre search hook -def post_search(request, ctx): - if ctx['search'].pageno > 1: +def post_search(request, search): + if search.search_query.pageno > 1: return True - if ctx['search'].query == 'ip': + if search.search_query.query == 'ip': x_forwarded_for = request.headers.getlist("X-Forwarded-For") if x_forwarded_for: ip = x_forwarded_for[0] else: ip = request.remote_addr - ctx['result_container'].answers.clear() - ctx['result_container'].answers.add(ip) - elif p.match(ctx['search'].query): + search.result_container.answers.clear() + search.result_container.answers.add(ip) + elif p.match(search.search_query.query): ua = request.user_agent - ctx['result_container'].answers.clear() - ctx['result_container'].answers.add(ua) + search.result_container.answers.clear() + search.result_container.answers.add(ua) return True diff --git a/searx/plugins/tracker_url_remover.py b/searx/plugins/tracker_url_remover.py @@ -28,8 +28,8 @@ description = gettext('Remove trackers arguments from the returned URL') default_on = True -def on_result(request, ctx): - query = ctx['result']['parsed_url'].query +def on_result(request, search, result): + query = result['parsed_url'].query if query == "": return True @@ -37,8 +37,8 @@ def on_result(request, ctx): for reg in regexes: query = reg.sub('', query) - if query != ctx['result']['parsed_url'].query: - ctx['result']['parsed_url'] = ctx['result']['parsed_url']._replace(query=query) - ctx['result']['url'] = urlunparse(ctx['result']['parsed_url']) + if query != result['parsed_url'].query: + result['parsed_url'] = result['parsed_url']._replace(query=query) + result['url'] = urlunparse(result['parsed_url']) return True diff --git a/searx/search.py b/searx/search.py @@ -357,35 +357,23 @@ class Search(object): return self.result_container -def search_with_plugins(do_search, search_query, request, request_data, result_container): - """Search using the do_search function and with plugins filtering. - Standalone function to have a well define locals(). - result_container contains the results after the function call. - """ - search = search_query - - if plugins.call('pre_search', request, locals()): - do_search() - - plugins.call('post_search', request, locals()) - - results = result_container.get_ordered_results() - - for result in results: - plugins.call('on_result', request, locals()) - - class SearchWithPlugins(Search): + """Similar to the Search class but call the plugins.""" + def __init__(self, search_query, request): super(SearchWithPlugins, self).__init__(search_query) self.request = request - self.request_data = request.request_data def search(self): - - def do_search(): + if plugins.call('pre_search', self.request, self): super(SearchWithPlugins, self).search() - search_with_plugins(do_search, self.search_query, self.request, self.request_data, self.result_container) + plugins.call('post_search', self.request, self) + + results = self.result_container.get_ordered_results() + + for result in results: + plugins.call('on_result', self.request, self, result) + return self.result_container diff --git a/tests/unit/test_plugins.py b/tests/unit/test_plugins.py @@ -6,8 +6,8 @@ from mock import Mock def get_search_mock(query, **kwargs): - return {'search': Mock(query=query, **kwargs), - 'result_container': Mock(answers=set())} + return Mock(search_query=Mock(query=query, **kwargs), + result_container=Mock(answers=set())) class PluginStoreTest(SearxTestCase): @@ -51,39 +51,39 @@ class SelfIPTest(SearxTestCase): request = Mock(user_plugins=store.plugins, remote_addr='127.0.0.1') request.headers.getlist.return_value = [] - ctx = get_search_mock(query='ip', pageno=1) - store.call('post_search', request, ctx) - self.assertTrue('127.0.0.1' in ctx['result_container'].answers) + search = get_search_mock(query='ip', pageno=1) + store.call('post_search', request, search) + self.assertTrue('127.0.0.1' in search.result_container.answers) - ctx = get_search_mock(query='ip', pageno=2) - store.call('post_search', request, ctx) - self.assertFalse('127.0.0.1' in ctx['result_container'].answers) + search = get_search_mock(query='ip', pageno=2) + store.call('post_search', request, search) + self.assertFalse('127.0.0.1' in search.result_container.answers) # User agent test request = Mock(user_plugins=store.plugins, user_agent='Mock') request.headers.getlist.return_value = [] - ctx = get_search_mock(query='user-agent', pageno=1) - store.call('post_search', request, ctx) - self.assertTrue('Mock' in ctx['result_container'].answers) + search = get_search_mock(query='user-agent', pageno=1) + store.call('post_search', request, search) + self.assertTrue('Mock' in search.result_container.answers) - ctx = get_search_mock(query='user-agent', pageno=2) - store.call('post_search', request, ctx) - self.assertFalse('Mock' in ctx['result_container'].answers) + search = get_search_mock(query='user-agent', pageno=2) + store.call('post_search', request, search) + self.assertFalse('Mock' in search.result_container.answers) - ctx = get_search_mock(query='user-agent', pageno=1) - store.call('post_search', request, ctx) - self.assertTrue('Mock' in ctx['result_container'].answers) + search = get_search_mock(query='user-agent', pageno=1) + store.call('post_search', request, search) + self.assertTrue('Mock' in search.result_container.answers) - ctx = get_search_mock(query='user-agent', pageno=2) - store.call('post_search', request, ctx) - self.assertFalse('Mock' in ctx['result_container'].answers) + search = get_search_mock(query='user-agent', pageno=2) + store.call('post_search', request, search) + self.assertFalse('Mock' in search.result_container.answers) - ctx = get_search_mock(query='What is my User-Agent?', pageno=1) - store.call('post_search', request, ctx) - self.assertTrue('Mock' in ctx['result_container'].answers) + search = get_search_mock(query='What is my User-Agent?', pageno=1) + store.call('post_search', request, search) + self.assertTrue('Mock' in search.result_container.answers) - ctx = get_search_mock(query='What is my User-Agent?', pageno=2) - store.call('post_search', request, ctx) - self.assertFalse('Mock' in ctx['result_container'].answers) + search = get_search_mock(query='What is my User-Agent?', pageno=2) + store.call('post_search', request, search) + self.assertFalse('Mock' in search.result_container.answers)