Merge branch 'searxng:master' into elasticsearch-custom-query

searx/engines/mojeek.py

@@ -6,7 +6,7 @@ from urllib.parse import urlencode
 from lxml import html
 
 from dateutil.relativedelta import relativedelta
-from searx.utils import eval_xpath, eval_xpath_list, extract_text, gen_useragent
+from searx.utils import eval_xpath, eval_xpath_list, extract_text
 
 about = {
     'website': 'https://mojeek.com',
@@ -63,7 +63,6 @@ def request(query, params):
         logger.debug(args["since"])
 
     params['url'] = f"{base_url}/search?{urlencode(args)}"
-    params['headers'] = {'User-Agent': gen_useragent()}
 
     return params
 

tests/unit/engines/__init__.py

@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+# pylint: disable=missing-module-docstring

tests/unit/engines/test_command.py

@@ -21,7 +21,7 @@ from searx.engines import command as command_engine
 from tests import SearxTestCase
 
 
-class TestCommandEngine(SearxTestCase):
+class TestCommandEngine(SearxTestCase):  # pylint: disable=missing-class-docstring
     def test_basic_seq_command_engine(self):
         ls_engine = command_engine
         ls_engine.command = ['seq', '{{QUERY}}']
@@ -33,10 +33,10 @@ class TestCommandEngine(SearxTestCase):
             {'number': '4', 'template': 'key-value.html'},
             {'number': '5', 'template': 'key-value.html'},
         ]
-        results = ls_engine.search('5'.encode('utf-8'), {'pageno': 1})
+        results = ls_engine.search('5', {'pageno': 1})
         self.assertEqual(results, expected_results)
 
-    def test_delimiter_parsing_command_engine(self):
+    def test_delimiter_parsing(self):
         searx_logs = '''DEBUG:searx.webapp:static directory is /home/n/p/searx/searx/static
 DEBUG:searx.webapp:templates directory is /home/n/p/searx/searx/templates
 DEBUG:searx.engines:soundcloud engine: Starting background initialization
@@ -140,10 +140,10 @@ INFO:werkzeug: * Debugger PIN: 299-578-362'''
         ]
 
         for i in [0, 1]:
-            results = echo_engine.search(''.encode('utf-8'), {'pageno': i + 1})
+            results = echo_engine.search('', {'pageno': i + 1})
             self.assertEqual(results, expected_results_by_page[i])
 
-    def test_regex_parsing_command_engine(self):
+    def test_regex_parsing(self):
         txt = '''commit 35f9a8c81d162a361b826bbcd4a1081a4fbe76a7
 Author: Noémi Ványi <sitbackandwait@gmail.com>
 Date:   Tue Oct 15 11:31:33 2019 +0200
@@ -168,11 +168,12 @@ commit '''
         git_log_engine.result_separator = '\n\ncommit '
         git_log_engine.delimiter = {}
         git_log_engine.parse_regex = {
-            'commit': '\w{40}',
+            'commit': r'\w{40}',
-            'author': '[\w* ]* <\w*@?\w*\.?\w*>',
+            'author': r'[\w* ]* <\w*@?\w*\.?\w*>',
-            'date': 'Date: .*',
+            'date': r'Date: .*',
-            'message': '\n\n.*$',
+            'message': r'\n\n.*$',
         }
+        git_log_engine.init({"command": git_log_engine.command, "parse_regex": git_log_engine.parse_regex})
         expected_results = [
             {
                 'commit': '35f9a8c81d162a361b826bbcd4a1081a4fbe76a7',
@@ -197,7 +198,7 @@ commit '''
             },
         ]
 
-        results = git_log_engine.search(''.encode('utf-8'), {'pageno': 1})
+        results = git_log_engine.search('', {'pageno': 1})
         self.assertEqual(results, expected_results)
 
     def test_working_dir_path_query(self):
@@ -207,7 +208,7 @@ commit '''
         ls_engine.delimiter = {'chars': ' ', 'keys': ['file']}
         ls_engine.query_type = 'path'
 
-        results = ls_engine.search('.'.encode(), {'pageno': 1})
+        results = ls_engine.search('.', {'pageno': 1})
         self.assertTrue(len(results) != 0)
 
         forbidden_paths = [
@@ -218,7 +219,7 @@ commit '''
             '/var',
         ]
         for forbidden_path in forbidden_paths:
-            self.assertRaises(ValueError, ls_engine.search, '..'.encode(), {'pageno': 1})
+            self.assertRaises(ValueError, ls_engine.search, forbidden_path, {'pageno': 1})
 
     def test_enum_queries(self):
         echo_engine = command_engine
@@ -227,7 +228,7 @@ commit '''
         echo_engine.query_enum = ['i-am-allowed-to-say-this', 'and-that']
 
         for allowed in echo_engine.query_enum:
-            results = echo_engine.search(allowed.encode(), {'pageno': 1})
+            results = echo_engine.search(allowed, {'pageno': 1})
             self.assertTrue(len(results) != 0)
 
         forbidden_queries = [
@@ -236,4 +237,4 @@ commit '''
             'prohibited',
         ]
         for forbidden in forbidden_queries:
-            self.assertRaises(ValueError, echo_engine.search, forbidden.encode(), {'pageno': 1})
+            self.assertRaises(ValueError, echo_engine.search, forbidden, {'pageno': 1})

tests/unit/engines/test_xpath.py

@@ -7,25 +7,43 @@ from searx.engines import xpath
 from tests import SearxTestCase
 
 
-class TestXpathEngine(SearxTestCase):
+class TestXpathEngine(SearxTestCase):  # pylint: disable=missing-class-docstring
+    html = """
+    <div>
+        <div class="search_result">
+            <a class="result" href="https://result1.com">Result 1</a>
+            <p class="content">Content 1</p>
+            <a class="cached" href="https://cachedresult1.com">Cache</a>
+        </div>
+        <div class="search_result">
+            <a class="result" href="https://result2.com">Result 2</a>
+            <p class="content">Content 2</p>
+            <a class="cached" href="https://cachedresult2.com">Cache</a>
+        </div>
+    </div>
+    """
+
     def test_request(self):
         xpath.search_url = 'https://url.com/{query}'
         xpath.categories = []
         xpath.paging = False
         query = 'test_query'
         dicto = defaultdict(dict)
+        dicto['language'] = 'all'
+        dicto['pageno'] = 1
         params = xpath.request(query, dicto)
         self.assertIn('url', params)
-        self.assertEquals('https://url.com/test_query', params['url'])
+        self.assertEqual('https://url.com/test_query', params['url'])
 
         xpath.search_url = 'https://url.com/q={query}&p={pageno}'
         xpath.paging = True
         query = 'test_query'
         dicto = defaultdict(dict)
+        dicto['language'] = 'all'
         dicto['pageno'] = 1
         params = xpath.request(query, dicto)
         self.assertIn('url', params)
-        self.assertEquals('https://url.com/q=test_query&p=1', params['url'])
+        self.assertEqual('https://url.com/q=test_query&p=1', params['url'])
 
     def test_response(self):
         # without results_xpath
@@ -38,24 +56,10 @@ class TestXpathEngine(SearxTestCase):
         self.assertRaises(AttributeError, xpath.response, '')
         self.assertRaises(AttributeError, xpath.response, '[]')
 
-        response = mock.Mock(text='<html></html>')
+        response = mock.Mock(text='<html></html>', status_code=200)
         self.assertEqual(xpath.response(response), [])
 
-        html = u"""
+        response = mock.Mock(text=self.html, status_code=200)
-        <div>
-            <div class="search_result">
-                <a class="result" href="https://result1.com">Result 1</a>
-                <p class="content">Content 1</p>
-                <a class="cached" href="https://cachedresult1.com">Cache</a>
-            </div>
-            <div class="search_result">
-                <a class="result" href="https://result2.com">Result 2</a>
-                <p class="content">Content 2</p>
-                <a class="cached" href="https://cachedresult2.com">Cache</a>
-            </div>
-        </div>
-        """
-        response = mock.Mock(text=html)
         results = xpath.response(response)
         self.assertEqual(type(results), list)
         self.assertEqual(len(results), 2)
@@ -80,6 +84,7 @@ class TestXpathEngine(SearxTestCase):
         results = xpath.response(response)
         self.assertTrue(results[0]['is_onion'])
 
+    def test_response_results_xpath(self):
         # with results_xpath
         xpath.results_xpath = '//div[@class="search_result"]'
         xpath.url_xpath = './/a[@class="result"]/@href'
@@ -93,10 +98,10 @@ class TestXpathEngine(SearxTestCase):
         self.assertRaises(AttributeError, xpath.response, '')
         self.assertRaises(AttributeError, xpath.response, '[]')
 
-        response = mock.Mock(text='<html></html>')
+        response = mock.Mock(text='<html></html>', status_code=200)
         self.assertEqual(xpath.response(response), [])
 
-        response = mock.Mock(text=html)
+        response = mock.Mock(text=self.html, status_code=200)
         results = xpath.response(response)
         self.assertEqual(type(results), list)
         self.assertEqual(len(results), 2)

tests/unit/processors/__init__.py

@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+# pylint: disable=missing-module-docstring

tests/unit/processors/test_online.py

@@ -0,0 +1,53 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+# pylint: disable=missing-module-docstring
+
+from searx.search import SearchQuery, EngineRef
+from searx.search.processors import online
+from searx.engines import load_engines
+from searx import engines
+
+from tests import SearxTestCase
+
+TEST_ENGINE_NAME = 'dummy engine'
+TEST_ENGINE = {
+    'name': TEST_ENGINE_NAME,
+    'engine': 'dummy',
+    'categories': 'general',
+    'shortcut': 'du',
+    'timeout': 3.0,
+    'tokens': [],
+}
+
+
+class TestOnlineProcessor(SearxTestCase):  # pylint: disable=missing-class-docstring
+
+    def setUp(self):
+        load_engines([TEST_ENGINE])
+
+    def tearDown(self):
+        load_engines([])
+
+    def _get_params(self, online_processor, search_query, engine_category):
+        params = online_processor.get_params(search_query, engine_category)
+        self.assertIsNotNone(params)
+        assert params is not None
+        return params
+
+    def test_get_params_default_params(self):
+        engine = engines.engines[TEST_ENGINE_NAME]
+        online_processor = online.OnlineProcessor(engine, TEST_ENGINE_NAME)
+        search_query = SearchQuery('test', [EngineRef(TEST_ENGINE_NAME, 'general')], 'all', 0, 1, None, None, None)
+        params = self._get_params(online_processor, search_query, 'general')
+        self.assertIn('method', params)
+        self.assertIn('headers', params)
+        self.assertIn('data', params)
+        self.assertIn('url', params)
+        self.assertIn('cookies', params)
+        self.assertIn('auth', params)
+
+    def test_get_params_useragent(self):
+        engine = engines.engines[TEST_ENGINE_NAME]
+        online_processor = online.OnlineProcessor(engine, TEST_ENGINE_NAME)
+        search_query = SearchQuery('test', [EngineRef(TEST_ENGINE_NAME, 'general')], 'all', 0, 1, None, None, None)
+        params = self._get_params(online_processor, search_query, 'general')
+        self.assertIn('User-Agent', params['headers'])

tests/unit/test_query.py

@@ -247,6 +247,7 @@ class TestBang(SearxTestCase):  # pylint:disable=missing-class-docstring
                 self.assertEqual(query.user_query_parts, TestBang.THE_QUERY.split(' '))
 
     def test_specific(self):
+        load_engines(TEST_ENGINES)
         for bang in TestBang.SPECIFIC_BANGS:
             with self.subTest(msg="Check bang is specific", bang=bang):
                 query_text = TestBang.THE_QUERY + ' ' + bang