Modify analyser tests to reflect the use of extractor

3 files changed, 29 insertions, 56 deletions

diff --git a/tests/analyser.py b/tests/analyser.py
index e91e72f..c698942 100644
--- a/tests/analyser.py
+++ b/tests/analyser.py
@@ -1,4 +1,3 @@
-from hashlib import md5
 from typing import Any, Dict
 from unittest import TestCase
 from unittest.mock import MagicMock, patch
@@ -9,27 +8,9 @@ from pymongo.collection import Collection
 from components.subscriptions.main import Subscription
 from components.subscriptions.typing import SubsDict
 from components.videos import VideoTuple
-from components.ytdlp import obtain_vid_info
 from data_analyser.utils import analyse_video, analyse_subscription, analyse_collection
-
-def url_based_extract_info(url: str, **_: Any) -> Dict[str, Any]:
-    hash = md5(url.encode()).hexdigest()
-    hash_int = int(hash[:4], 16)
-    seconds = hash_int %  60
-    minutes = hash_int // 60
-    hours   = minutes  // 60
-    minutes = minutes  %  60
-    if hours:
-        duration_string = "%d:%02d:%02d" % (hours, minutes, seconds)
-    elif minutes:
-        duration_string = "%d:%02d" % (minutes, seconds)
-    else:
-        duration_string = "%d" % (seconds, )
-    return {
-        "duration_string": duration_string,
-        "other_key": "other_value",
-        "other key": "other value",
-    }
+from .utils.vid_url_to_html import obtain_vid_duration
+from .utils.get_random_vid_info import get_random_vid_duration
 
 class TestAnalyser(TestCase):
     def setUp(self) -> None:
@@ -45,69 +26,52 @@ class TestAnalyser(TestCase):
         self.sub1.insert()
         self.sub1.fetch()
 
-        self.mock_ydl_class = patch('components.ytdlp.YoutubeDL').start()
-        self.mock_ydl_instance = MagicMock()
-        self.mock_ydl_class.return_value.__enter__.return_value = self.mock_ydl_instance
-        self.mock_ydl_instance.extract_info.side_effect = url_based_extract_info
+        self.mock_vid_duration = patch('data_analyser.utils.obtain_vid_duration').start()
+        self.mock_vid_duration.side_effect = obtain_vid_duration
         self.addCleanup(patch.stopall)
 
-    def test_obtain_vid_info(self) -> None:
-        result = obtain_vid_info(self.sub1.videos[0].link)
-        expected_duration = url_based_extract_info(self.sub1.videos[0].link)["duration_string"]
-        self.assertEqual(result, {"duration_string": expected_duration})
-
-    def test_obtain_vid_info_with_exception(self) -> None:
-        self.mock_ydl_instance.extract_info.side_effect = Exception("Network error")
-
-        result = obtain_vid_info(self.sub1.videos[0].link)
-        expected_keys = {"duration_string"}
-        actual_keys = set(result.keys())
-        self.assertEqual(expected_keys, actual_keys,
-                         "Expected keys %s, got %s." % (expected_keys, actual_keys))
-        self.assertRegex(result["duration_string"], "^-1:[0-5][0-9]$")
-
     def test_analyse_video(self) -> None:
         modified_vid = analyse_video(self.sub1.videos[1])
 
-        expected_duration = url_based_extract_info(self.sub1.videos[1].link)["duration_string"]
+        expected_duration = get_random_vid_duration(self.sub1.videos[1].link)
         self.assertIsInstance(modified_vid, VideoTuple)
-        self.assertEqual(modified_vid.duration_string, expected_duration)
+        self.assertEqual(modified_vid.duration, expected_duration)
         self.assertTrue(modified_vid.analysed)
 
     def test_analyse_video_with_exception(self) -> None:
-        self.mock_ydl_instance.extract_info.side_effect = Exception("Network error")
+        self.mock_vid_duration.side_effect = Exception("Network error")
 
         modified_vid = analyse_video(self.sub1.videos[0])
         self.assertIsInstance(modified_vid, VideoTuple)
-        self.assertRegex(modified_vid.duration_string, "^-1:[0-5][0-9]$")
+        self.assertLess(modified_vid.duration, 0)
         self.assertTrue(modified_vid.analysed)
 
     def test_analyse_subscription(self) -> None:
         self.assertTrue(analyse_subscription(self.sub1))
         for vid in self.sub1.videos:
-            expected_duration = url_based_extract_info(vid.link)["duration_string"]
-            self.assertEqual(vid.duration_string, expected_duration)
+            expected_duration = get_random_vid_duration(vid.link)
+            self.assertEqual(vid.duration, expected_duration)
             self.assertTrue(vid.analysed)
 
     def test_analyse_subscription_with_further_fetch(self) -> None:
         self.assertTrue(analyse_subscription(self.sub1))
         for vid in self.sub1.videos:
-            expected_duration = url_based_extract_info(vid.link)["duration_string"]
-            self.assertEqual(vid.duration_string, expected_duration)
+            expected_duration = get_random_vid_duration(vid.link)
+            self.assertEqual(vid.duration, expected_duration)
             self.assertTrue(vid.analysed)
         self.sub1.link = r"tests/data/feed@ytnnews24@002.xml"
         self.sub1.fetch()
         self.assertTrue(analyse_subscription(self.sub1))
         for vid in self.sub1.videos:
-            expected_duration = url_based_extract_info(vid.link)["duration_string"]
-            self.assertEqual(vid.duration_string, expected_duration)
+            expected_duration = get_random_vid_duration(vid.link)
+            self.assertEqual(vid.duration, expected_duration)
             self.assertTrue(vid.analysed)
 
     def test_analyse_subscription_without_further_fetch(self) -> None:
         self.assertTrue(analyse_subscription(self.sub1))
         for vid in self.sub1.videos:
-            expected_duration = url_based_extract_info(vid.link)["duration_string"]
-            self.assertEqual(vid.duration_string, expected_duration)
+            expected_duration = get_random_vid_duration(vid.link)
+            self.assertEqual(vid.duration, expected_duration)
             self.assertTrue(vid.analysed)
         self.assertFalse(analyse_subscription(self.sub1))
 
@@ -115,8 +79,8 @@ class TestAnalyser(TestCase):
         self.assertEqual(analyse_collection(self.collection), 1)
         for sub_dict in self.collection.find():
             for vid in map(VideoTuple._make, sub_dict["videos"]):
-                expected_duration = url_based_extract_info(vid.link)["duration_string"]
-                self.assertEqual(vid.duration_string, expected_duration)
+                expected_duration = get_random_vid_duration(vid.link)
+                self.assertEqual(vid.duration, expected_duration)
                 self.assertTrue(vid.analysed)
         self.sub2 = Subscription(
             _id="yt:channel:7YOGHUfC1Tb6E4pudI9STA",
@@ -130,8 +94,8 @@ class TestAnalyser(TestCase):
         for sub_dict in self.collection.find():
             sub = Subscription(**sub_dict)
             for vid in map(VideoTuple._make, sub_dict["videos"]):
-                expected_duration = url_based_extract_info(vid.link)["duration_string"]
-                self.assertEqual(vid.duration_string, expected_duration)
+                expected_duration = get_random_vid_duration(vid.link)
+                self.assertEqual(vid.duration, expected_duration)
                 self.assertTrue(vid.analysed)
 
     def tearDown(self) -> None:
diff --git a/tests/utils/get_random_vid_info.py b/tests/utils/get_random_vid_info.py
new file mode 100644
index 0000000..f85a226
--- /dev/null
+++ b/tests/utils/get_random_vid_info.py
@@ -0,0 +1,5 @@
+from hashlib import md5
+
+def get_random_vid_duration(url: str) -> int:
+    hash = md5(url.encode()).hexdigest()
+    return int(hash[:4], 16)
diff --git a/tests/utils/vid_url_to_html.py b/tests/utils/vid_url_to_html.py
index 355e830..60f6ccc 100644
--- a/tests/utils/vid_url_to_html.py
+++ b/tests/utils/vid_url_to_html.py
@@ -1,6 +1,7 @@
 from urllib.parse import urlparse, parse_qs
 
 from components.extractor.check_url import is_video
+from .get_random_vid_info import get_random_vid_duration
 
 def extract_vid_id(url: str) -> str:
     if not is_video(url):
@@ -16,3 +17,6 @@ def extract_vid_id(url: str) -> str:
 def get_vid_html_from_url(url: str) -> str:
     with open(f'tests/data/video@{extract_vid_id(url)}.html', 'r') as file:
         return file.read()
+
+def obtain_vid_duration(url:str, html: str = '') -> int:
+    return get_random_vid_duration(url)