From 161e8c56a48d886ee3717bf2f9f0a157dc22cadf Mon Sep 17 00:00:00 2001
From: Dan Niles <56271899+dan-niles@users.noreply.github.com>
Date: Mon, 15 Jul 2024 16:36:37 +0530
Subject: [PATCH 1/3] Add e2e tests for zim metadata, channel.json, videos, and
 playlists

---
 scraper/tests-integration/integration.py | 68 ++++++++++++++++++++++--
 1 file changed, 63 insertions(+), 5 deletions(-)

diff --git a/scraper/tests-integration/integration.py b/scraper/tests-integration/integration.py
index 85602f4..6a2c2ed 100644
--- a/scraper/tests-integration/integration.py
+++ b/scraper/tests-integration/integration.py
@@ -1,3 +1,4 @@
+import json
 import os
 
 from zimscraperlib.zim import Archive
@@ -18,12 +19,69 @@ def test_zim_main_page():
     assert main_entry.get_redirect_entry().path == "index.html"
 
 
-def test_zim_scraper():
+def test_zim_metadata():
     """Ensure scraper and zim title are present in metadata"""
 
     zim_fh = Archive(ZIM_FILE_PATH)
-    scraper = zim_fh.get_text_metadata("scraper")
-    zim_title = zim_fh.get_text_metadata("Title")
 
-    assert "youtube2zim " in scraper
-    assert "openZIM_testing" in zim_title
+    assert "youtube2zim " in zim_fh.get_text_metadata("scraper")
+    assert "openZIM_testing" in zim_fh.get_text_metadata("Title")
+    assert "-" in zim_fh.get_text_metadata("Description")
+    assert "en" in zim_fh.get_text_metadata("Language")
+    assert "openZIM" in zim_fh.get_text_metadata("Publisher")
+    assert "openZIM_testing" in zim_fh.get_text_metadata("Creator")
+
+    assert zim_fh.get_item("profile.jpg").mimetype == "image/jpeg"
+    assert zim_fh.get_item("favicon.png").mimetype == "image/png"
+
+
+def test_zim_channel_json():
+    """Ensure channel.json exists and is valid"""
+
+    zim_fh = Archive(ZIM_FILE_PATH)
+    assert zim_fh.get_item("channel.json").mimetype == "application/json"
+    channel_json = zim_fh.get_content("channel.json")
+    channel_json = json.loads(channel_json)
+
+    assert channel_json["id"] == "UC8elThf5TGMpQfQc_VE917Q"
+    assert channel_json["channelName"] == "openZIM_testing"
+    assert channel_json["collectionType"] == "channel"
+    assert channel_json["mainPlaylist"] == "uploads_from_openzim_testing-917Q"
+
+
+def test_zim_videos():
+    """Ensure videos and video thumbnails are present in ZIM file"""
+
+    zim_fh = Archive(ZIM_FILE_PATH)
+    videos_json_list = [
+        "cloudy_sky_time_lapse_4k_free_footage_video_gopro_11-k02q.json",
+        "coffee_machine-DYvY.json",
+        "marvel_studios_avengers_endgame_official_trailer-TcMB.json",
+        "timelapse-9Tgo.json",
+    ]
+
+    for video_json_file in videos_json_list:
+        json_path = "videos/" + video_json_file
+        assert zim_fh.get_item(json_path).mimetype == "application/json"
+
+        video_json = zim_fh.get_content(json_path)
+        video_json = json.loads(video_json)
+
+        assert zim_fh.get_item(video_json["videoPath"]).mimetype == "video/webm"
+        assert zim_fh.get_item(video_json["thumbnailPath"]).mimetype == "image/webp"
+
+
+def test_zim_playlists():
+    """Ensure playlists json files are present in ZIM file"""
+
+    zim_fh = Archive(ZIM_FILE_PATH)
+    playlists_json_list = [
+        "coffee-O2wS.json",
+        "timelapses-QgGI.json",
+        "trailers-5Gph.json",
+        "uploads_from_openzim_testing-917Q.json",
+    ]
+
+    for playlist_json_file in playlists_json_list:
+        json_path = "playlists/" + playlist_json_file
+        assert zim_fh.get_item(json_path).mimetype == "application/json"

From 2f296a414105fd7233cf27d95e8d7a8081ca41db Mon Sep 17 00:00:00 2001
From: Dan Niles <56271899+dan-niles@users.noreply.github.com>
Date: Mon, 15 Jul 2024 22:15:50 +0530
Subject: [PATCH 2/3] Add subtitle file validation in integration tests

---
 scraper/tests-integration/integration.py | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/scraper/tests-integration/integration.py b/scraper/tests-integration/integration.py
index 6a2c2ed..1a0d4ae 100644
--- a/scraper/tests-integration/integration.py
+++ b/scraper/tests-integration/integration.py
@@ -70,6 +70,14 @@ def test_zim_videos():
         assert zim_fh.get_item(video_json["videoPath"]).mimetype == "video/webm"
         assert zim_fh.get_item(video_json["thumbnailPath"]).mimetype == "image/webp"
 
+        for subtitle in video_json["subtitleList"]:
+            assert (
+                zim_fh.get_item(
+                    video_json["subtitlePath"] + f"/video.{subtitle["code"]}.vtt"
+                ).mimetype
+                == "text/vtt"
+            )
+
 
 def test_zim_playlists():
     """Ensure playlists json files are present in ZIM file"""

From 0fcb9b3a66578d15327a1e512d59a9e60c8a291b Mon Sep 17 00:00:00 2001
From: Dan Niles <56271899+dan-niles@users.noreply.github.com>
Date: Tue, 16 Jul 2024 08:14:11 +0530
Subject: [PATCH 3/3] Update CHANGELOG mentioning addition of integration tests

---
 CHANGELOG | 1 +
 1 file changed, 1 insertion(+)

diff --git a/CHANGELOG b/CHANGELOG
index e043b50..15e8135 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -29,6 +29,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Add videos, subtitles, thumbnails and channel branding to the ZIM "on the fly" (#209)
 - Remove `--no-zim`, `--keep` CLI arguments
 - Add support to index content from `zimui` JSON files in the ZIM using custom `IndexData` (#224)
+- Add integration tests to check the content of the ZIM created by the scraper (#268)
 
 ## [2.3.0] - 2024-05-22