Serve YouTube URLs of session recordings

Author: egeakman

data/examples/README.md

@@ -1,4 +1,4 @@
-# Explaining the output data
+# Explaining the Output Data
 
 **Note:** Some of the fields may be `null` or empty (`""`).
 
@@ -27,19 +27,16 @@
         "delivery": "in-person",
         "resources": [
             {
-                "resource": "https://example.com/notebook.ipynb",
-                "description": "Notebook used in the talk"
-            },
-            {
-                "resource": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
-                "description": "Video of the robot in action"
+                "resource": "https://example.com/slides.pdf",
+                "description": "Slides for the session"
             }
         ...
         ],
         "room": "South Hall 2A",
-        "start": "2024-07-10T14:00:00+02:00",
-        "end": "2024-07-10T15:00:00+02:00",
-        "website_url": "https://ep2024.europython.eu/session/example-talk/",
+        "start": "2099-07-10T14:00:00+02:00",
+        "end": "2099-07-10T15:00:00+02:00",
+        "website_url": "https://ep2099.europython.eu/session/example-talk/",
+        "youtube_url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ&pp=ygUJcmljayByb2xs",
         "sessions_in_parallel": [
         "F7G8H9",
         ...
@@ -81,6 +78,7 @@ The fields are as follows:
 | `start`                | `string (datetime ISO format)` \| `null`  | Start time of the session                                     |
 | `end`                  | `string (datetime ISO format)` \| `null`  | End time of the session                                       |
 | `website_url`          | `string`                                  | URL of the session on the conference website                  |
+| `youtube_url`          | `string` \| `null`                        | URL of the session's video on YouTube                         |
 | `sessions_in_parallel` | `array[string]` \| `null`                 | List of codes of sessions happening in parallel               |
 | `sessions_after`       | `array[string]` \| `null`                 | List of codes of sessions happening after this session        |
 | `sessions_before`      | `array[string]` \| `null`                 | List of codes of sessions happening before this session       |
@@ -137,3 +135,84 @@ The fields are as follows:
 | `twitter_url`  | `string` \| `null` | URL of the speaker's Twitter profile                                  |
 | `mastodon_url` | `string` \| `null` | URL of the speaker's Mastodon profile                                 |
 | `website_url`  | `string`           | URL of the speaker's profile on the conference website                |
+
+ 
+
+## `schedule.json`
+
+<details>
+<summary>Example schedule data JSON</summary>
+
+```json
+{
+    "days": {
+        "2099-07-08": {
+            "events": [
+                {
+                    "code": "LMN123",
+                    "title": "Welcome and Keynote",
+                    "speakers": [],
+                    "session_type": "Announcements",
+                    "slug": "welcome-keynote",
+                    "track": null,
+                    "level": "beginner",
+                    "rooms": [
+                        "Room A",
+                        "Room B"
+                    ],
+                    "start": "2099-07-08T08:00:00+02:00",
+                    "duration": 60,
+                    "tweet": "",
+                    "website_url": "https://ep2099.europython.eu/session/welcome-keynote"
+                },
+                {
+                    "code": "OPQ456",
+                    "title": "Advanced Python Techniques",
+                    "speakers": [
+                        {
+                            "avatar": "https://pretalx.com/media/avatars/picture.jpg",
+                            "code": "RST789",
+                            "name": "John Doe",
+                            "slug": "john-doe",
+                            "website_url": "https://ep2099.europython.eu/speaker/john-doe"
+                        }
+                    ],
+                    "session_type": "Tutorial",
+                    "slug": "advanced-python-techniques",
+                    "track": "CPython Internals",
+                    "level": "advanced",
+                    "rooms": [
+                        "Room C"
+                    ],
+                    "start": "2099-07-08T10:00:00+02:00",
+                    "duration": 90,
+                    "tweet": "",
+                    "website_url": "https://ep2099.europython.eu/advanced-python-techniques"
+                }
+            ]
+        }
+    }
+}
+```
+</details>
+
+&nbsp;
+
+The fields are as follows:
+
+| Key            | Type                        | Notes                                                      |
+|----------------|-----------------------------|------------------------------------------------------------|
+| `days`         | `object`                    | Contains schedule by date                                  |
+| `events`       | `array[object]`             | List of events for a particular day                        |
+| `code`         | `string`                    | Unique identifier for the event                            |
+| `title`        | `string`                    | Title of the event                                         |
+| `speakers`     | `array[object]`             | List of speakers for the event (if applicable)             |
+| `session_type` | `string`                    | Type of event (e.g. Announcements, Workshop, etc.)         |
+| `slug`         | `string`                    | URL-friendly version of the event title                    |
+| `track`        | `string` \| `null`          | Track associated with the event (e.g. Web, PyData, etc.)   |
+| `level`        | `string`                    | Level of the event (beginner, intermediate, advanced)       |
+| `rooms`        | `array[string]`             | List of rooms the event is being held in                   |
+| `start`        | `string (datetime ISO)`      | Start time of the event                                    |
+| `duration`     | `integer`                   | Duration of the event in minutes                           |
+| `tweet`        | `string` \| `null`          | Tweet-length description of the event                      |
+| `website_url`  | `string`                    | URL of the event on the conference website                 |

src/download.py

@@ -19,6 +19,7 @@
     # saving us later time with joining the answers.
     "submissions?questions=all&state=confirmed",
     "speakers?questions=all",
+    "p/youtube",
 ]
 
 Config.raw_path.mkdir(parents=True, exist_ok=True)
@@ -45,7 +46,8 @@
 
     pbar.close()
 
-    filename = resource.split("?")[0]  # To get rid of "?questions"
+    # To get the resource name without extra parameters
+    filename = resource.split("?")[0].split("/")[-1]
     filename = f"{filename}_latest.json"
     filepath = Config.raw_path / filename
 

src/models/europython.py

@@ -142,6 +142,7 @@ class EuroPythonSession(BaseModel):
     next_session: str | None = None
     prev_session: str | None = None
     slot_count: int = Field(..., exclude=True)
+    youtube_url: str | None = None
 
     @field_validator("room", mode="before")
     @classmethod

src/transform.py

@@ -16,12 +16,16 @@
     )
     pretalx_schedule = Parse.schedule(Config.raw_path / "schedule_latest.json")
 
+    # Parse the YouTube data
+    youtube_data = Parse.youtube(Config.raw_path / "youtube_latest.json")
+
     print("Computing timing relationships...")
     TimingRelationships.compute(pretalx_submissions.values())
 
     print("Transforming the data...")
     ep_sessions = Transform.pretalx_submissions_to_europython_sessions(
-        pretalx_submissions
+        pretalx_submissions,
+        youtube_data,
     )
     ep_speakers = Transform.pretalx_speakers_to_europython_speakers(pretalx_speakers)
     ep_schedule = Transform.pretalx_schedule_to_europython_schedule(

src/utils/parse.py

@@ -61,3 +61,14 @@ def schedule(input_file: Path | str) -> PretalxSchedule:
             schedule = PretalxSchedule.model_validate(js)
 
         return schedule
+
+    @staticmethod
+    def youtube(input_file: Path | str) -> dict[str, str]:
+        """
+        Returns the Session code to YouTube URL mapping
+        """
+        with open(input_file) as fd:
+            js = json.load(fd)
+            youtube_data = {s["submission"]: s["youtube_link"] for s in js}
+
+        return youtube_data

src/utils/transform.py

@@ -15,6 +15,7 @@ class Transform:
     @staticmethod
     def pretalx_submissions_to_europython_sessions(
         submissions: dict[str, PretalxSubmission],
+        youtube_data: dict[str, str],
     ) -> dict[str, EuroPythonSession]:
         """
         Transforms the given Pretalx submissions to EuroPython sessions
@@ -58,6 +59,7 @@ def pretalx_submissions_to_europython_sessions(
                 next_session=TimingRelationships.get_next_session(submission.code),
                 prev_session=TimingRelationships.get_prev_session(submission.code),
                 slot_count=submission.slot_count,
+                youtube_url=youtube_data.get(submission.code),
             )
             ep_sessions[code] = ep_session