chore: update env config, docker-compose, requirements and bucket init

.env

@@ -14,7 +14,7 @@ RUN_INIT=true
 #   - full: Full setup including demo school and users (infra → demo-school → demo-users → gais-data)
 #   - infra,demo-school,demo-users: Custom combination (comma-separated)
 #   - infra,gais-data: Infrastructure + GAIS data import
-INIT_MODE=full
+INIT_MODE=infra,demo-school,demo-users
 
 ## =====================================================
 ## APP INFORMATION & METADATA
@@ -93,7 +93,7 @@ GOOGLE_CLIENT_SECRETS_FILE=Users/kcar/ClassroomCopilot/backend/app/secrets/googl
 # External Service Endpoints
 TIKA_URL=https://tika.kevlarai.com
 TIKA_TIMEOUT=300
-DOCLING_URL=http://ubuntu-server:5001
+DOCLING_URL=https://docling.kevlarai.com
 
 ## =====================================================
 ## DOCUMENT STRUCTURE DISCOVERY & ANALYSIS
@@ -200,8 +200,8 @@ DOCLING_VLM_DO_PICTURE_DESCRIPTION=true
 ## =====================================================
 VITE_APP_URL=https://app.classroomcopilot.ai
 APP_API_URL=https://api.classroomcopilot.ai
-APP_GRAPH_URL=https://graph.classroomcopilot.ai
-APP_BOLT_URL=bolt://bolt.classroomcopilot.ai
+APP_GRAPH_URL=https://192.168.0.208
+APP_BOLT_URL=bolt://192.168.0.208
 
 ## =====================================================
 ## REDIS CONFIGURATION & ENVIRONMENT ISOLATION

docker-compose.yml

@@ -10,7 +10,7 @@ services:
     networks:
       - kevlarai-network
     healthcheck:
-      test: ["CMD", "redis-cli", "ping"]
+      test: [ "CMD", "redis-cli", "ping" ]
       interval: 5s
       timeout: 3s
       retries: 5
@@ -28,7 +28,7 @@ services:
       - RUN_INIT=true
       - INIT_MODE=${INIT_MODE:-infra} # Set via .env or override: INIT_MODE=infra,gais-data
       - INIT_ONLY=true # Exit after init, don't start server
-    command: ["./docker-entrypoint.sh", "init-only"]
+    command: [ "./docker-entrypoint.sh", "init-only" ]
     depends_on:
       redis:
         condition: service_healthy
@@ -49,7 +49,7 @@ services:
       - RUN_INIT=${RUN_INIT:-false} # Set to 'true' to run init on startup
       - INIT_MODE=${INIT_MODE:-infra} # Which init tasks to run
     ports:
-      - 8000:8000
+      - 8080:8080
     depends_on:
       redis:
         condition: service_healthy
@@ -59,6 +59,7 @@ services:
 volumes:
   redis-data:
 
+
 networks:
   kevlarai-network:
     name: kevlarai-network

requirements.txt

@@ -53,7 +53,7 @@ w3lib
 scikit-learn
 
 # Google APIs
-youtube-transcript-api
+youtube-transcript-api>=1.2.3
 google-api-python-client
 google-auth-oauthlib
 

routers/external/youtube.py

@@ -28,9 +28,9 @@ youtube = build('youtube', 'v3', developerKey=os.getenv('YOUTUBE_API_KEY'))
 @router.get("/youtube-proxy")
 async def youtube_proxy(videoId: str):
     try:
-        # Fetch transcript using youtube-transcript-api
-        transcript = YouTubeTranscriptApi.get_transcript(videoId, languages=['en'])
-        transcript_lines = [{"start": entry["start"], "duration": entry["duration"], "text": entry["text"]} for entry in transcript]
+        transcript_list = YouTubeTranscriptApi().list(videoId)
+        transcript = transcript_list.find_transcript(['en']).fetch()
+        transcript_lines = [{"start": entry.start, "duration": entry.duration, "text": entry.text} for entry in transcript]
 
         # Fetch video details using YouTube Data API
         video_response = youtube.videos().list(

run/initialization/buckets.py

@@ -46,6 +46,15 @@ def initialize_buckets() -> dict:
                     file_size_limit=1000 * 1024 * 1024,  # 1GB
                 )
             },
+            # Exam Board files
+            {
+                "id": "cc.examboards",
+                "options": CreateBucketOptions(
+                    name="Classroom Copilot Exam Board Files",
+                    public=False,
+                    file_size_limit=1000 * 1024 * 1024,  # 1GB
+                )
+            },
         ]
         
         results = {}