fix podcast generation

surfsense_backend/.env.example

@@ -1,5 +1,12 @@
 DATABASE_URL=postgresql+asyncpg://postgres:postgres@localhost:5432/surfsense
 
+SECRET_KEY=SECRET
+NEXT_FRONTEND_URL=http://localhost:3000
+
+#Celery Config
+CELERY_BROKER_URL=redis://localhost:6379/0
+CELERY_RESULT_BACKEND=redis://localhost:6379/0
+
 #Celery Config
 CELERY_BROKER_URL=redis://localhost:6379/0
 CELERY_RESULT_BACKEND=redis://localhost:6379/0

surfsense_backend/alembic/versions/32_add_podcast_staleness_detection.py

@@ -1,50 +0,0 @@
-"""Add podcast staleness detection columns to chats and podcasts tables
-
-This feature allows the system to detect when a podcast is outdated compared to the current state of the chat it was generated from, enabling users to regenerate podcasts when needed.
-
-Revision ID: 32
-Revises: 31
-"""
-
-from collections.abc import Sequence
-
-import sqlalchemy as sa
-
-from alembic import op
-
-# revision identifiers
-revision: str = "32"
-down_revision: str | None = "31"
-branch_labels: str | Sequence[str] | None = None
-depends_on: str | Sequence[str] | None = None
-
-
-def upgrade() -> None:
-    """Add state_version, chat_state_version, and chat_id to chats and podcasts tables."""
-
-    # Add state_version column to chats table with default value of 1
-    op.add_column(
-        "chats",
-        sa.Column("state_version", sa.BigInteger(), nullable=False, server_default="1"),
-    )
-
-    # Add chat_state_version column to podcasts table (nullable, set when podcast is generated)
-    op.add_column(
-        "podcasts", sa.Column("chat_state_version", sa.BigInteger(), nullable=True)
-    )
-
-    # Add chat_id column to podcasts table (nullable, set when podcast is generated from a chat)
-    op.add_column("podcasts", sa.Column("chat_id", sa.Integer(), nullable=True))
-
-
-def downgrade() -> None:
-    """Remove state_version, chat_state_version, and chat_id columns."""
-
-    # Remove chat_state_version from podcasts table
-    op.drop_column("podcasts", "chat_state_version")
-
-    # Remove chat_id from podcasts table
-    op.drop_column("podcasts", "chat_id")
-
-    # Remove state_version from chats table
-    op.drop_column("chats", "state_version")

surfsense_backend/alembic/versions/34_add_podcast_staleness_detection.py

@@ -0,0 +1,60 @@
+"""Add podcast staleness detection columns to chats and podcasts tables
+
+This feature allows the system to detect when a podcast is outdated compared to
+the current state of the chat it was generated from, enabling users to regenerate
+podcasts when needed.
+
+Revision ID: 34
+Revises: 33
+"""
+
+from collections.abc import Sequence
+
+from alembic import op
+
+# revision identifiers
+revision: str = "34"
+down_revision: str | None = "33"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    """Add columns only if they don't already exist (safe for re-runs)."""
+
+    # Add 'state_version' column to chats table (default 1)
+    op.execute("""
+        ALTER TABLE chats
+        ADD COLUMN IF NOT EXISTS state_version BIGINT DEFAULT 1 NOT NULL
+    """)
+
+    # Add 'chat_state_version' column to podcasts table
+    op.execute("""
+        ALTER TABLE podcasts
+        ADD COLUMN IF NOT EXISTS chat_state_version BIGINT
+    """)
+
+    # Add 'chat_id' column to podcasts table
+    op.execute("""
+        ALTER TABLE podcasts
+        ADD COLUMN IF NOT EXISTS chat_id INTEGER
+    """)
+
+
+def downgrade() -> None:
+    """Remove columns only if they exist."""
+
+    op.execute("""
+        ALTER TABLE podcasts
+        DROP COLUMN IF EXISTS chat_state_version
+    """)
+
+    op.execute("""
+        ALTER TABLE podcasts
+        DROP COLUMN IF EXISTS chat_id
+    """)
+
+    op.execute("""
+        ALTER TABLE chats
+        DROP COLUMN IF EXISTS state_version
+    """)

surfsense_backend/app/agents/podcaster/nodes.py

@@ -29,6 +29,7 @@ async def create_podcast_transcript(
     configuration = Configuration.from_runnable_config(config)
     user_id = configuration.user_id
     search_space_id = configuration.search_space_id
+    podcast_title = configuration.podcast_title
 
     # Get user's long context LLM
     llm = await get_user_long_context_llm(state.db_session, user_id, search_space_id)
@@ -37,8 +38,8 @@ async def create_podcast_transcript(
         print(error_message)
         raise RuntimeError(error_message)
 
-    # Get the prompt
-    prompt = get_podcast_generation_prompt()
+    # Get the prompt with podcast_title as user_prompt
+    prompt = get_podcast_generation_prompt(user_prompt=podcast_title)
 
     # Create the messages
     messages = [

surfsense_backend/app/agents/podcaster/prompts.py

@@ -1,11 +1,19 @@
 import datetime
 
 
-def get_podcast_generation_prompt():
+def get_podcast_generation_prompt(user_prompt: str | None = None):
+    user_prompt_section = ""
+    if user_prompt:
+        user_prompt_section = f"""
+<user_instructions>
+{user_prompt}
+</user_instructions>
+"""
+
     return f"""
 Today's date: {datetime.datetime.now().strftime("%Y-%m-%d")}
 <podcast_generation_system>
-You are a master podcast scriptwriter, adept at transforming diverse input content into a lively, engaging, and natural-sounding conversation between two distinct podcast hosts. Your primary objective is to craft authentic, flowing dialogue that captures the spontaneity and chemistry of a real podcast discussion, completely avoiding any hint of robotic scripting or stiff formality. Think dynamic interplay, not just information delivery.
+You are a master podcast scriptwriter, adept at transforming diverse input content into a lively, engaging, and natural-sounding conversation between two distinct podcast hosts. Your primary objective is to craft authentic, flowing dialogue that captures the spontaneity and chemistry of a real podcast discussion, completely avoiding any hint of robotic scripting or stiff formality. Think dynamic interplay, not just information delivery.{user_prompt_section}
 
 <input>
 - '<source_content>': A block of text containing the information to be discussed in the podcast. This could be research findings, an article summary, a detailed outline, user chat history related to the topic, or any other relevant raw information. The content might be unstructured but serves as the factual basis for the podcast dialogue.

surfsense_backend/app/tasks/celery_tasks/podcast_tasks.py

@@ -46,7 +46,7 @@ def generate_chat_podcast_task(
     Args:
         chat_id: ID of the chat to generate podcast from
         search_space_id: ID of the search space
-        podcast_title: Title for the podcast
+        podcast_title: Title for the podcast (used as user prompt for generation)
         user_id: ID of the user
     """
     loop = asyncio.new_event_loop()

surfsense_backend/app/tasks/podcast_tasks.py

@@ -96,7 +96,7 @@ async def generate_chat_podcast(
 
         config = {
             "configurable": {
-                "podcast_title": "SurfSense",
+                "podcast_title": podcast_title,
                 "user_id": str(user_id),
                 "search_space_id": search_space_id,
             }