introduced blocknote editor

surfsense_backend/alembic/versions/38_add_blocknote_fields_to_documents.py

@@ -0,0 +1,41 @@
+"""38_add_blocknote_fields_to_documents
+
+Revision ID: 38
+Revises: 37
+
+"""
+from collections.abc import Sequence
+
+import sqlalchemy as sa
+from sqlalchemy.dialects import postgresql
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision: str = '38'
+down_revision: str | None = '37'
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    """Upgrade schema - Add BlockNote fields only."""
+    
+    op.add_column(
+        'documents',
+        sa.Column('blocknote_document', postgresql.JSONB(astext_type=sa.Text()), nullable=True),
+    )
+    op.add_column(
+        'documents',
+        sa.Column('content_needs_reindexing', sa.Boolean(), nullable=False, server_default=sa.false()),
+    )
+    op.add_column(
+        'documents',
+        sa.Column('last_edited_at', sa.TIMESTAMP(timezone=True), nullable=True)
+    )
+
+def downgrade() -> None:
+    """Downgrade schema - Remove BlockNote fields."""
+    op.drop_column('documents', 'last_edited_at')
+    op.drop_column('documents', 'content_needs_reindexing')
+    op.drop_column('documents', 'blocknote_document')

surfsense_backend/app/db.py

@@ -20,7 +20,7 @@ from sqlalchemy import (
     UniqueConstraint,
     text,
 )
-from sqlalchemy.dialects.postgresql import UUID
+from sqlalchemy.dialects.postgresql import JSONB, UUID
 from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
 from sqlalchemy.orm import DeclarativeBase, Mapped, declared_attr, relationship
 
@@ -178,6 +178,17 @@ class Document(BaseModel, TimestampMixin):
     content_hash = Column(String, nullable=False, index=True, unique=True)
     unique_identifier_hash = Column(String, nullable=True, index=True, unique=True)
     embedding = Column(Vector(config.embedding_model_instance.dimension))
+    
+    # BlockNote live editing state (NULL when never edited)
+    blocknote_document = Column(JSONB, nullable=True)
+    
+    # blocknote background reindex flag
+    content_needs_reindexing = Column(
+        Boolean, nullable=False, default=False, server_default=text("false")
+    )
+    
+    # Track when blocknote document was last edited
+    last_edited_at = Column(TIMESTAMP(timezone=True), nullable=True)
 
     search_space_id = Column(
         Integer, ForeignKey("searchspaces.id", ondelete="CASCADE"), nullable=False

surfsense_backend/app/routes/__init__.py

@@ -5,6 +5,7 @@ from .airtable_add_connector_route import (
 )
 from .chats_routes import router as chats_router
 from .documents_routes import router as documents_router
+from .editor_routes import router as editor_router
 from .google_calendar_add_connector_route import (
     router as google_calendar_add_connector_router,
 )
@@ -21,6 +22,7 @@ from .search_spaces_routes import router as search_spaces_router
 router = APIRouter()
 
 router.include_router(search_spaces_router)
+router.include_router(editor_router)
 router.include_router(documents_router)
 router.include_router(podcasts_router)
 router.include_router(chats_router)

surfsense_backend/app/routes/editor_routes.py

@@ -0,0 +1,161 @@
+"""
+Editor routes for BlockNote document editing.
+"""
+from datetime import UTC, datetime
+from typing import Any
+
+from fastapi import APIRouter, Depends, HTTPException
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.db import Document, SearchSpace, User, get_async_session
+from app.users import current_active_user
+from app.utils.blocknote_converter import (
+    convert_blocknote_to_markdown,
+    convert_markdown_to_blocknote,
+)
+
+router = APIRouter()
+
+
+@router.get("/documents/{document_id}/editor-content")
+async def get_editor_content(
+    document_id: int,
+    session: AsyncSession = Depends(get_async_session),
+    user: User = Depends(current_active_user),
+):
+    """
+    Get document content for editing.
+    
+    Returns BlockNote JSON document. If blocknote_document is NULL,
+    attempts to convert from `content` - though this won't work well
+    for old documents that only have summaries.
+    """
+    result = await session.execute(
+        select(Document)
+        .join(SearchSpace)
+        .filter(Document.id == document_id, SearchSpace.user_id == user.id)
+    )
+    document = result.scalars().first()  # ✅ Changed from scalar_one_or_none()
+    
+    if not document:
+        raise HTTPException(status_code=404, detail="Document not found")
+    
+    # If blocknote_document exists, return it
+    if document.blocknote_document:
+        return {
+            "document_id": document.id,
+            "title": document.title,
+            "blocknote_document": document.blocknote_document,
+            "last_edited_at": document.last_edited_at.isoformat() if document.last_edited_at else None,
+        }
+    
+    # For old documents without blocknote_document, return error
+    # (Can't convert summary back to full document)
+    raise HTTPException(
+        status_code=400,
+        detail="This document was uploaded before editing was enabled. Please re-upload to enable editing."
+    )
+
+
+@router.put("/documents/{document_id}/blocknote-content")
+async def update_blocknote_content(
+    document_id: int,
+    data: dict[str, Any],
+    session: AsyncSession = Depends(get_async_session),
+    user: User = Depends(current_active_user),
+):
+    """
+    Auto-save BlockNote document during editing.
+    Only updates blocknote_document field, not content.
+    """
+    result = await session.execute(
+        select(Document)
+        .join(SearchSpace)
+        .filter(Document.id == document_id, SearchSpace.user_id == user.id)
+    )
+    document = result.scalars().first()  # ✅ Changed from scalar_one_or_none()
+    
+    if not document:
+        raise HTTPException(status_code=404, detail="Document not found")
+    
+    blocknote_document = data.get("blocknote_document")
+    if not blocknote_document:
+        raise HTTPException(status_code=400, detail="blocknote_document is required")
+    
+    # Update only blocknote_document and last_edited_at
+    document.blocknote_document = blocknote_document
+    document.last_edited_at = datetime.now(UTC)
+    
+    await session.commit()
+    await session.refresh(document)
+    
+    return {"status": "saved", "last_edited_at": document.last_edited_at.isoformat()}
+
+
+# did not implement reindexing (for now)
+# @router.post("/documents/{document_id}/finalize-edit")
+# async def finalize_edit(
+#     document_id: int,
+#     session: AsyncSession = Depends(get_async_session),
+#     user: User = Depends(current_active_user),
+# ):
+#     """
+#     Finalize document editing: convert BlockNote to markdown,
+#     update content (summary), and trigger reindexing.
+#     """
+#     result = await session.execute(
+#         select(Document)
+#         .join(SearchSpace)
+#         .filter(Document.id == document_id, SearchSpace.user_id == user.id)
+#     )
+#     document = result.scalars().first()
+    
+#     if not document:
+#         raise HTTPException(status_code=404, detail="Document not found")
+    
+#     if not document.blocknote_document:
+#         raise HTTPException(
+#             status_code=400,
+#             detail="Document has no BlockNote content to finalize"
+#         )
+    
+#     # 1. Convert BlockNote JSON → Markdown
+#     full_markdown = await convert_blocknote_to_markdown(document.blocknote_document)
+    
+#     if not full_markdown:
+#         raise HTTPException(
+#             status_code=500,
+#             detail="Failed to convert BlockNote document to markdown"
+#         )
+    
+#     # 2. Generate new summary from full markdown
+#     from app.services.llm_service import get_user_long_context_llm
+#     from app.utils.document_converters import generate_document_summary
+    
+#     user_llm = await get_user_long_context_llm(session, str(user.id), document.search_space_id)
+#     if not user_llm:
+#         raise HTTPException(
+#             status_code=500,
+#             detail="No LLM configured for summary generation"
+#         )
+    
+#     document_metadata = document.document_metadata or {}
+#     summary_content, summary_embedding = await generate_document_summary(
+#         full_markdown, user_llm, document_metadata
+#     )
+    
+#     # 3. Update document fields
+#     document.content = summary_content
+#     document.embedding = summary_embedding
+#     document.content_needs_reindexing = True  # Trigger chunk regeneration
+#     document.last_edited_at = datetime.now(UTC)
+    
+#     await session.commit()
+    
+#     return {
+#         "status": "finalized",
+#         "message": "Document saved. Summary and chunks will be regenerated in the background.",
+#         "content_needs_reindexing": True,
+#     }
+    

surfsense_backend/app/tasks/document_processors/extension_processor.py

@@ -144,6 +144,16 @@ async def add_extension_received_document(
 
         # Process chunks
         chunks = await create_document_chunks(content.pageContent)
+        
+        from app.utils.blocknote_converter import convert_markdown_to_blocknote
+        
+        # Convert markdown to BlockNote JSON
+        blocknote_json = await convert_markdown_to_blocknote(combined_document_string)
+        if not blocknote_json:
+            logging.warning(
+                f"Failed to convert extension document '{content.metadata.VisitedWebPageTitle}' "
+                f"to BlockNote JSON, document will not be editable"
+            )
 
         # Update or create document
         if existing_document:
@@ -154,6 +164,7 @@ async def add_extension_received_document(
             existing_document.embedding = summary_embedding
             existing_document.document_metadata = content.metadata.model_dump()
             existing_document.chunks = chunks
+            existing_document.blocknote_document = blocknote_json
 
             await session.commit()
             await session.refresh(existing_document)
@@ -170,6 +181,7 @@ async def add_extension_received_document(
                 chunks=chunks,
                 content_hash=content_hash,
                 unique_identifier_hash=unique_identifier_hash,
+                blocknote_document=blocknote_json,
             )
 
             session.add(document)

surfsense_backend/app/tasks/document_processors/file_processors.py

@@ -99,6 +99,14 @@ async def add_received_file_document_using_unstructured(
 
         # Process chunks
         chunks = await create_document_chunks(file_in_markdown)
+        
+        from app.utils.blocknote_converter import convert_markdown_to_blocknote
+        
+        # Convert markdown to BlockNote JSON
+        blocknote_json = await convert_markdown_to_blocknote(file_in_markdown)
+        if not blocknote_json:
+            logging.warning(f"Failed to convert {file_name} to BlockNote JSON, document will not be editable")
+            
 
         # Update or create document
         if existing_document:
@@ -112,6 +120,7 @@ async def add_received_file_document_using_unstructured(
                 "ETL_SERVICE": "UNSTRUCTURED",
             }
             existing_document.chunks = chunks
+            existing_document.blocknote_document = blocknote_json
 
             await session.commit()
             await session.refresh(existing_document)
@@ -131,6 +140,7 @@ async def add_received_file_document_using_unstructured(
                 chunks=chunks,
                 content_hash=content_hash,
                 unique_identifier_hash=unique_identifier_hash,
+                blocknote_document=blocknote_json,
             )
 
             session.add(document)
@@ -213,6 +223,14 @@ async def add_received_file_document_using_llamacloud(
 
         # Process chunks
         chunks = await create_document_chunks(file_in_markdown)
+        
+        from app.utils.blocknote_converter import convert_markdown_to_blocknote
+
+        # Convert markdown to BlockNote JSON
+        blocknote_json = await convert_markdown_to_blocknote(file_in_markdown)
+        if not blocknote_json:
+            logging.warning(f"Failed to convert {file_name} to BlockNote JSON, document will not be editable")
+            
 
         # Update or create document
         if existing_document:
@@ -226,6 +244,7 @@ async def add_received_file_document_using_llamacloud(
                 "ETL_SERVICE": "LLAMACLOUD",
             }
             existing_document.chunks = chunks
+            existing_document.blocknote_document = blocknote_json
 
             await session.commit()
             await session.refresh(existing_document)
@@ -245,6 +264,7 @@ async def add_received_file_document_using_llamacloud(
                 chunks=chunks,
                 content_hash=content_hash,
                 unique_identifier_hash=unique_identifier_hash,
+                blocknote_document=blocknote_json,
             )
 
             session.add(document)
@@ -352,6 +372,14 @@ async def add_received_file_document_using_docling(
 
         # Process chunks
         chunks = await create_document_chunks(file_in_markdown)
+        
+        from app.utils.blocknote_converter import convert_markdown_to_blocknote
+        
+        # Convert markdown to BlockNote JSON
+        blocknote_json = await convert_markdown_to_blocknote(file_in_markdown)
+        if not blocknote_json:
+            logging.warning(f"Failed to convert {file_name} to BlockNote JSON, document will not be editable")
+
 
         # Update or create document
         if existing_document:
@@ -365,6 +393,7 @@ async def add_received_file_document_using_docling(
                 "ETL_SERVICE": "DOCLING",
             }
             existing_document.chunks = chunks
+            existing_document.blocknote_document = blocknote_json
 
             await session.commit()
             await session.refresh(existing_document)
@@ -384,6 +413,7 @@ async def add_received_file_document_using_docling(
                 chunks=chunks,
                 content_hash=content_hash,
                 unique_identifier_hash=unique_identifier_hash,
+                blocknote_document=blocknote_json,
             )
 
         session.add(document)

surfsense_backend/app/tasks/document_processors/markdown_processor.py

@@ -109,6 +109,14 @@ async def add_received_markdown_file_document(
 
         # Process chunks
         chunks = await create_document_chunks(file_in_markdown)
+        
+        from app.utils.blocknote_converter import convert_markdown_to_blocknote
+        
+        # Convert to BlockNote JSON
+        blocknote_json = await convert_markdown_to_blocknote(file_in_markdown)
+        if not blocknote_json:
+            logging.warning(f"Failed to convert {file_name} to BlockNote JSON, document will not be editable")
+
 
         # Update or create document
         if existing_document:
@@ -121,6 +129,7 @@ async def add_received_markdown_file_document(
                 "FILE_NAME": file_name,
             }
             existing_document.chunks = chunks
+            existing_document.blocknote_document = blocknote_json
 
             await session.commit()
             await session.refresh(existing_document)
@@ -139,6 +148,7 @@ async def add_received_markdown_file_document(
                 chunks=chunks,
                 content_hash=content_hash,
                 unique_identifier_hash=unique_identifier_hash,
+                blocknote_document=blocknote_json,
             )
 
             session.add(document)

surfsense_backend/app/tasks/document_processors/url_crawler.py

@@ -247,6 +247,16 @@ async def add_crawled_url_document(
             f"Processing content chunks for URL: {url}",
             {"stage": "chunk_processing"},
         )
+        
+        from app.utils.blocknote_converter import convert_markdown_to_blocknote
+
+        # Convert markdown to BlockNote JSON
+        blocknote_json = await convert_markdown_to_blocknote(combined_document_string)
+        if not blocknote_json:
+            logging.warning(
+                f"Failed to convert crawled URL '{url}' to BlockNote JSON, "
+                "document will not be editable"
+            )
 
         chunks = await create_document_chunks(content_in_markdown)
 
@@ -267,6 +277,7 @@ async def add_crawled_url_document(
             existing_document.embedding = summary_embedding
             existing_document.document_metadata = url_crawled[0].metadata
             existing_document.chunks = chunks
+            existing_document.blocknote_document = blocknote_json
 
             document = existing_document
         else:
@@ -289,6 +300,7 @@ async def add_crawled_url_document(
                 chunks=chunks,
                 content_hash=content_hash,
                 unique_identifier_hash=unique_identifier_hash,
+                blocknote_document=blocknote_json,
             )
 
             session.add(document)

surfsense_backend/app/tasks/document_processors/youtube_processor.py

@@ -290,6 +290,16 @@ async def add_youtube_video_document(
             f"Processing content chunks for video: {video_data.get('title', 'YouTube Video')}",
             {"stage": "chunk_processing"},
         )
+        
+        from app.utils.blocknote_converter import convert_markdown_to_blocknote
+
+        # Convert transcript to BlockNote JSON
+        blocknote_json = await convert_markdown_to_blocknote(combined_document_string)
+        if not blocknote_json:
+            logging.warning(
+                f"Failed to convert YouTube video '{video_id}' to BlockNote JSON, "
+                "document will not be editable"
+            )
 
         chunks = await create_document_chunks(combined_document_string)
 
@@ -314,6 +324,7 @@ async def add_youtube_video_document(
                 "thumbnail": video_data.get("thumbnail_url", ""),
             }
             existing_document.chunks = chunks
+            existing_document.blocknote_document = blocknote_json
 
             await session.commit()
             await session.refresh(existing_document)
@@ -342,6 +353,7 @@ async def add_youtube_video_document(
                 search_space_id=search_space_id,
                 content_hash=content_hash,
                 unique_identifier_hash=unique_identifier_hash,
+                blocknote_document=blocknote_json,
             )
 
             session.add(document)

surfsense_backend/app/utils/blocknote_converter.py

@@ -0,0 +1,113 @@
+import logging
+from typing import Any
+
+import httpx
+
+from app.config import config
+
+logger = logging.getLogger(__name__)
+
+
+async def convert_markdown_to_blocknote(markdown: str) -> dict[str, Any] | None:
+    """
+    Convert markdown to BlockNote JSON via Next.js API.
+    
+    Args:
+        markdown: Markdown string to convert
+        
+    Returns:
+        BlockNote document as dict, or None if conversion fails
+    """
+    if not markdown or not markdown.strip():
+        logger.warning("Empty markdown provided for conversion")
+        return None
+    
+    if not markdown or len(markdown) < 10:
+        logger.warning("Markdown became too short after sanitization")
+        # Return a minimal BlockNote document
+        return [
+            {
+                "type": "paragraph",
+                "content": [
+                    {
+                        "type": "text",
+                        "text": "Document content could not be converted for editing.",
+                        "styles": {}
+                    }
+                ],
+                "children": []
+            }
+        ]
+    
+    async with httpx.AsyncClient() as client:
+        try:
+            response = await client.post(
+                f"{config.NEXT_FRONTEND_URL}/api/convert-to-blocknote",
+                json={"markdown": markdown},
+                timeout=30.0,
+            )
+            response.raise_for_status()
+            data = response.json()
+            blocknote_document = data.get("blocknote_document")
+            
+            if blocknote_document:
+                logger.info(f"Successfully converted markdown to BlockNote (original: {len(markdown)} chars, sanitized: {len(markdown)} chars)")
+                return blocknote_document
+            else:
+                logger.warning("Next.js API returned empty blocknote_document")
+                return None
+                
+        except httpx.TimeoutException:
+            logger.error("Timeout converting markdown to BlockNote after 30s")
+            return None
+        except httpx.HTTPStatusError as e:
+            logger.error(f"HTTP error converting markdown to BlockNote: {e.response.status_code} - {e.response.text}")
+            # Log first 1000 chars of problematic markdown for debugging
+            logger.debug(f"Problematic markdown sample: {markdown[:1000]}")
+            return None
+        except Exception as e:
+            logger.error(f"Failed to convert markdown to BlockNote: {e}", exc_info=True)
+            return None
+
+async def convert_blocknote_to_markdown(blocknote_document: dict[str, Any] | list[dict[str, Any]]) -> str | None:
+    """
+    Convert BlockNote JSON to markdown via Next.js API.
+    
+    Args:
+        blocknote_document: BlockNote document as dict or list of blocks
+        
+    Returns:
+        Markdown string, or None if conversion fails
+    """
+    if not blocknote_document:
+        logger.warning("Empty BlockNote document provided for conversion")
+        return None
+    
+    async with httpx.AsyncClient() as client:
+        try:
+            response = await client.post(
+                f"{config.NEXT_FRONTEND_URL}/api/convert-to-markdown",
+                json={"blocknote_document": blocknote_document},
+                timeout=30.0,
+            )
+            response.raise_for_status()
+            data = response.json()
+            markdown = data.get("markdown")
+            
+            if markdown:
+                logger.info(f"Successfully converted BlockNote to markdown ({len(markdown)} chars)")
+                return markdown
+            else:
+                logger.warning("Next.js API returned empty markdown")
+                return None
+                
+        except httpx.TimeoutException:
+            logger.error("Timeout converting BlockNote to markdown after 30s")
+            return None
+        except httpx.HTTPStatusError as e:
+            logger.error(f"HTTP error converting BlockNote to markdown: {e.response.status_code} - {e.response.text}")
+            return None
+        except Exception as e:
+            logger.error(f"Failed to convert BlockNote to markdown: {e}", exc_info=True)
+            return None
+