From 8f1533337cda6dc5f7698e59de18bbe1056c0e2c Mon Sep 17 00:00:00 2001
From: bsiggel <bsiggel@bitbylaw.com>
Date: Thu, 12 Mar 2026 22:35:48 +0000
Subject: [PATCH] feat: Enhance AI Knowledge sync process with full sync mode
 and attachment handling

---
 services/aiknowledge_sync_utils.py           | 25 +++++++++++++++++---
 steps/vmh/aiknowledge_full_sync_cron_step.py |  7 ++----
 steps/vmh/aiknowledge_sync_event_step.py     |  4 +++-
 3 files changed, 27 insertions(+), 9 deletions(-)

diff --git a/services/aiknowledge_sync_utils.py b/services/aiknowledge_sync_utils.py
index 7a7283e..c069a9c 100644
--- a/services/aiknowledge_sync_utils.py
+++ b/services/aiknowledge_sync_utils.py
@@ -227,7 +227,7 @@ class AIKnowledgeSync(BaseSyncUtils):
                     })
 
                 # Sync documents
-                await self._sync_knowledge_documents(knowledge_id, collection_id, ctx)
+                await self._sync_knowledge_documents(knowledge_id, collection_id, ctx, full_sync=full_sync)
 
             else:
                 ctx.logger.error(f"❌ Unknown activationStatus: {activation_status}")
@@ -327,14 +327,33 @@ class AIKnowledgeSync(BaseSyncUtils):
                 
                 ctx.logger.info(f"   🔄 Syncing: {reason}")
                 
+                # Get complete document entity with attachment info
+                doc_entity = await espocrm.get_entity('CDokumente', doc_id)
+                attachment_id = doc_entity.get('dokumentId')
+                
+                if not attachment_id:
+                    ctx.logger.error(f"   ❌ No attachment ID found for document {doc_id}")
+                    failed += 1
+                    continue
+                
+                # Get attachment details for MIME type
+                try:
+                    attachment = await espocrm.get_entity('Attachment', attachment_id)
+                    mime_type = attachment.get('type', 'application/octet-stream')
+                    file_size = attachment.get('size', 0)
+                except Exception as e:
+                    ctx.logger.warn(f"   ⚠️  Failed to get attachment details: {e}, using defaults")
+                    mime_type = 'application/octet-stream'
+                    file_size = 0
+                
+                ctx.logger.info(f"   📎 Attachment: {attachment_id} ({mime_type}, {file_size} bytes)")
+                
                 # Download document
-                attachment_id = doc.get('documentId')  # TODO: Get correct attachment ID from CDokumente
                 file_content = await espocrm.download_attachment(attachment_id)
                 ctx.logger.info(f"   📥 Downloaded {len(file_content)} bytes")
                 
                 # Upload to XAI
                 filename = doc_name
-                mime_type = 'application/octet-stream'  # TODO: Get from attachment
                 
                 xai_file_id = await xai.upload_file(file_content, filename, mime_type)
                 ctx.logger.info(f"   📤 Uploaded to XAI: {xai_file_id}")
diff --git a/steps/vmh/aiknowledge_full_sync_cron_step.py b/steps/vmh/aiknowledge_full_sync_cron_step.py
index 18450db..cbb1fe8 100644
--- a/steps/vmh/aiknowledge_full_sync_cron_step.py
+++ b/steps/vmh/aiknowledge_full_sync_cron_step.py
@@ -69,11 +69,8 @@ async def handler(input_data: None, ctx: FlowContext[Any]) -> None:
                 'topic': 'aiknowledge.sync',
                 'data': {
                     'knowledge_id': entity['id'],
-                    'source': 'daily_full_sync'
-                }
-            })
-            
-            ctx.logger.info(
+                        'source': 'daily_full_sync',
+                        'full_sync': True  # Enable Blake3 verification
                 f"📤 [{i}/{total}] Enqueued: {entity['name']} "
                 f"(syncStatus={entity.get('syncStatus')})"
             )
diff --git a/steps/vmh/aiknowledge_sync_event_step.py b/steps/vmh/aiknowledge_sync_event_step.py
index f3ca95b..d1df249 100644
--- a/steps/vmh/aiknowledge_sync_event_step.py
+++ b/steps/vmh/aiknowledge_sync_event_step.py
@@ -36,6 +36,7 @@ async def handler(event_data: Dict[str, Any], ctx: FlowContext[Any]) -> None:
     # Extract data
     knowledge_id = event_data.get('knowledge_id')
     source = event_data.get('source', 'unknown')
+    full_sync = event_data.get('full_sync', False)  # Blake3 verification mode
     
     if not knowledge_id:
         ctx.logger.error("❌ Missing knowledge_id in event data")
@@ -43,6 +44,7 @@ async def handler(event_data: Dict[str, Any], ctx: FlowContext[Any]) -> None:
     
     ctx.logger.info(f"📋 Knowledge ID: {knowledge_id}")
     ctx.logger.info(f"📋 Source: {source}")
+    ctx.logger.info(f"📋 Full Sync Mode: {full_sync}")
     ctx.logger.info("=" * 80)
     
     # Get Redis for locking
@@ -61,7 +63,7 @@ async def handler(event_data: Dict[str, Any], ctx: FlowContext[Any]) -> None:
     
     try:
         # Perform sync
-        await sync_utils.sync_knowledge_to_xai(knowledge_id, ctx)
+        await sync_utils.sync_knowledge_to_xai(knowledge_id, ctx, full_sync=full_sync)
         
         ctx.logger.info("=" * 80)
         ctx.logger.info("✅ AI KNOWLEDGE SYNC COMPLETED")