From a250335485676ec9fe316db83fe88d4109bb72a2 Mon Sep 17 00:00:00 2001
From: jyong <718720800@qq.com>
Date: Tue, 24 Dec 2024 15:26:10 +0800
Subject: [PATCH 1/2] notion and website import fix

---
 .../entities/knowledge_entities/knowledge_entities.py     | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/api/services/entities/knowledge_entities/knowledge_entities.py b/api/services/entities/knowledge_entities/knowledge_entities.py
index c3bf7c0a7ee796..8d6a246b6428d0 100644
--- a/api/services/entities/knowledge_entities/knowledge_entities.py
+++ b/api/services/entities/knowledge_entities/knowledge_entities.py
@@ -16,10 +16,16 @@ class ParentMode(str, Enum):
     PARAGRAPH = "paragraph"
 
 
+class NotionIcon(BaseModel):
+    type: str
+    url: Optional[str] = None
+    emoji: Optional[str] = None
+
+
 class NotionPage(BaseModel):
     page_id: str
     page_name: str
-    page_icon: Optional[str]
+    page_icon: Optional[NotionIcon] = None
     type: str
 
 

From 98cd57f1036ad14e91d75b5efad65150a8c3eee0 Mon Sep 17 00:00:00 2001
From: jyong <718720800@qq.com>
Date: Tue, 24 Dec 2024 15:30:39 +0800
Subject: [PATCH 2/2] notion and website import fix

---
 api/services/dataset_service.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/api/services/dataset_service.py b/api/services/dataset_service.py
index 7f2a82425667e7..6ca1ed9225c1dd 100644
--- a/api/services/dataset_service.py
+++ b/api/services/dataset_service.py
@@ -737,10 +737,10 @@ def save_document_with_dataset_id(
                 elif knowledge_config.data_source.info_list.data_source_type == "notion_import":
                     notion_info_list = knowledge_config.data_source.info_list.notion_info_list
                     for notion_info in notion_info_list:
-                        count = count + len(notion_info["pages"])
+                        count = count + len(notion_info.pages)
                 elif knowledge_config.data_source.info_list.data_source_type == "website_crawl":
                     website_info = knowledge_config.data_source.info_list.website_info_list
-                    count = len(website_info["urls"])
+                    count = len(website_info.urls)
                 batch_upload_limit = int(dify_config.BATCH_UPLOAD_LIMIT)
                 if count > batch_upload_limit:
                     raise ValueError(f"You have reached the batch upload limit of {batch_upload_limit}.")
@@ -898,7 +898,7 @@ def save_document_with_dataset_id(
                                 data_source_info = {
                                     "notion_workspace_id": workspace_id,
                                     "notion_page_id": page.page_id,
-                                    "notion_page_icon": page.page_icon,
+                                    "notion_page_icon": page.page_icon.model_dump(),
                                     "type": page.type,
                                 }
                                 document = DocumentService.build_document(
@@ -1141,7 +1141,7 @@ def save_document_without_dataset_id(tenant_id: str, knowledge_config: Knowledge
             elif knowledge_config.data_source.info_list.data_source_type == "notion_import":
                 notion_info_list = knowledge_config.data_source.info_list.notion_info_list
                 for notion_info in notion_info_list:
-                    count = count + len(notion_info["pages"])
+                    count = count + len(notion_info.pages)
             elif knowledge_config.data_source.info_list.data_source_type == "website_crawl":
                 website_info = knowledge_config.data_source.info_list.website_info_list
                 count = len(website_info.urls)