Feat/support parent child chunk (#12092)

api/services/entities/knowledge_entities/knowledge_entities.py

@@ -1,4 +1,5 @@
-from typing import Optional
+from enum import Enum
+from typing import Literal, Optional
 
 from pydantic import BaseModel
 
@@ -8,3 +9,112 @@ class SegmentUpdateEntity(BaseModel):
     answer: Optional[str] = None
     keywords: Optional[list[str]] = None
     enabled: Optional[bool] = None
+
+
+class ParentMode(str, Enum):
+    FULL_DOC = "full-doc"
+    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[NotionIcon] = None
+    type: str
+
+
+class NotionInfo(BaseModel):
+    workspace_id: str
+    pages: list[NotionPage]
+
+
+class WebsiteInfo(BaseModel):
+    provider: str
+    job_id: str
+    urls: list[str]
+    only_main_content: bool = True
+
+
+class FileInfo(BaseModel):
+    file_ids: list[str]
+
+
+class InfoList(BaseModel):
+    data_source_type: Literal["upload_file", "notion_import", "website_crawl"]
+    notion_info_list: Optional[list[NotionInfo]] = None
+    file_info_list: Optional[FileInfo] = None
+    website_info_list: Optional[WebsiteInfo] = None
+
+
+class DataSource(BaseModel):
+    info_list: InfoList
+
+
+class PreProcessingRule(BaseModel):
+    id: str
+    enabled: bool
+
+
+class Segmentation(BaseModel):
+    separator: str = "\n"
+    max_tokens: int
+    chunk_overlap: int = 0
+
+
+class Rule(BaseModel):
+    pre_processing_rules: Optional[list[PreProcessingRule]] = None
+    segmentation: Optional[Segmentation] = None
+    parent_mode: Optional[Literal["full-doc", "paragraph"]] = None
+    subchunk_segmentation: Optional[Segmentation] = None
+
+
+class ProcessRule(BaseModel):
+    mode: Literal["automatic", "custom", "hierarchical"]
+    rules: Optional[Rule] = None
+
+
+class RerankingModel(BaseModel):
+    reranking_provider_name: Optional[str] = None
+    reranking_model_name: Optional[str] = None
+
+
+class RetrievalModel(BaseModel):
+    search_method: Literal["hybrid_search", "semantic_search", "full_text_search"]
+    reranking_enable: bool
+    reranking_model: Optional[RerankingModel] = None
+    top_k: int
+    score_threshold_enabled: bool
+    score_threshold: Optional[float] = None
+
+
+class KnowledgeConfig(BaseModel):
+    original_document_id: Optional[str] = None
+    duplicate: bool = True
+    indexing_technique: Literal["high_quality", "economy"]
+    data_source: Optional[DataSource] = None
+    process_rule: Optional[ProcessRule] = None
+    retrieval_model: Optional[RetrievalModel] = None
+    doc_form: str = "text_model"
+    doc_language: str = "English"
+    embedding_model: Optional[str] = None
+    embedding_model_provider: Optional[str] = None
+    name: Optional[str] = None
+
+
+class SegmentUpdateArgs(BaseModel):
+    content: Optional[str] = None
+    answer: Optional[str] = None
+    keywords: Optional[list[str]] = None
+    regenerate_child_chunks: bool = False
+    enabled: Optional[bool] = None
+
+
+class ChildChunkUpdateArgs(BaseModel):
+    id: Optional[str] = None
+    content: str