o Ղi/R@sUdZddlmZmZmZmZddlZddlZddlm Z ddl m Z ddl m Z eeZGdddZdaeeed <d efd d ZdS) zP Graph Builder Service Orchestrates knowledge graph construction from documents )DictAnyOptionalListN) Neo4jClient)EntityExtractor)RelationExtractorc@seZdZdZdededefddZ       d'ded e d e d e e d ed ede ede e de e de e de e ede ede e effddZ  d(ded e d e e d ed ede e de e fddZde ded ed edef ddZdede d ed efddZdeded ed efddZded e e e efd ed ede e eff d!d"Zded e e e efd ed ede e eff d#d"Z d)d ed ed$e de e effd%d&ZdS)* GraphBuilderzb Knowledge graph builder service Coordinates entity extraction and graph construction neo4j_cliententity_extractorrelation_extractorcCs||_||_||_dSN)r r r )selfr r r rH/lsinfo/ai/hellotax_ai/base_platform/app/services/graph/graph_builder.py__init__s zGraphBuilder.__init__N document_idtitlecontentsummary tenant_idkb_id category_idtag_ids doc_number doc_statussupersedes_doc_ids parent_doc_idreturnc Csztd|d||j|||||| | d|jj||| d}|dg}|dg}|||||}|j||||||j ||||j |||||rZ| |||||rd| ||||| rr| D] }|j ||qh| rd}|j j||| ||dd td |d|d t|d d ||t|dWSty}ztd|d|d|t|dWYd}~Sd}~ww)a Build knowledge graph for a single document Args: document_id: Document ID title: Document title content: Document content summary: Document summary tenant_id: Tenant ID kb_id: Knowledge base ID category_id: Optional category ID tag_ids: Optional list of tag IDs Returns: Build result with entity and relation counts zBuilding graph for document : )rr)rr db_sessionentities relationsa; MATCH (parent:Document {id: $parent_id, tenant_id: $tenant_id, kb_id: $kb_id}) MATCH (att:Document {id: $doc_id, tenant_id: $tenant_id, kb_id: $kb_id}) MERGE (parent)-[:HAS_ATTACHMENT]->(att) MERGE (att)-[:IS_ATTACHMENT_OF]->(parent) )doc_id parent_idrr parametersz&Graph built successfully for document z entities, z relationsT)successr entity_countrelation_countz#Failed to build graph for document Fr'rerrorN)loggerinfo_create_document_noder extract_entitiesget_create_entity_nodesr create_entity_relationscreate_document_relationsinfer_document_references _link_tags_link_categoryr create_supersedes_relation execute_writelen Exceptionr+str)rrrrrrrrrr rrrrextraction_resultr!r"r( target_idqueryerrrbuild_document_graph sx      z!GraphBuilder.build_document_graphc CsD|dur|nd}d}|jj||||pd||pd|pdd|ddS)z'Create or update document node in Neo4jNra MERGE (d:Document {id: $doc_id, tenant_id: $tenant_id, kb_id: $kb_id}) SET d.title = $title, d.summary = $summary, d.doc_number = $doc_number, d.doc_status = $doc_status, d.updated_at = datetime() effective)r#rrrrrr&rr r8) rrrrrrrrr>rrrr.s   z"GraphBuilder._create_document_noder!cCs.|sdSd}|jj||||d|dt|S)zs Create entity nodes and link to document Returns: Number of entities created ra UNWIND $entities AS entity MERGE (e:Entity { name: entity.name, tenant_id: $tenant_id, kb_id: $kb_id }) SET e.type = entity.type, e.description = entity.description, e.confidence = entity.salience WITH e, entity MATCH (d:Document {id: $doc_id, tenant_id: $tenant_id, kb_id: $kb_id}) MERGE (d)-[r:CONTAINS]->(e) SET r.salience = entity.salience, r.frequency = 1 )r!r#rrC)r r8r9)rr!rrrr>rrrr1s  z!GraphBuilder._create_entity_nodescC"d}|jj|||||dddS)zLink document to tagsz MATCH (d:Document {id: $doc_id, tenant_id: $tenant_id, kb_id: $kb_id}) UNWIND $tag_ids AS tag_id MERGE (t:Tag {id: tag_id, tenant_id: $tenant_id, kb_id: $kb_id}) MERGE (d)-[:HAS_TAG]->(t) )r#rrrr%NrD)rrrrrr>rrrr5s zGraphBuilder._link_tagscCrE)zLink document to categoryz MATCH (d:Document {id: $doc_id, tenant_id: $tenant_id, kb_id: $kb_id}) MERGE (c:Category {id: $cat_id, tenant_id: $tenant_id, kb_id: $kb_id}) MERGE (d)-[:IN_CATEGORY]->(c) )r#cat_idrrr%NrD)rrrrrr>rrrr6s zGraphBuilder._link_categorychunksc Cztd|dt|dd}d}|D]L}|d}|s+td|ddq|jj|||d d |d d |dd|d d |d|||did|didd |d7}q|D]^}|d}|dg} |ru| svqd| D]I} t| tsqx| d} | sqxz|jj || | d| dd||d|d7}Wqxt y} ztd|d| d| WYd} ~ qxd} ~ wwqdtd|d|d|dd |||d!WSt y} zt d"|d| d |t | d#WYd} ~ Sd} ~ ww$aS Build chunk-level knowledge graph with REFERENCES relationships Args: document_id: Document ID chunks: List of chunk dictionaries with metadata tenant_id: Tenant ID kb_id: Knowledge base ID Returns: Build result with chunk and reference counts z"Building chunk graph for document rz chunksrchunk_idz"Chunk missing chunk_id, skipping: chunk_indexunknowntextrA chunk_level is_parentFparent_chunk_idmetadatadoc_typer) rJr chunk_textrNrKrOrPrrrRr referencestarget_doc_numberarticle_number confidenceg?)source_chunk_idrVtarget_articlerXrrz Failed to create reference from z to Nz,Chunk graph built successfully for document z chunks, z referencesT)r'r chunk_countreference_countz)Failed to build chunk graph for document r* r,r-r9r0warningr create_chunk_node isinstancedictcreate_chunk_referencer:r+r; rrrGrrr[r\chunkrJrUrefrVr?rrrbuild_chunk_graph            (zGraphBuilder.build_chunk_graphc CrHrIr]rcrrrrfyrg documentsc Csddlm}m}ddlm}td||}zcz(|||j |k } | r3d| _ | d} d} d} |D]} z~|j | d| d| d| d ||| d | d g|d }|d r| d7} | |d7} |||j | dk }|rd|_ |d|_| n| d7} |||j | dk }|rd|_ | td| dd|d d|ddWq;ty}z| d7} td| dd|WYd}~q;d}~wwz |j||Wnty}ztd|WYd}~nd}~ww|||j |k } | r%| dkrdnd| _ | | _| td|d| d| d| d d|| | | dWW|Sty}}z,td|d|z|||j |k } | rrd| _ | WWYd}~ww|w)a* Rebuild entire knowledge base graph Args: kb_id: Knowledge base ID tenant_id: Tenant ID documents: List of documents to process db_session: Database session for updating status Returns: Build statistics r) KnowledgeBaseKnowledgeDocument) SessionLocalz$Rebuilding graph for knowledge base buildingidrrrrr) rrrrrrrrr r'rTr( completedfailedzProcessed document rz , entities: zFailed to process document Nz*Failed to create co-occurrence relations: zGraph rebuild completed for kb z success, z failed, z total entitiesT)r'r success_count failed_counttotal_entitieszGraph rebuild failed for kb )app.models.knowledge_baserirjapp.db.sessionrkr,r-r>filterrmfirst graph_statuscommitr@r0r(r:r+r create_co_occurrence_relationsclose)rrrrhr rirjrkdbkbrprqrrdocresultdoc_objr?rrrrebuild_knowledge_base_graphs         0&   z)GraphBuilder.rebuild_knowledge_base_graph)NNNNNNN)NNr )__name__ __module__ __qualname____doc__rrrrintr;rlistrrrr@r.r1r5r6rfrrrrrr s       } $ .    `  e r graph_builderrcCs>tdurddlm}ddlm}ddlm}t|||atS)z$Get or create graph builder instanceNr)r )r )get_relation_extractor)rapp.services.graph.neo4j_clientr #app.services.graph.entity_extractorr %app.services.graph.relation_extractorrr )r r rrrrget_graph_builderWs   r)rtypingrrrrlogginguuidrrrrrr getLoggerrr,r r__annotations__rrrrrs    H