Upload files to "/"

2026-05-08 13:13:13 +00:00
parent 575625cf01
commit 9eac9227b2
4 changed files with 732 additions and 0 deletions
--- a/pkg_26ai_auto_load.sql
+++ b/pkg_26ai_auto_load.sql
@@ -0,0 +1,95 @@
+create or replace package pkg_26ai_auto_load
+as
+
+/*
+
+-- Teste
+
+    -- gerar arquivos a serem vetorizados
+    begin
+    PKG_26AI_AUTO_LOAD.prc_refresh_files(p_mimetype => 'PDF',p_embedding_name=>'COHERE');
+    PKG_26AI_AUTO_LOAD.prc_refresh_files(p_mimetype => 'JPG',p_embedding_name=>'COHERE');
+    PKG_26AI_AUTO_LOAD.prc_refresh_files(p_mimetype => 'CLOB',p_embedding_name=>'COHERE');
+    end; 
+
+    -- liste de dados a serem vetorizados. Status incial: NULL
+    select * from tb_26ai_auto_load
+
+    -- reserva de dados para associar aos jobs que executarao, marcado status como R
+    -- CUIDADO: nao preciso executar manulamnete opois faz parte da rotina de embedding AUTOMATICAMENTE
+    declare
+      v_name SYS.ODCIVARCHAR2LIST;
+    begin
+    PKG_26AI_AUTO_LOAD.prc_reserve_files (
+    	  p_limit    => 10,	                           
+          p_worker_id => 1 ,
+          p_oci_cred  => 'OCI_CRED',
+          p_bucket  => '',
+          p_docs  =>v_name
+    	)  ;
+    end; 
+
+    -- liste de dados a serem vetorizados. Status reservado por worker id em R
+    select * from tb_26ai_auto_load
+
+    -- Embedding 
+    -- CUIDADO: ja faz o reserve de acordo com p_worker_id e p_limit
+    declare
+      v_name SYS.ODCIVARCHAR2LIST;
+    begin
+    PKG_26AI_AUTO_LOAD.proc_embed_files (
+    	  p_limit    => 10,	                           
+          p_worker_id => 1 ,
+          p_stop_process_list  => 'N'
+    	)  ;
+    end; 
+
+    -- liste status
+    select * from tb_26ai_auto_load  
+
+    -- erro
+    select * from tb_26ai_auto_load_debug
+
+    -- vector
+    select * from tb_26ai_auto_load_vector
+
+    -- em caso de erro nos TESTES
+    begin
+    update tb_26ai_auto_load set ERROR_MSG= null, status = NULL, WORKER_ID = null where status is not null ;
+    delete tb_26ai_auto_load_debug;
+    delete tb_26ai_auto_load_vector;
+    end;
+
+
+*/
+   c_embedding_name CONSTANT VARCHAR2(10) := 'ONNX-E5'; -- requer ajuste do JSON usado nas rotinas de embedding no BODY
+
+   PROCEDURE  prc_refresh_files(p_oci_cred  IN VARCHAR2 default 'OCI_CRED',
+                                p_bucket in varchar2 default null,
+                                p_mimetype in varchar2 default 'PDF',
+                                 p_embedding_name  in varchar2 default 'COHERE');
+									 
+   PROCEDURE  prc_reserve_files (
+	  p_limit     IN NUMBER,	                           
+      p_worker_id IN VARCHAR2,
+      p_oci_cred  IN VARCHAR2 default 'OCI_CRED',
+      p_bucket in varchar2 ,
+      p_docs  OUT SYS.ODCIVARCHAR2LIST,
+      p_mimetype in varchar2 default 'PDF',
+      p_embedding_name  in varchar2 default 'COHERE'	) ;
+	  
+	PROCEDURE proc_embed_files(p_limit in number default 10,
+	                           p_worker_id in number,
+							   p_stop_process_list in varchar2 default 'N',
+                               p_mimetype in varchar2 default 'PDF',
+                               p_embedding_name  in varchar2 default 'COHERE');
+
+	PROCEDURE proc_remove_jobs;
+
+	PROCEDURE proc_add_jobs(p_limit in number, 
+                            p_total_jobs in number,
+                            p_mimetype in varchar2 default 'PDF',
+                            p_embedding_name  in varchar2 default 'COHERE');
+  
+end;
+/