Update database resources for pipeline outputs

This commit is contained in:
Umang Mistry
2024-06-21 16:43:42 -05:00
parent 6718bcc573
commit 98acf3833c
3 changed files with 14 additions and 18 deletions
@@ -7,6 +7,7 @@ CREATE TABLE IF NOT EXISTS STG.PROMPT_CONFIG (
PROMPT VARCHAR,
FM_MODEL_ID VARCHAR,
GROUP_ID NUMERIC,
DEPENDENT_FIELDS VARCHAR,
FIELD_MAX_LENGTH NUMBER,
SAMPLE_VALUE VARCHAR,
CONTRACT_SECTION VARCHAR
@@ -1,15 +1,11 @@
-- CREATING THE CONFIG TABLE
CREATE TABLE IF NOT EXISTS STG.DOCZY_PIPELINE_RAW_OUTPUT (
DOCUMENT_ID VARCHAR(16777216) NOT NULL,
SF_DB_COL_NAME VARCHAR(255),
ACTUAL_VALUE_STORED VARCHAR(16777216),
CREATE TABLE IF NOT EXISTS STG.DOCZY_PIPELINE_RAW_OUTPUT_AC (
CONTRACT_NAME VARCHAR(16777216) NOT NULL,
FIELD_NAME VARCHAR(255),
RAW_VALUE VARCHAR(16777216),
NEW_EXTRACTED_VALUE VARCHAR(16777216),
SNIPPET VARCHAR(16777216),
ORIGINAL_PAGE_NUMBER VARCHAR(255),
NEW_PAGE_NUMBER VARCHAR(255),
RESULT VARCHAR(16777216),
BATCH_ID VARCHAR(255),
CREATED_TIME TIMESTAMP_NTZ(9),
MODIFIED_TIME TIMESTAMP_NTZ(9)
PAGE_NUMBER VARCHAR(255),
REVISED_PROMPT VARCHAR(16777216),
BATCH_ID VARCHAR(255)
);
@@ -1,4 +1,4 @@
CREATE OR REPLACE PROCEDURE STG.LOAD_DOCZY_PIPELINE_RAW_OUTPUTS(file_name VARCHAR)
CREATE OR REPLACE PROCEDURE STG.LOAD_DOCZY_PIPELINE_RAW_OUTPUTS(file_name VARCHAR, bucket_name VARCHAR)
RETURNS STRING
LANGUAGE SQL
EXECUTE AS CALLER
@@ -14,26 +14,25 @@ BEGIN
-- Create or replace stage with dynamic file name
-- TODO: Update for UAT
-- s3://doczyai-use2-u-infra-s3-raw-data-ingestion/
EXECUTE IMMEDIATE 'CREATE OR REPLACE STAGE STG.DOCZY_PIPELINE_RAW_OUTPUT_STAGE
STORAGE_INTEGRATION = dev_bucket_integration
URL = ''s3://doczy-dev-infra-raw-data-ingestion/doczy_pipeline_output/' || :file_name || '''
URL = ''s3://doczyai-use2-u-infra-s3-raw-data-ingestion/doczy_pipeline_output/' || :file_name || '''
FILE_FORMAT = (FORMAT_NAME = ''STG.CSV_HEADER'');';
call stg.log_audit(:procedure_name, 'Section 1', 99, 'END');
call stg.log_audit(:procedure_name, 'Section 2', 99, 'START');
COPY INTO STG.DOCZY_PIPELINE_RAW_OUTPUT FROM (
COPY INTO STG.DOCZY_PIPELINE_RAW_OUTPUT_AC FROM (
SELECT
NULLIF(TRIM($1),'') AS CONTRACT_NAME,
NULLIF(TRIM($2),'') AS SF_DB_COL_NAME,
NULLIF(TRIM($3),'') AS ACTUAL_VALUE_STORED,
NULLIF(TRIM($2),'') AS FIELD_NAME,
NULLIF(TRIM($4),'') AS RAW_VALUE,
NULLIF(TRIM($5),'') AS NEW_EXTRACTED_VALUE,
NULLIF(TRIM($6),'') AS SNIPPET,
NULLIF(TRIM($7),'') AS ORIGINAL_PAGE_NUMBER,
NULLIF(TRIM($8),'') AS NEW_PAGE_NUMBER,
NULLIF(TRIM($9),'') AS RESULT
NULLIF(TRIM($8),'') AS PAGE_NUMBER,
NULLIF(TRIM($9),'') AS REVISED_PROMPT
NULLIF(TRIM($10),'') AS BATCH_ID
FROM @STG.DOCZY_PIPELINE_RAW_OUTPUT_STAGE