6c8b22eac9
- Created new package CT_MRDS.DATA_EXPORTER (v2.6.3) for comprehensive data export capabilities, including CSV and Parquet formats with OCI integration. - Implemented version history tracking and enhanced error handling for export processes. - Updated installation scripts to include pRegisterExport parameter for registering exports in A_WORKFLOW_HISTORY and A_SOURCE_FILE_RECEIVED.
227 lines
13 KiB
Plaintext
227 lines
13 KiB
Plaintext
create or replace PACKAGE CT_MRDS.DATA_EXPORTER
|
|
AUTHID CURRENT_USER
|
|
AS
|
|
/**
|
|
* Data Export Package: Provides comprehensive data export capabilities to various formats (CSV, Parquet)
|
|
* with support for cloud storage integration via Oracle Cloud Infrastructure (OCI).
|
|
* The structure of comment is used by GET_PACKAGE_DOCUMENTATION function
|
|
* which returns documentation text for confluence page (to Copy-Paste it).
|
|
**/
|
|
|
|
-- Package Version Information (Semantic Versioning: MAJOR.MINOR.PATCH)
|
|
PACKAGE_VERSION CONSTANT VARCHAR2(10) := '2.7.0';
|
|
PACKAGE_BUILD_DATE CONSTANT VARCHAR2(20) := '2026-02-09 20:00:00';
|
|
PACKAGE_AUTHOR CONSTANT VARCHAR2(100) := 'Grzegorz Michalski';
|
|
|
|
cgBL CONSTANT VARCHAR2(2) := CHR(13)||CHR(10);
|
|
|
|
-- Version History (Latest changes first)
|
|
VERSION_HISTORY CONSTANT VARCHAR2(4000) :=
|
|
'v2.7.0 (2026-02-09): NEW FEATURE - Added pRegisterExport parameter to EXPORT_TABLE_DATA_TO_CSV_BY_DATE. When TRUE, successfully exported files are registered in A_WORKFLOW_HISTORY (one record per YEAR/MONTH) and A_SOURCE_FILE_RECEIVED tables for tracking and audit purposes.' || cgBL ||
|
|
'v2.6.3 (2026-01-28): COMPILATION FIX - Resolved ORA-00904 error in EXPORT_PARTITION_PARALLEL. SQLERRM and DBMS_UTILITY.FORMAT_ERROR_BACKTRACE cannot be used directly in SQL UPDATE statements. Now properly assigned to vgMsgTmp variable before UPDATE.' || cgBL ||
|
|
'v2.6.2 (2026-01-28): CRITICAL FIX - Race condition when multiple exports run simultaneously. Changed DELETE to filter by age (>24h) instead of deleting all COMPLETED chunks. Prevents concurrent sessions from deleting each other chunks. Session-safe cleanup with TASK_NAME filtering. Enables true parallel execution of multiple export jobs.' || cgBL ||
|
|
'v2.6.1 (2026-01-28): Added DELETE_FAILED_EXPORT_FILE procedure to clean up partial/corrupted files before retry. When partition fails mid-export, partial file is deleted before retry to prevent Oracle from creating _1 suffixed duplicates. Ensures clean retry without orphaned files in OCI bucket.' || cgBL ||
|
|
'v2.6.0 (2026-01-28): CRITICAL FIX - Added STATUS tracking to A_PARALLEL_EXPORT_CHUNKS table to prevent data duplication on retry. System now restarts ONLY failed partitions instead of re-exporting all data. Added ERROR_MESSAGE and EXPORT_TIMESTAMP columns for better error handling and monitoring. Prevents duplicate file creation when parallel tasks fail (e.g., 22 partitions with 16 threads, 3 failures no longer duplicates 19 successful exports).' || cgBL ||
|
|
'v2.5.0 (2026-01-26): Added recorddelimiter parameter with CRLF (CHR(13)||CHR(10)) for CSV exports to ensure Windows-compatible line endings. Improves cross-platform compatibility when CSV files are opened in Windows applications (Notepad, Excel).' || cgBL ||
|
|
'v2.4.0 (2026-01-11): Added pTemplateTableName parameter for per-column date format configuration. Implements dynamic query building with TO_CHAR for each date/timestamp column using FILE_MANAGER.GET_DATE_FORMAT. Supports 3-tier hierarchy: column-specific, template DEFAULT, global fallback. Eliminates single dateformat limitation of DBMS_CLOUD.EXPORT_DATA.' || cgBL ||
|
|
'v2.3.0 (2025-12-20): Added parallel partition processing using DBMS_PARALLEL_EXECUTE. New pParallelDegree parameter (1-16, default 1) for EXPORT_TABLE_DATA_BY_DATE and EXPORT_TABLE_DATA_TO_CSV_BY_DATE procedures. Each year/month partition processed in separate thread for improved performance.' || cgBL ||
|
|
'v2.2.0 (2025-12-19): DRY refactoring - extracted shared helper functions (sanitizeFilename, VALIDATE_TABLE_AND_COLUMNS, GET_PARTITIONS, EXPORT_SINGLE_PARTITION worker procedure). Reduced code duplication by ~400 lines. Prepared architecture for v2.3.0 parallel processing.' || cgBL ||
|
|
'v2.1.1 (2025-12-04): Fixed JOIN column reference A_WORKFLOW_HISTORY_KEY -> A_ETL_LOAD_SET_KEY, added consistent column mapping and dynamic column list to EXPORT_TABLE_DATA procedure, enhanced DEBUG logging for all export operations' || cgBL ||
|
|
'v2.1.0 (2025-10-22): Added version tracking and PARTITION_YEAR/PARTITION_MONTH support' || cgBL ||
|
|
'v2.0.0 (2025-10-01): Separated export functionality from FILE_MANAGER package';
|
|
|
|
vgMsgTmp VARCHAR2(32000);
|
|
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
-- TYPE DEFINITIONS FOR PARTITION HANDLING
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
|
|
/**
|
|
* Record type for year/month partition information
|
|
**/
|
|
TYPE partition_rec IS RECORD (
|
|
year VARCHAR2(4),
|
|
month VARCHAR2(2)
|
|
);
|
|
|
|
/**
|
|
* Table type for collection of partition records
|
|
**/
|
|
TYPE partition_tab IS TABLE OF partition_rec;
|
|
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
-- INTERNAL PARALLEL PROCESSING CALLBACK
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
|
|
/**
|
|
* @name EXPORT_PARTITION_PARALLEL
|
|
* @desc Internal callback procedure for DBMS_PARALLEL_EXECUTE.
|
|
* Processes single partition (year/month) chunk in parallel task.
|
|
* Called by DBMS_PARALLEL_EXECUTE framework for each chunk.
|
|
* This procedure is PUBLIC because DBMS_PARALLEL_EXECUTE requires it,
|
|
* but should NOT be called directly by external code.
|
|
* @param pStartId - Chunk start ID (CHUNK_ID from A_PARALLEL_EXPORT_CHUNKS table)
|
|
* @param pEndId - Chunk end ID (same as pStartId for single-row chunks)
|
|
**/
|
|
PROCEDURE EXPORT_PARTITION_PARALLEL (
|
|
pStartId IN NUMBER,
|
|
pEndId IN NUMBER
|
|
);
|
|
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
-- MAIN EXPORT PROCEDURES
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
|
|
/**
|
|
* @name EXPORT_TABLE_DATA
|
|
* @desc Wrapper procedure for DBMS_CLOUD.EXPORT_DATA.
|
|
* Exports data into CSV file on OCI infrustructure.
|
|
* pBucketArea parameter accepts: 'INBOX', 'ODS', 'DATA', 'ARCHIVE'
|
|
* @example
|
|
* begin
|
|
* DATA_EXPORTER.EXPORT_TABLE_DATA(
|
|
* pSchemaName => 'CT_MRDS',
|
|
* pTableName => 'MY_TABLE',
|
|
* pKeyColumnName => 'A_ETL_LOAD_SET_KEY_FK',
|
|
* pBucketArea => 'DATA',
|
|
* pFolderName => 'csv_exports'
|
|
* );
|
|
* end;
|
|
**/
|
|
PROCEDURE EXPORT_TABLE_DATA (
|
|
pSchemaName IN VARCHAR2,
|
|
pTableName IN VARCHAR2,
|
|
pKeyColumnName IN VARCHAR2,
|
|
pBucketArea IN VARCHAR2,
|
|
pFolderName IN VARCHAR2,
|
|
pCredentialName IN VARCHAR2 default ENV_MANAGER.gvCredentialName
|
|
);
|
|
|
|
|
|
|
|
/**
|
|
* @name EXPORT_TABLE_DATA_BY_DATE
|
|
* @desc Wrapper procedure for DBMS_CLOUD.EXPORT_DATA.
|
|
* Exports data into PARQUET files on OCI infrustructure.
|
|
* Each YEAR_MONTH pair goes to seperate file (implicit partitioning).
|
|
* Allows specifying custom column list or uses T.* if pColumnList is NULL.
|
|
* Validates that all columns in pColumnList exist in the target table.
|
|
* Automatically adds 'T.' prefix to column names in pColumnList.
|
|
* Supports parallel partition processing via pParallelDegree parameter (default 1, range 1-16).
|
|
* pBucketArea parameter accepts: 'INBOX', 'ODS', 'DATA', 'ARCHIVE'
|
|
* @example
|
|
* begin
|
|
* DATA_EXPORTER.EXPORT_TABLE_DATA_BY_DATE(
|
|
* pSchemaName => 'CT_MRDS',
|
|
* pTableName => 'MY_TABLE',
|
|
* pKeyColumnName => 'A_ETL_LOAD_SET_KEY_FK',
|
|
* pBucketArea => 'DATA',
|
|
* pFolderName => 'parquet_exports',
|
|
* pColumnList => 'COLUMN1, COLUMN2, COLUMN3', -- Optional
|
|
* pMinDate => DATE '2024-01-01',
|
|
* pMaxDate => SYSDATE,
|
|
* pParallelDegree => 8 -- Optional, default 1, range 1-16
|
|
* );
|
|
* end;
|
|
**/
|
|
PROCEDURE EXPORT_TABLE_DATA_BY_DATE (
|
|
pSchemaName IN VARCHAR2,
|
|
pTableName IN VARCHAR2,
|
|
pKeyColumnName IN VARCHAR2,
|
|
pBucketArea IN VARCHAR2,
|
|
pFolderName IN VARCHAR2,
|
|
pColumnList IN VARCHAR2 default NULL,
|
|
pMinDate IN DATE default DATE '1900-01-01',
|
|
pMaxDate IN DATE default SYSDATE,
|
|
pParallelDegree IN NUMBER default 1,
|
|
pTemplateTableName IN VARCHAR2 default NULL,
|
|
pCredentialName IN VARCHAR2 default ENV_MANAGER.gvCredentialName
|
|
);
|
|
|
|
|
|
|
|
/**
|
|
* @name EXPORT_TABLE_DATA_TO_CSV_BY_DATE
|
|
* @desc Exports data to separate CSV files partitioned by year and month.
|
|
* Creates one CSV file for each year/month combination found in the data.
|
|
* Uses the same date filtering mechanism with CT_ODS.A_LOAD_HISTORY as EXPORT_TABLE_DATA_BY_DATE,
|
|
* but exports to CSV format instead of Parquet.
|
|
* Supports parallel partition processing via pParallelDegree parameter (1-16).
|
|
* File naming pattern: {pFileName}_YYYYMM.csv or {TABLENAME}_YYYYMM.csv (if pFileName is NULL)
|
|
* When pRegisterExport=TRUE, successfully exported files are registered in:
|
|
* - CT_MRDS.A_WORKFLOW_HISTORY (one record per YEAR/MONTH with export timestamp)
|
|
* - CT_MRDS.A_SOURCE_FILE_RECEIVED (tracks file location and partition info)
|
|
* @example
|
|
* begin
|
|
* -- With custom filename
|
|
* DATA_EXPORTER.EXPORT_TABLE_DATA_TO_CSV_BY_DATE(
|
|
* pSchemaName => 'CT_MRDS',
|
|
* pTableName => 'MY_TABLE',
|
|
* pKeyColumnName => 'A_ETL_LOAD_SET_KEY_FK',
|
|
* pBucketArea => 'DATA',
|
|
* pFolderName => 'exports',
|
|
* pFileName => 'my_export.csv',
|
|
* pMinDate => DATE '2024-01-01',
|
|
* pMaxDate => SYSDATE,
|
|
* pParallelDegree => 8, -- Optional, default 1, range 1-16
|
|
* pRegisterExport => TRUE -- Optional, default FALSE, registers to A_WORKFLOW_HISTORY and A_SOURCE_FILE_RECEIVED
|
|
* );
|
|
*
|
|
* -- With auto-generated filename (based on table name only)
|
|
* DATA_EXPORTER.EXPORT_TABLE_DATA_TO_CSV_BY_DATE(
|
|
* pSchemaName => 'OU_TOP',
|
|
* pTableName => 'AGGREGATED_ALLOTMENT',
|
|
* pKeyColumnName => 'A_ETL_LOAD_SET_KEY_FK',
|
|
* pBucketArea => 'ARCHIVE',
|
|
* pFolderName => 'exports',
|
|
* pMinDate => DATE '2025-09-01',
|
|
* pMaxDate => DATE '2025-09-17',
|
|
* pRegisterExport => TRUE -- Registers each export to tracking tables
|
|
* );
|
|
* -- This will create files like: AGGREGATED_ALLOTMENT_202509.csv, etc.
|
|
* pBucketArea parameter accepts: 'INBOX', 'ODS', 'DATA', 'ARCHIVE'
|
|
* end;
|
|
**/
|
|
PROCEDURE EXPORT_TABLE_DATA_TO_CSV_BY_DATE (
|
|
pSchemaName IN VARCHAR2,
|
|
pTableName IN VARCHAR2,
|
|
pKeyColumnName IN VARCHAR2,
|
|
pBucketArea IN VARCHAR2,
|
|
pFolderName IN VARCHAR2,
|
|
pFileName IN VARCHAR2 DEFAULT NULL,
|
|
pColumnList IN VARCHAR2 default NULL,
|
|
pMinDate IN DATE default DATE '1900-01-01',
|
|
pMaxDate IN DATE default SYSDATE,
|
|
pParallelDegree IN NUMBER default 1,
|
|
pTemplateTableName IN VARCHAR2 default NULL,
|
|
pMaxFileSize IN NUMBER default 104857600,
|
|
pRegisterExport IN BOOLEAN default FALSE,
|
|
pCredentialName IN VARCHAR2 default ENV_MANAGER.gvCredentialName
|
|
);
|
|
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
-- VERSION MANAGEMENT FUNCTIONS
|
|
---------------------------------------------------------------------------------------------------------------------------
|
|
|
|
/**
|
|
* Returns the current package version number
|
|
* return: Version string in format X.Y.Z (e.g., '2.1.0')
|
|
**/
|
|
FUNCTION GET_VERSION RETURN VARCHAR2;
|
|
|
|
/**
|
|
* Returns comprehensive build information including version, date, and author
|
|
* return: Formatted string with complete build details
|
|
**/
|
|
FUNCTION GET_BUILD_INFO RETURN VARCHAR2;
|
|
|
|
/**
|
|
* Returns the version history with recent changes
|
|
* return: Multi-line string with version history
|
|
**/
|
|
FUNCTION GET_VERSION_HISTORY RETURN VARCHAR2;
|
|
|
|
END;
|
|
|
|
/
|