90 DBMS_JSON

The DBMS_JSON package provides an interface for data-guide operations.

This chapter contains the following topics:

90.1 DBMS_JSON Overview

Package DBMS_JSON provides subprograms for manipulating JavaScript Object Notation (JSON) data that is stored in Oracle Database.

90.2 DBMS_JSON Security Model

PUBLIC is granted the EXECUTE privilege on package DBMS_JSON. Its subprograms execute with invoker's rights privileges.

90.3 DBMS_JSON Constants

The DBMS_JSON package uses these constants to define the JSON schema types and data-guide formatting options.

Table 90-1 DBMS_JSON Constants Defined for JSON Data-Guide Formatting

Name Value Description
FORMAT_FLAT 2 Display flat format
FORMAT_HIERARCHICAL 1 Display hierarchical format
PRETTY 1 Use appropriate indention to improve readability

Table 90-2 DBMS_JSON Constants for JSON Schema Types

Name Type Value Description
TYPE_ARRAY NUMBER(2) 6 A JSON array
TYPE_BOOLEAN NUMBER(2) 2 A JSON boolean
TYPE_OBJECT NUMBER(2) 5 A JSON object
TYPE_NULL NUMBER(2) 1 The JSON NULL value
TYPE_NUMBER NUMBER(2) 3 A JSON number
TYPE_STRING NUMBER(2) 4 A JSON string

90.4 Summary of DBMS_JSON Subprograms

This table lists the DBMS_JSON subprograms and briefly describes them.

DBMS_JSON Package Subprograms

Subprogram Description
ADD_VIRTUAL_COLUMNS Procedure

Add virtual columns based on data-guide information.

This has no effect when running on the shard catalog server — no virtual column is added.

CREATE_VIEW Procedure Create a view with relational columns and scalar JSON fields as specified in a data guide.
CREATE_VIEW_ON_PATH Procedure

Create a view based on data-guide information, with relational columns, top-level scalar types, and fully expanded sub-tree under a given path.

When running on the shard catalog server this raises an error stating that the data guide is empty.

DROP_VIRTUAL_COLUMNS Procedure

Drop virtual columns created by procedure add_virtual_columns.

This has no effect when running on the shard catalog server.

GET_INDEX_DATAGUIDE Function

Get JSON data guide from a data guide-enabled JSON search index.

When running on the shard catalog server this returns a single empty row as result.

RENAME_COLUMN Procedure

Set the preferred name for a view column or a virtual column creating using a data guide.

This has no effect when running on the shard catalog server.

Note:

In the context of sharding, each individual shard maintains its own data-guide information, which is obtained from the JSON documents stored in that shard. When running on individual shard, procedures in this package that use data-guide information use only the information that is maintained for that shard.

90.4.1 ADD_VIRTUAL_COLUMNS Procedure

This procedure adds virtual columns based on the data guide.

The virtual column name is the value of o:preferred_vc_name in the data guide. The procedure ignores JSON objects, arrays, and fields under arrays in the data guide. Before it adds virtual columns, procedure add_virtual_columns first drops any existing virtual columns that were projected from fields in the same JSON column by a previous invocation of add_virtual_columns or by data-guide change-trigger procedure add_vc (in effect, it does what procedure DBMS_JSON.drop_virtual_columns does).

Syntax

DBMS_JSON.ADD_VIRTUAL_COLUMNS (
    tablename  IN  VARCHAR2, 	                
    jcolname   IN  VARCHAR2, 	                
    dataguide  IN  CLOB);

For the following signature you must have a data guide-enabled search index on the JSON column. (This is not needed for the previous signature.)

DBMS_JSON.ADD_VIRTUAL_COLUMNS (
    tablename  IN  VARCHAR2, 	                
    jcolname   IN  VARCHAR2, 	
    frequency      NUMBER    DEFAULT 0,                
    hidden         BOOLEAN   DEFAULT FALSE);

Parameters

Table 90-3 ADD_VIRTUAL_COLUMNS Procedure Parameters

Parameter Description
tablename Name of the table containing JSON column jcolname.
jcolname Name of the JSON column in table tablename that contains the data from which to create the virtual column.
frequency Sets the minimum frequency threshold to display JSON columns. A frequency of 0 means display all JSON columns. Also, all JSON columns are displayed if statistics have not been collected, effectively overriding any value set by this parameter.
hidden TRUE means the added virtual column is hidden; FALSE means it is not. The default is FALSE.
dataguide The data guide. When o:hidden in the data guide for a particular JSON field is set to TRUE, the corresponding virtual column is added as a hidden column. The default value of o:hidden is FALSE.

Usage Notes

Procedure DBMS_STATS.GATHER_STATS collects statistics in the data guide. If the frequency statistic has not been collected, frequency is NULL. Setting the frequency to a value greater than zero means do not include columns for which there are no frequency statistics collected (statistic is NULL), unless DBMS_STATS.GATHER_STATS has never been executed. In that case, the frequency parameter is ignored and all columns are displayed in the view.

90.4.2 CREATE_VIEW Procedure

This procedure creates a view with relational columns, using scalar JSON fields as specified in the data guide. A data guide-enabled JSON search index is not required for this procedure; the data guide is passed to the procedure.

Syntax

(Optional) Enter syntax information here.

PROCEDURE CREATE_VIEW (
   viewname VARCHAR2,                       
   tablename VARCHAR2,                       
   jcolname VARCHAR2,                       
   dataguide CLOB);

Parameters

Table 90-4 DBMS_JSON.CREATE_VIEW Procedure Parameters

Parameter Description
viewname Name of the view.
tablename Name of the table containing JSON column jcolname.
jcolname Name of the JSON column in table tablename that is used to create the view.
dataguide The data guide.

90.4.3 CREATE_VIEW_ON_PATH Procedure

This procedure creates a view with relational columns, using top-level scalar values and the scalar values in the expanded sub-tree under a given path. The JSON column must have a data guide-enabled search index.

Syntax

PROCEDURE CREATE_VIEW_ON_PATH(
   viewname VARCHAR2, 
   tablename VARCHAR2, 
   jcolname VARCHAR2, 
   path VARCHAR2,
   frequency NUMBER DEFAULT 0);

Parameters

Table 90-5 CREATE_VIEW_ON_PATH Procedure Parameters

Parameter Description
viewname Name of the view.
tablename Name of the table containing JSON column jcolname.
jcolname Name of the JSON column in table tablename that is used to create the view. The column must have a data guide-enabled JSON search index, or else an error is raised.
path The path of the JSON field to be expanded. It uses JSON path-expression syntax. It expands the descendants under the specified path, and creates view columns for each scalar value in the resulting sub-tree. The path $ creates a view starting from the JSON document root.
frequency The minimum frequency threshold for displaying the JSON columns. A frequency of 0 means display all JSON columns. All JSON columns are also displayed if statistics have not been collected, effectively overriding any value set by this parameter. The view only displays JSON fields with frequency greater than the given frequency. It does not display JSON fields added after collecting statistics if the given frequency is greater than 0, if their statistic columns are NULL.

90.4.4 DROP_VIRTUAL_COLUMNS Procedure

Drop all virtual columns that were added using PL/SQL procedure DBMS_JSON.add_virtual_columns or using data-guide change-trigger procedure add_vc.

Syntax

PROCEDURE DROP_VIRTUAL_COLUMNS(
   tablename VARCHAR2,                                
   jcolname VARCHAR2);

Parameters

Table 90-6 DBMS_JSON.DROP_VIRTUAL_COLUMNS Procedure Parameters

Parameter Description
tablename Name of the table containing JSON column jcolname.
jcolname Name of the JSON column in table tablename.

90.4.5 GET_INDEX_DATAGUIDE Function

GET_INDEX_DATAGUIDE gets JSON data guide from data guide-enabled JSON search index.

Syntax

FUNCTION GET_INDEX_DATAGUIDE(
   tablename VARCHAR2,
   jcolname VARCHAR2,
   format NUMBER,
   pretty NUMBER DEFAULT 0) 
   RETURN CLOB;

Parameters

Table 90-7 DBMS_JSON.GET_INDEX_DATAGUIDE Procedure Parameters

Parameter Description
tablename Name of the table containing JSON column jcolname.
jcolname Name of the JSON column in table tablename that has a data guide-enabled JSON search index.
format The data-guide format:
  • FORMAT_HIERARCHICAL — hierarchical format
  • FORMAT_FLAT — flat format
pretty A value of DBMS_JSON.PRETTY means pretty-print the data guide, using indention to improve readability.

Example 90-1 Example Get Data Guide in Hierarchical Pretty Format

This example returns the data guide in hierarchical format.

SELECT DBMS_JSON.GET_INDEX_DATAGUIDE(‘T1’, ‘PO’, DBMS_JSON.FORMAT_HIERARCHICAL, DBMS_JSON.PRETTY) 
FROM DUAL;

90.4.6 RENAME_COLUMN Procedure

This procedure sets the preferred name for a JSON column, to be used by the create view, or add virtual columns procedure.

Syntax

PROCEDURE RENAME_COLUMN(
   tablename VARCHAR2,
   jcolname VARCHAR2,
   path VARCHAR2,
   type NUMBER,
   preferred_name VARCHAR2);

Parameters

Table 90-8 RENAME_COLUMN Procedure Parameters

Parameter Description
tablename Name of the table containing JSON column jcolname.
jcolname Name of the JSON column in table tablename. It must have a data guide-enabled JSON search index, or else an error is raised.
path Path to the JSON field on which to set the preferred column name.
type The type of the JSON field targeted by path. Two JSON fields can have the same path if they are of different types. Possible values:
  • TYPE_NULL
  • TYPE_STRING
  • TYPE_NUMBER
  • TYPE_BOOLEAN
  • TYPE_OBJECT
  • TYPE_ARRAY
preferred_name Preferred name for the JSON field specified by path. If there is a name conflict, a system generated name is used instead.

Example 90-2 Example Renaming a Column

This example renames a field to item_name.

EXEC DBMS_JSON.RENAME_COLUMN(‘T1’, ‘PO’, ‘$.purchaseOrder.items.name’, DBMS_JSON.TYPE_STRING, ‘item_name’);