1150

Teradata Database
Teradata JSON
Release 15.0
B035-1150-015K
June 2014
The product or products described in this book are licensed products of Teradata Corporation or its affiliates.
Teradata, Active Data Warehousing, Active Enterprise Intelligence, Applications-Within, Aprimo Marketing Studio, Aster, BYNET,
Claraview, DecisionCast, Gridscale, MyCommerce, SQL-MapReduce, Teradata Decision Experts, "Teradata Labs" logo, Teradata
ServiceConnect, Teradata Source Experts, WebAnalyst, and Xkoto are trademarks or registered trademarks of Teradata Corporation or its
affiliates in the United States and other countries.
Adaptec and SCSISelect are trademarks or registered trademarks of Adaptec, Inc.
AMD Opteron and Opteron are trademarks of Advanced Micro Devices, Inc.
Apache, Apache Hadoop, Hadoop, and the yellow elephant logo are either registered trademarks or trademarks of the Apache Software
Foundation in the United States and/or other countries.
Apple, Mac, and OS X all are registered trademarks of Apple Inc.
Axeda is a registered trademark of Axeda Corporation. Axeda Agents, Axeda Applications, Axeda Policy Manager, Axeda Enterprise, Axeda
Access, Axeda Software Management, Axeda Service, Axeda ServiceLink, and Firewall-Friendly are trademarks and Maximum Results and
Maximum Support are servicemarks of Axeda Corporation.
Data Domain, EMC, PowerPath, SRDF, and Symmetrix are registered trademarks of EMC Corporation.
GoldenGate is a trademark of Oracle.
Hewlett-Packard and HP are registered trademarks of Hewlett-Packard Company.
Hortonworks, the Hortonworks logo and other Hortonworks trademarks are trademarks of Hortonworks Inc. in the United States and other
countries.
Intel, Pentium, and XEON are registered trademarks of Intel Corporation.
IBM, CICS, RACF, Tivoli, and z/OS are registered trademarks of International Business Machines Corporation.
Linux is a registered trademark of Linus Torvalds.
LSI is a registered trademark of LSI Corporation.
Microsoft, Active Directory, Windows, Windows NT, and Windows Server are registered trademarks of Microsoft Corporation in the United
States and other countries.
NetVault is a trademark or registered trademark of Dell Inc. in the United States and/or other countries.
Novell and SUSE are registered trademarks of Novell, Inc., in the United States and other countries.
Oracle, Java, and Solaris are registered trademarks of Oracle and/or its affiliates.
QLogic and SANbox are trademarks or registered trademarks of QLogic Corporation.
Quantum and the Quantum logo are trademarks of Quantum Corporation, registered in the U.S.A. and other countries.
Red Hat is a trademark of Red Hat, Inc., registered in the U.S. and other countries. Used under license.
SAS and SAS/C are trademarks or registered trademarks of SAS Institute Inc.
SPARC is a registered trademark of SPARC International, Inc.
Symantec, NetBackup, and VERITAS are trademarks or registered trademarks of Symantec Corporation or its affiliates in the United States
and other countries.
Unicode is a registered trademark of Unicode, Inc. in the United States and other countries.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Other product and company names mentioned herein may be the trademarks of their respective owners.
The information contained in this document is provided on an "as-is" basis, without warranty of any kind, either express
or implied, including the implied warranties of merchantability, fitness for a particular purpose, or non-infringement.
Some jurisdictions do not allow the exclusion of implied warranties, so the above exclusion may not apply to you. In no
event will Teradata Corporation be liable for any indirect, direct, special, incidental, or consequential damages, including
lost profits or lost savings, even if expressly advised of the possibility of such damages.
The information contained in this document may contain references or cross-references to features, functions, products, or services that are
not announced or available in your country. Such references do not imply that Teradata Corporation intends to announce such features,
functions, products, or services in your country. Please consult your local Teradata Corporation representative for those features, functions,
products, or services available in your country.
Information contained in this document may contain technical inaccuracies or typographical errors. Information may be changed or
updated without notice. Teradata Corporation may also make improvements or changes in the products or services described in this
information at any time without notice.
To maintain the quality of our products and services, we would like your comments on the accuracy, clarity, organization, and value of this
document. Please e-mail: [email protected]
Any comments or materials (collectively referred to as "Feedback") sent to Teradata Corporation will be deemed non-confidential. Teradata
Corporation will have no obligation of any kind with respect to Feedback and will be free to use, reproduce, disclose, exhibit, display,
transform, create derivative works of, and distribute the Feedback and derivative works thereof without limitation on a royalty-free basis.
Further, Teradata Corporation will be free to use any ideas, concepts, know-how, or techniques contained in such Feedback for any purpose
whatsoever, including developing, manufacturing, or marketing products or services incorporating Feedback.
Copyright 2014 by Teradata. All Rights Reserved.
Preface
Purpose
Teradata JSON describes Teradata support for JSON data, including the JSON data type and
the functions and methods available for processing, shredding, and publishing JSON data.
Use Teradata JSON with the other books in the SQL book set.
Audience
Database administrators, application programmers, and other technical personnel
responsible for designing, maintaining, and using Teradata Database will find Teradata
JSON useful.
Supported Software Releases and Operating

Systems
This book supports Teradata Database 15.0.
Teradata Database 15.0 is supported on:
SUSE Linux Enterprise Server (SLES) 10 SP3
SUSE Linux Enterprise Server (SLES) 11 SP1
Teradata Database client applications support other operating systems.
Prerequisites
If you are not familiar with Teradata Database, you will find it useful to read Introduction to
Teradata, B035-1091 before reading Teradata JSON.
You should be familiar with basic relational database management technology and SQL. You
should also have basic knowledge of programming languages, such as JavaScript, C++, and
so on, and common programming language operations.
Teradata JSON
Preface
Changes to this Book
Changes to this Book

Release
Description
Teradata Database 15.0
Initial publication.
March 2014
April 2014
Added the following information:

Teradata Database client product support for
the JSON data type.
Loading JSON data using load utilities.
A warning that a JSON value can be truncated
if the size of the value exceeds the size of the
target column.
A recommendation that users do not use
algorithmic and block-level compression
together.
Additional examples for the JSON_KEYS
table operator.
June 2014
Added information about extracting JSON

data using SELECT and the JSON methods.
Revised the error message returned when a
JSON value is too large to store in the defined
JSON type.
Additional Information
URL
Description
www.info.teradata.com
Use the Teradata Information Products Publishing Library site to:

View or download a manual:
Under Online Publications, select General Search.
Enter your search criteria and click Search.
Download a documentation CD-ROM:
Under Online Publications, select General Search.
In the Title or Keyword field, enter CD-ROM, and click Search.
www.teradata.com
The Teradata home page provides links to numerous sources of

information about Teradata. Links include:
Executive reports, white papers, case studies of customer
experiences with Teradata, and thought leadership
Technical information, solutions, and expert advice
Press releases, mentions and media resources
Teradata JSON
Preface
Product Safety Information
URL
Description
www.teradata.com/TEN/
Teradata Customer Education delivers training that builds skills and

capabilities for our customers, enabling them to maximize their
Teradata investment.
tays.teradata.com
Use Teradata @ Your Service to access Orange Books, technical alerts,

and knowledge repositories, view and join forums, and download
software patches.
Teradata Developer
Exchange
Teradata Developer Exchange provides articles on using Teradata

products, technical discussion forums, and code downloads.
To maintain the quality of our products and services, we would like your comments on the
accuracy, clarity, organization, and value of this document. Please email [email protected].
Product Safety Information

This document may contain information addressing product safety practices related to data
or property damage, identified by the word Notice. A notice indicates a situation which, if
not avoided, could result in damage to property, such as equipment or data, but not related
to personal injury.
Example
Notice: Improper use of the Reconfiguration utility can result in data loss.
Teradata Database Optional Features

This book may include descriptions of the following optional Teradata Database features and
products:
Teradata Columnar
Teradata QueryGrid: Teradata Database-to-Hadoop
Teradata QueryGrid: Teradata Database-to-Oracle Database
Teradata Row-Level Security
Teradata Temporal
Teradata Virtual Storage (VS)
You may not use these features without the appropriate licenses. The fact that these features
may be included in product media or downloads, or described in documentation that you
receive, does not authorize you to use them without the appropriate licenses.
Contact your Teradata sales representative to purchase and enable optional features.
Teradata JSON
Preface
Teradata Database Optional Features
Teradata JSON
CHAPTER 1
The JSON Data Type
Teradata Support for JSON

JSON (Javascript Object Notation) is a text-based, data interchange format, often used in
web applications to transmit data. JSON has been widely adopted by web application
developers because compared to XML it is easier to read and write for humans and easier to
parse and generate for machines. JSON documents can be stored and processed in Teradata
Database.
Teradata Database can store JSON records as a JSON document or store JSON records in
relational format. Teradata Database provides the following support for JSON data.
Methods, functions, and stored procedures that operate on the JSON data type, such as
parsing and validation.
Shredding functionality that allows you to extract values from JSON documents up to
16MB in size and store the extracted data in relational format.
Publishing functionality that allows you to publish the results of SQL queries in JSON
format.
Schema-less or dynamic schema with the ability to add a new attribute without changing
the schema. Data with new attributes is immediately available for querying. Rows
without the new column can be filtered out.
Use existing join indexing structures on extracted portions of the JSON data type.
Apply advanced analytics to JSON data.
Functionality to convert an ST_Geometry object into a GeoJSON value and a GeoJSON
value into an ST_Geometry object.
Allows JSON data of varying maximum length and JSON data can be internally
compressed.
Collect statistics on extracted portions of the JSON data type.
Use standard SQL to query JSON data.
JSONPath provides simple traversal and regular expressions with wildcards to filter and
navigate complex JSON documents.
Client Support for JSON

The following table describes the support provided by the Teradata Database client products
for the JSON data type.
Teradata JSON
Client Product
JSON Support Provided
CLI
Full native DBS support.
Chapter 1 The JSON Data Type

Client Product
JSON Support Provided
ODBC
The ODBC specification does not have a unique data type code for JSON.
Therefore, the ODBC driver maps the JSON data type to
SQL_LONGVARCHAR or SQL_WLONGVARCHAR, which are the
ODBC CLOB data types. The metadata clearly differentiates between a
Teradata CLOB data type mapped to SQL_LONGVARCHAR and a
Teradata JSON data type mapped to SQL_LONGVARCHAR.
The ODBC driver supports LOB Input, Output and InputOutput
parameters. Therefore, it can load JSON data. Also the Catalog (Data
Dictionary) functions support JSON.
JDBC
VARCHAR values or CLOB values can be inserted into JSON

destination columns.
JSON values cannot be directly retrieved from the ResultSet of a query.
An SQLException is thrown with Error 1245 "Unable to obtain data
value because the Teradata Database indicated that the data type is
ambiguous".
The JSON columns in the select list of a query must be CAST to
VARCHAR or CLOB.
.NET Data Provider
No support.
Teradata Parallel
Transporter (TPT)
JSON columns are treated exactly like CLOB columns and subject to the
same limitations. JSON columns cannot exceed 16 MB (16,776,192 LATIN
characters or 8,388,096 UNICODE characters). TPT accommodates the
JSON keyword in object schema but internally converts it to CLOB.
Import and export are both fully supported.
BTEQ
The JSON keyword cannot be used in the USING data statement;

therefore, JSON values must be referred to as either CLOB or VARCHAR.
For VARCHAR, the JSON value cannot exceed 64 KB (64000 LATIN
characters or 32000 UNICODE characters).
Currently, BTEQ does not support deferred mode LOB transfer for server
to client. Only non-deferred mode JSON transfer is supported for server to
client, and the maximum size of an output row is limited to 64 KB.
Standalone Utilities
No support.
For more information about the Teradata Database client products, see the following books:
Teradata Call-Level Interface Version 2 Reference for Mainframe-Attached Systems,
B035-2417
Teradata Call-Level Interface Version 2 Reference for Workstation-Attached Systems,
B035-2418
ODBC Driver for Teradata User Guide, B035-2509
Teradata JDBC Driver User Guide, B035-2403
Teradata Parallel Transporter Reference, B035-2436
Teradata Parallel Transporter User Guide, B035-2445
Basic Teradata Query Reference, B035-2414
Teradata JSON

Terminology
A JSON document or JSON message is any string that conforms to the JSON format. When
discussing JSON values in the SQL context, JSON documents are referred to as an instance
of the JSON data type or simply as a JSON instance.
A JSON document structured as an object is encapsulated in {}. A JSON document
structured as an array is encapsulated in []. In the context of SQL, they are both JSON data
type instances. For details, see JSON String Syntax
When we discuss the serialized form (such as the example below), we call it a JSON
document. To describe the structure of the JSON document, we say "a JSON document
structured as an array or object" or simply JSON array and JSON object. The following is an
example of a JSON document.
{
"name": "Product",
"properties": {
"id": {
"type": "number",
"description": "Product identifier",
"required": true
},
"name": {
"type": "string",
"description": "Name of the product",
"required": true
},
"price": {
"type": "number",
"minimum": 0,
"required": true
},
"tags": {
"type": "array",
"items": {
"type": "string"
}
},
"stock": {
"type": "object",
"properties": {
"warehouse": {
"type": "number"
},
"retail": {
"type": "number"
}
}
}
}
}
Teradata JSON

Overview of the JSON Data Type
Overview of the JSON Data Type

JSON is a data type provided by the Teradata Database. You can use it the same way as other
SQL data types.
JSON content is stored in Teradata Database in an optimized format depending on the
size of the data.
The user is not responsible for executing the CREATE TYPE statement for the JSON data
type. JSON data types do not need to be created via DDL by the user as the JSON data
type exists in the database.
The JSON data type cannot be created, dropped, or altered by the user.
There are two points of commonality between all uses of the JSON data type in any valid
location:
The maximum length of any instance of the JSON data type is variable. You may set
the maximum length of the JSON instance, otherwise the default maximum length for
the character set is used.
The CHARACTER SET for the text of the JSON data type can be either UNICODE or
LATIN. If you don't specify the character set, the default character set of the user is
used.
Standards Compliance
The Teradata JSON data type is compliant with the standard for JSON syntax defined by
IETF RFC 4627. The standard is available at http://www.ietf.org/rfc/rfc4627.txt and http://
www.json.org/.
JSON Data Type Syntax

Syntax
The following shows the syntax when you use a JSON data type in a data type declaration
phrase. For example, this syntax is used when defining a table column to be JSON type.
JSON
( integer )
CHARACTER SET
UNICODE
attributes
LATIN
Syntax Elements
integer
A positive integer value that specifies the maximum length in characters of the JSON
type.
If you do not specify a maximum length, the default maximum length for the character
set is used. If specified, the length is subject to a minimum of two characters and cannot
10
Teradata JSON

Maximum Length of a JSON Instance
be greater than the maximum of 16776192 LATIN characters or 8388096 UNICODE

characters.
CHARACTER SET
The character set for the JSON type can be UNICODE or LATIN. If you do not specify a
character set, the default character set for the user is used.
attributes
The following data type attributes are supported for the JSON type.
NULL and NOT NULL
FORMAT
TITLE
NAMED
DEFAULT NULL
COMPRESS USING and DECOMPRESS USING
For details on these data type attributes, see SQL Data Types and Literals, B035-1143 .

The JSON type has a default maximum length. If no maximum length is specified, the
default maximum length for the character set is chosen. If specified, the length is subject to a
minimum of two characters and cannot be greater than the absolute maximum for the
character set. If the character set is not specified, the character set of the user is used.
The maximum length of the JSON text depends on the character set being used:
If the character set is UNICODE, the maximum length is 8388096 characters. This is
equivalent to 16776192 bytes, which is the maximum possible length for the JSON type.
If the character set is LATIN, the maximum length is 16776192 characters. This is
A shorter maximum length can be specified wherever the JSON type is used, for example, as
the data type of a table column or as a parameter to a function.
If you exceed the maximum length, you will get the following error:
*** JSON value is too large to store in the defined JSON type.
For example, if you try to insert data that is larger than the maximum length defined for a
JSON column, or pass data that is larger than what is supported by a JSON parameter, you
will get this error.
JSON data is stored inline and in LOB subtables depending on the size of the data.
1. If the maximum length specified is less than or equal to 64000 bytes (64000 LATIN
characters or 32000 UNICODE characters), then the data in all rows of the table are
stored inline.
2. If the maximum length specified exceeds 64000 bytes, then rows with less than 4K bytes
of data are stored inline, and rows with more than 4K bytes of data are stored in a LOB
subtable. A table with a column defined in this manner will have a few rows stored inline
and a few rows in a LOB subtable, depending on data size.
Teradata JSON
11

Examples: Creating Tables with JSON Columns with No Maximum Length

Specified or No Character Set Defined
The following examples demonstrate the maximum length and character set specifications.
Example: Create a table with a JSON type column, with no maximum length specified and no
character set specified:
CREATE TABLE json_table(id INTEGER, json_j1 JSON);
Result: This creates a table with LATIN character set with a maximum length of that
character set, 16776192 LATIN characters. When a character set is not specified for the JSON
type, the default character set for the user is used. The result for this example assumes the
user had LATIN as their default character set.
Example: Create a table with a JSON type column, with no maximum length specified and
specify UNICODE character set:
CREATE TABLE json_table(id INTEGER, json_j1 JSON CHARACTER SET
UNICODE);
Result: This creates a table with UNICODE character set and a maximum length of that
character set, 8388096 UNICODE characters.
Example: Create a table with a JSON type column, with a maximum length specified and no
CREATE TABLE json_table(id INTEGER, json_j1 JSON(100000));
Result: This creates a table with a maximum length of 100000 LATIN characters. Note, the
result for this example assumes the user had LATIN as their default character set.
Example: Create a table with a JSON type column, with a maximum length specified and
UNICODE character set specified:
CREATE TABLE json_table(id INTEGER, json_j1 JSON(100000) CHARACTER SET
UNICODE);
Result: This creates a table with a maximum length of 100000 UNICODE characters.
Example: Create a table with JSON type columns, with a maximum length specified that
exceeds the allowed length and no character set specified:
CREATE TABLE json_table(id INTEGER, json_j1 JSON(64000), json_j2
JSON(12000));
Result: This fails because the maximum possible amount of data stored in the row could grow
to approximately 76000 bytes. This exceeds the maximum row size, as described in item 1
earlier.
Example: Create a table with JSON type columns, with a maximum length specified and no
CREATE TABLE json_table(id INTEGER, json_j1 JSON(64001), json_j2
JSON(12000));
Result: This succeeds because the maximum possible amount of data stored in the row is
~16000 bytes which is within the maximum row size. This is because the json_j1 column has
12
Teradata JSON

the storage scheme described in item 2 earlier, in which a maximum of 4K bytes will be
stored in the row.
Example: Creating a Table with JSON Columns

The following table defines five columns that are the JSON type, each with a different
maximum length.
CREATE TABLE employee (
id INTEGER,
json1 JSON(20),
json2 JSON(25),
json3 JSON(30),
json4 JSON(1000),
json5 JSON(5000));
Examples: Creating a Function with JSON Type Parameters

The following shows functions with an input or return parameter that is a JSON type.
CREATE FUNCTION json1
(p1 JSON(1000))
RETURNS VARCHAR(100)
NO SQL
PARAMETER STYLE SQL
CALLED ON NULL INPUT
DETERMINISTIC
LANGUAGE C
EXTERNAL NAME 'CS!json1!json1.c!F!json1';
void json1 (
JSON_HANDLE
*json_handle,
VARCHAR_LATIN *result,
int
*indicator_ary,
int
*indicator_result,
char
sqlstate[6],
SQL_TEXT
extname[129],
SQL_TEXT
specific_name[129],
SQL_TEXT
error_message[257])
{
/* body function */
}
CREATE FUNCTION json2
(p1 VARCHAR(100))
RETURNS JSON(100) NO SQL
PARAMETER STYLE SQL
DETERMINISTIC
LANGUAGE C
EXTERNAL NAME 'CS!json2!json2.c!F!json2';
Teradata JSON
13

Supported Character Sets
void json2 (
JSON_HANDLE
*json_handle,
int
*indicator_ary,
int
*indicator_result,
char
sqlstate[6],
SQL_TEXT
extname[129],
SQL_TEXT
specific_name[129],
SQL_TEXT
error_message[257])
{
/* body function */
}
Example: Error: JSON Value Is Too Large to Store in the Defined JSON
Type
In this example, an error is returned when data being inserted into a JSON column is larger
than the maximum length defined.
The smallJSONTable table in this example has a JSON column with a maximum length of 10
LATIN characters.
CREATE TABLE smallJSONTable(id INTEGER, j JSON(10));
The following INSERT statement succeeds because the data inserted into the JSON column is
less than 10 characters.
INSERT INTO smallJSONTable(1, '{"a":1}');
*** Insert completed. One row added.
*** Total elapsed time was 1 second.
The following INSERT statement fails because '{"a":12345}' is greater than the maximum
length of 10 characters.
INSERT INTO smallJSONTable(1, '{"a":12345}');
*** Failure JSON value is too large to store in the defined JSON
type.
Supported Character Sets

You can only use the UNICODE or LATIN character sets with the JSON data type. Other
character sets such as GRAPHIC, KANJI1, or KANJISIS are not supported.
In each particular use of the JSON data type, you can specify the character set using the
CHARACTER SET clause. If you do not specify a character set, then the default character set
for the user is used.
14
Teradata JSON

Character Set Handling
Character Set Handling

The JSON data type accepts all available session character sets. The data is translated to
either the UNICODE or LATIN character set, depending on the definition of the JSON type
instance where the data is being used. Therefore, any characters used must be translatable to
either UNICODE or LATIN.
Any character set that must be translated is subject to the size restrictions of the UNICODE
or LATIN character set. The data in some character sets may appear to be the correct size,
but when translated to one of the two character sets, the data may be too large and result in a
size error.
If the data transformed from a JSON instance needs to be in a particular session character
set, the database will perform the translation, assuming that all of the characters can be
translated to that character set.
About JSON Type Constructor

Teradata provides a constructor for creating JSON type instances.
NEW JSON Constructor Expression

Purpose
The NEW JSON constructor allocates a JSON type instance. The constructor can be used
without arguments (default constructor) or with arguments to set the value and the
character set of the JSON instance. The resulting JSON instance can be used to insert a JSON
type into a column or as a JSON type argument to a function or method.
Syntax
NEW JSON
)
'JSON_string '
, LATIN
, UNICODE
Syntax Elements
'JSON_string'
The value of the resulting JSON type.
The string must conform to JSON syntax as described in JSON String Syntax.
LATIN or UNICODE
The character set of the resulting JSON type.
If you do not specify a character set, the default character set of the user is used.
Rules and Restrictions
When inserting a JSON type into a column or using the JSON type as an argument to a
function or method, you can pass in a string of any size as long as it is less than or equal to
Teradata JSON
15

About JSON Type Transform
the maximum possible length of the resulting JSON type, which is 16776192 LATIN
characters or 8388096 UNICODE characters. If the data being inserted is longer than the
maximum length specified for a particular JSON instance, an error is reported.
Usage Notes
In the default constructor no arguments are passed to the constructor expression. NEW
JSON() initializes an empty JSON type value with the character set based on the default
character set of the user. The data is set to a null string.
You can append a JSON entity reference to the end of a constructor expression as described
in JSON Entity Reference.
Example: Default JSON Constructor
NEW JSON();
Examples: JSON Constructor with Arguments

NEW JSON ('{"name" : "cameron", "age" : 24}')
NEW JSON ('{"name" : "cameron", "age" : 24}', LATIN)
About JSON Type Transform

Teradata automatically generates the fromsql and tosql functionality associated with the
transform of a newly created JSON type. The JSON string is transformed to/from a
CLOB(length) value, where length depends on the data. The character set is either UNICODE
or LATIN depending on the JSON type instance. The format of the transformed output string
conforms to the standard JSON string syntax.
Transform Input/Output String Format

For the JSON type, the following two items use the same string format.
tosql transform input parameter
fromsql transform output value
The format is a string that conforms to the standard JSON string syntax. It is represented as
CLOB(16776192) CHARACTER SET LATIN or CLOB(8388096) CHARACTER SET
UNICODE.
Related Topics
JSON String Syntax, on page 30

Maximum Length of a JSON Instance, on page 11
About JSON Type Cast

Teradata provides casting functionality for a JSON type.
16
Teradata JSON

About JSON Type Ordering
The JSON type can be cast to all other forms of the JSON type.
The JSON type can be cast to a JSON type of a different character set, such as JSON
LATIN from JSON UNICODE.
The JSON type can be cast to and from VARCHAR and CLOB of the same character set:
VARCHAR(32000) CHARACTER SET UNICODE
VARCHAR(64000) CHARACTER SET LATIN
CLOB(8388096) CHARACTER SET UNICODE
CLOB(16776192) CHARACTER SET LATIN
The JSON type can be cast to a CHAR of the same character set:
CHAR(32000) CHARACTER SET UNICODE
CHAR(64000) CHARACTER SET LATIN
The casting functionality can be implicitly invoked, and the format of the data cast to/from
in the text conforms to JSON syntax.
If any truncation occurs as a result of the cast, an error is reported.
About JSON Type Ordering

Ordering, comparison, or grouping are not allowed on a JSON type, so no default ordering is
provided. If a JSON column is used in a SET table, it is not included in the determination of
the uniqueness of a row. Therefore, if all other columns in a row are equivalent to another
row, the two rows are deemed equivalent.
You cannot use a JSON type in these clauses: GROUP BY, ORDER BY, PARTITION BY,
WHERE, ON, SET, DISTINCT, HAVING, QUALIFY, IN, CUBE, GROUPING SETS or
ROLLUP.
You can use JSON methods to isolate individual portions of a JSON instance for
comparison. You can also cast the JSON type to a predefined type that can have relational
comparisons performed on it.
About JSON Type Usage

The JSON data type can be used similarly to other data types. You can specify the JSON data
type in these cases:
In table definitions
For input parameters or the return type in user-defined functions (UDFs). However, the
JSON type cannot be used in user-defined table operators.
For input parameters or the return type in user-defined methods (UDMs)
For IN, OUT, or INOUT parameters in external stored procedures
For IN, OUT, or INOUT parameters in stored procedures
In local variable definitions in stored procedures
As an attribute of a structured user-defined type (UDT), but not as the base type of a
distinct UDT
Teradata JSON
17

Using JSON Type with UDFs

You can create a UDF containing one or more parameters, or return type, that is a JSON data
type. The JSON type is supported on the following types of UDFs:
Scalar, aggregate, and table UDFs
SQL UDFs
For SQL UDFs, the RETURN clause can be an SQL statement that evaluates to the JSON data
type.
The JSON data type is only supported for UDFs and external stored procedures written in C
or C++. You cannot use the JSON type with UDFs or procedures written in Java.
Example: UDF Parameter Style SQL
This example shows an SQL definition for a UDF using parameter style SQL with a JSON
parameter, and a C function that shows the corresponding parameter list.
/* Parameter Style SQL */
CREATE FUNCTION MyJSONUDF (a1 JSON(100))
NO SQL
PARAMETER STYLE SQL
DETERMINISTIC
LANGUAGE C
EXTERNAL NAME 'CS!MyJSONUDF!MyJSONUDF.c!F!MyJSONUDF';
/* C source file name: myJSONUDF.c */
void MyJSONUDF (
JSON_HANDLE
*json_handle,
int
*indicator_ary,
int
*indicator_result,
char
sqlstate[6],
SQL_TEXT
extname[129],
SQL_TEXT
specific_name[129],
SQL_TEXT
error_message[257])
{
/* body function */
}
Example: UDF Parameter Style TD_GENERAL
This example shows an SQL definition for a UDF using parameter style TD_GENERAL with
a JSON parameter, and a C function that shows the corresponding parameter list.
/* Parameter Style TD_GENERAL */
CREATE FUNCTION MyJSONUDF2 (a1 JSON(100))
NO SQL
PARAMETER STYLE TD_GENERAL
DETERMINISTIC
18
Teradata JSON

LANGUAGE C
EXTERNAL NAME 'CS!MyJSONUDF2!MyJSONUDF2.c!F!MyJSONUDF2';
/* C source file name: myJSONUDF.c */
void MyJSONUDF2 (
JSON_HANDLE
*json_handle,
char
sqlstate[6]
{
/* body function */
}
Example: SQL UDF with RETURN clause
This example shows an SQL UDF with a RETURN expression that evaluates to JSON type.
CREATE FUNCTION New_Funct ()
RETURNS JSON(100)
LANGUAGE SQL
CONTAINS SQL
DETERMINISTIC
SPECIFIC New_Funct
SQL SECURITY DEFINER
COLLATION INVOKER
INLINE TYPE 1
RETURN (new JSON('[1,2,3]'));
Using the JSON Type with Procedures (External Form)

You can create an external stored procedure containing one or more parameters that are a
JSON type. You can use the JSON type to define IN, OUT, or INOUT parameters.
The JSON type is only supported for UDFs and external stored procedures written in C or C
++. You cannot use the JSON type with UDFs or procedures written in Java.
Example: Procedure with Parameter Style SQL
This example shows an SQL definition for an external stored procedure using parameter
style SQL with an IN JSON parameter, and a C function that shows the corresponding
parameter list.
/* Parameter Style SQL */
CREATE PROCEDURE myJSONXSP( IN a JSON(100),
OUT phonenum VARCHAR(100) )
LANGUAGE C
NO SQL
PARAMETER STYLE SQL
EXTERNAL NAME 'CS!myJSONXSP!myJSONXSP.c';
/* C source file name: myJSONXSP.c */
void myJSONXSP (JSON_HANDLE *json_handle,
Teradata JSON
19

int *indicator_ary,
int *indicator_result,
char sqlstate[6],
SQL_TEXT extname[129],
SQL_TEXT specific_name[129],
SQL_TEXT error_message[257])
{
/* body function */
}
Example: Procedure with Parameter Style TD_General
This example shows an SQL definition for an external stored procedure using parameter style
TD_GENERAL with an IN JSON parameter, and a C function that shows the corresponding
parameter list.
/* Parameter Style TD_GENERAL */
CREATE PROCEDURE myJSONXSP2 ( IN a1 JSON(100),
OUT phonenum VARCHAR(100))
NO SQL
LANGUAGE C
EXTERNAL NAME 'CS!myJSONXSP2!myJSONXSP2.c!F!myJSONXSP2';
/* C source file name: myJSONXSP2.c */
void myJSONXSP2 (
JSON_HANDLE
*json_handle,
char
sqlstate[6])
{
/* body function */
}
Using the JSON Type with Procedures (SQL Form)

Stored Procedures with JSON Type Parameters
You can create a stored procedure containing one or more parameters that are a JSON type.
You can use the JSON type to define IN, OUT, or INOUT parameters.
Example: Stored Procedures with JSON Type Parameters
The following statement defines a stored procedure with an IN parameter that is a JSON type
with a maximum length of 100.
CREATE PROCEDURE my_tdsp1 (IN p1 JSON(100))
...
;
The following statement defines a stored procedure with an OUT parameter that is a JSON
type with a maximum length of 100.
20
Teradata JSON

REPLACE PROCEDURE my_tdsp2 (IN p1 INTEGER, OUT p2 JSON(100))
...
;
Stored Procedures with JSON Type Local Variables
You can create a stored procedure containing one or more local variables that are JSON data
type. You can use the DEFAULT clause with the declaration of these local variables:
DEFAULT NULL
DEFAULT value
If you specify DEFAULT value, you can use a JSON constructor expression to initialize the
local variable. See About JSON Type Constructor.
You can use the stored procedure semantics, including cursors, to access a particular value,
name/value pair, or nested documents structured as objects or arrays. You can use this with
the JSON methods and functions that allow JSONPath request syntax for access to specific
portions of a JSON instance.
Example: Stored Procedures with JSON Type Local Variables
The following stored procedure contains local variables, local1 and local2, that are of JSON
type with a maximum length of 100. The local1 variable is initialized using the NEW JSON
constructor method.
CREATE PROCEDURE TestJSON_TDSP1
(IN id INTEGER, OUT outName VARCHAR(20))
BEGIN
DECLARE local1 JSON(100) DEFAULT
NEW JSON('{"name":"Cameron", "age":24}');
DECLARE local2 JSON(100);
;
...
END;
Using the JSON Type with UDMs

You can create a user-defined method (UDM) containing one or more parameters, or return
type, that is a JSON type.
Note: You cannot create methods for the JSON type. The only methods a JSON type can use
are those that are automatically created by Teradata Database. Therefore, you cannot specify
a JSON type name in the FOR clause of a CREATE/REPLACE METHOD statement.
Example: UDM with Parameter Style SQL
The example shows an SQL definition for a UDM with a JSON parameter, and a C function
that defines the method.
/* Parameter style SQL: */
CREATE INSTANCE METHOD JSONMethod (p1 JSON(100))
FOR Some_UDT
RETURNS INTEGER
Teradata JSON
21

FOR Some_UDT
NO SQL
PARAMETER STYLE SQL
DETERMINISTIC
LANGUAGE C
EXTERNAL NAME 'CS!JSONMethod!JSONMethod.c!F!JSONMethod';
/* C source file name: JSONMethod.c */
void JSONMethod (
UDT_HANDLE
JSON_HANDLE
INTEGER
int
int
int
char
SQL_TEXT
SQL_TEXT
SQL_TEXT
{
/* body function
}
*someUdt,
*jsonval,
*result,
*indicator_this,
*indicator_aryval,
*indicator_result,
sqlstate[6],
extname[129],
specific_name[129],
error_message[257])
*/
Example: UDM with Parameter Style TD_GENERAL
The example shows an SQL definition for a UDM using parameter style TD_GENERAL with
a JSON parameter, and a C function that defines the method.
/* Parameter style TD_GENERAL */
CREATE INSTANCE METHOD JSONMethod (p1 JSON(100))
FOR Some_UDT
RETURNS INTEGER
FOR Some_UDT
NO SQL
DETERMINISTIC
LANGUAGE C
EXTERNAL NAME 'CS!JSONMethod!JSONMethod.c!F!JSONMethod';
/* C source file name: JSONMethod.c */
void JSONMethod (
UDT_HANDLE
JSON_HANDLE
INTEGER
char
{
/* body function
}
22
*someUdt,
*jsonval,
*result,
sqlstate[6])
*/
Teradata JSON

Using JSON Type with UDTs

You can specify the JSON data type as an attribute of a structured user-defined type (UDT),
but not as the base type of a distinct UDT. When specifying a JSON attribute, you can define
the size and character set of the JSON type.
Related Topics
FNC Library Routines That Support the JSON Type, on page 27

Example: Creating a Structured UDT with a JSON Type Attribute
This example shows how to create a structured user-defined type (UDT) which has a JSON
type attribute. The routines in this example are created in the SYSUDTLIB database.
Therefore, the user must have the UDTMETHOD privilege on the SYSUDTLIB database.
SQL Definition
This section shows the SQL DDL statements necessary to create the structured UDT.
Create a Structured UDT with a JSON Attribute
The following statement creates a structured UDT named judt with a JSON attribute named
Att1. The maximum length of the JSON attribute is 100000 characters, and the character set
of the JSON attribute is UNICODE.
CREATE TYPE judt AS (Att1 JSON(100000) CHARACTER SET UNICODE) NOT FINAL
CONSTRUCTOR METHOD judt (p1 JSON(100000) CHARACTER SET UNICODE)
RETURNS judt
SELF AS RESULT
SPECIFIC judt_cstr
LANGUAGE C
RETURNS NULL ON NULL INPUT
DETERMINISTIC
NO SQL;
Create the Constructor Method for the UDT

The following statement creates the constructor method used to initialize an instance of the
judt UDT.
CREATE CONSTRUCTOR METHOD judt (p1 JSON(100000) CHARACTER SET UNICODE)
FOR judt
EXTERNAL NAME 'CS!judt_cstr!judt_constructor.c!F!judt_cstr';
Create the Transform Functionality for the UDT

The following statements create the tosql and fromsql transform routines.
REPLACE FUNCTION SYSUDTLIB.judt_fromsql
(judt)
RETURNS CLOB AS LOCATOR CHARACTER SET UNICODE
NO SQL
PARAMETER STYLE SQL
DETERMINISTIC
Teradata JSON
23

LANGUAGE C
EXTERNAL NAME 'CS!judt_fromsql!judt_fromsql.c!F!judt_fromsql';
CREATE FUNCTION SYSUDTLIB.judt_tosql
(CLOB AS LOCATOR CHARACTER SET UNICODE)
RETURNS judt
NO SQL
DETERMINISTIC
LANGUAGE C
EXTERNAL NAME 'CS!judt_tosql!judt_tosql.c!F!judt_tosql';
The following statement associates the tosql and fromsql transform routines with the judt
UDT.
CREATE TRANSFORM FOR judt
judt_io (TO SQL WITH SPECIFIC FUNCTION SYSUDTLIB.judt_tosql,
FROM SQL WITH SPECIFIC FUNCTION SYSUDTLIB.judt_fromsql);
Create the Ordering Functionality for the UDT

The following statements create a map ordering routine used to compare the judt values.
CREATE FUNCTION SYSUDTLIB.judt_order
(p1 judt)
RETURNS INTEGER
NO SQL
PARAMETER STYLE SQL
DETERMINISTIC
LANGUAGE C
EXTERNAL NAME 'CS!judt_order!judt_order.c!F!judt_order';
CREATE ORDERING FOR judt
ORDER FULL BY MAP WITH FUNCTION SYSUDTLIB.judt_order;
Create the Casting Functionality for the UDT

The following statements define the casting behavior to and from judt UDT and CLOB.
CREATE CAST (judt AS CLOB CHARACTER SET UNICODE)
WITH FUNCTION judt_fromsql(judt) AS ASSIGNMENT;
CREATE CAST (CLOB CHARACTER SET UNICODE AS judt)
WITH FUNCTION judt_tosql AS ASSIGNMENT;
C Source Files
This section shows the C code for the methods and functions created in the previous section.
This is just sample code so there is no meaningful logic in the tosql or Ordering functions.
However, based on the examples for the Constructor and fromsql routines, you can enhance
the previous routines to perform the necessary functions.
judt_constructor.c
#define SQL_TEXT Latin_Text
#include <sqltypes_td.h>
24
Teradata JSON

#include <string.h>
#include <stdio.h>
#define buffer_size 64000
void judt_cstr( UDT_HANDLE
*inUdt,
JSON_HANDLE
*file1,
UDT_HANDLE
*resultUdt,
char
sqlstate[6])
{
char trunc_err[6] = "25001";
int actualInputLength = 0;
BYTE input1[buffer_size] = {0};
int inputMaxLength = 0;
charset_et inputCharSet = 0;
int inNumLobs= 0;
FNC_GetJSONInfo(*file1,&inputMaxLength,&inputCharSet,&inNumLobs);
if (inNumLobs == 0)
{
FNC_GetInternalValue(*file1,input1,buffer_size, &actualInputLength);
FNC_SetStructuredAttribute(*resultUdt, "Att1", input1, 0, actualInputLength);
}
else
{
LOB_LOCATOR inLOB;
LOB_RESULT_LOCATOR outLob;
LOB_CONTEXT_ID id;
FNC_LobLength_t readlen, writelen;
int trunc_err = 0;
FNC_GetJSONInputLob(*file1,&inLOB);
FNC_GetStructuredResultLobAttribute(*resultUdt, "Att1", &outLob);
FNC_LobOpen(inLOB, &id, 0, 0);
actualInputLength = FNC_GetLobLength(inLOB);
while(FNC_LobRead(id, input1, buffer_size, &readlen) == 0
&& !trunc_err )
{
trunc_err = FNC_LobAppend(outLob, input1, readlen, &writelen);
}
}
}
judt_fromsql.c
#include <string.h>
void judt_fromsql(UDT_HANDLE
LOB_RESULT_LOCATOR
int
int
char
{
Teradata JSON
*udt,
*result,
*inNull,
*outNull,
sqlstate[6])
25

int nullIndicator,length;
BYTE temp[buffer_size];
if (*inNull != -1)
{
FNC_LobLength_t readlen, writelen;
attribute_info_t attrInfo;
FNC_GetStructuredAttributeInfo(*udt,0,sizeof(attrInfo), &attrInfo);
if (attrInfo.lob_length == 0)
{
FNC_GetStructuredAttribute(*udt, "Att1", temp, buffer_size, &nullIndicator,
&length);
readlen = length;
FNC_LobAppend(*result, temp, readlen, &writelen);
}
else
{
LOB_LOCATOR inLob;
LOB_CONTEXT_ID id;
int trunc_err = 0;
int remBufSize = buffer_size;
BYTE *input1Ptr = temp;
readlen = 0;
FNC_GetStructuredInputLobAttribute(*udt, "Att1", &nullIndicator, &inLob);
FNC_LobOpen(inLob, &id, 0, 0);
length = FNC_GetLobLength(inLob);
while(FNC_LobRead(id, temp, buffer_size, &readlen) == 0 && !trunc_err )
{
trunc_err = FNC_LobAppend(*result, temp, readlen, &writelen);
}
FNC_LobClose(id);
}
*outNull = 0;
}
else *outNull = -1;
}
judt_tosql.c
#include <string.h>
void judt_tosql (LOB_LOCATOR
*p1,
UDT_HANDLE
*result,
char
sqlstate[6])
{
/* Using the LOB FNC routines, read from 'p1' and load the data
into the JSON attribute, depending on its length. See judt_cstr() for
an example of loading the JSON attribute. */
}
26
Teradata JSON

FNC Library Routines That Support the JSON Type
judt_order.c
#include "sqltypes_td.h"
void judt_order
UDT_HANDLE
INTEGER
int
int
char
SQL_TEXT
SQL_TEXT
SQL_TEXT
(
*UDT,
*result,
*indicator_udt,
*indicator_result,
sqlstate[6],
extname[129],
specific_name[129],
error_message[257])
{
/* Read out as much data as necessary, using either
FNC_GetStructuredAttribute or FNC_GetStructuredInputLobAttribute + LOB
FNC routines, following the example in judt_fromsql. Then use this data
to make the determination about the value of this instance in terms of
ordering. */
}
Examples: Using the judt Type

The following shows uses of the newly created judt type.
CREATE TABLE judtTable(id INTEGER, j1 judt);
INSERT INTO judtTable(1, NEW judt(NEW JSON('{"name":"Cameron"}', UNICODE)));
INSERT INTO judtTable(2, NEW judt('{"name":"Melissa"}'));
SELECT * FROM judtTable ORDER BY 1;
Result:
id
----------1
2
j1
---------------------------------------------------------{"name":"Cameron"}
{"name":"Melissa"}
FNC Library Routines That Support the JSON

Type
When developing UDFs, UDMs, or external stored procedures that are defined with JSON
type parameters or return values, you can use the following JSON type interface functions to
access or set the values of the JSON type parameters, or get information about a JSON type
instance.
Teradata JSON
27

FNC Library Routines That Support the JSON Type
FNC Library Routine
Description
FNC_GetInternalValue
Retrieves the string representation of a JSON type instance that is used

as an input argument to an external routine. The string returned is
UNICODE or LATIN text depending on how the instance was defined.
You can use this function only when the JSON data is not stored as a
LOB.
FNC_GetJSONInfo
Retrieves the maximum length and character set of a JSON type

instance that is used as an input argument to an external routine. The
function also indicates whether or not the JSON data is stored as a LOB.
FNC_GetJSONInputLob
Returns a LOB_LOCATOR for a JSON instance which has its data

stored as a LOB. You can use this LOB_LOCATOR with LOB FNC
routines to read the data from the JSON instance.
FNC_GetJSONResultLob
Returns a LOB_RESULT_LOCATOR to a LOB associated with a JSON

instance. You can use this locator with LOB FNC routines to write data
to the JSON instance defined as a return value of an external routine.
FNC_SetInternalValue
Sets the string representation of a JSON type instance that is defined as

an output parameter or return value to an external routine. The string is
UNICODE or LATIN text depending on how the instance was defined.
You can use this function only when the JSON data will not be stored as
a LOB.
Use the JSON_HANDLE C data type to pass a JSON type instance as an argument to an
external routine. Similarly, use JSON_HANDLE to return a JSON type result from an
external routine. JSON_HANDLE is defined in sqltypes_td.h as: typedef int
JSON_HANDLE;
You can specify the JSON data type as an attribute of a structured UDT. When passing a
structured UDT that includes a JSON attribute to or from an external routine, you can use
the following interface functions to access or set the value of the JSON attribute, or get
information about a JSON attribute.
28
FNC Library Routine
Description
FNC_GetStructuredAttribute
Returns a string representation of a JSON

attribute. You can use this function only when the
data of the JSON attribute is not stored as a LOB.
FNC_GetStructuredAttributeInfo and
FNC_GetStructuredAttributeInfo_EON
Retrieves the following information about the

JSON attribute:
max_length = the maximum possible length
in bytes
charset_code = the character set
lob_length = the maximum possible length in
bytes if the data is stored as a LOB; otherwise,
0.
Teradata JSON

Restrictions for the JSON Type
FNC Library Routine
Description
FNC_GetStructuredInputLobAttribute
Returns a LOB_LOCATOR which represents the

LOB where the data of a JSON attribute is stored.
You can use this LOB_LOCATOR with the LOB
FNC routines to read the data from the JSON
attribute.
FNC_GetStructuredResultLobAttribute
Returns a LOB_RESULT_LOCATOR which

represents a LOB where the data of a JSON
attribute may be stored. You can use this
LOB_RESULT_LOCATOR with LOB FNC
routines to write the data to the JSON attribute.
FNC_SetStructuredAttribute
Sets the string representation of a JSON

document to a JSON attribute. You can use this
function only when the JSON data will not be
stored as a LOB.
For details about the JSON type interface functions, see SQL External Routine Programming,
B035-1147.
Restrictions for the JSON Type

You cannot specify the JSON type in the following situations:
As the data type for input or output parameters for UDFs or external stored
procedures written in Java
In user-defined table operators
As the base type of a distinct UDT
As the element type of an ARRAY type
You cannot create methods for the JSON type. Therefore, you cannot specify a JSON type
name in the FOR clause of a CREATE/REPLACE METHOD statement.
You cannot specify a JSON type column as part of an index.
You cannot use JSON type columns in clauses that depend on ordering or comparison,
such as ORDER BY, GROUP BY, or HAVING.
You cannot update the entities of a JSON instance. Therefore, you cannot use a JSON
entity reference in the target portion of a SET clause.
Although the external representation of JSON values is of the character string type, string
operations are not allowed directly on the JSON values. JSON can be serialized or cast to
generate a string representation before you apply the string operation.
Extracted JSON values can be used in arithmetic expressions. You can cast the extracted
JSON value to the appropriate SQL type to perform the computation.
Extracted JSON values can be used in relational comparison operations (such as >, <, or
=). You can cast the extracted JSON value to the appropriate SQL type to perform the
comparison.
JSON type columns cannot do the following.
Occur in queue tables
Participate in joins; however, portions of the JSON column can be joined and JSON
columns can be in the SELECT list
Teradata JSON
29

JSON String Syntax
JSON String Syntax

A string has valid JSON syntax if it conforms to the JSON standard syntax as specified in
http://www.json.org/. Syntax for the JSON type conforms to this standard, which is shown
below.
Syntax
object
array
object
}
name/value pair
,
name/value pair
name/value pair
"string "
value
value
"string "
number
object
array
null
true
false
array
]
value
,
value
Syntax Elements
object
An unordered collection of name/value pairs.
array
An ordered sequence of values.
name/value pair
An attribute and its associated value.
string
A JSON string literal enclosed in double quotation marks.
number
A JSON numeric literal.
30
Teradata JSON

Rules for JSON Data
Rules for JSON Data

JSON allows Unicode characters to be embedded in their hexadecimal formats in a
character string using the '\u' string as an escape sequence. This is allowed within the
JSON type, but the '\u' hex Unicode character is not interpreted; it is merely stored as is.
White space outside the root of the JSON object or array is trimmed for all instances of
the JSON type. White space characters within the root of the JSON object or array are
considered significant and are not removed.
There is a maximum limit of nesting imposed on a JSON instance. The limit is
configurable up to a maximum of 32000 using the JSON_MaxDepth DBS Control Field.
The default is 512.
Note: A nested object or array counts against this limit. For details about the
JSON_MaxDepth field, see DBS Control in Utilities, B035-1102.
You can specify exponential numbers using scientific notation. The range of valid
numbers is between -1e308 and 1e308, noninclusive. Exponents can have a value between
-308 and 308, noninclusive. Any number or exponent specified outside of the valid range,
whether explicitly stated or understood to be outside of that range based on the value of
the mantissa, will result in an error.
Teradata JSON
31

Rules for JSON Data
32
Teradata JSON
CHAPTER 2
Operations on the JSON Type
Creating a Table to Store JSON Data

You can create or alter a table containing one or more JSON type columns.
You can use the CREATE TABLE statement to create a table that contains one or more
JSON type columns.
Note: You cannot use a JSON type column in an index definition.
You can use the ALTER TABLE statement to add, drop, or rename a JSON type column.
You can specify the same CREATE TABLE or ALTER TABLE options that are permitted
on the UDT types on the JSON types.
You cannot use ALTER TABLE to change the maximum length of a JSON type.
Example: CREATE TABLE Statement

This example uses the CREATE TABLE statement to create the "temp_json" table containing
the JSON type column "json_data."
CREATE TABLE temp_json (id INTEGER,
json_data JSON(100));
Example: ALTER TABLE Statement

This example uses the ALTER TABLE statement to alter the "temp" table by adding the JSON
type column "json_data."
CREATE TABLE temp (id INTEGER);
ALTER TABLE temp ADD json_data JSON(50);
For details about CREATE TABLE and ALTER TABLE, see SQL Data Definition Language Syntax and Examples, B035-1144
Using Algorithmic Compression on JSON Columns

You can use the following compression functions to perform algorithmic compression
(ALC) on JSON type columns:
JSON_COMPRESS
JSON_DECOMPRESS
TD_LZ_COMPRESS
TD_LZ_DECOMPRESS
Teradata JSON
33
Chapter 2 Operations on the JSON Type

Storing JSON Data
You can use TD_LZ_COMPRESS to compress JSON data; however, Teradata recommends
that you use JSON_COMPRESS instead because the JSON_COMPRESS function is
optimized for compressing JSON data.
JSON_COMPRESS and JSON_DECOMPRESS can be used to compress JSON type columns
only. These functions cannot be used to compress columns of other data types.
You cannot create your own compression and decompression user-defined functions to
perform algorithmic compression on JSON type columns. You must use the functions
previously listed.
Note: Using ALC together with block-level compression (BLC) may degrade performance, so
this practice is not recommended. For more information on compression use cases and
examples, see Teradata Orange Book Block-Level Compression in Teradata. Access Orange
Books through Teradata @ Your Service: http://tays.teradata.com.
For more information about compression functions, see SQL Functions, Operators,
Expressions, and Predicates, B035-1145 .
For information about the COMPRESS and DECOMPRESS phrases, see SQL Data Types and
Literals, B035-1143 .
Example: JSON_COMPRESS and JSON_DECOMPRESS Functions
In this example, the JSON data in the "json_col" column is compressed using the
JSON_COMPRESS function. The compressed data is uncompressed using the
JSON_DECOMPRESS function.
CREATE TABLE temp (
id
INTEGER,
json_col JSON(1000)
CHARACTER SET LATIN
COMPRESS USING JSON_COMPRESS
DECOMPRESS USING JSON_DECOMPRESS);
Storing JSON Data

You can use the following ways to store JSON data.
Store JSON data in tables containing columns defined with the JSON data type.
Use JSON shredding to extract values from JSON documents and use the values to update
tables in the database.
Validating JSON Data

JSON data is automatically validated by default. You can use the DBS Control
DisableJSONValidation field or the SET SESSION JSON IGNORE ERRORS attribute to
disable validation if the data is from a reliable source.
You can also use the JSON_CHECK function to verify the JSON string is compliant with
JSON syntax. If the JSON string is valid, the JSON_CHECK function will return OK.
34
Teradata JSON

Storing JSON Data
Loading JSON Data Using Load Utilities

Loading JSON data is the same as loading LOB or XML data, and you use the "JSON"
keyword in the load script to define the column type. Like LOB data, you can load in inline
mode if the data is less than 64 KB in size (64000 LATIN characters or 32000 UNICODE
characters). Otherwise, you must specify the column as "JSON AS DEFERRED BY NAME"
(with the data residing in external files), where the client and Teradata Database make
multiple calls to move the data.
You can use the following load utilities for loading and unloading JSON data:
The Teradata Parallel Transporter (TPT) SQL Inserter operator supports the loading of
JSON data into a table.
The TPT SQL Selector operator supports the unloading of JSON data from a table.
The TPT Data Connector operator supports the writing and reading of JSON data to and
from a file.
The following shows how JSON syntax is used in the TPT schema definition:
DEFINE SCHEMA TEST_SCHEMA
DESCRIPTION 'PRODUCT INFORMATION SCHEMA'
(
COL1_INT
INTEGER,
COL2_CHAR
CHAR(10),
COL3_JSON
JSON(1000),
COL4_BIGINT
BIGINT,
COL5_JSON
JSON(16776192) AS DEFERRED BY NAME,
COL6_VARCHAR VARCHAR(20)
);
The following shows a sample USING clause generated by the TPT SQL Inserter operator:
USING COL1_INT(INTEGER),
COL2_CHAR(CHAR(10)),
COL3_JSON(CLOB(1000)),
COL4_BIGINT(BIGINT),
COL5_JSON(CLOB(16776192) AS DEFERRED),
COL6_VARCHAR(VARCHAR(20))
INSERT INTO target_table
VALUES (:COL1_INT, :COL2_CHAR, :COL3_JSON, :COL4_BIGINT,
:COL5_JSON, :COL6_VARCHAR);
You cannot load JSON data using FastLoad, MultiLoad, or FastExport protocols using either
the legacy stand-alone load tools or Parallel Transporter. However, if the JSON data is less
than 64 KB, and if the target table defines the column as CHAR or VARCHAR, then you can
load the JSON data using these utilities. Similarly, you can load the JSON data into a staging
table with the columns defined as CHAR or VARCHAR, and then use INSERT-SELECT to
load the data into JSON columns of a target table.
For more information about the TPT load utility, see the following books:
Teradata Parallel Transporter Reference, B035-2436
Teradata Parallel Transporter User Guide, B035-2445
Teradata JSON
35

Storing JSON Data
Inserting Values into a JSON Column

You can insert a value into a JSON column using one of the following statements.
INSERT. During the INSERT operation, a JSON column value may be initialized by an
expression which evaluates to the JSON type. Typical examples include the return value of
a UDF which has a JSON return type, the JSON type constructor, or a string that is cast to
a JSON type (in which the data conforms to the JSON string syntax).
In the following INSERT forms, you can use a JSON type constructor or an expression
which evaluates to a JSON type as the source value.
Positional list
Assignment list
Named list
INSERT...SELECT. During the INSERT...SELECT operation, the source table with a JSON
column whose actual data length is compatible with the JSON column of the target table
may be selected and used as the source value. If the source data is larger than the
maximum possible length of the target JSON column, an error is reported.
If the source JSON column is UNICODE and the target JSON column is LATIN, and the
source JSON instance contains a character that cannot be represented in LATIN, an error
is reported.
MERGE. You can use the MERGE statement to insert a value based on the WHEN NOT
MATCHED THEN INSERT clause. You can use the WHEN MATCHED THEN UPDATE
clause to modify a JSON column.
UPDATE (Upsert Form). You can use the UPDATE (Upsert Form) to update JSON
column values in a specified row and, if the row does not exist, it inserts the row into the
table with a specified set of initial column values.
Examples: Inserting Values into a JSON Column

Example: INSERT Statement
The example creates a table with a JSON column, allocates and initializes a JSON instance
using the JSON constructor, then inserts the JSON and integer values into the table.
CREATE TABLE my_table (eno INTEGER, edata JSON(100));
INSERT INTO my_table VALUES(1,
NEW JSON('{"name" : "Cameron", "age" : 24}'));
The example inserts a JSON string into a table that contains a JSON column.
'{"name" : "Cameron", "age" : 24}');
Note: If the string is not formatted correctly, an error will be reported.

'{"name" : "Cameron"');
*** Failure 7548: Syntax error in JSON string: expected a '}'.
36
Teradata JSON

Migrating Data to the JSON Type
Example: INSERT...SELECT Statement
The example creates two tables, then inserts JSON data into the second table from the first
table.
CREATE TABLE my_table2 (eno INTEGER, edata JSON(20));
NEW JSON('{"name" : "Cam"}'));
INSERT INTO my_Table2
SELECT * FROM my_table;
Note: If the JSON data is too large to fit in the column an error is reported.
NEW JSON('{"name" : "Cameron", "age" : 24}'));
INSERT INTO my_Table2
SELECT * FROM my_table;
*** Failure 7548: Data too large for this JSON instance.
Migrating Data to the JSON Type

You can migrate data to the JSON type following these steps:
1 (Optional) If converting from XML to JSON, the XML data must be stored in CLOB or
VARCHAR columns.
2 Verify that the intended JSON data is well-formed and conforms to the rules of JSON
formatting.
3 Create new versions of the tables using the JSON type for columns that will hold the
JSON data.
4 Insert the JSON text (for example, the JSON constructor or string) into the JSON
columns.
JSON cannot be loaded with FastLoad, MultiLoad, or FastExport due to restrictions on
all LOB types.
JSON Entity Reference

In Javascript, the native context for JSON, you can traverse JSON data using dot notation
and array indexing, depending on the structure of the JSON instance. This is called a JSON
entity reference, and you can use the following syntax to retrieve entities of a JSON instance.
Teradata JSON
37

Syntax
JSON_expr
Object Member
Array Element
Object Member
nonreserved_word
" string_literal "
Array Element
[ integer ]
[ integer ]
Syntax Elements
JSON_expr
An expression that evaluates to a JSON data type.
Object Member
A Teradata nonreserved word or a string literal enclosed in double quotation marks.
integer
A positive integer value that is within the range of the JSON instance being traversed.
Return Value
When the desired entity is found, the resulting data is a VARCHAR value. If more than one
entity is found which satisfies the specified syntax, the result is a VARCHAR representation
of a JSON array composed of the entities found.
The default length of the return value is 4096 characters, but you can use the
JSON_AttributeSize DBS Control field to change the default length. An error is returned if
the result exceeds the specified length. The maximum length that can be set for the return
value is 32000 characters.
For details about the JSON_AttributeSize DBS Control field, see Utilities, B035-1102.
If the entity is not found or if the JSON instance being referenced is null, a Teradata NULL
value is returned.
If the result is an empty string, an empty string is returned.
Restrictions
You cannot use a JSON entity reference in the target portion of a SET clause because you
cannot update the entities of a JSON instance.
38
Teradata JSON

Ambiguity Between a JSON Entity Reference and References to Other

Database Objects
The syntax for a JSON entity reference, a Teradata ARRAY type element reference, and a
fully or partially qualified column reference is similar. This creates potential ambiguity in the
following cases:
1. Ambiguity between a Teradata ARRAY type element reference and a JSON type array
reference
2. Ambiguity between a fully or partially qualified column reference and a JSON type dot
notation reference
When there is ambiguity, the default interpretation of the syntax is as follows:
For case 1: The syntax is interpreted as a Teradata ARRAY type element reference.
For case 2: The syntax is interpreted as a fully or partially qualified column reference.
Handling of these potential ambiguities is described in more detail in the following sections.
To ensure that a JSON entity reference is not ambiguous, you should fully qualify a JSON
column being referenced. Alternatively, you can use the JSONExtract, JSONExtractValue, or
JSONExtractLargeValue methods to retrieve entities of a JSON instance.
ARRAY Type Element Reference vs. JSON Type Array Reference

When parsing a request that features syntax matching both a Teradata ARRAY type element
reference and a JSON type array reference, the following logic is employed:
1. The syntax is first treated as an element reference on a Teradata ARRAY type.
2. If it is determined that the source is not an ARRAY type, Teradata checks to see if it is an
instance of the JSON type.
3. If it is a JSON type instance, Teradata interprets the syntax as a JSON type array
reference.
4. If it is not a JSON type instance, an error is returned indicating an incorrect array
element reference.
Column Reference vs. JSON Type Dot Notation Reference

When parsing a request that features syntax matching both a fully or partially qualified
column reference and a JSON type dot notation reference, the following logic is employed.
Column Reference with a Specified Table
The following logic is employed to differentiate between a standard column reference in the
format table.column and a reference to an entity in a JSON instance in the format
name1.name2.
1. If standard resolution is successful, the reference is interpreted as a standard
table.column reference.
2. Otherwise, if there is one source table with a JSON column named name1, the reference
is interpreted as a reference to an entity called name2 on a JSON column called name1.
If there is more than one source table with a JSON column named name1, an ambiguity
error is returned.
Teradata JSON
39

3. If there are no source tables with a JSON column named name1, an error is returned.
Column Reference with a Specified Table and Database
The following logic is employed to differentiate between a standard column reference in the
format database.table.column and a reference to an entity in a JSON instance in the format
name1.name2.name3...nameN.
1. If standard resolution is successful and there are more than 3 names, the reference is
interpreted as a reference to an entity called name4nameN on a JSON column called
name1.name2.name3. An error is returned if the column is not a JSON column.
If there are 3 names, the reference is interpreted as a standard database.table.column
reference.
Otherwise
if standard resolution is not successful, the standard disambiguation logic is
2.
used as follows:
a. If a source table named name1 exists and it has a JSON column named name2, the
reference is interpreted as a reference to an entity called name3nameN on a JSON
column called name1.name2. Otherwise, an error is returned indicating that column
name2 was not found.
If
there is one source table with a JSON column named name1, the reference is
b.
interpreted as a reference to an entity called name2nameN on a JSON column
called name1 that is present in one source table.
If there is more than one source table with a JSON column named name1, an
ambiguity error is returned.
If
there is a table in the current database named name1 with a JSON column named
c.
name2, the reference is interpreted as a reference to an entity called name3nameN
on a JSON column called CurrentDatabase.name1.name2.
d. Otherwise, processing continues with standard error handling.
Setting Up the JSON Entity Reference Examples

Create and populate table(s) to use in subsequent example(s).
CREATE TABLE test.jsonTable(id INTEGER, jsonCol JSON(100));
CREATE TABLE test.jsonTable2(id INTEGER, jsonCol JSON(100));
CREATE TABLE test.jsonTable3(id INTEGER, jsonLatCol JSON(100) CHARACTER SET LATIN);
INSERT INTO test.jsonTable(1, new JSON('{"name" : "Cameron", "numbers" : [1,2,3,
[1,2]]}'));
INSERT INTO test.jsonTable(2, new JSON('{"name" : "Cameron", "name" : "Lewis"}'));
INSERT INTO test.jsonTable(3,new JSON('{"name" : {"first" : "Cameron", "last" :
"Lewis"}}'));
INSERT INTO test.jsonTable2(1, new JSON('{"name" : "Cameron", "numbers" : [1,2,3,
[1,2]], "source" : "jsonTable2"}'));
INSERT INTO test.jsonTable3(1, new JSON('{"name" : "Cameron", "numbers" : [1,2,3,
[1,2]], "source" : "jsonTable3"}'));
40
Teradata JSON

Examples: Unambiguous JSON Entity Reference

The following shows unambiguous examples of JSON entity reference usage.
Example
SELECT jsonCol.name
FROM test.jsonTable
WHERE id=1;
Result:
jsonCol.name
-----------Cameron
Example
SELECT jsonCol.numbers
FROM test.jsonTable
WHERE id=1;
Result:
jsonCol.numbers
--------------[1, 2, 3, [1, 2]]
Example
SELECT jsonCol.numbers[1]
FROM test.jsonTable
WHERE id=1;
Result:
jsonCol.numbers[1]
-----------------2
Example
SELECT new JSON('{"name" : "Cameron"}').name;
Result:
new JSON('{"name" : "Cameron"}').name
------------------------------------Cameron
Example
SELECT jsonCol.name
FROM test.jsonTable
WHERE id=2;
Result:
Teradata JSON
41

jsonCol.name
-----------["Cameron", "Lewis"]
Example
SELECT id, jsonCol.numbers
FROM test.jsonTable
WHERE jsonCol.name='Cameron'
ORDER BY id;
Result:
id
jsonCol.numbers
-----------------------1
[1, 2, 3, [1, 2]]
Example
SELECT id, jsonCol.numbers
FROM test.jsonTable
WHERE id < 3
ORDER BY id;
Result:
id
jsonCol.numbers
-----------------------1
[1, 2, 3, [1, 2]]
2
?
/* There are no numbers in this JSON */
Examples: Ambiguous JSON Entity Reference

Example
In the following query, T.jsonCol and T.id are interpreted as column references.
SELECT T.id, T.jsonCol
FROM test.jsonTable T
WHERE id < 3
ORDER BY 1;
Result:
id
jsonCol
-------------1
{"name" : "Cameron", "numbers" : [1,2,3,[1,2]]}
2
{"name" : "Cameron", "name" : "Lewis"}
Example
In the following query, jsonCol.name is interpreted as a JSON entity reference.
SELECT jsonCol.name
FROM test.jsonTable
WHERE id=1;
Result:
42
Teradata JSON

id
jsonCol.name
------------------1
Cameron
Example
The following query returns an error because there is more than one source table with a
JSON column named jsonCol.
SELECT jsonCol.name
FROM test.jsonTable, test.jsonTable2;
Result:
*** Failure 3809 Column 'jsonCol' is ambiguous.
Example
The following query shows a JSON entity reference specified as a fully qualified column
reference.
SELECT id, test.jsonTable.jsonCol.name
FROM test.jsonTable
WHERE id=1;
Result:
id
jsonTable.jsonCol.name
---------------------------------1
Cameron
Example
The following shows an incorrect JSON entity reference specified as a fully qualified column
reference.
SELECT test.jsonTable.id.name
FROM test.jsonTable
WHERE id=1;
The query returns an error.

*** Failure 3706 Syntax error: Invalid use of JSON entity reference
syntax on non-JSON type.
Example
In the following query, jsonTable.jsonCol.name is a JSON entity reference that looks
like a database.table.column reference.
SELECT id, jsonTable.jsonCol.name
FROM test.jsonTable
WHERE id=1;
Result:
id
jsonTable.jsonCol.name
----------------------------1
Cameron
Teradata JSON
43

Example
Incorrect JSON entity reference
SELECT jsonTable.id.name
FROM test.jsonTable
WHERE id=1;
Result:
*** Failure 3802 Database 'jsonTable' does not exist.
Example
In the following query, jsonCol.name."first" is interpreted as an entity reference on the
jsonCol column of the source table, test.jsonTable.
SELECT T.id, jsonCol.name."first"
FROM test.jsonTable T, test.jsonTable3 T3
ORDER BY T.id;
Result:
id
jsonCol.name.first
------------------------1
?
2
?
3
Cameron
Example
In the following query, the reference to jsonCol is ambiguous because both source tables
have a JSON column named jsonCol.
SELECT T.id, jsonCol.name."first"
FROM test.jsonTable T, test.jsonTable2 T2
ORDER BY T.id;
The query returns an error.

*** Failure 3809 Column 'jsonCol' is ambiguous.
Example
In this example, jsonTable2 is in the current database and it has a JSON column called
jsonCol, so jsonTable2.jsonCol.name is interpreted as a JSON entity reference.
SELECT jsonTable2.id, jsonTable2.jsonCol.name
FROM test.jsonTable3;
Result:
id
jsonCol.name
-----------------------------1
Cameron
44
Teradata JSON

JSONPath Request Syntax

You can use JSONPath to request information about a particular portion of a JSON instance.
The entity desired can be any portion of a JSON instance, such as a name/value pair, an
object, an array, an array element, or a value. Also, several JSON functions and methods
accept a JSONPath formatted string as input and operate on the specified entity.
The principles of JSONPath are analogous to those of XPath for XML.
Teradata does not provide complete validation of the syntax so malformed query strings can
produce undesirable results. Teradata only provides the following validations:
Validate the presence of the '$' root character.
Validate the omission of the following characters:
Any ';' characters
Three or more consecutive '.' characters. For example: ''
Syntax
The following shows the JSONPath syntax as specified in the JSONPath Specification located
at http://goessner.net/articles/JsonPath/.
'
'
$
children
children
child specification
..
options
child specification
*
name_string
options
options
index
expression
filter
index
*
integer
:
, integer
: integer : integer
: integer
Teradata JSON
45

expression
( @.LENGTH
)
+
integer
*
/
filter
?(@.element_string
)
number comparison
=~ string
number comparison
<=
integer
<
>
>=
==
!=
Syntax Elements
$
The root object or element.
children
The descent operator ('.' or '.. ') followed by a child specification or options.
child specification
The wildcard character ('*') which matches all objects or elements.
A string specifying the name of a particular object or element and associated options
if needed.
options
An index, an expression, or a filter.
integer
A signed integer.
expression
In this context, LENGTH is the length of the current JSON array, equal to the number of
elements in the array.
@
The current object or element.
filter
Applies a filter (script) expression.
element_string
A string specifying the name of an element.
46
Teradata JSON

=~ string
String comparison expression.
Example: Elements of the JSONPath Syntax

This example uses the following JSON instance to illustrate particular elements of the
JSONPath syntax. The corresponding table provides explanations of the syntax elements and
usage examples.
{
"customer" : "CustomerName",
"orderID" : 3,
"price" : "$100000.00",
"items" :
[
{ "ID" : 1, "name" : "disk", "amt" : 10 },
{ "ID" : 2, "name" : "RAM", "amt" : 20 },
{ "ID" : 3, "name" : "monitor", "amt" : 30 },
{ "ID" : 4, "name" : "keyboard", "amt" : 40 },
{ "ID" : 5, "name" : "camera", "amt" : 50 },
{ "ID" : 6, "name" : "button", "amt" : 60 },
{ "ID" : 7, "name" : "mouse", "amt" : 70 },
{ "ID" : 8, "name" : "pen", "amt" : 80 }
]
}
JSONPath
Description
Example
Explanation of Example
Result
The root object/element
$.customer
The name of the customer
CustomerName
The current object/element $.items[(@.length The last item in the order. {"ID":8,"name":"pen","amt":
-1)]
80}
The use of the 'length'
keyword in this context is
interpreted as the length of
the current JSON array
and is treated as a property
of the JSON array. This is
only interpreted in this
manner if 'length' occurs
immediately after the '@.'
syntax. If the word 'length'
is found later in the
expression (for example,
'@.firstChild.length'), it is
interpreted as the name of
a child of some entity, not
as a property of that entity.
..
Recursive descent
Teradata JSON
$..name
All item names
["disk","RAM","monitor",
"keyboard","camera","butto
n","mouse","pen"]
47

Retrieving JSON Data Using SELECT
JSONPath
Description
Example
Explanation of Example
Wildcard
$.items[0].*
All descriptions of the first [1,"disk",10]

item of the order
All objects/elements
regardless of their names
Result
[]
The native array operator
$.items[0]
The first item
{"ID":1,"name":"disk","amt":
10}
[start,end]
List of indexes
$.items[0,1]
The first two items
[{"ID":
1,"name":"disk","amt":10},
{"ID":
2,"name":"RAM","amt":20}]
[start:end:step]
Array slice operator
$.items[0:4:2]
All items from 1-5 (not

[{"ID":
inclusive on the end index) 1,"name":"disk","amt":10},
by a step of 2 (That is,
{"ID":
items 1 and 3)
3,"name":"monitor","amt":
30}]
$.items[?
(@.amt<50)]
Filter all items of which a

quantity less than 50 was
ordered
If you do not specify start,

the default is the first
index.
If you do not specify end,
the default is the last index.
If you do not specify step,
the default is a step of 1.
?( )
Applies a filter (script)

expression
[{"ID":
1,"name":"disk","amt":10},
{"ID":
2,"name":"RAM","amt":20},
{"ID":
3,"name":"monitor","amt":
30},
{"ID":
4,"name":"keyboard","amt":
40}]
()
Script expression, using the $.items[(@.length The last item in the order
underlying script engine
-1)]
{"ID":8,"name":"pen","amt":
80}
Retrieving JSON Data Using SELECT

You can use the SELECT statement to retrieve data from JSON columns.
In field mode, the SELECT result is always a string, even if it is filtered using one of the JSON
methods.
If a string is not the desired format, you must cast the result set to the proper data type it
represents.
In non-field mode, the JSON type is returned to the client.
48
Teradata JSON

Extracting JSON Data Using SELECT and JSON Methods
Note: You cannot include a JSON column in the ORDER BY, HAVING or GROUP BY
clauses of a SELECT statement.
In the SELECT or WHERE clause, you can add a JSON entity reference to the end of a
column reference or any expression which evaluates to a JSON type.
Setting Up the SELECT Statement Example

INSERT INTO my_table (1, NEW JSON('{"name" : "Cameron", "phoneNumber" : 8584852612}'));
INSERT INTO my_table (2, NEW JSON('{"name" : "Justin", "phoneNumber" : 8584852611}'));
SELECT edata FROM my_table;
Result:
'{"name" : "Justin", "phoneNumber" : 8584852611}'
'{"name" : "Cameron", "phoneNumber" : 8584852612}'
Example: Retrieve JSON Data Using the SELECT Statement

The example uses the JSONExtractValue method to extract data where the name is
Cameron.
SELECT eno, edata
FROM my_table
WHERE edata.JSONExtractValue('$.name') = 'Cameron'
ORDER BY 1;
Result:
eno edata
-------1
'{"name" : "Cameron", "phoneNumber" : 8584852612}'
Extracting JSON Data Using SELECT and JSON

Methods
JSON data can be extracted from JSON documents stored in the Teradata database using the
SELECT statement or by using JSON methods: JSONExtract, JSONExtractValue,
JSONExtractLargeValue, JSON_TABLE, JSON_SHRED_BATCH, and
JSON_SHRED_BATCH_U.
To adhere to the proposed ANSI SQL/JSON standard, the extraction methods return one
value that matches the desired path. The first result found is returned, except if one of the
following elements is present in the JSONPath query string.
Recursive descent operation
Teradata JSON
49

Modifying JSON Columns
Wildcard
Name or index list
Index slice
Filter
Note: The search is optimized in that it does not always need to search the entire document.
However, this means that the following scenario is possible.
Disable JSON validation
Insert malformed JSON data. For example, something similar to this:
{"name":"Cameron" 123456}
Enable JSON validation

Attempt to extract the value of the key "name"
In this case, the correct result of "Cameron" is returned, because the result was found before
passing through the entire document and reaching the malformed data. If any malformed
data is encountered during the search, before the result is reached, a syntax error is reported.

You can use the following statements to modify a JSON column:
UPDATE. You can use the UPDATE statement to modify values of a JSON column in the
SET clause.
Note: You can only update the JSON column, not individual portions of the JSON
instance. Also, when referencing a JSON column in its entirety, the input format is the
same as the INSERT statement.
MERGE. You can use the MERGE statement to:
Modify a JSON column based on the WHEN MATCHED THEN UPDATE clause.
Insert values into a JSON column based on the WHEN NOT MATCHED THEN
INSERT clause.
Note: The following MERGE statement clauses are supported on JSON functions and
methods:
The ON clause can be used to condition on a JSON instance.
The USING clause can be used to select the JSON column value.
Setting Up the Update Examples

INSERT INTO my_table (1, NEW JSON('{"name" : "Cameron", "age" : 24}'));
Example: UPDATE and SET

The following statement sets the edata column value and updates the table.
50
Teradata JSON

UPDATE my_table
SET edata = NEW JSON('{"name" : "Justin"}')
WHERE edata.JSONExtractValue('$.name') = 'Cameron';
Result: To see the result, run: SELECT edata FROM my_table;

edata
----{"name" : "Justin"}
Example: UPDATE Column Value

The following statement sets the edata column value and updates the table.
UPDATE my_table
SET edata = '{"name" : "George"}';
Result: To see the name has been updated, run: SELECT edata FROM my_table;
edata
----{"name" : "George"}
Setting Up the Merge Examples

INSERT INTO my_table(1,'{"name" : "Justin", "phoneNumber" :
8584852611}');
INSERT INTO my_table(2,'{"name" : "Cameron", "phoneNumber" :
8584852612}');
CREATE TABLE my_table2 (eno INTEGER, edata JSON(100));
INSERT INTO my_table2(1,'{"age":24}');
INSERT INTO my_table2(3,'{"age":34}');
Example: MERGE When Matched Then Update

The example shows how to use the MERGE WHEN MATCHED THEN UPDATE clause.
MERGE INTO my_table
USING (SELECT eno, edata FROM my_table2) AS S(a,b)
ON eno = S.a AND
CAST (edata.JSONExtractValue('$.name') AS VARCHAR(20)) = 'Justin'
WHEN MATCHED THEN UPDATE
SET edata = '{"name" : "Kevin", "phoneNumber" : 8584852613}';
To see the result of the query, run: SELECT * FROM my_table ORDER BY eno;
eno edata
--------1
{"name" : "Kevin", "phoneNumber" : 8584852613}
2
{"name" : "Cameron", "phoneNumber" : 8584852612}
Teradata JSON
51

Using the JSON Type in a DELETE or ABORT Statement
Example: MERGE When Not Matched Then Insert

The example shows how to use the MERGE WHEN NOT MATCHED THEN INSERT
clause.
MERGE INTO my_table
USING (SELECT eno, edata FROM my_table2) AS S(a,b)
ON eno = S.a AND
CAST (edata.JSONExtractValue('$.name') AS VARCHAR(20)) = 'Mike'
WHEN NOT MATCHED THEN INSERT (eno, edata)
VALUES (S.a, NEW JSON('{"name" : "Mike", "phoneNumber" :
8584852614}'));
To see the result of the query, run: SELECT * FROM my_table ORDER BY eno;
eno edata
----------1
{"name" : "Kevin", "phoneNumber" : 8584852613}
2
{"name" : "Cameron", "phoneNumber" : 8584852612}
2
{"name" : "Mike", "phoneNumber" : 8584852614}
Using the JSON Type in a DELETE or ABORT

Statement
You can use the DELETE and ABORT statements with the JSON types in the WHERE clause.
These types can be cast to a predefined type that can have relational comparisons performed
on it, or these types can make use of the system defined functions or methods to isolate
individual portions for comparison.
Setting Up the DELETE Example

INSERT INTO my_table(1, '{"name":"Cameron","age":24}');
Example: Using the DELETE Statement

Selectively delete data that meets the criteria.
DELETE my_table WHERE CAST (edata.JSONExtractValue('$.age') AS
INTEGER) = 24;
To see the result of the DELETE, run: SELECT * FROM my_table;

*** No rows found
Setting Up the ABORT Example

52
Teradata JSON

Creating a Join Index Using Portions of a JSON Instance
INSERT INTO my_table(1, '{"name":"Cameron","age":24}');
Example: Using the ABORT Statement

Abort on a portion of the JSON instance.
ABORT 'JSON Abort'
FROM my_table
WHERE CAST (edata.JSONExtractValue('$.age') AS INTEGER) = 24;
Result:
*** Failure 3513 JSON Abort.
Creating a Join Index Using Portions of a JSON

Instance
You can create a join index that contains extracted portions of a JSON instance.
Use the following to extract portions of the JSON data.
JSON Entity Reference syntax
JSONExtractValue or JSONExtractLargeValue methods
Setting Up Join Index Examples

CREATE TABLE json_table(id INTEGER, j JSON(100));
INSERT INTO json_table(1, NEW JSON('{"name":"Cameron", "age":25}'));
INSERT INTO json_table(2, NEW JSON('{"name":"Melissa", "age":24}'));
Example: Use JOIN INDEX with a JSON Entity Reference

The example shows how to use JOIN INDEX with portions of a JSON instance specified in a
JSON Entity Reference.
CREATE JOIN INDEX json_table_ji AS
SELECT j.name AS Name,
j.age AS Age
FROM json_table;
Example: Use JOIN INDEX with JSONExtractValue

The example creates a JOIN INDEX with portions of a JSON instance extracted by
JSONExtractValue.
Teradata JSON
53

Collecting Statistics on Portions of a JSON Instance
CREATE JOIN INDEX json_table_ji AS
SELECT j.JSONExtractValue('$.name') AS Name,
j.JSONExtractValue('$.age') AS Age
FROM json_table;
Collecting Statistics on Portions of a JSON

Instance
You can collect statistics on extracted portions of the JSON type using the JSON Entity
Reference syntax or the JSONExtractValue method.
Note: You cannot collect statistics on an entire JSON instance.
Setting Up Collect Statistics Examples

CREATE TABLE json_table(id INTEGER, j JSON(100));
INSERT INTO json_table(1, NEW JSON('{"name":"Cameron", "age":25}'));
INSERT INTO json_table(2, NEW JSON('{"name":"Melissa", "age":24}'));
Example: Use COLLECT STATISTICS with a JSON Entity Reference

The example shows how to use COLLECT STATISTICS with portions of a JSON instance
specified in a JSON Entity Reference.
COLLECT STATISTICS COLUMN j.name AS name_stats ON json_table;
Example: Use COLLECT STATISTICS with JSONExtractValue

The example uses COLLECT STATISTICS with portions of a JSON instance extracted by
JSONExtractValue.
COLLECT STATISTICS COLUMN j.JSONExtractValue('$.name') AS name_stats
ON json_table;
JSON Methods, Functions, External Stored

Procedures, and Table Operators
You can perform the following common operations on the JSON type to access or
manipulate JSON data.
54
Teradata JSON

JSON Methods, Functions, External Stored Procedures, and Table Operators
Methods
Combine: This method takes two JSON documents and combines them into a JSON
document structured as array or JSON document structured as an object.
ExistValue: This method allows you to specify a name or path in JSONPath syntax to
determine if that name or path exists in the JSON document.
JSONExtract: The JSONExtract method extracts data from a JSON instance. The desired
data is specified in a JSONPath expression. The result is a JSON array composed of the
values found, or NULL if there are no matches.
JSONExtractValue: This method allows you to retrieve the text representation of the
value of an entity in a JSON instance, specified using JSONPath syntax.
JSONExtractValue returns a VARCHAR. The returned VARCHAR length defaults to 4K,
but this can be increased to 32000 characters (not bytes) in DBS Control using the
JSON_AttributeSize flag.
JSONExtractLargeValue: This method is the same as JSONExtractValue, except for the
return size and type. For LOB based JSON objects this method returns a CLOB of
16776192 characters for CHARACTER SET LATIN or 8388096 characters for
CHARACTER SET UNICODE.
Functions
ARRAY_TO_JSON: This function allows any Teradata ARRAY type to be converted to a
JSON type composed of an array.
GeoJSONFromGeom: This function converts an ST_Geometry object into a JSON
document that conforms to the GeoJSON standards.
GeomFromGeoJSON: This function converts a JSON document that conforms to the
GeoJSON standards into an ST_Geometry object.
JSON_AGG: This aggregate function takes a variable number of input parameters and
packages them into a JSON document.
JSON_COMPOSE: This scalar function takes a variable number of input parameters and
packages them into a JSON document. This function provides a complex composition of
a JSON document when used in conjunction with the JSON_AGG function.
JSON_CHECK: This function checks a string for valid JSON syntax and provides an
informative error message about the cause of the syntax failure if the string is invalid.
Table Operators
JSON_KEYS: This table operator parses a JSON instance, from either CHAR or
VARCHAR input and returns a list of key names.
JSON_TABLE: This table operator creates a temporary table based on all, or a subset, of
the data in a JSON object.
SQL Stored Procedures

JSON_SHRED_BATCH and JSON_SHRED_BATCH_U are SQL stored procedures that
allow you to extract values from JSON documents and use the extracted data to populate
existing relational tables. This provides a flexible form of loading data from JSON format
into the relational model.
Teradata JSON
55

JSON Methods, Functions, External Stored Procedures, and Table Operators
56
Teradata JSON
CHAPTER 3
JSON Methods
Combine
Purpose
The Combine method takes two JSON documents (specified by the JSON expression),
combines them, and returns a JSON document structured as an array or structured as an
object.
Syntax
JSON_expr.Combine ( JSON_expr
)
, 'ARRAY'
, 'OBJECT'
Syntax Elements
JSON_expr
An expression that evaluates to a JSON data type; for example, this can be a column in
a table, a constant, or the result of a function or method.
'ARRAY' or 'OBJECT'
Optional. Explicitly specifies the result type as 'ARRAY' or 'OBJECT'.
If 'ARRAY' is specified, the result is a JSON document structured as an array. If
'OBJECT' is specified the result is a JSON document structured as an object.
When specifying 'OBJECT' both the JSON documents being combined must be JSON
objects, otherwise an error is reported.

An error occurs if the result of the combination is too large.
The JSON type has a default maximum length. If no maximum length is specified, the
default maximum length for the character set is chosen. If specified, the length is subject to a
minimum of two characters and cannot be greater than the absolute maximum for the
character set. If the character set is not specified, the character set of the user is used.
The maximum length of the JSON text depends on the character set being used:
If the character set is UNICODE, the maximum length is 8388096 characters. This is
Teradata JSON
57
Chapter 3 JSON Methods

Combine
If the character set is LATIN, the maximum length is 16776192 characters. This is
A shorter maximum length can be specified wherever the JSON type is used, for example, as
the data type of a table column or as a parameter to a function.
If the result of the Combine method is used to insert into a column of a table or as a
parameter to a UDF, UDM, external stored procedure (XSP), Teradata Stored Procedure
(TDSP), and so on, the resulting length is subject to the length of that defined JSON instance.
If it is too large, an error will result.
Usage Notes
The result of a combining two JSON documents is a JSON document containing data from
both the input documents. The resulting document is structured as a JSON object or a JSON
array. The JSON documents being combined can be structured differently than each other;
for example, a JSON document structured as an array can be combined with a JSON
document structured as an object and the resulting combination is structured as a JSON
array. The following explains the result of various combinations.
If the optional 'ARRAY' parameter is specified in the command:
If both input JSON documents are structured as JSON arrays, the result is an array
with values from each JSON document. For example, if j1 = [1,2] and j2 = [3,4], the
combination is [1,2,3,4].
If one of the JSON documents is structured as a JSON array and the other is structured
as a JSON object, the result is a JSON array composed of all the elements of the JSON
document structured as an array, plus one more element, which is the entire JSON
document structured as an object (the second document). For example, if j1 = [1,2]
and j2 = {"name" : "Jane"}, the combination is [1,2, { "name" : "Jane" } ].
If both JSON documents are structured as JSON objects, the result is a JSON
document structured as an array composed of JSON documents structured as objects
from each JSON object being combined. For example, if j1 = { "name" : "Harry" } and j2
= { "name" : "Jane" }, the combination is [ { "name" : "Harry" }, {"name" : "Jane"} ].
If the optional 'OBJECT' parameter is specified, the result is a combined JSON document
structured as an object containing all the members of each input object. For example, if j1
= { "name" : "Jane" , "age" : "30" } and j2 = { "name" : "Harry", "age" : "41" }, the combination
is { "name" : "Jane" , "age" : "30" , "name" : "Harry" , "age" : "41" }.
If either JSON document being combined is structured as an array an error is reported.
If 'ARRAY' or 'OBJECT' is not specified in the command:
If both JSON documents are structured as JSON arrays, the result is a JSON document
structured as an array with values from each JSON document.
If either JSON document is structured as an array, the result is a JSON document
structured as an array, which is composed of all the elements of the JSON document
which is an array, plus one more element which is the entire JSON document
structured as an object; this is the same behavior as specifying 'ARRAY'.
If both JSON documents are structured as objects, the result is a combined JSON
document structured as an object containing all the members of each input object; this
is the same behavior as specifying 'OBJECT'.
If one of the JSON documents being combined is NULL, the result is a non-NULL JSON
document.
If both JSON documents being combined are NULL, the result is a Teradata NULL value.
58
Teradata JSON

Combine
If one of the JSON documents being combined has CHARACTER SET UNICODE, the
resulting JSON instance has text in CHARACTER SET UNICODE.
If both JSON documents being combined has text in CHARACTER SET LATIN, the
resulting JSON instance has text in CHARACTER SET LATIN.
Setting Up the Combine Examples

Create and populate two tables containing JSON objects and INTEGER data. The examples
in the following sections refer to this table and data.
CREATE TABLE my_table (eno INTEGER,
edata JSON(100) CHARACTER SET UNICODE,
edata2 JSON(100) CHARACTER SET LATIN);
CREATE TABLE my_table2 (eno INTEGER,
edata JSON(100) CHARACTER SET UNICODE,
edata2 JSON(100) CHARACTER SET LATIN);
INSERT INTO my_table (1, NEW JSON('{"name" : "Cameron"}'), NEW
JSON('{"name" : "Lewis"}'));
INSERT INTO my_table (2, NEW JSON('{"name" : "Melissa"}'), NEW
INSERT INTO my_table (3, NEW JSON('[1,2,3]'), NEW JSON('{"name" :
"Lewis"}'));
INSERT INTO my_table2 (1, NEW JSON('{"name" : "Cameron"}'), NEW
INSERT INTO my_table2 (2, NEW JSON('{"name" : "Melissa"}'), NEW
Example: UNICODE Combine LATIN

The example shows how to use the JSON Combine method to combine two JSON objects,
one of CHARACTER SET UNICODE and the other of CHARACTER SET LATIN, which
results in a JSON instance with text in CHARACTER SET UNICODE. This example also
uses the 'OBJECT' parameter to specify a JSON object as the result.
Note: The example uses the table(s) created earlier.
/* Explicit statement of result type as JSON 'OBJECT' */
SELECT edata.Combine(edata2, 'OBJECT') FROM my_table WHERE eno < 3;
Result: When combining LATIN and UNICODE, the result is a JSON instance with text in
edata.Combine(edata2)
--------------------{"name" : "Cameron", "name" : "Lewis"}
{"name" : "Melissa", "name" : "Lewis"}
Teradata JSON
59

Combine
Example: UNICODE Combine UNICODE

The example shows the result when two JSON objects are combined with both JSON objects
having CHARACTER SET UNICODE, which results in a JSON instance with text in
/* Result of UNICODE combine UNICODE is UNICODE */
SELECT MA.edata.Combine(MB.edata)
FROM my_table AS MA, my_table2 AS MB
WHERE MA.eno < 3;
Result: When combining UNICODE and UNICODE, the result is a JSON instance with text
in CHARACTER SET UNICODE.
edata.Combine(edata)
-------------------{"name" : "Cameron","name"
{"name" : "Cameron","name"
{"name" : "Melissa","name"
{"name" : "Melissa","name"
:
:
:
:
"Cameron"}
"Melissa"}
"Cameron"}
"Melissa"}
Example: LATIN Combine LATIN

The example shows how to use the JSON Combine method to combine two JSON objects of
CHARACTER SET LATIN.
/* Result of LATIN combine LATIN is LATIN. */
SELECT MA.edata2.Combine(MB.edata2)
WHERE MA.eno < 3;
Result: The result is a JSON instance with text in CHARACTER SET LATIN.
edata2.Combine(edata2)
--------------------{"name" : "Lewis","name"
{"name" : "Lewis","name"
:
:
:
:
"Lewis"}
"Lewis"}
"Lewis"}
"Lewis"}
Example: Result Is Implicitly a JSON ARRAY

The example shows a JSON array combined with a JSON object; the result type is not
specified.
60
Teradata JSON

Combine
/* Result is implicitly a JSON ARRAY */
SELECT edata.Combine(edata2) FROM my_table WHERE eno=3;
Result: The result of combining a JSON array with a JSON object is implicitly a JSON
ARRAY.
--------------------[1,2,3, {"name" : "Lewis"}]
Example: Explicit Statement of Result Type as JSON ARRAY

The example shows how to use the JSON Combine method with the 'ARRAY' parameter to
explicitly set the result to a JSON array.
/* Explicit statement of result type as JSON ARRAY */
SELECT MA.edata2.Combine(MB.edata2, 'ARRAY')
WHERE MA.eno < 3;
Result: The result of using the 'ARRAY' parameter is a JSON array.

edata2.Combine(edata2)
---------------------[{"name" : "Lewis"},{"name"
[{"name" : "Lewis"},{"name"
:
:
:
:
"Lewis"}]
"Lewis"}]
"Lewis"}]
"Lewis"}]
Example: Combine Two JSON ARRAYs and Explicitly State the Result Is a
JSON ARRAY
The following examples show how to use the JSON Combine method with the 'ARRAY'
parameter to explicitly set the result to a JSON array.
The example specifies the 'ARRAY' parameter. Note, the data being combined are both JSON
array instances, so the result is implicitly a JSON ARRAY even if the 'ARRAY' parameter is
not specified.
/* Explicit statement of result type as JSON ARRAY. */
SELECT edata.Combine(edata, 'ARRAY') FROM my_table WHERE eno=3;
Result: The result is an array of values.

edata.Combine(edata)
[1,2,3,1,2,3]
Teradata JSON
61

ExistValue
Example: Combine a JSON ARRAY and a JSON Expression and Specify the
Result Is a JSON ARRAY
Combine a JSON array and an expression that evaluates to a JSON array and specify
'ARRAY' as the result. Note, the result is implicitly a JSON array even if the 'ARRAY'
parameter is not specified.
/* Explicit statement of result type as JSON 'ARRAY' */
SELECT edata.Combine(NEW JSON('[1,2,[3,4]]'), 'ARRAY') FROM my_table
WHERE eno=3;
Result: The result is an array of several elements, with the last element an array itself.
[1,2,3,1,2,[3,4]]
Example: Error - Combine Two JSON ARRAYs and State the Result Is a
JSON OBJECT
The example specifies the 'OBJECT' parameter; however, the data being combined are both
JSON array instances, so the result must be a JSON ARRAY. This example results in an error
because 'OBJECT' was specified.
/*Error case*/
SELECT edata.Combine(edata2, 'OBJECT') FROM my_table WHERE eno=3;
The example results in an error similar to this: *** Failure 7548: Invalid options
for JSON Combine method.
ExistValue
Purpose
The ExistValue method determines if a name represented by a JSONPath-formatted string
exists in a JSON instance.
Syntax
JSON_expr.ExistValue (JSONPath_expr )
62
Teradata JSON

ExistValue
Syntax Elements
JSON_expr
JSONPath_expr
A name in JSONPath syntax.
The name can be either UNICODE or LATIN, depending on the character set of the
JSON type that invoked this method. If the parameter character set does not match the
character set of the JSON type, Teradata attempts to translate the parameter character
set to the correct character set.
JSONPath_expr cannot be NULL. If the expression is NULL an error is reported.
Functional Description
ExistValue determines if the name specified by JSONPath_expr exists in the JSON instance
specified by JSON_expr.
Return Value
1, if the specified name is found at least once in the JSON instance.
0, if the name is not found.
NULL, if the JSON_expr argument is null.
Setting Up the ExistValue Examples

INSERT INTO my_table (1, NEW JSON('{"name" : "Cameron", "age" : 24,
"schools" : [ {"name" : "Lake", "type" : "elementary"}, {"name" :
"Madison", "type" : "middle"}, {"name" : "Rancho", "type" : "high"},
{"name" : "UCI", "type" : "college"} ] }'));
Example: Use ExistValue to Check for a Specific Child

The following query results in 'True' if the JSON column contains a child named 'schools'
which is an array, and the second element of the array has a child named 'type'.
SELECT eno, 'True'
FROM my_table
WHERE edata.ExistValue('$.schools[1].type') = 1;
Result:
ENO
'True'
-----------------1
True
Teradata JSON
63

Comparison of JSONExtract and JSONExtractValue
Example: ExistValue Does Not Find the Specified Child

The following query results in 'True' if the JSON column contains a child named 'schools'
which is an array, and the second element of the array has a child named 'location'.
SELECT eno, 'True'
FROM my_table
WHERE edata.ExistValue('$.schools[1].location') = 1;
Result:
*** No rows found ***
The JSON column contains a child named 'schools' which is an array, and the second element
of the array does not have a child named 'location'; therefore, the method did not return any
rows.
Comparison of JSONExtract and

JSONExtractValue
The JSONExtract, JSONExtractValue, and JSONExtractLargeValue methods extract values
from a JSON instance.
JSONExtract
A built-in method of the JSON type that returns an array containing values specified
by the JSONPath query string.
JSONExtractValue
A built-in method of the JSON type that allows a user to retrieve the text
representation (VARCHAR) of the value of an entity in a JSON object specified using
JSONPath syntax.
JSONExtractLargeValue
A built-in method of the JSON type. JSONExtractLargeValue is the same as
JSONExtractValue, except for the return type and size. This method returns a CLOB.
The return types of the methods are different.
JSONExtract results in a JSON array of the values found in the JSON instance.
JSONExtractValue returns VARCHAR, unless the search finds two or more paths that
meet the criteria, then the result is a JSON array formatted as a string. The string can be
cast to other appropriate data types.
By default, the return value is VARCHAR(4096) of the desired character set; however, the
size is tunable in JSON_AttributeSize in DBS Control. If the result cannot be contained in
the return length of this method (even after increasing the return length to the maximum
of 32000 characters), users should invoke the JSONExtractLargeValue method which
functions identically to JSONExtractValue, but returns a CLOB.
64
Teradata JSON

JSONExtract
JSONExtractLargeValue returns a CLOB of 16776192 characters for CHARACTER SET

LATIN or 8388096 characters for CHARACTER SET UNICODE, depending on the
character set of the JSON that invoked it.
All methods can result in NULL.
The reason to choose one method over the other depends on how you want to use the result.
For example, if you want to extract one value from a JSON instance and treat the value as
some predefined type which is castable from VARCHAR, use JSONExtractValue. If you want
to extract a nested JSON object/array and do further processing on the data, treating it as a
JSON object, use JSONExtract.
JSONExtract
Purpose
The JSONExtract method operates on a JSON instance, to extract data identified by the
JSONPath formatted string. If one or more entities are found, the result of this method is a
JSON array composed of the values found; otherwise, NULL is returned.
Syntax
JSON_expr. JSONExtract ( JSONPath_expr )
Syntax Elements
JSON_expr
JSONPath_expr
An expression to extract information about a particular portion of a JSON instance. For
example, $.employees.info[*] provides all the information about each employee.
The desired information can be any portion of a JSON instance; for example, a name/
value pair, object, array, array element, or a value.
The JSONPath expression must be in JSONPath syntax.
JSONPath_expr cannot be NULL. If the expression is NULL, an error is reported.
JSONExtract searches the JSON object specified by JSON_expr and retrieves the data that
matches the entity name specified by JSONPath_expr.
Return Value
A JSON array whose elements are all the matches for the JSONPath_expr in the JSON
instance.
NULL, if the entity was not found in the JSON object.
NULL, if the JSON_expr argument is null.
Teradata JSON
65

JSONExtract
If the JSONPath_expr is null, an error is returned.

The return value can be either UNICODE or LATIN, depending on the character set of the
JSON type that invoked this method. If the parameter character set does not match the
character set of the JSON type, Teradata attempts to translate the parameter character set to
the correct character set.
Setting Up the JSONExtract Examples

INSERT INTO my_table (1, NEW JSON('[{"firstName" :
"age" : 24, "schools" : [ {"name" : "Lake", "type"
"Madison", "type" : "middle"}, {"name" : "Rancho",
"UCI", "type" : "college"} ], "job" : "programmer"
"Cameron", "lastName" : "Lewis",

: "elementary"}, {"name" :
"type" : "high"}, {"name" :
}]'));
INSERT INTO my_table (2, NEW JSON('[{"firstName" : "Melissa", "lastName" : "Lewis",

"age" : 23, "schools" : [ {"name" : "Lake", "type" : "elementary"}, {"name" :
"Madison", "type" : "middle"}, {"name" : "Rancho", "type" : "high"}, {"name" : "Mira
Costa", "type" : "college"} ] }]'));
INSERT INTO my_table (3, NEW JSON('[{"firstName" : "Alex", "lastName" : "Smoley",
"Madison", "type" : "middle"}, {"name" : "Rancho", "type" : "high"}, {"name" :
"CSUSM", "type" : "college"} ], "job" : "CPA" }]'));
INSERT INTO my_table (4, NEW JSON('[{"firstName" : "David", "lastName" : "Perpich",
"Madison", "type" : "middle"}, {"name" : "Rancho", "type" : "high"} ], "job" :
"small business owner"}]'));
Example: JSONExtract with Several Results

The example uses the JSONExtract method to extract unfiltered results from a table. This
example shows multiple results being returned.
/* JSONExtract with several results - get the name of everyone. */
SELECT eno, edata.JSONExtract('$..firstName')
FROM my_Table
ORDER BY 1;
Result: The result(s) are arrays of values.

ENO edata.JSONExtract()
-------------------------1 [ "Cameron" ]
2 [ "Melissa" ]
66
Teradata JSON

JSONExtract
3
4
[ "Alex" ]
[ "David" ]
Example: JSONExtract with Filtered Results

The example uses the JSONExtract method to extract filtered results from a table.
/* JSONExtract with filtered results - get the name of everyone older
than 23. */
SELECT eno, edata.JSONExtract('$.[?(@.age > 23)].firstName')
FROM my_table
ORDER BY 1;
Result:
--------------------------1 [ "Cameron" ]
2 ?
3 [ "Alex" ]
4 [ "David" ]
Example: JSONExtract with NULL Results

The example uses the JSONExtract method to extract NULL results from a table.
/* JSONExtract with some NULL results get everyone's job. */
SELECT eno, edata.JSONExtract('$..job')
FROM my_table
ORDER BY 1;
Result: A NULL value is returned for a person in the table who does not have a job.
-------------------------1 [ "programmer" ]
2 ?
3 [ "CPA" ]
4 [ "small business owner" ]
Teradata JSON
67

JSONExtractValue and JSONExtractLargeValue

Purpose
The JSONExtractValue and JSONExtractLargeValue methods retrieve the text representation
of the value of an entity in a JSON instance. JSONExtractLargeValue takes the same input
parameters and operates the same as JSONExtractValue; however, the return type and size
are different.
Syntax
JSON_expr.JSONExtractValue
(JSONPath_expr )
JSON_expr.JSONExtractLargeValue
Syntax Elements
JSON_expr
JSONPath_expr
An expression to extract information about a particular portion of a JSON instance. For
example, $.employees.info[*] provides all the information about each employee.
The desired information can be any portion of a JSON instance; for example, a name/
value pair, object, array, array element, or a value.
The JSONPath expression must be in JSONPath syntax.
JSONPath_expr cannot be NULL. If the expression is NULL, an error is reported.
These methods search the JSON object specified by JSON_expr and get the value of the entity
name specified by JSONPath_expr. The entity name is represented by a JSONPath formatted
string.
Return Value
A string that is the value of the entity, if the entity was found in the JSON instance.
A JSON array, formatted as a string, if the result of the search found two or more paths
that meet the search criteria specified by JSONPath_expr. Each element of the array
represents one of the values found.
An empty string if the result is an empty string.
A Teradata NULL, if the result is a JSON null.
A Teradata NULL, if the entity was not found in the JSON object.
A Teradata NULL, if the JSON_expr argument is null.
JSONExtractValue returns a VARCHAR of the desired attribute. The returned length defaults
to 4K, but this can be increased to 32000 characters (not bytes) in DBS Control using the
JSON_AttributeSize flag. If the result of the method is too large for the buffer, an error will be
reported.
68
Teradata JSON

JSONExtractLargeValue extracts a value up to the maximum size of the type. It returns a

CLOB of 16776192 characters for CHARACTER SET LATIN or 8388096 characters for
The return value of both methods can be UNICODE or LATIN, depending on the character
set of the JSON type that invoked it. If the parameter character set does not match the
character set of the JSON type, Teradata attempts to translate the parameter character set to
the correct character set.
You can cast the return string to any data type that supports casting.
Usage
JSONExtractLargeValue should only be used when the results of the extraction are large
(greater than 32000 characters); otherwise use JSONExtractValue.
Setting Up the JSONExtractValue Examples

INSERT INTO my_table (1, NEW JSON('[{"firstName" :
"age" : 24, "schools" : [ {"name" : "Lake", "type"
"Madison", "type" : "middle"}, {"name" : "Rancho",
"UCI", "type" : "college"} ], "job" : "programmer"
"Cameron", "lastName" : "Lewis",

: "elementary"}, {"name" :
"type" : "high"}, {"name" :
}]'));
INSERT INTO my_table (2, NEW JSON('[{"firstName" : "Melissa", "lastName" : "Lewis",

"Mira Costa", "type" : "college"} ] }]'));
INSERT INTO my_table (3, NEW JSON('[{"firstName" : "Alex", "lastName" : "Smoley",
"CSUSM", "type" : "college"} ], "job" : "CPA" }]'));
INSERT INTO my_table (4, NEW JSON('[{"firstName" : "David", "lastName" : "Perpich",
"Madison", "type" : "middle"}, {"name" : "Rancho", "type" : "high"} ], "job" :
"small business owner"}]'));
Example: JSONExtractValue with Multiple Results

This example shows how to use the JSONExtractValue method to return multiple results.
This query gets the name of any school that has a type with a value like 'college'.
Note: JSONExtractLargeValue can be substituted for JSONExtractValue.
SELECT eno, edata.JSONExtractValue('$..schools[?(@.type == "college")].name')
FROM my_table
ORDER BY 1;
Teradata JSON
69

Result:
ENO edata.JSONExtractValue()
----------------------------1
UCI
2
Mira Costa
3
CSUSM
4
?
Example: JSONExtractValue with Filtered Results

This example shows how to use the JSONExtractValue method to filter results. This query
gets the name of all persons older than 23.
SELECT eno, edata.JSONExtractValue('$.[?(@.age > 23)].firstName')
FROM my_table
ORDER BY 1;
Result:
----------------------------1
Cameron
2
?
3
Alex
4
David
Example: JSONExtractValue with NULL Results

This example shows the JSONExtractValue method returning NULL results. This query gets
the job associated with each person.
SELECT eno, edata.JSONExtractValue('$..job')
FROM my_table
ORDER BY 1;
Result:
--------------------------------1
programmer
2
?
3
CPA
4
small business owner
Example: JSONExtractValue Resulting in Multiple Found Values in an Array

This example shows how to use JSONExtractValue to return multiple values in an array. This
query gets the name of every school for each person.
70
Teradata JSON


SELECT eno, edata.JSONExtractValue('$..schools..name')
FROM my_table
ORDER BY 1;
Result:
-------------------------------------------------1
["Lake", "Madison", "Rancho", "UCI"]
2
["Lake", "Madison", "Rancho", "Mira Costa"]
3
["Lake", "Madison", "Rancho", "CSUSM"]
4
["Lake", "Madison", "Rancho"]
Teradata JSON
71

72
Teradata JSON
CHAPTER 4
JSON Functions and Operators
ARRAY_TO_JSON
Purpose
The ARRAY_TO_JSON function converts a Teradata ARRAY type to a JSON type composed
of an ARRAY.
Syntax
ARRAY_TO_JSON ( ARRAY_expr )
TD_SYSFNLIB.
If you specify a RETURNS clause, you must enclose the function call in parenthesis.
(
ARRAY_TO_JSON ( ARRAY_expr )
RETURNS clause
TD_SYSFNLIB.
RETURNS clause
RETURNS data_type
( integer )
CHARACTER SET
UNICODE
LATIN
RETURNS STYLE column_expr
Syntax Elements
TD_SYSFNLIB
The name of the database where the function is located.
ARRAY_expr
An expression that evaluates to an ARRAY data type.
ARRAY_expr specifies the array to be converted to the JSON type.
RETURNS data_type
Specifies that data_type is the return type of the function.
data_type can only be JSON.
integer
type.
Teradata JSON
73
Chapter 4 JSON Functions and Operators

ARRAY_TO_JSON
If you do not specify a maximum length, the default maximum length for the character
set is used. If specified, the length is subject to a minimum of two characters and cannot
be greater than the maximum of 16776192 LATIN characters or 8388096 UNICODE
characters.
CHARACTER SET
The character set for the return value of the function, which can be LATIN or
UNICODE.
If you do not specify a character set, the default character set for the user is used.
Specifies that the return type of the function is the same as the data type of the specified
column. The data type of the column must be JSON.
column_expr can be any valid table or view column reference.
The ARRAY data type is mapped to a JSON-formatted string composed of an array, which
can also be a multidimensional array. If the data type of the ARRAY is a numeric predefined
type, the array element maps to a numeric type in the JSON instance. For all other types, the
value added to the JSON instance is the transformed value of each element of the ARRAY,
which is stored in the JSON instance as a string. Note, the JSON array should have the same
number of elements as the ARRAY type.
Return Value
The return type of this function is JSON.
You can use the RETURNS clause to specify the maximum length and character set of the
JSON type.
If you do not specify a RETURNS clause, the return type defaults to JSON data type with
UNICODE character set and a return value length of 64000 bytes, which supports up to
32000 UNICODE characters.
ARRAY_TO_JSON returns NULL if the ARRAY_expr argument is null.

If the ARRAY type is based on a user-defined type (UDT), you must provide a transform that
outputs data in valid JSON syntax in order to use the ARRAY_TO_JSON function.
Otherwise, validation of the JSON array will fail and the function returns an error.
Usage Notes
ARRAY_TO_JSON can be particularly powerful when used in conjunction with the
ARRAY_AGG function, which allows columns of a table to be aggregated into an ARRAY
object. You can then use ARRAY_TO_JSON to convert the aggregated ARRAY into a JSON
array.
Related Topics
74
Teradata JSON

ARRAY_TO_JSON
Setting Up the ARRAY_TO_JSON Examples

Note: The user must have permission to CREATE TYPE. For example, GRANT UDTTYPE
on sysudtlib to <userID> WITH GRANT OPTION;
CREATE TYPE intarr5 AS INTEGER ARRAY[5];
CREATE TYPE intarr33 AS INTEGER ARRAY[3][3];
CREATE TYPE varchararr5 AS VARCHAR(50) ARRAY[5];
CREATE TABLE arrayTable (id INTEGER, a intarr5, b intarr33);
INSERT INTO arrayTable(1, new intarr5(1,2,3,4,5),
new intarr33(1,2,3,4,5,6,7,8,9));
To show the contents of arrayTable, run: SELECT * FROM arrayTable;

id
a
b
------------------------------------------1
(1,2,3,4,5)
(1,2,3,4,5,6,7,8,9)
Example: Array to JSON Type Conversion Using the ARRAY_TO_JSON

Function
Use the ARRAY_TO_JSON function to convert an array to a JSON type composed of an
ARRAY and return the data in both LATIN and UNICODE character sets.
SELECT (ARRAY_TO_JSON(a) RETURNS JSON(20) CHARACTER SET LATIN),
(ARRAY_TO_JSON(b) RETURNS JSON(100) CHARACTER SET UNICODE)
FROM arrayTable;
Result:
ARRAY_TO_JSON(a)
ARRAY_TO_JSON(b)
---------------------------------------[1,2,3,4,5]
[[1,2,3], [4,5,6], [7,8,9]]
Setting Up the ARRAY_TO_JSON Using ARRAY_AGG Examples

CREATE TABLE employeeTable(empId
age INTEGER,
pos CHAR(10),
salary INTEGER);
INSERT INTO employeeTable(1, 24,
Teradata JSON
INTEGER,
'engineer',
'engineer',
'engineer',
'engineer',
'salesman',
'salesman',
'salesman',
'salesman',
50000);
100000);
50000);
75000);
50000);
100000);
50000);
75000);
75

ARRAY_TO_JSON
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
employeeTable(9, 40, 'manager', 75000);

employeeTable(13, 50, 'executive', 125000);
Example: Use ARRAY_AGG

Aggregate the job positions into arrays containing the ages of the employees in those jobs.
SELECT pos, ARRAY_AGG(age ORDER BY empId, NEW intarr5())
FROM employeeTable
GROUP BY pos
ORDER BY pos;
Result:
pos
ARRAY_AGG(age ORDER BY empId ASC, NEW intarr5())
-----------------------------------------------------------engineer
(24,34,25,21)
executive
(50,51,52,52,60)
manager
(40,41,45,48)
salesman
(31,32,33,40)
Example: Use ARRAY_AGG as Input to ARRAY_TO_JSON

Use the aggregated positions and ages from ARRAY_AGG as input to ARRAY_TO_JSON to
obtain a JSON object.
SELECT pos, ARRAY_TO_JSON(ARRAY_AGG(age ORDER BY empId, NEW intarr5()))
FROM employeeTable
GROUP BY pos
ORDER BY pos;
Result:
pos
ARRAY_TO_JSON(ARRAY_AGG())
--------------------------------------engineer
[24,34,25,21]
executive
[50,51,52,52,60]
manager
[40,41,45,48]
salesman
[31,32,33,40]
76
Teradata JSON

ARRAY_TO_JSON
Example: Use ARRAY_AGG to Aggregate Salaries

Use ARRAY_AGG to aggregate the salaries of the job positions.
SELECT salary, ARRAY_AGG(pos ORDER BY empId, NEW varchararr5())
FROM employeeTable
GROUP BY salary
ORDER BY salary;
Result:
salary
ARRAY_AGG(pos ORDER BY empId ASC, NEW varchararr5())
------------------------------------------------------------50000 ('engineer','engineer','salesman','salesman','manager')
75000 ('engineer','salesman','manager')
100000 ('engineer','salesman','manager')
125000 ('manager','executive')
150000 ('executive','executive')
200000 ('executive')
1000000 ('executive')
Example: Use ARRAY_AGG to Aggregate Salaries as Input to

ARRAY_TO_JSON
Use ARRAY_AGG to aggregate the salaries of the job positions as input to
ARRAY_TO_JSON, to obtain JSON objects as output.
SELECT salary, (ARRAY_TO_JSON(ARRAY_AGG(pos ORDER BY empId, NEW
varchararr5()))
RETURNS JSON(100) CHARACTER SET LATIN)
FROM employeeTable
GROUP BY salary
ORDER BY salary;
Result:
salary
ARRAY_TO_JSON(ARRAY_AGG())
----------------------------------50000
["engineer","engineer","salesman","salesman","manager"]
75000
["engineer","salesman","manager"]
100000
["engineer","salesman","manager"]
125000
["manager","executive"]
150000
["executive","executive"]
200000
["executive"]
1000000
["executive"]
Teradata JSON
77

GeoJSONFromGeom
GeoJSONFromGeom
Purpose
The GeoJSONFromGeom function converts an ST_Geometry object into a JSON document
that conforms to the GeoJSON standards.
Syntax
GeoJSONFromGeom
( geom_expr
TD_SYSFNLIB.
)
, precision
(
GeoJSONFromGeom
( geom_expr
TD_SYSFNLIB.
RETURNS clause
, precision
RETURNS clause
RETURNS data_type
( integer )
CHARACTER SET
UNICODE
LATIN
Syntax Elements
TD_SYSFNLIB
geom_expr
An expression which evaluates to an ST_Geometry object that represents a Point,
MultiPoint, LineString, MultiLineString, Polygon, MultiPolygon, or
GeometryCollection.
precision
An integer specifying the maximum number of decimal places in the coordinate values.
If this argument is NULL or not specified, the default precision is 15.
RETURNS data_type
data_type can be VARCHAR, CLOB, or JSON.
integer
Specifies the maximum length of the return type.
If you do not specify a maximum length, the default is the maximum length supported
by the return data type.
CHARACTER SET
The character set for the return value of the function, which can be LATIN or
UNICODE.
If you do not specify a character set, the default character set for the user is used.
78
Teradata JSON

GeoJSONFromGeom

column. The data type of the column must be VARCHAR, CLOB, or JSON.
Return Value
The result of this function is a JSON document which has a format that conforms to the
GeoJSON standards as specified in http://geojson.org/geojson-spec.html.
If you do not specify a RETURNS clause, the default return type is JSON with UNICODE as
the character set. The length of the return value is 64000 bytes, which supports up to 32000
UNICODE characters. This can result in a failure if the underlying ST_Geometry object
exceeds 64000 bytes when converted to a GeoJSON value.
Note: The length of the converted GeoJSON value is greater than the underlying length of
the ST_Geometry object.
The usual rules apply for the maximum length of each data type.
Data Type
Character Set
Maximum Length (characters)
VARCHAR
LATIN
64000
UNICODE
32000
LATIN
2097088000
UNICODE
1048544000
LATIN
16776192
UNICODE
8388096
CLOB
JSON
ANSI Compliance
This function is not compliant with the ANSI SQL:2011 standard.
Related Topics
Examples: GeoJSONFromGeom
These examples show the following conversions of ST_Geometry objects to JSON
documents that conform to GeoJSON standards:
Conversion of Point ST_Geometry objects to JSON documents that have a data type of
VARCHAR, CLOB, or JSON.
Conversion of various ST_Geometry objects to JSON documents that have a data type of
VARCHAR(2000) CHARACTER SET LATIN.
Example: Conversion to a JSON Document with a VARCHAR Type
SELECT (GeoJSONFromGeom(new ST_Geometry('Point(45.12345 85.67891)'))
RETURNS VARCHAR(2000) CHARACTER SET LATIN);
Teradata JSON
79

GeoJSONFromGeom
Result:
GeoJSONFromGeom( NEW ST_GEOMETRY('Point(45.12345 85.67891)'))
------------------------------------------------------------{ "type": "Point", "coordinates": [ 45.12345, 85.67891 ] }
Example: Conversion to a JSON Document with a CLOB Type

RETURNS CLOB CHARACTER SET UNICODE);
Result:
Example: Conversion to a JSON Document with a JSON Type

RETURNS JSON(2000) CHARACTER SET LATIN);
Result:
Example: Convert LineString ST_Geometry Object

SELECT (GeoJSONFromGeom (new ST_Geometry('LineString(10 20, 50 80, 200 50)') )
Result:
GeoJSONFromGeom( NEW ST_GEOMETRY('LineString(10 20, 50 80, 200 50)'))
--------------------------------------------------------------------{ "type": "LineString", "coordinates": [ [ 10.0, 20.0 ], [ 50.0, 80.0 ],
[ 200.0, 50.0 ] ] }
Example: Convert Polygon ST_Geometry Object

SELECT (GeoJSONFromGeom (new ST_Geometry('Polygon((0 0, 0 10, 10 10, 10 0, 0 0))',
4326) )
Result:
GeoJSONFromGeom( NEW ST_GEOMETRY('Polygon((0 0, 0 10, 10 10, 10 0, 0 0))', 4326))
--------------------------------------------------------------------------------{ "type": "Polygon", "coordinates": [ [ [ 0.0, 0.0 ], [ 0.0, 10.0 ], [ 10.0, 10.0 ],
[ 10.0, 0.0 ], [ 0.0, 0.0 ] ] ] }
Example: Convert MultiPoint ST_Geometry Object

SELECT (GeoJSONFromGeom (new ST_Geometry('MultiPoint(10 20, 50 80, 200 50)',
4326) )
80
Teradata JSON

GeoJSONFromGeom
Result:
GeoJSONFromGeom( NEW ST_GEOMETRY('MultiPoint(10 20, 50 80, 200 50)', 4326))
--------------------------------------------------------------------------{ "type": "MultiPoint", "coordinates": [ [ 10.0, 20.0 ], [ 50.0, 80.0 ],
[ 200.0, 50.0 ] ] }
Example: Convert MultiLineString ST_Geometry Object

SELECT (GeoJSONFromGeom (new ST_Geometry(
'MultiLineString((10 20, 50 80, 200 50), (0 100, 10 220, 20 240))', 4326) )
Result:
GeoJSONFromGeom( NEW ST_GEOMETRY('MultiLineString((10 20, 50 80, 200 50),
(0 100, 10 220, 20 240))', 4326))
------------------------------------------------------------------------{ "type": "MultiLineString", "coordinates": [ [ [ 10.0, 20.0 ], [ 50.0, 80.0 ],
[ 200.0, 50.0 ] ], [ [ 0.0, 100.0 ], [ 10.0, 220.0 ], [ 20.0, 240.0 ] ] ] }
Example: Convert MultiPolygon ST_Geometry Object

'MultiPolygon(((0 0, 0 10, 10 10, 10 0, 0 0)), ((0 50, 0 100, 100 100, 100 50, 0 50)))',
4326) )
Result:
GeoJSONFromGeom( NEW ST_GEOMETRY('MultiPolygon(((0 0, 0 10, 10 10, 10 0, 0 0)),
((0 50, 0 100, 100 100, 100 50, 0 50)))', 4326))
------------------------------------------------------------------------------{ "type": "MultiPolygon", "coordinates": [ [ [ [ 0.0, 0.0 ], [ 0.0, 10.0 ],
[ 10.0, 10.0 ], [ 10.0, 0.0 ], [ 0.0, 0.0 ] ] ], [ [ [ 0.0, 50.0 ], [ 0.0, 100.0 ],
[ 100.0, 100.0 ], [ 100.0, 50.0 ], [ 0.0, 50.0 ] ] ] ] }
Example: Convert GeometryCollection ST_Geometry Object

'GeometryCollection(point(10 20), linestring(50 80, 200 50))', 4326) )
Result:
GeoJSONFromGeom( NEW ST_GEOMETRY('GeometryCollection(point(10 20),
linestring(50 80, 200 50))', 4326))
-----------------------------------------------------------------{ "type": "GeometryCollection", "geometries": [ { "type": "Point", "coordinates":
[ 10.0, 20.0 ] }, { "type": "LineString", "coordinates": [ [ 50.0, 80.0 ],
[ 200.0, 50.0 ] ] } ] }
Teradata JSON
81

GeomFromGeoJSON
GeomFromGeoJSON
Purpose
The GeomFromGeoJSON function converts a JSON document that conforms to the
GeoJSON standards into an ST_Geometry object.
Syntax
GeomFromGeoJSON
( geojson_expr , asrid )
TD_SYSFNLIB.
Syntax Elements
TD_SYSFNLIB
geojson_expr
An expression which evaluates to a JSON document conforming to the GeoJSON
standards as specified in http://geojson.org/geojson-spec.html. The GeoJSON text string
can be represented as a VARCHAR, CLOB, or JSON instance in the LATIN or
UNICODE character set.
This GeoJSON text string must represent a geometry object. The value of the type
member must be one of these strings: "Point", "MultiPoint", "LineString",
"MultiLineString", "Polygon", "MultiPolygon", or "GeometryCollection".
asrid
An integer which specifies the Spatial Reference System (SRS) identifier assigned to the
returned ST_Geometry object.
Return Value
An ST_Geometry object which contains the data that was stored in the JSON document.
ANSI Compliance
This function is not compliant with the ANSI SQL:2011 standard.
Examples: GeomFromGeoJSON
Examples: Valid GeoJSON Geometry Objects
The following show examples of valid GeoJSON geometry objects. The objects are valid based
on the GeoJSON format specification which you can access at: http://geojson.org/geojsonspec.html.
Point
{ "type": "Point", "coordinates": [100.0, 0.0] }
LineString
82
Teradata JSON

GeomFromGeoJSON
{ "type": "LineString",
"coordinates": [ [100.0, 0.0], [101.0, 1.0] ]
}
Polygon (without Holes)

{ "type": "Polygon",
"coordinates": [
[ [100.0, 0.0], [101.0, 0.0], [101.0, 1.0], [100.0, 1.0], [100.0, 0.0] ]
]
}
Polygon (with Holes)

{ "type": "Polygon",
"coordinates": [
[ [100.0, 0.0], [101.0, 0.0], [101.0, 1.0], [100.0, 1.0], [100.0, 0.0] ],
[ [100.2, 0.2], [100.8, 0.2], [100.8, 0.8], [100.2, 0.8], [100.2, 0.2] ]
]
}
MultiPoint
{ "type": "MultiPoint",
"coordinates": [ [100.0, 0.0], [101.0, 1.0] ]
}
MultiLineString
{ "type": "MultiLineString",
"coordinates": [
[ [100.0, 0.0], [101.0, 1.0] ],
[ [102.0, 2.0], [103.0, 3.0] ]
]
}
MultiPolygon
{ "type": "MultiPolygon",
"coordinates": [
[[[102.0, 2.0], [103.0, 2.0], [103.0, 3.0], [102.0, 3.0], [102.0, 2.0]]],
[[[100.0, 0.0], [101.0, 0.0], [101.0, 1.0], [100.0, 1.0], [100.0, 0.0]],
[[ 100.2, 0.2], [100.8, 0.2], [100.8, 0.8], [100.2, 0.8], [100.2, 0.2]]]
]
}
GeometryCollection
{ "type": "GeometryCollection",
"geometries": [
{ "type": "Point",
"coordinates": [100.0, 0.0]
},
{ "type": "LineString",
"coordinates": [ [101.0, 0.0], [102.0, 1.0] ]
}
Teradata JSON
83

JSON_CHECK
]
}
Example: Converting a GeoJSON Value into an ST_Geometry Object

SELECT GeomFromGeoJSON('{ "type": "Point", "coordinates": [100.0, 0.0] }', 4326);
Result:
Result
-----POINT (100 0)
Example: Error - Passing a JSON Value That Is Not GeoJSON Compliant
In this example, an error is returned because "Poit" is not valid JSON geometry data.
SELECT GeomFromGeoJSON('{ "type": "Poit", "coordinates": [100.0, 0.0] }', 4326);
Result: The query returns an error.

*** Failure 9134 GeomFromGeoJSON: conversion from JSON text failed, invalid GeoJSON
input.
Statement# 1, Info =0
JSON_CHECK
Purpose
The JSON_CHECK function checks a string for valid JSON syntax and provides an
informative error message about the cause of the syntax failure if the string is invalid. It does
not create a JSON instance.
This function can be used to validate text before loading JSON data in order to save time in
case of syntax errors.
Syntax
JSON_CHECK
( ' { string } ' )
TD_SYSFNLIB.
Syntax Elements
TD_SYSFNLIB
'{string}'
The string to be tested for compliance to JSON syntax.
CHAR, VARCHAR, and CLOB are allowed as input types.
LATIN and UNICODE are allowed for all data types.
84
Teradata JSON

JSON_KEYS
A maximum of 8388096 UNICODE characters or 16776192 LATIN characters may be

passed in for validation.
Return Value
'OK', if the string is valid JSON syntax.
'INVALID: error message', if the string is not valid JSON syntax. The error message
provides an informative message about the cause of the syntax failure.
NULL, if the string is null.
Usage Notes
You can use this function to validate before loading a large amount of JSON data. This
prevents the rollback of an entire transaction in the event of a JSON syntax error and
provides the necessary information to fix any syntax errors that may be present.
Related Topics
JSON String Syntax, on page 30
Examples: JSON_CHECK
Valid JSON String Argument
The string passed to JSON_CHECK in this query has valid JSON syntax; therefore, the
function returns 'OK'.
SELECT JSON_CHECK('{"name" : "Cameron", "age" : 24, "schools" :
[ {"name" : "Lake", "type" : "elementary"}, {"name" : "Madison",
"type" : "middle"}, {"name" : "Rancho", "type" : "high"}, {"name" :
"UCI", "type" : "college"} ], "job" : "programmer" }');
Invalid JSON String Argument

The string passed to JSON_CHECK in this query is not valid JSON syntax; therefore, the
function returns the error message: INVALID: Expected something like whitespace or '"'
between ',' and the end of the string.
SELECT JSON_Check('{"name" : "Cameron",');
JSON_KEYS
Purpose
The JSON_KEYS table operator parses a JSON instance (represented as CHAR, VARCHAR,
CLOB, or JSON) and returns a list of key names.
Teradata JSON
85

JSON_KEYS
Syntax
JSON_KEYS
( ON ( json_expr )
TD_SYSFNLIB.
USING
name ( value )
correlation_name
AS
Syntax Elements
TD_SYSFNLIB
json_expr
An expression that evaluates to correct JSON syntax. The expression can be a CHAR,
VARCHAR, CLOB, or JSON representation of a JSON data type in LATIN or
UNICODE.
USING name (value)
AS
USING DEPTH (value)

An optional clause that restricts the depth of the search. value is a positive integer
that specifies the maximum depth for the search.
The default value is the maximum depth of the JSON instance.
For example, setting the depth to 1 gets the top level keys in the JSON instance.
USING QUOTES (value)
An optional clause that specifies whether or not the key names in the result are
enclosed in double quotation marks as follows:
If value is 'Y' or 'y', the key names in the result are enclosed in double quotation
marks. This is the default if the USING QUOTES clause is not specified.
If value is 'N' or 'n', the key names in the result are not enclosed in double
quotation marks.
An optional keyword introducing correlation_name.

correlation_name
An alias for the input column specified by the ON clause.
ANSI SQL refers to aliases as correlation names. They are also referred to as range
variables.
Return Values
JSON_KEYS performs a search on the JSON instance to the specified depth and returns a
VARCHAR column of key names.
If a JSON array is present in the document, one result per index of each array is generated.
All results are given according to their "path" in the JSON document. Therefore, the output
corresponding to a nested key will contain all of the parent keys, in addition to itself.
If you specify the USING QUOTES ('Y') clause, the key names in the result set are enclosed
in double quotation marks. This is the default behavior and allows you to copy and paste a
86
Teradata JSON

JSON_KEYS
path into an SQL statement, using the path as an entity reference on a JSON document,
without any potential for improper use of SQL keywords.
If you specify the USING QUOTES ('N') clause, the key names in the result set are not
enclosed in double quotation marks. This allows you to use the output as input to one of the
JSON extraction methods, such as JSONExtractValue.
Setting Up the JSON_Keys Examples

CREATE TABLE json_table(
id int,
json_data JSON(32000)
);
INSERT json_table(1, '
{
"coord": {
"lon": 145.766663,
"lat": -16.91667
},
"sys": {
"country": "AU",
"sunrise": 1375216946,
"sunset": 1375257851
},
"weather": [
{
"id": 800,
"main": "Clear",
"description": "Sky is Clear",
"icon": "01n"
}
],
"base": "global stations",
"main": {
"temp": 288.97,
"humidity": 99,
"pressure": 1007,
"temp_min": 288.71,
"temp_max": 289.15
},
"wind": {
"speed": 5.35,
"deg": 145.001
},
"clouds": {
"all": 0
},
"dt": 1375292971,
"id": 2172797,
"name": "Cairns",
"cod": 200
}
Teradata JSON
87

JSON_KEYS
');
INSERT json_table(2, '
{
"coord": {
"lon": 245.766663,
"lat": -16.91667
},
"sys": {
"country": "US",
"sunrise": 1375216946,
"sunset": 1375257851
},
"weather": [
{
"id": 800,
"main": "Clear",
"description": "Sky is Cloudy",
"icon": "01n"
}
],
"base": "global stations",
"main": {
"temp": 288.97,
"humidity": 99,
"pressure": 1007,
"temp_min": 288.71,
"temp_max": 289.15,
"temp_scale" : "Fahrenheit"
},
"wind": {
"speed": 5.35,
"deg": 145.001
},
"clouds": {
"all":0
},
"dt": 1375292971,
"id": 2172797,
"name": "Los Angeles",
"cod": 200
}
');
Example: Use JSON_Keys to Obtain Key Names with No Depth Specified

The example uses JSON_KEYS to obtain the key names of a JSON object. No depth is
specified, so the entire depth is searched and the keys at all levels are returned.
SELECT * from JSON_KEYS
(
ON (SELECT json_data FROM json_table))
AS json_data;
88
Teradata JSON

JSON_KEYS
Result:
KEYS
---coord
coord.lon
coord.lat
sys
sys.country
sys.sunrise
sys.sunset
weather
weather[0].id
weather[0].main
weather[0].description
weather[0].icon
base
main
main.temp
main.humidity
main.pressure
main.temp_min
main.temp_max
wind
wind.speed
wind.deg
clouds
clouds.all
dt
id
name
cod
Example: Use JSON_Keys to Obtain Key Names for the Top Level Depth
The example uses JSON_KEYS to obtain the key names from the top level depth.
SELECT * from JSON_KEYS
(
ON (SELECT json_data FROM json_table) USING DEPTH(1))
AS json_data;
Result:
JSONKEYS
-------coord
sys
weather
base
main
wind
clouds
dt
id
Teradata JSON
89

JSON_KEYS
name
cod
Examples: Using the USING QUOTES Clause

Example: USING QUOTES ('Y')
In this example, JSON_KEYS is invoked with the USING QUOTES ('Y') clause; therefore, the
key names returned in the result are enclosed in double quotation marks.
SELECT * FROM JSON_KEYS
(ON (SELECT NEW JSON('{ "x":{"a":{"b":3}}
QUOTES('Y'))
AS json_data;
, "y" : "b"}')) USING
Result:
KEYS
---"x"
"x"."a"
"x"."a"."b"
"y"
Example: USING QUOTES ('N')

In this example, JSON_KEYS is invoked with the USING QUOTES ('N') clause; therefore, the
key names returned in the result are not enclosed in double quotation marks.
QUOTES('N'))
AS json_data;
, "y" : "b"}')) USING
Result:
KEYS
---x
x.a
x.a.b
y
Example: USING QUOTES Is Not Specified

In this example, JSON_KEYS is invoked without specifying the USING QUOTES clause. The
default behavior is to return the key names enclosed in double quotation marks.
AS json_data;
, "y" : "b"}')))
Result:
KEYS
---"x"
90
Teradata JSON

JSON_KEYS
"x"."a"
"x"."a"."b"
"y"
Example: Get an Ordered List of Unique Keys From Documents

The example uses JSON_KEYS to get an ordered list of all unique keys from all documents of
all rows of a table.
SELECT distinct(JSONKeys) from JSON_KEYS
(
ON (SELECT json_data FROM json_table))
AS json_data
ORDER BY 1;
Result:
JSONKeys
-------"base"
"clouds"
"clouds"."all"
"cod"
"coord"
"coord"."lat"
"coord"."lon"
"dt"
"id"
"main"
"main"."humidity"
"main"."pressure"
"main"."temp"
"main"."temp_max"
"main"."temp_min"
"main"."temp_scale"
"name"
"sys"
"sys"."country"
"sys"."sunrise"
"sys"."sunset"
"weather"
"weather"[0]
"weather"[0]."description"
"weather"[0]."icon"
"weather"[0]."id"
"weather"[0]."main"
"wind"
"wind"."deg"
"wind"."speed"
Teradata JSON
91

JSON_KEYS
Example: Get a List of Unique Keys Without Quotes

The example uses JSON_KEYS to get an ordered list of all unique keys from all documents of
all rows of a table. The key names returned in the result are not enclosed in double quotation
marks.
SELECT distinct(JSONKeys) from JSON_KEYS
(
ON (SELECT json_data FROM json_table) USING QUOTES('N'))
AS json_data
ORDER BY 1;
Result:
JSONKeys
-------base
clouds
clouds.all
cod
coord
coord.lat
coord.lon
dt
id
main
main.humidity
main.pressure
main.temp
main.temp_max
main.temp_min
main.temp_scale
name
sys
sys.country
sys.sunrise
sys.sunset
weather
weather[0]
weather[0].icon
weather[0].id
weather[0].main
wind
wind.deg
wind.speed
Example: Use JSON_KEYS with JSONExtractValue to Extract All Values

The example uses JSON_KEYS with JSONExtractValue to extract all values of the JSON
document.
SELECT CAST(JSONKeys AS VARCHAR(30)),
T.json_data.JSONExtractValue('$.'||JSONKeys) from json_table T, JSON_KEYS
92
Teradata JSON

JSON_KEYS
(
ON (SELECT json_data FROM json_table WHERE id=1) USING QUOTES('N'))
AS json_data
where T.id=1
ORDER BY 1;
Result:
JSONKeys
json_data.JSONEXTRACTVALUE(('$.'||JSONKeys))
--------------------------------------------------------------------------base
global stations
clouds
{"all":0}
clouds.all
0
cod
200
coord
{"lon":145.766663,"lat":-16.91667}
coord.lat
-16.91667
coord.lon
145.766663
dt
1375292971
id
2172797
main
{"temp":288.97,"humidity":99,"pressure":
1007,"temp_min":288.71,"temp_max":289.15}
main.humidity
99
main.pressure
1007
main.temp
288.97
main.temp_max
289.15
main.temp_min
288.71
name
Cairns
sys
{"country":"AU","sunrise":1375216946,"sunset":
1375257851}
sys.country
AU
sys.sunrise
1375216946
sys.sunset
1375257851
weather
[{"id":800,"main":"Clear","description":"Sky is
Clear","icon":"01n"}]
weather[0]
{"id":800,"main":"Clear","description":"Sky is
Clear","icon":"01n"}
Sky is Clear
weather[0].icon
01n
weather[0].id
800
weather[0].main
Clear
wind
{"speed":5.35,"deg":145.001}
wind.deg
145.001
wind.speed
5.35
SELECT CAST(JSONKeys AS VARCHAR(30)),
T.json_data.JSONExtractValue('$.'||JSONKeys) from json_table T, JSON_KEYS
(
ON (SELECT json_data FROM json_table WHERE id=2) USING QUOTES('N'))
AS json_data
where T.id=2
ORDER BY 1;
Result:
JSONKeys
json_data.JSONEXTRACTVALUE(('$.'||JSONKeys))
---------------------------------------------------------------------------
Teradata JSON
93

JSON_KEYS
base
global stations
clouds
{"all":0}
clouds.all
0
cod
200
coord
{"lon":245.766663,"lat":-16.91667}
coord.lat
-16.91667
coord.lon
245.766663
dt
1375292971
id
2172797
main
{"temp":288.97,"humidity":99,"pressure":
1007,"temp_min":288.71,"temp_max":289.15,"temp_scale":"Fahrenheit"}
main.humidity
99
main.pressure
1007
main.temp
288.97
main.temp_max
289.15
main.temp_min
288.71
main.temp_scale
Fahrenheit
name
Los Angeles
sys
{"country":"US","sunrise":1375216946,"sunset":
1375257851}
sys.country
US
sys.sunrise
1375216946
sys.sunset
1375257851
weather
[{"id":800,"main":"Clear","description":"Sky is
Cloudy","icon":"01n"}]
weather[0]
{"id":800,"main":"Clear","description":"Sky is
Cloudy","icon":"01n"}
Sky is Cloudy
weather[0].icon
01n
weather[0].id
800
weather[0].main
Clear
wind
{"speed":5.35,"deg":145.001}
wind.deg
145.001
wind.speed
5.35
94
Teradata JSON
CHAPTER 5
JSON Shredding
About JSON Shredding

Teradata provides shredding functionality that allows you to extract values from JSON
documents and store the extracted data in relational format. Data from a JSON instance can
be used to INSERT, UPDATE, UPSERT, or DELETE from one or more existing tables.
Functions are provided to create a temporary table and/or update tables based on a single
JSON instance or a group of JSON instances.
JSON_TABLE
This is a table operator which takes a JSON instance and creates a temporary table
based on a subset, or all, of the data contained in the JSON instance.
JSON_SHRED_BATCH and JSON_SHRED_BATCH_U
The JSON_SHRED_BATCH and JSON_SHRED_BATCH_U stored procedures load
relational tables with data extracted from JSON instances.
JSON_SHRED_BATCH operates on LATIN character set data and
JSON_SHRED_BATCH_U operates on UNICODE data. Otherwise, the two
procedures are the same.
JSON_TABLE
Purpose
JSON_TABLE creates a temporary table based on all, or a subset, of the data in a JSON
object.
Syntax
JSON_TABLE ( ON ( json_documents_retrieving_expr )
USING rowexpr ( row_expr_literal )
TD_SYSFNLIB.
A
colexpr ( column_expr_literal ) )
correlation_name
,
AS
(
Teradata JSON
column_name
95
Chapter 5 JSON Shredding

JSON_TABLE
Syntax Elements
json_documents_retrieving_expr
A query expression that must result in at least two columns. The first column is an ID
that identifies a JSON document. The ID is returned in the output row to correlate
output rows to the input rows, and in case of errors, it is used to identify the input row
that caused the error. The second column is the JSON document itself. Other columns
can be included, but they must be after the JSON document column. Extra columns are
returned in the output row without being modified from the input.
This is an example of the json_documents_retrieving_expr.
SELECT id, jsonDoc FROM jsonTable;
This is an example json_documents_retrieving_expr query using extra columns. The

extra columns are returned by the table operator and are output as they are.
SELECT id, orderJson, orderDate, orderSite FROM orderJsnTable;
If the json_documents_retrieving_expr parameter is NULL, the result of the function is a

table with no rows.
row_expr_literal
A LATIN or UNICODE string literal that conforms to JSONPath syntax.
This parameter is the JSONPath query that is evaluated against the JSON document
returned in the second column of the json_documents_retrieving_expr.
The row_expr_literal query returns an array of objects. One output row is returned for
each object in the array.
If row_expr_literal is NULL an error is reported.
column_expr_literal
A LATIN or UNICODE literal that represents a JSON array. This array contains one or
more JSON objects, each of which defines a column to be produced by the
JSON_TABLE operator.
If this parameter is NULL an error is reported.
AS
Optional keyword introducing correlation_name.
correlation_name
An alias for the table that is referenced by json_documents_retrieving_expr.
ANSI SQL refers to table aliases as correlation names. They are also referred to as range
variables.
column_name
An optional list of one or more column names.
JSON_TABLE takes a JSON instance and creates a temporary table based on all, or a subset,
of the data in the instance. The JSON instance is retrieved from a table with an SQL
statement. JSONPath syntax is used to define the portions of the JSON instance to retrieve.
96
Teradata JSON

JSON_TABLE
Return Value
The JSON_TABLE table operator produces output rows which conform to the row format
defined by the column_expr_literal. That literal describes the columns in the output row and
their data types.
The rows returned by JSON_TABLE have the following columns:
The first column returned contains the JSON document ID obtained from the first
column in the json_documents_retrieving_expression.
The next N columns returned are generated based on the colexpr parameter, where N is
the number of objects in the JSON array represented by the column_expression_literal.
If json_documents_retrieving_expr returns more than two columns, all the extra columns
from the third column onward are added to the output row without being modified.
The column_expr_literal parameter requires a mapping of the columns in the
row_expr_literal to the columns of the output table of this function. Each column in the
column_expr_literal is defined by a JSON instance that must conform to one of the following
structures.
Ordinal Column
An Ordinal Column contains an integer sequence number. The sequence number is
not guaranteed to be unique in itself, but the combination of the id column, the first
column of the output row, and the ordinal column is unique.
{ "ordinal" : true}
Normal Column Relative to Row

The value of this column is determined by executing a JSONPath query. The query is
evaluated with an object returned by the ROWEXPR JSONPath query as the root
object.
{jsonpath : <jsonpath query string>,
type : <data type of output column> }
Normal Column Relative to Root

The value of this column is determined by executing a JSONPath query. The query is
evaluated with the object returned by json_documents_retrieving_expression, the
second column, as the root object. This type of column is identified by setting
"fromRoot" to true.
{jsonpath : <jsonpath query string>,
type : <data type of output column>,
fromRoot:true }
JSON_TABLE does not support UDTs and LOB types in the output so the JSON data type
cannot be the type for the output columns. The data type of the columns of the output table
may be any non-LOB predefined Teradata type. The supported data types for JSON_TABLE
output are listed next.
Supported Data Types

The following lists the supported data types for JSON_TABLE output. The square brackets
indicate optional values in the type specification.
Teradata JSON
97

JSON_TABLE
CHAR(n)
VARCHAR(n)
BYTE(n)
VARBYTE(n)
BYTEINT
SMALLINT
INTEGER
FLOAT/REAL
DECIMAL/NUMERIC [(n,[m])]
NUMBER
DATE
TIME [(fractional_seconds_precision)]
TIME [(fractional_seconds_precision)] WITH TIME ZONE
TIMESTAMP [(fractional_seconds_precision)]
TIMESTAMP [(fractional_seconds_precision)] WITH TIME ZONE
INTERVAL YEAR [(precision)]
INTERVAL YEAR [(precision)] TO MONTH
INTERVAL MONTH [(precision)]
INTERVAL DAY[(precision)]
INTERVAL DAY[(precision)] TO HOUR
INTERVAL DAY[(precision)] TO MINUTE
INTERVAL DAY [(precision)] TO SECOND
INTERVAL HOUR [(precision)]
INTERVAL HOUR [(precision)] TO MINUTE
INTERVAL HOUR [(precision)] TO SECOND
INTERVAL MINUTE [(precision)]
INTERVAL MINUTE [(precision)] TO SECOND
INTERVAL SECOND [(fractional_seconds_precision)]

This function resides in TD_SYSFNLIB. The ID column of the query result can be any
number type or character type excluding CLOB. It is the responsibility of the user to make
the ID column unique. If it is not unique JSON_TABLE does not fail, but it is difficult to tell
which rows came from which JSON document in the resulting output.
The result is subject to the maximum row size, and the query must not exceed the maximum
allowable size for a query.
98
Teradata JSON

JSON_TABLE
Note: When a JSON value is shredded to populate a CHAR, VARCHAR, or VARBYTE

column, if the size of the value is larger than the size of the target column, the value is
truncated to fit the column.
Setting Up the JSON_TABLE Examples

CREATE TABLE my_table (id INTEGER, jsonCol JSON(1000));
INSERT INTO my_table (1, NEW JSON('{"name" : "Cameron", "age" : 24,
"schools" : [ {"name" : "Lake", "type" : "elementary"},
{"name" : "Madison", "type" : "middle"}, {"name" : "Rancho", "type" : "high"},
{"name" : "UCI", "type" : "college"} ], "job" : "programmer" }'));
INSERT INTO my_table (2, NEW JSON('{"name" : "Melissa", "age" : 23,
{"name" : "Madison", "type" : "middle"},
{"name" : "Rancho", "type" : "high"},
{"name" : "Mira Costa", "type" : "college"} ] }'));
INSERT INTO my_table (3, NEW JSON('{"name" : "Alex", "age" : 25,
{"name" : "Rancho", "type" : "high"},
{"name" : "CSUSM", "type" : "college"} ], "job" : "CPA" }'));
INSERT INTO my_table (4, NEW JSON('{"name" : "David", "age" : 25,
{"name" : "Rancho", "type" : "high"} ],
"job" : "small business owner"}'));
Example: JSON_TABLE Using Normal Column

The example shows JSON_Table using normal column.
SELECT * FROM JSON_Table
(ON (SELECT id, jsonCol FROM my_table WHERE id=1)
USING rowexpr('$.schools[*]')
colexpr('[ {"jsonpath" : "$.name",
"type" : "CHAR(20)"},
{"jsonpath" : "$.type",
"type" : "VARCHAR(20)"}]')
) AS JT(id, schoolName, "type");
Result:
id schoolName
type
-----------------------------1
Lake
elementary
1
Madison
middle
Teradata JSON
99

JSON_TABLE
1
1
Rancho
UCI
high
college
The example shows JSON_Table using the fromRoot attribute.

( ON (SELECT id, jsonCol FROM my_table WHERE id=1)
"type" : "CHAR(20)"},
"type" : "VARCHAR(20)"},
{"jsonpath" : "$.name",
"type" : "VARCHAR(20)",
"fromRoot":true} ]')
) AS JT(id, schoolName, "type", studentName);
Result:
id schoolName
type
studentName
-------------------------------------------------1
Lake
elementary
Cameron
1
Madison
middle
Cameron
1
Rancho
high
Cameron
1
UCI
college
Cameron
Example: JSON_TABLE Using Extra Columns

The example shows JSON_TABLE using extra columns, in addition to the required columns
of ID and JSON document. The column_expression_literal parameter requires a mapping of
the columns in the row_expression_literal to the columns of the output table of this function,
as this example demonstrates. To simplify this example, constants are used for the state and
nation columns.
(ON (SELECT id, jsonCol, 'CA' AS state, 'USA' AS nation
FROM my_table WHERE id=1)
"type" : "CHAR(20)"},
"type" : "VARCHAR(20)"}]')
) AS JT(id, name, "type", State, Nation);
Result:
id name
type
State Nation
-----------------------------------------------1
Lake
elementary
CA
USA
1
Madison middle
CA
USA
100
Teradata JSON

1
1
Rancho
UCI
high
college
CA
CA
USA
USA
Example: JSON_TABLE Using Ordinal Column

The example shows JSON_TABLE using ordinal column.
SELECT * FROM JSON_TABLE
(ON (SELECT id, jsonCol, 'CA' AS state, 'USA' AS nation
FROM my_table)
colexpr('[ {"ordinal" : true},
{"jsonpath" : "$.name",
"type" : "CHAR ( 12 )"},
"type" : "VARCHAR ( 12 )"}]'))
AS JT(idcol, ordnum, res1, res2, State, Nation);
Result:
idcol
----------3
4
3
4
3
4
3
1
1
1
1
2
2
2
2
ordnum
----------0
0
1
1
2
2
3
4
5
6
7
8
9
10
11
res1
-----------Lake
Lake
Madison
Madison
Rancho
Rancho
CSUSM
Lake
Madison
Rancho
UCI
Lake
Madison
Rancho
Mira Costa
res2
-----------elementary
elementary
middle
middle
high
high
college
elementary
middle
high
college
elementary
middle
high
college
State
----CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
Nation
-----USA
USA
USA
USA
USA
USA
USA
USA
USA
USA
USA
USA
USA
USA
USA
JSON_SHRED_BATCH and
JSON_SHRED_BATCH_U
Purpose
JSON_SHRED_BATCH and JSON_SHRED_BATCH_U are SQL stored procedures that use
any number of JSON instances to populate existing tables, providing a flexible form of
loading data from the JSON format into a relational model. Two shred procedures are
provided; however, the only difference between them is the character set of the data. To
explain the functionality, we only describe JSON_SHRED_BATCH (the version that
Teradata JSON
101

operates on LATIN character set data), but the explanation applies equally to
JSON_SHRED_BATCH_U (the UNICODE version).
The batch shredding procedures map into a number of successive calls to JSON_TABLE to
create a conglomerate temporary table, the values of which can be assigned to existing tables.
Syntax
,
JSON_SHRED_BATCH
SYSLIB.
input_query
, : result_code
shred statement
JSON_SHRED_BATCH_U
shred statement
row expression
,
,
column expression
table expression
query expression
row expression
rowexpr : JSONPath_expr
column expression
,
colexpr :
temp_column_name : JSONPath_expr
type : data_type
, fromRoot
: : true
query expression
,
queryexpr :
temp_column_name :
column_type
table expression
,
tables : [
table_name : {
metadata
column_assignment
metadata
metadata : { operation :
insert
}
,
update
merge
, keys : [
, filter : filter_expression
table_column_name
delete
column assignment
,
columns : {
table_column_name :
temp_column_name
temp_expr
numeric_constant
[ string_constant ]
boolean_constant
null
102
Teradata JSON

Syntax Elements
CALL JSON_SHRED_BATCH
The following describes the parameters used by the JSON_SHRED_BATCH and

JSON_SHRED_BATCH_U procedure calls.
input query
input query
A string input parameter which specifies a query that results in a group of JSON
instances from which the user can perform shredding. Extra columns can result and be
referred to in the shred statement.
If this parameter is NULL, an error is reported.
The input query parameter can operate on one or more JSON objects in a source table. This
query is mapped to a JSON_TABLE function call. Since JSON_TABLE requires that the first
two columns specified be an ID value and a JSON object, respectively, the input query
parameter also requires the first two columns to be an ID value and a JSON object.
The following are examples of an input query string.
'SELECT id, empPersonalInfo, site FROM test.json_table'
'SELECT JSONDOCID, JSONDT1, a, b FROM jsonshred.JSON_TABLE3 WHERE
JSONID=100'
JSONID (uppercase or lowercase) is a keyword. It is a temporary column name used for the
JSON document ID value. JSONID is allowed in the input query and table expression clauses.
You cannot use JSONID as a temp_column_name in "colexpr" or "queryexpr".
The execution of JSON_TABLE on multiple JSON objects requires a join between the result
of one invocation and the source table. In order to avoid a full table join, we require an ID
column to be specified in the input query parameter, so that a join condition can be built off
that column.
The data types in the queryexpr (discussed later) must match the actual data type of the
columns specified in the input query. No explicit cast will be added, so the data must be
implicitly castable to the data type defined in the query expr, if not the exact data type. Any
errors encountered will result in a failed shred, and will be reported to the user.
If there is a problem encountered during the execution of JSON_TABLE, the ID column is
used in the error message to isolate which row caused the problem.
shred statement
The shred statement element defines the mapping of the JSON instance, that resulted from
the input query, to where the data will be loaded in the user table(s).
If the shred statement is NULL an error is reported.
All keywords in the shred statement must be specified in lowercase.
The following sections discuss the structure and syntax of the shred statement. Multiple
shred statements can be run, but there are performance impacts.
Teradata JSON
103

row expression
The following describes the variables used by the row expression.

"rowexpr" :
Required, literal entry.
Must be lowercase.
JSONPath expr
An expression in JSONPath syntax to extract information about a particular portion of
a JSON instance. For example, $.schools[*] identifies all schools.
column expression
The following describes the variables used by column expression.

"colexpr":
Must be lowercase.
temp_column_name
Any user-defined name for the temporary column. The temporary column name must
be unique or an error is reported.
Note: The names are not case sensitive. For example, col1 and COL1 will fail because
they are used in internal queries and are not unique.
You cannot use JSONID and ROWINDEX (uppercase or lowercase) here.
JSONPath expr
A JSONPath expression in JSONPath syntax that requests information about a
particular portion of a JSON object. For example, $.name.
"type":
Must be lowercase.
data_type
Non-LOB Teradata predefined type, such as INTEGER or VARCHAR. For a list of
supported types, see Supported Data Types.
"fromRoot" : true
Optional. You must use the fromRoot attribute for non-relative paths. Each column
expression is assumed to be relative to the row expression ("rowexpr": "JSONPATH
expression"), unless specified as not relative. A relative expression starts with one of the
names of a child of the row expression, whereas a non-relative expression starts with the
root in JSONPath, $.
true is a literal and must be lowercase.
Note: Attempting to set fromRoot to false raises an error.
The user is responsible for mapping elements of the column expression to acceptable data
types. No explicit casting is needed; the data is implicitly cast to the desired data type.
However, if the data does not correctly cast to the desired data type an error is reported,
therefore, care should be taken when determining the elements and data types of the column
expression. If the result of the expression is an array or object (instead of a single value), the
only acceptable data types are CHAR or VARCHAR of suitable length.
104
Teradata JSON

The data type of the temporary columns in the output table of JSON_TABLE must be
specified. This is enforced with JSON_SHRED_BATCH and JSON_SHRED_BATCH _U in
the column expression. It is necessary for the user to provide this information so that the data
may be correctly interpreted and used with the target table(s).
query expression
The following describes the query expression variables.

"queryexpr" :
Must be lowercase.
"temp_column_name"
Any user-defined name for the temporary column. The temporary column name must
be unique or an error is reported.
Note: The names are not case sensitive. For example, col1 and COL1 will fail because
they are used in internal queries and are not unique.
You cannot use JSONID and ROWINDEX (uppercase or lowercase) here.
"column_type"
Data type of the temp_column_name column.
The queryexpr is used to define the data types of the data selected in the input query, for the
columns that are not the ID value or the JSON object. It is mandatory that the extra columns
be referenced in the queryexpr to specify the data type desired.
All the columns in the input query, from the third column onward, must be replicated in the
same order in the queryexpr.
Note: The order of the columns is important while the names of the columns are not.
The data types in the queryexpr should match the actual data type of the columns specified
in the input query. No explicit cast is added, so the data must be implicitly castable to the
data type defined in the queryexpr, if not the exact data type. Any errors encountered will
result in a failed shred, and are reported to the user.
The following example demonstrates the importance of the ordering. Notice that both
columns qrycol1 and qrycol2 are included in the queryexpr. Note, qrycol2 in the queryexpr
refers to qrycol1 in the input query, and qrycol1 in the queryexpr refers to qrycol2 in the
input query. As stated, order is significant, not the names.
CALL SYSLIB.JSON_SHRED_BATCH(
'SELECT JSONDOCID, JSONDT1, qrycol1, qrycol2 FROM
jsonshred.JSON_TABLE3 WHERE JSONID=100',
'[ { "rowexpr" : "$.population.profile",
"colexpr" : [{"col1" : "$.name.first", "type" : "VARCHAR(30)"},
{"col2" : "$.address.zip", "type" : "NUMBER(5,0)"}],
"queryexpr" : [{ "qrycol2" : "VARCHAR(20)"}, { "qrycol1" :
"VARCHAR(20)"}],
"tables" : [
{"jsonshred.JSON_SHRED_TABLE1" : {
"metadata" : { "operation" : "insert" },
"columns" : {"EmpID":"JSONID*10",
"NAME":"col1","STATE":"qrycol1", "ZIP":"col2"}
}
Teradata JSON
105

}
]
}
]',res );
JSONID and ROWINDEX (uppercase or lowercase) are not allowed in colexpr and queryexpr
because they are fixed temporary column names. A syntax error is reported if they are used
in those clauses.
table expression
The following describes the variable used by table expression.

"tables" :
Must be lowercase.
"table_name"
The fully qualified name of an existing database table. The user must have permission to
access the table.
JSONID and ROWINDEX (uppercase or lowercase) are keywords. They are used to track the
input JSON document ID value (the first column of the input query) and the index number
for an input row, respectively. JSONID and ROWINDEX may be referenced in the table
expression clause as a source value for the shredding operation.
Note: In the process of shredding, a volatile table is created for each shred statement. A table
can have a maximum of 2048 columns, so all the columns together from all the table
mappings should not exceed 2044 columns (there are four internal columns). You can have 1
to N target tables, which can each have 1 to N columns, but the total number of all columns
must not exceed 2044.
metadata
The following describes the metadata variables.

Note: All literals must be lowercase.
"metadata" :
Must be lowercase.
"operation" :
"insert" | "update" | "merge" | "delete"
Operation to perform.
In a MERGE operation, the target table must have a primary index, and the primary
index has to be a member of the keys in the metadata.
"keys":
Using keys is optional. If used, "keys": is a required, literal entry.
Note: All names given in the keys clause must be present in the column assignment
clause.
The keys are used to perform the join between the temporary table created by the row
and column expressions and the target table. This should be used carefully as it can
106
Teradata JSON

drastically affect performance. In the case of a MERGE operation, the target table must
have a primary index, and the primary index has to be a member of the specified keys.
"table_column_name"
The name of any column in the table referenced by table_name, which is the name of an
existing table the executor has access too. table_name is referenced in the table
expression of JSON_SHRED_BATCH, as well.
"filter":
Filtering is optional. If used, "filter": is a required, literal entry.
filter_expression
SQL statement referencing elements of the column or query expressions.
Example filter statement: "filter" : "empId<5000"
column assignment
The following describes the column assignment variables.

"columns" :
"table_column_name'"
The name of any column in the table referenced by table_name. table_name is the name
of any existing table that the executor of the procedure has access to. It must be the fully
qualified table name. (table_name is also described in table expression.)
"temp_column_name"
A temp_column_name defined in "colexpr" or "queryexpr". The temporary column
name must be unique or an error is reported. Note: Temporary column names are not
case sensitive, so col1 and COL1 are not unique and will cause an error.
"temp_expr"
Teradata SQL expression.
numeric_constant
Any JSON supported Numeric value.
["string_constant"]
Any JSON supported string value.
Example string constant: "company" : ["Teradata"]
boolean_constant
true or false
true and false are JSON keywords and must be lowercase.
null
JSON null.
null is a JSON keyword and must be lowercase.
result code
result code
An output parameter representing the result of the shred operation. A value of 0
indicates success. All non-zero values indicate specific error conditions and an
appropriate error message is returned.
Teradata JSON
107

Return Values
The return value is an output parameter representing the result of the shred operation.
A value of 0 indicates success.
All non-zero values indicate specific error conditions, and an appropriate error message is
returned.
Usage Notes
When the character set of the data to be shredded is:
LATIN, use the JSON_SHRED_BATCH procedure
UNICODE, use the JSON_SHRED_BATCH_U procedure
Other than the difference regarding the character set of the data, the functionality of the two
procedures is identical.
Note: When a JSON value is shredded to populate a CHAR, VARCHAR, or VARBYTE
column, if the size of the value is larger than the size of the target column, the value is
truncated to fit the column.
The JSON_SHRED_BATCH query provides flexibility between the source JSON instance and
the table(s) the source data is loaded into. This flexibility allows for efficient and non-efficient
queries, depending on the query itself and how the mapping (shred statement) is performed.
The following guidelines assist in achieving the optimal performance with these procedures.
For each shred statement, a JSON_TABLE function call is made, to shred the JSON object
into a temporary table based on the row expression and column expressions. The resulting
temporary table may be used to assign values to any column of any table for which the
user has the proper privileges. The best performing queries optimize the mapping such
that each shred statement updates the maximum possible number of tables. Only if
complications of the mapping (such as hierarchical relationships) make it impossible to
map a shredding to an actual column should another shred statement be included in the
query.
The performance is largely dependent upon the usage of the procedure. If the mapping
minimizes the number of separate queries needed, it will perform best. It is not always the
case that everything can fit into one shred statement; for this reason multiple statements
are allowed.
This procedure allows INSERT, UPDATE, MERGE and DELETE operations, which can be
specified in the operation portion of the metadata portion of the statement. The keys in
the metadata statement are used to perform the join between the temporary table created
by the row/column expressions and the target table. This should be used carefully as it can
drastically affect performance. In a MERGE operation, the target table must have a
primary index, and the primary index has to be a member of the keys in the metadata.
In order to avoid a full table join, we require an ID column to be specified in the input
query parameter, so that a join condition can be built off that column.
Columns of a target table may be assigned values in the temporary table created by the row
and column expressions, constants, or the results of SQL expressions. The use of an SQL
expression requires the user to submit a proper SQL statement (in terms of syntax and actual
results of the query). This is a powerful and flexible way to manipulate the data in a target
table, but can cause a problem if queries are not structured properly. Any errors reported by
the DBS based on an SQL expression will be reported to the user and cause the query to fail.
108
Teradata JSON

Columns of the temporary table created by the row and column expressions and the extra
columns created by the input query may be used in the SQL expression.
In the process of shredding, a volatile table is created for each shred statement. A table can
have a maximum of 2048 columns, so all the columns together from all the table mappings
should not exceed 2044 columns (there are four internal columns). You can have 1 to N
target tables, which can each have 1 to N columns, but the total number of all columns must
not exceed 2044.
All keywords in the shred statement must be specified in lowercase.
The names assigned to the temporary columns (temp_column_name) and the names of extra
columns created by the input query must be unique. They can be referenced in the table
expression clause, so there cannot be any ambiguity. Note, names are not case-sensitive. If a
non-unique name is detected, an error is reported. For example, col1 and COL1 will fail
because they are used in internal queries and are not unique.
Note: All the names given in the keys clause must be present in the column assignment
clause.
You must specify the data type of the temporary column in the output table in the column
expression. It is necessary to provide this information so that the data may be correctly
interpreted and used with the target table(s).
JSONID and ROWINDEX (uppercase or lowercase) are keywords. They are used to track the
input JSON document ID value (the first column of the input query) and the index number
for an input row, respectively. JSONID and ROWINDEX are not allowed in colexpr and
queryexpr because they are fixed temporary column names. A syntax error is reported if they
are used in those clauses. However, they may be referenced in the table expression clause as a
source value for the shredding operation.
Supported Data Types

Teradata supports shredding to columns of the following data types.
Teradata JSON
CHAR
VARCHAR
CHARACTER(n)
CHARACTER SET GRAPHIC
VARCHAR(n) CHARACTER
SET GRAPHIC
CLOB
BYTE
VARBYTE
BLOB
BYTEINT
SMALLINT
INTEGER
BIGINT
FLOAT (for data types where

applicable, precision is
supported)
DECIMAL
NUMBER
DATE
TIME
TIME WITH TIME ZONE
TIMESTAMP
TIMESTAMP WITH TIME

ZONE
INTERVAL YEAR
INTERVAL YEAR TO MONTH INTERVAL MONTH
INTERVAL DAY
INTERVAL DAY TO HOUR
INTERVAL DAY TO SECOND
INTERVAL DAY TO MINUTE
109

INTERVAL HOUR
INTERVAL HOUR TO
MINUTE
INTERVAL HOUR TO
SECOND
INTERVAL MINUTE
INTERVAL MINUTE TO
SECOND
INTERVAL SECOND
PERIOD (DATE)
PERIOD (TIME)
PERIOD (TIME WITH TIME

ZONE)
PERIOD (TIMESTAMP)
PERIOD (TIMESTAMP WITH XML

TIME ZONE)
JSON
Privileges
JSON_SHRED_BATCH resides in the SYSLIB database. The user executing the
JSON_SHRED_BATCH procedures requires privileges on the tables being updated, which
include the following:
The database where the procedure is executing must have all privileges on SYSUDTLIB,
SYSLIB, and the database where the target table exists and EXECUTE PROCEDURE on
SYSLIB.
SYSLIB must have all privileges on the database which is executing the procedure.
For example, if the database where the procedure is executing and where the target tables
exists is called JSONShred, then the following statements will assign the required privileges:
GRANT
GRANT
GRANT
GRANT
GRANT
ALL ON SYSUDTLIB TO JSONShred;

ALL ON SYSLIB TO JSONShred;
EXECUTE PROCEDURE ON SYSLIB TO JSONSHRED;
ALL ON JSONShred TO JSONShred;
ALL ON JSONShred TO SYSLIB;
Setting up the JSON_SHRED_BATCH Examples

CREATE TABLE emp_table (
empID INTEGER,
company
VARCHAR(50),
empName VARCHAR(20),
empAge
INTEGER,
dept VARCHAR(20),
startDate DATE FORMAT 'YY/MM/DD',
site VARCHAR(20))
PRIMARY INDEX (company, empID);
CREATE MULTISET TABLE dept_table (
dept
VARCHAR(20),
description
VARCHAR(200),
empID
INTEGER);
CREATE TABLE json_table (
id INTEGER,
110
Teradata JSON

empPersonalInfo
JSON(1000),
empCompanyInfo JSON(1000),
site VARCHAR(20));
Insert sample data into the table.

INSERT INTO json_table (1,
'{"employees" : {
"company" : "Teradata",
"info" :
[
{ "id" : 1,
"name" : "Cameron",
"age" : 24,
"dept" : "engineering"},
{ "id" : 2,
"name" : "Justin",
"age" : 30,
"dept" : "engineering"},
{ "id" : 3,
"name" : "Melissa",
"age" : 24,
"dept" : "marketing"}
]}}',
'{"startDates" : {
"info" :
[
{"id" : 1, "startDate" : "2015/02/10"},
{"id" : 2, "startDate" : "2015/02/07"},
{"id" : 3, "startDate" : null}
]}}', 'RB'
);
Example: JSON_SHRED_BATCH Extracts from a JSON Object and Inserts

into a Table
The example populates a table using a JSON instance as the source data. The example shreds
the JSON document to extract values from it and inserts the data into the employee table
(emp_table).
Note: Before running the example GRANT to SYSLIB privileges to the database in which the
example table resides. For example: GRANT <privilege> ON <database> TO SYSLIB;
'SELECT id, empPersonalInfo, site
FROM test.json_table',
'[{"rowexpr" : "$.employees.info[*]",
"colexpr" : [{"col1" : "$.id",
"type" : "INTEGER"},
{"col2" : "$.employees.company",
"fromRoot" : true},
{"col3" : "$.name",
Teradata JSON
111

{"col4" : "$.age",
{"col5" : "$.dept",
"type" : "VARCHAR(10)"}],
"queryexpr" : [{"site" : "VARCHAR(20)"}],
"tables" : [
{"test.emp_table" : {
"metadata" : {
"operation" : "insert"
},
"columns" : {"empID" : "col1*100",
"company" : "col2",
"empName" : "col3",
"empAge" : "col4",
"dept" : "col5",
"startDate" : "CURRENT_DATE",
"site" : "site" }
}
}]
}]', :res );
The result of the shred populates the emp_table table with three rows, corresponding to the
three items in the JSON object used as source data.
To see the result, run: SELECT empID, company, empName, empAge, startDate,
site FROM emp_table ORDER BY empID;
empID company
empName empAge
startDate site
---------------------------------------------100
Teradata Cameron 24
13/09/19 RB
200
Teradata Justin
34
13/09/19 RB
300
Teradata Melissa 24
13/09/19 RB
Example: Use JSON_SHRED_BATCH to Update a Table from a JSON Object

The example uses the JSON_SHRED_BATCH Update operation to update a table from a
JSON instance.
Assume some new data comes in which provides the actual start date of the three employees
(previously we loaded the table with a default value). We can update that specific value using
JSON_SHRED_BATCH with the query below.
'SELECT id, empCompanyInfo FROM test.json_table',
'[
{
"rowexpr" : "$.startDates.info[*]",
"colexpr" : [
{"col1" : "$.id",
112
Teradata JSON

{"col2" : "$.startDates.company",
"fromRoot" : true},
{"col3" : "$.startDate",
"type" : "VARCHAR(20)"}
],
"tables" : [
{
"test.emp_table" : {
"metadata" : {
"operation" : "update",
"keys" : [ "empID", "company" ]
},
"columns" : { "empID" : "col1*100",
"company" : "col2",
"startDate" : "col3" }
}
}
]
}
]', :res );
Result: To view the updated data in the employee table, run: SELECT empID, company,
empName, empAge, startDate, site FROM emp_table ORDER BY empID;
empID company
empName empAge startDate site
---------------------------------------------100
Teradata Cameron 24
15/02/10 RB
200
Teradata Justin
34
15/02/07 RB
300
Teradata Melissa 24
15/02/07 RB
Example: Populate Multiple Tables from a JSON Object Using

JSON_SHRED_BATCH
The example uses a single JSON_SHRED_BATCH call to populate two tables with data.
'SELECT id, empPersonalInfo, site FROM test.json_table',
'[
{
"rowexpr" : "$.employees.info[*]",
"colexpr" : [
{"col1" : "$.id",
{"col2" : "$.employees.company",
"fromRoot" : true},
{"col3" : "$.name",
{"col4" : "$.age",
Teradata JSON
113

{"col5" : "$.dept",
"type" : "VARCHAR(20)"}
],
"queryexpr" : [
{"site" : "VARCHAR(20)"}
],
"tables" : [
{
"test.emp_table" : {
"columns" : {
"empID" : "col1*100",
"company" : "col2",
"empName" : "col3",
"empAge" : "col4",
"dept" : "col5",
"startDate" : "CURRENT_DATE",
"site" : "site" }
}
},
{
"test.dept_table" : {
"columns" : {
"dept" : "col5",
"description" : ["CONSTANT DESCRIPTION"],
"empID" : "col1"
}
}
}
]
}
]', :res );
The result of the above shred will populate the emp_table and dept_table tables with three
rows, corresponding to the three items in the JSON object used as source data.
Result: To view the data inserted into the employee table, run: SELECT * FROM
emp_table ORDER BY empID;
empID
----100
200
300
company
-----------Teradata
Teradata
Teradata
empName
--------Cameron
Justin
Melissa
empAge
-----24
30
24
dept
-----------engineering
engineering
marketing
startDate
---------15/02/10
15/02/07
?
site
----RB
RB
RB
Result: To view the data inserted into the department table, run: SELECT * FROM
dept_table ORDER BY dept, empID;
dept
description
empID
-------------------------------------engineering CONSTANT DESCRIPTION
1
engineering CONSTANT DESCRIPTION
2
marketing
CONSTANT DESCRIPTION
3
114
Teradata JSON

Setting Up the JSON_SHRED_BATCH JSONID and ROWINDEX Keyword

Example
CREATE TABLE JSONDocs(
JSONDocId INTEGER,
jsnCol JSON(10000),
site VARCHAR(200),
country VARCHAR(200)
) UNIQUE PRIMARY INDEX (JSONDocId);
INSERT INTO JSONDocs VALUES('1',
NEW JSON('{"employees" : {"company" : "Teradata","info" : [
{"id" : 1, "name" : "Cameron", "dept" : "engineering"},
{"id" : 2, "name" : "Justin","dept" : "engineering"}
]}}', LATIN),'HYD','USA');
INSERT INTO JSONDocs VALUES('2',
NEW JSON('{"employees" : {"company" : "Teradata","info" : [
{"id" : 3, "name" : "Madhu", "dept" : "engineering"},
{"id" : 4, "name" : "Srini","dept" : "engineering"}
]}}', LATIN),'HYD','USA');
CREATE TABLE jsonshred.Teradata_Employees(rowIndex INTEGER, empId
INTEGER, empName varchar(30), company varchar(30),dept varchar(10),
jsonDocId INTEGER,site varchar(10), country varchar(10));
Example: JSON_SHRED_BATCH Using JSONID and ROWINDEX Keywords

The example shows the use of the JSONID and ROWINDEX keywords.
'SELECT * FROM jsonshred.JSONDocs',
NEW JSON('[
{
"rowexpr" : "$.employees.info[*]",
"colexpr" : [{"col1" : "$..id", "type" : "INTEGER"},
{"col2" : "$.employees.company", "type" :
"VARCHAR(15)","fromRoot":true},
{"col3" : "$..name", "type" : "VARCHAR(20)"},
{"col4" : "$..dept", "type" : "VARCHAR(20)"}],
"queryexpr" : [{"qrycol1" : "varchar(20)"},{"qrycol2" :
"varchar(20)"}],
"tables" : [
{"jsonshred.Teradata_Employees" : {
"metadata" : { "operation" : "insert","keys" :
["empId"] },
"columns" : {"rowIndex" : "ROWINDEX",
"empId" : "col1+1000",
"empName" : "col3",
Teradata JSON
115

"jsonDocId" : "JSONID",
"company" : "col2",
"dept" : "col4",
"site" : "qrycol1",
"country" : "qrycol2" }
}
}
]
}]', LATIN), :res);
To see the result, run: SELECT * from jsonshred.Teradata_Employees order by

rowindex;
rowIndex
-------0
1
2
3
116
empId
-----1001
1002
1003
1004
empName
-----------Cameron
Justin
Madhu
Srini
company
----------Teradata
Teradata
Teradata
Teradata
dept
---------engineerin
engineerin
engineerin
engineerin
jsonDocId
--------1
1
2
2
site
-----HYD
HYD
HYD
HYD
country
------USA
USA
USA
USA
Teradata JSON
CHAPTER 6
JSON Publishing
About JSON Publishing

Teradata provides JSON publishing functionality to take a variable number of parameters
and compose a JSON document. A user may create simple or complex queries to generate
the JSON document.
Comparison of JSON_AGG and JSON_COMPOSE

JSON_AGG and JSON_Compose are similar functions that take a variable number of input
parameters and packages them into a JSON document. Both commands use the same syntax
and input parameters.
The difference is JSON_AGG is an aggregate function and JSON_COMPOSE is a scalar
function. JSON_COMPOSE can call JSON_AGG to provide a more complex composition of
a JSON document than the JSON_AGG function can do by itself.
JSON_AGG
Purpose
The JSON_AGG function returns a JSON document composed of aggregated values from
each input parameter. The input parameters can be a column reference or an expression.
Each input parameter results in a name/value pair in the returned JSON document.
Syntax
,
JSON_AGG (
param
)
( FORMAT ' format string ' )
AS name
,
( JSON_AGG (
param
Teradata JSON
RETURNS clause
AS name
117
Chapter 6 JSON Publishing

JSON_AGG
RETURNS clause
RETURNS data_type
( integer )
CHARACTER SET
UNICODE
LATIN
Syntax Elements
param
An input parameter that can be any supported data type, a column reference, constant,
or expression that evaluates to some value. A variable number of these parameters are
accepted and each input parameter results in a name/value pair in the returned JSON
document.
FORMAT 'format string'
format string is any allowable format string in Teradata.
For an example using the format string see Example: Use JSON_COMPOSE with
Subqueries and GROUP BY.
AS name
name is any allowable name in Teradata.
The string created conforms to the JSON standard escaping scheme. A subset of
UNICODE characters are required to be escaped by the '\' character. This is not the case
for strings in Teradata. Thus, when porting a Teradata string to a JSON string, proper
JSON escape characters are used where necessary. This also applies to the values of the
JSON instance and to the JSON_COMPOSE function. If the character set is LATIN, '\'
escaped characters must be part of that character set; otherwise a syntax error is
reported.
RETURNS data_type
data_type must be JSON for this function.
integer
type. If specified, the length is subject to a minimum of two characters and cannot be
greater than the absolute maximum allowed for the function. Shorter lengths may be
specified.
Note: As an aggregate function, JSON_AGG supports up to 64000 bytes, which is 32000
UNICODE characters or 64000 LATIN characters. The RETURNS clause can specify a
larger return value, but the actual data returned by JSON_AGG is 64000 bytes. If the
data length is greater than this an error is returned. Note, JSON_COMPOSE can specify
larger values than JSON_AGG.
If you do not specify a RETURNS clause, the return type defaults to JSON(32000)
CHARACTER SET UNICODE. In other words, the default return type is a JSON data
type with UNICODE character set and a return value length of 32000 characters.
CHARACTER SET UNICODE | LATIN
The character set for the data type in the RETURNS data_type clause.
The character set can be LATIN or UNICODE.
118
Teradata JSON

JSON_AGG

Return Value
By default, JSON_AGG returns a JSON document in character set UNICODE and a
maximum length of 32000 UNICODE characters (64000 bytes), unless otherwise specified
with the optional RETURNS clause.
A hierarchical relationship is not possible with this function. The resulting JSON instance is
flat, with each input parameter corresponding to one child of the result. The resulting
document will be in the following format.
{
name1 : data1,
name2 : data2,
...,
nameN : dataN,
}
If one of the values used to compose the JSON document is a Teradata NULL, it is returned
in the JSON instance as a JSON null.

The input parameter may be any data type allowed in Teradata. Predefined types will be
mapped to JSON data types according to this mapping.
Numeric data types in Teradata map to a number in JSON.
Null values in Teradata map to null values in JSON.
All other data types in Teradata map to a string in JSON.
Structured LOB UDTs are not allowed in JSON_AGG.
Note: All non-predefined types, such as all UDTs, use their transformed value to populate
the resulting JSON document. The user is required to provide a transform which outputs
data in valid JSON syntax in order to function properly. Otherwise, validation of the JSON
instance will fail and the function returns an error message. The data type in the JSON value
is determined according to the mapping above, based on the predefined data type of the
result of the transform. All non-LOB predefined types will be formatted according to the
optional FORMAT clause specified for any particular column, or, if that is not present, the
default format for the particular data type.
The param name can be specified using the optional "AS" clause for each parameter. If the
optional portion is NULL, the names of the parameters that make up the resulting JSON
document will be given according to current naming conventions of expressions in Teradata.
Usage Notes
The GROUP BY clause can be used in the SELECT statement which invokes the JSON_AGG
function. Existing rules for the GROUP BY clause and aggregate functions apply to
JSON_AGG. When this is used, the resulting JSON document is structured as an array with
Teradata JSON
119

JSON_AGG
objects as its elements that represent members of the resulting group. Each group is in a
different output row.
If one of the values used to compose the JSON object is a Teradata NULL, it is returned in the
JSON instance as a JSON null.
Setting Up the JSON_AGG Examples

empID
INTEGER,
company
VARCHAR(50),
empName
VARCHAR(20),
empAge
INTEGER);
INSERT INTO emp_table(1, 'Teradata', 'Cameron', 24);
INSERT INTO emp_table(2, 'Teradata', 'Justin', 34);
INSERT INTO emp_table(3, 'Teradata', 'Someone', 24);
Example: Using JSON_AGG without GROUP BY and No Parameter Names

The example selects columns from a table and uses JSON_AGG to compose those columns
into JSON objects.
SELECT JSON_agg(empID, company, empName, empAge)
FROM emp_table;
Result:
JSON_agg
-------[{ "empID" : 1, "company" : "Teradata", "empName" : "Cameron", "empAge" : 24 },
{ "empID" : 2, "company" : "Teradata", "empName" : "Justin", "empAge" : 34 },
{ "empID" : 3, "company" : "Teradata", "empName" : "Someone", "empAge" : 24 }]
Example: Using JSON_AGG without GROUP BY and with Parameter Names

The example shows how to use JSON_Agg to assign parameter names in the resulting JSON
object.
SELECT JSON_agg(empID AS id, company, empName AS name, empAge AS age)
FROM emp_table;
Result:
JSON_agg
-------{ "id" : 1, "company" : "Teradata", "name" : "Cameron", "age" : 24 },
{ "id" : 2, "company" : "Teradata", "name" : "Justin", "age" : 34 },
{ "id" : 3, "company" : "Teradata", "name" : "Someone", "age" : 24 }
120
Teradata JSON

JSON_COMPOSE
Example: Using JSON_AGG with GROUP BY and with All Parameter Names
The example shows how to use JSON_Agg to assign parameter names in the resulting JSON
instance and use the GROUP BY clause.
SELECT JSON_agg(empID AS id, empName AS name, empAge AS age)
FROM emp_table
GROUP BY company;
Result: The query returns one line of output (note, output line is wrapped).
JSON_agg(empID AS id,empName AS name,empAge AS age)
--------------------------------------------------[{"id":3,"name":"Someone","age":24},
{"id":1,"name":"Cameron","age":24},
{"id":2,"name":"Justin","age":34}]
Example: Using JSON_AGG with Multiple GROUP BY and Parameter Names

The example shows how to assign parameter names and use the GROUP BY clause.
SELECT JSON_agg(empID AS id, empName AS name)
FROM emp_table
GROUP BY company, empAge;
Result:
JSON_AGG(empID AS id,empName AS name)
------------------------------------------------------------------------{"id":2,"name":"Justin"}
[{"id":3,"name":"Someone"},{"id":1,"name":"Cameron"}]
JSON_COMPOSE
Purpose
JSON_COMPOSE creates a JSON document composed of the input parameters specified.
This function provides a complex composition of a JSON document when used in
conjunction with the JSON_AGG function.
Syntax
,
JSON_COMPOSE (
param
)
AS name
Teradata JSON
121

JSON_COMPOSE
,
param
( JSON_COMPOSE (
)
RETURNS clause
AS name
RETURNS clause
RETURNS data_type
( integer )
CHARACTER SET
UNICODE
LATIN
Syntax Elements
param
An input parameter that can be a column reference, constant, or expression that
evaluates to some value. A variable number of these parameters are accepted, and each
input parameter results in a name/value pair in the returned JSON document.
FORMAT 'format string'
format string is any allowable format string in Teradata.
For an example using the format string see Example: Use JSON_COMPOSE with
Subqueries and GROUP BY.
AS name
name is any allowable name in Teradata.
The string created conforms to the JSON standard escaping scheme. A subset of
UNICODE characters are required to be escaped by the '\' character. This is not the case
for strings in Teradata. Thus, when porting a Teradata string to a JSON string, proper
JSON escape characters are used where necessary. This also applies to the values of the
JSON instance and to the JSON_AGG function. If the character set is LATIN, '\' escaped
characters must be part of that character set; otherwise a syntax error is reported.
RETURNS data_type
data_type must be JSON for this function.
integer
type. If specified, the length is subject to a minimum of two characters and cannot be
greater than the absolute maximum for the character set. Shorter lengths may be
specified.
If you do not specify a RETURNS clause, the return type defaults to JSON(32000)
CHARACTER SET UNICODE. In other words, the default return type is a JSON data
type with UNICODE character set and a return value length of 32000 characters.
If you specify the optional RETURNS clause, the maximum lengths allowed are
16776192 characters for LATIN and 8388096 characters for UNICODE.
CHARACTER SET UNICODE | LATIN
The character set for the data type in the RETURNS data_type clause.
The character set can be LATIN or UNICODE.
122
Teradata JSON

JSON_COMPOSE

Return Value
By default, JSON_COMPOSE returns a LOB based JSON document with character set
UNICODE and a maximum length of 32000 UNICODE characters (64000 bytes), unless
otherwise specified with the optional RETURNS clause.
A hierarchical relationship is not possible with this function. The resulting JSON document
is flat, with each input parameter corresponding to one child of the result. The resulting
document will be in this format:
{
name1 : data1,
name2 : data2,
...,
nameN : dataN,
}
If one of the values used to compose the JSON document is a Teradata NULL, it is returned
in the JSON instance as a JSON null.

The input parameter may be any data type allowed in Teradata. Predefined types will be
mapped to JSON data types according to this mapping:
Numeric data types in Teradata map to a number in JSON.
Null values in Teradata map to null values in JSON.
All other data types in Teradata map to a string in JSON.
Note: All non-predefined types, such as, all UDTs, use their transformed value to populate
the resulting JSON document. The user is required to provide a transform which outputs
data in valid JSON syntax in order to function properly. Otherwise, validation of the JSON
instance will fail and the function returns an error message. The data type in the JSON value
is determined according to the mapping above, based on the predefined data type of the
result of the transform. All non-LOB predefined types will be formatted according to the
optional FORMAT clause specified for any particular column, or, if that is not present, the
default format for the particular data type.
The param name can be specified using the optional "AS" clause for each parameter. If the
optional portion is NULL, the names of the parameters that make up the resulting JSON
document will be given according to current naming conventions of expressions in Teradata.
Usage Notes
JSON_COMPOSE is most useful when used in conjunction with JSON_AGG. JSON_AGG
is limited in that it provides groups as identified by the GROUP BY clause, but it does not
provide the value that was used to create the group. To obtain this, use JSON_AGG in a
subquery that results in a derived table, and reference the result of JSON_AGG as one of the
parameters to the JSON_COMPOSE function. To ensure the values being grouped on are
Teradata JSON
123

JSON_COMPOSE
included with the proper groups, the columns used in the GROUP BY clause of the subquery
with the JSON_AGG function should be used as parameters to the JSON_COMPOSE
function along with the result of JSON_AGG. In this way, the values being grouped on will be
included alongside the group.
Setting Up the JSON_COMPOSE Basic Examples

empID INTEGER,
company VARCHAR(50),
empName VARCHAR(20),
empAge INTEGER);
INSERT INTO emp_table(1, 'Teradata', 'Cameron', 24);
INSERT INTO emp_table(2, 'Teradata', 'Justin', 34);
INSERT INTO emp_table(3, 'Apple', 'Someone', 34);
Example: Use JSON_COMPOSE to Extract Values from a Table and

Compose a JSON Object
Use JSON_COMPOSE to create a JSON object from the values selected from a table.
SELECT JSON_Compose(T.company, T.employees)
FROM
(
SELECT company, JSON_agg(empID AS id,
empName AS name,
empAge AS age) AS employees
FROM emp_table
GROUP BY company
) AS T;
Result:
JSON_Compose
-----------{
"company" :
"employees"
{ "id" :
{ "id" :
]
}
{
"company" :
"employees"
{ "id" :
]
}
124
"Teradata",
: [
1, "name" : "Cameron", "age" : 24 },
2, "name" : "Justin", "age" : 34 }
"Apple",
: [
3, "name" : "Someone", "age" : 24 }
Teradata JSON

JSON_COMPOSE
Example: Use JSON_COMPOSE with JSON_AGG

The example uses JSON_AGG to aggregate employee ID and employee name, and uses
JSON_COMPOSE to create a JSON object from the values extracted and aggregated.
SELECT JSON_Compose(T.company, T.empAge AS age, T.employees)
FROM
(
SELECT company, empAge,
JSON_agg(empID AS id,
empName AS name) AS employees
FROM emp_table
GROUP BY company, empAge
) AS T;
Result:
JSON_Compose
-----------{
"age" : 24,
"employees" : [
{ "id" : 1, "name" : "Cameron" }
]
}
{
"age" : 34,
"employees" : [
{ "id" : 2, "name" : "Justin" }
]
}
{
"company" : "Apple",
"age" : 24,
"employees" : [ { "id" : 3, "name" : "Someone" } ]
}
Setting up the JSON_COMPOSE Advanced Examples

CREATE TABLE order_table (
orderID
INTEGER,
customer
VARCHAR(50),
price
INTEGER);
CREATE MULTISET TABLE item_table (
orderID
INTEGER,
itemID
INTEGER,
itemName
VARCHAR(50),
amount
INTEGER);
INSERT INTO order_table(1, 'Teradata', 1000);
Teradata JSON
125

JSON_COMPOSE
INSERT INTO order_table(2, 'Teradata', 10000);
INSERT INTO order_table(3, 'Apple', 100000);
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INSERT
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
INTO
item_table(1,
item_table(1,
item_table(2,
item_table(2,
item_table(2,
item_table(2,
item_table(3,
item_table(3,
item_table(3,
item_table(3,
item_table(3,
item_table(3,
item_table(3,
item_table(3,
1,
2,
1,
2,
3,
4,
1,
2,
3,
4,
5,
6,
7,
8,
'disk', 100);
'RAM', 200);
'disk', 10);
'RAM', 20);
'monitor', 30);
'keyboard', 40);
'disk', 10);
'RAM', 20);
'monitor', 30);
'keyboard', 40);
'camera', 50);
'button', 60);
'mouse', 70);
'pen', 80);
Example: Use JSON_COMPOSE with Subqueries and GROUP BY

The example uses JSON_COMPOSE to select items from tables and group the output, with a
subquery that uses JSON_AGG
SELECT JSON_Compose(O.customer,
O.orderID,
O.price,
I.JA AS items)
FROM
(
SELECT customer AS customer,
orderId AS orderID,
price (FORMAT '$(9).9(2)') AS price
FROM order_table
) AS O,
(
SELECT orderID AS orderID,
JSON_Agg(itemID AS ID,
itemName AS name,
amount AS amt) AS JA
FROM item_table
GROUP BY orderID
) AS I
WHERE O.orderID = I.orderID;
Result:
JSON_Compose
-----------{
"customer" : "Teradata",
"orderID" : 1,
"price" : "$1000.00",
"items" :
[
{ "ID" : 1, "name" : "disk", "amt" : 100 },
126
Teradata JSON

JSON_COMPOSE
{ "ID" : 2, "name" : "RAM", "amt" : 200 },
]
}
{
"orderID" : 2,
"price" : "$10000.00",
"items" :
[
{ "ID" : 1, "name" : "disk", "amt" : 10 },
{ "ID" : 2, "name" : "RAM", "amt" : 20 },
{ "ID" : 4, "name" : "keyboard", "amt" : 40 }
]
}
{
"customer" : "Apple",
"orderID" : 3,
"price" : "$100000.00",
"items" :
[
{ "ID" : 1, "name" : "disk", "amt" : 10 },
{ "ID" : 2, "name" : "RAM", "amt" : 20 },
{ "ID" : 4, "name" : "keyboard", "amt" : 40 },
{ "ID" : 5, "name" : "camera", "amt" : 50 },
{ "ID" : 6, "name" : "button", "amt" : 60 },
{ "ID" : 7, "name" : "mouse", "amt" : 70 },
{ "ID" : 8, "name" : "pen", "amt" : 80 }
]
}
Example: Use JSON_COMPOSE with Multiple Subqueries and GROUP BY

Compose a JSON object that is grouped by customer AND orderID and has all items
associated with each orderID.
SELECT JSON_Compose(T.customer,
T.JA AS orders)
FROM
(
SELECT O.customer AS customer,
JSON_Agg(O.orderID, O.price, I.JA AS items) AS JA
FROM
(
SELECT customer AS customer,
orderId AS orderID,
price AS price
FROM order_table
) AS O,
(
SELECT orderID AS orderID,
JSON_Agg(itemID AS ID,
itemName AS name,
amount AS amt) AS JA
Teradata JSON
127

JSON_COMPOSE
FROM item_table
GROUP BY orderID
) AS I
WHERE O.orderID = I.orderID
GROUP BY O.customer
) AS T;
Result:
> JSON_Compose
-----------{
"orders" :
[
{
"orderID" : 1,
"price" : 1000,
"items" :
[
{ "ID" : 1, "name" :
{ "ID" : 2, "name" :
]
},
{
"orderID" : 2,
"price" : 10000,
"items" :
[
{ "ID" : 1, "name" :
{ "ID" : 2, "name" :
{ "ID" : 3, "name" :
{ "ID" : 4, "name" :
]
}
]
}
{
"customer" : "Apple",
"orders" :
[
{
"orderID" : 3,
"price" : 100000,
"items" :
[
{ "ID" : 1, "name" :
{ "ID" : 2, "name" :
{ "ID" : 3, "name" :
{ "ID" : 4, "name" :
{ "ID" : 5, "name" :
{ "ID" : 6, "name" :
{ "ID" : 7, "name" :
{ "ID" : 8, "name" :
]
}
128
"disk", "amt" : 100 },

"RAM", "amt" : 200 }
"disk", "amt" : 10 },
"RAM", "amt" : 20 },
"monitor", "amt" : 30 },
"keyboard", "amt" : 40 }
"disk", "amt" : 10 },
"RAM", "amt" : 20 },
"monitor", "amt" : 30 },
"keyboard", "amt" : 40 },
"camera", "amt" : 50 },
"button", "amt" : 60 },
"mouse", "amt" : 70 },
"pen", "amt" : 80 }
Teradata JSON

JSON_COMPOSE
]
}
Teradata JSON
129

JSON_COMPOSE
130
Teradata JSON
APPENDIX A
Notation Conventions
Syntax Diagram Conventions

Notation Conventions
Item
Definition and Comments
Letter
An uppercase or lowercase alphabetic character ranging from A through Z.
Number
A digit ranging from 0 through 9.

Do not use commas when typing a number with more than 3 digits.
Word
Keywords and variables.

UPPERCASE LETTERS represent a keyword.
Syntax diagrams show all keywords in uppercase, unless operating system
restrictions require them to be in lowercase.
lowercase letters represent a keyword that you must type in lowercase, such
as a Linux command.
Mixed Case letters represent exceptions to uppercase and lowercase rules.
The exceptions are noted in the syntax explanation.
lowercase italic letters represent a variable such as a column or table name.
Substitute the variable with a proper value.

lowercase bold letters represent an excerpt from the diagram.
The excerpt is defined immediately following the diagram that contains it.
UNDERLINED LETTERS represent the default value.
This applies to both uppercase and lowercase words.
Spaces
Use one space between items such as keywords or variables.
Punctuation
Type all punctuation exactly as it appears in the diagram.
Paths
The main path along the syntax diagram begins at the left with a keyword, and proceeds, left
to right, to the vertical bar, which marks the end of the diagram. Paths that do not have an
arrow or a vertical bar only show portions of the syntax.
The only part of a path that reads from right to left is a loop.
Teradata JSON
131
Appendix A Notation Conventions

Continuation Links
Paths that are too long for one line use continuation links. Continuation links are circled
letters indicating the beginning and end of a link:
A
When you see a circled letter in a syntax diagram, go to the corresponding circled letter and
continue reading.
Required Entries
Required entries appear on the main path:
SHOW
If you can choose from more than one entry, the choices appear vertically, in a stack. The first
entry appears on the main path:
SHOW
CONTROLS
VERSIONS
Optional Entries
You may choose to include or disregard optional entries. Optional entries appear below the
main path:
SHOW
CONTROLS
If you can optionally choose from more than one entry, all the choices appear below the main
path:
READ
SHARE
ACCESS
Some commands and statements treat one of the optional choices as a default value. This
value is UNDERLINED. It is presumed to be selected if you type the command or statement
without specifying one of the options.
Strings
String literals appear in apostrophes:
'msgtext '
Abbreviations
If a keyword or a reserved word has a valid abbreviation, the unabbreviated form always
appears on the main path. The shortest valid abbreviation appears beneath.
132
Teradata JSON

SHOW
CONTROLS
CONTROL
In the above syntax, the following formats are valid:

SHOW CONTROLS
SHOW CONTROL
Loops
A loop is an entry or a group of entries that you can repeat one or more times. Syntax
diagrams show loops as a return path above the main path, over the item or items that you
can repeat:
,
,
(
cname
3
4
)
Read loops from right to left.

The following conventions apply to loops:
Item
Description
Example
maximum
number of
entries allowed
The number appears in a circle on the return path.
In the example, you may type cname a maximum of

four times.
minimum
number of
entries allowed
The number appears in a square on the return path.
In the example, you must type at least three groups

of column names.
separator
The character appears on the return path.
character
If the diagram does not show a separator character,
required
use one blank space.
between entries
In the example, the separator character is a comma.
delimiter
The beginning and end characters appear outside
character
the return path.
required around
Generally, a space is not needed between delimiter
entries
characters and entries.
In the example, the delimiter characters are the left

and right parentheses.
Excerpts
Sometimes a piece of a syntax phrase is too large to fit into the diagram. Such a phrase is
indicated by a break in the path, marked by (|) terminators on each side of the break. The
name for the excerpted piece appears between the terminators in boldface type.
The boldface excerpt name and the excerpted phrase appears immediately after the main
diagram. The excerpted phrase starts and ends with a plain horizontal line:
Teradata JSON
133

LOCKING
excerpt
HAVING
con
excerpt
where_cond
,
cname
,
col_pos
Multiple Legitimate Phrases

In a syntax diagram, it is possible for any number of phrases to be legitimate:
dbname
DATABASE
tname
TABLE
vname
VIEW
In this example, any of the following phrases are legitimate:

dbname
DATABASE dbname
tname
TABLE tname
vname
VIEW vname
134
Teradata JSON

Character Shorthand Notation Used in This Book
Sample Syntax Diagram

,
CREATE VIEW
viewname
AS
cname
CV
LOCKING
LOCK
ACCESS
dbname
A
DATABASE
FOR
SHARE
IN
tname
READ
TABLE
WRITE
EXCLUSIVE
vname
VIEW
EXCL
,
SEL
B
MODE
expr
,
FROM
qual_cond
tname
.aname
C
HAVING cond
;
qual_cond
,
WHERE cond
GROUP BY
cname
,
col_pos

This book uses the Unicode naming convention for characters. For example, the lowercase
character a is more formally specified as either LATIN CAPITAL LETTER A or U+0041.
The U+xxxx notation refers to a particular code point in the Unicode standard, where xxxx
stands for the hexadecimal representation of the 16-bit value defined in the standard.
In parts of the book, it is convenient to use a symbol to represent a special character, or a
particular class of characters. This is particularly true in discussion of the following Japanese
character encodings:
KanjiEBCDIC
KanjiEUC
KanjiShift-JIS
These encodings are further defined in International Character Set Support.
Character Symbols
The symbols, along with character sets with which they are used, are defined in the following
table.
Teradata JSON
Symbol
Encoding
Meaning
a-z
A-Z
Any
Any single byte Latin letter or digit.
135

Symbol
Encoding
Meaning
a-z
A-Z
0-9
Any
Any fullwidth Latin letter or digit.
<
KanjiEBCDIC
Shift Out [SO] (0x0E).
0-9
Indicates transition from single to multibyte character in

KanjiEBCDIC.
>
KanjiEBCDIC
Shift In [SI] (0x0F).

Indicates transition from multibyte to single byte KanjiEBCDIC.
Any
Any multibyte character.

The encoding depends on the current character set.
For KanjiEUC, code set 3 characters are always preceded by ss3.
Any
Any single byte Hankaku Katakana character.

In KanjiEUC, it must be preceded by ss2, forming an individual
multibyte character.
Any
Represents the graphic pad character.
Any
Represents a single or multibyte pad character, depending on

context.
ss2
KanjiEUC
Represents the EUC code set 2 introducer (0x8E).
ss3
KanjiEUC
Represents the EUC code set 3 introducer (0x8F).
For example, string TEST, where each letter is intended to be a fullwidth character, is
written as TEST. Occasionally, when encoding is important, hexadecimal representation is
used.
For example, the following mixed single byte/multibyte character data in KanjiEBCDIC
character set
LMN<TEST>QRS
is represented as:
D3 D4 D5 0E 42E3 42C5 42E2 42E3 0F D8 D9 E2
Pad Characters
The following table lists the pad characters for the various character data types.
136
Server Character Set
Pad Character Name
Pad Character Value
LATIN
SPACE
0x20
UNICODE
SPACE
U+0020
GRAPHIC
IDEOGRAPHIC SPACE
U+3000
Teradata JSON

Teradata JSON
Server Character Set
Pad Character Name
Pad Character Value
KANJISJIS
ASCII SPACE
0x20
KANJI1
ASCII SPACE
0x20
137

138
Teradata JSON
APPENDIX B
External Representations for the JSON Type
Data Type Encoding

Some of the client and server interfacing parcels such as DataInfo (parcel flavor 71),
DataInfoX (parcel flavor 146), PrepInfo (parcel flavor 86), PrepInfoX (parcel flavor 125) and
StatementInfo (parcel flavor 169) return the data type of the field. All parcels that contain
the data type information use the following encoding for the JSON data type. The encoding
numbers defined follow the pattern for existing data types. For example, Nullable number is
non-Nullable value + 1 and stored procedure IN parameter type number is 500 + nonnullable number.
Data Type
NULL Property
Stored Procedure Parameter Type
Non-nullable
Nullable
IN
INOUT
OUT
Large JSON Data

Inline
880
881
1380
1381
1382
Large JSON Data

Locator
884
885
1384
1385
1386
Large JSON Data

Deferred
888
889
1388
1389
1390
These codes are sent from server to client, and are accepted by server from client in the
parcels described in the following sections. The only restriction is the type may not be used
in the USING clause. VARCHAR/CLOB can be used in the USING clause and when
necessary, this data is implicitly cast to the JSON type.
SQL Capabilities Parcel

The SQL Capabilities parcel includes a flag called SQLCap_JSON which indicates whether or
not the JSON data type is supported in the database.
typedef
struct PclCfgSQLCapFeatType {
PclCfgFeatureType
PclCfgFeatureLenType
byte /* 0 */
byte /* 1 */
.
.
.
byte /* 20 */
Teradata JSON
SQLCap_Feature;
SQLCap_Length;
SQLCap_UPSERT;
SQLCap_ArraySupport;
padbyte_boolean;
139
Appendix B External Representations for the JSON Type

Database Limits (ConfigResponse) Parcel
byte /* 21 */
} PclCfgSQLCapFeatType;
SQLCap_JSON;
The SQLCap_JSON flag has the following values:

0 indicates that the JSON data type is not supported.
1 indicates that the JSON data type is supported.
Database Limits (ConfigResponse) Parcel

The Database Limits parcel includes a flag that indicates the maximum length in bytes of the
JSON data type. This value is 16776192.
7
+0 (char)
'0'
'1'
+1 (byte)
+2 (16-bit int)
0
+4 (32-bit int)
+8 (32-bit int)
.
.
.
+124 (32-bit int)
+128 (32-bit int)
-Database-limit
-Maximum parcel size
-32767
-65535
UNUSED
-Maximum segment count
-Segments not supported
-Maximum segment size
-Max. avail. bytes in perm row
-Max. avail. bytes in response row

-Max. bytes in a JSON
StatementInfo Parcel
The following fields of the StatementInfo Parcel contain information relevant to a particular
instance of the JSON data type and show typical values that are expected for the JSON data
type:
Data Type Code = JSON Data Type, according to the table in the Data Type Encoding
section.
UDT indicator = 0 (JSON data type is treated as a system built-in type)
Fully qualified type name length = 0
Fully qualified type name = ""
Field Size = the maximum possible length in bytes for this particular JSON instance
Character Set Code = 1 or 2, depending on the character set for this particular JSON
instance
Maximum number of characters = number of characters of the JSON column - the same
as number in the column definition
Case Sensitive: 'Y' (JSON is case specific)
Example: Metadata Parcel Sequence

This example shows a statement and the associated Metadata Parcel Sequence.
140
Teradata JSON

Consider the following table, data, and query.

CREATE SET TABLE jsontab ,NO FALLBACK ,
NO BEFORE JOURNAL,
NO AFTER JOURNAL,
CHECKSUM = DEFAULT,
DEFAULT MERGEBLOCKRATIO
(
id INTEGER,
jsoncol JSON(543000) CHARACTER SET UNICODE)
PRIMARY INDEX ( id );
INSERT INTO jsontab VALUES(1, NEW JSON('{"name":"Cameron"}', LATIN));
SELECT jsoncol FROM jsontab;
When executing the SELECT statement, the StatementInfo parcel looks like the following:
Database Name
test_db
Table/View Name
jsontab
Column Name
jsoncol
Column Index
As Name
Title
jsoncol
Format
Default Value
Is Identity Column
Is Definitely Writable
Is Nullable
Is Searchable
Is Writable
Data Type Code
881 (JSON Nullable)
UDT Indicator
UDT Name
UDT Misc
Teradata JSON
Max Data Length
1086000 (for UTF16 Session Character Set; 543000 * 2 [export

factor])
Digits
Interval Digits
Fractional Digits
Character Set Code
141

Max Number of Characters
543000
Is CaseSensitive
Is Signed
Is Key Column
Is Unique
Is Expression
Is Sortable
Parameter Type
Struct Depth
Is Temporal Column
UDT Attribute Name
142
Server Data Type Code
Array Number of Dims
Teradata JSON

1150

Uploaded by

Copyright:

Available Formats

1150

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1150

Uploaded by

Copyright:

Available Formats

Teradata Database

Copyright 2014 by Teradata. All Rights Reserved.

Supported Software Releases and Operating

Changes to this Book

Teradata Database 15.0

Teradata Database 15.0

Added the following information:

Added information about extracting JSON

Use the Teradata Information Products Publishing Library site to:

The Teradata home page provides links to numerous sources of

Teradata Customer Education delivers training that builds skills and

Use Teradata @ Your Service to access Orange Books, technical alerts,

Teradata Developer Exchange provides articles on using Teradata

Product Safety Information

Teradata Database Optional Features

The JSON Data Type

Teradata Support for JSON

Client Support for JSON

JSON Support Provided

Full native DBS support.

Chapter 1 The JSON Data Type

JSON Support Provided

VARCHAR values or CLOB values can be inserted into JSON

.NET Data Provider

The JSON keyword cannot be used in the USING data statement;

Chapter 1 The JSON Data Type

Chapter 1 The JSON Data Type

Overview of the JSON Data Type

JSON Data Type Syntax

Chapter 1 The JSON Data Type

be greater than the maximum of 16776192 LATIN characters or 8388096 UNICODE

Maximum Length of a JSON Instance

Chapter 1 The JSON Data Type

Examples: Creating Tables with JSON Columns with No Maximum Length

Chapter 1 The JSON Data Type

Example: Creating a Table with JSON Columns

Examples: Creating a Function with JSON Type Parameters

Chapter 1 The JSON Data Type

Supported Character Sets

Chapter 1 The JSON Data Type

Character Set Handling

About JSON Type Constructor

NEW JSON Constructor Expression

Chapter 1 The JSON Data Type

Examples: JSON Constructor with Arguments

About JSON Type Transform

Transform Input/Output String Format

JSON String Syntax, on page 30

About JSON Type Cast

Chapter 1 The JSON Data Type

About JSON Type Ordering

About JSON Type Usage

Chapter 1 The JSON Data Type

Using JSON Type with UDFs

Example: UDF Parameter Style TD_GENERAL

Chapter 1 The JSON Data Type

Example: SQL UDF with RETURN clause

Using the JSON Type with Procedures (External Form)

Chapter 1 The JSON Data Type

Example: Procedure with Parameter Style TD_General