Table Functions¶

A table function returns a set of rows for each input row. The returned set can contain zero, one, or more rows. Each row can contain one or more columns.

Table functions are sometimes called “tabular functions”.

In this Topic:

What Are Table Functions?
Using a Table Function
- Using a Table Function in the FROM clause
- Using a Table as Input to a Table Function
List of System-Defined Table Functions
Syntax
Usage Notes

What Are Table Functions?¶

Table functions are typically used when a function returns multiple rows for each individual input.

Each time that a table function is called, it can return a different number of rows. For example, a function record_high_temperatures_for_date(), which returns a list of record high temperatures for a specified date, might return 0 rows on April 10, 1 row on June 10, and 40 rows on August 20.

Simple Examples of Table Functions¶

The following are appropriate as table functions:

A function that accepts an account number and a date, and returns all charges billed to that account on that date. (More than one charge might have been billed on a particular date.)
A function that accepts a user ID and returns the database roles assigned to that user. (A user might have multiple roles, including “sysadmin” and “useradmin”.)

Functions in Which Each Output Row Depends Upon Multiple Input Rows¶

Table functions can be grouped into two categories based on the number of input rows that affect each output row:

1-to-N
M-to-N

The functions described earlier are 1-to-N table functions: each output row depends upon only one input row. For example, a function record_high_temperatures_for_date() might produce multiple output rows (one for each city that hit a record on that date). Each output row for a specific input date depends only on that date; each output row is independent of the rows for every other date.

Snowflake also supports M-to-N table functions: each output row can depend upon multiple input rows. For example, if a function generates a moving average of stock prices, that function uses stock prices from multiple input rows (multiple dates) to generate each output row.

More generally, in an M-to-N function, a group of M input rows produces a group of N output rows. M can be one or more rows. N can be zero, one, or more rows.

For example, in a 10-day moving average, M is 10. N is 1 because each group of 10 input rows produces one average price.

Built-in Table Functions vs User-Defined Table Functions¶

Snowflake provides hundreds of built-in functions, many of which are table functions. Built-in table functions are listed in System-Defined Table Functions.

Users can also write their own functions, called user-defined functions or “UDFs”. Some UDFs are scalar; some are tabular. User-defined table functions are called “UDTFs”. For information about UDFs (including UDTFs), see UDFs (User-Defined Functions).

Built-in table functions and user-defined table functions generally follow the same rules; for example, they are called the same way from SQL statements.

Using a Table Function¶

Using a Table Function in the FROM clause¶

A table contains a set of rows. Similarly, a table function returns a set of rows. Both tables and table functions are used in contexts that expect a set of rows. Specifically, table functions are used in the FROM clause of a SQL statement.

To help the SQL compiler recognize a table function as a source of rows, Snowflake requires that the table function call be wrapped by the TABLE() keyword.

For example, the following statement calls a table function named record_high_temperatures_for_date(), which takes a DATE value as an argument:

SELECT city_name, temperature
    FROM TABLE(record_high_temperatures_for_date('2021-06-27'::DATE))
    ORDER BY city_name;

For more information about the syntax of TABLE(), see Table Literals.

Table functions, like functions in general, can accept zero, one, or multiple input arguments in each invocation. Each argument must be a scalar expression.

For more details about the syntax of table function calls, see Syntax (in this topic).

Using a Table as Input to a Table Function¶

The argument to a table function can be a literal or an expression, such as a column of a table. For example, the SELECT statement below passes values from a table as arguments to a table function:

create table dates_of_interest (d date);
insert into dates_of_interest (d) values
      ('2021-06-21'::DATE),
      ('2022-06-21'::DATE);

create function record_high_temperatures_for_date(d date)
    returns table (city varchar, temperature float)
    ... ;

select doi.date as "Date", record_temperatures.city, record_temperatures.temperature
    from dates_of_interest as doi,
         table(record_high_temperatures_for_date(doi.date)) as record_temperatures;

The arguments to a table function can come from other table-like sources, including views and other table functions.

List of System-Defined Table Functions¶

Snowflake provides the following system-defined (i.e. built-in) table functions:

Sub-category	Function	Notes
Data Loading	INFER_SCHEMA	For more information, see Loading Data into Snowflake.
	VALIDATE
Data Generation	GENERATOR
Data Conversion	SPLIT_TO_TABLE
	STRTOK_SPLIT_TO_TABLE
Object Modeling	GET_OBJECT_REFERENCES
Semi-structured Queries	FLATTEN	For more information, see Querying Semi-structured Data.
Query Results	RESULT_SCAN	Can be used to perform SQL operations on the output from another SQL operation (e.g. SHOW).
Historical & Usage Information (Snowflake Information Schema, Account Usage):
User Login	LOGIN_HISTORY , LOGIN_HISTORY_BY_USER
Queries	QUERY_HISTORY , QUERY_HISTORY_BY_*
Warehouse & Storage Usage	DATABASE_STORAGE_USAGE_HISTORY
	WAREHOUSE_LOAD_HISTORY
	WAREHOUSE_METERING_HISTORY
	STAGE_STORAGE_USAGE_HISTORY
Column-level & Row-level Security	POLICY_REFERENCES
Object Tagging	TAG_REFERENCES	Information Schema table function.
	TAG_REFERENCES_ALL_COLUMNS	Information Schema table function.
	TAG_REFERENCES_WITH_LINEAGE	Account Usage table function.
Database Replication	DATABASE_REFRESH_HISTORY	For more information, see Replicating Databases Across Multiple Accounts.
	DATABASE_REFRESH_PROGRESS , DATABASE_REFRESH_PROGRESS_BY_JOB
	REPLICATION_USAGE_HISTORY
Data Loading & Transfer	COPY_HISTORY
	DATA_TRANSFER_HISTORY
	PIPE_USAGE_HISTORY
	STAGE_DIRECTORY_FILE_REGISTRATION_HISTORY
	VALIDATE_PIPE_LOAD
Data Clustering (within Tables)	AUTOMATIC_CLUSTERING_HISTORY	For more information, see Automatic Clustering.
External Functions	EXTERNAL_FUNCTIONS_HISTORY	For more information, see External Functions.
External Tables	AUTO_REFRESH_REGISTRATION_HISTORY	For more information, see Working with External Tables.
	EXTERNAL_TABLE_FILES
	EXTERNAL_TABLE_FILE_REGISTRATION_HISTORY
Materialized Views Maintenance	MATERIALIZED_VIEW_REFRESH_HISTORY	For more information, see Working with Materialized Views.
SCIM Maintenance	REST_EVENT_HISTORY	For more information, see Auditing with SCIM
Search Optimization Maintenance	SEARCH_OPTIMIZATION_HISTORY	For more information, see Using the Search Optimization Service.
Tasks	COMPLETE_TASK_GRAPHS	For more information, see Executing SQL Statements on a Schedule Using Tasks.
	CURRENT_TASK_GRAPHS
	SERVERLESS_TASK_HISTORY
	TASK_DEPENDENTS
	TASK_HISTORY

Syntax¶

SELECT ...
  FROM [ <input_table> [ [AS] <alias_1> ] ,
         [ LATERAL ]
       ]
       TABLE( <table_function>( [ <arg_1> [, ... ] ] ) ) [ [ AS ] <alias_2> ];

For function-specific syntax, see the documentation for the individual system-defined table functions.

Usage Notes¶

Table functions can also be applied to a set of rows using the LATERAL construct.
To enable using table expressions, Snowflake supports ANSI/ISO standard syntax for table expressions in the FROM clause of queries and subqueries. This syntax is used to indicate that an expression returns a collection of rows instead of a single row.
This ANSI/ISO syntax is valid only in the FROM clause of the SELECT list. You cannot omit these keywords and parentheses from a collection subquery specification in any other context.