Reference v11
The sections that follow describe ecpgPlus language elements:
- C-Preprocessor Directives
- Supported C Data Types
- Type Codes
- The SQLDA Structure
- ECPGPlus Statements
C-preprocessor Directives
The ECPGPlus C-preprocessor enforces two behaviors that are dependent on the mode in which you invoke ECPGPlus:
PROC
mode- non-
PROC
mode
Compiling in PROC Mode
In PROC
mode, ECPGPlus allows you to:
- Declare host variables outside of an
EXEC SQL BEGIN/END DECLARE SECTION
. - Use any C variable as a host variable as long as it is of a data type compatible with ECPG.
When you invoke ECPGPlus in PROC
mode (by including the -C PROC
keywords), the ECPG compiler honors the following C-preprocessor directives:
Pre-processor directives are used to effect or direct the code that is received by the compiler. For example, using the following code sample:
If you invoke ECPGPlus with the following command-line arguments:
ECPGPlus will copy the entire fragment (without change) to the output file, but will only send the following tokens to the ECPG parser:
On the other hand, if you invoke ECPGPlus with the following command-line arguments:
The ECPG parser will receive the following tokens:
If your code uses preprocessor directives to filter the code that is sent to the compiler, the complete code is retained in the original code, while the ECPG parser sees only the processed token stream.
You can also use compatible syntax when executing the following preprocessor directives with an EXEC
directive:
For example, if your code includes the following:
If you invoke ECPGPlus with the following command-line arguments:
ECPGPlus will send the following tokens to the output file, and the ECPG parser:
Note
The EXEC ORACLE
pre-processor directives only work if you specify -C PROC
on the ECPG command line.
Using the SELECT_ERROR Precompiler Option
When using ECPGPlus in compatible mode, you can use the SELECT_ERROR
precompiler option to instruct your program how to handle result sets that contain more rows than the host variable can accommodate. The syntax is:
The default value is YES
; a SELECT
statement will return an error message if the result set exceeds the capacity of the host variable. Specify NO
to instruct the program to suppress error messages when a SELECT
statement returns more rows than a host variable can accommodate.
Use SELECT_ERROR
with the EXEC ORACLE OPTION
directive.
Compiling in non-PROC Mode
If you do not include the -C PROC
command-line option:
- C preprocessor directives are copied to the output file without change.
- You must declare the type and name of each C variable that you intend to use as a host variable within an
EXEC SQL BEGIN/END DECLARE
section.
When invoked in non-PROC
mode, ECPG implements the behavior described in the PostgreSQL Core documentation.
Supported C Data Types
An ECPGPlus application must deal with two sets of data types: SQL data types (such as SMALLINT
, DOUBLE PRECISION
and CHARACTER VARYING
) and C data types (like short
, double
and varchar[n]
). When an application fetches data from the server, ECPGPlus will map each SQL data type to the type of the C variable into which the data is returned.
In general, ECPGPlus can convert most SQL server types into similar C types, but not all combinations are valid. For example, ECPGPlus will try to convert a SQL character value into a C integer value, but the conversion may fail (at execution time) if the SQL character value contains non-numeric characters. The reverse is also true; when an application sends a value to the server, ECPGPlus will try to convert the C data type into the required SQL type. Again, the conversion may fail (at execution time) if the C value cannot be converted into the required SQL type.
ECPGPlus can convert any SQL type into C character values (char[n]
or varchar[n])
. Although it is safe to convert any SQL type to/from char[n]
or varchar[n]
, it is often convenient to use more natural C types such as int
, double
, or float
.
The supported C data types are:
short
int
unsigned int
long long int
float
double
char[n+1]
varchar[n+1]
bool
- and any equivalent created by a
typedef
In addition to the numeric and character types supported by C, the pgtypeslib
run-time library offers custom data types (and functions to operate on those types) for dealing with date/time and exact numeric values:
timestamp
interval
date
decimal
numeric
To use a data type supplied by pgtypeslib
, you must #include
the proper header file.
Type Codes
The following table contains the type codes for external data types. An external data type is used to indicate the type of a C host variable. When an application binds a value to a parameter or binds a buffer to a SELECT
-list item, the type code in the corresponding SQLDA descriptor (descriptor->T[column])
should be set to one of the following values:
Type Code | Host Variable Type (C Data Type) |
---|---|
1, 2, 8, 11, 12, 15, 23, 24, 91, 94, 95, 96, 97 | char[] |
3 | int |
4, 7, 21 | float |
5, 6 | null-terminated string (char[length+1]) |
9 | varchar |
22 | double |
68 | unsigned int |
The following table contains the type codes for internal data types. An internal type code is used to indicate the type of a value as it resides in the database. The DESCRIBE SELECT LIST
statement populates the data type array (descriptor->T[column])
using the following values.
Internal Type Code | Server Type |
---|---|
1 | VARCHAR2 |
2 | NUMBER |
8 | LONG |
11 | ROWID |
12 | DATE |
23 | RAW |
24 | LONG RAW |
96 | CHAR |
100 | BINARY FLOAT |
101 | BINARY DOUBLE |
104 | UROWID |
187 | TIMESTAMP |
188 | TIMESTAMP W/TIMEZONE |
189 | INTERVAL YEAR TO MONTH |
190 | INTERVAL DAY TO SECOND |
232 | TIMESTAMP LOCAL_TZ |
The SQLDA Structure
Oracle Dynamic SQL method 4 uses the SQLDA data structure to hold the data and metadata for a dynamic SQL statement. A SQLDA structure can describe a set of input parameters corresponding to the parameter markers found in the text of a dynamic statement or the result set of a dynamic statement. The layout of the SQLDA structure is:
Parameters
N - maximum number of entries
The N
structure member contains the maximum number of entries that the SQLDA may describe. This member is populated by the sqlald()
function when you allocate the SQLDA structure. Before using a descriptor in an OPEN
or FETCH
statement, you must set N
to the actual number of values described.
V - data values
The V
structure member is a pointer to an array of data values.
- For a
SELECT
-list descriptor,V
points to an array of values returned by aFETCH
statement (each member in the array corresponds to a column in the result set). - For a bind descriptor,
V
points to an array of parameter values (you must populate the values in this array before opening a cursor that uses the descriptor).
Your application must allocate the space required to hold each value. Refer to displayResultSet () function for an example of how to allocate space for SELECT
-list values.
L - length of each data value
The L
structure member is a pointer to an array of lengths. Each member of this array must indicate the amount of memory available in the corresponding member of the V
array. For example, if V[5]
points to a buffer large enough to hold a 20-byte NULL-terminated string, L[5]
should contain the value 21 (20 bytes for the characters in the string plus 1 byte for the NULL-terminator). Your application must set each member of the L
array.
T - data types
The T
structure member points to an array of data types, one for each column (or parameter) described by the descriptor.
- For a bind descriptor, you must set each member of the
T
array to tell ECPGPlus the data type of each parameter. - For a
SELECT
-list descriptor, theDESCRIBE SELECT LIST
statement sets each member of theT
array to reflect the type of data found in the corresponding column.
You may change any member of the T
array before executing a FETCH
statement to force ECPGPlus to convert the corresponding value to a specific data type. For example, if the DESCRIBE SELECT LIST
statement indicates that a given column is of type DATE
, you may change the corresponding T
member to request that the next FETCH
statement return that value in the form of a NULL-terminated string. Each member of the T
array is a numeric type code (see Type Codes for a list of type codes). The type codes returned by a DESCRIBE SELECT LIST
statement differ from those expected by a FETCH
statement. After executing a DESCRIBE SELECT LIST
statement, each member of T
encodes a data type and a flag indicating whether the corresponding column is nullable. You can use the sqlnul()
function to extract the type code and nullable flag from a member of the T array. The signature of the sqlnul()
function is as follows:
For example, to find the type code and nullable flag for the third column of a descriptor named results, you would invoke sqlnul()
as follows:
I - indicator variables
The I
structure member points to an array of indicator variables. This array is allocated for you when your application calls the sqlald()
function to allocate the descriptor.
- For a
SELECT
-list descriptor, each member of theI
array indicates whether the corresponding column contains a NULL (non-zero) or non-NULL (zero) value. - For a bind parameter, your application must set each member of the
I
array to indicate whether the corresponding parameter value is NULL.
F - number of entries
The F
structure member indicates how many values are described by the descriptor (the N
structure member indicates the maximum number of values which may be described by the descriptor; F
indicates the actual number of values). The value of the F
member is set by ECPGPlus when you execute a DESCRIBE
statement. F
may be positive, negative, or zero.
- For a
SELECT
-list descriptor,F
will contain a positive value if the number of columns in the result set is equal to or less than the maximum number of values permitted by the descriptor (as determined by theN
structure member); 0 if the statement is not aSELECT
statement, or a negative value if the query returns more columns than allowed by theN
structure member. - For a bind descriptor,
F
will contain a positive number if the number of parameters found in the statement is less than or equal to the maximum number of values permitted by the descriptor (as determined by theN
structure member); 0 if the statement contains no parameters markers, or a negative value if the statement contains more parameter markers than allowed by theN
structure member.
If F
contains a positive number (after executing a DESCRIBE
statement), that number reflects the count of columns in the result set (for a SELECT
-list descriptor) or the number of parameter markers found in the statement (for a bind descriptor). If F
contains a negative value, you may compute the absolute value of F
to discover how many values (or parameter markers) are required. For example, if F
contains -24
after describing a SELECT
list, you know that the query returns 24 columns.
S - column/parameter names
The S
structure member points to an array of NULL-terminated strings.
- For a
SELECT
-list descriptor, theDESCRIBE SELECT LIST
statement sets each member of this array to the name of the corresponding column in the result set. - For a bind descriptor, the
DESCRIBE BIND VARIABLES
statement sets each member of this array to the name of the corresponding bind variable.
In this release, the name of each bind variable is determined by the left-to-right order of the parameter marker within the query - for example, the name of the first parameter is always ?0
, the name of the second parameter is always ?1
, and so on.
M - maximum column/parameter name length
The M
structure member points to an array of lengths. Each member in this array specifies the maximum length of the corresponding member of the S
array (that is, M[0]
specifies the maximum length of the column/parameter name found at S[0]
). This array is populated by the sqlald()
function.
C - actual column/parameter name length
The C
structure member points to an array of lengths. Each member in this array specifies the actual length of the corresponding member of the S
array (that is, C[0]
specifies the actual length of the column/parameter name found at S[0]
).
This array is populated by the DESCRIBE
statement.
X - indicator variable names
The X
structure member points to an array of NULL-terminated strings -each string represents the name of a NULL indicator for the corresponding value.
This array is not used by ECPGPlus, but is provided for compatibility with Pro*C applications.
Y - maximum indicator name length
The Y
structure member points to an array of lengths. Each member in this array specifies the maximum length of the corresponding member of the X
array (that is, Y[0]
specifies the maximum length of the indicator name found at X[0]
).
This array is not used by ECPGPlus, but is provided for compatibility with Pro*C applications.
Z - actual indicator name length
The Z
structure member points to an array of lengths. Each member in this array specifies the actual length of the corresponding member of the X
array (that is, Z[0]
specifies the actual length of the indicator name found at X[0]
).
This array is not used by ECPGPlus, but is provided for compatibility with Pro*C applications.
ECPGPlus Statements
An embedded SQL statement allows your client application to interact with the server, while an embedded directive is an instruction to the ECPGPlus compiler.
You can embed any Advanced Server SQL statement in a C program. Each statement should begin with the keywords EXEC SQL
, and must be terminated with a semi-colon (;). Within the C program, a SQL statement takes the form:
Where sql_command_body
represents a standard SQL statement. You can use a host variable anywhere that the SQL statement expects a value expression. For more information about substituting host variables for value expressions, refer to Declaring Host Variables.
ECPGPlus extends the PostgreSQL server-side syntax for some statements; for those statements, syntax differences are outlined in the following reference sections. For a complete reference to the supported syntax of other SQL commands, refer to the PostgreSQL Core Documentation available at:
https://www.postgresql.org/docs/11/static/sql-commands.html
ALLOCATE DESCRIPTOR
Use the ALLOCATE DESCRIPTOR
statement to allocate an SQL descriptor area:
Where:
array_size
is a variable that specifies the number of array elements to allocate for the descriptor. array_size
may be an INTEGER
value or a host variable.
descriptor_name
is the host variable that contains the name of the descriptor, or the name of the descriptor. This value may take the form of an identifier, a quoted string literal, or of a host variable.
variable_count
specifies the maximum number of host variables in the descriptor. The default value of variable_count
is 100
.
The following code fragment allocates a descriptor named emp_query
that may be processed as an array (emp_array)
:
CALL
Use the CALL
statement to invoke a procedure or function on the server. The CALL
statement works only on Advanced Server. The CALL
statement comes in two forms; the first form is used to call a function:
The second form is used to call a procedure:
Where:
program_name
is the name of the stored procedure or function that the CALL
statement invokes. The program name may be schema-qualified or package-qualified (or both); if you do not specify the schema or package in which the program resides, ECPGPlus will use the value of search_path
to locate the program.
actual_arguments
specifies a comma-separated list of arguments required by the program. Note that each actual_argument
corresponds to a formal argument expected by the program. Each formal argument may be an IN
parameter, an OUT
parameter, or an INOUT
parameter.
:ret_variable
specifies a host variable that will receive the value returned if the program is a function.
:ret_indicator
specifies a host variable that will receive the indicator value returned, if the program is a function.
For example, the following statement invokes the get_job_desc
function with the value contained in the :ename
host variable, and captures the value returned by that function in the :job
host variable:
CLOSE
Use the CLOSE
statement to close a cursor, and free any resources currently in use by the cursor. A client application cannot fetch rows from a closed cursor. The syntax of the CLOSE
statement is:
Where:
cursor_name
is the name of the cursor closed by the statement. The cursor name may take the form of an identifier or of a host variable.
The OPEN
statement initializes a cursor. Once initialized, a cursor result set will remain unchanged unless the cursor is re-opened. You do not need to CLOSE
a cursor before re-opening it.
To manually close a cursor named emp_cursor
, use the command:
A cursor is automatically closed when an application terminates.
COMMIT
Use the COMMIT
statement to complete the current transaction, making all changes permanent and visible to other users. The syntax is:
Where:
database_name
is the name of the database (or host variable that contains the name of the database) in which the work resides. This value may take the form of an unquoted string literal, or of a host variable.
For compatibility, ECPGPlus accepts the COMMENT
clause without error but does not store any text included with the COMMENT
clause.
Include the RELEASE
clause to close the current connection after performing the commit.
For example, the following command commits all work performed on the dept
database and closes the current connection:
By default, statements are committed only when a client application performs a COMMIT
statement. Include the -t
option when invoking ECPGPlus to specify that a client application should invoke AUTOCOMMIT
functionality. You can also control AUTOCOMMIT
functionality in a client application with the following statements:
and
CONNECT
Use the CONNECT
statement to establish a connection to a database. The CONNECT
statement is available in two forms - one form is compatible with Oracle databases, the other is not.
The first form is compatible with Oracle databases:
Where:
user_name
is a host variable that contains the role that the client application will use to connect to the server.
password
is a host variable that contains the password associated with that role.
connection_id
is a host variable that contains a slash-delimited user name and password used to connect to the database.
Include the AT
clause to specify the database to which the connection is established. database_name
is the name of the database to which the client is connecting; specify the value in the form of a variable, or as a string literal.
Include the USING
clause to specify a host variable that contains a null-terminated string identifying the database to which the connection will be established.
The ALTER AUTHORIZATION
clause is supported for syntax compatibility only; ECPGPlus parses the ALTER AUTHORIZATION
clause, and reports a warning.
Using the first form of the CONNECT
statement, a client application might establish a connection with a host variable named user
that contains the identity of the connecting role, and a host variable named password
that contains the associated password using the following command:
A client application could also use the first form of the CONNECT
statement to establish a connection using a single host variable named :connection_id
. In the following example, connection_id
contains the slash-delimited role name and associated password for the user:
The syntax of the second form of the CONNECT
statement is:
Where credentials
is one of the following:
In the second form:
database_name
is the name or identity of the database to which the client is connecting. Specify database_name
as a variable, or as a string literal, in one of the following forms:
Where:
hostname
is the name or IP address of the server on which the database resides.
port
is the port on which the server listens.
You can also specify a value of DEFAULT
to establish a connection with the default database, using the default role name. If you specify DEFAULT
as the target database, do not include a connection_name
or credentials
.
connection_name
is the name of the connection to the database. connection_name
should take the form of an identifier (that is, not a string literal or a variable). You can open multiple connections, by providing a unique connection_name
for each connection.
If you do not specify a name for a connection, ecpglib
assigns a name of DEFAULT
to the connection. You can refer to the connection by name (DEFAULT
) in any EXEC SQL
statement.
CURRENT
is the most recently opened or the connection mentioned in the most-recent SET CONNECTION TO
statement. If you do not refer to a connection by name in an EXEC SQL
statement, ECPG assumes the name of the connection to be CURRENT
.
user_name
is the role used to establish the connection with the Advanced Server database. The privileges of the specified role will be applied to all commands performed through the connection.
password
is the password associated with the specified user_name
.
The following code fragment uses the second form of the CONNECT
statement to establish a connection to a database named edb
, using the role alice
and the password associated with that role, 1safepwd
:
The name of the connection is acctg_conn
; you can use the connection name when changing the connection name using the SET CONNECTION
statement.
DEALLOCATE DESCRIPTOR
Use the DEALLOCATE DESCRIPTOR
statement to free memory in use by an allocated descriptor. The syntax of the statement is:
Where:
descriptor_name
is the name of the descriptor. This value may take the form of a quoted string literal, or of a host variable.
The following example deallocates a descriptor named emp_query
:
DECLARE CURSOR
Use the DECLARE CURSOR
statement to define a cursor. The syntax of the statement is: