Direct Path Loading Functions

Table 17-4 lists the direct path loading functions that are described in this section.

Table 17-4 Direct Path Loading Functions

Function Purpose

"OCIDirPathAbort()"

Terminate a direct path operation

"OCIDirPathColArrayEntryGet()"

Get a specified entry in a column array

"OCIDirPathColArrayEntrySet()"

Set a specified entry in a column array to a specific value

"OCIDirPathColArrayReset()"

Reset the row array state

"OCIDirPathColArrayRowGet()"

Get the base row pointers for a specified row number

"OCIDirPathColArrayToStream()"

Convert from a column array to a direct path stream format

"OCIDirPathDataSave()"

Do a data savepoint, or commit the loaded data and finish the load operation

"OCIDirPathFinish()"

Finish and commit the loaded data

"OCIDirPathFlushRow()"

Deprecated.

"OCIDirPathLoadStream()"

Load the data converted to direct path stream format

"OCIDirPathPrepare()"

Prepare direct path interface to convert or load rows

"OCIDirPathStreamReset()"

Reset the direct path stream state


OCIDirPathAbort()

Purpose

Terminates a direct path operation.

Syntax

sword OCIDirPathAbort   ( OCIDirPathCtx          *dpctx,
                          OCIError               *errhp ); 

Parameters

dpctx (IN)

Direct path context handle.

errhp (IN/OUT)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

Comments

All state that was maintained by the server on behalf of the direct path operation is destroyed by a termination. For a direct path load, the data loaded before the terminate operation is not visible to any queries. However, the data may still consume space in the segments that are being loaded. Any load completion operations, such as index maintenance operations, are not performed.

OCIDirPathColArrayEntryGet()

Purpose

Gets a specified entry in a column array.

Syntax

sword OCIDirPathColArrayEntryGet ( OCIDirPathColArray   *dpca,
                                   OCIError             *errhp,
                                   ub4                  rownum,
                                   ub2                  colIdx,
                                   ub1                  **cvalpp,
                                   ub4                  *clenp,
                                   ub1                  *cflgp );

Parameters

dpca (IN/OUT)

Direct path column array handle.

errhp (IN)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

rownum (IN)

Zero-based row offset.

colIdx (IN)

The column's index used when building the column parameter list.

cvalpp (IN/OUT)

Pointer to pointer to column data.

clenp (IN/OUT)

Pointer to length of column data.

cflgp (IN/OUT)

Pointer to column flag.

One of these values is returned:

  • OCI_DIRPATH_COL_COMPLETE - All data for the column is present.

  • OCI_DIRPATH_COL_NULL - Column is NULL.

  • OCI_DIRPATH_COL_PARTIAL - Partial column data is being supplied.

Comments

If cflgp is set to OCI_DIRPATH_COL_NULL, the cvalpp and clenp parameters are not set by this operation.

OCIDirPathColArrayEntrySet()

Purpose

Sets a specified entry in a column array to the supplied values.

Syntax

sword OCIDirPathColArrayEntrySet ( OCIDirPathColArray   *dpca,
                                   OCIError             *errhp,
                                   ub4                  rownum,
                                   ub2                  colIdx,
                                   ub1                  *cvalp,
                                   ub4                  clen,
                                   ub1                  cflg );

Parameters

dpca (IN/OUT)

Direct path column array handle.

errhp (IN)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

rownum (IN)

Zero-based row offset.

colIdx (IN)

The column's index used when building the column parameter list.

cvalp (IN)

Pointer to column data.

clen (IN)

Length of column data.

cflg (IN)

Column flag. One of these values is returned:

  • OCI_DIRPATH_COL_COMPLETE - All data for the column is present.

  • OCI_DIRPATH_COL_NULL - Column is NULL.

  • OCI_DIRPATH_COL_PARTIAL - Partial column data is being supplied.

Comments

If cflg is set to OCI_DIRPATH_COL_NULL, the cvalp and clen parameters are not used.

Example

This example sets the source of data for the first row in a column array to addr, with a length of len. In this example, the column is identified by colId.

err = OCIDirPathColArrayEntrySet(dpca, errhp, (ub2)0, colId, addr, len,
          OCI_DIRPATH_COL_COMPLETE);

OCIDirPathColArrayReset()

Purpose

Resets the column array state.

Syntax

sword OCIDirPathColArrayReset ( OCIDirPathColArray   *dpca,
                                OCIError             *errhp );

Parameters

dpca (IN)

Direct path column array handle.

errhp (IN)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

Comments

Resetting the column array state is necessary when piecing in a large column and an error occurs in the middle of loading the column. Do not reset the column array if the last OCIDirPathColArrayReset() call returned OCI_NEED_DATA or OCI_CONTINUE. That is, you are in the middle of a row conversion. Use OCI_DIRPATH_COL_ERROR to purge the current row for OCI_NEED_DATA.

OCIDirPathColArrayRowGet()

Purpose

Gets the column array row pointers for a given row number.

Syntax

sword OCIDirPathColArrayRowGet ( OCIDirPathColArray   *dpca,
                                 OCIError             *errhp,
                                 ub4                  rownum,
                                 ub1                  ***cvalppp,
                                 ub4                  **clenpp,
                                 ub1                  **cflgpp );

Parameters

dpca (IN/OUT)

Direct path column array handle.

errhp (IN)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

rownum (IN)

Zero-based row offset

cvalppp (IN/OUT)

Pointer to vector of pointers to column data

clenpp (IN/OUT)

Pointer to vector of column data lengths

cflgpp (IN/OUT)

Pointer to vector of column flags

Comments

Returns pointers to column array entries for the given row. This allows the application to do simple pointer arithmetic to iterate across the columns of the specific row. You can use this interface to efficiently get or set the column array entries of a row, as opposed to calling OCIDirPathColArrayEntrySet() for every column. The application is also responsible for not dereferencing memory beyond the column array boundaries. The dimensions of the column array are available as attributes of the column array.

OCIDirPathColArrayToStream()

Purpose

Converts from column array format to a direct path stream format.

Syntax

sword OCIDirPathColArrayToStream ( OCIDirPathColArray     *dpca,
                                   OCIDirPathCtx  const   *dpctx,
                                   OCIDirPathStream       *dpstr,
                                   OCIError               *errhp, 
                                   ub4                    rowcnt,
                                   ub4                    rowoff );

Parameters

dpca (IN)

Direct path column array handle.

dpctx (IN)

Direct path context handle for the object being loaded.

dpstr (IN/OUT)

Direct path stream handle.

errhp (IN)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

rowcnt (IN)

Number of rows in the column array.

rowoff (IN)

Starting index in the column array.

Comments

This interface is used to convert a column array representation of data in its external format to a direct path stream format. The converted format is suitable for loading with OCIDirPathLoadStream().

The column data in direct path stream format is converted to its Oracle Database internal representation. All conversions are done on the client side of the two-task interface; all conversion errors occur synchronously with the call to this interface. Information concerning which row and column an error occurred on is available as an attribute of the column array handle.

Note that in a threaded environment, concurrent OCIDirPathColArrayToStream() operations can be referencing the same direct path context handle. However, the direct path context handle is not modified by this interface.

The return codes for this call are:

  • OCI_SUCCESS - All data in the column array was successfully converted to stream format. The column array attribute OCI_ATTR_ROW_COUNT is the number of rows processed.

  • OCI_ERROR - An error occurred during conversion; the error handle contains the error information. The column array attribute OCI_ATTR_ROW_COUNT is the number of rows successfully converted in the last call. The attribute OCI_ATTR_COL_COUNT contains the column index into the column array for the column that caused the error. A stream must always be loaded after column array to stream conversion returns OCI_ERROR. It cannot be reset or converted to until it is loaded.

  • OCI_CONTINUE - Not all of the data in the column array could be converted to stream format. The stream buffer is not large enough to contain all of the column array data. The caller should either load the data, save the data to a file, or use another stream and call OCIDirPathColArrayToStream() again to convert the remainder of the column array data. The column array attribute OCI_ATTR_ROW_COUNT is the number of rows successfully converted in the last call. The row offset must be updated for the next conversion; internal state does keep track of the column to continue conversion from. The OCI_ATTR_ROW_COUNT value must be added to the previous row offset by the caller.

  • OCI_NEED_DATA - All of the data in the column array was successfully converted, but a partial column was encountered. The caller should load the resulting stream, and supply the remainder of the row, iteratively if necessary. The column array attribute OCI_ATTR_ROW_COUNT is the number of rows successfully converted in the last call. The attribute OCI_ATTR_COL_COUNT contains the column index into the column array for the column that is marked partial.

OCIDirPathDataSave()

Purpose

Depending on the action requested, does a data savepoint, or commits the loaded data and finishes the direct path load operation.

Syntax

sword OCIDirPathDataSave ( OCIDirPathCtx          *dpctx,
                           OCIError               *errhp,
                           ub4                    action  ); 

Parameters

dpctx (IN)

Direct path context handle for the object loaded.

errhp (IN/OUT)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

action (IN)

Values for action parameter to OCIDirPathDataSave() are as follows:

  • OCI_DIRPATH_DATASAVE_SAVEONLY - To execute a data savepoint only

  • OCI_DIRPATH_DATASAVE_FINISH - To commit the loaded data and call the direct finishing function

Comments

A return value of OCI_SUCCESS indicates that the backend has properly executed a data savepoint or executed the finishing logic.

Executing a data savepoint is not allowed for LOBs.

Executing the finishing logic is different from properly terminating the load, because resources allocated are not freed.

OCIDirPathFinish()

Purpose

Finishes the direct path load operation.

Syntax

sword OCIDirPathFinish (   OCIDirPathCtx          *dpctx,
                           OCIError               *errhp ); 

Parameters

dpctx (IN)

Direct path context handle for the object loaded.

errhp (IN/OUT)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

Comments

After the load has completed, and the loaded data is to be committed, the direct path finishing function is called. Finish is not allowed until all streams have been loaded, and there is not a partially loaded row.

A return value of OCI_SUCCESS indicates that the backend has properly terminated the load.

OCIDirPathFlushRow()

Purpose

Flushes a partially loaded row from the server. This function is deprecated.

Syntax

sword OCIDirPathFlushRow (   OCIDirPathCtx          *dpctx,
                             OCIError               *errhp ); 

Parameters

dpctx (IN)

Direct path context handle for the object loaded.

errhp (IN/OUT)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

Comments

This function is necessary when part of a row is loaded, but a conversion error occurs on the next piece being processed by the application. Only the row currently in partial state is discarded. If the server is not currently processing a partial row for the object associated with the direct path context, this function does nothing.

OCIDirPathLoadStream()

Purpose

Loads the data converted to direct path stream format.

Syntax

sword OCIDirPathLoadStream (   OCIDirPathCtx          *dpctx,
                               OCIDirPathStream       *dpstr,
                               OCIError               *errhp ); 

Parameters

dpctx (IN)

Direct path context handle for the object loaded.

dpstr (IN)

Direct path stream handle for the stream to load.

errhp (IN)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

Comments

When the interface returns an error, information concerning the row in the column array that sourced the stream can be obtained as an attribute of the direct path stream. Also, the offset into the stream where the error occurred can be obtained as an attribute of the stream.

Return codes for this function are:

  • OCI_SUCCESS - All data in the stream was successfully loaded.

  • OCI_ERROR - An error occurred while loading the data. The problem could be a partition mapping error, a NULL constraint violation, a function-based index evaluation error, or an out of space condition, such as cannot allocate extent. OCI_ATTR_ROW_COUNT is the number of rows successfully loaded in the last call.

  • OCI_NEED_DATA - Last row was not complete. The caller must supply another row piece. If the stream was sourced from a column array, the attribute OCI_ATTR_ROW_COUNT is the number of complete rows successfully loaded in the last call.

  • OCI_NO_DATA - Attempt to load an empty stream or a stream that has been completely processed.

A stream must be repeatedly loaded until OCI_SUCCESS, OCI_NEED_DATA, or OCI_NO_DATA is returned. For example, a stream cannot be reset if OCI_ERROR is returned from OCIDirPathLoadStream().

OCIDirPathPrepare()

Purpose

Prepares the direct path load interface before any rows can be converted or loaded.

Syntax

sword OCIDirPathPrepare (   OCIDirPathCtx          *dpctx,
                            OCISvcCtx              *svchp,
                            OCIError               *errhp ); 

Parameters

dpctx (IN)

Direct path context handle for the object loaded.

svchp (IN)

Service context.

errhp (IN/OUT)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

Comments

After the name of the object to be operated on is set, the external attributes of the column data are set, and all load options are set, the direct path interface must be prepared with OCIDirPathPrepare() before any rows can be converted or loaded.

A return value of OCI_SUCCESS indicates that the backend has been properly initialized for a direct path load operation. A nonzero return indicates an error. Possible errors are:

  • Invalid context

  • Not connected to a server

  • Object name not set

  • Already prepared (cannot prepare twice)

  • Object not suitable for a direct path operation

OCIDirPathStreamReset()

Purpose

Resets the direct path stream state.

Syntax

sword OCIDirPathStreamReset ( OCIDirPathStream       *dpstr,
                              OCIError               *errhp );

Parameters

dpstr (IN)

Direct path stream handle.

errhp (IN)

An error handle that you can pass to OCIErrorGet() for diagnostic information when there is an error.

Comments

A direct path stream maintains the state that indicates where the next OCIDirPathColArrayToStream() call should start writing into the stream. Normally, data is appended to the end of the stream. A stream cannot be reset until it is successfully loaded (the loading returned OCI_SUCCESS, OCI_NEED_DATA, or OCI_NO_DATA).