EXPORT - medium - ABAP Keyword Documentation

ABAP - Keyword Documentation → ABAP - Reference → Processing External Data → Data Clusters → EXPORT →

EXPORT - medium

Syntax

... { DATA BUFFER xstr }
  | { INTERNAL TABLE itab }
  | { MEMORY ID id }
  | { DATABASE      dbtab(ar) [FROM wa] [CLIENT cl] ID id }
  | { SHARED MEMORY dbtab(ar) [FROM wa] [CLIENT cl] ID id }
  | { SHARED BUFFER dbtab(ar) [FROM wa] [CLIENT cl] ID id } ...

Alternatives:

1. ... DATA BUFFER xstr

2. ... INTERNAL TABLE itab

3. ... MEMORY ID id

4. ... DATABASE dbtab(ar) [FROM wa] [CLIENT cl] ID id

5. ... SHARED MEMORY dbtab(ar) [FROM wa] [CLIENT cl] ID id

6. ... SHARED BUFFER dbtab(ar) [FROM wa] [CLIENT cl] ID id

Effect

The exported data cluster can be stored in a byte string xstr, in an internal table itab, in the ABAP memory, in a database table dbtab, or in a shared memory area (SHARED MEMORY or BUFFER specified) .

Alternative 1

... DATA BUFFER xstr

Effect

If DATA BUFFER is specified, the data cluster is written to the elementary data object xstr, which has to be of the type xstring. The previous content of xstr is overwritten completely.

Notes

A data object xstr filled by EXPORT TO DATA BUFFER contains exactly one data cluster.
A common application of the addition DATA BUFFER is to store the new data cluster in a field of a database table with the corresponding data type. In this case, consider compressing the data cluster using the addition COMPRESSION, since, by default, compression as a medium is only activated if DATABASE is specified directly.
The content of a data object filled by EXPORT TO DATA BUFFER can only be evaluated using IMPORT FROM DATA BUFFER. In other evaluations, for example when comparing data clusters, the result is not defined. For example, the undefined content of alignment gaps in structures can produce different data clusters with structures that otherwise have the same content.

Alternative 2

... INTERNAL TABLE itab

Effect

If INTERNAL TABLE is specified, the data cluster is stored in the internal table itab. The previous content of itab is overwritten completely.

The first column of itab must have the data type s or i and the second column must have the type x. Depending on the width of the second column, the data is stored across multiple table rows if necessary. The first row contains the length occupied in the second row. The only table category allowed for itab are standard tables without secondary table keys.

Notes

An internal table itab filled by EXPORT TO INTERNAL TABLE contains exactly one data cluster.
The content of an internal table filled by EXPORT TO INTERNAL TABLE can only be evaluated using IMPORT FROM INTERNAL TABLE for the same reasons as for EXPORT TO DATA BUFFER.
The variant EXPORT TO DATA BUFFER is preferred over the variant EXPORT TO INTERNAL TABLE because it is easier to handle. An export to an internal table is of benefit only for very large data clusters and if the available memory is almost used up. This is because its memory is requested block by block, while the memory for a string must always exist as a single entity.

Alternative 3

... MEMORY ID id

Effect

If MEMORY is specified, the data cluster is written in to ABAP Memory with the ID specified in id. id expects a flat character-like data object containing a case-sensitive ID with a maximum of 60 characters. Any existing data clusters with the same ID id are completely overwritten. The ID in id identifies a data cluster in the repository and can be read again using the same identification.

Notes

A data cluster in the ABAP memory is available to all programs within a call sequence, which makes it possible to pass data to called programs.
Outside of classes, an obsolete short form exists, in which the addition ID can be omitted. However, this is prone to errors, since all EXPORT statements without ID overwrite the same data cluster.

Example

Two fields with two different IDs "P1" and "P2" with the dynamic variant of the cluster definition are written to the ABAP memory. After the statement IMPORT is executed, the contents of the fields text1 and text2 are swapped.

TYPES:
  BEGIN OF tab_type,
    para TYPE string,
    dobj TYPE string,
  END OF tab_type.

DATA:
  id    TYPE c LENGTH 10 VALUE 'TEXTS',
  text1 TYPE string VALUE `IKE`,
  text2 TYPE string VALUE `TINA`,
  line  TYPE tab_type,
  itab  TYPE STANDARD TABLE OF tab_type.

line-para = 'P1'.
line-dobj = 'TEXT1'.
APPEND line TO itab.

line-para = 'P2'.
line-dobj = 'TEXT2'.
APPEND line TO itab.

EXPORT (itab)     TO MEMORY ID id.
IMPORT p1 = text2
       p2 = text1 FROM MEMORY ID id.

Alternative 4

... DATABASE dbtab(ar) [FROM wa] [CLIENT cl] ID id

Effect

If DATABASE is specified, the data cluster with the ID id is stored in the database table dbtab and committed by the next database commit. In ABAP Dictionary, the database table must be defined with a fixed INDX-like structure. id expects a flat character-like data object containing an ID no longer than the key fields of the INDX-like table defined between the columns RELID and SRTF2. The ID is case-sensitive.

The two-character area ar, which must be specified directly, splits up the rows of the database table into multiple areas, so that data clusters with the same ID id can exist more than once in the database table.

After FROM, a work area wa can be specified that must have the same data type as the database table dbtab. In an export, the current values of the components of wa, which are located between the fields SRTF2 and CLUSTR, are written to all rows occupied by the data cluster of the database table. If the addition FROM wa is not specified within classes, no data transport takes place in these database fields. If the addition FROM wa is not specified outside of classes, but the statement TABLES is used to declare a table work area for the database table dbtab, the current values of the corresponding components of the table work area dbtab are written to the rows of the database table in the export.

If the database table dbtab is client-specific, a flat character-like field cl can be specified after the addition CLIENT. This field contains a client ID. If the addition is not specified, the current client is used. The column MANDT of every row in the database table occupied by the data cluster is filled by this client ID in the export.

Notes

The statement EXPORT TO DATABASE sets a database lock until the next database commit or rollback. If used incorrectly, this can produce a deadlock.
Data clusters in databases are not converted when migrating from a non-Unicode database to a Unicode system. In a Unicode system, therefore, data clusters may sometimes exist that contain non-Unicode characters. These characters are automatically converted to Unicode characters in each import. When data is exported in Unicode systems, any Unicode characters in the stored data objects are saved in accordance with the relevant platform.
It is still possible to use a table work area implicitly (instead of using FROM wa explicitly). This should be considered an obsolete short form, however.
Since each client represents a self-contained unit, the addition CLIENT must not be used in application programs.

Example

An internal table itab with the name tab and the ID "TABLE" is exported to the area "SQ"
of the database table DEMO_INDX_TABLE, filling the freely selectable components from the structure wa_indx.

TYPES:
  BEGIN OF tab_type,
    col1 TYPE i,
    col2 TYPE i,
  END OF tab_type.

DATA:
  wa_indx TYPE demo_indx_table,
  wa_itab TYPE tab_type,
  itab  TYPE STANDARD TABLE OF tab_type.

WHILE sy-index < 100.
  wa_itab-col1 = sy-index.
  wa_itab-col2 = sy-index ** 2.
  APPEND wa_itab TO itab.
ENDWHILE.

wa_indx-timestamp = sy-datum && sy-uzeit.
wa_indx-userid    = sy-uname.

EXPORT tab = itab
  TO DATABASE demo_indx_table(SQ)
  FROM wa_indx
  ID 'TABLE'.

Alternative 5

... SHARED MEMORY dbtab(ar) [FROM wa] [CLIENT cl] ID id

Alternative 6

... SHARED BUFFER dbtab(ar) [FROM wa] [CLIENT cl] ID id

Effect

If SHARED MEMORY or SHARED BUFFER is specified, the data cluster is stored in cross-transaction application buffers of the shared memory on the application server. All programs of the same application server have access to these buffers.

The two application buffers differ in respect to how the system behaves when reaching the memory limit. Both application buffers can be filled to an internal maximum limit, which can be adjusted using the profile parameter rsdb/esm/buffersize_kb (SHARED MEMORY) and rsdb/obj/buffersize (SHARED BUFFER). Before the maximum limit of the SHARED MEMORY buffer is reached, space must be freed using the statement DELETE FROM SHARED MEMORY; otherwise a handleable exception is raised. The buffer of SHARED BUFFER is cleared automatically by a displacement when it reaches the maximum limit. This procedure deletes the least used data objects from the buffer.

When storing the data, the system creates a memory table in the application buffer. The row structure of this table is defined using dbtab. For dbtab, a database table from ABAP Dictionary must be specified that has the same structure if stored in the database table itself. The row area ar, the work area wa, the optional client cl, and the ID id have the same significance for the memory table as if saved in a database table, with the exception that the length of the ID in id is limited to 59 or 62 characters depending on whether the addition CLIENT is specified or not.

Notes

When storing data in the shared memory, reference is made to a database table, even if the data is not stored in the table itself, but in an appropriately structured memory table.
The length of the key fields of the INDX-like table addressed between the columns RELID and SRTF2 cannot exceed 59 or 62 characters, depending on whether a client column exists.
When data is exported, any data clusters that have the same client cl, row area ar, and ID id are overwritten. If an existing data cluster is to be overwritten by a bigger one when using SHARED MEMORY, and this would exceed the memory limit, this only deletes the existing data cluster.
Instead of saving data clusters in the shared memory, it is best to use shared objects. Shared objects enable objects with complex dependencies to be saved, can be edited like normal objects, and enable multiple consumers to access the shared memory without any copying effort.