This driver supports reading and writing data in Oracle Spatial (8.1.7 or
later) Object-Relational format. The Oracle Spatial driver is not
normally built into OGR, but may be built in on platforms where the
Oracle client libraries are available.
When opening a database, it's name should be specified in the form
"OCI:userid/password@database_instance:table,table". The list of tables
is optional. The database_instance portion may be omitted
when accessing the default local database instance.
If the list of tables is not provided, then all tables appearing in
ALL_SDO_GEOM_METADATA will be treated by OGR as layers with the table
names as the layer names. Non-spatial tables or spatial tables not listed
in the ALL_SDO_GEOM_METADATA table are not accessible unless explicitly
listed in the datasource name. Even in databases where all desired layers
are in the ALL_SDO_GEOM_METADATA table, it may be desirable to list only
the tables to be used as this can substantially reduce initialization time
in databases with many tables.
If the table has an integer column called OGR_FID it will be used as the
feature id by OGR (and it will not appear as a regular attribute). When
loading data into Oracle Spatial OGR will always create the OGR_FID field.
By default, the Oracle driver passes SQL statements directly to Oracle rather
than evaluating them internally when using the ExecuteSQL() call on the
OGRDataSource, or the -sql command option to ogr2ogr. Attribute query
expressions are also passed through to Oracle.
As well two special commands are supported via the ExecuteSQL() interface.
These are "DELLAYER:<table_name>" to delete a layer,
and "VALLAYER:<table_name>" to apply the
SDO_GEOM.VALIDATE_GEOMETRY() check to a layer. Internally these
pseudo-commands are translated into more complex SQL commands for Oracle.
It's also possible to request the driver to handle SQL commands with OGR SQL engine,
by passing "OGRSQL" string to the ExecuteSQL() method, as name of the SQL dialect.
- The type recognition logic is currently somewhat impoverished. No
effort is made to preserve real width information for integer and real
- Various types such as objects, and BLOBs in Oracle will be completely
ignored by OGR.
- Currently the OGR transaction semantics are not properly mapped onto
transaction semantics in Oracle.
- If an attribute called OGR_FID exists in the schema for tables being
read, it will be used as the FID. Random (FID based) reads on tables without
an identified (and indexed) FID field can be very slow. To force use of a
particular field name the OCI_FID configuration variable (i.e. environment
variable) can be set to the target field name.
- Curved geometry types are converted to linestrings or linear rings in six degree segments when reading. The driver has no support for writing curved geometries.
- There is no support for point cloud (SDO_PC), TIN (SDO_TIN) and annotation text data types in Oracle Spatial.
- It might be necessary to define the environment variable NLS_LANG to
"American_America.UTF8" to avoid issues with floating point numbers being
truncated to integer on non-English environments.
- For developers: when running the driver under the memory error detection
tool Valgrind, specifying the database_instance, typically to localhost, or with
the TWO_TASK environment variable seems to be
compulsory, otherwise "TNS:permission denied" errors will be reported)
The Oracle Spatial driver does not support creation of new datasets (database
instances), but it does allow creation of new layers within an
Upon closing the OGRDataSource newly created layers will have a spatial
index automatically built. At this point the USER_SDO_GEOM_METADATA table
will also be updated with bounds for the table based on the features that
have actually been written. One consequence of this is that once a layer
has been loaded it is generally not possible to load additional features
outside the original extents without manually modifying the DIMINFO information
in USER_SDO_GEOM_METADATA and rebuilding the spatial index.
Layer Creation Options
- OVERWRITE: This may be "YES" to force an existing layer (=table) of the
same desired name to be destroyed before creating the requested layer.
The default value is “NO"
- TRUNCATE: This may be "YES" to force the existing table to
be reused, but to first truncate all records in the table, preserving
indexes or dependencies. The default value is “NO".
- LAUNDER: This may be "YES" to force new fields created on this
layer to have their field names "laundered" into a form more compatible with
Oracle. This converts to upper case and converts some special characters
like "-" and "#" to "_". The default value is "NO".
- PRECISION: This may be "YES" to force new fields created on this
layer to try and represent the width and precision information, if available
using NUMBER(width,precision) or VARCHAR2(width) types. If "NO" then the types
NUMBER, INTEGER and VARCHAR2 will be used instead. The default is "YES".
- DIM: This may be set to 2 or 3 to force the dimension of the
created layer. Prior to GDAL 2.2, 3 is used by default. Starting with GDAL 2.2,
the dimension of the layer geometry type is used by default.
- SPATIAL_INDEX: This may be set to FALSE to disable creation of a spatial
index when a layer load is complete. By default an index is created if
any of the layer features have valid geometries. The default is "YES".
Note: option was called INDEX in releases before GDAL 2
- INDEX_PARAMETERS: This may be set to pass creation parameters
when the spatial index is created. For instance setting INDEX_PARAMETERS
to SDO_RTR_PCTFREE=0 would cause the rtree index to be created without any
empty space. By default no parameters are passed causing a default R-Tree
spatial index to be created.
- ADD_LAYER_GTYPE=YES/NO: (starting with GDAL 2.0) This may be
set to NO to disable the constraints on the geometry type in the spatial index,
through the layer_gtype keyword in the PARAMETERS clause of the CREATE INDEX.
Layers of type MultiPoint, MultiLineString or MultiPolygon will also accept
single geometry type (Point, LineString, Polygon). Defaults to YES.
- DIMINFO_X: This may be set to xmin,xmax,xres values to
control the X dimension info written into the USER_SDO_GEOM_METADATA table.
By default extents are collected from the actual data written.
- DIMINFO_Y: This may be set to ymin,ymax,yres values to
control the Y dimension info written into the USER_SDO_GEOM_METADATA table.
By default extents are collected from the actual data written.
- DIMINFO_Z: This may be set to zmin,zmax,zres values to
control the Z dimension info written into the USER_SDO_GEOM_METADATA table.
By default fixed values of -100000,100000,0.002 are used for layers with
a third dimension.
- SRID: By default this driver will attempt to find an existing
row in the MDSYS.CS_SRS table with a well known text coordinate system
exactly matching the one for this dataset. If one is not found, a new
row will be added to this table. The SRID creation option allows the user
to force use of an existing Oracle SRID item even it if does not exactly
match the WKT the driver expects.
- MULTI_LOAD: If enabled new features will be created in
groups of 100 per SQL INSERT command, instead of each feature being a separate
INSERT command. Having this enabled is the fastest way to load data quickly.
Multi-load mode is enabled by default, and may be forced off for existing
layers or for new layers by setting to NO. The number of rows in each group is defined by
MULTI_LOAD_COUNT. To load one row at a time, set MULTI_LOAD to NO.
- MULTI_LOAD_COUNT: Define the number of features on each ARRAY INSERT command,
instead of the default 100 item defined by MULTI_LOAD. Since each array insert will commit
a transaction, this options shouldn't be combined with ogr2ogr "-gt N".
Use "-gt unlimited" preferably when using MULTI_LOAD_COUNT. The default is 100.
If neither MULTI_LOAD nor MULTI_LOAD_COUNT are specified, then the loading happens
in groups of 100 rows.
- FIRST_ID: Define the first numeric value of the id column on the first
rows. It's also work as a open option when used to append or update an existing dataset.
- NO_LOGGING: Define that the table and the geometry will be create
with nologging attributes.
- LOADER_FILE: If this option is set, all feature information will
be written to a file suitable for use with SQL*Loader instead of inserted
directly in the database. The layer itself is still created in the database
immediately. The SQL*Loader support is experimental, and generally
MULTI_LOAD enabled mode should be used instead when trying for optimal
- GEOMETRY_NAME: By default OGR creates new tables with the
geometry column named ORA_GEOMETRY. If you wish to use a different name,
it can be supplied with the GEOMETRY_NAME layer creation option.
Layer Open Options
- FIRST_ID: See Layer Create Options comments on FIRST_ID.
- MULTI_LOAD_COUNT: See Layer Create Options comments on MULTI_LOAD_COUNT.
- WORKSPACE: Define what user workspace to use.
Simple translation of a shapefile into Oracle. The table 'ABC' will
be created with the features from abc.shp and attributes from abc.dbf.
% ogr2ogr -f OCI OCI:email@example.com abc.shp
This second example loads a political boundaries layer from VPF (via the
OGDI driver), and renames the layer from the
cryptic OGDI layer name to something more sensible. If an existing table
of the desired name exists it is overwritten.
% ogr2ogr -f OCI OCI:warmerda/password \
-lco OVERWRITE=yes -nln polbndl_bnd 'polbndl@bnd(*)_line'
This example shows using ogrinfo to evaluate an SQL query statement
within Oracle. More sophisticated Oracle Spatial specific queries may also be
used via the -sql commandline switch to ogrinfo.
ogrinfo -ro OCI:warmerda/password -sql "SELECT pop_1994 from canada where province_name = 'Alberta'"
I would like to thank SRC, LLC
for it's financial support of the development of this driver.