django/docs/ref/contrib/gis/gdal.txt

========
GDAL API
========

.. module:: django.contrib.gis.gdal
    :synopsis: GeoDjango's high-level interface to the GDAL library.

`GDAL`__ stands for **Geospatial Data Abstraction Library**,
and is a veritable "Swiss army knife" of GIS data functionality.  A subset
of GDAL is the `OGR`__ Simple Features Library, which specializes
in reading and writing vector geographic data in a variety of standard
formats.

GeoDjango provides a high-level Python interface for some of the
capabilities of OGR, including the reading and coordinate transformation
of vector spatial data and minimal support for GDAL's features with respect
to raster (image) data.

.. note::

    Although the module is named ``gdal``, GeoDjango only supports some of the
    capabilities of OGR and GDAL's raster features at this time.

__ http://www.gdal.org/
__ http://www.gdal.org/ogr_arch.html

Overview
========

.. _gdal_sample_data:

Sample Data
-----------

The GDAL/OGR tools described here are designed to help you read in
your geospatial data, in order for most of them to be useful you have
to have some data to work with.  If you're starting out and don't yet
have any data of your own to use, GeoDjango tests contain a number of
simple data sets that you can use for testing. You can download them here::

    $ wget https://raw.githubusercontent.com/django/django/master/tests/gis_tests/data/cities/cities.{shp,prj,shx,dbf}
    $ wget https://raw.githubusercontent.com/django/django/master/tests/gis_tests/data/rasters/raster.tif

Vector Data Source Objects
==========================

``DataSource``
--------------

:class:`DataSource` is a wrapper for the OGR data source object that
supports reading data from a variety of OGR-supported geospatial file
formats and data sources using a simple, consistent interface.  Each
data source is represented by a :class:`DataSource` object which contains
one or more layers of data.  Each layer, represented by a :class:`Layer`
object, contains some number of geographic features (:class:`Feature`),
information about the type of features contained in that layer (e.g.
points, polygons, etc.), as well as the names and types of any
additional fields (:class:`Field`) of data that may be associated with
each feature in that layer.

.. class:: DataSource(ds_input, encoding='utf-8')

    The constructor for ``DataSource`` only requires one parameter: the path of
    the file you want to read. However, OGR also supports a variety of more
    complex data sources, including databases, that may be accessed by passing
    a special name string instead of a path. For more information, see the
    `OGR Vector Formats`__ documentation. The :attr:`name` property of a
    ``DataSource`` instance gives the OGR name of the underlying data source
    that it is using.

    The optional ``encoding`` parameter allows you to specify a non-standard
    encoding of the strings in the source. This is typically useful when you
    obtain ``DjangoUnicodeDecodeError`` exceptions while reading field values.

    Once you've created your ``DataSource``, you can find out how many layers
    of data it contains by accessing the :attr:`layer_count` property, or
    (equivalently) by using the ``len()`` function. For information on
    accessing the layers of data themselves, see the next section::

        >>> from django.contrib.gis.gdal import DataSource
        >>> ds = DataSource('/path/to/your/cities.shp')
        >>> ds.name
        '/path/to/your/cities.shp'
        >>> ds.layer_count                  # This file only contains one layer
        1

    .. attribute:: layer_count

    Returns the number of layers in the data source.

    .. attribute:: name

    Returns the name of the data source.

__ http://www.gdal.org/ogr_formats.html

``Layer``
---------

.. class:: Layer

    ``Layer`` is a wrapper for a layer of data in a ``DataSource`` object. You
    never create a ``Layer`` object directly. Instead, you retrieve them from
    a :class:`DataSource` object, which is essentially a standard Python
    container of ``Layer`` objects. For example, you can access a specific
    layer by its index (e.g. ``ds[0]`` to access the first layer), or you can
    iterate over all the layers in the container in a ``for`` loop. The
    ``Layer`` itself acts as a container for geometric features.

    Typically, all the features in a given layer have the same geometry type.
    The :attr:`geom_type` property of a layer is an :class:`OGRGeomType` that
    identifies the feature type. We can use it to print out some basic
    information about each layer in a :class:`DataSource`::

        >>> for layer in ds:
        ...     print('Layer "%s": %i %ss' % (layer.name, len(layer), layer.geom_type.name))
        ...
        Layer "cities": 3 Points

    The example output is from the cities data source, loaded above, which
    evidently contains one layer, called ``"cities"``, which contains three
    point features. For simplicity, the examples below assume that you've
    stored that layer in the variable ``layer``::

        >>> layer = ds[0]

    .. attribute:: name

    Returns the name of this layer in the data source.

        >>> layer.name
        'cities'

    .. attribute:: num_feat

    Returns the number of features in the layer. Same as ``len(layer)``::

        >>> layer.num_feat
        3

    .. attribute:: geom_type

    Returns the geometry type of the layer, as an :class:`OGRGeomType` object::

        >>> layer.geom_type.name
        'Point'

    .. attribute:: num_fields

    Returns the number of fields in the layer, i.e the number of fields of
    data associated with each feature in the layer::

        >>> layer.num_fields
        4

    .. attribute:: fields

    Returns a list of the names of each of the fields in this layer::

        >>> layer.fields
        ['Name', 'Population', 'Density', 'Created']

    .. attribute field_types

    Returns a list of the data types of each of the fields in this layer. These
    are subclasses of ``Field``, discussed below::

        >>> [ft.__name__ for ft in layer.field_types]
        ['OFTString', 'OFTReal', 'OFTReal', 'OFTDate']

    .. attribute:: field_widths

    Returns a list of the maximum field widths for each of the fields in this
    layer::

        >>> layer.field_widths
        [80, 11, 24, 10]

    .. attribute:: field_precisions

    Returns a list of the numeric precisions for each of the fields in this
    layer. This is meaningless (and set to zero) for non-numeric fields::

        >>> layer.field_precisions
        [0, 0, 15, 0]

    .. attribute:: extent

    Returns the spatial extent of this layer, as an :class:`Envelope` object::

        >>> layer.extent.tuple
        (-104.609252, 29.763374, -95.23506, 38.971823)

    .. attribute:: srs

    Property that returns the :class:`SpatialReference` associated with this
    layer::

        >>> print(layer.srs)
        GEOGCS["GCS_WGS_1984",
            DATUM["WGS_1984",
                SPHEROID["WGS_1984",6378137,298.257223563]],
            PRIMEM["Greenwich",0],
            UNIT["Degree",0.017453292519943295]]

    If the :class:`Layer` has no spatial reference information associated
    with it, ``None`` is returned.

    .. attribute:: spatial_filter

    Property that may be used to retrieve or set a spatial filter for this
    layer. A spatial filter can only be set with an :class:`OGRGeometry`
    instance, a 4-tuple extent, or ``None``. When set with something other than
    ``None``, only features that intersect the filter will be returned when
    iterating over the layer::

        >>> print(layer.spatial_filter)
        None
        >>> print(len(layer))
        3
        >>> [feat.get('Name') for feat in layer]
        ['Pueblo', 'Lawrence', 'Houston']
        >>> ks_extent = (-102.051, 36.99, -94.59, 40.00) # Extent for state of Kansas
        >>> layer.spatial_filter = ks_extent
        >>> len(layer)
        1
        >>> [feat.get('Name') for feat in layer]
        ['Lawrence']
        >>> layer.spatial_filter = None
        >>> len(layer)
        3

    .. method:: get_fields()

    A method that returns a list of the values of a given field for each
    feature in the layer::

        >>> layer.get_fields('Name')
        ['Pueblo', 'Lawrence', 'Houston']

    .. method:: get_geoms(geos=False)

    A method that returns a list containing the geometry of each feature in the
    layer. If the optional argument ``geos`` is set to ``True`` then the
    geometries are converted to :class:`~django.contrib.gis.geos.GEOSGeometry`
    objects. Otherwise, they are returned as :class:`OGRGeometry` objects::

        >>> [pt.tuple for pt in layer.get_geoms()]
        [(-104.609252, 38.255001), (-95.23506, 38.971823), (-95.363151, 29.763374)]

    .. method:: test_capability(capability)

    Returns a boolean indicating whether this layer supports the given
    capability (a string).  Examples of valid capability strings include:
    ``'RandomRead'``, ``'SequentialWrite'``, ``'RandomWrite'``,
    ``'FastSpatialFilter'``, ``'FastFeatureCount'``, ``'FastGetExtent'``,
    ``'CreateField'``, ``'Transactions'``, ``'DeleteFeature'``, and
    ``'FastSetNextByIndex'``.

``Feature``
-----------

.. class:: Feature

    ``Feature`` wraps an OGR feature. You never create a ``Feature`` object
    directly. Instead, you retrieve them from a :class:`Layer` object. Each
    feature consists of a geometry and a set of fields containing additional
    properties. The geometry of a field is accessible via its ``geom`` property,
    which returns an :class:`OGRGeometry` object. A ``Feature`` behaves like a
    standard Python container for its fields, which it returns as :class:`Field`
    objects: you can access a field directly by its index or name, or you can
    iterate over a feature's fields, e.g. in a ``for`` loop.

    .. attribute:: geom

    Returns the geometry for this feature, as an ``OGRGeometry`` object::

        >>> city.geom.tuple
        (-104.609252, 38.255001)

    .. attribute:: get

    A method that returns the value of the given field (specified by name)
    for this feature, **not** a ``Field`` wrapper object::

        >>> city.get('Population')
        102121

    .. attribute:: geom_type

    Returns the type of geometry for this feature, as an :class:`OGRGeomType`
    object. This will be the same for all features in a given layer and is
    equivalent to the :attr:`Layer.geom_type` property of the :class:`Layer`
    object the feature came from.

    .. attribute:: num_fields

    Returns the number of fields of data associated with the feature. This will
    be the same for all features in a given layer and is equivalent to the
    :attr:`Layer.num_fields` property of the :class:`Layer` object the feature
    came from.

    .. attribute:: fields

    Returns a list of the names of the fields of data associated with the
    feature. This will be the same for all features in a given layer and is
    equivalent to the :attr:`Layer.fields` property of the :class:`Layer`
    object the feature came from.

    .. attribute:: fid

    Returns the feature identifier within the layer::

        >>> city.fid
        0

    .. attribute:: layer_name

    Returns the name of the :class:`Layer` that the feature came from. This
    will be the same for all features in a given layer::

        >>> city.layer_name
        'cities'

    .. attribute:: index

    A method that returns the index of the given field name. This will be the
    same for all features in a given layer::

        >>> city.index('Population')
        1

``Field``
---------

.. class:: Field

    .. attribute:: name

    Returns the name of this field::

        >>> city['Name'].name
        'Name'

    .. attribute:: type

    Returns the OGR type of this field, as an integer. The ``FIELD_CLASSES``
    dictionary maps these values onto subclasses of ``Field``::

        >>> city['Density'].type
        2

    .. attribute:: type_name

    Returns a string with the name of the data type of this field::

        >>> city['Name'].type_name
        'String'

    .. attribute:: value

    Returns the value of this field. The ``Field`` class itself returns the
    value as a string, but each subclass returns the value in the most
    appropriate form::

        >>> city['Population'].value
        102121

    .. attribute:: width

    Returns the width of this field::

        >>> city['Name'].width
        80

    .. attribute:: precision

    Returns the numeric precision of this field. This is meaningless (and set
    to zero) for non-numeric fields::

        >>> city['Density'].precision
        15

    .. method:: as_double()

    Returns the value of the field as a double (float)::

        >>> city['Density'].as_double()
        874.7

    .. method:: as_int()

    Returns the value of the field as an integer::

        >>> city['Population'].as_int()
        102121

    .. method:: as_string()

    Returns the value of the field as a string::

        >>> city['Name'].as_string()
        'Pueblo'

    .. method:: as_datetime()

    Returns the value of the field as a tuple of date and time components::

        >>> city['Created'].as_datetime()
        (c_long(1999), c_long(5), c_long(23), c_long(0), c_long(0), c_long(0), c_long(0))

``Driver``
----------

.. class:: Driver(dr_input)

    The ``Driver`` class is used internally to wrap an OGR :class:`DataSource`
    driver.

    .. attribute:: driver_count

    Returns the number of OGR vector drivers currently registered.

OGR Geometries
==============

``OGRGeometry``
---------------

:class:`OGRGeometry` objects share similar functionality with
:class:`~django.contrib.gis.geos.GEOSGeometry` objects and are thin wrappers
around OGR's internal geometry representation. Thus, they allow for more
efficient access to data when using :class:`DataSource`. Unlike its GEOS
counterpart, :class:`OGRGeometry` supports spatial reference systems and
coordinate transformation::

    >>> from django.contrib.gis.gdal import OGRGeometry
    >>> polygon = OGRGeometry('POLYGON((0 0, 5 0, 5 5, 0 5))')

.. class:: OGRGeometry(geom_input, srs=None)

    This object is a wrapper for the `OGR Geometry`__ class. These objects are
    instantiated directly from the given ``geom_input`` parameter, which may be
    a string containing WKT, HEX, GeoJSON, a ``buffer`` containing WKB data, or
    an :class:`OGRGeomType` object. These objects are also returned from the
    :class:`Feature.geom` attribute, when reading vector data from
    :class:`Layer` (which is in turn a part of a :class:`DataSource`).

    __ http://www.gdal.org/classOGRGeometry.html

    .. classmethod:: from_gml(gml_string)

    .. versionadded:: 1.11

    Constructs an :class:`OGRGeometry` from the given GML string.

    .. classmethod:: from_bbox(bbox)

    Constructs a :class:`Polygon` from the given bounding-box (a 4-tuple).

    .. method:: __len__()

    Returns the number of points in a :class:`LineString`, the number of rings
    in a :class:`Polygon`, or the number of geometries in a
    :class:`GeometryCollection`. Not applicable to other geometry types.

    .. method:: __iter__()

    Iterates over the points in a :class:`LineString`, the rings in a
    :class:`Polygon`, or the geometries in a :class:`GeometryCollection`.
    Not applicable to other geometry types.

    .. method:: __getitem__()

    Returns the point at the specified index for a :class:`LineString`, the
    interior ring at the specified index for a :class:`Polygon`, or the geometry
    at the specified index in a :class:`GeometryCollection`. Not applicable to
    other geometry types.

    .. attribute:: dimension

    Returns the number of coordinated dimensions of the geometry, i.e. 0
    for points, 1 for lines, and so forth::

        >> polygon.dimension
        2

    .. attribute:: coord_dim

    Returns or sets the coordinate dimension of this geometry. For example, the
    value would be 2 for two-dimensional geometries.

    .. attribute:: geom_count

    Returns the number of elements in this geometry::

        >>> polygon.geom_count
        1

    .. attribute:: point_count

    Returns the number of points used to describe this geometry::

        >>> polygon.point_count
        4

    .. attribute:: num_points

    Alias for :attr:`point_count`.

    .. attribute:: num_coords

    Alias for :attr:`point_count`.

    .. attribute:: geom_type

    Returns the type of this geometry, as an :class:`OGRGeomType` object.

    .. attribute:: geom_name

    Returns the name of the type of this geometry::

        >>> polygon.geom_name
        'POLYGON'

    .. attribute:: area

    Returns the area of this geometry, or 0 for geometries that do not contain
    an area::

        >>> polygon.area
        25.0

    .. attribute:: envelope

    Returns the envelope of this geometry, as an :class:`Envelope` object.

    .. attribute:: extent

    Returns the envelope of this geometry as a 4-tuple, instead of as an
    :class:`Envelope` object::

        >>> point.extent
        (0.0, 0.0, 5.0, 5.0)

    .. attribute:: srs

    This property controls the spatial reference for this geometry, or
    ``None`` if no spatial reference system has been assigned to it.
    If assigned, accessing this property returns a :class:`SpatialReference`
    object.  It may be set with another :class:`SpatialReference` object,
    or any input that :class:`SpatialReference` accepts. Example::

        >>> city.geom.srs.name
        'GCS_WGS_1984'

    .. attribute:: srid

    Returns or sets the spatial reference identifier corresponding to
    :class:`SpatialReference` of this geometry.  Returns ``None`` if
    there is no spatial reference information associated with this
    geometry, or if an SRID cannot be determined.

    .. attribute:: geos

    Returns a :class:`~django.contrib.gis.geos.GEOSGeometry` object
    corresponding to this geometry.

    .. attribute:: gml

    Returns a string representation of this geometry in GML format::

        >>> OGRGeometry('POINT(1 2)').gml
        '<gml:Point><gml:coordinates>1,2</gml:coordinates></gml:Point>'

    .. attribute:: hex

    Returns a string representation of this geometry in HEX WKB format::

        >>> OGRGeometry('POINT(1 2)').hex
        '0101000000000000000000F03F0000000000000040'

    .. attribute:: json

    Returns a string representation of this geometry in JSON format::

        >>> OGRGeometry('POINT(1 2)').json
        '{ "type": "Point", "coordinates": [ 1.000000, 2.000000 ] }'

    .. attribute:: kml

    Returns a string representation of this geometry in KML format.

    .. attribute:: wkb_size

    Returns the size of the WKB buffer needed to hold a WKB representation
    of this geometry::

        >>> OGRGeometry('POINT(1 2)').wkb_size
        21

    .. attribute:: wkb

    Returns a ``buffer`` containing a WKB representation of this geometry.

    .. attribute:: wkt

    Returns a string representation of this geometry in WKT format.

    .. attribute:: ewkt

    Returns the EWKT representation of this geometry.

    .. method:: clone()

    Returns a new :class:`OGRGeometry` clone of this geometry object.

    .. method:: close_rings()

    If there are any rings within this geometry that have not been closed,
    this routine will do so by adding the starting point to the end::

        >>> triangle = OGRGeometry('LINEARRING (0 0,0 1,1 0)')
        >>> triangle.close_rings()
        >>> triangle.wkt
        'LINEARRING (0 0,0 1,1 0,0 0)'

    .. method:: transform(coord_trans, clone=False)

    Transforms this geometry to a different spatial reference system. May take
    a :class:`CoordTransform` object, a :class:`SpatialReference` object, or
    any other input accepted by :class:`SpatialReference` (including spatial
    reference WKT and PROJ.4 strings, or an integer SRID).

    By default nothing is returned and the geometry is transformed in-place.
    However, if the ``clone`` keyword is set to ``True`` then a transformed
    clone of this geometry is returned instead.

    .. method:: intersects(other)

    Returns ``True`` if this geometry intersects the other, otherwise returns
    ``False``.

    .. method:: equals(other)

    Returns ``True`` if this geometry is equivalent to the other, otherwise
    returns ``False``.

    .. method:: disjoint(other)

    Returns ``True`` if this geometry is spatially disjoint to (i.e. does
    not intersect) the other, otherwise returns ``False``.

    .. method:: touches(other)

    Returns ``True`` if this geometry touches the other, otherwise returns
    ``False``.

    .. method:: crosses(other)

    Returns ``True`` if this geometry crosses the other, otherwise returns
    ``False``.

    .. method:: within(other)

    Returns ``True`` if this geometry is contained within the other, otherwise
    returns ``False``.

    .. method:: contains(other)

    Returns ``True`` if this geometry contains the other, otherwise returns
    ``False``.

    .. method:: overlaps(other)

    Returns ``True`` if this geometry overlaps the other, otherwise returns
    ``False``.

    .. method:: boundary()

    The boundary of this geometry, as a new :class:`OGRGeometry` object.

    .. attribute:: convex_hull

    The smallest convex polygon that contains this geometry, as a new
    :class:`OGRGeometry` object.

    .. method:: difference()

    Returns the region consisting of the difference of this geometry and
    the other, as a new :class:`OGRGeometry` object.

    .. method:: intersection()

    Returns the region consisting of the intersection of this geometry and
    the other, as a new :class:`OGRGeometry` object.

    .. method:: sym_difference()

    Returns the region consisting of the symmetric difference of this
    geometry and the other, as a new :class:`OGRGeometry` object.

    .. method:: union()

    Returns the region consisting of the union of this geometry and
    the other, as a new :class:`OGRGeometry` object.

    .. attribute:: tuple

    Returns the coordinates of a point geometry as a tuple, the
    coordinates of a line geometry as a tuple of tuples, and so forth::

        >>> OGRGeometry('POINT (1 2)').tuple
        (1.0, 2.0)
        >>> OGRGeometry('LINESTRING (1 2,3 4)').tuple
        ((1.0, 2.0), (3.0, 4.0))

    .. attribute:: coords

    An alias for :attr:`tuple`.

.. class:: Point

    .. attribute:: x

    Returns the X coordinate of this point::

        >>> OGRGeometry('POINT (1 2)').x
        1.0

    .. attribute:: y

    Returns the Y coordinate of this point::

        >>> OGRGeometry('POINT (1 2)').y
        2.0

    .. attribute:: z

    Returns the Z coordinate of this point, or ``None`` if the point does not
    have a Z coordinate::

        >>> OGRGeometry('POINT (1 2 3)').z
        3.0

.. class:: LineString

    .. attribute:: x

    Returns a list of X coordinates in this line::

        >>> OGRGeometry('LINESTRING (1 2,3 4)').x
        [1.0, 3.0]

    .. attribute:: y

    Returns a list of Y coordinates in this line::

        >>> OGRGeometry('LINESTRING (1 2,3 4)').y
        [2.0, 4.0]

    .. attribute:: z

    Returns a list of Z coordinates in this line, or ``None`` if the line does
    not have Z coordinates::

        >>> OGRGeometry('LINESTRING (1 2 3,4 5 6)').z
        [3.0, 6.0]


.. class:: Polygon

    .. attribute:: shell

    Returns the shell or exterior ring of this polygon, as a ``LinearRing``
    geometry.

    .. attribute:: exterior_ring

    An alias for :attr:`shell`.

    .. attribute:: centroid

    Returns a :class:`Point` representing the centroid of this polygon.

.. class:: GeometryCollection

    .. method:: add(geom)

    Adds a geometry to this geometry collection. Not applicable to other
    geometry types.

``OGRGeomType``
---------------

.. class:: OGRGeomType(type_input)

    This class allows for the representation of an OGR geometry type
    in any of several ways::

        >>> from django.contrib.gis.gdal import OGRGeomType
        >>> gt1 = OGRGeomType(3)             # Using an integer for the type
        >>> gt2 = OGRGeomType('Polygon')     # Using a string
        >>> gt3 = OGRGeomType('POLYGON')     # It's case-insensitive
        >>> print(gt1 == 3, gt1 == 'Polygon') # Equivalence works w/non-OGRGeomType objects
        True True

    .. attribute:: name

    Returns a short-hand string form of the OGR Geometry type::

        >>> gt1.name
        'Polygon'

    .. attribute:: num

    Returns the number corresponding to the OGR geometry type::

        >>> gt1.num
        3

    .. attribute:: django

    Returns the Django field type (a subclass of GeometryField) to use for
    storing this OGR type, or ``None`` if there is no appropriate Django type::

        >>> gt1.django
        'PolygonField'

``Envelope``
------------

.. class:: Envelope(*args)

    Represents an OGR Envelope structure that contains the minimum and maximum
    X, Y coordinates for a rectangle bounding box. The naming of the variables
    is compatible with the OGR Envelope C structure.

    .. attribute:: min_x

    The value of the minimum X coordinate.

    .. attribute:: min_y

    The value of the maximum X coordinate.

    .. attribute:: max_x

    The value of the minimum Y coordinate.

    .. attribute:: max_y

    The value of the maximum Y coordinate.

    .. attribute:: ur

    The upper-right coordinate, as a tuple.

    .. attribute:: ll

    The lower-left coordinate, as a tuple.

    .. attribute:: tuple

    A tuple representing the envelope.

    .. attribute:: wkt

    A string representing this envelope as a polygon in WKT format.

    .. method:: expand_to_include(*args)

Coordinate System Objects
=========================

``SpatialReference``
--------------------

.. class:: SpatialReference(srs_input)

    Spatial reference objects are initialized on the given ``srs_input``,
    which may be one of the following:

    * OGC Well Known Text (WKT) (a string)
    * EPSG code (integer or string)
    * PROJ.4 string
    * A shorthand string for well-known standards (``'WGS84'``, ``'WGS72'``,
      ``'NAD27'``, ``'NAD83'``)

    Example::

        >>> wgs84 = SpatialReference('WGS84') # shorthand string
        >>> wgs84 = SpatialReference(4326) # EPSG code
        >>> wgs84 = SpatialReference('EPSG:4326') # EPSG string
        >>> proj4 = '+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs '
        >>> wgs84 = SpatialReference(proj4) # PROJ.4 string
        >>> wgs84 = SpatialReference("""GEOGCS["WGS 84",
        DATUM["WGS_1984",
             SPHEROID["WGS 84",6378137,298.257223563,
                 AUTHORITY["EPSG","7030"]],
             AUTHORITY["EPSG","6326"]],
         PRIMEM["Greenwich",0,
             AUTHORITY["EPSG","8901"]],
         UNIT["degree",0.01745329251994328,
             AUTHORITY["EPSG","9122"]],
         AUTHORITY["EPSG","4326"]]""") # OGC WKT

    .. method:: __getitem__(target)

    Returns the value of the given string attribute node, ``None`` if the node
    doesn't exist. Can also take a tuple as a parameter, (target, child), where
    child is the index of the attribute in the WKT. For example::

        >>> wkt = 'GEOGCS["WGS 84", DATUM["WGS_1984, ... AUTHORITY["EPSG","4326"]]')
        >>> srs = SpatialReference(wkt) # could also use 'WGS84', or 4326
        >>> print(srs['GEOGCS'])
        WGS 84
        >>> print(srs['DATUM'])
        WGS_1984
        >>> print(srs['AUTHORITY'])
        EPSG
        >>> print(srs['AUTHORITY', 1]) # The authority value
        4326
        >>> print(srs['TOWGS84', 4]) # the fourth value in this wkt
        0
        >>> print(srs['UNIT|AUTHORITY']) # For the units authority, have to use the pipe symbol.
        EPSG
        >>> print(srs['UNIT|AUTHORITY', 1]) # The authority value for the units
        9122

    .. method:: attr_value(target, index=0)

    The attribute value for the given target node (e.g. ``'PROJCS'``).
    The index keyword specifies an index of the child node to return.

    .. method:: auth_name(target)

    Returns the authority name for the given string target node.

    .. method:: auth_code(target)

    Returns the authority code for the given string target node.

    .. method:: clone()

    Returns a clone of this spatial reference object.

    .. method:: identify_epsg()

    This method inspects the WKT of this ``SpatialReference`` and will add EPSG
    authority nodes where an EPSG identifier is applicable.

    .. method:: from_esri()

    Morphs this SpatialReference from ESRI's format to EPSG

    .. method:: to_esri()

    Morphs this SpatialReference to ESRI's format.

    .. method:: validate()

    Checks to see if the given spatial reference is valid, if not
    an exception will be raised.

    .. method:: import_epsg(epsg)

    Import spatial reference from EPSG code.

    .. method:: import_proj(proj)

    Import spatial reference from PROJ.4 string.

    .. method:: import_user_input(user_input)

    .. method:: import_wkt(wkt)

    Import spatial reference from WKT.

    .. method:: import_xml(xml)

    Import spatial reference from XML.

    .. attribute:: name

    Returns the name of this Spatial Reference.

    .. attribute:: srid

    Returns the SRID of top-level authority, or ``None`` if undefined.

    .. attribute:: linear_name

    Returns the name of the linear units.

    .. attribute:: linear_units

    Returns the value of the linear units.

    .. attribute:: angular_name

    Returns the name of the angular units."

    .. attribute:: angular_units

    Returns the value of the angular units.

    .. attribute:: units

    Returns a 2-tuple of the units value and the units name and will
    automatically determines whether to return the linear or angular units.

    .. attribute:: ellipsoid

    Returns a tuple of the ellipsoid parameters for this spatial reference:
    (semimajor axis, semiminor axis, and inverse flattening).

    .. attribute:: semi_major

    Returns the semi major axis of the ellipsoid for this spatial reference.

    .. attribute:: semi_minor

    Returns the semi minor axis of the ellipsoid for this spatial reference.

    .. attribute:: inverse_flattening

    Returns the inverse flattening of the ellipsoid for this spatial reference.

    .. attribute:: geographic

    Returns ``True`` if this spatial reference is geographic (root node is
    ``GEOGCS``).

    .. attribute:: local

    Returns ``True`` if this spatial reference is local (root node is
    ``LOCAL_CS``).

    .. attribute:: projected

    Returns ``True`` if this spatial reference is a projected coordinate system
    (root node is ``PROJCS``).

    .. attribute:: wkt

    Returns the WKT representation of this spatial reference.

    .. attribute:: pretty_wkt

    Returns the 'pretty' representation of the WKT.

    .. attribute:: proj

    Returns the PROJ.4 representation for this spatial reference.

    .. attribute:: proj4

    Alias for :attr:`SpatialReference.proj`.

    .. attribute:: xml

    Returns the XML representation of this spatial reference.

``CoordTransform``
------------------

.. class:: CoordTransform(source, target)

Represents a coordinate system transform. It is initialized with two
:class:`SpatialReference`, representing the source and target coordinate
systems, respectively. These objects should be used when performing the same
coordinate transformation repeatedly on different geometries::

    >>> ct = CoordTransform(SpatialReference('WGS84'), SpatialReference('NAD83'))
    >>> for feat in layer:
    ...     geom = feat.geom # getting clone of feature geometry
    ...     geom.transform(ct) # transforming

.. _raster-data-source-objects:

Raster Data Objects
===================

``GDALRaster``
----------------

:class:`GDALRaster` is a wrapper for the GDAL raster source object that
supports reading data from a variety of GDAL-supported geospatial file
formats and data sources using a simple, consistent interface.  Each
data source is represented by a :class:`GDALRaster` object which contains
one or more layers of data named bands.  Each band, represented by a
:class:`GDALBand` object, contains georeferenced image data. For example, an RGB
image is represented as three bands: one for red, one for green, and one for
blue.

.. note::

    For raster data there is no difference between a raster instance and its
    data source. Unlike for the Geometry objects, :class:`GDALRaster` objects are
    always a data source. Temporary rasters can be instantiated in memory
    using the corresponding driver, but they will be of the same class as file-based
    raster sources.

.. class:: GDALRaster(ds_input, write=False)

    The constructor for ``GDALRaster`` accepts two parameters. The first
    parameter defines the raster source, and the second parameter defines if a
    raster should be opened in write mode. For newly-created rasters, the second
    parameter is ignored and the new raster is always created in write mode.

    The first parameter can take three forms: a string representing a file
    path, a dictionary with values defining a new raster, or a bytes object
    representing a raster file.

    If the input is a file path, the raster is opened from there. If the input
    is raw data in a dictionary, the parameters ``width``, ``height``, and
    ``srid`` are required. If the input is a bytes object, it will be opened
    using a GDAL virtual filesystem.

    For a detailed description of how to create rasters using dictionary input,
    see :ref:`gdal-raster-ds-input`. For a detailed description of how to
    create rasters in the virtual filesystem, see :ref:`gdal-raster-vsimem`.

    The following example shows how rasters can be created from different input
    sources (using the sample data from the GeoDjango tests; see also the
    :ref:`gdal_sample_data` section).

        >>> from django.contrib.gis.gdal import GDALRaster
        >>> rst = GDALRaster('/path/to/your/raster.tif', write=False)
        >>> rst.name
        '/path/to/your/raster.tif'
        >>> rst.width, rst.height  # This file has 163 x 174 pixels
        (163, 174)
        >>> rst = GDALRaster({  # Creates an in-memory raster
        ...     'srid': 4326,
        ...     'width': 4,
        ...     'height': 4,
        ...     'datatype': 1,
        ...     'bands': [{
        ...         'data': (2, 3),
        ...         'offset': (1, 1),
        ...         'size': (2, 2),
        ...         'shape': (2, 1),
        ...         'nodata_value': 5,
        ...     }]
        ... })
        >>> rst.srs.srid
        4326
        >>> rst.width, rst.height
        (4, 4)
        >>> rst.bands[0].data()
        array([[5, 5, 5, 5],
               [5, 2, 3, 5],
               [5, 2, 3, 5],
               [5, 5, 5, 5]], dtype=uint8)
        >>> rst_file = open('/path/to/your/raster.tif', 'rb')
        >>> rst_bytes = rst_file.read()
        >>> rst = GDALRaster(rst_bytes)
        >>> rst.is_vsi_based
        True
        >>> rst.name  # Stored in a random path in the vsimem filesystem.
        '/vsimem/da300bdb-129d-49a8-b336-e410a9428dad'

    .. versionchanged:: 1.11

        Added the ability to pass the ``size``, ``shape``, and ``offset``
        parameters when creating :class:`GDALRaster` objects. The parameters
        can be passed through the ``ds_input`` dictionary. This allows to
        finely control initial pixel values. The functionality is similar to
        the :meth:`GDALBand.data()<django.contrib.gis.gdal.GDALBand.data>`
        method.

    .. versionchanged:: 2.0

       Added the ability to read and write rasters in GDAL's memory-based
       virtual filesystem. ``GDALRaster`` objects can now be converted to and
       from binary data in-memory.

    .. attribute:: name

        The name of the source which is equivalent to the input file path or the name
        provided upon instantiation.

            >>> GDALRaster({'width': 10, 'height': 10, 'name': 'myraster', 'srid': 4326}).name
            'myraster'

    .. attribute:: driver

        The name of the GDAL driver used to handle the input file. For ``GDALRaster``\s created
        from a file, the driver type is detected automatically. The creation of rasters from
        scratch is a in-memory raster by default (``'MEM'``), but can be altered as
        needed. For instance, use ``GTiff`` for a ``GeoTiff`` file. For a list of file types,
        see also the `GDAL Raster Formats`__ list.

        __ http://www.gdal.org/formats_list.html

        An in-memory raster is created through the following example:

            >>> GDALRaster({'width': 10, 'height': 10, 'srid': 4326}).driver.name
            'MEM'

        A file based GeoTiff raster is created through the following example:

            >>> import tempfile
            >>> rstfile = tempfile.NamedTemporaryFile(suffix='.tif')
            >>> rst = GDALRaster({'driver': 'GTiff', 'name': rstfile.name, 'srid': 4326,
            ...                   'width': 255, 'height': 255, 'nr_of_bands': 1})
            >>> rst.name
            '/tmp/tmp7x9H4J.tif'           # The exact filename will be different on your computer
            >>> rst.driver.name
            'GTiff'

    .. attribute:: width

        The width of the source in pixels (X-axis).

            >>> GDALRaster({'width': 10, 'height': 20, 'srid': 4326}).width
            10

    .. attribute:: height

        The height of the source in pixels (Y-axis).

            >>> GDALRaster({'width': 10, 'height': 20, 'srid': 4326}).height
            20

    .. attribute:: srs

        The spatial reference system of the raster, as a
        :class:`SpatialReference` instance. The SRS can be changed by
        setting it to an other :class:`SpatialReference` or providing any input
        that is accepted by the :class:`SpatialReference` constructor.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.srs.srid
            4326
            >>> rst.srs = 3086
            >>> rst.srs.srid
            3086

    .. attribute:: srid

        The Spatial Reference System Identifier (SRID) of the raster. This
        property is a shortcut to getting or setting the SRID through the
        :attr:`srs` attribute.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.srid
            4326
            >>> rst.srid = 3086
            >>> rst.srid
            3086
            >>> rst.srs.srid  # This is equivalent
            3086

    .. attribute:: geotransform

        The affine transformation matrix used to georeference the source, as a
        tuple of six coefficients which map pixel/line coordinates into
        georeferenced space using the following relationship::

            Xgeo = GT(0) + Xpixel*GT(1) + Yline*GT(2)
            Ygeo = GT(3) + Xpixel*GT(4) + Yline*GT(5)

        The same values can be retrieved by accessing the :attr:`origin`
        (indices 0 and 3), :attr:`scale` (indices 1 and 5) and :attr:`skew`
        (indices 2 and 4) properties.

        The default is ``[0.0, 1.0, 0.0, 0.0, 0.0, -1.0]``.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.geotransform
            [0.0, 1.0, 0.0, 0.0, 0.0, -1.0]

    .. attribute:: origin

        Coordinates of the top left origin of the raster in the spatial
        reference system of the source, as a point object with ``x`` and ``y``
        members.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.origin
            [0.0, 0.0]
            >>> rst.origin.x = 1
            >>> rst.origin
            [1.0, 0.0]

    .. attribute:: scale

        Pixel width and height used for georeferencing the raster, as a as a
        point object with ``x`` and ``y``  members. See :attr:`geotransform`
        for more information.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.scale
            [1.0, -1.0]
            >>> rst.scale.x = 2
            >>> rst.scale
            [2.0, -1.0]

    .. attribute:: skew

        Skew coefficients used to georeference the raster, as a point object
        with ``x`` and ``y``  members. In case of north up images, these
        coefficients are both ``0``.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.skew
            [0.0, 0.0]
            >>> rst.skew.x = 3
            >>> rst.skew
            [3.0, 0.0]

    .. attribute:: extent

        Extent (boundary values) of the raster source, as a 4-tuple
        ``(xmin, ymin, xmax, ymax)`` in the spatial reference system of the
        source.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.extent
            (0.0, -20.0, 10.0, 0.0)
            >>> rst.origin.x = 100
            >>> rst.extent
            (100.0, -20.0, 110.0, 0.0)

    .. attribute:: bands

        List of all bands of the source, as :class:`GDALBand` instances.

            >>> rst = GDALRaster({"width": 1, "height": 2, 'srid': 4326,
            ...                   "bands": [{"data": [0, 1]}, {"data": [2, 3]}]})
            >>> len(rst.bands)
            2
            >>> rst.bands[1].data()
            array([[ 2.,  3.]], dtype=float32)

    .. method:: warp(ds_input, resampling='NearestNeighbour', max_error=0.0)

        Returns a warped version of this raster.

        The warping parameters can be specified through the ``ds_input``
        argument. The use of ``ds_input`` is analogous to the corresponding
        argument of the class constructor. It is a dictionary with the
        characteristics of the target raster. Allowed dictionary key values are
        width, height, SRID, origin, scale, skew, datatype, driver, and name
        (filename).

        By default, the warp functions keeps most parameters equal to the
        values of the original source raster, so only parameters that should be
        changed need to be specified. Note that this includes the driver, so
        for file-based rasters the warp function will create a new raster on
        disk.

        The only parameter that is set differently from the source raster is the
        name. The default value of the the raster name is the name of the source
        raster appended with ``'_copy' + source_driver_name``. For file-based
        rasters it is recommended to provide the file path of the target raster.

        The resampling algorithm used for warping can be specified with the
        ``resampling`` argument. The default is ``NearestNeighbor``, and the
        other allowed values are ``Bilinear``, ``Cubic``, ``CubicSpline``,
        ``Lanczos``, ``Average``, and ``Mode``.

        The ``max_error`` argument can be used to specify the maximum error
        measured in input pixels that is allowed in approximating the
        transformation. The default is 0.0 for exact calculations.

        For users familiar with ``GDAL``, this function has a similar
        functionality to the ``gdalwarp`` command-line utility.

        For example, the warp function can be used for aggregating a raster to
        the double of its original pixel scale:

            >>> rst = GDALRaster({
            ...     "width": 6, "height": 6, "srid": 3086,
            ...     "origin": [500000, 400000],
            ...     "scale": [100, -100],
            ...     "bands": [{"data": range(36), "nodata_value": 99}]
            ... })
            >>> target = rst.warp({"scale": [200, -200], "width": 3, "height": 3})
            >>> target.bands[0].data()
            array([[  7.,   9.,  11.],
                   [ 19.,  21.,  23.],
                   [ 31.,  33.,  35.]], dtype=float32)

    .. method:: transform(srid, driver=None, name=None, resampling='NearestNeighbour', max_error=0.0)

        Returns a transformed version of this raster with the specified SRID.

        This function transforms the current raster into a new spatial reference
        system that can be specified with an ``srid``. It calculates the bounds
        and scale of the current raster in the new spatial reference system and
        warps the raster using the :attr:`~GDALRaster.warp` function.

        By default, the driver of the source raster is used and the name of the
        raster is the original name appended with
        ``'_copy' + source_driver_name``. A different driver or name can be
        specified with the ``driver`` and ``name`` arguments.

        The default resampling algorithm is ``NearestNeighbour`` but can be
        changed using the ``resampling`` argument. The default maximum allowed
        error for resampling is 0.0 and can be changed using the ``max_error``
        argument. Consult the :attr:`~GDALRaster.warp` documentation for detail
        on those arguments.

            >>> rst = GDALRaster({
            ...     "width": 6, "height": 6, "srid": 3086,
            ...     "origin": [500000, 400000],
            ...     "scale": [100, -100],
            ...     "bands": [{"data": range(36), "nodata_value": 99}]
            ... })
            >>> target = rst.transform(4326)
            >>> target.origin
            [-82.98492744885776, 27.601924753080144]

    .. attribute:: info

        .. versionadded:: 2.0

        Returns a string with a summary of the raster. This is equivalent to
        the `gdalinfo`__ command line utility.

        __ http://www.gdal.org/gdalinfo.html

    .. attribute:: metadata

        .. versionadded:: 2.0

        The metadata of this raster, represented as a nested dictionary. The
        first-level key is the metadata domain. The second-level contains the
        metadata item names and values from each domain.

        To set or update a metadata item, pass the corresponding metadata item
        to the method using the nested structure described above. Only keys
        that are in the specified dictionary are updated; the rest of the
        metadata remains unchanged.

        To remove a metadata item, use ``None`` as the metadata value.

            >>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
            >>> rst.metadata
            {}
            >>> rst.metadata = {'DEFAULT': {'OWNER': 'Django', 'VERSION': '1.0'}}
            >>> rst.metadata
            {'DEFAULT': {'OWNER': 'Django', 'VERSION': '1.0'}}
            >>> rst.metadata = {'DEFAULT': {'OWNER': None, 'VERSION': '2.0'}}
            >>> rst.metadata
            {'DEFAULT': {'VERSION': '2.0'}}

    .. attribute:: vsi_buffer

       .. versionadded:: 2.0

        A ``bytes`` representation of this raster. Returns ``None`` for rasters
        that are not stored in GDAL's virtual filesystem.

    .. attribute:: is_vsi_based

       .. versionadded:: 2.0

        A boolean indicating if this raster is stored in GDAL's virtual
        filesystem.

``GDALBand``
------------

.. class:: GDALBand

    ``GDALBand`` instances are not created explicitly, but rather obtained
    from a :class:`GDALRaster` object, through its :attr:`~GDALRaster.bands`
    attribute. The GDALBands contain the actual pixel values of the raster.

    .. attribute:: description

        The name or description of the band, if any.

    .. attribute:: width

        The width of the band in pixels (X-axis).

    .. attribute:: height

        The height of the band in pixels (Y-axis).

    .. attribute:: pixel_count

        The total number of pixels in this band. Is equal to ``width * height``.

    .. method:: statistics(refresh=False, approximate=False)

        Compute statistics on the pixel values of this band. The return value
        is a tuple with the following structure:
        ``(minimum, maximum, mean, standard deviation)``.

        If the ``approximate`` argument is set to ``True``, the statistics may
        be computed based on overviews or a subset of image tiles.

        If the ``refresh`` argument is set to ``True``, the statistics will be
        computed from the data directly, and the cache will be updated with the
        result.

        If a persistent cache value is found, that value is returned. For
        raster formats using Persistent Auxiliary Metadata (PAM) services, the
        statistics might be cached in an auxiliary file. In some cases this
        metadata might be out of sync with the pixel values or cause values
        from a previous call to be returned which don't reflect the value of
        the ``approximate`` argument. In such cases, use the ``refresh``
        argument to get updated values and store them in the cache.

        For empty bands (where all pixel values are "no data"), all statistics
        are returned as ``None``.

        The statistics can also be retrieved directly by accessing the
        :attr:`min`, :attr:`max`, :attr:`mean`, and :attr:`std` properties.

    .. attribute:: min

        The minimum pixel value of the band (excluding the "no data" value).

    .. attribute:: max

        The maximum pixel value of the band (excluding the "no data" value).

    .. attribute:: mean

        The mean of all pixel values of the band (excluding the "no data"
        value).

    .. attribute:: std

        The standard deviation of all pixel values of the band (excluding the
        "no data" value).

    .. attribute:: nodata_value

        The "no data" value for a band is generally a special marker value used
        to mark pixels that are not valid data. Such pixels should generally not
        be displayed, nor contribute to analysis operations.

        To delete an existing "no data" value, set this property to ``None``
        (requires GDAL ≥ 2.1).

    .. method:: datatype(as_string=False)

        The data type contained in the band, as an integer constant between 0
        (Unknown) and 11. If ``as_string`` is ``True``, the data type is
        returned as a string with the following possible values:
        ``GDT_Unknown``, ``GDT_Byte``, ``GDT_UInt16``, ``GDT_Int16``,
        ``GDT_UInt32``, ``GDT_Int32``, ``GDT_Float32``, ``GDT_Float64``,
        ``GDT_CInt16``, ``GDT_CInt32``, ``GDT_CFloat32``, and ``GDT_CFloat64``.

    .. method:: data(data=None, offset=None, size=None, shape=None)

        The accessor to the pixel values of the ``GDALBand``. Returns the complete
        data array if no parameters are provided. A subset of the pixel array can
        be requested by specifying an offset and block size as tuples.

        If NumPy is available, the data is returned as NumPy array. For performance
        reasons, it is highly recommended to use NumPy.

        Data is written to the ``GDALBand`` if the ``data`` parameter is provided.
        The input can be of one of the following types - packed string, buffer, list,
        array, and NumPy array. The number of items in the input should normally
        correspond to the total number of pixels in the band, or to the number
        of pixels for a specific block of pixel values if the ``offset`` and
        ``size`` parameters are provided.

        If the number of items in the input is different from the target pixel
        block, the ``shape`` parameter must be specified. The shape is a tuple
        that specifies the width and height of the input data in pixels. The
        data is then replicated to update the pixel values of the selected
        block. This is useful to fill an entire band with a single value, for
        instance.

        For example:

            >>> rst = GDALRaster({'width': 4, 'height': 4, 'srid': 4326, 'datatype': 1, 'nr_of_bands': 1})
            >>> bnd = rst.bands[0]
            >>> bnd.data(range(16))
            >>> bnd.data()
            array([[ 0,  1,  2,  3],
                   [ 4,  5,  6,  7],
                   [ 8,  9, 10, 11],
                   [12, 13, 14, 15]], dtype=int8)
            >>> bnd.data(offset=(1, 1), size=(2, 2))
            array([[ 5,  6],
                   [ 9, 10]], dtype=int8)
            >>> bnd.data(data=[-1, -2, -3, -4], offset=(1, 1), size=(2, 2))
            >>> bnd.data()
            array([[ 0,  1,  2,  3],
                   [ 4, -1, -2,  7],
                   [ 8, -3, -4, 11],
                   [12, 13, 14, 15]], dtype=int8)
            >>> bnd.data(data='\x9d\xa8\xb3\xbe', offset=(1, 1), size=(2, 2))
            >>> bnd.data()
            array([[  0,   1,   2,   3],
                   [  4, -99, -88,   7],
                   [  8, -77, -66,  11],
                   [ 12,  13,  14,  15]], dtype=int8)
            >>> bnd.data([1], shape=(1, 1))
            >>> bnd.data()
            array([[1, 1, 1, 1],
                   [1, 1, 1, 1],
                   [1, 1, 1, 1],
                   [1, 1, 1, 1]], dtype=uint8)
            >>> bnd.data(range(4), shape=(1, 4))
            array([[0, 0, 0, 0],
                   [1, 1, 1, 1],
                   [2, 2, 2, 2],
                   [3, 3, 3, 3]], dtype=uint8)

    .. attribute:: metadata

        .. versionadded:: 2.0

        The metadata of this band. The functionality is identical to
        :attr:`GDALRaster.metadata`.

.. _gdal-raster-ds-input:

Creating rasters from data
--------------------------

This section describes how to create rasters from scratch using the
``ds_input`` parameter.

A new raster is created when a ``dict`` is passed to the :class:`GDALRaster`
constructor. The dictionary contains defining parameters of the new raster,
such as the origin, size, or spatial reference system. The dictionary can also
contain pixel data and information about the format of the new raster. The
resulting raster can therefore be file-based or memory-based, depending on the
driver specified.

There's no standard for describing raster data in a dictionary or JSON flavor.
The definition of the dictionary input to the :class:`GDALRaster` class is
therefore specific to Django. It's inspired by the `geojson`__ format, but the
``geojson`` standard is currently limited to vector formats.

Examples of using the different keys when creating rasters can be found in the
documentation of the corresponding attributes and methods of the
:class:`GDALRaster` and :class:`GDALBand` classes.

__ http://geojson.org

The ``ds_input`` dictionary
~~~~~~~~~~~~~~~~~~~~~~~~~~~

Only a few keys are required in the ``ds_input`` dictionary to create a raster:
``width``, ``height``, and ``srid``. All other parameters have default values
(see the table below). The list of keys that can be passed in the ``ds_input``
dictionary is closely related but not identical to the :class:`GDALRaster`
properties. Many of the parameters are mapped directly to those properties;
the others are described below.

The following table describes all keys that can be set in the ``ds_input``
dictionary.

================= ======== ==================================================
Key               Default  Usage
================= ======== ==================================================
``srid``          required Mapped to the :attr:`~GDALRaster.srid` attribute
``width``         required Mapped to the :attr:`~GDALRaster.width` attribute
``height``        required Mapped to the :attr:`~GDALRaster.height` attribute
``driver``        ``MEM``  Mapped to the :attr:`~GDALRaster.driver` attribute
``name``          ``''``   See below
``origin``        ``0``    Mapped to the :attr:`~GDALRaster.origin` attribute
``scale``         ``0``    Mapped to the :attr:`~GDALRaster.scale` attribute
``skew``          ``0``    Mapped to the :attr:`~GDALRaster.width` attribute
``bands``         ``[]``   See below
``nr_of_bands``   ``0``    See below
``datatype``      ``6``    See below
``papsz_options`` ``{}``   See below
================= ======== ==================================================

.. object:: name

    String representing the name of the raster. When creating a file-based
    raster, this parameter must be the file path for the new raster. If the
    name starts with ``/vsimem/``, the raster is created in GDAL's virtual
    filesystem.

.. object:: datatype

    Integer representing the data type for all the bands. Defaults to ``6``
    (Float32). All bands of a new raster are required to have the same datatype.
    The value mapping is:

    ===== =============== ===============================
    Value GDAL Pixel Type Description
    ===== =============== ===============================
    1     GDT_Byte        Eight bit unsigned integer
    2     GDT_UInt16      Sixteen bit unsigned integer
    3     GDT_Int16       Sixteen bit signed integer
    4     GDT_UInt32      Thirty-two bit unsigned integer
    5     GDT_Int32       Thirty-two bit signed integer
    6     GDT_Float32     Thirty-two bit floating point
    7     GDT_Float64     Sixty-four bit floating point
    ===== =============== ===============================

.. object:: nr_of_bands

    Integer representing the number of bands of the raster. A raster can be
    created without passing band data upon creation. If the number of bands
    isn't specified, it's automatically calculated from the length of the
    ``bands`` input. The number of bands can't be changed after creation.

.. object:: bands

    A list of ``band_input`` dictionaries with band input data. The resulting
    band indices are the same as in the list provided. The definition of the
    band input dictionary is given below. If band data isn't provided, the
    raster bands values are instantiated as an array of zeros and the "no
    data" value is set to ``None``.

.. object:: papsz_options

    .. versionadded:: 2.0

    A dictionary with raster creation options. The key-value pairs of the
    input dictionary are passed to the driver on creation of the raster.

    The available options are driver-specific and are described in the
    documentation of each driver.

    The values in the dictionary are not case-sensitive and are automatically
    converted to the correct string format upon creation.

    The following example uses some of the options available for the
    `GTiff driver`__. The result is a compressed signed byte raster with an
    internal tiling scheme. The internal tiles have a block size of 23 by 23::

        >>> GDALRaster({
        ...    'driver': 'GTiff',
        ...    'name': '/path/to/new/file.tif',
        ...    'srid': 4326,
        ...    'width': 255,
        ...    'height': 255,
        ...    'nr_of_bands': 1,
        ...    'papsz_options': {
        ...        'compress': 'packbits',
        ...        'pixeltype': 'signedbyte',
        ...        'tiled': 'yes',
        ...        'blockxsize': 23,
        ...        'blockysize': 23,
        ...    }
        ... })

__ http://www.gdal.org/frmt_gtiff.html

The ``band_input`` dictionary
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The ``bands`` key in the ``ds_input`` dictionary is a list of ``band_input``
dictionaries. Each ``band_input`` dictionary can contain pixel values and the
"no data" value to be set on the bands of the new raster. The data array can
have the full size of the new raster or be smaller. For arrays that are smaller
than the full raster, the ``size``, ``shape``, and ``offset`` keys  control the
pixel values. The corresponding keys are passed to the :meth:`~GDALBand.data`
method. Their functionality is the same as setting the band data with that
method. The following table describes the keys that can be used.

================ ================================= ======================================================
Key              Default                           Usage
================ ================================= ======================================================
``nodata_value`` ``None``                          Mapped to the :attr:`~GDALBand.nodata_value` attribute
``data``         Same as ``nodata_value`` or ``0`` Passed to the :meth:`~GDALBand.data` method
``size``         ``(with, height)`` of raster      Passed to the :meth:`~GDALBand.data` method
``shape``        Same as size                      Passed to the :meth:`~GDALBand.data` method
``offset``       ``(0, 0)``                        Passed to the :meth:`~GDALBand.data` method
================ ================================= ======================================================

.. _gdal-raster-vsimem:

Using GDAL's Virtual Filesystem
-------------------------------

GDAL has an internal memory-based filesystem, which allows treating blocks of
memory as files. It can be used to read and write :class:`GDALRaster` objects
to and from binary file buffers.

This is useful in web contexts where rasters might be obtained as a buffer
from a remote storage or returned from a view without being written to disk.

:class:`GDALRaster` objects are created in the virtual filesystem when a
``bytes`` object is provided as input, or when the file path starts with
``/vsimem/``.

Input provided as ``bytes`` has to be a full binary representation of a file.
For instance::

    # Read a raster as a file object from a remote source.
    >>> from urllib.request import urlopen
    >>> dat = urlopen('http://example.com/raster.tif').read()
    # Instantiate a raster from the bytes object.
    >>> rst = GDALRaster(dat)
    # The name starts with /vsimem/, indicating that the raster lives in the
    # virtual filesystem.
    >>> rst.name
    '/vsimem/da300bdb-129d-49a8-b336-e410a9428dad'

To create a new virtual file-based raster from scratch, use the ``ds_input``
dictionary representation and provide a ``name`` argument that starts with
``/vsimem/`` (for detail of the dictionary representation, see
:ref:`gdal-raster-ds-input`). For virtual file-based rasters, the
:attr:`~GDALRaster.vsi_buffer` attribute returns the ``bytes`` representation
of the raster.

Here's how to create a raster and return it as a file in an
:class:`~django.http.HttpResponse`::

    >>> from django.http import HttpResponse
    >>> rst = GDALRaster({
    ...     'name': '/vsimem/temporarymemfile',
    ...     'driver': 'tif',
    ...     'width': 6, 'height': 6, 'srid': 3086,
    ...     'origin': [500000, 400000],
    ...     'scale': [100, -100],
    ...     'bands': [{'data': range(36), 'nodata_value': 99}]
    ... })
    >>> HttpResponse(rast.vsi_buffer, 'image/tiff')

Settings
========

.. setting:: GDAL_LIBRARY_PATH

``GDAL_LIBRARY_PATH``
---------------------

A string specifying the location of the GDAL library.  Typically,
this setting is only used if the GDAL library is in a non-standard
location (e.g., ``/home/john/lib/libgdal.so``).

Exceptions
==========

.. exception:: GDALException

    The base GDAL exception, indicating a GDAL-related error.

.. exception:: SRSException

    An exception raised when an error occurs when constructing or using a
    spatial reference system object.