Clone GDAL/OGR DataSource/DataSet to in-memory instance #gdal #ogr #python

clone_to_mem.py

from osgeo import gdal, ogr

def clone_data_to_mem(ds, name=''):
  if (isinstance(ds, gdal.Dataset)):
    return clone_raster_to_mem(ds, name)
  elif (isinstance(ds, ogr.DataSource)):
    return clone_vector_to_mem(ds, name)
  else:
    raise TypeError('Data source must be of GDAL dataset or OGR datasource')

def clone_vector_to_mem(vector_ds, name=''):
  driver = ogr.GetDriverByName('Memory')
  return driver.CopyDataSource(vector_ds, name)

def clone_raster_to_mem(raster_ds, name=''):
  driver = gdal.GetDriverByName('MEM')
  return driver.CopyDataSource(raster_ds, name)

Useful for cloning a disk backed data into an in-memory mutable data, so you can work on it in memory, before committing it to disk.

At some point you may need to create a target datasource, and then copy the memory data back into it.

In GDAL 2.x, there's no need to use OGR as well, as GDALDataset when used with OpenEx will support both raster and vector data.