Create a TileDB dense or sparse array from a given data.frame Object

The supplied data.frame object is (currently) limited to integer, numeric, or character. In addition, three datetime columns are supported with the R representations of Date, POSIXct and nanotime.

fromDataFrame(
  obj,
  uri,
  col_index = NULL,
  sparse = FALSE,
  allows_dups = sparse,
  cell_order = "COL_MAJOR",
  tile_order = "COL_MAJOR",
  filter = "ZSTD",
  capacity = 10000L,
  tile_domain = NULL,
  tile_extent = NULL,
  debug = FALSE
)

Arguments

obj	A data.frame object.
uri	A character variable with an Array URI.
col_index	An optional column index, either numeric with a column index, or character with a column name, designating an index column; default is NULL implying an index column is added when the array is created
sparse	A logical switch to select sparse or dense (the default)
allows_dups	A logical switch to select if duplicate values are allowed or not, default is the same value as ‘sparse’.
cell_order	A character variable with one of the TileDB cell order values, default is “COL_MAJOR”.
tile_order	A character variable with one of the TileDB tile order values, default is “COL_MAJOR”.
filter	A character variable vector, defaults to ‘ZSTD’, for one or more filters to be applied to each attribute;
capacity	A integer value with the schema capacity, default is 10000.
tile_domain	An integer vector or list or NULL. If an integer vector of size two it specifies the integer domain of the row dimension; if a list then a named element is used for the dimension of the same name; or if NULL the row dimension of the obj is used.
tile_extent	An integer value for the tile extent of the row dimensions; if NULL the row dimension of the obj is used. Note that the tile_extent cannot exceed the tile domain.
debug	Logical flag to select additional output

Value

Null, invisibly.

Details

The created (dense or sparse) array will have as many attributes as there are columns in the data.frame. Each attribute will be a single column. For a sparse array, one or more columns have to be designated as dimensions.

At present, factor variable are converted to character.

Examples

ctx <- tiledb_ctx(limitTileDBCores())
if (FALSE) {
uri <- tempfile()
## turn factor into character
irisdf <- within(iris, Species <- as.character(Species))
fromDataFrame(irisdf, uri)
arr <- tiledb_array(uri, as.data.frame=TRUE, sparse=FALSE)
newdf <- arr[]
all.equal(iris, newdf)
}