add.gdsn: Add a new GDS node

View source: R/gdsfmt-main.r

add.gdsnR Documentation

Add a new GDS node

Description

Add a new GDS node to the GDS file.

Usage

add.gdsn(node, name, val=NULL, storage=storage.mode(val), valdim=NULL,
    compress=c("", "ZIP", "ZIP_RA", "LZMA", "LZMA_RA", "LZ4", "LZ4_RA"),
    closezip=FALSE, check=TRUE, replace=FALSE, visible=TRUE, ...)

Arguments

node

an object of class gdsn.class or gds.class: "gdsn.class" – the node of hierarchical structure; "gds.class" – the root of hieracrchical structure

name

the variable name; if it is not specified, a temporary name is assigned

val

the R value can be integers, real numbers, characters, factor, logical or raw variable, list and data.frame

storage

to specify data type (not case-sensitive), signed integer: "int8", "int16", "int24", "int32", "int64", "sbit2", "sbit3", ..., "sbit16", "sbit24", "sbit32", "sbit64", "vl_int" (encoding variable-length signed integer); unsigned integer: "uint8", "uint16", "uint24", "uint32", "uint64", "bit1", "bit2", "bit3", ..., "bit15", "bit16", "bit24", "bit32", "bit64", "vl_uint" (encoding variable-length unsigned integer); floating-point number ( "float32", "float64" ); packed real number ( "packedreal8", "packedreal16", "packedreal24", "packedreal32": pack a floating-point number to a signed 8/16/24/32-bit integer with two attributes "offset" and "scale", representing “(signed int)*scale + offset”, where the minimum of the signed integer is used to represent NaN; "packedreal8u", "packedreal16u", "packedreal24u", "packedreal32u": pack a floating-point number to an unsigned 8/16/24/32-bit integer with two attributes "offset" and "scale", representing “(unsigned int)*scale + offset”, where the maximum of the unsigned integer is used to represent NaN ); sparse array ( "sp.int"(="sp.int32"), "sp.int8", "sp.int16", "sp.int32", "sp.int64", "sp.uint8", "sp.uint16", "sp.uint32", "sp.uint64", "sp.real"(="sp.real64"), "sp.real32", "sp.real64" ); string (variable-length: "string", "string16", "string32"; C [null-terminated] string: "cstring", "cstring16", "cstring32"; fixed-length: "fstring", "fstring16", "fstring32"); Or "char" (="int8"), "int"/"integer" (="int32"), "single" (="float32"), "float" (="float32"), "double" (="float64"), "character" (="string"), "logical", "list", "factor", "folder"; Or a gdsn.class object, the storage mode is set to be the same as the object specified by storage.

valdim

the dimension attribute for the array to be created, which is a vector of length one or more giving the maximal indices in each dimension

compress

the compression method can be "" (no compression), "ZIP", "ZIP.fast", "ZIP.def", "ZIP.max" or "ZIP.none" (original zlib); "ZIP_RA", "ZIP_RA.fast", "ZIP_RA.def", "ZIP_RA.max" or "ZIP_RA.none" (zlib with efficient random access); "LZ4", "LZ4.none", "LZ4.fast", "LZ4.hc" or "LZ4.max" (LZ4 compression/decompression library); "LZ4_RA", "LZ4_RA.none", "LZ4_RA.fast", "LZ4_RA.hc" or "LZ4_RA.max" (with efficient random access); "LZMA", "LZMA.fast", "LZMA.def", "LZMA.max", "LZMA_RA", "LZMA_RA.fast", "LZMA_RA.def", "LZMA_RA.max" (lzma compression/decompression algorithm). See details

closezip

if a compression method is specified, get into read mode after compression

check

if TRUE, a warning will be given when val is character and there are missing values in val. GDS format does not support missing characters NA, and any NA will be converted to a blank string ""

replace

if TRUE, replace the existing variable silently if possible

visible

FALSE – invisible/hidden, except print(, all=TRUE)

...

additional parameters for specific storage, see details

Details

val: if val is list or data.frame, the child node(s) will be added corresponding to objects in list or data.frame. If calling add.gdsn(node, name, val=NULL), then a label will be added which does not have any other data except the name and attributes. If val is raw-type, it is interpreted as 8-bit signed integer.

storage: the default value is storage.mode(val), "int" denotes signed integer, "uint" denotes unsigned integer, 8, 16, 24, 32 and 64 denote the number of bits. "bit1" to "bit32" denote the packed data types for 1 to 32 bits which are packed on disk, and "sbit2" to "sbit32" denote the corresponding signed integers. "float32" denotes single-precision number, and "float64" denotes double-precision number. "string" represents strings of 8-bit characters, "string16" represents strings of 16-bit characters following UTF16 industry standard, and "string32" represents a string of 32-bit characters following UTF32 industry standard. "folder" is to create a folder.

valdim: the values in data are taken to be those in the array with the leftmost subscript moving fastest. The last entry could be ZERO. If the total number of elements is zero, gdsfmt does not allocate storage space. NA is treated as 0.

compress: Z compression algorithm (http://www.zlib.net) can be used to deflate the data stored in the GDS file. "ZIP" option is equivalent to "ZIP.def". "ZIP.fast", "ZIP.def" and "ZIP.max" correspond to different compression levels.

To support efficient random access of Z stream, "ZIP_RA", "ZIP_RA.fast", "ZIP_RA.def" or "ZIP_RA.max" should be specified. "ZIP_RA" option is equivalent to "ZIP_RA.def:256K". The block size can be specified by following colon, and "16K", "32K", "64K", "128K", "256K", "512K", "1M", "2M", "4M" and "8M" are allowed, like "ZIP_RA:64K". The compression algorithm tries to keep each independent compressed data block to be about of the specified block size, like 64K.

LZ4 fast lossless compression algorithm is allowed when compress="LZ4" (https://github.com/lz4/lz4). Three compression levels can be specified, "LZ4.fast" (LZ4 fast mode), "LZ4.hc" (LZ4 high compression mode), "LZ4.max" (maximize the compression ratio). The block size can be specified by following colon, and "64K", "256K", "1M" and "4M" are allowed according to LZ4 frame format. "LZ4" is equivalent to "LZ4.hc:256K".

To support efficient random access of LZ4 stream, "LZ4_RA", "LZ4_RA.fast", "LZ4_RA.hc" or "ZIP_RA.max" should be specified. "LZ4_RA" option is equivalent to "LZ4_RA.hc:256K". The block size can be specified by following colon, and "16K", "32K", "64K", "128K", "256K", "512K", "1M", "2M", "4M" and "8M" are allowed, like "LZ4_RA:64K". The compression algorithm tries to keep each independent compressed data block to be about of the specified block size, like 64K.

LZMA compression algorithm (https://tukaani.org/xz/) is available since gdsfmt_v1.7.18, which has a higher compression ratio than ZIP algorithm. "LZMA", "LZMA.fast", "LZMA.def" and "LZMA.max" available. To support efficient random access of LZMA stream, "LZMA_RA", "LZMA_RA.fast", "LZMA_RA.def" and "LZMA_RA.max" can be used. The block size can be specified by following colon. "LZMA_RA" is equivalent to "LZMA_RA.def:256K".

To finish compressing, you should call readmode.gdsn to close the writing mode.

the parameter details with equivalent command lines can be found at compression.gdsn.

closezip: if compression option is specified, then enter a read mode after deflating the data. see readmode.gdsn.

...: if storage = "fstring", "fstring16" or "fstring32", users can set the max length of string in advance by maxlen=. If storage = "packedreal8", "packedreal8u", "packedreal16", "packedreal16u", "packedreal32" or "packedreal32u", users can define offset and scale to represent real numbers by “val*scale + offset” where “val” is a 8/16/32-bit integer. By default, offset=0, scale=0.01 for "packedreal8" and "packedreal8u", scale=0.0001 for "packedreal16" and "packedreal16u", scale=0.00001 for "packedreal24" and "packedreal24u", scale=0.000001 for "packedreal32" and "packedreal32u". For example, packedreal8:scale=1/127,offset=0, packedreal16:scale=1/32767,offset=0 for correlation [-1, 1]; packedreal8u:scale=1/254,offset=0, packedreal16u:scale=1/65534,offset=0 for a probability [0, 1].

Value

An object of class gdsn.class of the new node.

Author(s)

Xiuwen Zheng

References

http://zlib.net, https://github.com/lz4/lz4, https://tukaani.org/xz/

See Also

addfile.gdsn, addfolder.gdsn, compression.gdsn, index.gdsn, read.gdsn, readex.gdsn, write.gdsn, append.gdsn

Examples

# cteate a GDS file
f <- createfn.gds("test.gds")
L <- -2500:2499

##########################################################################
# commom types

add.gdsn(f, "label", NULL)
add.gdsn(f, "int", 1:10000, compress="ZIP", closezip=TRUE)
add.gdsn(f, "int.matrix", matrix(L, nrow=100, ncol=50))
add.gdsn(f, "double", seq(1, 1000, 0.4))
add.gdsn(f, "character", c("int", "double", "logical", "factor"))
add.gdsn(f, "logical", rep(c(TRUE, FALSE, NA), 50))
add.gdsn(f, "factor", as.factor(c(letters, NA, "AA", "CC")))
add.gdsn(f, "NA", rep(NA, 10))
add.gdsn(f, "NaN", c(rep(NaN, 20), 1:20))
add.gdsn(f, "bit2-matrix", matrix(L[1:5000], nrow=50, ncol=100),
    storage="bit2")
# list and data.frame
add.gdsn(f, "list", list(X=1:10, Y=seq(1, 10, 0.25)))
add.gdsn(f, "data.frame", data.frame(X=1:19, Y=seq(1, 10, 0.5)))


##########################################################################
# save a .RData object

obj <- list(X=1:10, Y=seq(1, 10, 0.1))
save(obj, file="tmp.RData")
addfile.gdsn(f, "tmp.RData", filename="tmp.RData")

f

read.gdsn(index.gdsn(f, "list"))
read.gdsn(index.gdsn(f, "list/Y"))
read.gdsn(index.gdsn(f, "data.frame"))


##########################################################################
# allocate the disk spaces

n1 <- add.gdsn(f, "n1", 1:100, valdim=c(10, 20))
read.gdsn(index.gdsn(f, "n1"))

n2 <- add.gdsn(f, "n2", matrix(1:100, 10, 10), valdim=c(15, 20))
read.gdsn(index.gdsn(f, "n2"))


##########################################################################
# replace variables

f

add.gdsn(f, "double", 1:100, storage="float", replace=TRUE)
f
read.gdsn(index.gdsn(f, "double"))


# close the GDS file
closefn.gds(f)


# delete the temporary file
unlink("test.gds", force=TRUE)

zhengxwen/gdsfmt documentation built on Sept. 27, 2024, 10:08 a.m.