X-Git-Url: https://git.auder.net/?p=epclust.git;a=blobdiff_plain;f=epclust%2FR%2Fde_serialize.R;h=8dde2581545ea442d37a60217cb1dbb8bce38158;hp=242e23a0b496b8ccf33134facf3c617c7c56af19;hb=4bcfdbee4e2157f232427a5bfdf240f34760110d;hpb=4efef8ccd1522278f53aa5ce265f3a6cfb6fbd9f

diff --git a/epclust/R/de_serialize.R b/epclust/R/de_serialize.R
index 242e23a..8dde258 100644
--- a/epclust/R/de_serialize.R
+++ b/epclust/R/de_serialize.R
@@ -1,10 +1,32 @@
-#data: matrix of double or connection
-serialize = function(data_ascii, data_bin_file, nb_per_chunk,
+#' @name de_serialize
+#' @rdname de_serialize
+#' @aliases binarize getDataInFile
+#'
+#' @title (De)Serialization of a matrix
+#'
+#' @description \code{binarize()} serializes a matrix or CSV file with minimal overhead,
+#'   into a binary file. \code{getDataInFile()} achieves the inverse task: she retrieves
+#'   (ASCII) data rows from indices in the binary file
+#'
+#' @param data_ascii Either a matrix or CSV file, with items in rows
+#' @param indices Indices of the lines to retrieve
+#' @param data_bin_file Name of binary file on output (\code{binarize})
+#'   or intput (\code{getDataInFile})
+#' @param nb_per_chunk Number of lines to process in one batch
+#' @inheritParams claws
+#'
+#' @return For \code{getDataInFile()}, the matrix with rows corresponding to the
+#'   requested indices
+NULL
+
+#' @rdname de_serialize
+#' @export
+binarize = function(data_ascii, data_bin_file, nb_per_chunk,
 	sep=",", nbytes=4, endian=.Platform$endian)
 {
 	if (is.character(data_ascii))
 		data_ascii = file(data_ascii, open="r")
-	else if (is(data_ascii,"connection") && !isOpen(data_ascii))
+	else if (methods::is(data_ascii,"connection") && !isOpen(data_ascii))
 		open(data_ascii)
 
 	first_write = (!file.exists(data_bin_file) || file.info(data_bin_file)$size == 0)
@@ -14,13 +36,13 @@ serialize = function(data_ascii, data_bin_file, nb_per_chunk,
 	if (first_write)
 	{
 		#number of items always on 8 bytes
-		writeBin(0L, data_bin, size=8) #,endian="little")
+		writeBin(0L, data_bin, size=8, endian=endian)
 		if (is.matrix(data_ascii))
 			data_length = ncol(data_ascii)
 		else #if (is(data, "connection"))
 		{
 			data_line = scan(data_ascii, double(), sep=sep, nlines=1, quiet=TRUE)
-			writeBin(data_line, data_bin, size=nbytes)
+			writeBin(data_line, data_bin, size=nbytes, endian=endian)
 			data_length = length(data_line)
 		}
 	}
@@ -43,34 +65,35 @@ serialize = function(data_ascii, data_bin_file, nb_per_chunk,
 			data_chunk = scan(data_ascii, double(), sep=sep, nlines=nb_per_chunk, quiet=TRUE)
 		if (length(data_chunk)==0)
 			break
-		writeBin(data_chunk, data_bin, size=nbytes)
+		writeBin(data_chunk, data_bin, size=nbytes, endian=endian)
 	}
 
 	if (first_write)
 	{
 		#ecrire file_size-1 / (nbytes*nbWritten) en 0 dans bin_data ! ignored == file_size
 		ignored = seek(data_bin, 0)
-		writeBin(data_length, data_bin, size=8)
+		writeBin(data_length, data_bin, size=8, endian=endian)
 	}
 	close(data_bin)
 
-	if (is(data_ascii,"connection"))
+	if (methods::is(data_ascii,"connection"))
 		close(data_ascii)
 }
 
-#read in binary file, always same structure
+#' @rdname de_serialize
+#' @export
 getDataInFile = function(indices, data_bin_file, nbytes=4, endian=.Platform$endian)
 {
 	data_bin = file(data_bin_file, "rb")
-	data_size = file.info(data_bin)$size
-	data_length = readBin(data_bin, "integer", 1, 8, endian)
+	data_size = file.info(data_bin_file)$size
+	data_length = readBin(data_bin, "integer", n=1, size=8, endian=endian)
 	#Ou t(sapply(...)) (+ rapide ?)
 	data_ascii = do.call( rbind, lapply( indices, function(i) {
 		offset = 8+(i-1)*data_length*nbytes
 		if (offset > data_size)
 			return (vector("double",0))
 		ignored = seek(data_bin, offset)
-		readBin(data_bin, "double", n=data_length, size=nbytes)
+		readBin(data_bin, "double", n=data_length, size=nbytes, endian=endian)
 	} ) )
 	close(data_bin)
 	if (ncol(data_ascii)>0) data_ascii else NULL