pkg/man/de_serialize.Rd

   1 % Generated by roxygen2: do not edit by hand
   2 % Please edit documentation in R/de_serialize.R
   3 \name{de_serialize}
   4 \alias{de_serialize}
   5 \alias{binarize}
   6 \alias{binarizeTransform}
   7 \alias{getDataInFile}
   8 \alias{binarize}
   9 \alias{binarizeTransform}
  10 \alias{getDataInFile}
  11 \title{(De)Serialization of a [big]matrix or data stream}
  12 \usage{
  13 binarize(data_ascii, data_bin_file, nb_per_chunk, sep = ",", nbytes = 4,
  14   endian = .Platform$endian)
  15
  16 binarizeTransform(getData, transform, data_bin_file, nb_per_chunk, nbytes = 4,
  17   endian = .Platform$endian)
  18
  19 getDataInFile(indices, data_bin_file, nbytes = 4, endian = .Platform$endian)
  20 }
  21 \arguments{
  22 \item{data_ascii}{Either a matrix or CSV file, with items in rows}
  23
  24 \item{data_bin_file}{Name of binary file on output (\code{binarize})
  25 or intput (\code{getDataInFile})}
  26
  27 \item{nb_per_chunk}{Number of lines to process in one batch}
  28
  29 \item{sep}{Separator in CSV input file (if any provided)}
  30
  31 \item{nbytes}{Number of bytes to serialize a floating-point number; 4 or 8}
  32
  33 \item{endian}{Endianness to use for (de)serialization. Use "little" or "big" for portability}
  34
  35 \item{getData}{Function to retrieve data chunks}
  36
  37 \item{transform}{Transformation function to apply on data chunks}
  38
  39 \item{indices}{Indices of the lines to retrieve}
  40 }
  41 \value{
  42 For \code{getDataInFile()}, the matrix with rows corresponding to the
  43   requested indices. \code{binarizeTransform} returns the number of processed lines.
  44   \code{binarize} is designed to serialize in several calls, thus returns nothing.
  45 }
  46 \description{
  47 \code{binarize()} serializes a matrix or CSV file with minimal overhead,
  48   into a binary file. \code{getDataInFile()} achieves the inverse task: she retrieves
  49   (ASCII) data rows from indices in the binary file. Finally,
  50   \code{binarizeTransform()} serialize transformations of all data chunks; to use it,
  51   a data-retrieval function must be provided, thus \code{binarize} will most likely be
  52   used first (and then a function defined to seek in generated binary file)
  53 }