FreeMat - FREAD File Read Function

FREAD File Read Function

Usage

Reads a block of binary data from the given file handle into a variable of a given shape and precision. The general use of the function is

  A = fread(handle,size,precision)

The handle argument must be a valid value returned by the fopen function, and accessable for reading. The size argument determines the number of values read from the file. The size argument is simply a vector indicating the size of the array A. The size argument can also contain a single inf dimension, indicating that FreeMat should calculate the size of the array along that dimension so as to read as much data as possible from the file (see the examples listed below for more details). The data is stored as columns in the file, not rows. Alternately, you can specify two return values to the fread function, in which case the second value contains the number of elements read

   [A,count] = fread(...)

where count is the number of elements in A. The third argument determines the type of the data. Legal values for this argument are listed below:

'uint8','uchar','unsigned char' for an unsigned, 8-bit integer.
'int8','char','integer*1' for a signed, 8-bit integer.
'uint16','unsigned short' for an unsigned, 16-bit integer.
'int16','short','integer*2' for a signed, 16-bit integer.
'uint32','unsigned int' for an unsigned, 32-bit integer.
'int32','int','integer*4' for a signed, 32-bit integer.
'single','float32','float','real*4' for a 32-bit floating point.
'double','float64','real*8' for a 64-bit floating point.

Starting with FreeMat 4, the format for the third argument has changed. If you specify only a type, such as 'float', the data is read in as single precision, but the output type is always 'double'. This behavior is consistent with Matlab. If you want the output type to match the input type (as was previous behavior in FreeMat), you must preface the precision string with a '*'. Thus, the precision string '*float' implies that data is read in as single precision, and the output is also single precision. The third option is to specify the input and output types explicitly. You can do this by specifiying a precision string of the form 'type1 => type2', where 'type1' is the input type and 'type2' is the output type. For example, if the input type is 'double' and the output type is to be a 'float', then a type spec of 'double => float' should be used.

Example

First, we create an array of 512 x 512 Gaussian-distributed float random variables, and then writing them to a file called test.dat.

--> A = float(randn(512));
--> fp = fopen('test.dat','w');
--> fwrite(fp,A);
--> fclose(fp);

Read as many floats as possible into a row vector

--> fp = fopen('test.dat','r');
--> x = fread(fp,[1,inf],'float');
--> fclose(fp);
--> who x
  Variable Name       Type   Flags             Size
              x    double                    [1 262144]

Note that x is a double array. This behavior is new to FreeMat 4. Read the same floats into a 2-D float array.

--> fp = fopen('test.dat','r');
--> x = fread(fp,[512,inf],'float');
--> fclose(fp);
--> who x
  Variable Name       Type   Flags             Size
              x    double                    [512 512]

To read them as a single precision float array, we can use the following form:

--> fp = fopen('test.dat','r');
--> x = fread(fp,[512,inf],'*float');
--> fclose(fp);
--> who x
  Variable Name       Type   Flags             Size
              x    single                    [512 512]

FreeMat v4.0 - Online Documentation

FREAD File Read Function

Usage

Example