STScI Logo

generic softools



generic -- generic preprocessor


generic [-k] [-o ofile] [-p prefix] [-t types] files


Allow the output files generated by generic to clobber any existing files.
-o ofile
The name of the output file. If this option is selected, only a single file can be processed.
-p prefix
A prefix to be prepended to the output filenames. This is useful when the output files are to be placed in a different directory.
-t types
The datatypes for which output is desired. One output file will be generated for each type specified, with generic automatically generating the output filename by appending the type character to the root filename of the input file. The type string is some subset of [ubscilrdx], where the type characters are as follows.
u	- C unsigned short
b	- C byte (char)
c	- SPP character
s	- SPP short
i	- SPP int
l	- SPP long
r	- SPP real
d	- SPP double
x	- SPP complex

This option cannot be used in combination with the -o option, and should not be used when generic code is expanded inline, rather than written into multiple output files.

The input file or files to be processed. Generic input files should have the extension ".gx" or ".gc", although this is not required. Only a single input file can be given if the -o option is specified.


The generic preprocessor is used to translate generic source code (code written to work for any datatype) into type dependent source code, suitable for compilation and insertion into a library. The generic source is translated for each datatype, producing a type dependent copy of the source code for each datatype. There are two primary modes of operation:

The generic source is embedded in a normal file, bracketed by $for and $endfor directives. There is one input file and one somewhat larger output file, with the generic code in the input file being replaced in the output file by several copies of the enclosed source, one for each datatype. This mode is most commonly used for modules to be linked in their entirety into an applications package. The "-o" parameter is used to specify the output filename.
The entire input file is generic. There may be multiple input files, and for each input file N output files are generated, one for each datatype specified with the "-t" parameter. The output filenames are automatically generated by appending the type character to the root filename of the input file. This mode is most commonly used for object libraries.

The generic preprocessor operates by token replacement (currently using a UNIX Lex lexical analyzer). The input stream is broken up into a stream of tokens. Each token is examined to see if it is in the following list, and the indicated action is taken if the token is matched. The generic preprocessor directives have the form "$NAME", where $ marks a generic directive, and where NAME is the name of the directive.

Replaced by the current type name, e.g., "int", "real", etc.
Replaced by the current type name in upper case, preceded by an X, e.g., "XINT", "XREAL", etc. This is used for generic C procedures meant to be called from SPP or Fortran.
Replaced by the numeric constant denoting indefinite for the current datatype.
These strings are not replaced, since the "INDEF" in this case is not generic.
Replaced by "SZ_INT", "SZ_REAL", etc.
Replaced by "TY_INT", "TY_REAL", etc.
Replaced by the string "PIXEL". This is used in doubly generic sources, where the first pass translates $PIXEL to PIXEL, and the second to the actual type string.
Replaced by the string "INDEF".
Replaced by one of the characters [ubcsilrdx].
Replaced by one of the characters [UBCSILRDX].
Replaced by the string "...", i.e., whatever is within the // delimiters. Used to disable generic preprocessing of arbitrary text.
Replaced by the corresponding real or double constant. For example, "1$f" translates as "1.0" for type real, but as "1.0D0" for type double.

$if (expression)
The conditional preprocessing facility. If the $IF tests false the code which follows is skipped over, and is not copied to the output file. Control transfers to the matching $ELSE or $ENDIF. The following may be used in the boolean expression:

"datatype"	denotes the current type
ubcsilrdx	any subset of these characters denotes
		    the corresponding datatype
sizeof()	the size of the specified type,
		    e.g., for comparisons

!= ==		the relational operators
 >  <  >= <=


	$if (datatype != dx)
	    (code to be compiled if type not d or x)

	$if (sizeof(i) <= sizeof(r))
	    (code to be compiled if size int <= real)

$IF constructs may be nested. The directive may appear anywhere on a line.

Marks the else clause of a $IF.
Marks the end of a $IF. One is required for every $IF.
$for (types)
For each of the listed types, output a translated copy of the code between the $FOR and the matching $ENDFOR. Nesting is permitted.

	$for (silrd)
	(any amount of generic code)
Marks the end of a $FOR statement.
Replaced by a single $.
C comments are not preprocessed.
Quoted strings are not preprocessed.
SPP comments are not preprocessed.
SPP Fortran escapes are not preprocessed.


1. Translate the generic source "aadd.gx" to produce the six output files "aadds.x", "aaddi.x", etc., in the subdirectory "ak", clobbering any existing files therein. The generic task is a bootstrap utility written in C and is implemented as a CL foreign task, hence the UNIX command syntax.

cl> generic -k -p ak/ -t silrdx aadd.gx

2. Perform an inline transformation ($FOR directive) of the source file "imsum.gx", producing the single file "imsum.x" as output.

cl> generic -k -o imsum.x imsum.gx

3. The following is a simple example of a typical generic source file. For additional examples, see the ".gx" sources in the VOPS, IMIO, IMAGES and other directories.

# ALIM -- Compute the limits (minimum and maximum values) of a vector.
# (this is a copy of the file vops$alim.gx).

procedure alim$t (a, npix, minval, maxval)

PIXEL	a[ARB], minval, maxval, value
int	npix, i

	minval = a[1]
	maxval = a[1]

	do i = 1, npix {
	    value = a[i]
	    $if (datatype == x)
		if (abs(value) < abs(minval))
		    minval = value
		else if (abs(value) > abs(maxval))
		    maxval = value
		if (value < minval)
		    minval = value
		else if (value > maxval)
		    maxval = value


xc, xyacc

Source Code · Search Form · STSDAS