Imager::Preprocess - simple preprocessor for handling multiple sample sizes


  /* in the source: */
  #code condition true to work with 8-bit samples
  ... code using preprocessor types/values ...

  # process and make #line directives
  perl -MImager::Preprocess -epreprocess foo.c

  # process and no #line directives
  perl -MImager::Preprocess -epreprocess -l foo.c


This is a simple preprocessor that aims to reduce duplication of source code when implementing an algorithm both for 8-bit samples and double samples in Imager.

Imager's Makefile.PL currently scans the MANIFEST for .im files and adds Makefile files to convert these to .c files.

The beginning of a sample-independent section of code is preceded by:

  #code expression

where expression should return true if processing should be done at 8-bits/sample.

You can also use a #code block around a function definition to produce 8-bit and double sample versions of a function. In this case #code has no expression and you will need to use IM_SUFFIX() to produce different function names.

The end of a sample-independent section of code is terminated by:


#code sections cannot be nested.

#/code without a starting #code is an error.

You can also define extra sample-size dependent macros with #!define:

    #define common-name eight-bit-name floating-point-name

The following types and values are defined in a #code section:

  • IM_GPIX(im, x, y, &col)

  • IM_GLIN(im, l, r, y, colors)

  • IM_PPIX(im, x, y, &col)

  • IM_PLIN(im, x, y, colors)

  • IM_GSAMP(im, l, r, y, samples, chans, chan_count)

    These correspond to the appropriate image function, eg. IM_GPIX() becomes i_gpix() or i_gpixf() as appropriate.

  • IM_ADAPT_COLORS(dest_channels, src_channels, colors, count)

    Call i_adapt_colors() or i_adapt_fcolors().

  • IM_FILL_COMBINE(fill) - retrieve the combine function from a fill object.

  • IM_FILL_FILLER(fill) - retrieve the fill_with_* function from a fill object.

  • IM_ABS(sample) - calculate the absolute value of an IM_WORK_T value.

  • IM_SAMPLE_MAX - maximum value for a sample

  • IM_SAMPLE_MAX2 - maximum value for a sample, squared

  • IM_SAMPLE_T - type of a sample (i_sample_t or i_fsample_t)

  • IM_COLOR - color type, either i_color or i_fcolor.

  • IM_WORK_T - working sample type, either int or double.

  • IM_Sf - format string for the sample type, "%d" or "%f".

  • IM_Wf - format string for the work type, "%d" or "%f".

  • IM_SUFFIX(identifier) - adds _8 or _double onto the end of identifier.

  • IM_EIGHT_BIT - this is a macro defined only in 8-bit/sample code.

Other types, functions and values may be added in the future.


Tony Cook <>