pre_analysis.txt

This program should be a command line utility which takes a bunch of images
as command line parameters and packs them together into a single image output:
a BMP is output. Together with the image output, there should be an output
file describing the BMP. For each input image, there should be an entry in this
output file telling where that input image is located in the output BMP image.

Output image can only be in BMP.

Output file describing the output image's sub regions (the input images) will
be in CSV. Converting CSV into something else should be trivial.

The program will be able to work with whatever SDL2_image can work with.

After the command line options. You simply specify all the image inputs.

The CSV format will be:

  name,x,y,w,h

Where x,y,w,h is the x,y within the output image that the input image is, and
w,h is the input image's width and height. The name field is based on the
file name:

  - Stripped out of extensions if it has any.
  - Anything non [A-Za-z0-9_] is converted into _ (if input file name is
  foo/bar2.bmp, then csv name is foo_bar2). This is to make it easier to use
  this in source code (it's often easy to convert a CSV text into some sort of
  struct/array value).

Command line options:

  -w WIDTH
    Maximum width (pixels) for the resulting BMP.
  -h HEIGHT
    Maximum height (pixels) for the resulting BMP.
  -r REPLACEMENT
    Character to be used instead of the default '_' to replace non C-identifier
    characters.
  -o BMP_OUTPUT_FILE_NAME
    The name of the BMP output file (default is out.bmp).
  -c CSV_OUTPUT_FILE_NAME
    The name of the CSV output file (default is out.csv).
  -l
    Loosely enforce maximum dimensions. If specified, the program will try to
    avoid going over maximum width/height, but if it finds it has to, it'll.

As soon as something not recognized as a command line option appears, no more
command line options can appear. So they should come first.

  # This will attempt to load file: 'a.bmp' and '-l'.
  imgpacker -w 1024 a.bmp -l

Issues
======
There will be no attempt to deal with too many images so that they wouldn't fit
into memory.