UTF-8 support for font-renedering using the sprites API

[josuah]

There will be a host-side fontgen.py script to generate a packed font file, to upload as either a file on the filesystem, either built into the firmware.

There will be a device-side font.py to render a font out of that font data onto a canvas, and use that canvas to send it to the FPGA.

[josuah]

This is only the font generation, along with a script for plotting it as text, but this means both sides of the implementation currently work:

$ wget http://unifoundry.com/pub/unifont/unifont-15.0.06/font-builds/unifont-15.0.06.bdf.gz
$ gunzip unifont-15.0.06.bdf.gz
$ python3 fontgen unifont-15.0.06.bdf unifont-15.0.06.bin
$ python3 fontdraw.py unifont-15.0.06.bin  "八つ橋"
八
        #     |
    #   #     |
    #   #     |
    #   #     |
    #   #     |
    #   #     |
    #    #    |
   #     #    |
   #     #    |
   #      #   |
  #       #   |
  #        #  |
 #         #  |
 #          # |
#            #|
つ
     #####  |
#  ##     # |
 ##        #|
           #|
           #|
           #|
          # |
        ##  |
     ###    |
橋
   #       ##  |
   #   ####    |
   #     #     |
   #  #########|
######  #   #  |
   #   # ### # |
  ##  #  # #  #|
  ###    ###   |
 # # #         |
 # # # ####### |
#  #   #     # |
   #   # ### # |
   #   # # # # |
   #   # ### # |
   #   #     # |
   #   #    ## |
$

[josuah]

This format has an index that can be read in Python array, and can tell where to find the data in the rest of the file. As opposed to a full memory-mapped bitmap like http://unifoundry.com/pub/unifont/unifont-15.0.01/unifont-15.0.01.bmp, this permits:

• to have a variable size font,
• to reduce storage size for small glyphs: . is smaller to store than 龍
• to remove glyphs from the font to reduce the font size,
• to stay fast, with a small index in RAM
• to minimize the amount of data to read from the font file
• to use multiple font files that will all be looked-up to find the data