#!/usr/bin/perl
# -*- tab-width: 4; -*-
# vi: set sw=2 ts=4 expandtab:

# Copyright 2019-2020 The Khronos Group Inc.
# SPDX-License-Identifier: Apache-2.0

use strict;
use warnings;

# N.B. 0 arguments, read stdin, write stdout.
# 1 argument, read ARGV[0], write stdout.
# 2 arguments, read ARGV[0], write ARGV[1].
my $infile = shift @ARGV;
my $outfile = shift @ARGV;
my $input;
if (defined $infile) {
    open($input, '<', $infile);
} else {
    $input = *STDIN;
}
if (defined $outfile) {
    open (my $output, '>', $outfile);
    select $output;
}

# Endianness is a parameter to the (non-block-compressed) generators
# This doesn't have to be a number: $bigEndian = "myBigEndianFlag" will drop this argument in the generated code
my $bigEndian = 0;

# Keep track of formats we've seen to avoid duplicates
my %foundFormats = ();

# Check if we've processed a format. Returns true for extension formats
# that have been promoted to core when the core format has been processed.
# Usually only _KHR and _EXT extensions are promoted so only those are
# checked.
sub formatProcessed {
    my $format =$_[0];
    my $format_noext = $format;
    $format_noext =~ s/_(EXT|KHR)$//;

    return (exists($foundFormats{$format})
            || exists($foundFormats{$format_noext}));
}

print "/* Copyright 2019-2020 The Khronos Group Inc. */\n";
print "/* SPDX-", "License-Identifier: Apache-2.0 */\n\n";
print "/***************************** Do not edit.  *****************************\n";
print "             Automatically generated by makevk2dfd.pl.\n";
print " *************************************************************************/\n\n";

# Loop over each line of input
# IMPORTANT: Do not use `<>` as that reads all files given on the
# command line.
while (my $line = <$input>) {
    # Match any format that starts with a channel description (some number of R, G, B, A or a number)
    # In PERL, "=~" performs a regex operation on the left argument
    # m/<regex>/ matches the regular expression

    my $format = "";
    my $suffix = "";
    my $channels = "";
    my $scheme = "";
    # VK_FORMAT_ is a simple 422 format
    if ($line =~ m/(VK_FORMAT_[RGBX0-9]+_422_UNORM(_4PACK16)?)/) {
        $format = $1; # Extract the format identifier from the rest of the line
        if (exists($foundFormats{$format})) { next }
        $foundFormats{$format} = 1;

        # VK_FORMAT_G8B8G8R8_422_UNORM,
        # VK_FORMAT_B8G8R8G8_422_UNORM,
        # VK_FORMAT_G10X6B10X6G10X6R10X6_422_UNORM_4PACK16,
        # VK_FORMAT_B10X6G10X6R10X6G10X6_422_UNORM_4PACK16,
        # VK_FORMAT_G12X4B12X4G12X4R12X4_422_UNORM_4PACK16,
        # VK_FORMAT_B12X4G12X4R12X4G12X4_422_UNORM_4PACK16,
        # VK_FORMAT_G16B16G16R16_422_UNORM,
        # VK_FORMAT_B16G16R16G16_422_UNORM,

        $format =~ m/VK_FORMAT_([RGBX0-9]+)_422_UNORM(_4PACK16)?/;
        my $formatChannels = $1;

        my $bits = "";
        my $shiftBits = "";
        my $position_xs = "";
        my $position_ys = "";
        my $last_green = 1;

        while ($formatChannels =~ m/([RGBX0-9]*)([RGB])([0-9]+)X?([0-9]+)?/) {
            # Processing from right to left for ease of regex expression
            my $channel = $2;
            my $bit = $3;
            my $padding = $4 ? $4 : "";

            if ($channels ne "") {
                $channels = ", " . $channels;
                $bits = ", " . $bits;
                $shiftBits = ", " . $shiftBits;
                $position_xs = ", " . $position_xs;
                $position_ys = ", " . $position_ys;
            }

            # The default 4:2:2 positioning is currently cosited-even.
            # Block size is 2x1: Sample positions coded as 1/128 on X and 1/256 on Y
            # Cosited-even:
            #     Position      Sample
            #      X    Y       X    Y
            # Y0: 0.5, 0.5  ->  64, 128
            # U : 0.5, 0.5  ->  64, 128
            # Y1: 1.5, 0.5  -> 192, 128
            # V : 0.5, 0.5  ->  64, 128
            # Midpoint:
            #     Position      Sample
            #      X    Y       X    Y
            # Y0: 0.5, 0.5  ->  64, 128
            # U : 1.0, 0.5  -> 128, 128
            # Y1: 1.5, 0.5  -> 192, 128
            # V : 1.0, 0.5  -> 128, 128

            if ($channel eq 'R') {
                $channels = "2" . $channels; # 2 = Cr / V
                $position_xs = "64" . $position_xs;
                $position_ys = "128" . $position_ys;
            } elsif ($channel eq 'G') {
                $channels = "0" . $channels; # 0 = Y
                $position_xs = ($last_green ? "192" : "64") . $position_xs;
                $position_ys = "128" . $position_ys;
                $last_green = 0;
            } elsif ($channel eq 'B') {
                $channels = "1" . $channels; # 1 = Cb / U
                $position_xs = "64" . $position_xs;
                $position_ys = "128" . $position_ys;
            }

            $bits = $bit . $bits;

            if ($padding ne "") { $shiftBits = $padding . $shiftBits; }
            else { $shiftBits = "0" . $shiftBits; }

            $formatChannels = $1;
        }

        print "case $format: {\n";
        print "    int channels[] = {$channels}; int bits[] = {$bits}; int shiftBits[] = {$shiftBits};\n";
        print "    int position_xs[] = {$position_xs}; int position_ys[] = {$position_ys};\n";
        print "    return createDFD422($bigEndian, 4, bits, shiftBits, channels, position_xs, position_ys, s_UNORM);\n}\n";

    } elsif ($line =~ m/VK_FORMAT_[RGBAE0-9]+_/) {
        # Set $format to the enum identifier
        ($line =~ m/(VK_FORMAT[A-Z0-9_]+)/);

        # $<number> holds the <number>'th parenthesised entry in the previous regex
        # (only one in this case)
        $format = $1;

        # Skip a format if we've already processed it
        if (!formatProcessed($format)) {

            if ($format =~ m/_E5B9G9R9/) {
                # Special case (assumed little-endian).
                print "case $format: {\n";
                print "    int bits[] = {0}; int channels[] = {0};\n";
                print "    return createDFDPacked(0, 6, bits, channels, s_UFLOAT);\n";
                print "}\n";
                $foundFormats{$format} = 1;
            } elsif ($format =~ m/_PACK/) {
                # Packed formats end "_PACK<n>" - is this format packed?

                # Extract the channel identifiers and suffix from the packed format
                $format =~ m/VK_FORMAT_([RGBA0-9]+)_([^_]+)_PACK[0-9]+/;

                # The first parenthesised bit of regex is the channels ("R5G5B5" etc.)
                $channels = $1;

                # The second parenthesised bit of regex is the suffix ("UNORM" etc.)
                $suffix = $2;

                # N.B. We don't care about the total bit count (after "PACK" in the name)

                # Create an empty array of channel names and corresponding bits
                my @packChannels = ();
                my @packBits = ();

                # Loop over channels, separating out the last letter followed by a number
                while ($channels =~ m/([RGBA0-9]*)([RGBA])([0-9]+)/) {

                    # Add the rightmost channel name to our array
                    push @packChannels, $2;

                    # Add the rightmost channel bits to the array
                    push @packBits, $3;

                    # Truncate the channels string to remove the channel we've processed
                    $channels = $1;
                }

                # The number of channels we've found is the array length we've built
                my $numChannels = @packChannels;

                # Packed output needs a C block for local variables
                print "case $format: {\n";

                # Start with a null list of channel ids
                my $channelIds = "";

                # Loop over the channel names we've found
                foreach (@packChannels) {

                    # Use a comma as a separator, so don't add it if the $channelIds string is empty
                    if ($channelIds ne "") { $channelIds .= ","; }

                    # Map the channel names to our internal numbering
                    if ($_ eq 'R') { $channelIds .= "0"; }
                    elsif ($_ eq 'G') { $channelIds .= "1"; }
                    elsif ($_ eq 'B') { $channelIds .= "2"; }
                    elsif ($_ eq 'A') { $channelIds .= "3"; }
                }

                # Channel bit counts are easier: we can use join() to make a comma-separated
                # string of the numbers in the array
                my $channelBits = join (',', @packBits);

                # Print initialisation for the two arrays we've created
                print "    int channels[] = {" . $channelIds . "}; ";
                print "int bits[] = {" . $channelBits . "};\n";

                # Now print the function call and close the block
                print "    return createDFDPacked($bigEndian, $numChannels, bits, channels, s_$suffix);\n";
                print "}\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;

                # If we're not packed, do we have a simple RGBA channel size list with a suffix?
                # N.B. We don't want to pick up downsampled or planar formats, which have more _-separated fields
                # - "$" matches the end of the format identifier
            } elsif ($format =~ m/VK_FORMAT_A([0-9]+)_([^_]+)(_[^_]+)?$/) {
                # Special case for alpha-only formats

                # Extract our "suffix" (e.g. "UNORM") and bytes per channel
                $suffix = $2;
                my $bytesPerChannel = $1 / 8;

                # Output the case entry
                print "case $format: return createDFDAlpha($bigEndian, $bytesPerChannel, s_$suffix);\n";
            } elsif ($format =~ m/VK_FORMAT_([RGBA0-9]+)_([^_]+)$/) {

                # Extract our "channels" (e.g. "B8G8R8") and "suffix" (e.g. "UNORM")
                $channels = $1;
                $suffix = $2;
                my $rbswap = 0;
                my $numChannels = 0;

                # Non-packed format either start with red (R8G8B8A8) or blue (B8G8R8A8)
                # We have a special case to notice when we start with blue
                if (substr($channels,0,1) eq "B") {

                    # Red and blue are swapped (B, G, R, A) - record this
                    # N.B. createDFDUnpacked() just knows this and R,G,B,A channel order, not arbitrary
                    $rbswap = 1;

                    # We know we saw "B" for blue, so we must also have red and green
                    $numChannels = 3;

                    # If we have "A" in the channels as well, we have four channels
                    if ($channels =~ m/A/) {
                        $numChannels = 4;
                    }
                } else {

                    # We didn't start "B", so we're in conventional order (R, G, B, A)
                    $rbswap = 0;

                    # Check for the channel names present and map that to the number of channels
                    if ($channels =~ m/A/) {
                        $numChannels = 4;
                    } elsif ($channels =~ m/B/) {
                        $numChannels = 3;
                    } elsif ($channels =~ m/G/) {
                        $numChannels = 2;
                    } else {
                        $numChannels = 1;
                    }
                }

                # In an unpacked format, all the channels are the same size, so we only need to check one
                $channels =~ m/R([0-9]+)/;

                # For unpacked, we need bytes per channel, not bits
                my $bytesPerChannel = $1 / 8;

                # Output the case entry
                print "case $format: return createDFDUnpacked($bigEndian, $numChannels, $bytesPerChannel, $rbswap, s_$suffix);\n";
                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;
            } elsif ($format =~ m/R16G16_SFIXED5_NV/) {
                # Currently only this 2-channel SFIXED5 format exists so an
                # explicit match is used.

                # Output the case entry
                print "case $format: return createDFDUnpacked($bigEndian, 2, 2, 0, s_SFIXED5);\n";
            }
        }

    # VK_FORMAT_ is a packed HDR formats with padding
    # R10X6_UNORM_PACK16
    # R10X6G10X6_UNORM_2PACK16
    # R10X6G10X6B10X6A10X6_UNORM_4PACK16
    # R12X4_UNORM_PACK16
    # R12X4G12X4_UNORM_2PACK16
    # R12X4G12X4B12X4A12X4_UNORM_4PACK16
    } elsif ($line =~ m/(VK_FORMAT_R10X6_UNORM_PACK16)/) {
        $format = $1; # Extract the format identifier from the rest of the line
        if (!exists($foundFormats{$format})) {
            $foundFormats{$format} = 1;
            print "case $format: {\n";
            print "    int channels[] = {0}; int bits[] = {10}; int shiftBits[] = {6};\n";
            print "    return createDFDPackedShifted($bigEndian, 1, bits, shiftBits, channels, s_UNORM);\n}\n"
        }
    } elsif ($line =~ m/(VK_FORMAT_R10X6G10X6_UNORM_2PACK16)/) {
        $format = $1; # Extract the format identifier from the rest of the line
        if (!exists($foundFormats{$format})) {
            $foundFormats{$format} = 1;
            print "case $format: {\n";
            print "    int channels[] = {0, 1}; int bits[] = {10, 10}; int shiftBits[] = {6, 6};\n";
            print "    return createDFDPackedShifted($bigEndian, 2, bits, shiftBits, channels, s_UNORM);\n}\n"
        }
    } elsif ($line =~ m/(VK_FORMAT_R10X6G10X6B10X6A10X6_UNORM_4PACK16)/) {
        $format = $1; # Extract the format identifier from the rest of the line
        if (!exists($foundFormats{$format})) {
            $foundFormats{$format} = 1;
            print "case $format: {\n";
            print "    int channels[] = {0, 1, 2, 3}; int bits[] = {10, 10, 10, 10}; int shiftBits[] = {6, 6, 6, 6};\n";
            print "    return createDFDPackedShifted($bigEndian, 4, bits, shiftBits, channels, s_UNORM);\n}\n"
        }
    } elsif ($line =~ m/(VK_FORMAT_R12X4_UNORM_PACK16)/) {
        $format = $1; # Extract the format identifier from the rest of the line
        if (!exists($foundFormats{$format})) {
            $foundFormats{$format} = 1;
            print "case $format: {\n";
            print "    int channels[] = {0}; int bits[] = {12}; int shiftBits[] = {4};\n";
            print "    return createDFDPackedShifted($bigEndian, 1, bits, shiftBits, channels, s_UNORM);\n}\n"
        }
    } elsif ($line =~ m/(VK_FORMAT_R12X4G12X4_UNORM_2PACK16)/) {
        $format = $1; # Extract the format identifier from the rest of the line
        if (!exists($foundFormats{$format})) {
            $foundFormats{$format} = 1;
            print "case $format: {\n";
            print "    int channels[] = {0, 1}; int bits[] = {12, 12}; int shiftBits[] = {4, 4};\n";
            print "    return createDFDPackedShifted($bigEndian, 2, bits, shiftBits, channels, s_UNORM);\n}\n"
        }
    } elsif ($line =~ m/(VK_FORMAT_R12X4G12X4B12X4A12X4_UNORM_4PACK16)/) {
        $format = $1; # Extract the format identifier from the rest of the line
        if (!exists($foundFormats{$format})) {
            $foundFormats{$format} = 1;
            print "case $format: {\n";
            print "    int channels[] = {0, 1, 2, 3}; int bits[] = {12, 12, 12, 12}; int shiftBits[] = {4, 4, 4, 4};\n";
            print "    return createDFDPackedShifted($bigEndian, 4, bits, shiftBits, channels, s_UNORM);\n}\n"
        }

    # If we weren't VK_FORMAT_ plus a channel, we might be a compressed
    # format, that ends "_BLOCK"
    } elsif ($line =~ m/(VK_FORMAT_[A-Z0-9x_]+_BLOCK(_[A-Z]+)?)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;

        # Skip a format if we've already processed it
        if (!formatProcessed($format)) {

            # Special-case BC1_RGB to separate it from BC1_RGBA
            if ($line =~ m/VK_FORMAT_BC1_RGB_([A-Z]+)_BLOCK/) {

                # Pull out the suffix ("UNORM" etc.)
                $suffix = $1;

                # Output the special case - a 4x4 BC1 block
                print "case $format: return createDFDCompressed(c_BC1_RGB, 4, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;

                # Special case BC1_RGBA (but still extract the suffix with a regex)
            } elsif ($line =~ m/VK_FORMAT_BC1_RGBA_([A-Z]+)_BLOCK/) {
                $suffix = $1;
                print "case $format: return createDFDCompressed(c_BC1_RGBA, 4, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;

                # All the other BC formats don't have a channel identifier in the name, so we regex match them
            } elsif ($line =~ m/VK_FORMAT_(BC(?:[2-57]|6H))_([A-Z]+)_BLOCK/) {
                $scheme = $1;
                $suffix = $2;
                print "case $format: return createDFDCompressed(c_$scheme, 4, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;

                # The ETC and EAC formats have two-part names (ETC2_R8G8B8, EAC_R11 etc.) starting with "E"
            } elsif ($line =~ m/VK_FORMAT_(E[^_]+_[^_]+)_([A-Z]+)_BLOCK/) {
                $scheme = $1;
                $suffix = $2;
                print "case $format: return createDFDCompressed(c_$scheme, 4, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;

                # Finally, ASTC and PVRTC, the only cases where the block size is a parameter
            } elsif ($line =~ m/VK_FORMAT_ASTC_([0-9]+)x([0-9]+)(x([0-9]+))?_([A-Z]+)_BLOCK(_EXT)?/) {
                my $w = $1;
                my $h = $2;
                my $d = $4 ? $4 : '1';
                $suffix = $5;
                print "case $format: return createDFDCompressed(c_ASTC, $w, $h, $d, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;
            } elsif ($line =~ m/VK_FORMAT_PVRTC1_2BPP_([A-Z]+)_BLOCK_IMG/) {

                # Pull out the suffix ("UNORM" etc.)
                $suffix = $1;

                # Output the special case - an 8x4 PVRTC block
                print "case $format: return createDFDCompressed(c_PVRTC, 8, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;
            } elsif ($line =~ m/VK_FORMAT_PVRTC1_4BPP_([A-Z]+)_BLOCK_IMG/) {

                # Pull out the suffix ("UNORM" etc.)
                $suffix = $1;

                # Output the special case - an 8x4 PVRTC block
                print "case $format: return createDFDCompressed(c_PVRTC, 4, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;
            } elsif ($line =~ m/VK_FORMAT_PVRTC2_2BPP_([A-Z]+)_BLOCK_IMG/) {

                # Pull out the suffix ("UNORM" etc.)
                $suffix = $1;

                # Output the special case - an 8x4 PVRTC block
                print "case $format: return createDFDCompressed(c_PVRTC2, 8, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;
            } elsif ($line =~ m/VK_FORMAT_PVRTC2_4BPP_([A-Z]+)_BLOCK_IMG/) {

                # Pull out the suffix ("UNORM" etc.)
                $suffix = $1;

                # Output the special case - an 8x4 PVRTC block
                print "case $format: return createDFDCompressed(c_PVRTC2, 4, 4, 1, s_$suffix);\n";

                # Add the format we've processed to our "done" hash
                $foundFormats{$format} = 1;
            }
        }
    } elsif ($line =~ m/(VK_FORMAT_X8_D24_UNORM_PACK32)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;
        if (!exists($foundFormats{$format})) {
            # Add the format we've processed to our "done" hash
            $foundFormats{$format} = 1;
            print "case $format: return createDFDDepthStencil(24,0,4);\n";
        }

    } elsif ($line =~ m/(VK_FORMAT_D32_SFLOAT_S8_UINT)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;
        if (!exists($foundFormats{$format})) {
            # Add the format we've processed to our "done" hash
            $foundFormats{$format} = 1;
            print "case $format: return createDFDDepthStencil(32,8,8);\n";
        }

    } elsif ($line =~ m/(VK_FORMAT_D16_UNORM_S8_UINT)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;
        if (!exists($foundFormats{$format})) {
            # Add the format we've processed to our "done" hash
            $foundFormats{$format} = 1;
            print "case $format: return createDFDDepthStencil(16,8,4);\n";
        }

    } elsif ($line =~ m/(VK_FORMAT_D24_UNORM_S8_UINT)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;
        if (!exists($foundFormats{$format})) {
            # Add the format we've processed to our "done" hash
            $foundFormats{$format} = 1;
            print "case $format: return createDFDDepthStencil(24,8,4);\n";
        }

    } elsif ($line =~ m/(VK_FORMAT_D32_SFLOAT)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;
        if (!exists($foundFormats{$format})) {
            # Add the format we've processed to our "done" hash
            $foundFormats{$format} = 1;
            print "case $format: return createDFDDepthStencil(32,0,4);\n";
        }
    } elsif ($line =~ m/(VK_FORMAT_S8_UINT)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;
        if (!exists($foundFormats{$format})) {
            # Add the format we've processed to our "done" hash
            $foundFormats{$format} = 1;
            print "case $format: return createDFDDepthStencil(0,8,1);\n";
        }
    } elsif ($line =~ m/(VK_FORMAT_D16_UNORM)/) {

        # Extract the format identifier from the rest of the line
        $format = $1;
        if (!exists($foundFormats{$format})) {
            # Add the format we've processed to our "done" hash
            $foundFormats{$format} = 1;
            print "case $format: return createDFDDepthStencil(16,0,2);\n";
        }
    }

    # ...and continue to the next line
}

# vim:ai:ts=4:sts=4:sw=2:expandtab