#-------------------------------------------------------------------------
#
# convert_conversion2dat.pl
#     Script to convert the ad hoc representation of conversions and their
#     functions into data entries.
#
#-------------------------------------------------------------------------

use strict;
use warnings;

use FindBin;
use lib "$FindBin::RealBin/../../postgresql/src/backend/catalog/";
use Catalog;

my $output_path = 'src/include/catalog';
my $full_tuples = 0;


# conversion_name source_encoding destination_encoding function object
my $CONVERSIONS = <<EOM;
ascii_to_mic	SQL_ASCII MULE_INTERNAL ascii_to_mic ascii_and_mic
mic_to_ascii	MULE_INTERNAL SQL_ASCII mic_to_ascii ascii_and_mic
koi8_r_to_mic	KOI8R MULE_INTERNAL koi8r_to_mic cyrillic_and_mic
mic_to_koi8_r	MULE_INTERNAL KOI8R mic_to_koi8r cyrillic_and_mic
iso_8859_5_to_mic	ISO-8859-5 MULE_INTERNAL iso_to_mic cyrillic_and_mic
mic_to_iso_8859_5	MULE_INTERNAL ISO-8859-5 mic_to_iso cyrillic_and_mic
windows_1251_to_mic	WIN1251 MULE_INTERNAL win1251_to_mic cyrillic_and_mic
mic_to_windows_1251	MULE_INTERNAL WIN1251 mic_to_win1251 cyrillic_and_mic
windows_866_to_mic	WIN866 MULE_INTERNAL win866_to_mic cyrillic_and_mic
mic_to_windows_866	MULE_INTERNAL WIN866 mic_to_win866 cyrillic_and_mic
koi8_r_to_windows_1251   KOI8R WIN1251 koi8r_to_win1251 cyrillic_and_mic
windows_1251_to_koi8_r   WIN1251 KOI8R win1251_to_koi8r cyrillic_and_mic
koi8_r_to_windows_866	KOI8R WIN866 koi8r_to_win866 cyrillic_and_mic
windows_866_to_koi8_r	WIN866 KOI8R win866_to_koi8r cyrillic_and_mic
windows_866_to_windows_1251	WIN866 WIN1251 win866_to_win1251 cyrillic_and_mic
windows_1251_to_windows_866	WIN1251	WIN866 win1251_to_win866 cyrillic_and_mic
iso_8859_5_to_koi8_r	ISO-8859-5 KOI8R iso_to_koi8r cyrillic_and_mic
koi8_r_to_iso_8859_5	KOI8R ISO-8859-5 koi8r_to_iso cyrillic_and_mic
iso_8859_5_to_windows_1251	ISO-8859-5 WIN1251 iso_to_win1251 cyrillic_and_mic
windows_1251_to_iso_8859_5	WIN1251 ISO-8859-5 win1251_to_iso cyrillic_and_mic
iso_8859_5_to_windows_866	ISO-8859-5 WIN866 iso_to_win866 cyrillic_and_mic
windows_866_to_iso_8859_5	WIN866 ISO-8859-5 win866_to_iso cyrillic_and_mic
euc_cn_to_mic	EUC_CN MULE_INTERNAL euc_cn_to_mic euc_cn_and_mic
mic_to_euc_cn	MULE_INTERNAL EUC_CN mic_to_euc_cn euc_cn_and_mic
euc_jp_to_sjis	EUC_JP SJIS euc_jp_to_sjis euc_jp_and_sjis
sjis_to_euc_jp	SJIS EUC_JP sjis_to_euc_jp euc_jp_and_sjis
euc_jp_to_mic	EUC_JP MULE_INTERNAL euc_jp_to_mic euc_jp_and_sjis
sjis_to_mic	SJIS MULE_INTERNAL sjis_to_mic euc_jp_and_sjis
mic_to_euc_jp	MULE_INTERNAL EUC_JP mic_to_euc_jp euc_jp_and_sjis
mic_to_sjis	MULE_INTERNAL SJIS mic_to_sjis euc_jp_and_sjis
euc_kr_to_mic	EUC_KR MULE_INTERNAL euc_kr_to_mic euc_kr_and_mic
mic_to_euc_kr	MULE_INTERNAL EUC_KR mic_to_euc_kr euc_kr_and_mic
euc_tw_to_big5	EUC_TW BIG5 euc_tw_to_big5 euc_tw_and_big5
big5_to_euc_tw	BIG5 EUC_TW big5_to_euc_tw euc_tw_and_big5
euc_tw_to_mic	EUC_TW MULE_INTERNAL euc_tw_to_mic euc_tw_and_big5
big5_to_mic	BIG5 MULE_INTERNAL big5_to_mic euc_tw_and_big5
mic_to_euc_tw	MULE_INTERNAL EUC_TW mic_to_euc_tw euc_tw_and_big5
mic_to_big5	MULE_INTERNAL BIG5 mic_to_big5 euc_tw_and_big5
iso_8859_2_to_mic	LATIN2 MULE_INTERNAL latin2_to_mic latin2_and_win1250
mic_to_iso_8859_2	MULE_INTERNAL LATIN2 mic_to_latin2 latin2_and_win1250
windows_1250_to_mic	WIN1250 MULE_INTERNAL win1250_to_mic latin2_and_win1250
mic_to_windows_1250	MULE_INTERNAL WIN1250 mic_to_win1250 latin2_and_win1250
iso_8859_2_to_windows_1250  LATIN2 WIN1250 latin2_to_win1250 latin2_and_win1250
windows_1250_to_iso_8859_2  WIN1250 LATIN2 win1250_to_latin2 latin2_and_win1250
iso_8859_1_to_mic	LATIN1 MULE_INTERNAL latin1_to_mic latin_and_mic
mic_to_iso_8859_1	MULE_INTERNAL LATIN1 mic_to_latin1 latin_and_mic
iso_8859_3_to_mic	LATIN3 MULE_INTERNAL latin3_to_mic latin_and_mic
mic_to_iso_8859_3	MULE_INTERNAL LATIN3 mic_to_latin3 latin_and_mic
iso_8859_4_to_mic	LATIN4 MULE_INTERNAL latin4_to_mic latin_and_mic
mic_to_iso_8859_4	MULE_INTERNAL LATIN4 mic_to_latin4 latin_and_mic
ascii_to_utf8 SQL_ASCII UTF8 ascii_to_utf8 utf8_and_ascii
utf8_to_ascii UTF8 SQL_ASCII utf8_to_ascii utf8_and_ascii
big5_to_utf8 BIG5 UTF8 big5_to_utf8 utf8_and_big5
utf8_to_big5 UTF8 BIG5 utf8_to_big5 utf8_and_big5
utf8_to_koi8_r	UTF8 KOI8R utf8_to_koi8r utf8_and_cyrillic
koi8_r_to_utf8	KOI8R UTF8 koi8r_to_utf8 utf8_and_cyrillic
utf8_to_koi8_u	UTF8 KOI8U utf8_to_koi8u utf8_and_cyrillic
koi8_u_to_utf8	KOI8U UTF8 koi8u_to_utf8 utf8_and_cyrillic
utf8_to_windows_866 UTF8 WIN866 utf8_to_win utf8_and_win
windows_866_to_utf8 WIN866 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_874 UTF8 WIN874 utf8_to_win utf8_and_win
windows_874_to_utf8 WIN874 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1250 UTF8 WIN1250 utf8_to_win utf8_and_win
windows_1250_to_utf8 WIN1250 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1251 UTF8 WIN1251 utf8_to_win utf8_and_win
windows_1251_to_utf8 WIN1251 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1252 UTF8 WIN1252 utf8_to_win utf8_and_win
windows_1252_to_utf8 WIN1252 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1253 UTF8 WIN1253 utf8_to_win utf8_and_win
windows_1253_to_utf8 WIN1253 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1254 UTF8 WIN1254 utf8_to_win utf8_and_win
windows_1254_to_utf8 WIN1254 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1255 UTF8 WIN1255 utf8_to_win utf8_and_win
windows_1255_to_utf8 WIN1255 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1256 UTF8 WIN1256 utf8_to_win utf8_and_win
windows_1256_to_utf8 WIN1256 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1257 UTF8 WIN1257 utf8_to_win utf8_and_win
windows_1257_to_utf8 WIN1257 UTF8 win_to_utf8 utf8_and_win
utf8_to_windows_1258 UTF8 WIN1258 utf8_to_win utf8_and_win
windows_1258_to_utf8 WIN1258 UTF8 win_to_utf8 utf8_and_win
euc_cn_to_utf8 EUC_CN UTF8 euc_cn_to_utf8 utf8_and_euc_cn
utf8_to_euc_cn UTF8 EUC_CN utf8_to_euc_cn utf8_and_euc_cn
euc_jp_to_utf8 EUC_JP UTF8 euc_jp_to_utf8 utf8_and_euc_jp
utf8_to_euc_jp UTF8 EUC_JP utf8_to_euc_jp utf8_and_euc_jp
euc_kr_to_utf8 EUC_KR UTF8 euc_kr_to_utf8 utf8_and_euc_kr
utf8_to_euc_kr UTF8 EUC_KR utf8_to_euc_kr utf8_and_euc_kr
euc_tw_to_utf8 EUC_TW UTF8 euc_tw_to_utf8 utf8_and_euc_tw
utf8_to_euc_tw UTF8 EUC_TW utf8_to_euc_tw utf8_and_euc_tw
gb18030_to_utf8 GB18030 UTF8 gb18030_to_utf8 utf8_and_gb18030
utf8_to_gb18030 UTF8 GB18030 utf8_to_gb18030 utf8_and_gb18030
gbk_to_utf8 GBK UTF8 gbk_to_utf8 utf8_and_gbk
utf8_to_gbk UTF8 GBK utf8_to_gbk utf8_and_gbk
utf8_to_iso_8859_2 UTF8 LATIN2 utf8_to_iso8859 utf8_and_iso8859
iso_8859_2_to_utf8 LATIN2 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_3 UTF8 LATIN3 utf8_to_iso8859 utf8_and_iso8859
iso_8859_3_to_utf8 LATIN3 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_4 UTF8 LATIN4 utf8_to_iso8859 utf8_and_iso8859
iso_8859_4_to_utf8 LATIN4 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_9 UTF8 LATIN5 utf8_to_iso8859 utf8_and_iso8859
iso_8859_9_to_utf8 LATIN5 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_10 UTF8 LATIN6 utf8_to_iso8859 utf8_and_iso8859
iso_8859_10_to_utf8 LATIN6 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_13 UTF8 LATIN7 utf8_to_iso8859 utf8_and_iso8859
iso_8859_13_to_utf8 LATIN7 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_14 UTF8 LATIN8 utf8_to_iso8859 utf8_and_iso8859
iso_8859_14_to_utf8 LATIN8 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_15 UTF8 LATIN9 utf8_to_iso8859 utf8_and_iso8859
iso_8859_15_to_utf8 LATIN9 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_16 UTF8 LATIN10 utf8_to_iso8859 utf8_and_iso8859
iso_8859_16_to_utf8 LATIN10 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_5 UTF8 ISO-8859-5 utf8_to_iso8859 utf8_and_iso8859
iso_8859_5_to_utf8 ISO-8859-5 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_6 UTF8 ISO-8859-6 utf8_to_iso8859 utf8_and_iso8859
iso_8859_6_to_utf8 ISO-8859-6 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_7 UTF8 ISO-8859-7 utf8_to_iso8859 utf8_and_iso8859
iso_8859_7_to_utf8 ISO-8859-7 UTF8 iso8859_to_utf8 utf8_and_iso8859
utf8_to_iso_8859_8 UTF8 ISO-8859-8 utf8_to_iso8859 utf8_and_iso8859
iso_8859_8_to_utf8 ISO-8859-8 UTF8 iso8859_to_utf8 utf8_and_iso8859
iso_8859_1_to_utf8 LATIN1 UTF8 iso8859_1_to_utf8 utf8_and_iso8859_1
utf8_to_iso_8859_1 UTF8 LATIN1 utf8_to_iso8859_1 utf8_and_iso8859_1
johab_to_utf8 JOHAB UTF8 johab_to_utf8 utf8_and_johab
utf8_to_johab UTF8 JOHAB utf8_to_johab utf8_and_johab
sjis_to_utf8 SJIS UTF8 sjis_to_utf8 utf8_and_sjis
utf8_to_sjis UTF8 SJIS utf8_to_sjis utf8_and_sjis
uhc_to_utf8 UHC UTF8 uhc_to_utf8 utf8_and_uhc
utf8_to_uhc UTF8 UHC utf8_to_uhc utf8_and_uhc
euc_jis_2004_to_utf8 EUC_JIS_2004 UTF8 euc_jis_2004_to_utf8 utf8_and_euc2004
utf8_to_euc_jis_2004 UTF8 EUC_JIS_2004 utf8_to_euc_jis_2004 utf8_and_euc2004
shift_jis_2004_to_utf8 SHIFT_JIS_2004 UTF8 shift_jis_2004_to_utf8 utf8_and_sjis2004
utf8_to_shift_jis_2004 UTF8 SHIFT_JIS_2004 utf8_to_shift_jis_2004 utf8_and_sjis2004
euc_jis_2004_to_shift_jis_2004 EUC_JIS_2004 SHIFT_JIS_2004 euc_jis_2004_to_shift_jis_2004 euc2004_sjis2004
shift_jis_2004_to_euc_jis_2004 SHIFT_JIS_2004 EUC_JIS_2004 shift_jis_2004_to_euc_jis_2004 euc2004_sjis2004
EOM


my %catalogs;
my %catalog_data;
my @catnames = qw(pg_proc pg_conversion);

$catalogs{pg_proc}       = Catalog::ParseHeader('src/include/catalog/pg_proc.h');
$catalogs{pg_conversion} = Catalog::ParseHeader('src/include/catalog/pg_conversion.h');

# Make sure output_path ends in a slash.
if ($output_path ne '' && substr($output_path, -1) ne '/')
{
	$output_path .= '/';
}

# Metadata of a catalog entry
my @METADATA = ('oid', 'oid_symbol', 'descr');


open my $fh, '<', \$CONVERSIONS or die $!;
my $proc_oid = 4600;
my $conv_oid = 4800;
my %seenit;
while (<$fh>) {
	chomp;
	my ($conversion_name, $source_encoding, $destination_encoding, $function, $object) = split /\s+/;

	my %proc_tuple;
	my %conv_tuple;

	$proc_tuple{oid} = $proc_oid;
	$proc_tuple{descr} =
		sprintf "internal conversion function for %s to %s",
		  $function eq 'iso8859_to_utf8' ? 'ISO-8859 2-16'
		: $function eq 'win_to_utf8'     ? 'WIN'
		:                                  $source_encoding,

		  $function eq 'utf8_to_iso8859' ? 'ISO-8859 2-16'
		: $function eq 'utf8_to_win'     ? 'WIN'
		:                                  $destination_encoding;

	$proc_tuple{proname} = $function;
	$proc_tuple{prolang} = '13';
	# Assume immutable, unlike the .sql script
	$proc_tuple{prorettype} = 'void';
	$proc_tuple{proargtypes} = 'int4 int4 cstring internal int4';
	$proc_tuple{prosrc} = $function;
	$proc_tuple{probin} = '$libdir/' . $object;

	$conv_tuple{oid} = $conv_oid;
	$conv_tuple{descr} =
		sprintf "conversion for %s to %s",
		  $source_encoding, $destination_encoding;

	# Make encoding names match pg_wchar.h
	$source_encoding      =~ s/-/_/g;
	$destination_encoding =~ s/-/_/g;
	$conv_tuple{conforencoding} = "PG_$source_encoding";
	$conv_tuple{contoencoding}  = "PG_$destination_encoding";

	$conv_tuple{conname} = $conversion_name;
	$conv_tuple{conproc} = $function;

	if (!$seenit{$proc_tuple{proname}})
	{
		push @{ $catalog_data{pg_proc} }, \%proc_tuple;
		$proc_oid++;
	}
	$seenit{$proc_tuple{proname}}++;

	push @{ $catalog_data{pg_conversion} }, \%conv_tuple;
	$conv_oid++;
}

# Write the data.
foreach my $catname (@catnames)
{
	my $catalog = $catalogs{$catname};
	my @attnames;
	my $schema = $catalog->{columns};

	foreach my $column (@$schema)
	{
		my $attname = $column->{name};
		push @attnames, $attname;
	}

	# Write to temp files and copy over later.
	my $datfile = "$output_path${catname}_tmp.dat";
	open my $dat, '>', $datfile
	  or die "can't open $datfile: $!";

	# Write the data.
	foreach my $data (@{ $catalog_data{$catname} })
	{

		# Hash ref representing a data entry.
		if (ref $data eq 'HASH')
		{
			my %values = %$data;

			print $dat "{";

			# Separate out metadata fields for readability.
			my $metadata_str = format_hash(\%values, @METADATA);
			if ($metadata_str)
			{
				print $dat $metadata_str;

				# User attributes start on next line.
				print $dat ",\n ";
			}

			my $data_str = format_hash(\%values, @attnames);
			print $dat $data_str;
			print $dat " },\n";
		}

	}
	close $dat;
}

# Format the individual elements of a Perl hash into a valid string
# representation. We do this ourselves, rather than use native Perl
# facilities, so we can keep control over the exact formatting of the
# data files.
sub format_hash
{
	my $data          = shift;
	my @orig_attnames = @_;

	# Copy attname to new array if it has a value, so we can determine
	# the last populated element. We do this because we may have default
	# values or empty metadata fields.
	my @attnames;
	foreach my $orig_attname (@orig_attnames)
	{
		push @attnames, $orig_attname
		  if defined $data->{$orig_attname};
	}

	# When calling this function, we ether have an open-bracket or a
	# leading space already.
	my $char_count = 1;

	my $threshold;
	my $hash_str      = '';
	my $element_count = 0;

	foreach my $attname (@attnames)
	{
		$element_count++;

		# To limit the line to 80 chars, we need to account for the
		# trailing characters.
		if ($element_count == $#attnames + 1)
		{
			# Last element, so allow space for ' },'
			$threshold = 77;
		}
		else
		{
			# Just need space for trailing comma
			$threshold = 79;
		}

		if ($element_count > 1)
		{
			$hash_str .= ',';
			$char_count++;
		}

		my $value = $data->{$attname};

		# Escape single quotes.
		$value =~ s/'/\\'/g;

		# Include a leading space in the key-value pair, since this will
		# always go after either a comma or an additional padding space on
		# the next line.
		my $element        = " $attname => '$value'";
		my $element_length = length($element);

		# If adding the element to the current line would expand the line
		# beyond 80 chars, put it on the next line. We don't do this for
		# the first element, since that would create a blank line.
		if ($element_count > 1 and $char_count + $element_length > $threshold)
		{

			# Put on next line with an additional space preceding. There
			# are now two spaces in front of the key-value pair, lining
			# it up with the line above it.
			$hash_str .= "\n $element";
			$char_count = $element_length + 1;
		}
		else
		{
			$hash_str .= $element;
			$char_count += $element_length;
		}
	}
	return $hash_str;
}