Generate GB18030 mappings from the Unicode Consortium's UCM file

Previously we built the .map files for GB18030 (version 2000) from an
XML file. The 2022 version for this encoding is only available as a
Unicode Character Mapping (UCM) file, so as preparatory refactoring
switch to this format as the source for building version 2000.

As we do with most input files for the conversion mappings, download
the file on demand. In order to generate the same mappings we have
now, we must download from a previous upstream commit, rather than
the head since the latter contains a correction not present in our
current .map files.

The XML file is still used by EUC_CN, so we cannot delete it from our
repository. GB18030 is a superset of EUC_CN, so it may be possible to
build EUC_CN from the same UCM file, but that is left for future work.

Author: Chao Li <lic@highgo.com>
Discussion: https://postgr.es/m/966d9fc.169.198741fe60b.Coremail.jiaoshuntian%40highgo.com

3 files changed, 29 insertions, 11 deletions

diff --git a/src/backend/utils/mb/Unicode/Makefile b/src/backend/utils/mb/Unicode/Makefile
index ad789b31e54..27424b2a001 100644
--- a/src/backend/utils/mb/Unicode/Makefile
+++ b/src/backend/utils/mb/Unicode/Makefile
@@ -54,7 +54,7 @@ $(eval $(call map_rule,euc_cn,UCS_to_EUC_CN.pl,gb-18030-2000.xml))
 $(eval $(call map_rule,euc_kr,UCS_to_EUC_KR.pl,KSX1001.TXT))
 $(eval $(call map_rule,euc_tw,UCS_to_EUC_TW.pl,CNS11643.TXT))
 $(eval $(call map_rule,sjis,UCS_to_SJIS.pl,CP932.TXT))
-$(eval $(call map_rule,gb18030,UCS_to_GB18030.pl,gb-18030-2000.xml))
+$(eval $(call map_rule,gb18030,UCS_to_GB18030.pl,gb-18030-2000.ucm))
 $(eval $(call map_rule,big5,UCS_to_BIG5.pl,CP950.TXT BIG5.TXT CP950.TXT))
 $(eval $(call map_rule,euc_jis_2004,UCS_to_EUC_JIS_2004.pl,euc-jis-2004-std.txt))
 $(eval $(call map_rule,shift_jis_2004,UCS_to_SHIFT_JIS_2004.pl,sjis-0213-2004-std.txt))
@@ -78,6 +78,9 @@ euc-jis-2004-std.txt sjis-0213-2004-std.txt:
 gb-18030-2000.xml windows-949-2000.xml:
 	$(DOWNLOAD) https://raw.githubusercontent.com/unicode-org/icu-data/master/charset/data/xml/$(@F)
 
+gb-18030-2000.ucm:
+	$(DOWNLOAD) https://raw.githubusercontent.com/unicode-org/icu-data/d9d3a6ed27bb98a7106763e940258f0be8cd995b/charset/data/ucm/$(@F)
+
 GB2312.TXT:
 	$(DOWNLOAD) 'http://trac.greenstone.org/browser/trunk/gsdl/unicode/MAPPINGS/EASTASIA/GB/GB2312.TXT?rev=1842&format=txt'
 
diff --git a/src/backend/utils/mb/Unicode/UCS_to_GB18030.pl b/src/backend/utils/mb/Unicode/UCS_to_GB18030.pl
index ddcbd6ef0c4..084fdf66af1 100755
--- a/src/backend/utils/mb/Unicode/UCS_to_GB18030.pl
+++ b/src/backend/utils/mb/Unicode/UCS_to_GB18030.pl
@@ -5,13 +5,14 @@
 # src/backend/utils/mb/Unicode/UCS_to_GB18030.pl
 #
 # Generate UTF-8 <--> GB18030 code conversion tables from
-# "gb-18030-2000.xml", obtained from
-# http://source.icu-project.org/repos/icu/data/trunk/charset/data/xml/
+# "gb-18030-2000.ucm", obtained from
+# https://github.com/unicode-org/icu-data/tree/main/charset/data/ucm
 #
 # The lines we care about in the source file look like
-#    <a u="009A" b="81 30 83 36"/>
-# where the "u" field is the Unicode code point in hex,
-# and the "b" field is the hex byte sequence for GB18030
+#   <UXXXX> \xYY[\xYY...] |n
+# where XXXX is the Unicode code point in hex,
+# and the \xYY... is the hex byte sequence for GB18030,
+# and n is a flag indicating the type of mapping.
 
 use strict;
 use warnings FATAL => 'all';
@@ -22,7 +23,7 @@ my $this_script = 'src/backend/utils/mb/Unicode/UCS_to_GB18030.pl';
 
 # Read the input
 
-my $in_file = "gb-18030-2000.xml";
+my $in_file = "gb-18030-2000.ucm";
 
 open(my $in, '<', $in_file) || die("cannot open $in_file");
 
@@ -30,9 +31,18 @@ my @mapping;
 
 while (<$in>)
 {
-	next if (!m/<a u="([0-9A-F]+)" b="([0-9A-F ]+)"/);
-	my ($u, $c) = ($1, $2);
-	$c =~ s/ //g;
+	# Mappings may have been removed by commenting out
+	next if /^#/;
+
+	next if !/^<U([0-9A-Fa-f]+)>\s+
+			((?:\\x[0-9A-Fa-f]{2})+)\s+
+			\|(\d+)/x;
+	my ($u, $c, $flag) = ($1, $2, $3);
+	$c =~ s/\\x//g;
+
+	# We only want round-trip mappings
+	next if ($flag ne '0');
+
 	my $ucs = hex($u);
 	my $code = hex($c);
 	if ($code >= 0x80 && $ucs >= 0x0080)
diff --git a/src/backend/utils/mb/conversion_procs/utf8_and_gb18030/utf8_and_gb18030.c b/src/backend/utils/mb/conversion_procs/utf8_and_gb18030/utf8_and_gb18030.c
index ffc9c58cd13..a512df93577 100644
--- a/src/backend/utils/mb/conversion_procs/utf8_and_gb18030/utf8_and_gb18030.c
+++ b/src/backend/utils/mb/conversion_procs/utf8_and_gb18030/utf8_and_gb18030.c
@@ -124,7 +124,12 @@ utf8word_to_unicode(uint32 c)
 /*
  * Perform mapping of GB18030 ranges to UTF8
  *
- * The ranges we need to convert are specified in gb-18030-2000.xml.
+ * General description, and the range we need to convert for U+10000 and up:
+ * https://htmlpreview.github.io/?https://github.com/unicode-org/icu-data/blob/main/charset/source/gb18030/gb18030.html
+ *
+ * Ranges up to U+FFFF:
+ * https://github.com/unicode-org/icu-data/blob/main/charset/source/gb18030/ranges.txt
+ *
  * All are ranges of 4-byte GB18030 codes.
  */
 static uint32