More 2-stage lookup tables

1 files changed, 113 insertions, 59 deletions

diff --git a/gen/prop/scf b/gen/prop/scf
index 8dcfcec..47cfc0a 100755
--- a/gen/prop/scf
+++ b/gen/prop/scf
@@ -1,59 +1,113 @@
-#!/bin/sh
-
-set -e
-cd "${0%/*}/../.."
-exec >lib/unicode/prop/uprop_get_scf.c
-
-gawk '
-BEGIN {
-	FS = "[ ;]+"
-
-	print "/* This file is autogenerated by gen/prop/scf; DO NOT EDIT. */"
-	print ""
-	print "#include \"_bsearch.h\""
-	print "#include \"macros.h\""
-	print "#include \"rune.h\""
-	print "#include \"unicode/prop.h\""
-	print ""
-}
-
-$0 !~ /^#/ && $2 ~ /[CS]/ {
-	map[strtonum("0X" $1)] = strtonum("0X" $3)
-}
-
-END {
-	print "static constexpr rune lookup_lat1[] = {"
-	for (i = 0; i < 0x100; i++) {
-		if (i % 8 == 0)
-			printf "\t"
-		printf "0x%03X,%s", map[i] ? map[i] : i, i % 8 == 7 ? "\n" : " "
-	}
-	print "};"
-	print ""
-	print "static const struct {"
-	print "\trune k, v;"
-	print "} lookup[] = {"
-
-	for (i = 0x100; i <= 0x10FFFF; i++) {
-		if (!map[i])
-			continue
-		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X)},\n", i, map[i]
-	}
-
-	print "};"
-	print ""
-	print "_MLIB_DEFINE_BSEARCH_KV(rune, lookup, ch)"
-	print ""
-	print "rune"
-	print "uprop_get_scf(rune ch, bool az_or_tr)"
-	print "{"
-	print "\tif (az_or_tr) {"
-	print "\t\tif (ch == \x27I\x27)"
-	print "\t\t\treturn U\x27ı\x27;"
-	print "\t\tif (ch == U\x27İ\x27)"
-	print "\t\t\treturn \x27i\x27;"
-	print "\t}"
-	print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup_kv(ch);"
-	print "}"
-}
-' data/CaseFolding
+#!/usr/bin/python3
+
+import math
+
+from lib import *
+
+
+longest = 0
+
+def parse(file: str) -> list[bool]:
+	global longest
+
+	xs = ['0'] * 0x110000
+	with open(file, 'r') as f:
+		for line in f.readlines():
+			if len(line.strip()) == 0 or line[0] == '#':
+				continue
+
+			parts = line.split(';')
+			if parts[1].strip() not in {'C', 'S'}:
+				continue
+			ranges = [int(x, 16) for x in parts[0].strip().split('..')]
+			prop = f'RUNE_C(0x{parts[2].split('#')[0].strip()})'
+			longest = max(longest, len(prop))
+
+			for i in range(ranges[0], ranges[len(ranges) - 1] + 1):
+				xs[i] = prop
+	return xs
+
+def genfile(cs: list[tuple[bool, ...]], blksize: int) -> None:
+	Cs = cs
+	cs = list(dict.fromkeys(Cs))
+
+	print('''\
+/* This file is autogenerated by gen/prop/scf; DO NOT EDIT. */
+
+#include <stdint.h>
+
+#include "rune.h"
+#include "unicode/prop.h"
+''')
+
+	print(f'static constexpr {typename(len(cs) - 1)} stage1[] = {{')
+	for i, c in enumerate(Cs):
+		print(f'%c%{len(str(len(cs) - 1))}d,' % ('\t' if i % 16 == 0 else ' ', cs.index(c)), end='')
+		if i % 16 == 15:
+			print()
+	print('};')
+
+	print()
+
+	ppc = columns(blksize, longest + 1)
+	print(f'static constexpr rune stage2[][{blksize}] = {{')
+	for c in cs:
+		for i in range(blksize // ppc):
+			print('\t{' if i == 0 else '\t ', end='')
+			for j in range(ppc):
+				print(c[i*ppc + j], end='')
+				if i < blksize // ppc - 1 or j < ppc - 1:
+					print(',', end='')
+				if j < ppc - 1:
+					print(' ' * (longest + 1 - len(c[i*ppc + j])), end='')
+			if i < blksize // ppc - 1:
+				print()
+		print('},')
+	print('};')
+
+	print()
+
+	print(f'''\
+rune
+uprop_get_scf(rune ch, bool az_tr)
+{{
+	if (az_tr) {{
+		if (ch == 'I')
+			return U'ı';
+		if (ch == U'İ')
+			return 'i';
+	}}
+	rune hc = stage2[stage1[ch / {blksize}]][ch % {blksize}];
+	return hc == 0 ? ch : hc;
+}}''')
+
+def main() -> None:
+	cwd_init()
+	xs = parse('data/CaseFolding')
+
+	blksize = -1
+	smallest = math.inf
+
+	for bs in powers_of_2():
+		if bs > len(xs):
+			break
+		Cs = [tuple(x) for x in chunks(xs, bs)]
+		cs = set(Cs)
+
+		sz_s1 = len(Cs) * isize(len(cs) - 1)
+		sz_s2 = len(cs) * bs * 4
+		sz = sz_s1 + sz_s2
+
+		if sz < smallest:
+			smallest = sz
+			blksize = bs
+
+	Cs = [tuple(x) for x in chunks(xs, blksize)]
+	with open('lib/unicode/prop/uprop_get_scf.c', 'w') as f:
+		sys.stdout = f
+		genfile(Cs, blksize)
+
+	report_size(len(xs), smallest)
+
+if __name__ == '__main__':
+	main()