From 1b068cedf3c113f27121182b29e096ccb3e1b9c1 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Fri, 3 May 2024 18:37:50 +0200
Subject: Add more 2-stage lookup tables

---
 gen/prop/nfXX_qc | 185 ++++++++++++++++++++++++++++++++-----------------------
 1 file changed, 109 insertions(+), 76 deletions(-)

(limited to 'gen/prop/nfXX_qc')

diff --git a/gen/prop/nfXX_qc b/gen/prop/nfXX_qc
index fed603e..24fcf45 100755
--- a/gen/prop/nfXX_qc
+++ b/gen/prop/nfXX_qc
@@ -1,77 +1,110 @@
-#!/bin/sh
-
-set -e
-cd "${0%/*}/../.."
-
-for x in c d kc kd
-do
-	gawk -v s=$x '
-	BEGIN {
-		FS = "[ ;]+"
-
-		_default = "NF" toupper(s) "_QC_Y"
-
-		print "/* This file is autogenerated by gen/prop/nfXX_qc; DO NOT EDIT. */"
-		print ""
-		print "/* The macros.h include may be unused */"
-		print ""
-		print "#include \"_bsearch.h\""
-		print "#include \"macros.h\""
-		print "#include \"rune.h\""
-		print "#include \"unicode/prop.h\""
-		print ""
-	}
-
-	/^[A-F0-9]/ && $2 == "NF" toupper(s) "_QC" {
-		n = split($1, a, /\.\./)
-		lo = strtonum("0X" a[1])
-		hi = strtonum("0X" a[n])
-
-		for (i = lo; i <= hi; i++)
-			props[i] = "NF" toupper(s) "_QC_" $3
-		if (lo < 0x100)
-			want_lat1_tbl = 1
-	}
-
-	END {
-		if (want_lat1_tbl) {
-			print "static constexpr enum uprop_nf" s "_qc lookup_lat1[] = {"
-			for (i = 0; i < 0x100; i++) {
-				if (i % 8 == 0)
-					printf "\t"
-				printf "%s,%s", props[i] ? props[i] : _default, \
-					i % 8 == 7 ? "\n" : " "
-			}
-			print "};"
-			print ""
-		}
-
-		print "static const struct {"
-		print "\trune lo, hi;"
-		print "\tenum uprop_nf" s "_qc val;"
-		print "} lookup[] = {"
-
-		for (i = want_lat1_tbl ? 0x100 : 0; i <= 0x10FFFF; i++) {
-			if (!props[i])
+#!/usr/bin/python3
+
+import math
+
+from lib import *
+
+
+longest = 0
+TYPES = ['c', 'd', 'kc', 'kd']
+
+def parse(file: str, _type: str) -> list[bool]:
+	global longest
+
+	_type = _type.upper()
+
+	xs = [f'NF{_type}_QC_Y'] * 0x110000
+	with open(file, 'r') as f:
+		for line in f.readlines():
+			if (
+				len(line.strip()) == 0
+				or line[0] == '#'
+				or f'NF{_type}_QC' not in line
+			):
 				continue
-			for (lo = i; props[lo] == props[i + 1]; i++)
-				;
-			printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[i]
-		}
-
-		print "};"
-		print ""
-		print "_MLIB_DEFINE_BSEARCH(enum uprop_nf" s "_qc, lookup, " _default ")"
-		print ""
-		print "enum uprop_nf" s "_qc"
-		print "uprop_get_nf" s "_qc(rune ch)"
-		print "{"
-		if (want_lat1_tbl)
-			print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
-		else
-			print "\treturn ch < lookup[0].lo ? " _default " : mlib_lookup(ch);"
-		print "}"
-	}
-	' data/DerivedNormalizationProps \
-	| sed 's/\s*$//' >lib/unicode/prop/uprop_get_nf${x}_qc.c
-done
+
+			parts = line.split(';')
+			ranges = [int(x, 16) for x in parts[0].strip().split('..')]
+			prop = f'NF{_type}_QC_' + 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, _type: str) -> None:
+	Cs = cs
+	cs = list(dict.fromkeys(Cs))
+
+	print('''\
+/* This file is autogenerated by gen/prop/nfXX_qc; DO NOT EDIT. */
+
+#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 enum uprop_nf{_type}_qc 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'''\
+enum uprop_nf{_type}_qc
+uprop_get_nf{_type}_qc(rune ch)
+{{
+	return stage2[stage1[ch / {blksize}]][ch % {blksize}];
+}}''')
+
+def main(_type: str) -> None:
+	cwd_init()
+	xs = parse('data/DerivedNormalizationProps', _type)
+
+	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
+		sz = sz_s1 + sz_s2
+
+		if sz < smallest:
+			smallest = sz
+			blksize = bs
+
+	Cs = [tuple(x) for x in chunks(xs, blksize)]
+	with open(f'lib/unicode/prop/uprop_get_nf{_type}_qc.c', 'w') as f:
+		sys.stdout = f
+		genfile(Cs, blksize, _type)
+
+	report_size(len(xs), smallest)
+
+if __name__ == '__main__':
+	for _type in TYPES:
+		longest = 0
+		main(_type)
-- 
cgit v1.2.3