Add more 2-stage lookup tables

1 files changed, 101 insertions, 65 deletions

diff --git a/gen/prop/lb b/gen/prop/lb
index 2abab6c..e003d13 100755
--- a/gen/prop/lb
+++ b/gen/prop/lb
@@ -1,65 +1,101 @@
-#!/bin/sh
-
-set -e
-cd "${0%/*}/../.."
-exec >lib/unicode/prop/uprop_get_lb.c
-
-gawk '
-BEGIN {
-	FS = "( *#.*| *; +)"
-
-	print "/* This file is autogenerated by gen/prop/lb; DO NOT EDIT. */"
-	print ""
-	print "#include \"_bsearch.h\""
-	print "#include \"macros.h\""
-	print "#include \"rune.h\""
-	print "#include \"unicode/prop.h\""
-	print ""
-}
-
-/^[^#]/ {
-	n = split($1, a, /\.\./)
-	lo = strtonum("0X" a[1])
-	hi = strtonum("0X" a[n])
-
-	for (i = lo; i <= hi; i++) {
-		gsub(/^; /, "", $2)
-		props[i] = "LB_" $2
-	}
-}
-
-END {
-	print "static constexpr enum uprop_lb lookup_lat1[] = {"
-	for (i = 0; i < 0x100; i++) {
-		if (i % 8 == 0)
-			printf "\t"
-		printf "%-6s%s", (props[i] ? props[i] : "LB_XX") ",", \
-			i % 8 == 7 ? "\n" : " "
-	}
-	print "};"
-	print ""
-	print "static const struct {"
-	print "\trune lo, hi;"
-	print "\tenum uprop_lb val;"
-	print "} lookup[] = {"
-
-	for (i = 0x100; i <= 0x10FFFF; i++) {
-		if (!props[i])
-			continue
-		lo = i
-		while (props[lo] == props[i + 1])
-			i++
-		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[i]
-	}
-
-	print "};"
-	print ""
-	print "_MLIB_DEFINE_BSEARCH(enum uprop_lb, lookup, LB_XX)"
-	print ""
-	print "enum uprop_lb"
-	print "uprop_get_lb(rune ch)"
-	print "{"
-	print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
-	print "}"
-}
-' data/DerivedLineBreak | sed 's/\s*$//'
+#!/usr/bin/python3
+
+import math
+
+from lib import *
+
+
+longest = 0
+
+def parse(file: str) -> list[bool]:
+	global longest
+
+	xs = ['LB_XX'] * 0x110000
+	with open(file, 'r') as f:
+		for line in f.readlines():
+			if len(line.strip()) == 0 or line[0] == '#':
+				continue
+
+			parts = line.split(';')
+			ranges = [int(x, 16) for x in parts[0].strip().split('..')]
+			prop = 'LB_' + parts[1].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/lb; 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_lb 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_lb
+uprop_get_lb(rune ch)
+{{
+	return stage2[stage1[ch / {blksize}]][ch % {blksize}];
+}}''')
+
+def main() -> None:
+	cwd_init()
+	xs = parse('data/DerivedLineBreak')
+
+	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('lib/unicode/prop/uprop_get_lb.c', 'w') as f:
+		sys.stdout = f
+		genfile(Cs, blksize)
+
+	report_size(len(xs), smallest)
+
+if __name__ == '__main__':
+	main()