Add uprop_get_wdth() and ucswdth()

2 files changed, 125 insertions, 0 deletions

diff --git a/gen/prop/wdth b/gen/prop/wdth
new file mode 100755
index 0000000..e987276
--- /dev/null
+++ b/gen/prop/wdth
@@ -0,0 +1,100 @@
+#!/usr/bin/python3
+
+import math
+import subprocess
+
+from lib import *
+
+
+longest = 2 # ‘-1’
+
+def parse() -> list[int]:
+	xs = [None] * 0x110000
+	p = subprocess.Popen(
+		['./gen/prop/wdth.c'],
+		stdout=subprocess.PIPE,
+		shell=True,
+	)
+	for line in p.stdout:
+		parts = line.decode().strip().split(' ')
+		if len(parts[1]) == 1:
+			parts[1] = ' ' + parts[1]
+		xs[int(parts[0], 16)] = parts[1]
+	return xs
+
+def genfile(cs: list[tuple[int, ...]], blksize: int) -> None:
+	Cs = cs
+	cs = list(dict.fromkeys(Cs))
+
+	print('''\
+/* This file is autogenerated by gen/prop/wdth; DO NOT EDIT. */
+
+#include <stdint.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 int8_t 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'''\
+int
+uprop_get_wdth(rune ch)
+{{
+	return stage2[stage1[ch / {blksize}]][ch % {blksize}];
+}}''')
+
+def main() -> None:
+	cwd_init()
+	xs = parse()
+
+	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_wdth.c', 'w') as f:
+		sys.stdout = f
+		genfile(Cs, blksize)
+
+	report_size(len(xs), smallest)
+
+if __name__ == '__main__':
+	main()
diff --git a/gen/prop/wdth.c b/gen/prop/wdth.c
new file mode 100755
index 0000000..a9379cf
--- /dev/null
+++ b/gen/prop/wdth.c
@@ -0,0 +1,25 @@
+#if 0
+cd "${0%/*}/../.."
+trap 'rm -f /tmp/wdth' EXIT
+cc -Iinclude -std=c23 -Wno-attributes -fsanitize=address,undefined \
+	-lunistring -o /tmp/wdth gen/prop/wdth.c libmlib.a
+/tmp/wdth
+exit 0
+#endif
+
+/* Cheating slightly because I am lazy; using libunistring to figure out the
+   values here */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include <rune.h>
+#include <uniwidth.h>
+
+int
+main(void)
+{
+	for (ucs4_t ch = 0; ch <= RUNE_MAX; ch++)
+		printf("%04" PRIXRUNE " %d\n", ch, uc_width(ch, "UTF-8"));
+	return EXIT_SUCCESS;
+}