From 0b12949fbadae1df31989affdae54af33f6a58ce Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Sat, 23 Mar 2024 19:05:20 +0100
Subject: Add unicode/gbrk

---
 gen/data-files | 18 +++++++++++
 gen/gbrk       | 97 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 115 insertions(+)
 create mode 100755 gen/data-files
 create mode 100755 gen/gbrk

(limited to 'gen')

diff --git a/gen/data-files b/gen/data-files
new file mode 100755
index 0000000..69ad9ba
--- /dev/null
+++ b/gen/data-files
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+cd "${0%/*}/.."
+
+mkdir -p data
+
+readonly URLS='
+https://www.unicode.org/Public/UCD/latest/ucd/auxiliary/GraphemeBreakProperty.txt
+https://www.unicode.org/Public/UCD/latest/ucd/DerivedCoreProperties.txt
+https://www.unicode.org/Public/UCD/latest/ucd/emoji/emoji-data.txt
+'
+
+for url in $URLS
+do
+	name="data/${url##*/}"
+	test -f "$name" || wget -q "$url" -O "$name" &
+done
+wait
diff --git a/gen/gbrk b/gen/gbrk
new file mode 100755
index 0000000..8216beb
--- /dev/null
+++ b/gen/gbrk
@@ -0,0 +1,97 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/.."
+exec >include/unicode/__gbrk_lookup.h
+
+cat <<C
+/* This file is autogenerated by gen/gbrk; DO NOT EDIT. */
+
+#ifndef MLIB_UNICODE___GBRK_LOOKUP_H
+#define MLIB_UNICODE___GBRK_LOOKUP_H
+
+/* clang-format off */
+
+#include "__rune.h"
+
+typedef enum {
+	GBP_OTHER = 0,
+
+	GBP_CTRL = 1 << 0, /* Control */
+	GBP_EXT  = 1 << 1, /* Extend */
+	GBP_PIC  = 1 << 2, /* Extended_Pictographic */
+	GBP_PREP = 1 << 3, /* Prepend */
+	GBP_RI   = 1 << 4, /* Regional_Indicator */
+	GBP_SM   = 1 << 5, /* SpacingMark */
+	GBP_ZWJ  = 1 << 6, /* ZWJ */
+
+	GBP_HNGL_L   = 1 <<  7, /* Hangul L */
+	GBP_HNGL_LV  = 1 <<  8, /* Hangul LV */
+	GBP_HNGL_LVT = 1 <<  9, /* Hangul LVT */
+	GBP_HNGL_T   = 1 << 10, /* Hangul T */
+	GBP_HNGL_V   = 1 << 11, /* Hangul V */
+
+	GBP_INDC_CNSNT = 1 << 12, /* Indic Consonant */
+	GBP_INDC_EXT   = 1 << 13, /* Indic Extend */
+	GBP_INDC_LNK   = 1 << 14, /* Indic Linker */
+} gbrk_prop;
+
+static const struct {
+	rune lo, hi;
+	gbrk_prop val;
+} gbrk_prop_tbl[] = {
+C
+
+gawk '
+BEGIN {
+	FS = "( *#.*| +; +)"
+	map["Control"]               = "CTRL"
+	map["Extend"]                = "EXT"
+	map["Extended_Pictographic"] = "PIC"
+	map["Prepend"]               = "PREP"
+	map["Regional_Indicator"]    = "RI"
+	map["SpacingMark"]           = "SM"
+	map["ZWJ"]                   = "ZWJ"
+
+	map["L"]   = "HNGL_L"
+	map["LV"]  = "HNGL_LV"
+	map["LVT"] = "HNGL_LVT"
+	map["T"]   = "HNGL_T"
+	map["V"]   = "HNGL_V"
+
+	map["InCB; Consonant"] = "INDC_CNSNT"
+	map["InCB; Extend"]    = "INDC_EXT"
+	map["InCB; Linker"]    = "INDC_LNK"
+}
+
+map[$2] {
+	n = split($1, a, /\.\./)
+	lo = strtonum("0X" a[1])
+	hi = strtonum("0X" a[n])
+
+	for (i = lo; i <= hi; i++) {
+		s = "GBP_" map[$2]
+		props[i] = props[i] ? props[i] " | " s : s
+	}
+}
+
+END {
+	for (i = 0; i <= 0x10FFFF; i++) {
+		if (!props[i])
+			continue
+		lo = i
+		while (props[lo] == props[i + 1])
+			i++
+		printf "\t{0x%06X, 0x%06X, %s},\n", lo, i, props[lo]
+	}
+}
+' data/GraphemeBreakProperty.txt \
+  data/DerivedCoreProperties.txt \
+  data/emoji-data.txt \
+| sort
+
+cat <<C
+};
+
+#endif /* !MLIB_UNICODE___GBRK_LOOKUP_H */
+C
-- 
cgit v1.2.3