From 03f69018340f0fae95556321a3d89dcb49910ebe Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Fri, 5 Apr 2024 15:35:58 +0200
Subject: Add more predicates to unicode/prop.h

---
 lib/unicode/prop/uprop_is_idst.c   |  7 +++++++
 lib/unicode/prop/uprop_is_idsu.c   |  7 +++++++
 lib/unicode/prop/uprop_is_join_c.c |  7 +++++++
 lib/unicode/prop/uprop_is_nchar.c  | 10 ++++++++++
 lib/unicode/prop/uprop_is_pat_ws.c | 30 ++++++++++++++++++++++++++++++
 lib/unicode/prop/uprop_is_ri.c     |  8 ++++++++
 6 files changed, 69 insertions(+)
 create mode 100644 lib/unicode/prop/uprop_is_idst.c
 create mode 100644 lib/unicode/prop/uprop_is_idsu.c
 create mode 100644 lib/unicode/prop/uprop_is_join_c.c
 create mode 100644 lib/unicode/prop/uprop_is_nchar.c
 create mode 100644 lib/unicode/prop/uprop_is_pat_ws.c
 create mode 100644 lib/unicode/prop/uprop_is_ri.c

(limited to 'lib/unicode/prop')

diff --git a/lib/unicode/prop/uprop_is_idst.c b/lib/unicode/prop/uprop_is_idst.c
new file mode 100644
index 0000000..b303838
--- /dev/null
+++ b/lib/unicode/prop/uprop_is_idst.c
@@ -0,0 +1,7 @@
+#include "unicode/prop.h"
+
+bool
+uprop_is_idst(rune ch)
+{
+	return ch == 0x2FF2 || ch == 0x2FF3;
+}
diff --git a/lib/unicode/prop/uprop_is_idsu.c b/lib/unicode/prop/uprop_is_idsu.c
new file mode 100644
index 0000000..66bc3c6
--- /dev/null
+++ b/lib/unicode/prop/uprop_is_idsu.c
@@ -0,0 +1,7 @@
+#include "unicode/prop.h"
+
+bool
+uprop_is_idsu(rune ch)
+{
+	return ch == 0x2FFE || ch == 0x2FFF;
+}
diff --git a/lib/unicode/prop/uprop_is_join_c.c b/lib/unicode/prop/uprop_is_join_c.c
new file mode 100644
index 0000000..91dd816
--- /dev/null
+++ b/lib/unicode/prop/uprop_is_join_c.c
@@ -0,0 +1,7 @@
+#include "unicode/prop.h"
+
+bool
+uprop_is_join_c(rune ch)
+{
+	return ch == 0x200C || ch == 0x200D;
+}
diff --git a/lib/unicode/prop/uprop_is_nchar.c b/lib/unicode/prop/uprop_is_nchar.c
new file mode 100644
index 0000000..7022d2e
--- /dev/null
+++ b/lib/unicode/prop/uprop_is_nchar.c
@@ -0,0 +1,10 @@
+#include "rune.h"
+#include "unicode/prop.h"
+
+bool
+rprop_is_nchar(rune ch)
+{
+	return (ch >= 0xFDD0 && ch <= 0xFDEF)
+	    || ((ch & 0xFFFE) == 0xFFFE
+	        && (ch & RUNE_C(0xFF0000)) <= RUNE_C(0x0F0000));
+}
diff --git a/lib/unicode/prop/uprop_is_pat_ws.c b/lib/unicode/prop/uprop_is_pat_ws.c
new file mode 100644
index 0000000..fe2f462
--- /dev/null
+++ b/lib/unicode/prop/uprop_is_pat_ws.c
@@ -0,0 +1,30 @@
+/* This file is autogenerated by gen/prop/bool-props; DO NOT EDIT. */
+
+#include "__bsearch.h"
+#include "bitset.h"
+#include "rune.h"
+#include "unicode/prop.h"
+
+/* clang-format off */
+
+static constexpr bitset(bs, LATIN1_MAX) = {
+	0x00, 0x3E, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+};
+
+static const struct {
+	rune lo, hi;
+} lookup[] = {
+	{RUNE_C(0x00200E), RUNE_C(0x00200F)},
+	{RUNE_C(0x002028), RUNE_C(0x002029)},
+};
+
+__MLIB_DEFINE_BSEARCH_CONTAINS(lookup)
+
+bool
+uprop_is_pat_ws(rune ch)
+{
+	return ch <= LATIN1_MAX ? TESTBIT(bs, ch) : mlib_lookup_contains(ch);
+}
diff --git a/lib/unicode/prop/uprop_is_ri.c b/lib/unicode/prop/uprop_is_ri.c
new file mode 100644
index 0000000..4ce6a75
--- /dev/null
+++ b/lib/unicode/prop/uprop_is_ri.c
@@ -0,0 +1,8 @@
+#include "rune.h"
+#include "unicode/prop.h"
+
+bool
+uprop_is_ri(rune ch)
+{
+	return ch >= RUNE_C(0x1F1E6) && ch <= RUNE_C(0x1F1FF);
+}
-- 
cgit v1.2.3