From 790aaa14406d45dd95dea3f7d6d00da5911c6584 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Sat, 4 May 2024 04:36:28 +0200
Subject: Replace u8split() with u8cut()

---
 include/mbstring.h     |  6 ++++--
 lib/mbstring/u8cut.c   | 19 +++++++++++++++++++
 lib/mbstring/u8split.c | 16 ----------------
 test/_brk-test.h       | 28 ++++++----------------------
 test/_case-test.h      |  6 +++---
 5 files changed, 32 insertions(+), 43 deletions(-)
 create mode 100644 lib/mbstring/u8cut.c
 delete mode 100644 lib/mbstring/u8split.c

diff --git a/include/mbstring.h b/include/mbstring.h
index ba654bb..d725e0d 100644
--- a/include/mbstring.h
+++ b/include/mbstring.h
@@ -20,12 +20,13 @@
 #define u8bytec(x) (((x) & 0xC0) == 0x80)
 /* clang-format on */
 
+constexpr int U8_LEN_MAX = 4;
 constexpr rune U8_1B_MAX = 0x00007F;
 constexpr rune U8_2B_MAX = 0x0007FF;
 constexpr rune U8_3B_MAX = 0x00FFFF;
 constexpr rune U8_4B_MAX = 0x10FFFF;
 
-constexpr int U8_LEN_MAX = 4;
+constexpr rune MBEND = 0x110000;
 
 #define PRIsU8          ".*s"
 #define SV_PRI_ARGS(sv) ((int)(sv).len), ((sv).p)
@@ -43,6 +44,7 @@ int u8tor(rune *, const char8_t *);
 [[nodiscard]] size_t u8cspn(struct u8view, const rune *, size_t);
 [[nodiscard]] size_t u8len(struct u8view);
 [[nodiscard]] size_t u8spn(struct u8view, const rune *, size_t);
-struct u8view u8split(struct u8view *, rune);
+rune u8cut(struct u8view *restrict, struct u8view *restrict, const rune *,
+           size_t);
 
 #endif /* !MLIB_MBSTRING_H */
diff --git a/lib/mbstring/u8cut.c b/lib/mbstring/u8cut.c
new file mode 100644
index 0000000..3dd9663
--- /dev/null
+++ b/lib/mbstring/u8cut.c
@@ -0,0 +1,19 @@
+#include "macros.h"
+#include "mbstring.h"
+
+rune
+u8cut(struct u8view *restrict x, struct u8view *restrict y, const rune *seps,
+      size_t n)
+{
+	ASSUME(y != nullptr);
+	ASSUME(seps != nullptr);
+	size_t off = u8cspn(*y, seps, n);
+	if (x != nullptr) {
+		x->p = y->p;
+		x->len = off;
+	}
+	VSHFT(y, off);
+	rune ch = MBEND;
+	u8next(&ch, y);
+	return ch;
+}
diff --git a/lib/mbstring/u8split.c b/lib/mbstring/u8split.c
deleted file mode 100644
index c26f48b..0000000
--- a/lib/mbstring/u8split.c
+++ /dev/null
@@ -1,16 +0,0 @@
-#include "mbstring.h"
-
-struct u8view
-u8split(struct u8view *rhs, rune ch)
-{
-	struct u8view lhs = {.p = rhs->p};
-	if ((rhs->p = u8chr(*rhs, ch)) == nullptr) {
-		lhs.len = rhs->len;
-		rhs->len = 0;
-	} else {
-		lhs.len = rhs->p - lhs.p;
-		rhs->len -= lhs.len;
-		u8next(nullptr, rhs);
-	}
-	return lhs;
-}
diff --git a/test/_brk-test.h b/test/_brk-test.h
index e468f23..3a66f23 100644
--- a/test/_brk-test.h
+++ b/test/_brk-test.h
@@ -19,7 +19,6 @@
 #define CNTFUNC  CONCAT(CONCAT(u8, BRKTYPE), cnt)
 
 static bool test(struct u8view, int);
-static int hexdigits(rune);
 
 int
 main(int, char **argv)
@@ -59,25 +58,21 @@ test(struct u8view sv, int id)
 
 	typedef dynarr(char8_t) item;
 	dynarr(item) items = {};
-	struct u8view sv_cpy = sv;
-
-	do {
-		rune op, ch;
 
-		u8next(&op, &sv_cpy);
+	rune op;
+	struct u8view sv_cpy = sv;
+	while ((op = u8cut(nullptr, &sv_cpy, U"×÷", 2)) != MBEND) {
+		rune ch;
 		sscanf(sv_cpy.p, "%" SCNxRUNE, &ch);
-		int off = hexdigits(ch);
-		off = MAX(4, off);
-		VSHFT(&sv_cpy, off);
 
-		char8_t buf[U8_LEN_MAX] = {};
+		char8_t buf[U8_LEN_MAX];
 		int w = rtou8(buf, sizeof(buf), ch);
 		total += w;
 
 		if (op == U'÷')
 			DAPUSH(&items, (item){});
 		DAEXTEND(&items.buf[items.len - 1], buf, w);
-	} while (sv_cpy.len > 0);
+	}
 
 	size_t off = 0;
 	char8_t *p = bufalloc(nullptr, 1, total);
@@ -114,14 +109,3 @@ test(struct u8view sv, int id)
 
 	return true;
 }
-
-int
-hexdigits(rune ch)
-{
-	int n = 0;
-	do {
-		ch /= 16;
-		n++;
-	} while (ch != 0);
-	return n;
-}
diff --git a/test/_case-test.h b/test/_case-test.h
index 12e3a57..c594554 100644
--- a/test/_case-test.h
+++ b/test/_case-test.h
@@ -54,9 +54,9 @@ test(const char8_t *line, int id)
 {
 	struct u8view sv = {line, strlen(line)};
 	struct u8view before, after, flags;
-	before = u8split(&sv, ';');
-	after  = u8split(&sv, ';');
-	flags  = u8split(&sv, ';');
+	u8cut(&before, &sv, U";", 1);
+	u8cut(&after,  &sv, U";", 1);
+	u8cut(&flags,  &sv, U";", 1);
 
 	enum caseflags cf = u8eq(flags, U8("ẞ"))  ? CF_ẞ
 	                  : u8eq(flags, U8("AZ")) ? CF_LANG_AZ
-- 
cgit v1.2.3