From ac1b4bcbaeaee7d2ef9132dcdc254f2d08691650 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Sat, 4 May 2024 04:01:45 +0200
Subject: Go all in on string views, and fix manuals

---
 lib/mbstring/u8chk.c    | 14 ++++++--------
 lib/mbstring/u8chr.c    | 32 ++++++++++++++++----------------
 lib/mbstring/u8cmp.c    |  4 ++--
 lib/mbstring/u8cspn.c   |  8 ++++----
 lib/mbstring/u8haspfx.c |  4 ++--
 lib/mbstring/u8hassfx.c |  4 ++--
 lib/mbstring/u8len.c    |  4 ++--
 lib/mbstring/u8next.c   | 15 +++++++--------
 lib/mbstring/u8rchr.c   | 40 ++++++++++++++++++++--------------------
 lib/mbstring/u8split.c  | 16 ++++++++--------
 lib/mbstring/u8spn.c    | 10 +++++-----
 11 files changed, 74 insertions(+), 77 deletions(-)

(limited to 'lib/mbstring')

diff --git a/lib/mbstring/u8chk.c b/lib/mbstring/u8chk.c
index 2566bac..20c4f3f 100644
--- a/lib/mbstring/u8chk.c
+++ b/lib/mbstring/u8chk.c
@@ -1,17 +1,15 @@
 #include "rune.h"
 #include "mbstring.h"
 
-char8_t *
-(u8chk)(const char8_t *s, size_t n)
+const char8_t *
+u8chk(struct u8view sv)
 {
-	while (n) {
-		rune ch;
-		int m = u8tor(&ch, s);
+	int w;
+	rune ch;
 
+	while (w = u8next(&ch, &sv)) {
 		if (ch == RUNE_ERROR)
-			return (char8_t *)s;
-		n -= m;
-		s += m;
+			return sv.p - w;
 	}
 
 	return nullptr;
diff --git a/lib/mbstring/u8chr.c b/lib/mbstring/u8chr.c
index 395a328..4831695 100644
--- a/lib/mbstring/u8chr.c
+++ b/lib/mbstring/u8chr.c
@@ -31,7 +31,7 @@
        TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
        SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
-static char8_t *
+static const char8_t *
 memmem2(const char8_t *h, size_t k, const char8_t *n)
 {
 	uint16_t hw, nw;
@@ -40,12 +40,12 @@ memmem2(const char8_t *h, size_t k, const char8_t *n)
 
 	for (h += 2, k -= 2; k; k--, hw = hw << 8 | *h++) {
 		if (hw == nw)
-			return (char8_t *)h - 2;
+			return h - 2;
 	}
-	return hw == nw ? (char8_t *)h - 2 : nullptr;
+	return hw == nw ? h - 2 : nullptr;
 }
 
-static char8_t *
+static const char8_t *
 memmem3(const char8_t *h, size_t k, const char8_t *n)
 {
 	uint32_t hw, nw;
@@ -54,12 +54,12 @@ memmem3(const char8_t *h, size_t k, const char8_t *n)
 
 	for (h += 3, k -= 3; k; k--, hw = (hw | *h++) << 8) {
 		if (hw == nw)
-			return (char8_t *)h - 3;
+			return h - 3;
 	}
-	return hw == nw ? (char8_t *)h - 3 : nullptr;
+	return hw == nw ? h - 3 : nullptr;
 }
 
-static char8_t *
+static const char8_t *
 memmem4(const char8_t *h, size_t k, const char8_t *n)
 {
 	uint32_t hw, nw;
@@ -68,28 +68,28 @@ memmem4(const char8_t *h, size_t k, const char8_t *n)
 
 	for (h += 4, k -= 4; k; k--, hw = hw << 8 | *h++) {
 		if (hw == nw)
-			return (char8_t *)h - 4;
+			return h - 4;
 	}
-	return hw == nw ? (char8_t *)h - 4 : nullptr;
+	return hw == nw ? h - 4 : nullptr;
 }
 
-char8_t *
-(u8chr)(const char8_t *s, size_t n, rune ch)
+const char8_t *
+u8chr(struct u8view sv, rune ch)
 {
 	char8_t buf[U8_LEN_MAX];
 	int m = rtou8(buf, sizeof(buf), ch);
 
-	if (n < (size_t)m)
+	if (sv.len < (size_t)m)
 		return nullptr;
 	switch (m) {
 	case 1:
-		return memchr(s, ch, n);
+		return memchr(sv.p, ch, sv.len);
 	case 2:
-		return memmem2(s, n, buf);
+		return memmem2(sv.p, sv.len, buf);
 	case 3:
-		return memmem3(s, n, buf);
+		return memmem3(sv.p, sv.len, buf);
 	case 4:
-		return memmem4(s, n, buf);
+		return memmem4(sv.p, sv.len, buf);
 	}
 
 	unreachable();
diff --git a/lib/mbstring/u8cmp.c b/lib/mbstring/u8cmp.c
index 8bd2400..0059020 100644
--- a/lib/mbstring/u8cmp.c
+++ b/lib/mbstring/u8cmp.c
@@ -3,7 +3,7 @@
 #include "mbstring.h"
 
 int
-u8cmp(const char8_t *x, size_t n, const char8_t *y, size_t m)
+u8cmp(struct u8view x, struct u8view y)
 {
-	return n != m ? (n > m ? +1 : -1) : memcmp(x, y, n);
+	return x.len != y.len ? (x.len > y.len ? +1 : -1) : memcmp(x.p, y.p, x.len);
 }
diff --git a/lib/mbstring/u8cspn.c b/lib/mbstring/u8cspn.c
index 4892de4..827373f 100644
--- a/lib/mbstring/u8cspn.c
+++ b/lib/mbstring/u8cspn.c
@@ -1,13 +1,13 @@
 #include "mbstring.h"
 
 size_t
-u8cspn(const char8_t *s, size_t n, const rune *p, size_t m)
+u8cspn(struct u8view sv, const rune *p, size_t n)
 {
 	rune ch;
-	size_t k, l;
+	size_t k, w;
 
-	for (k = 0; (l = u8next(&ch, &s, &n)); k += l) {
-		for (size_t i = 0; i < m; i++) {
+	for (k = 0; w = u8next(&ch, &sv); k += w) {
+		for (size_t i = 0; i < n; i++) {
 			if (p[i] == ch)
 				goto found;
 		}
diff --git a/lib/mbstring/u8haspfx.c b/lib/mbstring/u8haspfx.c
index b6cea50..c61efbb 100644
--- a/lib/mbstring/u8haspfx.c
+++ b/lib/mbstring/u8haspfx.c
@@ -4,7 +4,7 @@
 #include "mbstring.h"
 
 bool
-u8haspfx(const char8_t *s, size_t n, const char8_t *pfx, size_t m)
+u8haspfx(struct u8view sv, struct u8view pfx)
 {
-	return n >= m && memeq(s, pfx, m);
+	return sv.len >= pfx.len && memeq(sv.p, pfx.p, pfx.len);
 }
diff --git a/lib/mbstring/u8hassfx.c b/lib/mbstring/u8hassfx.c
index e31bb4b..8ea4456 100644
--- a/lib/mbstring/u8hassfx.c
+++ b/lib/mbstring/u8hassfx.c
@@ -4,7 +4,7 @@
 #include "mbstring.h"
 
 bool
-u8hassfx(const char8_t *s, size_t n, const char8_t *sfx, size_t m)
+u8hassfx(struct u8view sv, struct u8view sfx)
 {
-	return n >= m && memeq(s + n - m, sfx, m);
+	return sv.len >= sfx.len && memeq(sv.p + sv.len - sfx.len, sfx.p, sfx.len);
 }
diff --git a/lib/mbstring/u8len.c b/lib/mbstring/u8len.c
index 217ab66..23c55c5 100644
--- a/lib/mbstring/u8len.c
+++ b/lib/mbstring/u8len.c
@@ -1,10 +1,10 @@
 #include "mbstring.h"
 
 size_t
-u8len(const char8_t *s, size_t n)
+u8len(struct u8view sv)
 {
 	size_t m = 0;
-	while (u8next(nullptr, &s, &n))
+	while (u8next(nullptr, &sv))
 		m++;
 	return m;
 }
diff --git a/lib/mbstring/u8next.c b/lib/mbstring/u8next.c
index 82d2ad7..518de49 100644
--- a/lib/mbstring/u8next.c
+++ b/lib/mbstring/u8next.c
@@ -1,16 +1,15 @@
 #include "mbstring.h"
 
 int
-u8next(rune *ch, const char8_t **s, size_t *n)
+u8next(rune *ch, struct u8view *sv)
 {
-	rune _;
-	int m = 0;
+	int n = 0;
 
-	if (*n) {
-		m = u8tor(ch ? ch : &_, *s);
-		*n -= m;
-		*s += m;
+	if (sv->len) {
+		rune _;
+		n = u8tor(ch ? ch : &_, sv->p);
+		VSHFT(sv, n);
 	}
 
-	return m;
+	return n;
 }
diff --git a/lib/mbstring/u8rchr.c b/lib/mbstring/u8rchr.c
index 09aa111..825f8fd 100644
--- a/lib/mbstring/u8rchr.c
+++ b/lib/mbstring/u8rchr.c
@@ -3,17 +3,17 @@
 
 #include "mbstring.h"
 
-static char8_t *
+static const char8_t *
 memrchr1(const char8_t *s, size_t k, const char8_t *n)
 {
 	for (const char8_t *p = s + k - 1; k-- > 0; p--) {
 		if (*p == *n)
-			return (char8_t *)p;
+			return p;
 	}
 	return nullptr;
 }
 
-static char8_t *
+static const char8_t *
 memrchr2(const char8_t *h, size_t k, const char8_t *n)
 {
 	uint16_t hw, nw;
@@ -23,13 +23,13 @@ memrchr2(const char8_t *h, size_t k, const char8_t *n)
 
 	for (H -= 2, k -= 2; k; k--, hw = hw >> 8 | (*H-- << 8)) {
 		if (hw == nw)
-			return (char8_t *)H + 1;
+			return H + 1;
 	}
 
-	return hw == nw ? (char8_t *)H + 1 : nullptr;
+	return hw == nw ? H + 1 : nullptr;
 }
 
-static char8_t *
+static const char8_t *
 memrchr3(const char8_t *h, size_t k, const char8_t *n)
 {
 	uint32_t hw, nw;
@@ -41,13 +41,13 @@ memrchr3(const char8_t *h, size_t k, const char8_t *n)
 	     k--, hw = (hw >> 8 | (*H-- << 24)) & UINT32_C(0xFFFFFF00))
 	{
 		if (hw == nw)
-			return (char8_t *)H + 1;
+			return H + 1;
 	}
 
-	return hw == nw ? (char8_t *)H + 1 : nullptr;
+	return hw == nw ? H + 1 : nullptr;
 }
 
-static char8_t *
+static const char8_t *
 memrchr4(const char8_t *h, size_t k, const char8_t *n)
 {
 	uint32_t hw, nw;
@@ -57,29 +57,29 @@ memrchr4(const char8_t *h, size_t k, const char8_t *n)
 
 	for (H -= 4, k -= 4; k; k--, hw = hw >> 8 | (*H-- << 24)) {
 		if (hw == nw)
-			return (char8_t *)H + 1;
+			return H + 1;
 	}
 
-	return hw == nw ? (char8_t *)H + 1 : nullptr;
+	return hw == nw ? H + 1 : nullptr;
 }
 
-char8_t *
-(u8rchr)(const char8_t *s, size_t n, rune ch)
+const char8_t *
+u8rchr(struct u8view sv, rune ch)
 {
 	char8_t buf[U8_LEN_MAX];
-	int m = rtou8(buf, ch, sizeof(buf));
+	int n = rtou8(buf, ch, sizeof(buf));
 
-	if (n < (size_t)m)
+	if (sv.len < (size_t)n)
 		return nullptr;
-	switch (m) {
+	switch (n) {
 	case 1:
-		return (char8_t *)memrchr1(s, n, buf);
+		return memrchr1(sv.p, sv.len, buf);
 	case 2:
-		return (char8_t *)memrchr2(s, n, buf);
+		return memrchr2(sv.p, sv.len, buf);
 	case 3:
-		return (char8_t *)memrchr3(s, n, buf);
+		return memrchr3(sv.p, sv.len, buf);
 	case 4:
-		return (char8_t *)memrchr4(s, n, buf);
+		return memrchr4(sv.p, sv.len, buf);
 	}
 
 	unreachable();
diff --git a/lib/mbstring/u8split.c b/lib/mbstring/u8split.c
index 5ee3bc0..c26f48b 100644
--- a/lib/mbstring/u8split.c
+++ b/lib/mbstring/u8split.c
@@ -1,16 +1,16 @@
 #include "mbstring.h"
 
 struct u8view
-u8split(const char8_t **p, size_t *n, rune ch)
+u8split(struct u8view *rhs, rune ch)
 {
-	struct u8view lhs = {.p = *p};
-	if ((*p = u8chr(*p, *n, ch)) == nullptr) {
-		lhs.len = *n;
-		*n = 0;
+	struct u8view lhs = {.p = rhs->p};
+	if ((rhs->p = u8chr(*rhs, ch)) == nullptr) {
+		lhs.len = rhs->len;
+		rhs->len = 0;
 	} else {
-		lhs.len = *p - lhs.p;
-		*n -= lhs.len;
-		u8next(nullptr, p, n);
+		lhs.len = rhs->p - lhs.p;
+		rhs->len -= lhs.len;
+		u8next(nullptr, rhs);
 	}
 	return lhs;
 }
diff --git a/lib/mbstring/u8spn.c b/lib/mbstring/u8spn.c
index 1cf45f2..d41fcbc 100644
--- a/lib/mbstring/u8spn.c
+++ b/lib/mbstring/u8spn.c
@@ -1,15 +1,15 @@
 #include "mbstring.h"
 
 size_t
-u8spn(const char8_t *s, size_t n, const rune *p, size_t m)
+u8spn(struct u8view sv, const rune *p, size_t n)
 {
 	rune ch;
-	size_t k = 0, l;
+	size_t k = 0, w;
 
-	while ((l = u8next(&ch, &s, &n))) {
-		for (size_t i = 0; i < m; i++) {
+	while (w = u8next(&ch, &sv)) {
+		for (size_t i = 0; i < n; i++) {
 			if (p[i] == ch) {
-				k += l;
+				k += w;
 				goto found;
 			}
 		}
-- 
cgit v1.2.3