Make the rest of the casemapping functions take custom allocators

6 files changed, 81 insertions, 62 deletions

diff --git a/include/unicode/string.h b/include/unicode/string.h
index 7da4385..e3d93ad 100644
--- a/include/unicode/string.h
+++ b/include/unicode/string.h
@@ -7,8 +7,6 @@
 #include "_charN_t.h"
 #include "_u8view.h"
 
-#define mlib_warn_trunc nodiscard("don’t forget to check for truncation")
-
 /* clang-format off */
 
 enum [[clang::flag_enum]] caseflags {
@@ -30,20 +28,19 @@ size_t u8gnext(struct u8view *, struct u8view *);
 size_t u8wnext(struct u8view *, struct u8view *);
 size_t u8wnext_human(struct u8view *, struct u8view *);
 
-[[mlib_warn_trunc]]
-size_t u8casefold(char8_t *restrict, size_t, struct u8view, enum caseflags);
-[[mlib_warn_trunc]]
-size_t u8lower(char8_t *restrict, size_t, struct u8view, enum caseflags);
-[[mlib_warn_trunc]]
-size_t u8title(char8_t *restrict, size_t, struct u8view, enum caseflags);
-[[nodiscard]]
-char8_t *u8upper(size_t *, struct u8view, enum caseflags, alloc_fn, void *);
+[[nodiscard]] char8_t *u8casefold(size_t *, struct u8view, enum caseflags,
+                                  alloc_fn, void *);
+[[nodiscard]] char8_t *u8lower(size_t *, struct u8view, enum caseflags,
+                               alloc_fn, void *);
+[[nodiscard]] char8_t *u8title(size_t *, struct u8view, enum caseflags,
+                               alloc_fn, void *);
+[[nodiscard]] char8_t *u8upper(size_t *, struct u8view, enum caseflags,
+                               alloc_fn, void *);
 
+constexpr double U8CASEFOLD_SCALE = 3;
 constexpr double U8LOWER_SCALE = 1.5;
 constexpr double U8LOWER_SCALE_LT = 3;
 constexpr double U8TITLE_SCALE = 3;
 constexpr double U8UPPER_SCALE = 3;
 
-#undef mlib_warn_trunc
-
 #endif /* !MLIB_UNICODE_STRING_H */
diff --git a/lib/unicode/string/u8casefold.c b/lib/unicode/string/u8casefold.c
index 2ab7c7c..aba08f3 100644
--- a/lib/unicode/string/u8casefold.c
+++ b/lib/unicode/string/u8casefold.c
@@ -1,24 +1,32 @@
+#include <errno.h>
+#include <stdckdint.h>
+
 #include "mbstring.h"
 #include "unicode/prop.h"
 #include "unicode/string.h"
 
-size_t
-u8casefold(char8_t *restrict dst, size_t dstn, struct u8view sv,
-           enum caseflags flags)
+char8_t *
+u8casefold(size_t *dstn, struct u8view sv, enum caseflags flags, alloc_fn alloc,
+           void *alloc_ctx)
 {
+	size_t bufsz;
+	if (ckd_mul(&bufsz, sv.len, (size_t)U8CASEFOLD_SCALE)) {
+		errno = EOVERFLOW;
+		return nullptr;
+	}
+
+	char8_t *dst = alloc(alloc_ctx, nullptr, 0, bufsz, alignof(char8_t));
+	if (dst == nullptr)
+		return nullptr;
+
 	rune ch;
 	size_t n = 0;
-
 	while (u8next(&ch, &sv)) {
 		struct rview rv = uprop_get_cf(ch, flags & CF_LANG_AZ);
-		for (size_t i = 0; i < rv.len; i++) {
-			if (n >= dstn) {
-				char8_t buf[U8_LEN_MAX];
-				n += rtou8(buf, sizeof(buf), rv.p[i]);
-			} else
-				n += rtou8(dst + n, dstn - n, rv.p[i]);
-		}
+		for (size_t i = 0; i < rv.len; i++)
+			n += rtou8(dst + n, bufsz - n, rv.p[i]);
 	}
 
-	return n;
+	*dstn = n;
+	return alloc(alloc_ctx, dst, bufsz, n, alignof(char8_t));
 }
diff --git a/lib/unicode/string/u8lower.c b/lib/unicode/string/u8lower.c
index 907077b..47c32f5 100644
--- a/lib/unicode/string/u8lower.c
+++ b/lib/unicode/string/u8lower.c
@@ -1,3 +1,6 @@
+#include <errno.h>
+#include <stdckdint.h>
+
 #include "_attrs.h"
 #include "mbstring.h"
 #include "unicode/prop.h"
@@ -12,9 +15,9 @@ uprop_ccc_0_or_230(rune ch)
 	return x == 0 || x == 230;
 }
 
-size_t
-u8lower(char8_t *restrict dst, size_t dstn, struct u8view sv,
-        enum caseflags flags)
+char8_t *
+u8lower(size_t *dstn, struct u8view sv, enum caseflags flags, alloc_fn alloc,
+        void *alloc_ctx)
 {
 	struct lcctx ctx = {
 		.az_or_tr = flags & CF_LANG_AZ,
@@ -32,6 +35,17 @@ u8lower(char8_t *restrict dst, size_t dstn, struct u8view sv,
 
 	n = before_dot_cnt = more_above_cnt = 0;
 
+	size_t bufsz;
+	/* TODO: Also use U8LOWER_SCALE */
+	if (ckd_mul(&bufsz, sv.len, (size_t)U8LOWER_SCALE_LT)) {
+		errno = EOVERFLOW;
+		return nullptr;
+	}
+
+	char8_t *dst = alloc(alloc_ctx, nullptr, 0, bufsz, alignof(char8_t));
+	if (dst == nullptr)
+		return nullptr;
+
 	while (u8next(&ch, &sv)) {
 		rune next = 0;
 		if (sv.len > 0)
@@ -76,13 +90,8 @@ u8lower(char8_t *restrict dst, size_t dstn, struct u8view sv,
 		ctx.final_sigma = final_sigma.before && final_sigma.after == 0;
 
 		struct rview rv = uprop_get_lc(ch, ctx);
-		for (size_t i = 0; i < rv.len; i++) {
-			if (n >= dstn) {
-				char8_t buf[U8_LEN_MAX];
-				n += rtou8(buf, sizeof(buf), rv.p[i]);
-			} else
-				n += rtou8(dst + n, dstn - n, rv.p[i]);
-		}
+		for (size_t i = 0; i < rv.len; i++)
+			n += rtou8(dst + n, bufsz - n, rv.p[i]);
 
 		ctx.after_I = (ch == 'I') || (ctx.after_I && !uprop_ccc_0_or_230(ch));
 		if (uprop_is_cased(ch))
@@ -91,5 +100,6 @@ u8lower(char8_t *restrict dst, size_t dstn, struct u8view sv,
 			final_sigma.before = false;
 	}
 
-	return n;
+	*dstn = n;
+	return alloc(alloc_ctx, dst, bufsz, n, alignof(char8_t));
 }
diff --git a/lib/unicode/string/u8title.c b/lib/unicode/string/u8title.c
index f4d9b7e..1adf110 100644
--- a/lib/unicode/string/u8title.c
+++ b/lib/unicode/string/u8title.c
@@ -1,3 +1,6 @@
+#include <errno.h>
+#include <stdckdint.h>
+
 #include "_attrs.h"
 #include "mbstring.h"
 #include "unicode/prop.h"
@@ -13,9 +16,9 @@ uprop_ccc_0_or_230(rune ch)
 	return x == 0 || x == 230;
 }
 
-size_t
-u8title(char8_t *restrict dst, size_t dstn, struct u8view sv,
-        enum caseflags flags)
+char8_t *
+u8title(size_t *dstn, struct u8view sv, enum caseflags flags, alloc_fn alloc,
+        void *alloc_ctx)
 {
 	struct tcctx ctx_t;
 	struct lcctx ctx_l;
@@ -39,6 +42,16 @@ u8title(char8_t *restrict dst, size_t dstn, struct u8view sv,
 
 	n = before_dot_cnt = more_above_cnt = 0;
 
+	size_t bufsz;
+	if (ckd_mul(&bufsz, sv.len, (size_t)U8TITLE_SCALE)) {
+		errno = EOVERFLOW;
+		return nullptr;
+	}
+
+	char8_t *dst = alloc(alloc_ctx, nullptr, 0, bufsz, alignof(char8_t));
+	if (dst == nullptr)
+		return nullptr;
+
 	while (u8next(&ch, &sv)) {
 		if (sv.p > word.p + word.len) {
 			u8wnext(&word, &wcpy);
@@ -102,13 +115,8 @@ u8title(char8_t *restrict dst, size_t dstn, struct u8view sv,
 			}
 		}
 
-		for (size_t i = 0; i < rv.len; i++) {
-			if (n >= dstn) {
-				char8_t buf[U8_LEN_MAX];
-				n += rtou8(buf, sizeof(buf), rv.p[i]);
-			} else
-				n += rtou8(dst + n, dstn - n, rv.p[i]);
-		}
+		for (size_t i = 0; i < rv.len; i++)
+			n += rtou8(dst + n, bufsz - n, rv.p[i]);
 
 		ctx_l.after_I =
 			(ch == 'I') || (ctx_l.after_I && !uprop_ccc_0_or_230(ch));
@@ -123,5 +131,6 @@ u8title(char8_t *restrict dst, size_t dstn, struct u8view sv,
 			ctx_t.after_soft_dotted = false;
 	}
 
-	return n;
+	*dstn = n;
+	return alloc(alloc_ctx, dst, bufsz, n, alignof(char8_t));
 }
diff --git a/test/_case-test.h b/test/_case-test.h
index 8f18b69..684c6ec 100644
--- a/test/_case-test.h
+++ b/test/_case-test.h
@@ -52,7 +52,7 @@ main(int, char **argv)
 bool
 test(const char8_t *line, int id)
 {
-	struct u8view sv = {line, strlen(line)};
+	struct u8view mapped, sv = {line, strlen(line)};
 	struct u8view before, after, flags;
 	u8cut(&before, &sv, U";", 1);
 	u8cut(&after,  &sv, U";", 1);
@@ -63,27 +63,21 @@ test(const char8_t *line, int id)
 	                  : u8eq(flags, U8("LT")) ? CF_LANG_LT
 	                  : u8eq(flags, U8("NL")) ? CF_LANG_NL
 	                                          : 0;
-	char8_t *buf = bufalloc(nullptr, 1, after.len);
-	size_t bufsz = FUNC(nullptr, 0, before, cf);
-	if (bufsz != after.len) {
-		warn("case %d: expected %s buffer size of %zu but got %zu",
-		     id, STR(CASETYPE_VERB), after.len, bufsz);
-		return false;
-	}
 
-	bufsz = FUNC(buf, bufsz, before, cf);
-	if (bufsz != after.len) {
-		warn("case %d: expected %s length of %zu but got %zu",
-		     id, STR(CASETYPE_VERB), after.len, bufsz);
+	arena a = mkarena(0);
+	mapped.p = FUNC(&mapped.len, before, cf, alloc_arena, &a);
+
+	if (mapped.p == nullptr) {
+		warn("case %d: got null %s buffer", id, STR(CASETYPE_VERB));
 		return false;
 	}
 
-	if (!memeq(buf, after.p, bufsz)) {
-		warn("case %d: expected ‘%.*s’ but got ‘%.*s’",
-		     id, SV_PRI_ARGS(after), (int)bufsz, buf);
+	if (!u8eq(mapped, after)) {
+		warn("case %d: expected ‘%.*s’ but got ‘%.*s’", id, SV_PRI_ARGS(after),
+		     SV_PRI_ARGS(mapped));
 		return false;
 	}
 
-	free(buf);
+	arena_free(&a);
 	return true;
 }
diff --git a/test/run-tests b/test/run-tests
index 2562a32..f8554c0 100755
--- a/test/run-tests
+++ b/test/run-tests
@@ -16,6 +16,7 @@ readonly FLAGS='
 	-Wall -Wextra -Wpedantic
 	-Wno-pointer-sign
 	-Wno-attributes
+	-fsanitize=address,undefined
 '
 
 (cd ..; ./make)