Final utf8 api, and docs.

2 files changed, 9 insertions, 9 deletions

diff --git a/include/stc/csview.h b/include/stc/csview.h
index e65821bb..f3a0ae11 100644
--- a/include/stc/csview.h
+++ b/include/stc/csview.h
@@ -69,18 +69,18 @@ STC_INLINE bool         csview_ends_with(csview sv, csview sub)
                             { if (sub.size > sv.size) return false;

                               return !memcmp(sv.str + sv.size - sub.size, sub.str, sub.size); }

 STC_INLINE csview_iter  csview_begin(const csview* self)

-                            { return c_make(csview_iter){.cp = {self->str, utf8_codepoint_size(*self->str)}}; }

+                            { return c_make(csview_iter){.cp = {self->str, utf8_codepoint_size(self->str)}}; }

 STC_INLINE csview_iter  csview_end(const csview* self)

                             { return c_make(csview_iter){self->str + self->size}; }

 STC_INLINE void         csview_next(csview_iter* it)

-                            { it->ref += it->cp.size; it->cp.size = utf8_codepoint_size(*it->ref); }

+                            { it->ref += it->cp.size; it->cp.size = utf8_codepoint_size(it->ref); }

 

 /* utf8 */

-STC_INLINE bool csview_valid_utf8(csview sv)

-    { return utf8_valid(sv.str); }

+STC_INLINE bool utf8_valid_sv(csview sv)

+    { return utf8_size_n(sv.str, sv.size) != SIZE_MAX; }

 

-STC_INLINE size_t csview_size_utf8(csview sv)

-    { return utf8_size(sv.str); }

+STC_INLINE size_t utf8_size_sv(csview sv)

+    { return utf8_size_n(sv.str, sv.size); }

 

 STC_INLINE csview utf8_substr(const char* str, size_t pos, size_t n) {

     csview sv;

diff --git a/include/stc/utf8.h b/include/stc/utf8.h
index 93f8461a..4270c8c7 100644
--- a/include/stc/utf8.h
+++ b/include/stc/utf8.h
@@ -49,8 +49,8 @@ STC_INLINE uint32_t utf8_peek(const char *s) {
     return codepoint;

 }

 

-STC_INLINE size_t utf8_codepoint_size(char c) {

-    uint8_t u = (uint8_t)c;

+STC_INLINE size_t utf8_codepoint_size(const char* s) {

+    uint8_t u = *(const uint8_t *)s;

     size_t ret = (u & 0xF0) == 0xE0;

     ret += ret << 1;                       // 3

     ret |= u < 0x80;                       // 1

@@ -60,7 +60,7 @@ STC_INLINE size_t utf8_codepoint_size(char c) {
 }

 

 STC_INLINE const char *utf8_next(const char *s) {

-    const char* t = s + utf8_codepoint_size(*s);

+    const char* t = s + utf8_codepoint_size(s);

     

     uintptr_t p = (uintptr_t)t;

     p &= (uintptr_t) -(*s != 0);