Updated _advance() iter methods. Some have signed offsets.

utf8_peek()/utf8_peek_off() added/changed.

7 files changed, 37 insertions, 10 deletions

diff --git a/include/stc/ccommon.h b/include/stc/ccommon.h
index 69871728..29cb0761 100644
--- a/include/stc/ccommon.h
+++ b/include/stc/ccommon.h
@@ -30,6 +30,14 @@
 #include <string.h>
 #include <assert.h>
 
+#if SIZE_MAX == UINT32_MAX
+   typedef int32_t isize_t;
+#  define ISIZE_MAX INT32_MAX
+#elif SIZE_MAX == UINT64_MAX
+   typedef int64_t isize_t;
+#  define ISIZE_MAX INT64_MAX
+#endif
+
 #if defined(_MSC_VER)
 #  pragma warning(disable: 4116 4996) // unnamed type definition in parentheses
 #  define STC_FORCE_INLINE static __forceinline
diff --git a/include/stc/cdeq.h b/include/stc/cdeq.h
index b4f02081..49590477 100644
--- a/include/stc/cdeq.h
+++ b/include/stc/cdeq.h
@@ -154,7 +154,7 @@ STC_INLINE _cx_iter _cx_memb(_end)(const _cx_self* self)
 STC_INLINE void _cx_memb(_next)(_cx_iter* it)
     { if (++it->ref == it->end) it->ref = NULL; }
 
-STC_INLINE _cx_iter _cx_memb(_advance)(_cx_iter it, size_t n)
+STC_INLINE _cx_iter _cx_memb(_advance)(_cx_iter it, isize_t n)
     { if ((it.ref += n) >= it.end) it.ref = NULL; return it; }
 
 #if !defined _i_no_emplace
diff --git a/include/stc/cstack.h b/include/stc/cstack.h
index a0c6e042..c0611465 100644
--- a/include/stc/cstack.h
+++ b/include/stc/cstack.h
@@ -180,7 +180,7 @@ STC_INLINE _cx_iter _cx_memb(_end)(const _cx_self* self)
 STC_INLINE void _cx_memb(_next)(_cx_iter* it)
     { if (++it->ref == it->end) it->ref = NULL; }
 
-STC_INLINE _cx_iter _cx_memb(_advance)(_cx_iter it, size_t n)
+STC_INLINE _cx_iter _cx_memb(_advance)(_cx_iter it, isize_t n)
     { if ((it.ref += n) >= it.end) it.ref = NULL ; return it; }
 
 #include "template.h"
diff --git a/include/stc/cstr.h b/include/stc/cstr.h
index 1e7a351f..54d55d4c 100644
--- a/include/stc/cstr.h
+++ b/include/stc/cstr.h
@@ -238,6 +238,13 @@ STC_INLINE void cstr_next(cstr_iter* it) {
     it->u8.chr.size = utf8_chr_size(it->ref);
     if (it->ref == it->u8.end) it->ref = NULL;
 }
+STC_INLINE cstr_iter cstr_advance(cstr_iter it, isize_t pos) {
+    int inc = -1;
+    if (pos > 0) pos = -pos, inc = 1;
+    while (pos && it.ref != it.u8.end) pos += (*(it.ref += inc) & 0xC0) != 0x80;
+    it.u8.chr.size = utf8_chr_size(it.ref);
+    return it;
+}
 
 
 STC_INLINE void cstr_clear(cstr* self)
diff --git a/include/stc/csview.h b/include/stc/csview.h
index 54a4c4cc..2e90c69b 100644
--- a/include/stc/csview.h
+++ b/include/stc/csview.h
@@ -78,15 +78,21 @@ STC_INLINE csview_iter csview_begin(const csview* self) {
     return c_make(csview_iter){.u8 = {{self->str, utf8_chr_size(self->str)},
                                       self->str + self->size}};
 }
-
-STC_INLINE csview_iter csview_end(const csview* self) 
-    { return c_make(csview_iter){.u8 = {{NULL}, self->str + self->size}}; }
-
+STC_INLINE csview_iter csview_end(const csview* self) {
+    return c_make(csview_iter){.u8 = {{NULL}, self->str + self->size}};
+}
 STC_INLINE void csview_next(csview_iter* it) {
     it->ref += it->u8.chr.size;
     it->u8.chr.size = utf8_chr_size(it->ref);
     if (it->ref == it->u8.end) it->ref = NULL;
 }
+STC_INLINE csview_iter csview_advance(csview_iter it, isize_t pos) {
+    int inc = -1;
+    if (pos > 0) pos = -pos, inc = 1;
+    while (pos && it.ref != it.u8.end) pos += (*(it.ref += inc) & 0xC0) != 0x80;
+    it.u8.chr.size = utf8_chr_size(it.ref);
+    return it;
+}
 
 
 /* utf8 */
diff --git a/include/stc/cvec.h b/include/stc/cvec.h
index ff26de1e..effcc1d1 100644
--- a/include/stc/cvec.h
+++ b/include/stc/cvec.h
@@ -207,7 +207,7 @@ STC_INLINE _cx_iter _cx_memb(_end)(const _cx_self* self)
 STC_INLINE void _cx_memb(_next)(_cx_iter* it) 
     { if (++it->ref == it->end) it->ref = NULL; }
 
-STC_INLINE _cx_iter _cx_memb(_advance)(_cx_iter it, size_t n)
+STC_INLINE _cx_iter _cx_memb(_advance)(_cx_iter it, isize_t n)
     { if ((it.ref += n) >= it.end) it.ref = NULL; return it; }
 
 STC_INLINE size_t _cx_memb(_index)(const _cx_self* cx, _cx_iter it) 
diff --git a/include/stc/utf8.h b/include/stc/utf8.h
index c20b80cb..7a4ba921 100644
--- a/include/stc/utf8.h
+++ b/include/stc/utf8.h
@@ -17,13 +17,13 @@ extern uint32_t utf8_toupper(uint32_t c);
 extern bool     utf8_valid_n(const char* s, size_t nbytes);
 extern int      utf8_icmp_sv(csview s1, csview s2);
 extern unsigned utf8_encode(char *out, uint32_t c);
-extern uint32_t utf8_peek(const char *s, int u8pos);
+extern uint32_t utf8_peek_off(const char *s, int offset);
+
+/* following functions uses src/utf8code.c */
 
 STC_INLINE bool utf8_isupper(uint32_t c) { return utf8_tolower(c) != c; }
 STC_INLINE bool utf8_islower(uint32_t c) { return utf8_toupper(c) != c; }
 
-/* following functions uses src/utf8code.c */
-
 /* decode next utf8 codepoint. https://bjoern.hoehrmann.de/utf-8/decoder/dfa */
 typedef struct { uint32_t state, codep; } utf8_decode_t;
 
@@ -35,6 +35,12 @@ STC_INLINE uint32_t utf8_decode(utf8_decode_t* d, const uint32_t byte) {
     return d->state = utf8_dtab[256 + d->state + type];
 }
 
+STC_INLINE uint32_t utf8_peek(const char* s) {
+    utf8_decode_t d = {.state=0};
+    do { utf8_decode(&d, (uint8_t)*s++); } while (d.state);
+    return d.codep;
+}
+
 /* case-insensitive utf8 string comparison */
 STC_INLINE int utf8_icmp(const char* s1, const char* s2) {
     return utf8_icmp_sv(c_sv(s1, ~(size_t)0), c_sv(s2, ~(size_t)0));