fix UTF-8 tests + rename some minor things + prep string escape/uesca…

…pe code for re-use
facil-io · Oct 4, 2024 · 751157f · 751157f
1 parent 41ca1cc
commit 751157f
Show file tree

Hide file tree

Showing 8 changed files with 794 additions and 568 deletions.
diff --git a/fio-stl.h b/fio-stl.h
diff --git a/fio-stl.md b/fio-stl.md
@@ -5621,8 +5621,8 @@ Default reallocation callback implementation
 #### `FIO_STRING_ALLOC_COPY`
 
 ```c
-#define FIO_STRING_ALLOC_COPY fio_string_default_copy_and_reallocate
-void fio_string_default_copy_and_reallocate(fio_str_info_s *dest, size_t new_capa);
+#define FIO_STRING_ALLOC_COPY fio_string_default_allocate_copy
+void fio_string_default_allocate_copy(fio_str_info_s *dest, size_t new_capa);
 ```
 
 Default reallocation callback for memory that mustn't be freed.

diff --git a/fio-stl/000 core.h b/fio-stl/000 core.h
@@ -750,6 +750,10 @@ typedef struct fio_buf_info_s {
     .buf = fio___stack_mem___##name, .capa = (capacity)                        \
   }
 
+/** Tests to see if memory reallocation happened. */
+#define FIO_STR_INFO_TMP_IS_REALLOCATED(name)                                  \
+  (fio___stack_mem___##name != name.buf)
+
 /* *****************************************************************************
 UTF-8 Support (basic)
 ***************************************************************************** */
@@ -783,27 +787,25 @@ FIO_IFUNC size_t fio_utf8_char_len_unsafe(uint8_t c) {
 
 /** Returns the number of valid UTF-8 bytes used by first char at `str`. */
 FIO_IFUNC size_t fio_utf8_char_len(const void *str_) {
-  size_t r, tst;
+  size_t r, tst = 1;
   const uint8_t *s = (uint8_t *)str_;
   r = fio_utf8_char_len_unsafe(*s);
+  r &= 7;
   if (r < 2)
     return r;
-  tst = fio_utf8_char_len_unsafe(s[1]); /* avoid overflowing more than 1 char */
-  tst &= fio_utf8_char_len_unsafe(s[(tst >> 3) + (r > 2)]);
-  tst &= fio_utf8_char_len_unsafe(s[((tst >> 3) + (r > 3)) | (tst >> 3)]);
-  tst &= 8;
-  tst -= tst >> 3;
-  r &= tst;
+  tst += (fio_utf8_char_len_unsafe(s[tst]) >> 3) & (r > 3);
+  tst += (fio_utf8_char_len_unsafe(s[tst]) >> 3) & (r > 2);
+  tst += (fio_utf8_char_len_unsafe(s[tst]) >> 3);
+  if (r != tst)
+    r = 0;
   return r;
 }
 
 /** Writes code point to `dest` using UFT-8. Returns number of bytes written. */
 FIO_IFUNC size_t fio_utf8_write(void *dest_, uint32_t u) {
   const uint8_t len = fio_utf8_code_len(u);
   uint8_t *dest = (uint8_t *)dest_;
-  if (!len)
-    return len;
-  if (len == 1) {
+  if (len < 2) { /* writes, but doesn't report on len == 0 */
     *dest = u;
     return len;
   }

diff --git a/fio-stl/001 memalt.h b/fio-stl/001 memalt.h
@@ -309,7 +309,7 @@ SFUNC FIO___ASAN_AVOID size_t fio_strlen(const char *str) {
       goto found_nul_byte0;
     str += 8;
   }
-  str = FIO_PTR_MATH_RMASK(const char, str, 6);
+  str = FIO_PTR_MATH_RMASK(const char, str, 6); /* compiler hint */
   /* loop endlessly */
   for (;;) {
     for (size_t i = 0; i < 8; ++i) {