add string encoding utility functions

include/libtransistor/encoding.h

@@ -0,0 +1,32 @@
+/**
+ * @file libtransistor/encoding.h
+ * @brief String encoding functions
+ * Based on https://github.com/nothings/stb
+ */
+
+#pragma once
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include<stdint.h>
+#include<string.h>
+
+typedef uint16_t trn_char16_t;
+
+/**
+ * @brief Converts a UTF-8 string to UTF-16
+ * Writes at most n characters to out, returning NULL if there is an error.
+ */
+trn_char16_t *trn_utf8_to_utf16(trn_char16_t *out, const char *in, size_t n);
+
+/**
+ * @brief Converts a UTF-16 string to UTF-8
+ * Writes at most n characters to out, returning NULL if there is an error.
+ */
+char   *trn_utf16_to_utf8(char *out, const trn_char16_t *in, size_t n);
+
+#ifdef __cplusplus
+}
+#endif

include/libtransistor/nx.h

@@ -13,6 +13,7 @@ extern "C" {
 #include<libtransistor/types.h>
 #include<libtransistor/loader_config.h>
 #include<libtransistor/runtime_config.h>
+#include<libtransistor/encoding.h>
 #include<libtransistor/environment.h>
 #include<libtransistor/alloc_pages.h>
 #include<libtransistor/address_space.h>

lib/encoding.c

@@ -0,0 +1,93 @@
+#include<libtransistor/encoding.h>
+
+#include<stdint.h>
+#include<stdlib.h>
+#include<stddef.h>
+
+// based on stb.h
+
+trn_char16_t *trn_utf8_to_utf16(trn_char16_t *buffer, const char *ostr, size_t n)
+{
+   unsigned char *str = (unsigned char *) ostr;
+   uint32_t c;
+   size_t i=0;
+   --n;
+   while (*str) {
+      if (i >= n)
+         return NULL;
+      if (!(*str & 0x80))
+         buffer[i++] = *str++;
+      else if ((*str & 0xe0) == 0xc0) {
+         if (*str < 0xc2) return NULL;
+         c = (*str++ & 0x1f) << 6;
+         if ((*str & 0xc0) != 0x80) return NULL;
+         buffer[i++] = c + (*str++ & 0x3f);
+      } else if ((*str & 0xf0) == 0xe0) {
+         if (*str == 0xe0 && (str[1] < 0xa0 || str[1] > 0xbf)) return NULL;
+         if (*str == 0xed && str[1] > 0x9f) return NULL; // str[1] < 0x80 is checked below
+         c = (*str++ & 0x0f) << 12;
+         if ((*str & 0xc0) != 0x80) return NULL;
+         c += (*str++ & 0x3f) << 6;
+         if ((*str & 0xc0) != 0x80) return NULL;
+         buffer[i++] = c + (*str++ & 0x3f);
+      } else if ((*str & 0xf8) == 0xf0) {
+         if (*str > 0xf4) return NULL;
+         if (*str == 0xf0 && (str[1] < 0x90 || str[1] > 0xbf)) return NULL;
+         if (*str == 0xf4 && str[1] > 0x8f) return NULL; // str[1] < 0x80 is checked below
+         c = (*str++ & 0x07) << 18;
+         if ((*str & 0xc0) != 0x80) return NULL;
+         c += (*str++ & 0x3f) << 12;
+         if ((*str & 0xc0) != 0x80) return NULL;
+         c += (*str++ & 0x3f) << 6;
+         if ((*str & 0xc0) != 0x80) return NULL;
+         c += (*str++ & 0x3f);
+         // utf-8 encodings of values used in surrogate pairs are invalid
+         if ((c & 0xFFFFF800) == 0xD800) return NULL;
+         if (c >= 0x10000) {
+            c -= 0x10000;
+            if (i + 2 > n) return NULL;
+            buffer[i++] = 0xD800 | (0x3ff & (c >> 10));
+            buffer[i++] = 0xDC00 | (0x3ff & (c      ));
+         }
+      } else
+         return NULL;
+   }
+   buffer[i] = 0;
+   return buffer;
+}
+
+char *trn_utf16_to_utf8(char *buffer, const trn_char16_t *str, size_t n)
+{
+   size_t i=0;
+   --n;
+   while (*str) {
+      if (*str < 0x80) {
+         if (i+1 > n) return NULL;
+         buffer[i++] = (char) *str++;
+      } else if (*str < 0x800) {
+         if (i+2 > n) return NULL;
+         buffer[i++] = 0xc0 + (*str >> 6);
+         buffer[i++] = 0x80 + (*str & 0x3f);
+         str += 1;
+      } else if (*str >= 0xd800 && *str < 0xdc00) {
+         uint32_t c;
+         if (i+4 > n) return NULL;
+         c = ((str[0] - 0xd800) << 10) + ((str[1]) - 0xdc00) + 0x10000;
+         buffer[i++] = 0xf0 + (c >> 18);
+         buffer[i++] = 0x80 + ((c >> 12) & 0x3f);
+         buffer[i++] = 0x80 + ((c >>  6) & 0x3f);
+         buffer[i++] = 0x80 + ((c      ) & 0x3f);
+         str += 2;
+      } else if (*str >= 0xdc00 && *str < 0xe000) {
+         return NULL;
+      } else {
+         if (i+3 > n) return NULL;
+         buffer[i++] = 0xe0 + (*str >> 12);
+         buffer[i++] = 0x80 + ((*str >> 6) & 0x3f);
+         buffer[i++] = 0x80 + ((*str     ) & 0x3f);
+         str += 1;
+      }
+   }
+   buffer[i] = 0;
+   return buffer;
+}

mk/transistor_headers.mk

@@ -14,6 +14,7 @@ libtransistor_HEADER_NAMES := \
 	display/parcel.h \
 	display/rect.h \
 	display/surface.h \
+	encoding.h \
 	environment.h \
 	err.h \
 	err/modules.h \

mk/transistor_objects.mk

@@ -11,6 +11,7 @@ libtransistor_OBJECT_NAMES := \
 	display/graphic_buffer_queue.o \
 	display/parcel.o \
 	display/surface.o \
+	encoding.o \
 	environment.o \
 	err.o \
 	fs/blobfd.o \