| 153 | } |
| 154 | |
| 155 | UArray *UArray_asUTF8(const UArray *self) { |
| 156 | UArray *out = UArray_new(); |
| 157 | UArray_setItemType_(out, CTYPE_uint8_t); |
| 158 | UArray_setEncoding_(out, CENCODING_UTF8); |
| 159 | UArray_setSize_(out, self->size * 4); |
| 160 | |
| 161 | { |
| 162 | // ConversionFlags options = lenientConversion; |
| 163 | void *sourceStart = self->data; |
| 164 | // void *sourceEnd = self->data + self->size * self->itemSize; |
| 165 | UTF8 *targetStart = out->data; |
| 166 | // UTF8 *targetEnd = out->data + out->size * out->itemSize; |
| 167 | size_t outSize; |
| 168 | |
| 169 | switch (self->encoding) { |
| 170 | case CENCODING_ASCII: |
| 171 | UArray_copy_(out, self); |
| 172 | break; |
| 173 | case CENCODING_UTF8: |
| 174 | UArray_copy_(out, self); |
| 175 | break; |
| 176 | /* |
| 177 | case CENCODING_UTF16: |
| 178 | r = ConvertUTF16toUTF8((const UTF16 **)&sourceStart, (const UTF16 |
| 179 | *)sourceEnd, &targetStart, targetEnd, options); break; case CENCODING_UTF32: |
| 180 | r = ConvertUTF32toUTF8((const UTF32 **)&sourceStart, (const UTF32 |
| 181 | *)sourceEnd, &targetStart, targetEnd, options); break; |
| 182 | */ |
| 183 | case CENCODING_UCS2: |
| 184 | // should the size be the num of chars or num of bytes?????????? |
| 185 | outSize = ucs2encode(targetStart, sourceStart, self->size, |
| 186 | NULL); // ucs2 to utf8 |
| 187 | UArray_setSize_(out, outSize - 1); |
| 188 | break; |
| 189 | case CENCODING_UCS4: |
| 190 | // should the size be the num of chars or num of bytes?????????? |
| 191 | outSize = ucs4encode(targetStart, sourceStart, self->size, |
| 192 | NULL); // ucs4 to utf8 |
| 193 | UArray_setSize_(out, outSize - 1); |
| 194 | break; |
| 195 | case CENCODING_NUMBER: { |
| 196 | UArray *nas = UArray_asNumberArrayString(self); |
| 197 | UArray_free(out); |
| 198 | out = UArray_asUTF8(nas); |
| 199 | UArray_free(nas); |
| 200 | break; |
| 201 | } |
| 202 | default: |
| 203 | printf("UArray_asUTF8 - unknown source encoding\n"); |
| 204 | } |
| 205 | } |
| 206 | |
| 207 | return out; |
| 208 | } |
| 209 | |
| 210 | int UArray_isUTF8Compatible(const UArray *self) { |
| 211 | return (self->encoding == CENCODING_ASCII || |
no test coverage detected