libsvn_subr/utf8proc/utf8proc.c

116   utf8proc_uint32_t uc;
122   uc = *str++;
123   if (uc < 0x80) {
124     *dst = uc;
128   if ((uc - 0xc2) > (0xf4-0xc2)) return UTF8PROC_ERROR_INVALIDUTF8;
129   if (uc < 0xe0) {         /* 2-byte sequence */
132      *dst = ((uc & 0x1f)<<6) | (*str & 0x3f);
135   if (uc < 0xf0) {        /* 3-byte sequence */
139      if (uc == 0xed && *str > 0x9f)
141      uc = ((uc & 0xf)<<12) | ((*str & 0x3f)<<6) | (str[1] & 0x3f);
142      if (uc < 0x800)
144      *dst = uc;
152   if (uc == 0xf0) {
154   } else if (uc == 0xf4) {
157   *dst = ((uc & 7)<<18) | ((*str & 0x3f)<<12) | ((str[1] & 0x3f)<<6) | (str[2] & 0x3f);
161 UTF8PROC_DLLEXPORT utf8proc_bool utf8proc_codepoint_valid(utf8proc_int32_t uc) {
162     return (((utf8proc_uint32_t)uc)-0xd800 > 0x07ff) && ((utf8proc_uint32_t)uc < 0x110000);
165 UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_encode_char(utf8proc_int32_t uc, utf8proc_uint8_t *dst) {
166   if (uc < 0x00) {
168   } else if (uc < 0x80) {
169     dst[0] = (utf8proc_uint8_t) uc;
171   } else if (uc < 0x800) {
172     dst[0] = (utf8proc_uint8_t)(0xC0 + (uc >> 6));
173     dst[1] = (utf8proc_uint8_t)(0x80 + (uc & 0x3F));
177   } else if (uc < 0x10000) {
178     dst[0] = (utf8proc_uint8_t)(0xE0 + (uc >> 12));
179     dst[1] = (utf8proc_uint8_t)(0x80 + ((uc >> 6) & 0x3F));
180     dst[2] = (utf8proc_uint8_t)(0x80 + (uc & 0x3F));
182   } else if (uc < 0x110000) {
183     dst[0] = (utf8proc_uint8_t)(0xF0 + (uc >> 18));
184     dst[1] = (utf8proc_uint8_t)(0x80 + ((uc >> 12) & 0x3F));
185     dst[2] = (utf8proc_uint8_t)(0x80 + ((uc >> 6) & 0x3F));
186     dst[3] = (utf8proc_uint8_t)(0x80 + (uc & 0x3F));
191 /* internal "unsafe" version that does not check whether uc is in range */
192 static utf8proc_ssize_t unsafe_encode_char(utf8proc_int32_t uc, utf8proc_uint8_t *dst) {
193    if (uc < 0x00) {
195    } else if (uc < 0x80) {
196       dst[0] = (utf8proc_uint8_t)uc;
198    } else if (uc < 0x800) {
199       dst[0] = (utf8proc_uint8_t)(0xC0 + (uc >> 6));
200       dst[1] = (utf8proc_uint8_t)(0x80 + (uc & 0x3F));
202    } else if (uc == 0xFFFF) {
205    } else if (uc == 0xFFFE) {
208    } else if (uc < 0x10000) {
209       dst[0] = (utf8proc_uint8_t)(0xE0 + (uc >> 12));
210       dst[1] = (utf8proc_uint8_t)(0x80 + ((uc >> 6) & 0x3F));
211       dst[2] = (utf8proc_uint8_t)(0x80 + (uc & 0x3F));
213    } else if (uc < 0x110000) {
214       dst[0] = (utf8proc_uint8_t)(0xF0 + (uc >> 18));
215       dst[1] = (utf8proc_uint8_t)(0x80 + ((uc >> 12) & 0x3F));
216       dst[2] = (utf8proc_uint8_t)(0x80 + ((uc >> 6) & 0x3F));
217       dst[3] = (utf8proc_uint8_t)(0x80 + (uc & 0x3F));
222 /* internal "unsafe" version that does not check whether uc is in range */
223 static const utf8proc_property_t *unsafe_get_property(utf8proc_int32_t uc) {
224   /* ASSERT: uc >= 0 && uc < 0x110000 */
227       utf8proc_stage1table[uc >> 8] + (uc & 0xFF)
232 UTF8PROC_DLLEXPORT const utf8proc_property_t *utf8proc_get_property(utf8proc_int32_t uc) {
233   return uc < 0 || uc >= 0x110000 ? utf8proc_properties : unsafe_get_property(uc);
399 UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_decompose_char(utf8proc_int32_t uc, utf8proc_int32_t *dst, utf8proc_ssize_t bufsize, utf8proc_option_t options, int *last_boundclass) {
403   if (uc < 0 || uc >= 0x110000) return UTF8PROC_ERROR_NOTASSIGNED;
404   property = unsafe_get_property(uc);
406   hangul_sindex = uc - UTF8PROC_HANGUL_SBASE;
430     if (uc == 0x2018 || uc == 0x2019 || uc == 0x02BC || uc == 0x02C8)
432     if (category == UTF8PROC_CATEGORY_PD || uc == 0x2212)
434     if (uc == 0x2044 || uc == 0x2215) utf8proc_decompose_lump(0x002F);
435     if (uc == 0x2236) utf8proc_decompose_lump(0x003A);
436     if (uc == 0x2039 || uc == 0x2329 || uc == 0x3008)
438     if (uc == 0x203A || uc == 0x232A || uc == 0x3009)
440     if (uc == 0x2216) utf8proc_decompose_lump(0x005C);
441     if (uc == 0x02C4 || uc == 0x02C6 || uc == 0x2038 || uc == 0x2303)
443     if (category == UTF8PROC_CATEGORY_PC || uc == 0x02CD)
445     if (uc == 0x02CB) utf8proc_decompose_lump(0x0060);
446     if (uc == 0x2223) utf8proc_decompose_lump(0x007C);
447     if (uc == 0x223C) utf8proc_decompose_lump(0x007E);
476       if (bufsize >= 2) dst[1] = uc;
480   if (bufsize >= 1) *dst = uc;
504     utf8proc_int32_t uc;
510         rpos += utf8proc_iterate(str + rpos, -1, &uc);
512            as 'uc' is < 0 in case of error */
513         if (uc < 0) return UTF8PROC_ERROR_INVALIDUTF8;
515         if (uc == 0) break;
518         rpos += utf8proc_iterate(str + rpos, strlen - rpos, &uc);
519         if (uc < 0) return UTF8PROC_ERROR_INVALIDUTF8;
522         uc = custom_func(uc, custom_data);   /* user-specified custom mapping */
525         uc, buffer + wpos, (bufsize > wpos) ? (bufsize - wpos) : 0, options,
563     utf8proc_int32_t uc;
565       uc = buffer[rpos];
566       if (uc == 0x000D && rpos < length-1 && buffer[rpos+1] == 0x000A) rpos++;
567       if (uc == 0x000A || uc == 0x000D || uc == 0x0085 ||
568           ((options & UTF8PROC_STRIPCC) && (uc == 0x000B || uc == 0x000C))) {
583           (uc < 0x0020 || (uc >= 0x007F && uc < 0x00A0))) {
584         if (uc == 0x0009) buffer[wpos++] = 0x0020;
586         buffer[wpos++] = uc;
677     utf8proc_int32_t uc;
680             uc = buffer[rpos];
681             wpos += unsafe_encode_char(uc, ((utf8proc_uint8_t *)buffer) + wpos);
685             uc = buffer[rpos];
686             wpos += utf8proc_encode_char(uc, ((utf8proc_uint8_t *)buffer) + wpos);