jbe@0: /*
jbe@7:  *  Copyright (c) 2006-2007 Jan Behrens, FlexiGuided GmbH, Berlin
jbe@0:  *
jbe@7:  *  Permission is hereby granted, free of charge, to any person obtaining a
jbe@7:  *  copy of this software and associated documentation files (the "Software"),
jbe@7:  *  to deal in the Software without restriction, including without limitation
jbe@7:  *  the rights to use, copy, modify, merge, publish, distribute, sublicense,
jbe@7:  *  and/or sell copies of the Software, and to permit persons to whom the
jbe@7:  *  Software is furnished to do so, subject to the following conditions:
jbe@0:  *
jbe@7:  *  The above copyright notice and this permission notice shall be included in
jbe@7:  *  all copies or substantial portions of the Software.
jbe@0:  *
jbe@7:  *  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
jbe@7:  *  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
jbe@7:  *  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
jbe@7:  *  AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
jbe@7:  *  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
jbe@7:  *  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
jbe@7:  *  DEALINGS IN THE SOFTWARE.
jbe@7:  */
jbe@7: 
jbe@7: /*
jbe@0:  *  This library contains derived data from a modified version of the
jbe@0:  *  Unicode data files.
jbe@0:  *
jbe@0:  *  The original data files are available at
jbe@0:  *  http://www.unicode.org/Public/UNIDATA/
jbe@0:  *
jbe@0:  *  Please notice the copyright statement in the file "utf8proc_data.c".
jbe@0:  */
jbe@0: 
jbe@0: 
jbe@0: /*
jbe@0:  *  File name:    utf8proc.c
jbe@7:  *  Version:      1.1.1
jbe@7:  *  Last changed: 2007-07-22
jbe@0:  *
jbe@0:  *  Description:
jbe@0:  *  Implementation of libutf8proc.
jbe@0:  */
jbe@0: 
jbe@0: 
jbe@0: #include "utf8proc.h"
jbe@0: #include "utf8proc_data.c"
jbe@0: 
jbe@0: 
jbe@0: const int8_t utf8proc_utf8class[256] = {
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
jbe@0:   0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
jbe@0:   0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
jbe@0:   0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
jbe@0:   0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
jbe@0:   2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
jbe@0:   2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
jbe@0:   3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
jbe@0:   4, 4, 4, 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0 };
jbe@0: 
jbe@0: #define UTF8PROC_HANGUL_SBASE 0xAC00
jbe@0: #define UTF8PROC_HANGUL_LBASE 0x1100
jbe@0: #define UTF8PROC_HANGUL_VBASE 0x1161
jbe@0: #define UTF8PROC_HANGUL_TBASE 0x11A7
jbe@0: #define UTF8PROC_HANGUL_LCOUNT 19
jbe@0: #define UTF8PROC_HANGUL_VCOUNT 21
jbe@0: #define UTF8PROC_HANGUL_TCOUNT 28
jbe@0: #define UTF8PROC_HANGUL_NCOUNT 588
jbe@0: #define UTF8PROC_HANGUL_SCOUNT 11172
jbe@2: // END is exclusive
jbe@2: #define UTF8PROC_HANGUL_L_START  0x1100
jbe@2: #define UTF8PROC_HANGUL_L_END    0x115A
jbe@2: #define UTF8PROC_HANGUL_L_FILLER 0x115F
jbe@2: #define UTF8PROC_HANGUL_V_START  0x1160
jbe@2: #define UTF8PROC_HANGUL_V_END    0x11A3
jbe@2: #define UTF8PROC_HANGUL_T_START  0x11A8
jbe@2: #define UTF8PROC_HANGUL_T_END    0x11FA
jbe@2: #define UTF8PROC_HANGUL_S_START  0xAC00
jbe@2: #define UTF8PROC_HANGUL_S_END    0xD7A4
jbe@2: 
jbe@2: 
jbe@2: #define UTF8PROC_BOUNDCLASS_START    0
jbe@2: #define UTF8PROC_BOUNDCLASS_OTHER    1
jbe@2: #define UTF8PROC_BOUNDCLASS_CR       2
jbe@2: #define UTF8PROC_BOUNDCLASS_LF       3
jbe@2: #define UTF8PROC_BOUNDCLASS_CONTROL  4
jbe@2: #define UTF8PROC_BOUNDCLASS_EXTEND   5
jbe@2: #define UTF8PROC_BOUNDCLASS_L        6
jbe@2: #define UTF8PROC_BOUNDCLASS_V        7
jbe@2: #define UTF8PROC_BOUNDCLASS_T        8
jbe@2: #define UTF8PROC_BOUNDCLASS_LV       9
jbe@2: #define UTF8PROC_BOUNDCLASS_LVT     10
jbe@0: 
jbe@0: 
jbe@0: const char *utf8proc_errmsg(ssize_t errcode) {
jbe@0:   switch (errcode) {
jbe@0:     case UTF8PROC_ERROR_NOMEM:
jbe@0:     return "Memory for processing UTF-8 data could not be allocated.";
jbe@0:     case UTF8PROC_ERROR_OVERFLOW:
jbe@0:     return "UTF-8 string is too long to be processed.";
jbe@0:     case UTF8PROC_ERROR_INVALIDUTF8:
jbe@0:     return "Invalid UTF-8 string";
jbe@0:     case UTF8PROC_ERROR_NOTASSIGNED:
jbe@0:     return "Unassigned Unicode code point found in UTF-8 string.";
jbe@3:     case UTF8PROC_ERROR_INVALIDOPTS:
jbe@3:     return "Invalid options for UTF-8 processing chosen.";
jbe@0:     default:
jbe@0:     return "An unknown error occured while processing UTF-8 data.";
jbe@0:   }
jbe@0: }
jbe@0: 
jbe@7: ssize_t utf8proc_iterate(
jbe@7:   const uint8_t *str, ssize_t strlen, int32_t *dst
jbe@7: ) {
jbe@0:   int length;
jbe@0:   int i;
jbe@0:   int32_t uc = -1;
jbe@0:   *dst = -1;
jbe@0:   if (!strlen) return 0;
jbe@0:   length = utf8proc_utf8class[str[0]];
jbe@0:   if (!length) return UTF8PROC_ERROR_INVALIDUTF8;
jbe@0:   if (strlen >= 0 && length > strlen) return UTF8PROC_ERROR_INVALIDUTF8;
jbe@0:   for (i=1; i<length; i++) {
jbe@0:     if ((str[i] & 0xC0) != 0x80) return UTF8PROC_ERROR_INVALIDUTF8;
jbe@0:   }
jbe@0:   switch (length) {
jbe@0:     case 1:
jbe@0:     uc = str[0];
jbe@0:     break;
jbe@0:     case 2:
jbe@0:     uc = ((str[0] & 0x1F) <<  6) + (str[1] & 0x3F);
jbe@0:     if (uc < 0x80) uc = -1;
jbe@0:     break;
jbe@0:     case 3:
jbe@0:     uc = ((str[0] & 0x0F) << 12) + ((str[1] & 0x3F) <<  6)
jbe@0:       + (str[2] & 0x3F);
jbe@0:     if (uc < 0x800 || (uc >= 0xD800 && uc < 0xE000) ||
jbe@0:       (uc >= 0xFDD0 && uc < 0xFDF0)) uc = -1;
jbe@0:     break;
jbe@0:     case 4:
jbe@0:     uc = ((str[0] & 0x07) << 18) + ((str[1] & 0x3F) << 12)
jbe@0:       + ((str[2] & 0x3F) <<  6) + (str[3] & 0x3F);
jbe@0:     if (uc < 0x10000 || uc >= 0x110000) uc = -1;
jbe@0:     break;
jbe@0:   }
jbe@7:   if (uc < 0 || ((uc & 0xFFFF) >= 0xFFFE))
jbe@7:     return UTF8PROC_ERROR_INVALIDUTF8;
jbe@0:   *dst = uc;
jbe@0:   return length;
jbe@0: }
jbe@0: 
jbe@7: bool utf8proc_codepoint_valid(int32_t uc) {
jbe@7:   if (uc < 0 || uc >= 0x110000 ||
jbe@7:     ((uc & 0xFFFF) >= 0xFFFE) || (uc >= 0xD800 && uc < 0xE000) ||
jbe@7:     (uc >= 0xFDD0 && uc < 0xFDF0)) return false;
jbe@7:   else return true;
jbe@7: }
jbe@7: 
jbe@0: ssize_t utf8proc_encode_char(int32_t uc, uint8_t *dst) {
jbe@0:   if (uc < 0x00) {
jbe@0:     return 0;
jbe@0:   } else if (uc < 0x80) {
jbe@0:     dst[0] = uc;
jbe@0:     return 1;
jbe@0:   } else if (uc < 0x800) {
jbe@0:     dst[0] = 0xC0 + (uc >> 6);
jbe@0:     dst[1] = 0x80 + (uc & 0x3F);
jbe@0:     return 2;
jbe@2:   } else if (uc == 0xFFFF) {
jbe@2:     dst[0] = 0xFF;
jbe@2:     return 1;
jbe@2:   } else if (uc == 0xFFFE) {
jbe@2:     dst[0] = 0xFE;
jbe@2:     return 1;
jbe@0:   } else if (uc < 0x10000) {
jbe@0:     dst[0] = 0xE0 + (uc >> 12);
jbe@0:     dst[1] = 0x80 + ((uc >> 6) & 0x3F);
jbe@0:     dst[2] = 0x80 + (uc & 0x3F);
jbe@0:     return 3;
jbe@0:   } else if (uc < 0x110000) {
jbe@0:     dst[0] = 0xF0 + (uc >> 18);
jbe@0:     dst[1] = 0x80 + ((uc >> 12) & 0x3F);
jbe@0:     dst[2] = 0x80 + ((uc >> 6) & 0x3F);
jbe@0:     dst[3] = 0x80 + (uc & 0x3F);
jbe@0:     return 4;
jbe@0:   } else return 0;
jbe@0: }
jbe@0: 
jbe@0: const utf8proc_property_t *utf8proc_get_property(int32_t uc) {
jbe@0:   // ASSERT: uc >= 0 && uc < 0x110000
jbe@0:   return utf8proc_properties + (
jbe@0:     utf8proc_stage2table[
jbe@0:       utf8proc_stage1table[uc >> 8] + (uc & 0xFF)
jbe@0:     ]
jbe@0:   );
jbe@0: }
jbe@0: 
jbe@3: #define utf8proc_decompose_lump(replacement_uc) \
jbe@3:   return utf8proc_decompose_char((replacement_uc), dst, bufsize, \
jbe@3:   options & ~UTF8PROC_LUMP, last_boundclass)
jbe@3: 
jbe@0: ssize_t utf8proc_decompose_char(int32_t uc, int32_t *dst, ssize_t bufsize,
jbe@2:     int options, int *last_boundclass) {
jbe@0:   // ASSERT: uc >= 0 && uc < 0x110000
jbe@0:   const utf8proc_property_t *property;
jbe@3:   utf8proc_propval_t category;
jbe@0:   int32_t hangul_sindex;
jbe@0:   property = utf8proc_get_property(uc);
jbe@3:   category = property->category;
jbe@0:   hangul_sindex = uc - UTF8PROC_HANGUL_SBASE;
jbe@3:   if (options & (UTF8PROC_COMPOSE|UTF8PROC_DECOMPOSE)) {
jbe@3:     if (hangul_sindex >= 0 && hangul_sindex < UTF8PROC_HANGUL_SCOUNT) {
jbe@3:       int32_t hangul_tindex;
jbe@3:       if (bufsize >= 1) {
jbe@3:         dst[0] = UTF8PROC_HANGUL_LBASE +
jbe@3:           hangul_sindex / UTF8PROC_HANGUL_NCOUNT;
jbe@3:         if (bufsize >= 2) dst[1] = UTF8PROC_HANGUL_VBASE +
jbe@3:           (hangul_sindex % UTF8PROC_HANGUL_NCOUNT) / UTF8PROC_HANGUL_TCOUNT;
jbe@3:       }
jbe@3:       hangul_tindex = hangul_sindex % UTF8PROC_HANGUL_TCOUNT;
jbe@3:       if (!hangul_tindex) return 2;
jbe@3:       if (bufsize >= 3) dst[2] = UTF8PROC_HANGUL_TBASE + hangul_tindex;
jbe@3:       return 3;
jbe@0:     }
jbe@3:   }
jbe@3:   if (options & UTF8PROC_REJECTNA) {
jbe@3:     if (!category) return UTF8PROC_ERROR_NOTASSIGNED;
jbe@3:   }
jbe@3:   if (options & UTF8PROC_IGNORE) {
jbe@3:     if (property->ignorable) return 0;
jbe@3:   }
jbe@3:   if (options & UTF8PROC_LUMP) {
jbe@3:     if (category == UTF8PROC_CATEGORY_ZS) utf8proc_decompose_lump(0x0020);
jbe@3:     if (uc == 0x2018 || uc == 0x2019 || uc == 0x02BC || uc == 0x02C8)
jbe@3:       utf8proc_decompose_lump(0x0027);
jbe@3:     if (category == UTF8PROC_CATEGORY_PD || uc == 0x2212)
jbe@3:       utf8proc_decompose_lump(0x002D);
jbe@3:     if (uc == 0x2044 || uc == 0x2215) utf8proc_decompose_lump(0x002F);
jbe@3:     if (uc == 0x2236) utf8proc_decompose_lump(0x003A);
jbe@3:     if (uc == 0x2039 || uc == 0x2329 || uc == 0x3008)
jbe@3:       utf8proc_decompose_lump(0x003C);
jbe@3:     if (uc == 0x203A || uc == 0x232A || uc == 0x3009)
jbe@3:       utf8proc_decompose_lump(0x003E);
jbe@3:     if (uc == 0x2216) utf8proc_decompose_lump(0x005C);
jbe@3:     if (uc == 0x02C4 || uc == 0x02C6 || uc == 0x2038 || uc == 0x2303)
jbe@3:       utf8proc_decompose_lump(0x005E);
jbe@3:     if (category == UTF8PROC_CATEGORY_PC || uc == 0x02CD)
jbe@3:       utf8proc_decompose_lump(0x005F);
jbe@3:     if (uc == 0x02CB) utf8proc_decompose_lump(0x0060);
jbe@3:     if (uc == 0x2223) utf8proc_decompose_lump(0x007C);
jbe@3:     if (uc == 0x223C) utf8proc_decompose_lump(0x007E);
jbe@3:     if ((options & UTF8PROC_NLF2LS) && (options & UTF8PROC_NLF2PS)) {
jbe@3:       if (category == UTF8PROC_CATEGORY_ZL ||
jbe@3:           category == UTF8PROC_CATEGORY_ZP)
jbe@3:         utf8proc_decompose_lump(0x000A);
jbe@3:     }
jbe@3:   }
jbe@3:   if (options & UTF8PROC_STRIPMARK) {
jbe@3:     if (category == UTF8PROC_CATEGORY_MN ||
jbe@3:       category == UTF8PROC_CATEGORY_MC ||
jbe@3:       category == UTF8PROC_CATEGORY_ME) return 0;
jbe@3:   }
jbe@3:   if (options & UTF8PROC_CASEFOLD) {
jbe@3:     if (property->casefold_mapping) {
jbe@3:       const int32_t *casefold_entry;
jbe@3:       ssize_t written = 0;
jbe@3:       for (casefold_entry = property->casefold_mapping;
jbe@3:           *casefold_entry >= 0; casefold_entry++) {
jbe@3:         written += utf8proc_decompose_char(*casefold_entry, dst+written,
jbe@3:           (bufsize > written) ? (bufsize - written) : 0, options,
jbe@3:           last_boundclass);
jbe@3:         if (written < 0) return UTF8PROC_ERROR_OVERFLOW;
jbe@3:       }
jbe@3:       return written;
jbe@3:     }
jbe@3:   }
jbe@3:   if (options & (UTF8PROC_COMPOSE|UTF8PROC_DECOMPOSE)) {
jbe@3:     if (property->decomp_mapping &&
jbe@3:         (!property->decomp_type || (options & UTF8PROC_COMPAT))) {
jbe@3:       const int32_t *decomp_entry;
jbe@3:       ssize_t written = 0;
jbe@3:       for (decomp_entry = property->decomp_mapping;
jbe@3:           *decomp_entry >= 0; decomp_entry++) {
jbe@3:         written += utf8proc_decompose_char(*decomp_entry, dst+written,
jbe@3:           (bufsize > written) ? (bufsize - written) : 0, options,
jbe@2:         last_boundclass);
jbe@3:         if (written < 0) return UTF8PROC_ERROR_OVERFLOW;
jbe@3:       }
jbe@3:       return written;
jbe@0:     }
jbe@3:   }
jbe@3:   if (options & UTF8PROC_CHARBOUND) {
jbe@2:     bool boundary;
jbe@2:     int tbc, lbc;
jbe@2:     tbc =
jbe@2:       (uc == 0x000D) ? UTF8PROC_BOUNDCLASS_CR :
jbe@2:       (uc == 0x000A) ? UTF8PROC_BOUNDCLASS_LF :
jbe@2:       ((category == UTF8PROC_CATEGORY_ZL ||
jbe@2:         category == UTF8PROC_CATEGORY_ZP ||
jbe@2:         category == UTF8PROC_CATEGORY_CC ||
jbe@2:         category == UTF8PROC_CATEGORY_CF) &&
jbe@2:         !(uc == 0x200C || uc == 0x200D)) ? UTF8PROC_BOUNDCLASS_CONTROL :
jbe@2:       property->extend ? UTF8PROC_BOUNDCLASS_EXTEND :
jbe@2:       ((uc >= UTF8PROC_HANGUL_L_START && uc < UTF8PROC_HANGUL_L_END) ||
jbe@2:         uc == UTF8PROC_HANGUL_L_FILLER) ? UTF8PROC_BOUNDCLASS_L :
jbe@2:       (uc >= UTF8PROC_HANGUL_V_START && uc < UTF8PROC_HANGUL_V_END) ?
jbe@2:         UTF8PROC_BOUNDCLASS_V :
jbe@2:       (uc >= UTF8PROC_HANGUL_T_START && uc < UTF8PROC_HANGUL_T_END) ?
jbe@2:         UTF8PROC_BOUNDCLASS_T :
jbe@2:       (uc >= UTF8PROC_HANGUL_S_START && uc < UTF8PROC_HANGUL_S_END) ? (
jbe@2:         ((uc-UTF8PROC_HANGUL_SBASE) % UTF8PROC_HANGUL_TCOUNT == 0) ?
jbe@2:           UTF8PROC_BOUNDCLASS_LV : UTF8PROC_BOUNDCLASS_LVT
jbe@2:       ) :
jbe@2:       UTF8PROC_BOUNDCLASS_OTHER;
jbe@2:     lbc = *last_boundclass;
jbe@2:     boundary =
jbe@2:       (tbc == UTF8PROC_BOUNDCLASS_EXTEND) ? false :
jbe@2:       (lbc == UTF8PROC_BOUNDCLASS_START) ? true :
jbe@2:       (lbc == UTF8PROC_BOUNDCLASS_CR &&
jbe@2:        tbc == UTF8PROC_BOUNDCLASS_LF) ? false :
jbe@2:       (lbc == UTF8PROC_BOUNDCLASS_CONTROL) ? true :
jbe@2:       (tbc == UTF8PROC_BOUNDCLASS_CONTROL) ? true :
jbe@2:       (lbc == UTF8PROC_BOUNDCLASS_L &&
jbe@2:        (tbc == UTF8PROC_BOUNDCLASS_L ||
jbe@2:         tbc == UTF8PROC_BOUNDCLASS_V ||
jbe@2:         tbc == UTF8PROC_BOUNDCLASS_LV ||
jbe@2:         tbc == UTF8PROC_BOUNDCLASS_LVT)) ? false :
jbe@2:       ((lbc == UTF8PROC_BOUNDCLASS_LV ||
jbe@2:         lbc == UTF8PROC_BOUNDCLASS_V) &&
jbe@2:        (tbc == UTF8PROC_BOUNDCLASS_V ||
jbe@2:         tbc == UTF8PROC_BOUNDCLASS_T)) ? false :
jbe@2:       ((lbc == UTF8PROC_BOUNDCLASS_LVT ||
jbe@2:         lbc == UTF8PROC_BOUNDCLASS_T) &&
jbe@2:        tbc == UTF8PROC_BOUNDCLASS_T) ? false :
jbe@2:        true;
jbe@2:     *last_boundclass = tbc;
jbe@2:     if (boundary) {
jbe@2:       if (bufsize >= 1) dst[0] = 0xFFFF;
jbe@2:       if (bufsize >= 2) dst[1] = uc;
jbe@2:       return 2;
jbe@2:     }
jbe@0:   }
jbe@2:   if (bufsize >= 1) *dst = uc;
jbe@2:   return 1;
jbe@0: }
jbe@0: 
jbe@7: ssize_t utf8proc_decompose(
jbe@7:   const uint8_t *str, ssize_t strlen,
jbe@7:   int32_t *buffer, ssize_t bufsize, int options
jbe@7: ) {
jbe@0:   // strlen will be ignored, if UTF8PROC_NULLTERM is set in options
jbe@0:   ssize_t wpos = 0;
jbe@3:   if ((options & UTF8PROC_COMPOSE) && (options & UTF8PROC_DECOMPOSE))
jbe@3:     return UTF8PROC_ERROR_INVALIDOPTS;
jbe@3:   if ((options & UTF8PROC_STRIPMARK) &&
jbe@3:       !(options & UTF8PROC_COMPOSE) && !(options & UTF8PROC_DECOMPOSE))
jbe@3:     return UTF8PROC_ERROR_INVALIDOPTS;
jbe@0:   {
jbe@0:     int32_t uc;
jbe@0:     ssize_t rpos = 0;
jbe@0:     ssize_t decomp_result;
jbe@2:     int boundclass = UTF8PROC_BOUNDCLASS_START;
jbe@0:     while (1) {
jbe@0:       if (options & UTF8PROC_NULLTERM) {
jbe@0:         rpos += utf8proc_iterate(str + rpos, -1, &uc);
jbe@0:         // checking of return value is not neccessary,
jbe@0:         // as 'uc' is < 0 in case of error
jbe@0:         if (uc < 0) return UTF8PROC_ERROR_INVALIDUTF8;
jbe@0:         if (rpos < 0) return UTF8PROC_ERROR_OVERFLOW;
jbe@0:         if (uc == 0) break;
jbe@0:       } else {
jbe@0:         if (rpos >= strlen) break;
jbe@0:         rpos += utf8proc_iterate(str + rpos, strlen - rpos, &uc);
jbe@0:         if (uc < 0) return UTF8PROC_ERROR_INVALIDUTF8;
jbe@0:       }
jbe@0:       decomp_result = utf8proc_decompose_char(
jbe@2:         uc, buffer + wpos, (bufsize > wpos) ? (bufsize - wpos) : 0, options,
jbe@2:         &boundclass
jbe@0:       );
jbe@0:       if (decomp_result < 0) return decomp_result;
jbe@0:       wpos += decomp_result;
jbe@0:       // prohibiting integer overflows due to too long strings:
jbe@0:       if (wpos < 0 || wpos > SSIZE_MAX/sizeof(int32_t)/2)
jbe@0:         return UTF8PROC_ERROR_OVERFLOW;
jbe@0:     }
jbe@0:   }
jbe@2:   if ((options & (UTF8PROC_COMPOSE|UTF8PROC_DECOMPOSE)) && bufsize >= wpos) {
jbe@0:     ssize_t pos = 0;
jbe@0:     while (pos < wpos-1) {
jbe@0:       int32_t uc1, uc2;
jbe@0:       const utf8proc_property_t *property1, *property2;
jbe@0:       uc1 = buffer[pos];
jbe@0:       uc2 = buffer[pos+1];
jbe@0:       property1 = utf8proc_get_property(uc1);
jbe@0:       property2 = utf8proc_get_property(uc2);
jbe@0:       if (property1->combining_class > property2->combining_class &&
jbe@0:           property2->combining_class > 0) {
jbe@0:         buffer[pos] = uc2;
jbe@0:         buffer[pos+1] = uc1;
jbe@0:         if (pos > 0) pos--; else pos++;
jbe@0:       } else {
jbe@0:         pos++;
jbe@0:       }
jbe@0:     }
jbe@0:   }
jbe@0:   return wpos;
jbe@0: }
jbe@0: 
jbe@0: ssize_t utf8proc_reencode(int32_t *buffer, ssize_t length, int options) {
jbe@0:   // UTF8PROC_NULLTERM option will be ignored, 'length' is never ignored
jbe@0:   // ASSERT: 'buffer' has one spare byte of free space at the end!
jbe@0:   if (options & (UTF8PROC_NLF2LS | UTF8PROC_NLF2PS | UTF8PROC_STRIPCC)) {
jbe@0:     ssize_t rpos;
jbe@0:     ssize_t wpos = 0;
jbe@0:     int32_t uc;
jbe@0:     for (rpos = 0; rpos < length; rpos++) {
jbe@0:       uc = buffer[rpos];
jbe@0:       if (uc == 0x000D && rpos < length-1 && buffer[rpos+1] == 0x000A) rpos++;
jbe@0:       if (uc == 0x000A || uc == 0x000D || uc == 0x0085 ||
jbe@0:           ((options & UTF8PROC_STRIPCC) && (uc == 0x000B || uc == 0x000C))) {
jbe@0:         if (options & UTF8PROC_NLF2LS) {
jbe@0:           if (options & UTF8PROC_NLF2PS) {
jbe@0:             buffer[wpos++] = 0x000A;
jbe@0:           } else {
jbe@0:             buffer[wpos++] = 0x2028;
jbe@0:           }
jbe@0:         } else {
jbe@0:           if (options & UTF8PROC_NLF2PS) {
jbe@0:             buffer[wpos++] = 0x2029;
jbe@0:           } else {
jbe@0:             buffer[wpos++] = 0x0020;
jbe@0:           }
jbe@0:         }
jbe@0:       } else if ((options & UTF8PROC_STRIPCC) &&
jbe@0:           (uc < 0x0020 || (uc >= 0x007F && uc < 0x00A0))) {
jbe@0:         if (uc == 0x0009) buffer[wpos++] = 0x0020;
jbe@0:       } else {
jbe@0:         buffer[wpos++] = uc;
jbe@0:       }
jbe@0:     }
jbe@0:     length = wpos;
jbe@0:   }
jbe@0:   if (options & UTF8PROC_COMPOSE) {
jbe@0:     int32_t *starter = NULL;
jbe@0:     int32_t current_char;
jbe@0:     const utf8proc_property_t *starter_property = NULL, *current_property;
jbe@3:     utf8proc_propval_t max_combining_class = -1;
jbe@0:     ssize_t rpos;
jbe@0:     ssize_t wpos = 0;
jbe@0:     int32_t composition;
jbe@0:     for (rpos = 0; rpos < length; rpos++) {
jbe@0:       current_char = buffer[rpos];
jbe@0:       current_property = utf8proc_get_property(current_char);
jbe@0:       if (starter && current_property->combining_class > max_combining_class) {
jbe@0:         // combination perhaps possible
jbe@0:         int32_t hangul_lindex;
jbe@0:         int32_t hangul_sindex;
jbe@0:         hangul_lindex = *starter - UTF8PROC_HANGUL_LBASE;
jbe@0:         if (hangul_lindex >= 0 && hangul_lindex < UTF8PROC_HANGUL_LCOUNT) {
jbe@0:           int32_t hangul_vindex;
jbe@0:           hangul_vindex = current_char - UTF8PROC_HANGUL_VBASE;
jbe@0:           if (hangul_vindex >= 0 && hangul_vindex < UTF8PROC_HANGUL_VCOUNT) {
jbe@0:             *starter = UTF8PROC_HANGUL_SBASE +
jbe@0:               (hangul_lindex * UTF8PROC_HANGUL_VCOUNT + hangul_vindex) *
jbe@0:               UTF8PROC_HANGUL_TCOUNT;
jbe@0:             starter_property = NULL;
jbe@0:             continue;
jbe@0:           }
jbe@0:         }
jbe@0:         hangul_sindex = *starter - UTF8PROC_HANGUL_SBASE;
jbe@0:         if (hangul_sindex >= 0 && hangul_sindex < UTF8PROC_HANGUL_SCOUNT &&
jbe@0:             (hangul_sindex % UTF8PROC_HANGUL_TCOUNT) == 0) {
jbe@0:           int32_t hangul_tindex;
jbe@0:           hangul_tindex = current_char - UTF8PROC_HANGUL_TBASE;
jbe@0:           if (hangul_tindex >= 0 && hangul_tindex < UTF8PROC_HANGUL_TCOUNT) {
jbe@0:             *starter += hangul_tindex;
jbe@0:             starter_property = NULL;
jbe@0:             continue;
jbe@0:           }
jbe@0:         }
jbe@0:         if (!starter_property) {
jbe@0:           starter_property = utf8proc_get_property(*starter);
jbe@0:         }
jbe@0:         if (starter_property->comb1st_index >= 0 &&
jbe@0:             current_property->comb2nd_index >= 0) {
jbe@0:           composition = utf8proc_combinations[
jbe@0:             starter_property->comb1st_index +
jbe@0:             current_property->comb2nd_index
jbe@0:           ];
jbe@0:           if (composition >= 0 && (!(options & UTF8PROC_STABLE) ||
jbe@0:               !(utf8proc_get_property(composition)->comp_exclusion))) {
jbe@0:             *starter = composition;
jbe@0:             starter_property = NULL;
jbe@0:             continue;
jbe@0:           }
jbe@0:         }
jbe@0:       }
jbe@0:       buffer[wpos] = current_char;
jbe@0:       if (current_property->combining_class) {
jbe@0:         if (current_property->combining_class > max_combining_class) {
jbe@0:           max_combining_class = current_property->combining_class;
jbe@0:         }
jbe@0:       } else {
jbe@0:         starter = buffer + wpos;
jbe@0:         starter_property = NULL;
jbe@0:         max_combining_class = -1;
jbe@0:       }
jbe@0:       wpos++;
jbe@0:     }
jbe@0:     length = wpos;
jbe@0:   }
jbe@0:   {
jbe@0:     ssize_t rpos, wpos = 0;
jbe@0:     int32_t uc;
jbe@0:     for (rpos = 0; rpos < length; rpos++) {
jbe@0:       uc = buffer[rpos];
jbe@0:       wpos += utf8proc_encode_char(uc, ((uint8_t *)buffer) + wpos);
jbe@0:     }
jbe@0:     ((uint8_t *)buffer)[wpos] = 0;
jbe@0:     return wpos;
jbe@0:   }
jbe@0: }
jbe@0: 
jbe@7: ssize_t utf8proc_map(
jbe@7:   const uint8_t *str, ssize_t strlen, uint8_t **dstptr, int options
jbe@7: ) {
jbe@0:   int32_t *buffer;
jbe@0:   ssize_t result;
jbe@0:   *dstptr = NULL;
jbe@0:   result = utf8proc_decompose(str, strlen, NULL, 0, options);
jbe@0:   if (result < 0) return result;
jbe@0:   buffer = malloc(result * sizeof(int32_t) + 1);
jbe@0:   if (!buffer) return UTF8PROC_ERROR_NOMEM;
jbe@0:   result = utf8proc_decompose(str, strlen, buffer, result, options);
jbe@0:   if (result < 0) {
jbe@0:     free(buffer);
jbe@0:     return result;
jbe@0:   }
jbe@0:   result = utf8proc_reencode(buffer, result, options);
jbe@0:   if (result < 0) {
jbe@0:     free(buffer);
jbe@0:     return result;
jbe@0:   }
jbe@0:   {
jbe@0:     int32_t *newptr;
jbe@0:     newptr = realloc(buffer, result+1);
jbe@0:     if (newptr) buffer = newptr;
jbe@0:   }
jbe@0:   *dstptr = (uint8_t *)buffer;
jbe@0:   return result;
jbe@0: }
jbe@0: 
jbe@7: uint8_t *utf8proc_NFD(const uint8_t *str) {
jbe@0:   uint8_t *retval;
jbe@2:   utf8proc_map(str, 0, &retval, UTF8PROC_NULLTERM | UTF8PROC_STABLE |
jbe@2:     UTF8PROC_DECOMPOSE);
jbe@0:   return retval;
jbe@0: }
jbe@0: 
jbe@7: uint8_t *utf8proc_NFC(const uint8_t *str) {
jbe@0:   uint8_t *retval;
jbe@0:   utf8proc_map(str, 0, &retval, UTF8PROC_NULLTERM | UTF8PROC_STABLE |
jbe@0:     UTF8PROC_COMPOSE);
jbe@0:   return retval;
jbe@0: }
jbe@0: 
jbe@7: uint8_t *utf8proc_NFKD(const uint8_t *str) {
jbe@0:   uint8_t *retval;
jbe@0:   utf8proc_map(str, 0, &retval, UTF8PROC_NULLTERM | UTF8PROC_STABLE |
jbe@2:     UTF8PROC_DECOMPOSE | UTF8PROC_COMPAT);
jbe@0:   return retval;
jbe@0: }
jbe@0: 
jbe@7: uint8_t *utf8proc_NFKC(const uint8_t *str) {
jbe@0:   uint8_t *retval;
jbe@0:   utf8proc_map(str, 0, &retval, UTF8PROC_NULLTERM | UTF8PROC_STABLE |
jbe@0:     UTF8PROC_COMPOSE | UTF8PROC_COMPAT);
jbe@0:   return retval;
jbe@0: }
jbe@0: