1228753Smm/*- 2232153Smm * Copyright (c) 2003-2011 Tim Kientzle 3232153Smm * Copyright (c) 2011-2012 Michihiro NAKAJIMA 4228753Smm * All rights reserved. 5228753Smm * 6228753Smm * Redistribution and use in source and binary forms, with or without 7228753Smm * modification, are permitted provided that the following conditions 8228753Smm * are met: 9228753Smm * 1. Redistributions of source code must retain the above copyright 10228753Smm * notice, this list of conditions and the following disclaimer. 11228753Smm * 2. Redistributions in binary form must reproduce the above copyright 12228753Smm * notice, this list of conditions and the following disclaimer in the 13228753Smm * documentation and/or other materials provided with the distribution. 14228753Smm * 15228753Smm * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 16228753Smm * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 17228753Smm * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 18228753Smm * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 19228753Smm * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 20228753Smm * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 21228753Smm * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 22228753Smm * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 23228753Smm * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 24228753Smm * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 25228753Smm */ 26228753Smm 27228753Smm#include "archive_platform.h" 28228763Smm__FBSDID("$FreeBSD: stable/11/contrib/libarchive/libarchive/archive_string.c 368707 2020-12-16 22:25:20Z mm $"); 29228753Smm 30228753Smm/* 31228753Smm * Basic resizable string support, to simplify manipulating arbitrary-sized 32228753Smm * strings while minimizing heap activity. 33232153Smm * 34232153Smm * In particular, the buffer used by a string object is only grown, it 35232153Smm * never shrinks, so you can clear and reuse the same string object 36232153Smm * without incurring additional memory allocations. 37228753Smm */ 38228753Smm 39232153Smm#ifdef HAVE_ERRNO_H 40232153Smm#include <errno.h> 41232153Smm#endif 42232153Smm#ifdef HAVE_ICONV_H 43232153Smm#include <iconv.h> 44232153Smm#endif 45232153Smm#ifdef HAVE_LANGINFO_H 46232153Smm#include <langinfo.h> 47232153Smm#endif 48232153Smm#ifdef HAVE_LOCALCHARSET_H 49232153Smm#include <localcharset.h> 50232153Smm#endif 51228753Smm#ifdef HAVE_STDLIB_H 52228753Smm#include <stdlib.h> 53228753Smm#endif 54228753Smm#ifdef HAVE_STRING_H 55228753Smm#include <string.h> 56228753Smm#endif 57228753Smm#ifdef HAVE_WCHAR_H 58228753Smm#include <wchar.h> 59228753Smm#endif 60228753Smm#if defined(_WIN32) && !defined(__CYGWIN__) 61228753Smm#include <windows.h> 62232153Smm#include <locale.h> 63228753Smm#endif 64228753Smm 65232153Smm#include "archive_endian.h" 66228753Smm#include "archive_private.h" 67228753Smm#include "archive_string.h" 68232153Smm#include "archive_string_composition.h" 69228753Smm 70232153Smm#if !defined(HAVE_WMEMCPY) && !defined(wmemcpy) 71232153Smm#define wmemcpy(a,b,i) (wchar_t *)memcpy((a), (b), (i) * sizeof(wchar_t)) 72232153Smm#endif 73232153Smm 74299529Smm#if !defined(HAVE_WMEMMOVE) && !defined(wmemmove) 75299529Smm#define wmemmove(a,b,i) (wchar_t *)memmove((a), (b), (i) * sizeof(wchar_t)) 76299529Smm#endif 77299529Smm 78358088Smm#undef max 79358088Smm#define max(a, b) ((a)>(b)?(a):(b)) 80358088Smm 81232153Smmstruct archive_string_conv { 82232153Smm struct archive_string_conv *next; 83232153Smm char *from_charset; 84232153Smm char *to_charset; 85232153Smm unsigned from_cp; 86232153Smm unsigned to_cp; 87232153Smm /* Set 1 if from_charset and to_charset are the same. */ 88232153Smm int same; 89232153Smm int flag; 90232153Smm#define SCONV_TO_CHARSET 1 /* MBS is being converted to specified 91232153Smm * charset. */ 92232153Smm#define SCONV_FROM_CHARSET (1<<1) /* MBS is being converted from 93232153Smm * specified charset. */ 94232153Smm#define SCONV_BEST_EFFORT (1<<2) /* Copy at least ASCII code. */ 95232153Smm#define SCONV_WIN_CP (1<<3) /* Use Windows API for converting 96232153Smm * MBS. */ 97232153Smm#define SCONV_UTF8_LIBARCHIVE_2 (1<<4) /* Incorrect UTF-8 made by libarchive 98232153Smm * 2.x in the wrong assumption. */ 99232153Smm#define SCONV_NORMALIZATION_C (1<<6) /* Need normalization to be Form C. 100232153Smm * Before UTF-8 characters are actually 101232153Smm * processed. */ 102232153Smm#define SCONV_NORMALIZATION_D (1<<7) /* Need normalization to be Form D. 103232153Smm * Before UTF-8 characters are actually 104232153Smm * processed. 105232153Smm * Currently this only for MAC OS X. */ 106232153Smm#define SCONV_TO_UTF8 (1<<8) /* "to charset" side is UTF-8. */ 107232153Smm#define SCONV_FROM_UTF8 (1<<9) /* "from charset" side is UTF-8. */ 108232153Smm#define SCONV_TO_UTF16BE (1<<10) /* "to charset" side is UTF-16BE. */ 109232153Smm#define SCONV_FROM_UTF16BE (1<<11) /* "from charset" side is UTF-16BE. */ 110232153Smm#define SCONV_TO_UTF16LE (1<<12) /* "to charset" side is UTF-16LE. */ 111232153Smm#define SCONV_FROM_UTF16LE (1<<13) /* "from charset" side is UTF-16LE. */ 112232153Smm#define SCONV_TO_UTF16 (SCONV_TO_UTF16BE | SCONV_TO_UTF16LE) 113232153Smm#define SCONV_FROM_UTF16 (SCONV_FROM_UTF16BE | SCONV_FROM_UTF16LE) 114232153Smm 115232153Smm#if HAVE_ICONV 116232153Smm iconv_t cd; 117232153Smm iconv_t cd_w;/* Use at archive_mstring on 118232153Smm * Windows. */ 119232153Smm#endif 120232153Smm /* A temporary buffer for normalization. */ 121232153Smm struct archive_string utftmp; 122232153Smm int (*converter[2])(struct archive_string *, const void *, size_t, 123232153Smm struct archive_string_conv *); 124232153Smm int nconverter; 125232153Smm}; 126232153Smm 127232153Smm#define CP_C_LOCALE 0 /* "C" locale only for this file. */ 128232153Smm#define CP_UTF16LE 1200 129232153Smm#define CP_UTF16BE 1201 130232153Smm 131232153Smm#define IS_HIGH_SURROGATE_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDBFF) 132232153Smm#define IS_LOW_SURROGATE_LA(uc) ((uc) >= 0xDC00 && (uc) <= 0xDFFF) 133232153Smm#define IS_SURROGATE_PAIR_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDFFF) 134232153Smm#define UNICODE_MAX 0x10FFFF 135232153Smm#define UNICODE_R_CHAR 0xFFFD /* Replacement character. */ 136232153Smm/* Set U+FFFD(Replacement character) in UTF-8. */ 137299529Smmstatic const char utf8_replacement_char[] = {0xef, 0xbf, 0xbd}; 138232153Smm 139232153Smmstatic struct archive_string_conv *find_sconv_object(struct archive *, 140232153Smm const char *, const char *); 141232153Smmstatic void add_sconv_object(struct archive *, struct archive_string_conv *); 142232153Smmstatic struct archive_string_conv *create_sconv_object(const char *, 143232153Smm const char *, unsigned, int); 144232153Smmstatic void free_sconv_object(struct archive_string_conv *); 145232153Smmstatic struct archive_string_conv *get_sconv_object(struct archive *, 146232153Smm const char *, const char *, int); 147232153Smmstatic unsigned make_codepage_from_charset(const char *); 148232153Smmstatic unsigned get_current_codepage(void); 149232153Smmstatic unsigned get_current_oemcp(void); 150232153Smmstatic size_t mbsnbytes(const void *, size_t); 151232153Smmstatic size_t utf16nbytes(const void *, size_t); 152232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 153232153Smmstatic int archive_wstring_append_from_mbs_in_codepage( 154232153Smm struct archive_wstring *, const char *, size_t, 155232153Smm struct archive_string_conv *); 156232153Smmstatic int archive_string_append_from_wcs_in_codepage(struct archive_string *, 157232153Smm const wchar_t *, size_t, struct archive_string_conv *); 158232153Smmstatic int is_big_endian(void); 159232153Smmstatic int strncat_in_codepage(struct archive_string *, const void *, 160232153Smm size_t, struct archive_string_conv *); 161238856Smmstatic int win_strncat_from_utf16be(struct archive_string *, const void *, 162232153Smm size_t, struct archive_string_conv *); 163238856Smmstatic int win_strncat_from_utf16le(struct archive_string *, const void *, 164232153Smm size_t, struct archive_string_conv *); 165238856Smmstatic int win_strncat_to_utf16be(struct archive_string *, const void *, 166232153Smm size_t, struct archive_string_conv *); 167238856Smmstatic int win_strncat_to_utf16le(struct archive_string *, const void *, 168232153Smm size_t, struct archive_string_conv *); 169238856Smm#endif 170238856Smmstatic int best_effort_strncat_from_utf16be(struct archive_string *, 171238856Smm const void *, size_t, struct archive_string_conv *); 172238856Smmstatic int best_effort_strncat_from_utf16le(struct archive_string *, 173238856Smm const void *, size_t, struct archive_string_conv *); 174238856Smmstatic int best_effort_strncat_to_utf16be(struct archive_string *, 175238856Smm const void *, size_t, struct archive_string_conv *); 176238856Smmstatic int best_effort_strncat_to_utf16le(struct archive_string *, 177238856Smm const void *, size_t, struct archive_string_conv *); 178232153Smm#if defined(HAVE_ICONV) 179232153Smmstatic int iconv_strncat_in_locale(struct archive_string *, const void *, 180232153Smm size_t, struct archive_string_conv *); 181232153Smm#endif 182238856Smmstatic int best_effort_strncat_in_locale(struct archive_string *, 183238856Smm const void *, size_t, struct archive_string_conv *); 184232153Smmstatic int _utf8_to_unicode(uint32_t *, const char *, size_t); 185232153Smmstatic int utf8_to_unicode(uint32_t *, const char *, size_t); 186232153Smmstatic inline uint32_t combine_surrogate_pair(uint32_t, uint32_t); 187232153Smmstatic int cesu8_to_unicode(uint32_t *, const char *, size_t); 188232153Smmstatic size_t unicode_to_utf8(char *, size_t, uint32_t); 189232153Smmstatic int utf16_to_unicode(uint32_t *, const char *, size_t, int); 190232153Smmstatic size_t unicode_to_utf16be(char *, size_t, uint32_t); 191232153Smmstatic size_t unicode_to_utf16le(char *, size_t, uint32_t); 192232153Smmstatic int strncat_from_utf8_libarchive2(struct archive_string *, 193232153Smm const void *, size_t, struct archive_string_conv *); 194232153Smmstatic int strncat_from_utf8_to_utf8(struct archive_string *, const void *, 195232153Smm size_t, struct archive_string_conv *); 196232153Smmstatic int archive_string_normalize_C(struct archive_string *, const void *, 197232153Smm size_t, struct archive_string_conv *); 198232153Smmstatic int archive_string_normalize_D(struct archive_string *, const void *, 199232153Smm size_t, struct archive_string_conv *); 200232153Smmstatic int archive_string_append_unicode(struct archive_string *, 201232153Smm const void *, size_t, struct archive_string_conv *); 202232153Smm 203232153Smmstatic struct archive_string * 204232153Smmarchive_string_append(struct archive_string *as, const char *p, size_t s) 205228753Smm{ 206232153Smm if (archive_string_ensure(as, as->length + s + 1) == NULL) 207232153Smm return (NULL); 208318482Smm if (s) 209318482Smm memmove(as->s + as->length, p, s); 210228753Smm as->length += s; 211232153Smm as->s[as->length] = 0; 212228753Smm return (as); 213228753Smm} 214228753Smm 215232153Smmstatic struct archive_wstring * 216232153Smmarchive_wstring_append(struct archive_wstring *as, const wchar_t *p, size_t s) 217232153Smm{ 218232153Smm if (archive_wstring_ensure(as, as->length + s + 1) == NULL) 219232153Smm return (NULL); 220324417Smm if (s) 221324417Smm wmemmove(as->s + as->length, p, s); 222232153Smm as->length += s; 223232153Smm as->s[as->length] = 0; 224232153Smm return (as); 225232153Smm} 226232153Smm 227313570Smmstruct archive_string * 228313570Smmarchive_array_append(struct archive_string *as, const char *p, size_t s) 229313570Smm{ 230313570Smm return archive_string_append(as, p, s); 231313570Smm} 232313570Smm 233228753Smmvoid 234232153Smmarchive_string_concat(struct archive_string *dest, struct archive_string *src) 235228753Smm{ 236232153Smm if (archive_string_append(dest, src->s, src->length) == NULL) 237232153Smm __archive_errx(1, "Out of memory"); 238228753Smm} 239228753Smm 240228753Smmvoid 241238856Smmarchive_wstring_concat(struct archive_wstring *dest, 242238856Smm struct archive_wstring *src) 243228753Smm{ 244232153Smm if (archive_wstring_append(dest, src->s, src->length) == NULL) 245232153Smm __archive_errx(1, "Out of memory"); 246228753Smm} 247228753Smm 248228753Smmvoid 249232153Smmarchive_string_free(struct archive_string *as) 250228753Smm{ 251228753Smm as->length = 0; 252228753Smm as->buffer_length = 0; 253232153Smm free(as->s); 254232153Smm as->s = NULL; 255228753Smm} 256228753Smm 257232153Smmvoid 258232153Smmarchive_wstring_free(struct archive_wstring *as) 259232153Smm{ 260232153Smm as->length = 0; 261232153Smm as->buffer_length = 0; 262232153Smm free(as->s); 263232153Smm as->s = NULL; 264232153Smm} 265232153Smm 266232153Smmstruct archive_wstring * 267232153Smmarchive_wstring_ensure(struct archive_wstring *as, size_t s) 268232153Smm{ 269232153Smm return (struct archive_wstring *) 270232153Smm archive_string_ensure((struct archive_string *)as, 271232153Smm s * sizeof(wchar_t)); 272232153Smm} 273232153Smm 274228753Smm/* Returns NULL on any allocation failure. */ 275228753Smmstruct archive_string * 276232153Smmarchive_string_ensure(struct archive_string *as, size_t s) 277228753Smm{ 278232153Smm char *p; 279232153Smm size_t new_length; 280232153Smm 281228753Smm /* If buffer is already big enough, don't reallocate. */ 282228753Smm if (as->s && (s <= as->buffer_length)) 283228753Smm return (as); 284228753Smm 285228753Smm /* 286228753Smm * Growing the buffer at least exponentially ensures that 287228753Smm * append operations are always linear in the number of 288228753Smm * characters appended. Using a smaller growth rate for 289228753Smm * larger buffers reduces memory waste somewhat at the cost of 290228753Smm * a larger constant factor. 291228753Smm */ 292228753Smm if (as->buffer_length < 32) 293228753Smm /* Start with a minimum 32-character buffer. */ 294232153Smm new_length = 32; 295228753Smm else if (as->buffer_length < 8192) 296228753Smm /* Buffers under 8k are doubled for speed. */ 297232153Smm new_length = as->buffer_length + as->buffer_length; 298228753Smm else { 299228753Smm /* Buffers 8k and over grow by at least 25% each time. */ 300232153Smm new_length = as->buffer_length + as->buffer_length / 4; 301232153Smm /* Be safe: If size wraps, fail. */ 302232153Smm if (new_length < as->buffer_length) { 303232153Smm /* On failure, wipe the string and return NULL. */ 304232153Smm archive_string_free(as); 305232153Smm errno = ENOMEM;/* Make sure errno has ENOMEM. */ 306228753Smm return (NULL); 307228753Smm } 308228753Smm } 309228753Smm /* 310228753Smm * The computation above is a lower limit to how much we'll 311228753Smm * grow the buffer. In any case, we have to grow it enough to 312228753Smm * hold the request. 313228753Smm */ 314232153Smm if (new_length < s) 315232153Smm new_length = s; 316228753Smm /* Now we can reallocate the buffer. */ 317232153Smm p = (char *)realloc(as->s, new_length); 318232153Smm if (p == NULL) { 319232153Smm /* On failure, wipe the string and return NULL. */ 320232153Smm archive_string_free(as); 321232153Smm errno = ENOMEM;/* Make sure errno has ENOMEM. */ 322228753Smm return (NULL); 323232153Smm } 324232153Smm 325232153Smm as->s = p; 326232153Smm as->buffer_length = new_length; 327228753Smm return (as); 328228753Smm} 329228753Smm 330232153Smm/* 331232153Smm * TODO: See if there's a way to avoid scanning 332232153Smm * the source string twice. Then test to see 333232153Smm * if it actually helps (remember that we're almost 334232153Smm * always called with pretty short arguments, so 335232153Smm * such an optimization might not help). 336232153Smm */ 337228753Smmstruct archive_string * 338232153Smmarchive_strncat(struct archive_string *as, const void *_p, size_t n) 339228753Smm{ 340228753Smm size_t s; 341228753Smm const char *p, *pp; 342228753Smm 343228753Smm p = (const char *)_p; 344228753Smm 345228753Smm /* Like strlen(p), except won't examine positions beyond p[n]. */ 346228753Smm s = 0; 347228753Smm pp = p; 348228753Smm while (s < n && *pp) { 349228753Smm pp++; 350228753Smm s++; 351228753Smm } 352232153Smm if ((as = archive_string_append(as, p, s)) == NULL) 353232153Smm __archive_errx(1, "Out of memory"); 354232153Smm return (as); 355228753Smm} 356228753Smm 357232153Smmstruct archive_wstring * 358232153Smmarchive_wstrncat(struct archive_wstring *as, const wchar_t *p, size_t n) 359232153Smm{ 360232153Smm size_t s; 361232153Smm const wchar_t *pp; 362232153Smm 363232153Smm /* Like strlen(p), except won't examine positions beyond p[n]. */ 364232153Smm s = 0; 365232153Smm pp = p; 366232153Smm while (s < n && *pp) { 367232153Smm pp++; 368232153Smm s++; 369232153Smm } 370232153Smm if ((as = archive_wstring_append(as, p, s)) == NULL) 371232153Smm __archive_errx(1, "Out of memory"); 372232153Smm return (as); 373232153Smm} 374232153Smm 375228753Smmstruct archive_string * 376232153Smmarchive_strcat(struct archive_string *as, const void *p) 377228753Smm{ 378232153Smm /* strcat is just strncat without an effective limit. 379232153Smm * Assert that we'll never get called with a source 380232153Smm * string over 16MB. 381232153Smm * TODO: Review all uses of strcat in the source 382232153Smm * and try to replace them with strncat(). 383232153Smm */ 384232153Smm return archive_strncat(as, p, 0x1000000); 385228753Smm} 386228753Smm 387232153Smmstruct archive_wstring * 388232153Smmarchive_wstrcat(struct archive_wstring *as, const wchar_t *p) 389232153Smm{ 390232153Smm /* Ditto. */ 391232153Smm return archive_wstrncat(as, p, 0x1000000); 392232153Smm} 393232153Smm 394232153Smmstruct archive_string * 395232153Smmarchive_strappend_char(struct archive_string *as, char c) 396232153Smm{ 397232153Smm if ((as = archive_string_append(as, &c, 1)) == NULL) 398232153Smm __archive_errx(1, "Out of memory"); 399232153Smm return (as); 400232153Smm} 401232153Smm 402232153Smmstruct archive_wstring * 403232153Smmarchive_wstrappend_wchar(struct archive_wstring *as, wchar_t c) 404232153Smm{ 405232153Smm if ((as = archive_wstring_append(as, &c, 1)) == NULL) 406232153Smm __archive_errx(1, "Out of memory"); 407232153Smm return (as); 408232153Smm} 409232153Smm 410228753Smm/* 411232153Smm * Get the "current character set" name to use with iconv. 412232153Smm * On FreeBSD, the empty character set name "" chooses 413232153Smm * the correct character encoding for the current locale, 414232153Smm * so this isn't necessary. 415232153Smm * But iconv on Mac OS 10.6 doesn't seem to handle this correctly; 416232153Smm * on that system, we have to explicitly call nl_langinfo() 417232153Smm * to get the right name. Not sure about other platforms. 418232153Smm * 419232153Smm * NOTE: GNU libiconv does not recognize the character-set name 420232153Smm * which some platform nl_langinfo(CODESET) returns, so we should 421232153Smm * use locale_charset() instead of nl_langinfo(CODESET) for GNU libiconv. 422228753Smm */ 423232153Smmstatic const char * 424232153Smmdefault_iconv_charset(const char *charset) { 425232153Smm if (charset != NULL && charset[0] != '\0') 426232153Smm return charset; 427232153Smm#if HAVE_LOCALE_CHARSET && !defined(__APPLE__) 428232153Smm /* locale_charset() is broken on Mac OS */ 429232153Smm return locale_charset(); 430232153Smm#elif HAVE_NL_LANGINFO 431232153Smm return nl_langinfo(CODESET); 432232153Smm#else 433232153Smm return ""; 434232153Smm#endif 435232153Smm} 436232153Smm 437232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 438232153Smm 439232153Smm/* 440232153Smm * Convert MBS to WCS. 441232153Smm * Note: returns -1 if conversion fails. 442232153Smm */ 443232153Smmint 444232153Smmarchive_wstring_append_from_mbs(struct archive_wstring *dest, 445232153Smm const char *p, size_t len) 446228753Smm{ 447238856Smm return archive_wstring_append_from_mbs_in_codepage(dest, p, len, NULL); 448232153Smm} 449232153Smm 450232153Smmstatic int 451232153Smmarchive_wstring_append_from_mbs_in_codepage(struct archive_wstring *dest, 452232153Smm const char *s, size_t length, struct archive_string_conv *sc) 453232153Smm{ 454232153Smm int count, ret = 0; 455232153Smm UINT from_cp; 456232153Smm 457232153Smm if (sc != NULL) 458232153Smm from_cp = sc->from_cp; 459232153Smm else 460232153Smm from_cp = get_current_codepage(); 461232153Smm 462232153Smm if (from_cp == CP_C_LOCALE) { 463232153Smm /* 464353376Smm * "C" locale special processing. 465232153Smm */ 466232153Smm wchar_t *ws; 467232153Smm const unsigned char *mp; 468232153Smm 469232153Smm if (NULL == archive_wstring_ensure(dest, 470232153Smm dest->length + length + 1)) 471232153Smm return (-1); 472232153Smm 473232153Smm ws = dest->s + dest->length; 474232153Smm mp = (const unsigned char *)s; 475232153Smm count = 0; 476232153Smm while (count < (int)length && *mp) { 477232153Smm *ws++ = (wchar_t)*mp++; 478232153Smm count++; 479232153Smm } 480238856Smm } else if (sc != NULL && 481238856Smm (sc->flag & (SCONV_NORMALIZATION_C | SCONV_NORMALIZATION_D))) { 482232153Smm /* 483232153Smm * Normalize UTF-8 and UTF-16BE and convert it directly 484232153Smm * to UTF-16 as wchar_t. 485232153Smm */ 486232153Smm struct archive_string u16; 487232153Smm int saved_flag = sc->flag;/* save current flag. */ 488232153Smm 489232153Smm if (is_big_endian()) 490232153Smm sc->flag |= SCONV_TO_UTF16BE; 491232153Smm else 492232153Smm sc->flag |= SCONV_TO_UTF16LE; 493232153Smm 494232153Smm if (sc->flag & SCONV_FROM_UTF16) { 495232153Smm /* 496232153Smm * UTF-16BE/LE NFD ===> UTF-16 NFC 497238856Smm * UTF-16BE/LE NFC ===> UTF-16 NFD 498232153Smm */ 499248616Smm count = (int)utf16nbytes(s, length); 500232153Smm } else { 501232153Smm /* 502232153Smm * UTF-8 NFD ===> UTF-16 NFC 503238856Smm * UTF-8 NFC ===> UTF-16 NFD 504232153Smm */ 505248616Smm count = (int)mbsnbytes(s, length); 506232153Smm } 507232153Smm u16.s = (char *)dest->s; 508232153Smm u16.length = dest->length << 1;; 509232153Smm u16.buffer_length = dest->buffer_length; 510238856Smm if (sc->flag & SCONV_NORMALIZATION_C) 511238856Smm ret = archive_string_normalize_C(&u16, s, count, sc); 512238856Smm else 513238856Smm ret = archive_string_normalize_D(&u16, s, count, sc); 514232153Smm dest->s = (wchar_t *)u16.s; 515232153Smm dest->length = u16.length >> 1; 516232153Smm dest->buffer_length = u16.buffer_length; 517232153Smm sc->flag = saved_flag;/* restore the saved flag. */ 518232153Smm return (ret); 519232153Smm } else if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) { 520248616Smm count = (int)utf16nbytes(s, length); 521232153Smm count >>= 1; /* to be WCS length */ 522232153Smm /* Allocate memory for WCS. */ 523232153Smm if (NULL == archive_wstring_ensure(dest, 524232153Smm dest->length + count + 1)) 525232153Smm return (-1); 526238856Smm wmemcpy(dest->s + dest->length, (const wchar_t *)s, count); 527232153Smm if ((sc->flag & SCONV_FROM_UTF16BE) && !is_big_endian()) { 528232153Smm uint16_t *u16 = (uint16_t *)(dest->s + dest->length); 529232153Smm int b; 530232153Smm for (b = 0; b < count; b++) { 531232153Smm uint16_t val = archive_le16dec(u16+b); 532232153Smm archive_be16enc(u16+b, val); 533232153Smm } 534232153Smm } else if ((sc->flag & SCONV_FROM_UTF16LE) && is_big_endian()) { 535232153Smm uint16_t *u16 = (uint16_t *)(dest->s + dest->length); 536232153Smm int b; 537232153Smm for (b = 0; b < count; b++) { 538232153Smm uint16_t val = archive_be16dec(u16+b); 539232153Smm archive_le16enc(u16+b, val); 540232153Smm } 541232153Smm } 542232153Smm } else { 543232153Smm DWORD mbflag; 544238856Smm size_t buffsize; 545232153Smm 546232153Smm if (sc == NULL) 547232153Smm mbflag = 0; 548232153Smm else if (sc->flag & SCONV_FROM_CHARSET) { 549232153Smm /* Do not trust the length which comes from 550232153Smm * an archive file. */ 551232153Smm length = mbsnbytes(s, length); 552232153Smm mbflag = 0; 553232153Smm } else 554232153Smm mbflag = MB_PRECOMPOSED; 555232153Smm 556238856Smm buffsize = dest->length + length + 1; 557238856Smm do { 558238856Smm /* Allocate memory for WCS. */ 559238856Smm if (NULL == archive_wstring_ensure(dest, buffsize)) 560232153Smm return (-1); 561238856Smm /* Convert MBS to WCS. */ 562238856Smm count = MultiByteToWideChar(from_cp, 563248616Smm mbflag, s, (int)length, dest->s + dest->length, 564248616Smm (int)(dest->buffer_length >> 1) -1); 565238856Smm if (count == 0 && 566238856Smm GetLastError() == ERROR_INSUFFICIENT_BUFFER) { 567238856Smm /* Expand the WCS buffer. */ 568238856Smm buffsize = dest->buffer_length << 1; 569238856Smm continue; 570232153Smm } 571238856Smm if (count == 0 && length != 0) 572238856Smm ret = -1; 573302294Smm break; 574302294Smm } while (1); 575232153Smm } 576232153Smm dest->length += count; 577232153Smm dest->s[dest->length] = L'\0'; 578232153Smm return (ret); 579232153Smm} 580232153Smm 581232153Smm#else 582232153Smm 583232153Smm/* 584232153Smm * Convert MBS to WCS. 585232153Smm * Note: returns -1 if conversion fails. 586232153Smm */ 587232153Smmint 588232153Smmarchive_wstring_append_from_mbs(struct archive_wstring *dest, 589232153Smm const char *p, size_t len) 590232153Smm{ 591232153Smm size_t r; 592238856Smm int ret_val = 0; 593232153Smm /* 594232153Smm * No single byte will be more than one wide character, 595232153Smm * so this length estimate will always be big enough. 596232153Smm */ 597358088Smm // size_t wcs_length = len; 598232153Smm size_t mbs_length = len; 599232153Smm const char *mbs = p; 600232153Smm wchar_t *wcs; 601232153Smm#if HAVE_MBRTOWC 602232153Smm mbstate_t shift_state; 603232153Smm 604232153Smm memset(&shift_state, 0, sizeof(shift_state)); 605232153Smm#endif 606358088Smm /* 607358088Smm * As we decided to have wcs_length == mbs_length == len 608358088Smm * we can use len here instead of wcs_length 609358088Smm */ 610358088Smm if (NULL == archive_wstring_ensure(dest, dest->length + len + 1)) 611238856Smm return (-1); 612232153Smm wcs = dest->s + dest->length; 613232153Smm /* 614232153Smm * We cannot use mbsrtowcs/mbstowcs here because those may convert 615313570Smm * extra MBS when strlen(p) > len and one wide character consists of 616232153Smm * multi bytes. 617232153Smm */ 618238856Smm while (*mbs && mbs_length > 0) { 619358088Smm /* 620358088Smm * The buffer we allocated is always big enough. 621358088Smm * Keep this code path in a comment if we decide to choose 622358088Smm * smaller wcs_length in the future 623358088Smm */ 624358088Smm/* 625238856Smm if (wcs_length == 0) { 626238856Smm dest->length = wcs - dest->s; 627238856Smm dest->s[dest->length] = L'\0'; 628238856Smm wcs_length = mbs_length; 629238856Smm if (NULL == archive_wstring_ensure(dest, 630238856Smm dest->length + wcs_length + 1)) 631238856Smm return (-1); 632238856Smm wcs = dest->s + dest->length; 633238856Smm } 634358088Smm*/ 635232153Smm#if HAVE_MBRTOWC 636358088Smm r = mbrtowc(wcs, mbs, mbs_length, &shift_state); 637232153Smm#else 638358088Smm r = mbtowc(wcs, mbs, mbs_length); 639232153Smm#endif 640232153Smm if (r == (size_t)-1 || r == (size_t)-2) { 641238856Smm ret_val = -1; 642358088Smm break; 643232153Smm } 644232153Smm if (r == 0 || r > mbs_length) 645232153Smm break; 646232153Smm wcs++; 647358088Smm // wcs_length--; 648232153Smm mbs += r; 649232153Smm mbs_length -= r; 650232153Smm } 651232153Smm dest->length = wcs - dest->s; 652232153Smm dest->s[dest->length] = L'\0'; 653238856Smm return (ret_val); 654232153Smm} 655232153Smm 656232153Smm#endif 657232153Smm 658232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 659232153Smm 660232153Smm/* 661232153Smm * WCS ==> MBS. 662232153Smm * Note: returns -1 if conversion fails. 663232153Smm * 664232153Smm * Win32 builds use WideCharToMultiByte from the Windows API. 665232153Smm * (Maybe Cygwin should too? WideCharToMultiByte will know a 666232153Smm * lot more about local character encodings than the wcrtomb() 667232153Smm * wrapper is going to know.) 668232153Smm */ 669232153Smmint 670232153Smmarchive_string_append_from_wcs(struct archive_string *as, 671232153Smm const wchar_t *w, size_t len) 672232153Smm{ 673238856Smm return archive_string_append_from_wcs_in_codepage(as, w, len, NULL); 674232153Smm} 675232153Smm 676232153Smmstatic int 677232153Smmarchive_string_append_from_wcs_in_codepage(struct archive_string *as, 678232153Smm const wchar_t *ws, size_t len, struct archive_string_conv *sc) 679232153Smm{ 680232153Smm BOOL defchar_used, *dp; 681232153Smm int count, ret = 0; 682232153Smm UINT to_cp; 683232153Smm int wslen = (int)len; 684232153Smm 685232153Smm if (sc != NULL) 686232153Smm to_cp = sc->to_cp; 687232153Smm else 688232153Smm to_cp = get_current_codepage(); 689232153Smm 690232153Smm if (to_cp == CP_C_LOCALE) { 691232153Smm /* 692353376Smm * "C" locale special processing. 693232153Smm */ 694232153Smm const wchar_t *wp = ws; 695232153Smm char *p; 696232153Smm 697232153Smm if (NULL == archive_string_ensure(as, 698232153Smm as->length + wslen +1)) 699232153Smm return (-1); 700232153Smm p = as->s + as->length; 701232153Smm count = 0; 702232153Smm defchar_used = 0; 703232153Smm while (count < wslen && *wp) { 704232153Smm if (*wp > 255) { 705232153Smm *p++ = '?'; 706232153Smm wp++; 707232153Smm defchar_used = 1; 708232153Smm } else 709232153Smm *p++ = (char)*wp++; 710232153Smm count++; 711232153Smm } 712232153Smm } else if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) { 713232153Smm uint16_t *u16; 714232153Smm 715232153Smm if (NULL == 716232153Smm archive_string_ensure(as, as->length + len * 2 + 2)) 717232153Smm return (-1); 718232153Smm u16 = (uint16_t *)(as->s + as->length); 719232153Smm count = 0; 720232153Smm defchar_used = 0; 721232153Smm if (sc->flag & SCONV_TO_UTF16BE) { 722232153Smm while (count < (int)len && *ws) { 723232153Smm archive_be16enc(u16+count, *ws); 724232153Smm ws++; 725232153Smm count++; 726232153Smm } 727232153Smm } else { 728232153Smm while (count < (int)len && *ws) { 729232153Smm archive_le16enc(u16+count, *ws); 730232153Smm ws++; 731232153Smm count++; 732232153Smm } 733232153Smm } 734232153Smm count <<= 1; /* to be byte size */ 735232153Smm } else { 736232153Smm /* Make sure the MBS buffer has plenty to set. */ 737232153Smm if (NULL == 738232153Smm archive_string_ensure(as, as->length + len * 2 + 1)) 739232153Smm return (-1); 740232153Smm do { 741232153Smm defchar_used = 0; 742232153Smm if (to_cp == CP_UTF8 || sc == NULL) 743232153Smm dp = NULL; 744232153Smm else 745232153Smm dp = &defchar_used; 746232153Smm count = WideCharToMultiByte(to_cp, 0, ws, wslen, 747358088Smm as->s + as->length, 748358088Smm (int)as->buffer_length - as->length - 1, NULL, dp); 749232153Smm if (count == 0 && 750232153Smm GetLastError() == ERROR_INSUFFICIENT_BUFFER) { 751232153Smm /* Expand the MBS buffer and retry. */ 752232153Smm if (NULL == archive_string_ensure(as, 753232153Smm as->buffer_length + len)) 754232153Smm return (-1); 755232153Smm continue; 756232153Smm } 757232153Smm if (count == 0) 758232153Smm ret = -1; 759299529Smm break; 760299529Smm } while (1); 761232153Smm } 762232153Smm as->length += count; 763232153Smm as->s[as->length] = '\0'; 764232153Smm return (defchar_used?-1:ret); 765232153Smm} 766232153Smm 767232153Smm#elif defined(HAVE_WCTOMB) || defined(HAVE_WCRTOMB) 768232153Smm 769232153Smm/* 770232153Smm * Translates a wide character string into current locale character set 771232153Smm * and appends to the archive_string. Note: returns -1 if conversion 772232153Smm * fails. 773232153Smm */ 774232153Smmint 775232153Smmarchive_string_append_from_wcs(struct archive_string *as, 776232153Smm const wchar_t *w, size_t len) 777232153Smm{ 778232153Smm /* We cannot use the standard wcstombs() here because it 779232153Smm * cannot tell us how big the output buffer should be. So 780232153Smm * I've built a loop around wcrtomb() or wctomb() that 781232153Smm * converts a character at a time and resizes the string as 782232153Smm * needed. We prefer wcrtomb() when it's available because 783232153Smm * it's thread-safe. */ 784232153Smm int n, ret_val = 0; 785228753Smm char *p; 786232153Smm char *end; 787232153Smm#if HAVE_WCRTOMB 788232153Smm mbstate_t shift_state; 789228753Smm 790232153Smm memset(&shift_state, 0, sizeof(shift_state)); 791232153Smm#else 792232153Smm /* Clear the shift state before starting. */ 793232153Smm wctomb(NULL, L'\0'); 794232153Smm#endif 795228753Smm /* 796232153Smm * Allocate buffer for MBS. 797232153Smm * We need this allocation here since it is possible that 798232153Smm * as->s is still NULL. 799228753Smm */ 800232153Smm if (archive_string_ensure(as, as->length + len + 1) == NULL) 801238856Smm return (-1); 802232153Smm 803232153Smm p = as->s + as->length; 804232153Smm end = as->s + as->buffer_length - MB_CUR_MAX -1; 805232153Smm while (*w != L'\0' && len > 0) { 806232153Smm if (p >= end) { 807232153Smm as->length = p - as->s; 808232153Smm as->s[as->length] = '\0'; 809232153Smm /* Re-allocate buffer for MBS. */ 810232153Smm if (archive_string_ensure(as, 811358088Smm as->length + max(len * 2, 812358088Smm (size_t)MB_CUR_MAX) + 1) == NULL) 813238856Smm return (-1); 814232153Smm p = as->s + as->length; 815232153Smm end = as->s + as->buffer_length - MB_CUR_MAX -1; 816228753Smm } 817232153Smm#if HAVE_WCRTOMB 818232153Smm n = wcrtomb(p, *w++, &shift_state); 819232153Smm#else 820232153Smm n = wctomb(p, *w++); 821232153Smm#endif 822232153Smm if (n == -1) { 823232153Smm if (errno == EILSEQ) { 824232153Smm /* Skip an illegal wide char. */ 825232153Smm *p++ = '?'; 826232153Smm ret_val = -1; 827232153Smm } else { 828232153Smm ret_val = -1; 829232153Smm break; 830232153Smm } 831232153Smm } else 832232153Smm p += n; 833232153Smm len--; 834232153Smm } 835232153Smm as->length = p - as->s; 836232153Smm as->s[as->length] = '\0'; 837232153Smm return (ret_val); 838232153Smm} 839232153Smm 840232153Smm#else /* HAVE_WCTOMB || HAVE_WCRTOMB */ 841232153Smm 842232153Smm/* 843232153Smm * TODO: Test if __STDC_ISO_10646__ is defined. 844232153Smm * Non-Windows uses ISO C wcrtomb() or wctomb() to perform the conversion 845232153Smm * one character at a time. If a non-Windows platform doesn't have 846232153Smm * either of these, fall back to the built-in UTF8 conversion. 847232153Smm */ 848232153Smmint 849232153Smmarchive_string_append_from_wcs(struct archive_string *as, 850232153Smm const wchar_t *w, size_t len) 851232153Smm{ 852232153Smm (void)as;/* UNUSED */ 853232153Smm (void)w;/* UNUSED */ 854232153Smm (void)len;/* UNUSED */ 855238856Smm errno = ENOSYS; 856232153Smm return (-1); 857232153Smm} 858232153Smm 859232153Smm#endif /* HAVE_WCTOMB || HAVE_WCRTOMB */ 860232153Smm 861232153Smm/* 862232153Smm * Find a string conversion object by a pair of 'from' charset name 863232153Smm * and 'to' charset name from an archive object. 864232153Smm * Return NULL if not found. 865232153Smm */ 866232153Smmstatic struct archive_string_conv * 867232153Smmfind_sconv_object(struct archive *a, const char *fc, const char *tc) 868232153Smm{ 869232153Smm struct archive_string_conv *sc; 870232153Smm 871232153Smm if (a == NULL) 872232153Smm return (NULL); 873232153Smm 874232153Smm for (sc = a->sconv; sc != NULL; sc = sc->next) { 875232153Smm if (strcmp(sc->from_charset, fc) == 0 && 876232153Smm strcmp(sc->to_charset, tc) == 0) 877232153Smm break; 878232153Smm } 879232153Smm return (sc); 880232153Smm} 881232153Smm 882232153Smm/* 883232153Smm * Register a string object to an archive object. 884232153Smm */ 885232153Smmstatic void 886232153Smmadd_sconv_object(struct archive *a, struct archive_string_conv *sc) 887232153Smm{ 888232153Smm struct archive_string_conv **psc; 889232153Smm 890232153Smm /* Add a new sconv to sconv list. */ 891232153Smm psc = &(a->sconv); 892232153Smm while (*psc != NULL) 893232153Smm psc = &((*psc)->next); 894232153Smm *psc = sc; 895232153Smm} 896232153Smm 897232153Smmstatic void 898232153Smmadd_converter(struct archive_string_conv *sc, int (*converter) 899232153Smm (struct archive_string *, const void *, size_t, 900232153Smm struct archive_string_conv *)) 901232153Smm{ 902232153Smm if (sc == NULL || sc->nconverter >= 2) 903353376Smm __archive_errx(1, "Programming error"); 904232153Smm sc->converter[sc->nconverter++] = converter; 905232153Smm} 906232153Smm 907232153Smmstatic void 908232153Smmsetup_converter(struct archive_string_conv *sc) 909232153Smm{ 910232153Smm 911232153Smm /* Reset. */ 912232153Smm sc->nconverter = 0; 913232153Smm 914232153Smm /* 915232153Smm * Perform special sequence for the incorrect UTF-8 filenames 916232153Smm * made by libarchive2.x. 917232153Smm */ 918232153Smm if (sc->flag & SCONV_UTF8_LIBARCHIVE_2) { 919232153Smm add_converter(sc, strncat_from_utf8_libarchive2); 920232153Smm return; 921232153Smm } 922232153Smm 923232153Smm /* 924232153Smm * Convert a string to UTF-16BE/LE. 925232153Smm */ 926232153Smm if (sc->flag & SCONV_TO_UTF16) { 927232153Smm /* 928232153Smm * If the current locale is UTF-8, we can translate 929232153Smm * a UTF-8 string into a UTF-16BE string. 930232153Smm */ 931232153Smm if (sc->flag & SCONV_FROM_UTF8) { 932232153Smm add_converter(sc, archive_string_append_unicode); 933232153Smm return; 934228753Smm } 935232153Smm 936232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 937232153Smm if (sc->flag & SCONV_WIN_CP) { 938232153Smm if (sc->flag & SCONV_TO_UTF16BE) 939232153Smm add_converter(sc, win_strncat_to_utf16be); 940232153Smm else 941232153Smm add_converter(sc, win_strncat_to_utf16le); 942232153Smm return; 943232153Smm } 944232153Smm#endif 945232153Smm 946232153Smm#if defined(HAVE_ICONV) 947232153Smm if (sc->cd != (iconv_t)-1) { 948232153Smm add_converter(sc, iconv_strncat_in_locale); 949232153Smm return; 950232153Smm } 951232153Smm#endif 952232153Smm 953232153Smm if (sc->flag & SCONV_BEST_EFFORT) { 954232153Smm if (sc->flag & SCONV_TO_UTF16BE) 955238856Smm add_converter(sc, 956238856Smm best_effort_strncat_to_utf16be); 957232153Smm else 958238856Smm add_converter(sc, 959238856Smm best_effort_strncat_to_utf16le); 960232153Smm } else 961232153Smm /* Make sure we have no converter. */ 962232153Smm sc->nconverter = 0; 963232153Smm return; 964232153Smm } 965232153Smm 966232153Smm /* 967232153Smm * Convert a string from UTF-16BE/LE. 968232153Smm */ 969232153Smm if (sc->flag & SCONV_FROM_UTF16) { 970232153Smm /* 971232153Smm * At least we should normalize a UTF-16BE string. 972232153Smm */ 973232153Smm if (sc->flag & SCONV_NORMALIZATION_D) 974232153Smm add_converter(sc,archive_string_normalize_D); 975238856Smm else if (sc->flag & SCONV_NORMALIZATION_C) 976232153Smm add_converter(sc, archive_string_normalize_C); 977232153Smm 978232153Smm if (sc->flag & SCONV_TO_UTF8) { 979232153Smm /* 980232153Smm * If the current locale is UTF-8, we can translate 981232153Smm * a UTF-16BE/LE string into a UTF-8 string directly. 982232153Smm */ 983232153Smm if (!(sc->flag & 984232153Smm (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) 985232153Smm add_converter(sc, 986232153Smm archive_string_append_unicode); 987232153Smm return; 988232153Smm } 989232153Smm 990232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 991232153Smm if (sc->flag & SCONV_WIN_CP) { 992232153Smm if (sc->flag & SCONV_FROM_UTF16BE) 993232153Smm add_converter(sc, win_strncat_from_utf16be); 994232153Smm else 995232153Smm add_converter(sc, win_strncat_from_utf16le); 996232153Smm return; 997232153Smm } 998232153Smm#endif 999232153Smm 1000232153Smm#if defined(HAVE_ICONV) 1001232153Smm if (sc->cd != (iconv_t)-1) { 1002232153Smm add_converter(sc, iconv_strncat_in_locale); 1003232153Smm return; 1004232153Smm } 1005232153Smm#endif 1006232153Smm 1007232153Smm if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) 1008232153Smm == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) 1009232153Smm add_converter(sc, best_effort_strncat_from_utf16be); 1010232153Smm else if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) 1011232153Smm == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) 1012232153Smm add_converter(sc, best_effort_strncat_from_utf16le); 1013232153Smm else 1014232153Smm /* Make sure we have no converter. */ 1015232153Smm sc->nconverter = 0; 1016232153Smm return; 1017232153Smm } 1018232153Smm 1019232153Smm if (sc->flag & SCONV_FROM_UTF8) { 1020232153Smm /* 1021232153Smm * At least we should normalize a UTF-8 string. 1022232153Smm */ 1023232153Smm if (sc->flag & SCONV_NORMALIZATION_D) 1024232153Smm add_converter(sc,archive_string_normalize_D); 1025238856Smm else if (sc->flag & SCONV_NORMALIZATION_C) 1026232153Smm add_converter(sc, archive_string_normalize_C); 1027232153Smm 1028232153Smm /* 1029232153Smm * Copy UTF-8 string with a check of CESU-8. 1030232153Smm * Apparently, iconv does not check surrogate pairs in UTF-8 1031232153Smm * when both from-charset and to-charset are UTF-8, and then 1032232153Smm * we use our UTF-8 copy code. 1033232153Smm */ 1034232153Smm if (sc->flag & SCONV_TO_UTF8) { 1035232153Smm /* 1036232153Smm * If the current locale is UTF-8, we can translate 1037232153Smm * a UTF-16BE string into a UTF-8 string directly. 1038232153Smm */ 1039232153Smm if (!(sc->flag & 1040232153Smm (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) 1041232153Smm add_converter(sc, strncat_from_utf8_to_utf8); 1042232153Smm return; 1043232153Smm } 1044232153Smm } 1045232153Smm 1046232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 1047232153Smm /* 1048232153Smm * On Windows we can use Windows API for a string conversion. 1049232153Smm */ 1050232153Smm if (sc->flag & SCONV_WIN_CP) { 1051232153Smm add_converter(sc, strncat_in_codepage); 1052232153Smm return; 1053232153Smm } 1054232153Smm#endif 1055232153Smm 1056232153Smm#if HAVE_ICONV 1057232153Smm if (sc->cd != (iconv_t)-1) { 1058232153Smm add_converter(sc, iconv_strncat_in_locale); 1059238856Smm /* 1060238856Smm * iconv generally does not support UTF-8-MAC and so 1061238856Smm * we have to the output of iconv from NFC to NFD if 1062238856Smm * need. 1063238856Smm */ 1064238856Smm if ((sc->flag & SCONV_FROM_CHARSET) && 1065238856Smm (sc->flag & SCONV_TO_UTF8)) { 1066238856Smm if (sc->flag & SCONV_NORMALIZATION_D) 1067238856Smm add_converter(sc, archive_string_normalize_D); 1068238856Smm } 1069232153Smm return; 1070232153Smm } 1071232153Smm#endif 1072232153Smm 1073232153Smm /* 1074232153Smm * Try conversion in the best effort or no conversion. 1075232153Smm */ 1076232153Smm if ((sc->flag & SCONV_BEST_EFFORT) || sc->same) 1077232153Smm add_converter(sc, best_effort_strncat_in_locale); 1078232153Smm else 1079232153Smm /* Make sure we have no converter. */ 1080232153Smm sc->nconverter = 0; 1081232153Smm} 1082232153Smm 1083232153Smm/* 1084232153Smm * Return canonicalized charset-name but this supports just UTF-8, UTF-16BE 1085232153Smm * and CP932 which are referenced in create_sconv_object(). 1086232153Smm */ 1087232153Smmstatic const char * 1088232153Smmcanonical_charset_name(const char *charset) 1089232153Smm{ 1090232153Smm char cs[16]; 1091232153Smm char *p; 1092232153Smm const char *s; 1093232153Smm 1094232153Smm if (charset == NULL || charset[0] == '\0' 1095232153Smm || strlen(charset) > 15) 1096232153Smm return (charset); 1097232153Smm 1098232153Smm /* Copy name to uppercase. */ 1099232153Smm p = cs; 1100232153Smm s = charset; 1101232153Smm while (*s) { 1102232153Smm char c = *s++; 1103232153Smm if (c >= 'a' && c <= 'z') 1104232153Smm c -= 'a' - 'A'; 1105232153Smm *p++ = c; 1106232153Smm } 1107232153Smm *p++ = '\0'; 1108232153Smm 1109232153Smm if (strcmp(cs, "UTF-8") == 0 || 1110232153Smm strcmp(cs, "UTF8") == 0) 1111232153Smm return ("UTF-8"); 1112232153Smm if (strcmp(cs, "UTF-16BE") == 0 || 1113232153Smm strcmp(cs, "UTF16BE") == 0) 1114232153Smm return ("UTF-16BE"); 1115232153Smm if (strcmp(cs, "UTF-16LE") == 0 || 1116232153Smm strcmp(cs, "UTF16LE") == 0) 1117232153Smm return ("UTF-16LE"); 1118232153Smm if (strcmp(cs, "CP932") == 0) 1119232153Smm return ("CP932"); 1120232153Smm return (charset); 1121232153Smm} 1122232153Smm 1123232153Smm/* 1124232153Smm * Create a string conversion object. 1125232153Smm */ 1126232153Smmstatic struct archive_string_conv * 1127232153Smmcreate_sconv_object(const char *fc, const char *tc, 1128232153Smm unsigned current_codepage, int flag) 1129232153Smm{ 1130232153Smm struct archive_string_conv *sc; 1131232153Smm 1132232153Smm sc = calloc(1, sizeof(*sc)); 1133232153Smm if (sc == NULL) 1134232153Smm return (NULL); 1135232153Smm sc->next = NULL; 1136232153Smm sc->from_charset = strdup(fc); 1137232153Smm if (sc->from_charset == NULL) { 1138232153Smm free(sc); 1139232153Smm return (NULL); 1140232153Smm } 1141232153Smm sc->to_charset = strdup(tc); 1142232153Smm if (sc->to_charset == NULL) { 1143248616Smm free(sc->from_charset); 1144232153Smm free(sc); 1145232153Smm return (NULL); 1146232153Smm } 1147232153Smm archive_string_init(&sc->utftmp); 1148232153Smm 1149232153Smm if (flag & SCONV_TO_CHARSET) { 1150232153Smm /* 1151232153Smm * Convert characters from the current locale charset to 1152232153Smm * a specified charset. 1153232153Smm */ 1154232153Smm sc->from_cp = current_codepage; 1155232153Smm sc->to_cp = make_codepage_from_charset(tc); 1156232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 1157232153Smm if (IsValidCodePage(sc->to_cp)) 1158232153Smm flag |= SCONV_WIN_CP; 1159232153Smm#endif 1160232153Smm } else if (flag & SCONV_FROM_CHARSET) { 1161232153Smm /* 1162232153Smm * Convert characters from a specified charset to 1163232153Smm * the current locale charset. 1164232153Smm */ 1165232153Smm sc->to_cp = current_codepage; 1166232153Smm sc->from_cp = make_codepage_from_charset(fc); 1167232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 1168232153Smm if (IsValidCodePage(sc->from_cp)) 1169232153Smm flag |= SCONV_WIN_CP; 1170232153Smm#endif 1171232153Smm } 1172232153Smm 1173232153Smm /* 1174232153Smm * Check if "from charset" and "to charset" are the same. 1175232153Smm */ 1176232153Smm if (strcmp(fc, tc) == 0 || 1177232153Smm (sc->from_cp != (unsigned)-1 && sc->from_cp == sc->to_cp)) 1178232153Smm sc->same = 1; 1179232153Smm else 1180232153Smm sc->same = 0; 1181232153Smm 1182232153Smm /* 1183232153Smm * Mark if "from charset" or "to charset" are UTF-8 or UTF-16BE/LE. 1184232153Smm */ 1185232153Smm if (strcmp(tc, "UTF-8") == 0) 1186232153Smm flag |= SCONV_TO_UTF8; 1187232153Smm else if (strcmp(tc, "UTF-16BE") == 0) 1188232153Smm flag |= SCONV_TO_UTF16BE; 1189232153Smm else if (strcmp(tc, "UTF-16LE") == 0) 1190232153Smm flag |= SCONV_TO_UTF16LE; 1191232153Smm if (strcmp(fc, "UTF-8") == 0) 1192232153Smm flag |= SCONV_FROM_UTF8; 1193232153Smm else if (strcmp(fc, "UTF-16BE") == 0) 1194232153Smm flag |= SCONV_FROM_UTF16BE; 1195232153Smm else if (strcmp(fc, "UTF-16LE") == 0) 1196232153Smm flag |= SCONV_FROM_UTF16LE; 1197232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 1198232153Smm if (sc->to_cp == CP_UTF8) 1199232153Smm flag |= SCONV_TO_UTF8; 1200232153Smm else if (sc->to_cp == CP_UTF16BE) 1201232153Smm flag |= SCONV_TO_UTF16BE | SCONV_WIN_CP; 1202232153Smm else if (sc->to_cp == CP_UTF16LE) 1203232153Smm flag |= SCONV_TO_UTF16LE | SCONV_WIN_CP; 1204232153Smm if (sc->from_cp == CP_UTF8) 1205232153Smm flag |= SCONV_FROM_UTF8; 1206232153Smm else if (sc->from_cp == CP_UTF16BE) 1207232153Smm flag |= SCONV_FROM_UTF16BE | SCONV_WIN_CP; 1208232153Smm else if (sc->from_cp == CP_UTF16LE) 1209232153Smm flag |= SCONV_FROM_UTF16LE | SCONV_WIN_CP; 1210232153Smm#endif 1211232153Smm 1212232153Smm /* 1213232153Smm * Set a flag for Unicode NFD. Usually iconv cannot correctly 1214232153Smm * handle it. So we have to translate NFD characters to NFC ones 1215232153Smm * ourselves before iconv handles. Another reason is to prevent 1216232153Smm * that the same sight of two filenames, one is NFC and other 1217232153Smm * is NFD, would be in its directory. 1218232153Smm * On Mac OS X, although its filesystem layer automatically 1219232153Smm * convert filenames to NFD, it would be useful for filename 1220232153Smm * comparing to find out the same filenames that we normalize 1221232153Smm * that to be NFD ourselves. 1222232153Smm */ 1223232153Smm if ((flag & SCONV_FROM_CHARSET) && 1224232153Smm (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8))) { 1225232153Smm#if defined(__APPLE__) 1226238856Smm if (flag & SCONV_TO_UTF8) 1227238856Smm flag |= SCONV_NORMALIZATION_D; 1228238856Smm else 1229232153Smm#endif 1230232153Smm flag |= SCONV_NORMALIZATION_C; 1231232153Smm } 1232238856Smm#if defined(__APPLE__) 1233238856Smm /* 1234238856Smm * In case writing an archive file, make sure that a filename 1235238856Smm * going to be passed to iconv is a Unicode NFC string since 1236238856Smm * a filename in HFS Plus filesystem is a Unicode NFD one and 1237238856Smm * iconv cannot handle it with "UTF-8" charset. It is simpler 1238238856Smm * than a use of "UTF-8-MAC" charset. 1239238856Smm */ 1240238856Smm if ((flag & SCONV_TO_CHARSET) && 1241238856Smm (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && 1242238856Smm !(flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) 1243238856Smm flag |= SCONV_NORMALIZATION_C; 1244238856Smm /* 1245238856Smm * In case reading an archive file. make sure that a filename 1246238856Smm * will be passed to users is a Unicode NFD string in order to 1247238856Smm * correctly compare the filename with other one which comes 1248238856Smm * from HFS Plus filesystem. 1249238856Smm */ 1250238856Smm if ((flag & SCONV_FROM_CHARSET) && 1251238856Smm !(flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && 1252238856Smm (flag & SCONV_TO_UTF8)) 1253238856Smm flag |= SCONV_NORMALIZATION_D; 1254238856Smm#endif 1255232153Smm 1256232153Smm#if defined(HAVE_ICONV) 1257232153Smm sc->cd_w = (iconv_t)-1; 1258232153Smm /* 1259232153Smm * Create an iconv object. 1260232153Smm */ 1261232153Smm if (((flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) && 1262232153Smm (flag & (SCONV_FROM_UTF8 | SCONV_FROM_UTF16))) || 1263232153Smm (flag & SCONV_WIN_CP)) { 1264232153Smm /* This case we won't use iconv. */ 1265232153Smm sc->cd = (iconv_t)-1; 1266232153Smm } else { 1267232153Smm sc->cd = iconv_open(tc, fc); 1268232153Smm if (sc->cd == (iconv_t)-1 && (sc->flag & SCONV_BEST_EFFORT)) { 1269232153Smm /* 1270313570Smm * Unfortunately, all of iconv implements do support 1271232153Smm * "CP932" character-set, so we should use "SJIS" 1272232153Smm * instead if iconv_open failed. 1273232153Smm */ 1274232153Smm if (strcmp(tc, "CP932") == 0) 1275232153Smm sc->cd = iconv_open("SJIS", fc); 1276232153Smm else if (strcmp(fc, "CP932") == 0) 1277232153Smm sc->cd = iconv_open(tc, "SJIS"); 1278232153Smm } 1279232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 1280232153Smm /* 1281232153Smm * archive_mstring on Windows directly convert multi-bytes 1282232153Smm * into archive_wstring in order not to depend on locale 1283313570Smm * so that you can do a I18N programming. This will be 1284232153Smm * used only in archive_mstring_copy_mbs_len_l so far. 1285232153Smm */ 1286232153Smm if (flag & SCONV_FROM_CHARSET) { 1287232153Smm sc->cd_w = iconv_open("UTF-8", fc); 1288232153Smm if (sc->cd_w == (iconv_t)-1 && 1289232153Smm (sc->flag & SCONV_BEST_EFFORT)) { 1290232153Smm if (strcmp(fc, "CP932") == 0) 1291232153Smm sc->cd_w = iconv_open("UTF-8", "SJIS"); 1292232153Smm } 1293232153Smm } 1294232153Smm#endif /* _WIN32 && !__CYGWIN__ */ 1295232153Smm } 1296232153Smm#endif /* HAVE_ICONV */ 1297232153Smm 1298232153Smm sc->flag = flag; 1299232153Smm 1300232153Smm /* 1301238856Smm * Set up converters. 1302232153Smm */ 1303232153Smm setup_converter(sc); 1304232153Smm 1305232153Smm return (sc); 1306232153Smm} 1307232153Smm 1308232153Smm/* 1309232153Smm * Free a string conversion object. 1310232153Smm */ 1311232153Smmstatic void 1312232153Smmfree_sconv_object(struct archive_string_conv *sc) 1313232153Smm{ 1314232153Smm free(sc->from_charset); 1315232153Smm free(sc->to_charset); 1316232153Smm archive_string_free(&sc->utftmp); 1317232153Smm#if HAVE_ICONV 1318232153Smm if (sc->cd != (iconv_t)-1) 1319232153Smm iconv_close(sc->cd); 1320232153Smm if (sc->cd_w != (iconv_t)-1) 1321232153Smm iconv_close(sc->cd_w); 1322232153Smm#endif 1323232153Smm free(sc); 1324232153Smm} 1325232153Smm 1326232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 1327232153Smmstatic unsigned 1328232153Smmmy_atoi(const char *p) 1329232153Smm{ 1330232153Smm unsigned cp; 1331232153Smm 1332232153Smm cp = 0; 1333232153Smm while (*p) { 1334232153Smm if (*p >= '0' && *p <= '9') 1335232153Smm cp = cp * 10 + (*p - '0'); 1336232153Smm else 1337232153Smm return (-1); 1338232153Smm p++; 1339232153Smm } 1340232153Smm return (cp); 1341232153Smm} 1342232153Smm 1343232153Smm/* 1344232153Smm * Translate Charset name (as used by iconv) into CodePage (as used by Windows) 1345232153Smm * Return -1 if failed. 1346232153Smm * 1347232153Smm * Note: This translation code may be insufficient. 1348232153Smm */ 1349232153Smmstatic struct charset { 1350232153Smm const char *name; 1351232153Smm unsigned cp; 1352232153Smm} charsets[] = { 1353232153Smm /* MUST BE SORTED! */ 1354232153Smm {"ASCII", 1252}, 1355232153Smm {"ASMO-708", 708}, 1356232153Smm {"BIG5", 950}, 1357232153Smm {"CHINESE", 936}, 1358232153Smm {"CP367", 1252}, 1359232153Smm {"CP819", 1252}, 1360232153Smm {"CP1025", 21025}, 1361232153Smm {"DOS-720", 720}, 1362232153Smm {"DOS-862", 862}, 1363232153Smm {"EUC-CN", 51936}, 1364232153Smm {"EUC-JP", 51932}, 1365232153Smm {"EUC-KR", 949}, 1366232153Smm {"EUCCN", 51936}, 1367232153Smm {"EUCJP", 51932}, 1368232153Smm {"EUCKR", 949}, 1369232153Smm {"GB18030", 54936}, 1370232153Smm {"GB2312", 936}, 1371232153Smm {"HEBREW", 1255}, 1372232153Smm {"HZ-GB-2312", 52936}, 1373232153Smm {"IBM273", 20273}, 1374232153Smm {"IBM277", 20277}, 1375232153Smm {"IBM278", 20278}, 1376232153Smm {"IBM280", 20280}, 1377232153Smm {"IBM284", 20284}, 1378232153Smm {"IBM285", 20285}, 1379232153Smm {"IBM290", 20290}, 1380232153Smm {"IBM297", 20297}, 1381232153Smm {"IBM367", 1252}, 1382232153Smm {"IBM420", 20420}, 1383232153Smm {"IBM423", 20423}, 1384232153Smm {"IBM424", 20424}, 1385232153Smm {"IBM819", 1252}, 1386232153Smm {"IBM871", 20871}, 1387232153Smm {"IBM880", 20880}, 1388232153Smm {"IBM905", 20905}, 1389232153Smm {"IBM924", 20924}, 1390232153Smm {"ISO-8859-1", 28591}, 1391232153Smm {"ISO-8859-13", 28603}, 1392232153Smm {"ISO-8859-15", 28605}, 1393232153Smm {"ISO-8859-2", 28592}, 1394232153Smm {"ISO-8859-3", 28593}, 1395232153Smm {"ISO-8859-4", 28594}, 1396232153Smm {"ISO-8859-5", 28595}, 1397232153Smm {"ISO-8859-6", 28596}, 1398232153Smm {"ISO-8859-7", 28597}, 1399232153Smm {"ISO-8859-8", 28598}, 1400232153Smm {"ISO-8859-9", 28599}, 1401232153Smm {"ISO8859-1", 28591}, 1402232153Smm {"ISO8859-13", 28603}, 1403232153Smm {"ISO8859-15", 28605}, 1404232153Smm {"ISO8859-2", 28592}, 1405232153Smm {"ISO8859-3", 28593}, 1406232153Smm {"ISO8859-4", 28594}, 1407232153Smm {"ISO8859-5", 28595}, 1408232153Smm {"ISO8859-6", 28596}, 1409232153Smm {"ISO8859-7", 28597}, 1410232153Smm {"ISO8859-8", 28598}, 1411232153Smm {"ISO8859-9", 28599}, 1412232153Smm {"JOHAB", 1361}, 1413232153Smm {"KOI8-R", 20866}, 1414232153Smm {"KOI8-U", 21866}, 1415232153Smm {"KS_C_5601-1987", 949}, 1416232153Smm {"LATIN1", 1252}, 1417232153Smm {"LATIN2", 28592}, 1418232153Smm {"MACINTOSH", 10000}, 1419232153Smm {"SHIFT-JIS", 932}, 1420232153Smm {"SHIFT_JIS", 932}, 1421232153Smm {"SJIS", 932}, 1422232153Smm {"US", 1252}, 1423232153Smm {"US-ASCII", 1252}, 1424232153Smm {"UTF-16", 1200}, 1425232153Smm {"UTF-16BE", 1201}, 1426232153Smm {"UTF-16LE", 1200}, 1427232153Smm {"UTF-8", CP_UTF8}, 1428232153Smm {"X-EUROPA", 29001}, 1429232153Smm {"X-MAC-ARABIC", 10004}, 1430232153Smm {"X-MAC-CE", 10029}, 1431232153Smm {"X-MAC-CHINESEIMP", 10008}, 1432232153Smm {"X-MAC-CHINESETRAD", 10002}, 1433232153Smm {"X-MAC-CROATIAN", 10082}, 1434232153Smm {"X-MAC-CYRILLIC", 10007}, 1435232153Smm {"X-MAC-GREEK", 10006}, 1436232153Smm {"X-MAC-HEBREW", 10005}, 1437232153Smm {"X-MAC-ICELANDIC", 10079}, 1438232153Smm {"X-MAC-JAPANESE", 10001}, 1439232153Smm {"X-MAC-KOREAN", 10003}, 1440232153Smm {"X-MAC-ROMANIAN", 10010}, 1441232153Smm {"X-MAC-THAI", 10021}, 1442232153Smm {"X-MAC-TURKISH", 10081}, 1443232153Smm {"X-MAC-UKRAINIAN", 10017}, 1444232153Smm}; 1445232153Smmstatic unsigned 1446232153Smmmake_codepage_from_charset(const char *charset) 1447232153Smm{ 1448232153Smm char cs[16]; 1449232153Smm char *p; 1450232153Smm unsigned cp; 1451232153Smm int a, b; 1452232153Smm 1453232153Smm if (charset == NULL || strlen(charset) > 15) 1454232153Smm return -1; 1455232153Smm 1456232153Smm /* Copy name to uppercase. */ 1457232153Smm p = cs; 1458232153Smm while (*charset) { 1459232153Smm char c = *charset++; 1460232153Smm if (c >= 'a' && c <= 'z') 1461232153Smm c -= 'a' - 'A'; 1462232153Smm *p++ = c; 1463232153Smm } 1464232153Smm *p++ = '\0'; 1465232153Smm cp = -1; 1466232153Smm 1467232153Smm /* Look it up in the table first, so that we can easily 1468232153Smm * override CP367, which we map to 1252 instead of 367. */ 1469232153Smm a = 0; 1470232153Smm b = sizeof(charsets)/sizeof(charsets[0]); 1471232153Smm while (b > a) { 1472232153Smm int c = (b + a) / 2; 1473232153Smm int r = strcmp(charsets[c].name, cs); 1474232153Smm if (r < 0) 1475232153Smm a = c + 1; 1476232153Smm else if (r > 0) 1477232153Smm b = c; 1478232153Smm else 1479232153Smm return charsets[c].cp; 1480232153Smm } 1481232153Smm 1482232153Smm /* If it's not in the table, try to parse it. */ 1483232153Smm switch (*cs) { 1484232153Smm case 'C': 1485232153Smm if (cs[1] == 'P' && cs[2] >= '0' && cs[2] <= '9') { 1486232153Smm cp = my_atoi(cs + 2); 1487232153Smm } else if (strcmp(cs, "CP_ACP") == 0) 1488232153Smm cp = get_current_codepage(); 1489232153Smm else if (strcmp(cs, "CP_OEMCP") == 0) 1490232153Smm cp = get_current_oemcp(); 1491232153Smm break; 1492232153Smm case 'I': 1493232153Smm if (cs[1] == 'B' && cs[2] == 'M' && 1494232153Smm cs[3] >= '0' && cs[3] <= '9') { 1495232153Smm cp = my_atoi(cs + 3); 1496232153Smm } 1497232153Smm break; 1498232153Smm case 'W': 1499232153Smm if (strncmp(cs, "WINDOWS-", 8) == 0) { 1500232153Smm cp = my_atoi(cs + 8); 1501232153Smm if (cp != 874 && (cp < 1250 || cp > 1258)) 1502232153Smm cp = -1;/* This may invalid code. */ 1503232153Smm } 1504232153Smm break; 1505232153Smm } 1506232153Smm return (cp); 1507232153Smm} 1508232153Smm 1509232153Smm/* 1510232153Smm * Return ANSI Code Page of current locale set by setlocale(). 1511232153Smm */ 1512232153Smmstatic unsigned 1513232153Smmget_current_codepage(void) 1514232153Smm{ 1515232153Smm char *locale, *p; 1516232153Smm unsigned cp; 1517232153Smm 1518232153Smm locale = setlocale(LC_CTYPE, NULL); 1519232153Smm if (locale == NULL) 1520232153Smm return (GetACP()); 1521232153Smm if (locale[0] == 'C' && locale[1] == '\0') 1522232153Smm return (CP_C_LOCALE); 1523232153Smm p = strrchr(locale, '.'); 1524232153Smm if (p == NULL) 1525232153Smm return (GetACP()); 1526346104Smm if (strcmp(p+1, "utf8") == 0) 1527346104Smm return CP_UTF8; 1528232153Smm cp = my_atoi(p+1); 1529346104Smm if ((int)cp <= 0) 1530232153Smm return (GetACP()); 1531232153Smm return (cp); 1532232153Smm} 1533232153Smm 1534232153Smm/* 1535232153Smm * Translation table between Locale Name and ACP/OEMCP. 1536232153Smm */ 1537232153Smmstatic struct { 1538232153Smm unsigned acp; 1539232153Smm unsigned ocp; 1540232153Smm const char *locale; 1541232153Smm} acp_ocp_map[] = { 1542232153Smm { 950, 950, "Chinese_Taiwan" }, 1543232153Smm { 936, 936, "Chinese_People's Republic of China" }, 1544232153Smm { 950, 950, "Chinese_Taiwan" }, 1545232153Smm { 1250, 852, "Czech_Czech Republic" }, 1546232153Smm { 1252, 850, "Danish_Denmark" }, 1547232153Smm { 1252, 850, "Dutch_Netherlands" }, 1548232153Smm { 1252, 850, "Dutch_Belgium" }, 1549232153Smm { 1252, 437, "English_United States" }, 1550232153Smm { 1252, 850, "English_Australia" }, 1551232153Smm { 1252, 850, "English_Canada" }, 1552232153Smm { 1252, 850, "English_New Zealand" }, 1553232153Smm { 1252, 850, "English_United Kingdom" }, 1554232153Smm { 1252, 437, "English_United States" }, 1555232153Smm { 1252, 850, "Finnish_Finland" }, 1556232153Smm { 1252, 850, "French_France" }, 1557232153Smm { 1252, 850, "French_Belgium" }, 1558232153Smm { 1252, 850, "French_Canada" }, 1559232153Smm { 1252, 850, "French_Switzerland" }, 1560232153Smm { 1252, 850, "German_Germany" }, 1561232153Smm { 1252, 850, "German_Austria" }, 1562232153Smm { 1252, 850, "German_Switzerland" }, 1563232153Smm { 1253, 737, "Greek_Greece" }, 1564232153Smm { 1250, 852, "Hungarian_Hungary" }, 1565232153Smm { 1252, 850, "Icelandic_Iceland" }, 1566232153Smm { 1252, 850, "Italian_Italy" }, 1567232153Smm { 1252, 850, "Italian_Switzerland" }, 1568232153Smm { 932, 932, "Japanese_Japan" }, 1569232153Smm { 949, 949, "Korean_Korea" }, 1570232153Smm { 1252, 850, "Norwegian (BokmOl)_Norway" }, 1571232153Smm { 1252, 850, "Norwegian (BokmOl)_Norway" }, 1572232153Smm { 1252, 850, "Norwegian-Nynorsk_Norway" }, 1573232153Smm { 1250, 852, "Polish_Poland" }, 1574232153Smm { 1252, 850, "Portuguese_Portugal" }, 1575232153Smm { 1252, 850, "Portuguese_Brazil" }, 1576232153Smm { 1251, 866, "Russian_Russia" }, 1577232153Smm { 1250, 852, "Slovak_Slovakia" }, 1578232153Smm { 1252, 850, "Spanish_Spain" }, 1579232153Smm { 1252, 850, "Spanish_Mexico" }, 1580232153Smm { 1252, 850, "Spanish_Spain" }, 1581232153Smm { 1252, 850, "Swedish_Sweden" }, 1582232153Smm { 1254, 857, "Turkish_Turkey" }, 1583232153Smm { 0, 0, NULL} 1584232153Smm}; 1585232153Smm 1586232153Smm/* 1587232153Smm * Return OEM Code Page of current locale set by setlocale(). 1588232153Smm */ 1589232153Smmstatic unsigned 1590232153Smmget_current_oemcp(void) 1591232153Smm{ 1592232153Smm int i; 1593232153Smm char *locale, *p; 1594232153Smm size_t len; 1595232153Smm 1596232153Smm locale = setlocale(LC_CTYPE, NULL); 1597232153Smm if (locale == NULL) 1598232153Smm return (GetOEMCP()); 1599232153Smm if (locale[0] == 'C' && locale[1] == '\0') 1600232153Smm return (CP_C_LOCALE); 1601232153Smm 1602232153Smm p = strrchr(locale, '.'); 1603232153Smm if (p == NULL) 1604232153Smm return (GetOEMCP()); 1605232153Smm len = p - locale; 1606232153Smm for (i = 0; acp_ocp_map[i].acp; i++) { 1607232153Smm if (strncmp(acp_ocp_map[i].locale, locale, len) == 0) 1608232153Smm return (acp_ocp_map[i].ocp); 1609232153Smm } 1610232153Smm return (GetOEMCP()); 1611232153Smm} 1612232153Smm#else 1613232153Smm 1614232153Smm/* 1615232153Smm * POSIX platform does not use CodePage. 1616232153Smm */ 1617232153Smm 1618232153Smmstatic unsigned 1619232153Smmget_current_codepage(void) 1620232153Smm{ 1621232153Smm return (-1);/* Unknown */ 1622232153Smm} 1623232153Smmstatic unsigned 1624232153Smmmake_codepage_from_charset(const char *charset) 1625232153Smm{ 1626232153Smm (void)charset; /* UNUSED */ 1627232153Smm return (-1);/* Unknown */ 1628232153Smm} 1629232153Smmstatic unsigned 1630232153Smmget_current_oemcp(void) 1631232153Smm{ 1632232153Smm return (-1);/* Unknown */ 1633232153Smm} 1634232153Smm 1635232153Smm#endif /* defined(_WIN32) && !defined(__CYGWIN__) */ 1636232153Smm 1637232153Smm/* 1638232153Smm * Return a string conversion object. 1639232153Smm */ 1640232153Smmstatic struct archive_string_conv * 1641232153Smmget_sconv_object(struct archive *a, const char *fc, const char *tc, int flag) 1642232153Smm{ 1643232153Smm struct archive_string_conv *sc; 1644232153Smm unsigned current_codepage; 1645232153Smm 1646232153Smm /* Check if we have made the sconv object. */ 1647232153Smm sc = find_sconv_object(a, fc, tc); 1648232153Smm if (sc != NULL) 1649232153Smm return (sc); 1650232153Smm 1651232153Smm if (a == NULL) 1652232153Smm current_codepage = get_current_codepage(); 1653232153Smm else 1654232153Smm current_codepage = a->current_codepage; 1655232153Smm 1656232153Smm sc = create_sconv_object(canonical_charset_name(fc), 1657232153Smm canonical_charset_name(tc), current_codepage, flag); 1658232153Smm if (sc == NULL) { 1659232153Smm if (a != NULL) 1660232153Smm archive_set_error(a, ENOMEM, 1661232153Smm "Could not allocate memory for " 1662232153Smm "a string conversion object"); 1663232153Smm return (NULL); 1664232153Smm } 1665232153Smm 1666232153Smm /* 1667232153Smm * If there is no converter for current string conversion object, 1668232153Smm * we cannot handle this conversion. 1669232153Smm */ 1670232153Smm if (sc->nconverter == 0) { 1671232153Smm if (a != NULL) { 1672232153Smm#if HAVE_ICONV 1673232153Smm archive_set_error(a, ARCHIVE_ERRNO_MISC, 1674232153Smm "iconv_open failed : Cannot handle ``%s''", 1675232153Smm (flag & SCONV_TO_CHARSET)?tc:fc); 1676232153Smm#else 1677232153Smm archive_set_error(a, ARCHIVE_ERRNO_MISC, 1678232153Smm "A character-set conversion not fully supported " 1679232153Smm "on this platform"); 1680232153Smm#endif 1681232153Smm } 1682232153Smm /* Failed; free a sconv object. */ 1683232153Smm free_sconv_object(sc); 1684232153Smm return (NULL); 1685232153Smm } 1686232153Smm 1687232153Smm /* 1688232153Smm * Success! 1689232153Smm */ 1690232153Smm if (a != NULL) 1691232153Smm add_sconv_object(a, sc); 1692232153Smm return (sc); 1693232153Smm} 1694232153Smm 1695232153Smmstatic const char * 1696232153Smmget_current_charset(struct archive *a) 1697232153Smm{ 1698232153Smm const char *cur_charset; 1699232153Smm 1700232153Smm if (a == NULL) 1701232153Smm cur_charset = default_iconv_charset(""); 1702232153Smm else { 1703232153Smm cur_charset = default_iconv_charset(a->current_code); 1704232153Smm if (a->current_code == NULL) { 1705232153Smm a->current_code = strdup(cur_charset); 1706232153Smm a->current_codepage = get_current_codepage(); 1707232153Smm a->current_oemcp = get_current_oemcp(); 1708232153Smm } 1709232153Smm } 1710232153Smm return (cur_charset); 1711232153Smm} 1712232153Smm 1713232153Smm/* 1714232153Smm * Make and Return a string conversion object. 1715232153Smm * Return NULL if the platform does not support the specified conversion 1716232153Smm * and best_effort is 0. 1717232153Smm * If best_effort is set, A string conversion object must be returned 1718232153Smm * unless memory allocation for the object fails, but the conversion 1719232153Smm * might fail when non-ASCII code is found. 1720232153Smm */ 1721232153Smmstruct archive_string_conv * 1722232153Smmarchive_string_conversion_to_charset(struct archive *a, const char *charset, 1723232153Smm int best_effort) 1724232153Smm{ 1725232153Smm int flag = SCONV_TO_CHARSET; 1726232153Smm 1727232153Smm if (best_effort) 1728232153Smm flag |= SCONV_BEST_EFFORT; 1729232153Smm return (get_sconv_object(a, get_current_charset(a), charset, flag)); 1730232153Smm} 1731232153Smm 1732232153Smmstruct archive_string_conv * 1733232153Smmarchive_string_conversion_from_charset(struct archive *a, const char *charset, 1734232153Smm int best_effort) 1735232153Smm{ 1736232153Smm int flag = SCONV_FROM_CHARSET; 1737232153Smm 1738232153Smm if (best_effort) 1739232153Smm flag |= SCONV_BEST_EFFORT; 1740232153Smm return (get_sconv_object(a, charset, get_current_charset(a), flag)); 1741232153Smm} 1742232153Smm 1743232153Smm/* 1744232153Smm * archive_string_default_conversion_*_archive() are provided for Windows 1745232153Smm * platform because other archiver application use CP_OEMCP for 1746232153Smm * MultiByteToWideChar() and WideCharToMultiByte() for the filenames 1747232153Smm * in tar or zip files. But mbstowcs/wcstombs(CRT) usually use CP_ACP 1748232153Smm * unless you use setlocale(LC_ALL, ".OCP")(specify CP_OEMCP). 1749232153Smm * So we should make a string conversion between CP_ACP and CP_OEMCP 1750313570Smm * for compatibility. 1751232153Smm */ 1752232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 1753232153Smmstruct archive_string_conv * 1754232153Smmarchive_string_default_conversion_for_read(struct archive *a) 1755232153Smm{ 1756232153Smm const char *cur_charset = get_current_charset(a); 1757232153Smm char oemcp[16]; 1758232153Smm 1759232153Smm /* NOTE: a check of cur_charset is unneeded but we need 1760232153Smm * that get_current_charset() has been surely called at 1761232153Smm * this time whatever C compiler optimized. */ 1762232153Smm if (cur_charset != NULL && 1763232153Smm (a->current_codepage == CP_C_LOCALE || 1764232153Smm a->current_codepage == a->current_oemcp)) 1765232153Smm return (NULL);/* no conversion. */ 1766232153Smm 1767232153Smm _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); 1768232153Smm /* Make sure a null termination must be set. */ 1769232153Smm oemcp[sizeof(oemcp)-1] = '\0'; 1770232153Smm return (get_sconv_object(a, oemcp, cur_charset, 1771232153Smm SCONV_FROM_CHARSET)); 1772232153Smm} 1773232153Smm 1774232153Smmstruct archive_string_conv * 1775232153Smmarchive_string_default_conversion_for_write(struct archive *a) 1776232153Smm{ 1777232153Smm const char *cur_charset = get_current_charset(a); 1778232153Smm char oemcp[16]; 1779232153Smm 1780232153Smm /* NOTE: a check of cur_charset is unneeded but we need 1781232153Smm * that get_current_charset() has been surely called at 1782232153Smm * this time whatever C compiler optimized. */ 1783232153Smm if (cur_charset != NULL && 1784232153Smm (a->current_codepage == CP_C_LOCALE || 1785232153Smm a->current_codepage == a->current_oemcp)) 1786232153Smm return (NULL);/* no conversion. */ 1787232153Smm 1788232153Smm _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); 1789232153Smm /* Make sure a null termination must be set. */ 1790232153Smm oemcp[sizeof(oemcp)-1] = '\0'; 1791232153Smm return (get_sconv_object(a, cur_charset, oemcp, 1792232153Smm SCONV_TO_CHARSET)); 1793232153Smm} 1794232153Smm#else 1795232153Smmstruct archive_string_conv * 1796232153Smmarchive_string_default_conversion_for_read(struct archive *a) 1797232153Smm{ 1798232153Smm (void)a; /* UNUSED */ 1799232153Smm return (NULL); 1800232153Smm} 1801232153Smm 1802232153Smmstruct archive_string_conv * 1803232153Smmarchive_string_default_conversion_for_write(struct archive *a) 1804232153Smm{ 1805232153Smm (void)a; /* UNUSED */ 1806232153Smm return (NULL); 1807232153Smm} 1808232153Smm#endif 1809232153Smm 1810232153Smm/* 1811232153Smm * Dispose of all character conversion objects in the archive object. 1812232153Smm */ 1813232153Smmvoid 1814232153Smmarchive_string_conversion_free(struct archive *a) 1815232153Smm{ 1816232153Smm struct archive_string_conv *sc; 1817232153Smm struct archive_string_conv *sc_next; 1818232153Smm 1819232153Smm for (sc = a->sconv; sc != NULL; sc = sc_next) { 1820232153Smm sc_next = sc->next; 1821232153Smm free_sconv_object(sc); 1822232153Smm } 1823232153Smm a->sconv = NULL; 1824232153Smm free(a->current_code); 1825232153Smm a->current_code = NULL; 1826232153Smm} 1827232153Smm 1828232153Smm/* 1829232153Smm * Return a conversion charset name. 1830232153Smm */ 1831232153Smmconst char * 1832232153Smmarchive_string_conversion_charset_name(struct archive_string_conv *sc) 1833232153Smm{ 1834232153Smm if (sc->flag & SCONV_TO_CHARSET) 1835232153Smm return (sc->to_charset); 1836232153Smm else 1837232153Smm return (sc->from_charset); 1838232153Smm} 1839232153Smm 1840232153Smm/* 1841232153Smm * Change the behavior of a string conversion. 1842232153Smm */ 1843232153Smmvoid 1844232153Smmarchive_string_conversion_set_opt(struct archive_string_conv *sc, int opt) 1845232153Smm{ 1846232153Smm switch (opt) { 1847232153Smm /* 1848232153Smm * A filename in UTF-8 was made with libarchive 2.x in a wrong 1849232153Smm * assumption that wchar_t was Unicode. 1850232153Smm * This option enables simulating the assumption in order to read 1851311041Smm * that filename correctly. 1852232153Smm */ 1853232153Smm case SCONV_SET_OPT_UTF8_LIBARCHIVE2X: 1854232153Smm#if (defined(_WIN32) && !defined(__CYGWIN__)) \ 1855232153Smm || defined(__STDC_ISO_10646__) || defined(__APPLE__) 1856232153Smm /* 1857232153Smm * Nothing to do for it since wchar_t on these platforms 1858232153Smm * is really Unicode. 1859232153Smm */ 1860232153Smm (void)sc; /* UNUSED */ 1861232153Smm#else 1862232153Smm if ((sc->flag & SCONV_UTF8_LIBARCHIVE_2) == 0) { 1863232153Smm sc->flag |= SCONV_UTF8_LIBARCHIVE_2; 1864238856Smm /* Set up string converters. */ 1865232153Smm setup_converter(sc); 1866232153Smm } 1867232153Smm#endif 1868232153Smm break; 1869238856Smm case SCONV_SET_OPT_NORMALIZATION_C: 1870238856Smm if ((sc->flag & SCONV_NORMALIZATION_C) == 0) { 1871238856Smm sc->flag |= SCONV_NORMALIZATION_C; 1872238856Smm sc->flag &= ~SCONV_NORMALIZATION_D; 1873238856Smm /* Set up string converters. */ 1874238856Smm setup_converter(sc); 1875238856Smm } 1876238856Smm break; 1877238856Smm case SCONV_SET_OPT_NORMALIZATION_D: 1878238856Smm#if defined(HAVE_ICONV) 1879238856Smm /* 1880238856Smm * If iconv will take the string, do not change the 1881238856Smm * setting of the normalization. 1882238856Smm */ 1883238856Smm if (!(sc->flag & SCONV_WIN_CP) && 1884238856Smm (sc->flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && 1885238856Smm !(sc->flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) 1886238856Smm break; 1887238856Smm#endif 1888238856Smm if ((sc->flag & SCONV_NORMALIZATION_D) == 0) { 1889238856Smm sc->flag |= SCONV_NORMALIZATION_D; 1890238856Smm sc->flag &= ~SCONV_NORMALIZATION_C; 1891238856Smm /* Set up string converters. */ 1892238856Smm setup_converter(sc); 1893238856Smm } 1894238856Smm break; 1895232153Smm default: 1896232153Smm break; 1897232153Smm } 1898232153Smm} 1899232153Smm 1900232153Smm/* 1901232153Smm * 1902232153Smm * Copy one archive_string to another in locale conversion. 1903232153Smm * 1904238856Smm * archive_strncat_l(); 1905238856Smm * archive_strncpy_l(); 1906232153Smm * 1907232153Smm */ 1908232153Smm 1909232153Smmstatic size_t 1910232153Smmmbsnbytes(const void *_p, size_t n) 1911232153Smm{ 1912232153Smm size_t s; 1913232153Smm const char *p, *pp; 1914232153Smm 1915232153Smm if (_p == NULL) 1916232153Smm return (0); 1917232153Smm p = (const char *)_p; 1918232153Smm 1919232153Smm /* Like strlen(p), except won't examine positions beyond p[n]. */ 1920232153Smm s = 0; 1921232153Smm pp = p; 1922232153Smm while (s < n && *pp) { 1923232153Smm pp++; 1924232153Smm s++; 1925232153Smm } 1926232153Smm return (s); 1927232153Smm} 1928232153Smm 1929232153Smmstatic size_t 1930232153Smmutf16nbytes(const void *_p, size_t n) 1931232153Smm{ 1932232153Smm size_t s; 1933232153Smm const char *p, *pp; 1934232153Smm 1935232153Smm if (_p == NULL) 1936232153Smm return (0); 1937232153Smm p = (const char *)_p; 1938232153Smm 1939232153Smm /* Like strlen(p), except won't examine positions beyond p[n]. */ 1940232153Smm s = 0; 1941232153Smm pp = p; 1942232153Smm n >>= 1; 1943232153Smm while (s < n && (pp[0] || pp[1])) { 1944232153Smm pp += 2; 1945232153Smm s++; 1946232153Smm } 1947232153Smm return (s<<1); 1948232153Smm} 1949232153Smm 1950232153Smmint 1951238856Smmarchive_strncpy_l(struct archive_string *as, const void *_p, size_t n, 1952232153Smm struct archive_string_conv *sc) 1953232153Smm{ 1954232153Smm as->length = 0; 1955238856Smm return (archive_strncat_l(as, _p, n, sc)); 1956232153Smm} 1957232153Smm 1958232153Smmint 1959238856Smmarchive_strncat_l(struct archive_string *as, const void *_p, size_t n, 1960232153Smm struct archive_string_conv *sc) 1961232153Smm{ 1962232153Smm const void *s; 1963311041Smm size_t length = 0; 1964232153Smm int i, r = 0, r2; 1965232153Smm 1966311041Smm if (_p != NULL && n > 0) { 1967311041Smm if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) 1968311041Smm length = utf16nbytes(_p, n); 1969311041Smm else 1970311041Smm length = mbsnbytes(_p, n); 1971311041Smm } 1972311041Smm 1973232153Smm /* We must allocate memory even if there is no data for conversion 1974232153Smm * or copy. This simulates archive_string_append behavior. */ 1975311041Smm if (length == 0) { 1976232153Smm int tn = 1; 1977232153Smm if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) 1978232153Smm tn = 2; 1979232153Smm if (archive_string_ensure(as, as->length + tn) == NULL) 1980232153Smm return (-1); 1981232153Smm as->s[as->length] = 0; 1982232153Smm if (tn == 2) 1983232153Smm as->s[as->length+1] = 0; 1984232153Smm return (0); 1985232153Smm } 1986232153Smm 1987232153Smm /* 1988232153Smm * If sc is NULL, we just make a copy. 1989232153Smm */ 1990232153Smm if (sc == NULL) { 1991232153Smm if (archive_string_append(as, _p, length) == NULL) 1992232153Smm return (-1);/* No memory */ 1993232153Smm return (0); 1994232153Smm } 1995232153Smm 1996232153Smm s = _p; 1997232153Smm i = 0; 1998232153Smm if (sc->nconverter > 1) { 1999232153Smm sc->utftmp.length = 0; 2000232153Smm r2 = sc->converter[0](&(sc->utftmp), s, length, sc); 2001232153Smm if (r2 != 0 && errno == ENOMEM) 2002232153Smm return (r2); 2003232153Smm if (r > r2) 2004232153Smm r = r2; 2005232153Smm s = sc->utftmp.s; 2006232153Smm length = sc->utftmp.length; 2007232153Smm ++i; 2008232153Smm } 2009232153Smm r2 = sc->converter[i](as, s, length, sc); 2010232153Smm if (r > r2) 2011232153Smm r = r2; 2012232153Smm return (r); 2013232153Smm} 2014232153Smm 2015232153Smm#if HAVE_ICONV 2016232153Smm 2017232153Smm/* 2018311041Smm * Return -1 if conversion fails. 2019232153Smm */ 2020232153Smmstatic int 2021232153Smmiconv_strncat_in_locale(struct archive_string *as, const void *_p, 2022232153Smm size_t length, struct archive_string_conv *sc) 2023232153Smm{ 2024248616Smm ICONV_CONST char *itp; 2025232153Smm size_t remaining; 2026232153Smm iconv_t cd; 2027232153Smm char *outp; 2028232153Smm size_t avail, bs; 2029232153Smm int return_value = 0; /* success */ 2030232153Smm int to_size, from_size; 2031232153Smm 2032232153Smm if (sc->flag & SCONV_TO_UTF16) 2033232153Smm to_size = 2; 2034232153Smm else 2035232153Smm to_size = 1; 2036232153Smm if (sc->flag & SCONV_FROM_UTF16) 2037232153Smm from_size = 2; 2038232153Smm else 2039232153Smm from_size = 1; 2040232153Smm 2041232153Smm if (archive_string_ensure(as, as->length + length*2+to_size) == NULL) 2042232153Smm return (-1); 2043232153Smm 2044232153Smm cd = sc->cd; 2045248616Smm itp = (char *)(uintptr_t)_p; 2046232153Smm remaining = length; 2047232153Smm outp = as->s + as->length; 2048232153Smm avail = as->buffer_length - as->length - to_size; 2049232153Smm while (remaining >= (size_t)from_size) { 2050248616Smm size_t result = iconv(cd, &itp, &remaining, &outp, &avail); 2051232153Smm 2052232153Smm if (result != (size_t)-1) 2053232153Smm break; /* Conversion completed. */ 2054232153Smm 2055232153Smm if (errno == EILSEQ || errno == EINVAL) { 2056232153Smm /* 2057232153Smm * If an output charset is UTF-8 or UTF-16BE/LE, 2058232153Smm * unknown character should be U+FFFD 2059232153Smm * (replacement character). 2060232153Smm */ 2061232153Smm if (sc->flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) { 2062232153Smm size_t rbytes; 2063232153Smm if (sc->flag & SCONV_TO_UTF8) 2064299529Smm rbytes = sizeof(utf8_replacement_char); 2065232153Smm else 2066232153Smm rbytes = 2; 2067232153Smm 2068232153Smm if (avail < rbytes) { 2069232153Smm as->length = outp - as->s; 2070232153Smm bs = as->buffer_length + 2071232153Smm (remaining * to_size) + rbytes; 2072232153Smm if (NULL == 2073232153Smm archive_string_ensure(as, bs)) 2074232153Smm return (-1); 2075232153Smm outp = as->s + as->length; 2076232153Smm avail = as->buffer_length 2077232153Smm - as->length - to_size; 2078232153Smm } 2079232153Smm if (sc->flag & SCONV_TO_UTF8) 2080299529Smm memcpy(outp, utf8_replacement_char, sizeof(utf8_replacement_char)); 2081232153Smm else if (sc->flag & SCONV_TO_UTF16BE) 2082232153Smm archive_be16enc(outp, UNICODE_R_CHAR); 2083232153Smm else 2084232153Smm archive_le16enc(outp, UNICODE_R_CHAR); 2085232153Smm outp += rbytes; 2086232153Smm avail -= rbytes; 2087232153Smm } else { 2088232153Smm /* Skip the illegal input bytes. */ 2089232153Smm *outp++ = '?'; 2090232153Smm avail--; 2091232153Smm } 2092248616Smm itp += from_size; 2093232153Smm remaining -= from_size; 2094232153Smm return_value = -1; /* failure */ 2095228753Smm } else { 2096232153Smm /* E2BIG no output buffer, 2097232153Smm * Increase an output buffer. */ 2098232153Smm as->length = outp - as->s; 2099232153Smm bs = as->buffer_length + remaining * 2; 2100232153Smm if (NULL == archive_string_ensure(as, bs)) 2101232153Smm return (-1); 2102232153Smm outp = as->s + as->length; 2103232153Smm avail = as->buffer_length - as->length - to_size; 2104228753Smm } 2105228753Smm } 2106232153Smm as->length = outp - as->s; 2107232153Smm as->s[as->length] = 0; 2108232153Smm if (to_size == 2) 2109232153Smm as->s[as->length+1] = 0; 2110232153Smm return (return_value); 2111228753Smm} 2112228753Smm 2113232153Smm#endif /* HAVE_ICONV */ 2114232153Smm 2115232153Smm 2116232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 2117232153Smm 2118232153Smm/* 2119232153Smm * Translate a string from a some CodePage to an another CodePage by 2120311041Smm * Windows APIs, and copy the result. Return -1 if conversion fails. 2121232153Smm */ 2122228753Smmstatic int 2123232153Smmstrncat_in_codepage(struct archive_string *as, 2124232153Smm const void *_p, size_t length, struct archive_string_conv *sc) 2125228753Smm{ 2126232153Smm const char *s = (const char *)_p; 2127232153Smm struct archive_wstring aws; 2128232153Smm size_t l; 2129232153Smm int r, saved_flag; 2130228753Smm 2131232153Smm archive_string_init(&aws); 2132232153Smm saved_flag = sc->flag; 2133232153Smm sc->flag &= ~(SCONV_NORMALIZATION_D | SCONV_NORMALIZATION_C); 2134232153Smm r = archive_wstring_append_from_mbs_in_codepage(&aws, s, length, sc); 2135232153Smm sc->flag = saved_flag; 2136232153Smm if (r != 0) { 2137232153Smm archive_wstring_free(&aws); 2138232153Smm if (errno != ENOMEM) 2139232153Smm archive_string_append(as, s, length); 2140232153Smm return (-1); 2141232153Smm } 2142232153Smm 2143232153Smm l = as->length; 2144232153Smm r = archive_string_append_from_wcs_in_codepage( 2145232153Smm as, aws.s, aws.length, sc); 2146232153Smm if (r != 0 && errno != ENOMEM && l == as->length) 2147232153Smm archive_string_append(as, s, length); 2148232153Smm archive_wstring_free(&aws); 2149232153Smm return (r); 2150232153Smm} 2151232153Smm 2152232153Smm/* 2153232153Smm * Test whether MBS ==> WCS is okay. 2154232153Smm */ 2155232153Smmstatic int 2156232153Smminvalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) 2157232153Smm{ 2158232153Smm const char *p = (const char *)_p; 2159232153Smm unsigned codepage; 2160232153Smm DWORD mbflag = MB_ERR_INVALID_CHARS; 2161232153Smm 2162232153Smm if (sc->flag & SCONV_FROM_CHARSET) 2163232153Smm codepage = sc->to_cp; 2164232153Smm else 2165232153Smm codepage = sc->from_cp; 2166232153Smm 2167232153Smm if (codepage == CP_C_LOCALE) 2168232153Smm return (0); 2169232153Smm if (codepage != CP_UTF8) 2170232153Smm mbflag |= MB_PRECOMPOSED; 2171232153Smm 2172248616Smm if (MultiByteToWideChar(codepage, mbflag, p, (int)n, NULL, 0) == 0) 2173232153Smm return (-1); /* Invalid */ 2174232153Smm return (0); /* Okay */ 2175232153Smm} 2176232153Smm 2177232153Smm#else 2178232153Smm 2179232153Smm/* 2180232153Smm * Test whether MBS ==> WCS is okay. 2181232153Smm */ 2182232153Smmstatic int 2183232153Smminvalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) 2184232153Smm{ 2185232153Smm const char *p = (const char *)_p; 2186232153Smm size_t r; 2187232153Smm 2188232153Smm#if HAVE_MBRTOWC 2189232153Smm mbstate_t shift_state; 2190232153Smm 2191232153Smm memset(&shift_state, 0, sizeof(shift_state)); 2192232153Smm#else 2193232153Smm /* Clear the shift state before starting. */ 2194232153Smm mbtowc(NULL, NULL, 0); 2195232153Smm#endif 2196232153Smm while (n) { 2197232153Smm wchar_t wc; 2198232153Smm 2199232153Smm#if HAVE_MBRTOWC 2200232153Smm r = mbrtowc(&wc, p, n, &shift_state); 2201232153Smm#else 2202232153Smm r = mbtowc(&wc, p, n); 2203232153Smm#endif 2204232153Smm if (r == (size_t)-1 || r == (size_t)-2) 2205232153Smm return (-1);/* Invalid. */ 2206232153Smm if (r == 0) 2207232153Smm break; 2208232153Smm p += r; 2209232153Smm n -= r; 2210232153Smm } 2211238856Smm (void)sc; /* UNUSED */ 2212232153Smm return (0); /* All Okey. */ 2213232153Smm} 2214232153Smm 2215232153Smm#endif /* defined(_WIN32) && !defined(__CYGWIN__) */ 2216232153Smm 2217232153Smm/* 2218232153Smm * Basically returns -1 because we cannot make a conversion of charset 2219232153Smm * without iconv but in some cases this would return 0. 2220232153Smm * Returns 0 if all copied characters are ASCII. 2221232153Smm * Returns 0 if both from-locale and to-locale are the same and those 2222232153Smm * can be WCS with no error. 2223232153Smm */ 2224232153Smmstatic int 2225232153Smmbest_effort_strncat_in_locale(struct archive_string *as, const void *_p, 2226232153Smm size_t length, struct archive_string_conv *sc) 2227232153Smm{ 2228232153Smm size_t remaining; 2229248616Smm const uint8_t *itp; 2230232153Smm int return_value = 0; /* success */ 2231232153Smm 2232232153Smm /* 2233232153Smm * If both from-locale and to-locale is the same, this makes a copy. 2234232153Smm * And then this checks all copied MBS can be WCS if so returns 0. 2235232153Smm */ 2236232153Smm if (sc->same) { 2237232153Smm if (archive_string_append(as, _p, length) == NULL) 2238232153Smm return (-1);/* No memory */ 2239232153Smm return (invalid_mbs(_p, length, sc)); 2240232153Smm } 2241232153Smm 2242232153Smm /* 2243232153Smm * If a character is ASCII, this just copies it. If not, this 2244313570Smm * assigns '?' character instead but in UTF-8 locale this assigns 2245232153Smm * byte sequence 0xEF 0xBD 0xBD, which are code point U+FFFD, 2246232153Smm * a Replacement Character in Unicode. 2247232153Smm */ 2248232153Smm 2249232153Smm remaining = length; 2250248616Smm itp = (const uint8_t *)_p; 2251248616Smm while (*itp && remaining > 0) { 2252299529Smm if (*itp > 127) { 2253299529Smm // Non-ASCII: Substitute with suitable replacement 2254299529Smm if (sc->flag & SCONV_TO_UTF8) { 2255299529Smm if (archive_string_append(as, utf8_replacement_char, sizeof(utf8_replacement_char)) == NULL) { 2256299529Smm __archive_errx(1, "Out of memory"); 2257299529Smm } 2258299529Smm } else { 2259299529Smm archive_strappend_char(as, '?'); 2260232153Smm } 2261232153Smm return_value = -1; 2262232153Smm } else { 2263299529Smm archive_strappend_char(as, *itp); 2264232153Smm } 2265299529Smm ++itp; 2266232153Smm } 2267232153Smm return (return_value); 2268232153Smm} 2269232153Smm 2270232153Smm 2271232153Smm/* 2272232153Smm * Unicode conversion functions. 2273232153Smm * - UTF-8 <===> UTF-8 in removing surrogate pairs. 2274232153Smm * - UTF-8 NFD ===> UTF-8 NFC in removing surrogate pairs. 2275232153Smm * - UTF-8 made by libarchive 2.x ===> UTF-8. 2276232153Smm * - UTF-16BE <===> UTF-8. 2277232153Smm * 2278232153Smm */ 2279232153Smm 2280232153Smm/* 2281232153Smm * Utility to convert a single UTF-8 sequence. 2282232153Smm * 2283232153Smm * Usually return used bytes, return used byte in negative value when 2284232153Smm * a unicode character is replaced with U+FFFD. 2285232153Smm * See also http://unicode.org/review/pr-121.html Public Review Issue #121 2286232153Smm * Recommended Practice for Replacement Characters. 2287232153Smm */ 2288232153Smmstatic int 2289232153Smm_utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) 2290232153Smm{ 2291232153Smm static const char utf8_count[256] = { 2292232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ 2293232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ 2294232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ 2295232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ 2296232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ 2297232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ 2298232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ 2299232153Smm 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ 2300232153Smm 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ 2301232153Smm 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ 2302232153Smm 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ 2303232153Smm 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ 2304232153Smm 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ 2305232153Smm 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ 2306232153Smm 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ 2307232153Smm 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ 2308232153Smm }; 2309232153Smm int ch, i; 2310232153Smm int cnt; 2311232153Smm uint32_t wc; 2312232153Smm 2313232153Smm /* Sanity check. */ 2314232153Smm if (n == 0) 2315232153Smm return (0); 2316232153Smm /* 2317228753Smm * Decode 1-4 bytes depending on the value of the first byte. 2318228753Smm */ 2319232153Smm ch = (unsigned char)*s; 2320232153Smm if (ch == 0) 2321228753Smm return (0); /* Standard: return 0 for end-of-string. */ 2322232153Smm cnt = utf8_count[ch]; 2323232153Smm 2324311041Smm /* Invalid sequence or there are not plenty bytes. */ 2325232153Smm if ((int)n < cnt) { 2326248616Smm cnt = (int)n; 2327232153Smm for (i = 1; i < cnt; i++) { 2328232153Smm if ((s[i] & 0xc0) != 0x80) { 2329232153Smm cnt = i; 2330232153Smm break; 2331232153Smm } 2332232153Smm } 2333232153Smm goto invalid_sequence; 2334228753Smm } 2335232153Smm 2336232153Smm /* Make a Unicode code point from a single UTF-8 sequence. */ 2337232153Smm switch (cnt) { 2338232153Smm case 1: /* 1 byte sequence. */ 2339232153Smm *pwc = ch & 0x7f; 2340232153Smm return (cnt); 2341232153Smm case 2: /* 2 bytes sequence. */ 2342232153Smm if ((s[1] & 0xc0) != 0x80) { 2343232153Smm cnt = 1; 2344232153Smm goto invalid_sequence; 2345232153Smm } 2346232153Smm *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); 2347232153Smm return (cnt); 2348232153Smm case 3: /* 3 bytes sequence. */ 2349232153Smm if ((s[1] & 0xc0) != 0x80) { 2350232153Smm cnt = 1; 2351232153Smm goto invalid_sequence; 2352232153Smm } 2353232153Smm if ((s[2] & 0xc0) != 0x80) { 2354232153Smm cnt = 2; 2355232153Smm goto invalid_sequence; 2356232153Smm } 2357232153Smm wc = ((ch & 0x0f) << 12) 2358228753Smm | ((s[1] & 0x3f) << 6) 2359228753Smm | (s[2] & 0x3f); 2360232153Smm if (wc < 0x800) 2361232153Smm goto invalid_sequence;/* Overlong sequence. */ 2362232153Smm break; 2363232153Smm case 4: /* 4 bytes sequence. */ 2364232153Smm if ((s[1] & 0xc0) != 0x80) { 2365232153Smm cnt = 1; 2366232153Smm goto invalid_sequence; 2367232153Smm } 2368232153Smm if ((s[2] & 0xc0) != 0x80) { 2369232153Smm cnt = 2; 2370232153Smm goto invalid_sequence; 2371232153Smm } 2372232153Smm if ((s[3] & 0xc0) != 0x80) { 2373232153Smm cnt = 3; 2374232153Smm goto invalid_sequence; 2375232153Smm } 2376232153Smm wc = ((ch & 0x07) << 18) 2377228753Smm | ((s[1] & 0x3f) << 12) 2378228753Smm | ((s[2] & 0x3f) << 6) 2379228753Smm | (s[3] & 0x3f); 2380232153Smm if (wc < 0x10000) 2381232153Smm goto invalid_sequence;/* Overlong sequence. */ 2382232153Smm break; 2383232153Smm default: /* Others are all invalid sequence. */ 2384232153Smm if (ch == 0xc0 || ch == 0xc1) 2385232153Smm cnt = 2; 2386232153Smm else if (ch >= 0xf5 && ch <= 0xf7) 2387232153Smm cnt = 4; 2388232153Smm else if (ch >= 0xf8 && ch <= 0xfb) 2389232153Smm cnt = 5; 2390232153Smm else if (ch == 0xfc || ch == 0xfd) 2391232153Smm cnt = 6; 2392232153Smm else 2393232153Smm cnt = 1; 2394232153Smm if ((int)n < cnt) 2395248616Smm cnt = (int)n; 2396232153Smm for (i = 1; i < cnt; i++) { 2397232153Smm if ((s[i] & 0xc0) != 0x80) { 2398232153Smm cnt = i; 2399232153Smm break; 2400232153Smm } 2401232153Smm } 2402232153Smm goto invalid_sequence; 2403232153Smm } 2404232153Smm 2405311041Smm /* The code point larger than 0x10FFFF is not legal 2406232153Smm * Unicode values. */ 2407232153Smm if (wc > UNICODE_MAX) 2408232153Smm goto invalid_sequence; 2409232153Smm /* Correctly gets a Unicode, returns used bytes. */ 2410232153Smm *pwc = wc; 2411232153Smm return (cnt); 2412232153Smminvalid_sequence: 2413232153Smm *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ 2414232153Smm return (cnt * -1); 2415232153Smm} 2416232153Smm 2417232153Smmstatic int 2418232153Smmutf8_to_unicode(uint32_t *pwc, const char *s, size_t n) 2419232153Smm{ 2420232153Smm int cnt; 2421232153Smm 2422232153Smm cnt = _utf8_to_unicode(pwc, s, n); 2423311041Smm /* Any of Surrogate pair is not legal Unicode values. */ 2424232153Smm if (cnt == 3 && IS_SURROGATE_PAIR_LA(*pwc)) 2425232153Smm return (-3); 2426232153Smm return (cnt); 2427232153Smm} 2428232153Smm 2429232153Smmstatic inline uint32_t 2430232153Smmcombine_surrogate_pair(uint32_t uc, uint32_t uc2) 2431232153Smm{ 2432232153Smm uc -= 0xD800; 2433232153Smm uc *= 0x400; 2434232153Smm uc += uc2 - 0xDC00; 2435232153Smm uc += 0x10000; 2436232153Smm return (uc); 2437232153Smm} 2438232153Smm 2439232153Smm/* 2440232153Smm * Convert a single UTF-8/CESU-8 sequence to a Unicode code point in 2441232153Smm * removing surrogate pairs. 2442232153Smm * 2443232153Smm * CESU-8: The Compatibility Encoding Scheme for UTF-16. 2444232153Smm * 2445232153Smm * Usually return used bytes, return used byte in negative value when 2446232153Smm * a unicode character is replaced with U+FFFD. 2447232153Smm */ 2448232153Smmstatic int 2449232153Smmcesu8_to_unicode(uint32_t *pwc, const char *s, size_t n) 2450232153Smm{ 2451248616Smm uint32_t wc = 0; 2452232153Smm int cnt; 2453232153Smm 2454232153Smm cnt = _utf8_to_unicode(&wc, s, n); 2455232153Smm if (cnt == 3 && IS_HIGH_SURROGATE_LA(wc)) { 2456248616Smm uint32_t wc2 = 0; 2457232153Smm if (n - 3 < 3) { 2458232153Smm /* Invalid byte sequence. */ 2459232153Smm goto invalid_sequence; 2460232153Smm } 2461232153Smm cnt = _utf8_to_unicode(&wc2, s+3, n-3); 2462232153Smm if (cnt != 3 || !IS_LOW_SURROGATE_LA(wc2)) { 2463232153Smm /* Invalid byte sequence. */ 2464232153Smm goto invalid_sequence; 2465232153Smm } 2466232153Smm wc = combine_surrogate_pair(wc, wc2); 2467232153Smm cnt = 6; 2468232153Smm } else if (cnt == 3 && IS_LOW_SURROGATE_LA(wc)) { 2469232153Smm /* Invalid byte sequence. */ 2470232153Smm goto invalid_sequence; 2471232153Smm } 2472232153Smm *pwc = wc; 2473232153Smm return (cnt); 2474232153Smminvalid_sequence: 2475232153Smm *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ 2476232153Smm if (cnt > 0) 2477232153Smm cnt *= -1; 2478232153Smm return (cnt); 2479232153Smm} 2480232153Smm 2481232153Smm/* 2482232153Smm * Convert a Unicode code point to a single UTF-8 sequence. 2483232153Smm * 2484311041Smm * NOTE:This function does not check if the Unicode is legal or not. 2485232153Smm * Please you definitely check it before calling this. 2486232153Smm */ 2487232153Smmstatic size_t 2488232153Smmunicode_to_utf8(char *p, size_t remaining, uint32_t uc) 2489232153Smm{ 2490232153Smm char *_p = p; 2491232153Smm 2492299529Smm /* Invalid Unicode char maps to Replacement character */ 2493299529Smm if (uc > UNICODE_MAX) 2494299529Smm uc = UNICODE_R_CHAR; 2495232153Smm /* Translate code point to UTF8 */ 2496232153Smm if (uc <= 0x7f) { 2497232153Smm if (remaining == 0) 2498232153Smm return (0); 2499232153Smm *p++ = (char)uc; 2500232153Smm } else if (uc <= 0x7ff) { 2501232153Smm if (remaining < 2) 2502232153Smm return (0); 2503232153Smm *p++ = 0xc0 | ((uc >> 6) & 0x1f); 2504232153Smm *p++ = 0x80 | (uc & 0x3f); 2505232153Smm } else if (uc <= 0xffff) { 2506232153Smm if (remaining < 3) 2507232153Smm return (0); 2508232153Smm *p++ = 0xe0 | ((uc >> 12) & 0x0f); 2509232153Smm *p++ = 0x80 | ((uc >> 6) & 0x3f); 2510232153Smm *p++ = 0x80 | (uc & 0x3f); 2511299529Smm } else { 2512232153Smm if (remaining < 4) 2513232153Smm return (0); 2514232153Smm *p++ = 0xf0 | ((uc >> 18) & 0x07); 2515232153Smm *p++ = 0x80 | ((uc >> 12) & 0x3f); 2516232153Smm *p++ = 0x80 | ((uc >> 6) & 0x3f); 2517232153Smm *p++ = 0x80 | (uc & 0x3f); 2518232153Smm } 2519232153Smm return (p - _p); 2520232153Smm} 2521232153Smm 2522232153Smmstatic int 2523232153Smmutf16be_to_unicode(uint32_t *pwc, const char *s, size_t n) 2524232153Smm{ 2525232153Smm return (utf16_to_unicode(pwc, s, n, 1)); 2526232153Smm} 2527232153Smm 2528232153Smmstatic int 2529232153Smmutf16le_to_unicode(uint32_t *pwc, const char *s, size_t n) 2530232153Smm{ 2531232153Smm return (utf16_to_unicode(pwc, s, n, 0)); 2532232153Smm} 2533232153Smm 2534232153Smmstatic int 2535232153Smmutf16_to_unicode(uint32_t *pwc, const char *s, size_t n, int be) 2536232153Smm{ 2537232153Smm const char *utf16 = s; 2538232153Smm unsigned uc; 2539232153Smm 2540232153Smm if (n == 0) 2541232153Smm return (0); 2542232153Smm if (n == 1) { 2543232153Smm /* set the Replacement Character instead. */ 2544232153Smm *pwc = UNICODE_R_CHAR; 2545232153Smm return (-1); 2546232153Smm } 2547232153Smm 2548232153Smm if (be) 2549232153Smm uc = archive_be16dec(utf16); 2550232153Smm else 2551232153Smm uc = archive_le16dec(utf16); 2552232153Smm utf16 += 2; 2553232153Smm 2554232153Smm /* If this is a surrogate pair, assemble the full code point.*/ 2555232153Smm if (IS_HIGH_SURROGATE_LA(uc)) { 2556232153Smm unsigned uc2; 2557232153Smm 2558232153Smm if (n >= 4) { 2559232153Smm if (be) 2560232153Smm uc2 = archive_be16dec(utf16); 2561232153Smm else 2562232153Smm uc2 = archive_le16dec(utf16); 2563232153Smm } else 2564232153Smm uc2 = 0; 2565232153Smm if (IS_LOW_SURROGATE_LA(uc2)) { 2566232153Smm uc = combine_surrogate_pair(uc, uc2); 2567232153Smm utf16 += 2; 2568232153Smm } else { 2569232153Smm /* Undescribed code point should be U+FFFD 2570232153Smm * (replacement character). */ 2571232153Smm *pwc = UNICODE_R_CHAR; 2572232153Smm return (-2); 2573232153Smm } 2574232153Smm } 2575232153Smm 2576232153Smm /* 2577232153Smm * Surrogate pair values(0xd800 through 0xdfff) are only 2578313570Smm * used by UTF-16, so, after above calculation, the code 2579232153Smm * must not be surrogate values, and Unicode has no codes 2580311041Smm * larger than 0x10ffff. Thus, those are not legal Unicode 2581232153Smm * values. 2582232153Smm */ 2583232153Smm if (IS_SURROGATE_PAIR_LA(uc) || uc > UNICODE_MAX) { 2584232153Smm /* Undescribed code point should be U+FFFD 2585232153Smm * (replacement character). */ 2586232153Smm *pwc = UNICODE_R_CHAR; 2587232153Smm return (((int)(utf16 - s)) * -1); 2588232153Smm } 2589232153Smm *pwc = uc; 2590232153Smm return ((int)(utf16 - s)); 2591232153Smm} 2592232153Smm 2593232153Smmstatic size_t 2594232153Smmunicode_to_utf16be(char *p, size_t remaining, uint32_t uc) 2595232153Smm{ 2596232153Smm char *utf16 = p; 2597232153Smm 2598232153Smm if (uc > 0xffff) { 2599232153Smm /* We have a code point that won't fit into a 2600232153Smm * wchar_t; convert it to a surrogate pair. */ 2601232153Smm if (remaining < 4) 2602232153Smm return (0); 2603232153Smm uc -= 0x10000; 2604232153Smm archive_be16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); 2605232153Smm archive_be16enc(utf16+2, (uc & 0x3ff) + 0xDC00); 2606228753Smm return (4); 2607232153Smm } else { 2608232153Smm if (remaining < 2) 2609232153Smm return (0); 2610232153Smm archive_be16enc(utf16, uc); 2611232153Smm return (2); 2612232153Smm } 2613228753Smm} 2614228753Smm 2615232153Smmstatic size_t 2616232153Smmunicode_to_utf16le(char *p, size_t remaining, uint32_t uc) 2617232153Smm{ 2618232153Smm char *utf16 = p; 2619232153Smm 2620232153Smm if (uc > 0xffff) { 2621232153Smm /* We have a code point that won't fit into a 2622232153Smm * wchar_t; convert it to a surrogate pair. */ 2623232153Smm if (remaining < 4) 2624232153Smm return (0); 2625232153Smm uc -= 0x10000; 2626232153Smm archive_le16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); 2627232153Smm archive_le16enc(utf16+2, (uc & 0x3ff) + 0xDC00); 2628232153Smm return (4); 2629232153Smm } else { 2630232153Smm if (remaining < 2) 2631232153Smm return (0); 2632232153Smm archive_le16enc(utf16, uc); 2633232153Smm return (2); 2634232153Smm } 2635232153Smm} 2636232153Smm 2637228753Smm/* 2638232153Smm * Copy UTF-8 string in checking surrogate pair. 2639232153Smm * If any surrogate pair are found, it would be canonicalized. 2640228753Smm */ 2641232153Smmstatic int 2642238856Smmstrncat_from_utf8_to_utf8(struct archive_string *as, const void *_p, 2643238856Smm size_t len, struct archive_string_conv *sc) 2644228753Smm{ 2645232153Smm const char *s; 2646232153Smm char *p, *endp; 2647232153Smm int n, ret = 0; 2648228753Smm 2649232153Smm (void)sc; /* UNUSED */ 2650232153Smm 2651232153Smm if (archive_string_ensure(as, as->length + len + 1) == NULL) 2652232153Smm return (-1); 2653232153Smm 2654232153Smm s = (const char *)_p; 2655232153Smm p = as->s + as->length; 2656232153Smm endp = as->s + as->buffer_length -1; 2657232153Smm do { 2658232153Smm uint32_t uc; 2659232153Smm const char *ss = s; 2660232153Smm size_t w; 2661232153Smm 2662232153Smm /* 2663232153Smm * Forward byte sequence until a conversion of that is needed. 2664232153Smm */ 2665232153Smm while ((n = utf8_to_unicode(&uc, s, len)) > 0) { 2666232153Smm s += n; 2667232153Smm len -= n; 2668232153Smm } 2669232153Smm if (ss < s) { 2670232153Smm if (p + (s - ss) > endp) { 2671232153Smm as->length = p - as->s; 2672232153Smm if (archive_string_ensure(as, 2673232153Smm as->buffer_length + len + 1) == NULL) 2674232153Smm return (-1); 2675232153Smm p = as->s + as->length; 2676232153Smm endp = as->s + as->buffer_length -1; 2677232153Smm } 2678232153Smm 2679232153Smm memcpy(p, ss, s - ss); 2680232153Smm p += s - ss; 2681232153Smm } 2682232153Smm 2683232153Smm /* 2684232153Smm * If n is negative, current byte sequence needs a replacement. 2685232153Smm */ 2686228753Smm if (n < 0) { 2687232153Smm if (n == -3 && IS_SURROGATE_PAIR_LA(uc)) { 2688232153Smm /* Current byte sequence may be CESU-8. */ 2689232153Smm n = cesu8_to_unicode(&uc, s, len); 2690232153Smm } 2691228753Smm if (n < 0) { 2692232153Smm ret = -1; 2693232153Smm n *= -1;/* Use a replaced unicode character. */ 2694228753Smm } 2695232153Smm 2696232153Smm /* Rebuild UTF-8 byte sequence. */ 2697232153Smm while ((w = unicode_to_utf8(p, endp - p, uc)) == 0) { 2698232153Smm as->length = p - as->s; 2699232153Smm if (archive_string_ensure(as, 2700232153Smm as->buffer_length + len + 1) == NULL) 2701232153Smm return (-1); 2702232153Smm p = as->s + as->length; 2703232153Smm endp = as->s + as->buffer_length -1; 2704228753Smm } 2705232153Smm p += w; 2706232153Smm s += n; 2707232153Smm len -= n; 2708228753Smm } 2709232153Smm } while (n > 0); 2710232153Smm as->length = p - as->s; 2711232153Smm as->s[as->length] = '\0'; 2712232153Smm return (ret); 2713232153Smm} 2714232153Smm 2715232153Smmstatic int 2716232153Smmarchive_string_append_unicode(struct archive_string *as, const void *_p, 2717232153Smm size_t len, struct archive_string_conv *sc) 2718232153Smm{ 2719232153Smm const char *s; 2720232153Smm char *p, *endp; 2721232153Smm uint32_t uc; 2722232153Smm size_t w; 2723232153Smm int n, ret = 0, ts, tm; 2724232153Smm int (*parse)(uint32_t *, const char *, size_t); 2725232153Smm size_t (*unparse)(char *, size_t, uint32_t); 2726232153Smm 2727232153Smm if (sc->flag & SCONV_TO_UTF16BE) { 2728232153Smm unparse = unicode_to_utf16be; 2729232153Smm ts = 2; 2730232153Smm } else if (sc->flag & SCONV_TO_UTF16LE) { 2731232153Smm unparse = unicode_to_utf16le; 2732232153Smm ts = 2; 2733232153Smm } else if (sc->flag & SCONV_TO_UTF8) { 2734232153Smm unparse = unicode_to_utf8; 2735232153Smm ts = 1; 2736232153Smm } else { 2737232153Smm /* 2738232153Smm * This case is going to be converted to another 2739232153Smm * character-set through iconv. 2740232153Smm */ 2741232153Smm if (sc->flag & SCONV_FROM_UTF16BE) { 2742232153Smm unparse = unicode_to_utf16be; 2743232153Smm ts = 2; 2744232153Smm } else if (sc->flag & SCONV_FROM_UTF16LE) { 2745232153Smm unparse = unicode_to_utf16le; 2746232153Smm ts = 2; 2747232153Smm } else { 2748232153Smm unparse = unicode_to_utf8; 2749232153Smm ts = 1; 2750232153Smm } 2751228753Smm } 2752232153Smm 2753232153Smm if (sc->flag & SCONV_FROM_UTF16BE) { 2754232153Smm parse = utf16be_to_unicode; 2755232153Smm tm = 1; 2756232153Smm } else if (sc->flag & SCONV_FROM_UTF16LE) { 2757232153Smm parse = utf16le_to_unicode; 2758232153Smm tm = 1; 2759232153Smm } else { 2760232153Smm parse = cesu8_to_unicode; 2761232153Smm tm = ts; 2762232153Smm } 2763232153Smm 2764232153Smm if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) 2765232153Smm return (-1); 2766232153Smm 2767232153Smm s = (const char *)_p; 2768232153Smm p = as->s + as->length; 2769232153Smm endp = as->s + as->buffer_length - ts; 2770232153Smm while ((n = parse(&uc, s, len)) != 0) { 2771232153Smm if (n < 0) { 2772232153Smm /* Use a replaced unicode character. */ 2773232153Smm n *= -1; 2774232153Smm ret = -1; 2775232153Smm } 2776232153Smm s += n; 2777232153Smm len -= n; 2778232153Smm while ((w = unparse(p, endp - p, uc)) == 0) { 2779232153Smm /* There is not enough output buffer so 2780232153Smm * we have to expand it. */ 2781232153Smm as->length = p - as->s; 2782232153Smm if (archive_string_ensure(as, 2783232153Smm as->buffer_length + len * tm + ts) == NULL) 2784232153Smm return (-1); 2785232153Smm p = as->s + as->length; 2786232153Smm endp = as->s + as->buffer_length - ts; 2787232153Smm } 2788232153Smm p += w; 2789232153Smm } 2790232153Smm as->length = p - as->s; 2791232153Smm as->s[as->length] = '\0'; 2792232153Smm if (ts == 2) 2793232153Smm as->s[as->length+1] = '\0'; 2794232153Smm return (ret); 2795228753Smm} 2796228753Smm 2797232153Smm/* 2798232153Smm * Following Constants for Hangul compositions this information comes from 2799232153Smm * Unicode Standard Annex #15 http://unicode.org/reports/tr15/ 2800232153Smm */ 2801232153Smm#define HC_SBASE 0xAC00 2802232153Smm#define HC_LBASE 0x1100 2803232153Smm#define HC_VBASE 0x1161 2804232153Smm#define HC_TBASE 0x11A7 2805232153Smm#define HC_LCOUNT 19 2806232153Smm#define HC_VCOUNT 21 2807232153Smm#define HC_TCOUNT 28 2808232153Smm#define HC_NCOUNT (HC_VCOUNT * HC_TCOUNT) 2809232153Smm#define HC_SCOUNT (HC_LCOUNT * HC_NCOUNT) 2810228753Smm 2811232153Smmstatic uint32_t 2812232153Smmget_nfc(uint32_t uc, uint32_t uc2) 2813232153Smm{ 2814232153Smm int t, b; 2815232153Smm 2816232153Smm t = 0; 2817232153Smm b = sizeof(u_composition_table)/sizeof(u_composition_table[0]) -1; 2818232153Smm while (b >= t) { 2819232153Smm int m = (t + b) / 2; 2820232153Smm if (u_composition_table[m].cp1 < uc) 2821232153Smm t = m + 1; 2822232153Smm else if (u_composition_table[m].cp1 > uc) 2823232153Smm b = m - 1; 2824232153Smm else if (u_composition_table[m].cp2 < uc2) 2825232153Smm t = m + 1; 2826232153Smm else if (u_composition_table[m].cp2 > uc2) 2827232153Smm b = m - 1; 2828232153Smm else 2829232153Smm return (u_composition_table[m].nfc); 2830232153Smm } 2831232153Smm return (0); 2832232153Smm} 2833232153Smm 2834232153Smm#define FDC_MAX 10 /* The maximum number of Following Decomposable 2835232153Smm * Characters. */ 2836232153Smm 2837228753Smm/* 2838232153Smm * Update first code point. 2839232153Smm */ 2840232153Smm#define UPDATE_UC(new_uc) do { \ 2841232153Smm uc = new_uc; \ 2842232153Smm ucptr = NULL; \ 2843232153Smm} while (0) 2844232153Smm 2845232153Smm/* 2846232153Smm * Replace first code point with second code point. 2847232153Smm */ 2848232153Smm#define REPLACE_UC_WITH_UC2() do { \ 2849232153Smm uc = uc2; \ 2850232153Smm ucptr = uc2ptr; \ 2851232153Smm n = n2; \ 2852232153Smm} while (0) 2853232153Smm 2854232153Smm#define EXPAND_BUFFER() do { \ 2855232153Smm as->length = p - as->s; \ 2856232153Smm if (archive_string_ensure(as, \ 2857232153Smm as->buffer_length + len * tm + ts) == NULL)\ 2858232153Smm return (-1); \ 2859232153Smm p = as->s + as->length; \ 2860232153Smm endp = as->s + as->buffer_length - ts; \ 2861232153Smm} while (0) 2862232153Smm 2863232153Smm#define UNPARSE(p, endp, uc) do { \ 2864232153Smm while ((w = unparse(p, (endp) - (p), uc)) == 0) {\ 2865232153Smm EXPAND_BUFFER(); \ 2866232153Smm } \ 2867232153Smm p += w; \ 2868232153Smm} while (0) 2869232153Smm 2870232153Smm/* 2871232153Smm * Write first code point. 2872232153Smm * If the code point has not be changed from its original code, 2873232153Smm * this just copies it from its original buffer pointer. 2874232153Smm * If not, this converts it to UTF-8 byte sequence and copies it. 2875232153Smm */ 2876232153Smm#define WRITE_UC() do { \ 2877232153Smm if (ucptr) { \ 2878232153Smm if (p + n > endp) \ 2879232153Smm EXPAND_BUFFER(); \ 2880232153Smm switch (n) { \ 2881232153Smm case 4: \ 2882232153Smm *p++ = *ucptr++; \ 2883232153Smm /* FALL THROUGH */ \ 2884232153Smm case 3: \ 2885232153Smm *p++ = *ucptr++; \ 2886232153Smm /* FALL THROUGH */ \ 2887232153Smm case 2: \ 2888232153Smm *p++ = *ucptr++; \ 2889232153Smm /* FALL THROUGH */ \ 2890232153Smm case 1: \ 2891232153Smm *p++ = *ucptr; \ 2892232153Smm break; \ 2893232153Smm } \ 2894232153Smm ucptr = NULL; \ 2895232153Smm } else { \ 2896232153Smm UNPARSE(p, endp, uc); \ 2897232153Smm } \ 2898232153Smm} while (0) 2899232153Smm 2900232153Smm/* 2901232153Smm * Collect following decomposable code points. 2902232153Smm */ 2903232153Smm#define COLLECT_CPS(start) do { \ 2904232153Smm int _i; \ 2905232153Smm for (_i = start; _i < FDC_MAX ; _i++) { \ 2906232153Smm nx = parse(&ucx[_i], s, len); \ 2907232153Smm if (nx <= 0) \ 2908232153Smm break; \ 2909232153Smm cx = CCC(ucx[_i]); \ 2910232153Smm if (cl >= cx && cl != 228 && cx != 228)\ 2911232153Smm break; \ 2912232153Smm s += nx; \ 2913232153Smm len -= nx; \ 2914232153Smm cl = cx; \ 2915232153Smm ccx[_i] = cx; \ 2916232153Smm } \ 2917232153Smm if (_i >= FDC_MAX) { \ 2918232153Smm ret = -1; \ 2919232153Smm ucx_size = FDC_MAX; \ 2920232153Smm } else \ 2921232153Smm ucx_size = _i; \ 2922232153Smm} while (0) 2923232153Smm 2924232153Smm/* 2925232153Smm * Normalize UTF-8/UTF-16BE characters to Form C and copy the result. 2926228753Smm * 2927313570Smm * TODO: Convert composition exclusions, which are never converted 2928232153Smm * from NFC,NFD,NFKC and NFKD, to Form C. 2929228753Smm */ 2930232153Smmstatic int 2931232153Smmarchive_string_normalize_C(struct archive_string *as, const void *_p, 2932232153Smm size_t len, struct archive_string_conv *sc) 2933228753Smm{ 2934232153Smm const char *s = (const char *)_p; 2935232153Smm char *p, *endp; 2936232153Smm uint32_t uc, uc2; 2937232153Smm size_t w; 2938232153Smm int always_replace, n, n2, ret = 0, spair, ts, tm; 2939232153Smm int (*parse)(uint32_t *, const char *, size_t); 2940232153Smm size_t (*unparse)(char *, size_t, uint32_t); 2941228753Smm 2942232153Smm always_replace = 1; 2943232153Smm ts = 1;/* text size. */ 2944232153Smm if (sc->flag & SCONV_TO_UTF16BE) { 2945232153Smm unparse = unicode_to_utf16be; 2946232153Smm ts = 2; 2947232153Smm if (sc->flag & SCONV_FROM_UTF16BE) 2948232153Smm always_replace = 0; 2949232153Smm } else if (sc->flag & SCONV_TO_UTF16LE) { 2950232153Smm unparse = unicode_to_utf16le; 2951232153Smm ts = 2; 2952232153Smm if (sc->flag & SCONV_FROM_UTF16LE) 2953232153Smm always_replace = 0; 2954232153Smm } else if (sc->flag & SCONV_TO_UTF8) { 2955232153Smm unparse = unicode_to_utf8; 2956232153Smm if (sc->flag & SCONV_FROM_UTF8) 2957232153Smm always_replace = 0; 2958232153Smm } else { 2959232153Smm /* 2960232153Smm * This case is going to be converted to another 2961232153Smm * character-set through iconv. 2962232153Smm */ 2963232153Smm always_replace = 0; 2964232153Smm if (sc->flag & SCONV_FROM_UTF16BE) { 2965232153Smm unparse = unicode_to_utf16be; 2966232153Smm ts = 2; 2967232153Smm } else if (sc->flag & SCONV_FROM_UTF16LE) { 2968232153Smm unparse = unicode_to_utf16le; 2969232153Smm ts = 2; 2970232153Smm } else { 2971232153Smm unparse = unicode_to_utf8; 2972232153Smm } 2973232153Smm } 2974232153Smm 2975232153Smm if (sc->flag & SCONV_FROM_UTF16BE) { 2976232153Smm parse = utf16be_to_unicode; 2977232153Smm tm = 1; 2978232153Smm spair = 4;/* surrogate pair size in UTF-16. */ 2979232153Smm } else if (sc->flag & SCONV_FROM_UTF16LE) { 2980232153Smm parse = utf16le_to_unicode; 2981232153Smm tm = 1; 2982232153Smm spair = 4;/* surrogate pair size in UTF-16. */ 2983232153Smm } else { 2984232153Smm parse = cesu8_to_unicode; 2985232153Smm tm = ts; 2986232153Smm spair = 6;/* surrogate pair size in UTF-8. */ 2987232153Smm } 2988232153Smm 2989232153Smm if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) 2990232153Smm return (-1); 2991232153Smm 2992232153Smm p = as->s + as->length; 2993232153Smm endp = as->s + as->buffer_length - ts; 2994232153Smm while ((n = parse(&uc, s, len)) != 0) { 2995232153Smm const char *ucptr, *uc2ptr; 2996232153Smm 2997232153Smm if (n < 0) { 2998232153Smm /* Use a replaced unicode character. */ 2999232153Smm UNPARSE(p, endp, uc); 3000232153Smm s += n*-1; 3001232153Smm len -= n*-1; 3002232153Smm ret = -1; 3003232153Smm continue; 3004232153Smm } else if (n == spair || always_replace) 3005232153Smm /* uc is converted from a surrogate pair. 3006232153Smm * this should be treated as a changed code. */ 3007232153Smm ucptr = NULL; 3008232153Smm else 3009232153Smm ucptr = s; 3010232153Smm s += n; 3011232153Smm len -= n; 3012232153Smm 3013232153Smm /* Read second code point. */ 3014232153Smm while ((n2 = parse(&uc2, s, len)) > 0) { 3015232153Smm uint32_t ucx[FDC_MAX]; 3016232153Smm int ccx[FDC_MAX]; 3017232153Smm int cl, cx, i, nx, ucx_size; 3018232153Smm int LIndex,SIndex; 3019232153Smm uint32_t nfc; 3020232153Smm 3021232153Smm if (n2 == spair || always_replace) 3022232153Smm /* uc2 is converted from a surrogate pair. 3023232153Smm * this should be treated as a changed code. */ 3024232153Smm uc2ptr = NULL; 3025232153Smm else 3026232153Smm uc2ptr = s; 3027232153Smm s += n2; 3028232153Smm len -= n2; 3029232153Smm 3030232153Smm /* 3031232153Smm * If current second code point is out of decomposable 3032232153Smm * code points, finding compositions is unneeded. 3033232153Smm */ 3034232153Smm if (!IS_DECOMPOSABLE_BLOCK(uc2)) { 3035232153Smm WRITE_UC(); 3036232153Smm REPLACE_UC_WITH_UC2(); 3037232153Smm continue; 3038232153Smm } 3039232153Smm 3040232153Smm /* 3041232153Smm * Try to combine current code points. 3042232153Smm */ 3043232153Smm /* 3044232153Smm * We have to combine Hangul characters according to 3045232153Smm * http://uniicode.org/reports/tr15/#Hangul 3046232153Smm */ 3047232153Smm if (0 <= (LIndex = uc - HC_LBASE) && 3048232153Smm LIndex < HC_LCOUNT) { 3049232153Smm /* 3050232153Smm * Hangul Composition. 3051232153Smm * 1. Two current code points are L and V. 3052232153Smm */ 3053232153Smm int VIndex = uc2 - HC_VBASE; 3054232153Smm if (0 <= VIndex && VIndex < HC_VCOUNT) { 3055232153Smm /* Make syllable of form LV. */ 3056232153Smm UPDATE_UC(HC_SBASE + 3057232153Smm (LIndex * HC_VCOUNT + VIndex) * 3058232153Smm HC_TCOUNT); 3059232153Smm } else { 3060232153Smm WRITE_UC(); 3061232153Smm REPLACE_UC_WITH_UC2(); 3062232153Smm } 3063232153Smm continue; 3064232153Smm } else if (0 <= (SIndex = uc - HC_SBASE) && 3065232153Smm SIndex < HC_SCOUNT && (SIndex % HC_TCOUNT) == 0) { 3066232153Smm /* 3067232153Smm * Hangul Composition. 3068232153Smm * 2. Two current code points are LV and T. 3069232153Smm */ 3070232153Smm int TIndex = uc2 - HC_TBASE; 3071232153Smm if (0 < TIndex && TIndex < HC_TCOUNT) { 3072232153Smm /* Make syllable of form LVT. */ 3073232153Smm UPDATE_UC(uc + TIndex); 3074232153Smm } else { 3075232153Smm WRITE_UC(); 3076232153Smm REPLACE_UC_WITH_UC2(); 3077232153Smm } 3078232153Smm continue; 3079232153Smm } else if ((nfc = get_nfc(uc, uc2)) != 0) { 3080232153Smm /* A composition to current code points 3081232153Smm * is found. */ 3082232153Smm UPDATE_UC(nfc); 3083232153Smm continue; 3084232153Smm } else if ((cl = CCC(uc2)) == 0) { 3085232153Smm /* Clearly 'uc2' the second code point is not 3086232153Smm * a decomposable code. */ 3087232153Smm WRITE_UC(); 3088232153Smm REPLACE_UC_WITH_UC2(); 3089232153Smm continue; 3090232153Smm } 3091232153Smm 3092232153Smm /* 3093232153Smm * Collect following decomposable code points. 3094232153Smm */ 3095232153Smm cx = 0; 3096232153Smm ucx[0] = uc2; 3097232153Smm ccx[0] = cl; 3098232153Smm COLLECT_CPS(1); 3099232153Smm 3100232153Smm /* 3101232153Smm * Find a composed code in the collected code points. 3102232153Smm */ 3103232153Smm i = 1; 3104232153Smm while (i < ucx_size) { 3105232153Smm int j; 3106232153Smm 3107232153Smm if ((nfc = get_nfc(uc, ucx[i])) == 0) { 3108232153Smm i++; 3109232153Smm continue; 3110232153Smm } 3111232153Smm 3112232153Smm /* 3113232153Smm * nfc is composed of uc and ucx[i]. 3114232153Smm */ 3115232153Smm UPDATE_UC(nfc); 3116232153Smm 3117232153Smm /* 3118232153Smm * Remove ucx[i] by shifting 3119232153Smm * following code points. 3120232153Smm */ 3121232153Smm for (j = i; j+1 < ucx_size; j++) { 3122232153Smm ucx[j] = ucx[j+1]; 3123232153Smm ccx[j] = ccx[j+1]; 3124232153Smm } 3125232153Smm ucx_size --; 3126232153Smm 3127232153Smm /* 3128232153Smm * Collect following code points blocked 3129232153Smm * by ucx[i] the removed code point. 3130232153Smm */ 3131232153Smm if (ucx_size > 0 && i == ucx_size && 3132232153Smm nx > 0 && cx == cl) { 3133232153Smm cl = ccx[ucx_size-1]; 3134232153Smm COLLECT_CPS(ucx_size); 3135232153Smm } 3136232153Smm /* 3137232153Smm * Restart finding a composed code with 3138232153Smm * the updated uc from the top of the 3139232153Smm * collected code points. 3140232153Smm */ 3141232153Smm i = 0; 3142232153Smm } 3143232153Smm 3144232153Smm /* 3145232153Smm * Apparently the current code points are not 3146232153Smm * decomposed characters or already composed. 3147232153Smm */ 3148232153Smm WRITE_UC(); 3149232153Smm for (i = 0; i < ucx_size; i++) 3150232153Smm UNPARSE(p, endp, ucx[i]); 3151232153Smm 3152232153Smm /* 3153232153Smm * Flush out remaining canonical combining characters. 3154232153Smm */ 3155232153Smm if (nx > 0 && cx == cl && len > 0) { 3156232153Smm while ((nx = parse(&ucx[0], s, len)) 3157232153Smm > 0) { 3158232153Smm cx = CCC(ucx[0]); 3159232153Smm if (cl > cx) 3160232153Smm break; 3161232153Smm s += nx; 3162232153Smm len -= nx; 3163232153Smm cl = cx; 3164232153Smm UNPARSE(p, endp, ucx[0]); 3165232153Smm } 3166232153Smm } 3167232153Smm break; 3168232153Smm } 3169232153Smm if (n2 < 0) { 3170232153Smm WRITE_UC(); 3171232153Smm /* Use a replaced unicode character. */ 3172232153Smm UNPARSE(p, endp, uc2); 3173232153Smm s += n2*-1; 3174232153Smm len -= n2*-1; 3175232153Smm ret = -1; 3176232153Smm continue; 3177232153Smm } else if (n2 == 0) { 3178232153Smm WRITE_UC(); 3179232153Smm break; 3180232153Smm } 3181232153Smm } 3182232153Smm as->length = p - as->s; 3183232153Smm as->s[as->length] = '\0'; 3184232153Smm if (ts == 2) 3185232153Smm as->s[as->length+1] = '\0'; 3186232153Smm return (ret); 3187232153Smm} 3188232153Smm 3189238856Smmstatic int 3190238856Smmget_nfd(uint32_t *cp1, uint32_t *cp2, uint32_t uc) 3191238856Smm{ 3192238856Smm int t, b; 3193232153Smm 3194238856Smm /* 3195238856Smm * These are not converted to NFD on Mac OS. 3196238856Smm */ 3197238856Smm if ((uc >= 0x2000 && uc <= 0x2FFF) || 3198238856Smm (uc >= 0xF900 && uc <= 0xFAFF) || 3199238856Smm (uc >= 0x2F800 && uc <= 0x2FAFF)) 3200238856Smm return (0); 3201238856Smm /* 3202238856Smm * Those code points are not converted to NFD on Mac OS. 3203238856Smm * I do not know the reason because it is undocumented. 3204238856Smm * NFC NFD 3205238856Smm * 1109A ==> 11099 110BA 3206238856Smm * 1109C ==> 1109B 110BA 3207238856Smm * 110AB ==> 110A5 110BA 3208238856Smm */ 3209238856Smm if (uc == 0x1109A || uc == 0x1109C || uc == 0x110AB) 3210238856Smm return (0); 3211238856Smm 3212238856Smm t = 0; 3213238856Smm b = sizeof(u_decomposition_table)/sizeof(u_decomposition_table[0]) -1; 3214238856Smm while (b >= t) { 3215238856Smm int m = (t + b) / 2; 3216238856Smm if (u_decomposition_table[m].nfc < uc) 3217238856Smm t = m + 1; 3218238856Smm else if (u_decomposition_table[m].nfc > uc) 3219238856Smm b = m - 1; 3220238856Smm else { 3221238856Smm *cp1 = u_decomposition_table[m].cp1; 3222238856Smm *cp2 = u_decomposition_table[m].cp2; 3223238856Smm return (1); 3224238856Smm } 3225238856Smm } 3226238856Smm return (0); 3227238856Smm} 3228238856Smm 3229238856Smm#define REPLACE_UC_WITH(cp) do { \ 3230238856Smm uc = cp; \ 3231238856Smm ucptr = NULL; \ 3232238856Smm} while (0) 3233238856Smm 3234232153Smm/* 3235232153Smm * Normalize UTF-8 characters to Form D and copy the result. 3236232153Smm */ 3237232153Smmstatic int 3238232153Smmarchive_string_normalize_D(struct archive_string *as, const void *_p, 3239232153Smm size_t len, struct archive_string_conv *sc) 3240232153Smm{ 3241238856Smm const char *s = (const char *)_p; 3242238856Smm char *p, *endp; 3243238856Smm uint32_t uc, uc2; 3244238856Smm size_t w; 3245238856Smm int always_replace, n, n2, ret = 0, spair, ts, tm; 3246238856Smm int (*parse)(uint32_t *, const char *, size_t); 3247238856Smm size_t (*unparse)(char *, size_t, uint32_t); 3248232153Smm 3249238856Smm always_replace = 1; 3250238856Smm ts = 1;/* text size. */ 3251238856Smm if (sc->flag & SCONV_TO_UTF16BE) { 3252238856Smm unparse = unicode_to_utf16be; 3253238856Smm ts = 2; 3254238856Smm if (sc->flag & SCONV_FROM_UTF16BE) 3255238856Smm always_replace = 0; 3256238856Smm } else if (sc->flag & SCONV_TO_UTF16LE) { 3257238856Smm unparse = unicode_to_utf16le; 3258238856Smm ts = 2; 3259238856Smm if (sc->flag & SCONV_FROM_UTF16LE) 3260238856Smm always_replace = 0; 3261238856Smm } else if (sc->flag & SCONV_TO_UTF8) { 3262238856Smm unparse = unicode_to_utf8; 3263238856Smm if (sc->flag & SCONV_FROM_UTF8) 3264238856Smm always_replace = 0; 3265238856Smm } else { 3266238856Smm /* 3267238856Smm * This case is going to be converted to another 3268238856Smm * character-set through iconv. 3269238856Smm */ 3270238856Smm always_replace = 0; 3271238856Smm if (sc->flag & SCONV_FROM_UTF16BE) { 3272238856Smm unparse = unicode_to_utf16be; 3273238856Smm ts = 2; 3274238856Smm } else if (sc->flag & SCONV_FROM_UTF16LE) { 3275238856Smm unparse = unicode_to_utf16le; 3276238856Smm ts = 2; 3277238856Smm } else { 3278238856Smm unparse = unicode_to_utf8; 3279238856Smm } 3280228753Smm } 3281232153Smm 3282238856Smm if (sc->flag & SCONV_FROM_UTF16BE) { 3283238856Smm parse = utf16be_to_unicode; 3284238856Smm tm = 1; 3285238856Smm spair = 4;/* surrogate pair size in UTF-16. */ 3286238856Smm } else if (sc->flag & SCONV_FROM_UTF16LE) { 3287238856Smm parse = utf16le_to_unicode; 3288238856Smm tm = 1; 3289238856Smm spair = 4;/* surrogate pair size in UTF-16. */ 3290238856Smm } else { 3291238856Smm parse = cesu8_to_unicode; 3292238856Smm tm = ts; 3293238856Smm spair = 6;/* surrogate pair size in UTF-8. */ 3294238856Smm } 3295238856Smm 3296238856Smm if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) 3297232153Smm return (-1); 3298232153Smm 3299238856Smm p = as->s + as->length; 3300238856Smm endp = as->s + as->buffer_length - ts; 3301238856Smm while ((n = parse(&uc, s, len)) != 0) { 3302238856Smm const char *ucptr; 3303238856Smm uint32_t cp1, cp2; 3304238856Smm int SIndex; 3305238856Smm struct { 3306238856Smm uint32_t uc; 3307238856Smm int ccc; 3308238856Smm } fdc[FDC_MAX]; 3309238856Smm int fdi, fdj; 3310238856Smm int ccc; 3311232153Smm 3312238856Smmcheck_first_code: 3313238856Smm if (n < 0) { 3314238856Smm /* Use a replaced unicode character. */ 3315238856Smm UNPARSE(p, endp, uc); 3316238856Smm s += n*-1; 3317238856Smm len -= n*-1; 3318238856Smm ret = -1; 3319238856Smm continue; 3320238856Smm } else if (n == spair || always_replace) 3321238856Smm /* uc is converted from a surrogate pair. 3322238856Smm * this should be treated as a changed code. */ 3323238856Smm ucptr = NULL; 3324238856Smm else 3325238856Smm ucptr = s; 3326238856Smm s += n; 3327238856Smm len -= n; 3328232153Smm 3329238856Smm /* Hangul Decomposition. */ 3330238856Smm if ((SIndex = uc - HC_SBASE) >= 0 && SIndex < HC_SCOUNT) { 3331238856Smm int L = HC_LBASE + SIndex / HC_NCOUNT; 3332238856Smm int V = HC_VBASE + (SIndex % HC_NCOUNT) / HC_TCOUNT; 3333238856Smm int T = HC_TBASE + SIndex % HC_TCOUNT; 3334232153Smm 3335238856Smm REPLACE_UC_WITH(L); 3336238856Smm WRITE_UC(); 3337238856Smm REPLACE_UC_WITH(V); 3338238856Smm WRITE_UC(); 3339238856Smm if (T != HC_TBASE) { 3340238856Smm REPLACE_UC_WITH(T); 3341238856Smm WRITE_UC(); 3342238856Smm } 3343238856Smm continue; 3344238856Smm } 3345238856Smm if (IS_DECOMPOSABLE_BLOCK(uc) && CCC(uc) != 0) { 3346238856Smm WRITE_UC(); 3347238856Smm continue; 3348238856Smm } 3349232153Smm 3350238856Smm fdi = 0; 3351238856Smm while (get_nfd(&cp1, &cp2, uc) && fdi < FDC_MAX) { 3352238856Smm int k; 3353238856Smm 3354238856Smm for (k = fdi; k > 0; k--) 3355238856Smm fdc[k] = fdc[k-1]; 3356238856Smm fdc[0].ccc = CCC(cp2); 3357238856Smm fdc[0].uc = cp2; 3358238856Smm fdi++; 3359238856Smm REPLACE_UC_WITH(cp1); 3360238856Smm } 3361238856Smm 3362238856Smm /* Read following code points. */ 3363238856Smm while ((n2 = parse(&uc2, s, len)) > 0 && 3364238856Smm (ccc = CCC(uc2)) != 0 && fdi < FDC_MAX) { 3365238856Smm int j, k; 3366238856Smm 3367238856Smm s += n2; 3368238856Smm len -= n2; 3369238856Smm for (j = 0; j < fdi; j++) { 3370238856Smm if (fdc[j].ccc > ccc) 3371238856Smm break; 3372238856Smm } 3373238856Smm if (j < fdi) { 3374238856Smm for (k = fdi; k > j; k--) 3375238856Smm fdc[k] = fdc[k-1]; 3376238856Smm fdc[j].ccc = ccc; 3377238856Smm fdc[j].uc = uc2; 3378238856Smm } else { 3379238856Smm fdc[fdi].ccc = ccc; 3380238856Smm fdc[fdi].uc = uc2; 3381238856Smm } 3382238856Smm fdi++; 3383238856Smm } 3384238856Smm 3385238856Smm WRITE_UC(); 3386238856Smm for (fdj = 0; fdj < fdi; fdj++) { 3387238856Smm REPLACE_UC_WITH(fdc[fdj].uc); 3388238856Smm WRITE_UC(); 3389238856Smm } 3390238856Smm 3391238856Smm if (n2 == 0) 3392238856Smm break; 3393238856Smm REPLACE_UC_WITH(uc2); 3394238856Smm n = n2; 3395238856Smm goto check_first_code; 3396232153Smm } 3397238856Smm as->length = p - as->s; 3398238856Smm as->s[as->length] = '\0'; 3399238856Smm if (ts == 2) 3400238856Smm as->s[as->length+1] = '\0'; 3401232153Smm return (ret); 3402228753Smm} 3403228753Smm 3404228753Smm/* 3405232153Smm * libarchive 2.x made incorrect UTF-8 strings in the wrong assumption 3406232153Smm * that WCS is Unicode. It is true for several platforms but some are false. 3407232153Smm * And then people who did not use UTF-8 locale on the non Unicode WCS 3408232153Smm * platform and made a tar file with libarchive(mostly bsdtar) 2.x. Those 3409232153Smm * now cannot get right filename from libarchive 3.x and later since we 3410232153Smm * fixed the wrong assumption and it is incompatible to older its versions. 3411232153Smm * So we provide special option, "compat-2x.x", for resolving it. 3412232153Smm * That option enable the string conversion of libarchive 2.x. 3413228753Smm * 3414232153Smm * Translates the wrong UTF-8 string made by libarchive 2.x into current 3415232153Smm * locale character set and appends to the archive_string. 3416232153Smm * Note: returns -1 if conversion fails. 3417228753Smm */ 3418232153Smmstatic int 3419232153Smmstrncat_from_utf8_libarchive2(struct archive_string *as, 3420232153Smm const void *_p, size_t len, struct archive_string_conv *sc) 3421228753Smm{ 3422232153Smm const char *s; 3423228753Smm int n; 3424228753Smm char *p; 3425232153Smm char *end; 3426232153Smm uint32_t unicode; 3427228753Smm#if HAVE_WCRTOMB 3428228753Smm mbstate_t shift_state; 3429228753Smm 3430228753Smm memset(&shift_state, 0, sizeof(shift_state)); 3431228753Smm#else 3432228753Smm /* Clear the shift state before starting. */ 3433228753Smm wctomb(NULL, L'\0'); 3434228753Smm#endif 3435232153Smm (void)sc; /* UNUSED */ 3436228753Smm /* 3437232153Smm * Allocate buffer for MBS. 3438232153Smm * We need this allocation here since it is possible that 3439232153Smm * as->s is still NULL. 3440228753Smm */ 3441232153Smm if (archive_string_ensure(as, as->length + len + 1) == NULL) 3442232153Smm return (-1); 3443232153Smm 3444232153Smm s = (const char *)_p; 3445232153Smm p = as->s + as->length; 3446232153Smm end = as->s + as->buffer_length - MB_CUR_MAX -1; 3447232153Smm while ((n = _utf8_to_unicode(&unicode, s, len)) != 0) { 3448232153Smm wchar_t wc; 3449232153Smm 3450232153Smm if (p >= end) { 3451232153Smm as->length = p - as->s; 3452232153Smm /* Re-allocate buffer for MBS. */ 3453232153Smm if (archive_string_ensure(as, 3454358088Smm as->length + max(len * 2, 3455358088Smm (size_t)MB_CUR_MAX) + 1) == NULL) 3456232153Smm return (-1); 3457232153Smm p = as->s + as->length; 3458232153Smm end = as->s + as->buffer_length - MB_CUR_MAX -1; 3459228753Smm } 3460232153Smm 3461232153Smm /* 3462313570Smm * As libarchive 2.x, translates the UTF-8 characters into 3463232153Smm * wide-characters in the assumption that WCS is Unicode. 3464232153Smm */ 3465232153Smm if (n < 0) { 3466232153Smm n *= -1; 3467232153Smm wc = L'?'; 3468232153Smm } else 3469232153Smm wc = (wchar_t)unicode; 3470232153Smm 3471232153Smm s += n; 3472232153Smm len -= n; 3473232153Smm /* 3474232153Smm * Translates the wide-character into the current locale MBS. 3475232153Smm */ 3476228753Smm#if HAVE_WCRTOMB 3477248616Smm n = (int)wcrtomb(p, wc, &shift_state); 3478228753Smm#else 3479248616Smm n = (int)wctomb(p, wc); 3480228753Smm#endif 3481228753Smm if (n == -1) 3482232153Smm return (-1); 3483228753Smm p += n; 3484228753Smm } 3485232153Smm as->length = p - as->s; 3486232153Smm as->s[as->length] = '\0'; 3487232153Smm return (0); 3488232153Smm} 3489232153Smm 3490232153Smm 3491232153Smm/* 3492232153Smm * Conversion functions between current locale dependent MBS and UTF-16BE. 3493232153Smm * strncat_from_utf16be() : UTF-16BE --> MBS 3494232153Smm * strncat_to_utf16be() : MBS --> UTF16BE 3495232153Smm */ 3496232153Smm 3497232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 3498232153Smm 3499232153Smm/* 3500232153Smm * Convert a UTF-16BE/LE string to current locale and copy the result. 3501311041Smm * Return -1 if conversion fails. 3502232153Smm */ 3503232153Smmstatic int 3504232153Smmwin_strncat_from_utf16(struct archive_string *as, const void *_p, size_t bytes, 3505232153Smm struct archive_string_conv *sc, int be) 3506232153Smm{ 3507232153Smm struct archive_string tmp; 3508232153Smm const char *u16; 3509232153Smm int ll; 3510232153Smm BOOL defchar; 3511232153Smm char *mbs; 3512232153Smm size_t mbs_size, b; 3513232153Smm int ret = 0; 3514232153Smm 3515232153Smm bytes &= ~1; 3516232153Smm if (archive_string_ensure(as, as->length + bytes +1) == NULL) 3517232153Smm return (-1); 3518232153Smm 3519232153Smm mbs = as->s + as->length; 3520232153Smm mbs_size = as->buffer_length - as->length -1; 3521232153Smm 3522232153Smm if (sc->to_cp == CP_C_LOCALE) { 3523232153Smm /* 3524232153Smm * "C" locale special process. 3525232153Smm */ 3526232153Smm u16 = _p; 3527232153Smm ll = 0; 3528232153Smm for (b = 0; b < bytes; b += 2) { 3529232153Smm uint16_t val; 3530232153Smm if (be) 3531232153Smm val = archive_be16dec(u16+b); 3532232153Smm else 3533232153Smm val = archive_le16dec(u16+b); 3534232153Smm if (val > 255) { 3535232153Smm *mbs++ = '?'; 3536232153Smm ret = -1; 3537232153Smm } else 3538232153Smm *mbs++ = (char)(val&0xff); 3539232153Smm ll++; 3540232153Smm } 3541232153Smm as->length += ll; 3542232153Smm as->s[as->length] = '\0'; 3543232153Smm return (ret); 3544232153Smm } 3545232153Smm 3546232153Smm archive_string_init(&tmp); 3547232153Smm if (be) { 3548232153Smm if (is_big_endian()) { 3549232153Smm u16 = _p; 3550232153Smm } else { 3551232153Smm if (archive_string_ensure(&tmp, bytes+2) == NULL) 3552232153Smm return (-1); 3553232153Smm memcpy(tmp.s, _p, bytes); 3554232153Smm for (b = 0; b < bytes; b += 2) { 3555232153Smm uint16_t val = archive_be16dec(tmp.s+b); 3556232153Smm archive_le16enc(tmp.s+b, val); 3557232153Smm } 3558232153Smm u16 = tmp.s; 3559232153Smm } 3560232153Smm } else { 3561232153Smm if (!is_big_endian()) { 3562232153Smm u16 = _p; 3563232153Smm } else { 3564232153Smm if (archive_string_ensure(&tmp, bytes+2) == NULL) 3565232153Smm return (-1); 3566232153Smm memcpy(tmp.s, _p, bytes); 3567232153Smm for (b = 0; b < bytes; b += 2) { 3568232153Smm uint16_t val = archive_le16dec(tmp.s+b); 3569232153Smm archive_be16enc(tmp.s+b, val); 3570232153Smm } 3571232153Smm u16 = tmp.s; 3572232153Smm } 3573232153Smm } 3574232153Smm 3575232153Smm do { 3576232153Smm defchar = 0; 3577232153Smm ll = WideCharToMultiByte(sc->to_cp, 0, 3578248616Smm (LPCWSTR)u16, (int)bytes>>1, mbs, (int)mbs_size, 3579232153Smm NULL, &defchar); 3580302294Smm /* Exit loop if we succeeded */ 3581302294Smm if (ll != 0 || 3582302294Smm GetLastError() != ERROR_INSUFFICIENT_BUFFER) { 3583302294Smm break; 3584232153Smm } 3585302294Smm /* Else expand buffer and loop to try again. */ 3586302294Smm ll = WideCharToMultiByte(sc->to_cp, 0, 3587302294Smm (LPCWSTR)u16, (int)bytes, NULL, 0, NULL, NULL); 3588302294Smm if (archive_string_ensure(as, ll +1) == NULL) 3589302294Smm return (-1); 3590302294Smm mbs = as->s + as->length; 3591302294Smm mbs_size = as->buffer_length - as->length -1; 3592302294Smm } while (1); 3593232153Smm archive_string_free(&tmp); 3594232153Smm as->length += ll; 3595232153Smm as->s[as->length] = '\0'; 3596232153Smm if (ll == 0 || defchar) 3597232153Smm ret = -1; 3598232153Smm return (ret); 3599232153Smm} 3600232153Smm 3601232153Smmstatic int 3602238856Smmwin_strncat_from_utf16be(struct archive_string *as, const void *_p, 3603238856Smm size_t bytes, struct archive_string_conv *sc) 3604232153Smm{ 3605232153Smm return (win_strncat_from_utf16(as, _p, bytes, sc, 1)); 3606232153Smm} 3607232153Smm 3608232153Smmstatic int 3609238856Smmwin_strncat_from_utf16le(struct archive_string *as, const void *_p, 3610238856Smm size_t bytes, struct archive_string_conv *sc) 3611232153Smm{ 3612232153Smm return (win_strncat_from_utf16(as, _p, bytes, sc, 0)); 3613232153Smm} 3614232153Smm 3615232153Smmstatic int 3616232153Smmis_big_endian(void) 3617232153Smm{ 3618232153Smm uint16_t d = 1; 3619232153Smm 3620232153Smm return (archive_be16dec(&d) == 1); 3621232153Smm} 3622232153Smm 3623232153Smm/* 3624232153Smm * Convert a current locale string to UTF-16BE/LE and copy the result. 3625311041Smm * Return -1 if conversion fails. 3626232153Smm */ 3627232153Smmstatic int 3628238856Smmwin_strncat_to_utf16(struct archive_string *as16, const void *_p, 3629238856Smm size_t length, struct archive_string_conv *sc, int bigendian) 3630232153Smm{ 3631232153Smm const char *s = (const char *)_p; 3632232153Smm char *u16; 3633232153Smm size_t count, avail; 3634232153Smm 3635232153Smm if (archive_string_ensure(as16, 3636232153Smm as16->length + (length + 1) * 2) == NULL) 3637232153Smm return (-1); 3638232153Smm 3639232153Smm u16 = as16->s + as16->length; 3640232153Smm avail = as16->buffer_length - 2; 3641232153Smm if (sc->from_cp == CP_C_LOCALE) { 3642232153Smm /* 3643232153Smm * "C" locale special process. 3644232153Smm */ 3645232153Smm count = 0; 3646232153Smm while (count < length && *s) { 3647232153Smm if (bigendian) 3648232153Smm archive_be16enc(u16, *s); 3649232153Smm else 3650232153Smm archive_le16enc(u16, *s); 3651232153Smm u16 += 2; 3652232153Smm s++; 3653232153Smm count++; 3654232153Smm } 3655232153Smm as16->length += count << 1; 3656232153Smm as16->s[as16->length] = 0; 3657232153Smm as16->s[as16->length+1] = 0; 3658232153Smm return (0); 3659232153Smm } 3660232153Smm do { 3661232153Smm count = MultiByteToWideChar(sc->from_cp, 3662248616Smm MB_PRECOMPOSED, s, (int)length, (LPWSTR)u16, (int)avail>>1); 3663302294Smm /* Exit loop if we succeeded */ 3664302294Smm if (count != 0 || 3665302294Smm GetLastError() != ERROR_INSUFFICIENT_BUFFER) { 3666302294Smm break; 3667232153Smm } 3668302294Smm /* Expand buffer and try again */ 3669302294Smm count = MultiByteToWideChar(sc->from_cp, 3670302294Smm MB_PRECOMPOSED, s, (int)length, NULL, 0); 3671302294Smm if (archive_string_ensure(as16, (count +1) * 2) 3672302294Smm == NULL) 3673302294Smm return (-1); 3674302294Smm u16 = as16->s + as16->length; 3675302294Smm avail = as16->buffer_length - 2; 3676302294Smm } while (1); 3677232153Smm as16->length += count * 2; 3678232153Smm as16->s[as16->length] = 0; 3679232153Smm as16->s[as16->length+1] = 0; 3680232153Smm if (count == 0) 3681232153Smm return (-1); 3682232153Smm 3683232153Smm if (is_big_endian()) { 3684232153Smm if (!bigendian) { 3685232153Smm while (count > 0) { 3686232153Smm uint16_t v = archive_be16dec(u16); 3687232153Smm archive_le16enc(u16, v); 3688232153Smm u16 += 2; 3689232153Smm count--; 3690232153Smm } 3691232153Smm } 3692232153Smm } else { 3693232153Smm if (bigendian) { 3694232153Smm while (count > 0) { 3695232153Smm uint16_t v = archive_le16dec(u16); 3696232153Smm archive_be16enc(u16, v); 3697232153Smm u16 += 2; 3698232153Smm count--; 3699232153Smm } 3700232153Smm } 3701232153Smm } 3702232153Smm return (0); 3703232153Smm} 3704232153Smm 3705232153Smmstatic int 3706238856Smmwin_strncat_to_utf16be(struct archive_string *as16, const void *_p, 3707238856Smm size_t length, struct archive_string_conv *sc) 3708232153Smm{ 3709232153Smm return (win_strncat_to_utf16(as16, _p, length, sc, 1)); 3710232153Smm} 3711232153Smm 3712232153Smmstatic int 3713238856Smmwin_strncat_to_utf16le(struct archive_string *as16, const void *_p, 3714238856Smm size_t length, struct archive_string_conv *sc) 3715232153Smm{ 3716232153Smm return (win_strncat_to_utf16(as16, _p, length, sc, 0)); 3717232153Smm} 3718232153Smm 3719232153Smm#endif /* _WIN32 && !__CYGWIN__ */ 3720232153Smm 3721232153Smm/* 3722232153Smm * Do the best effort for conversions. 3723232153Smm * We cannot handle UTF-16BE character-set without such iconv, 3724232153Smm * but there is a chance if a string consists just ASCII code or 3725232153Smm * a current locale is UTF-8. 3726232153Smm */ 3727232153Smm 3728232153Smm/* 3729232153Smm * Convert a UTF-16BE string to current locale and copy the result. 3730311041Smm * Return -1 if conversion fails. 3731232153Smm */ 3732232153Smmstatic int 3733232153Smmbest_effort_strncat_from_utf16(struct archive_string *as, const void *_p, 3734232153Smm size_t bytes, struct archive_string_conv *sc, int be) 3735232153Smm{ 3736232153Smm const char *utf16 = (const char *)_p; 3737232153Smm char *mbs; 3738232153Smm uint32_t uc; 3739232153Smm int n, ret; 3740232153Smm 3741232153Smm (void)sc; /* UNUSED */ 3742232153Smm /* 3743232153Smm * Other case, we should do the best effort. 3744232153Smm * If all character are ASCII(<0x7f), we can convert it. 3745232153Smm * if not , we set a alternative character and return -1. 3746232153Smm */ 3747232153Smm ret = 0; 3748232153Smm if (archive_string_ensure(as, as->length + bytes +1) == NULL) 3749232153Smm return (-1); 3750232153Smm mbs = as->s + as->length; 3751232153Smm 3752232153Smm while ((n = utf16_to_unicode(&uc, utf16, bytes, be)) != 0) { 3753232153Smm if (n < 0) { 3754232153Smm n *= -1; 3755232153Smm ret = -1; 3756232153Smm } 3757232153Smm bytes -= n; 3758232153Smm utf16 += n; 3759232153Smm 3760232153Smm if (uc > 127) { 3761232153Smm /* We cannot handle it. */ 3762232153Smm *mbs++ = '?'; 3763232153Smm ret = -1; 3764232153Smm } else 3765232153Smm *mbs++ = (char)uc; 3766232153Smm } 3767232153Smm as->length = mbs - as->s; 3768232153Smm as->s[as->length] = '\0'; 3769232153Smm return (ret); 3770232153Smm} 3771232153Smm 3772232153Smmstatic int 3773232153Smmbest_effort_strncat_from_utf16be(struct archive_string *as, const void *_p, 3774232153Smm size_t bytes, struct archive_string_conv *sc) 3775232153Smm{ 3776232153Smm return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 1)); 3777232153Smm} 3778232153Smm 3779232153Smmstatic int 3780232153Smmbest_effort_strncat_from_utf16le(struct archive_string *as, const void *_p, 3781232153Smm size_t bytes, struct archive_string_conv *sc) 3782232153Smm{ 3783232153Smm return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 0)); 3784232153Smm} 3785232153Smm 3786232153Smm/* 3787232153Smm * Convert a current locale string to UTF-16BE/LE and copy the result. 3788311041Smm * Return -1 if conversion fails. 3789232153Smm */ 3790232153Smmstatic int 3791232153Smmbest_effort_strncat_to_utf16(struct archive_string *as16, const void *_p, 3792232153Smm size_t length, struct archive_string_conv *sc, int bigendian) 3793232153Smm{ 3794232153Smm const char *s = (const char *)_p; 3795232153Smm char *utf16; 3796232153Smm size_t remaining; 3797232153Smm int ret; 3798232153Smm 3799232153Smm (void)sc; /* UNUSED */ 3800232153Smm /* 3801232153Smm * Other case, we should do the best effort. 3802232153Smm * If all character are ASCII(<0x7f), we can convert it. 3803232153Smm * if not , we set a alternative character and return -1. 3804232153Smm */ 3805232153Smm ret = 0; 3806232153Smm remaining = length; 3807232153Smm 3808232153Smm if (archive_string_ensure(as16, 3809232153Smm as16->length + (length + 1) * 2) == NULL) 3810232153Smm return (-1); 3811232153Smm 3812232153Smm utf16 = as16->s + as16->length; 3813232153Smm while (remaining--) { 3814232153Smm unsigned c = *s++; 3815232153Smm if (c > 127) { 3816232153Smm /* We cannot handle it. */ 3817232153Smm c = UNICODE_R_CHAR; 3818232153Smm ret = -1; 3819232153Smm } 3820232153Smm if (bigendian) 3821232153Smm archive_be16enc(utf16, c); 3822232153Smm else 3823232153Smm archive_le16enc(utf16, c); 3824232153Smm utf16 += 2; 3825232153Smm } 3826232153Smm as16->length = utf16 - as16->s; 3827232153Smm as16->s[as16->length] = 0; 3828232153Smm as16->s[as16->length+1] = 0; 3829232153Smm return (ret); 3830232153Smm} 3831232153Smm 3832232153Smmstatic int 3833232153Smmbest_effort_strncat_to_utf16be(struct archive_string *as16, const void *_p, 3834232153Smm size_t length, struct archive_string_conv *sc) 3835232153Smm{ 3836232153Smm return (best_effort_strncat_to_utf16(as16, _p, length, sc, 1)); 3837232153Smm} 3838232153Smm 3839232153Smmstatic int 3840232153Smmbest_effort_strncat_to_utf16le(struct archive_string *as16, const void *_p, 3841232153Smm size_t length, struct archive_string_conv *sc) 3842232153Smm{ 3843232153Smm return (best_effort_strncat_to_utf16(as16, _p, length, sc, 0)); 3844232153Smm} 3845232153Smm 3846232153Smm 3847232153Smm/* 3848232153Smm * Multistring operations. 3849232153Smm */ 3850232153Smm 3851232153Smmvoid 3852232153Smmarchive_mstring_clean(struct archive_mstring *aes) 3853232153Smm{ 3854232153Smm archive_wstring_free(&(aes->aes_wcs)); 3855232153Smm archive_string_free(&(aes->aes_mbs)); 3856232153Smm archive_string_free(&(aes->aes_utf8)); 3857232153Smm archive_string_free(&(aes->aes_mbs_in_locale)); 3858232153Smm aes->aes_set = 0; 3859232153Smm} 3860232153Smm 3861232153Smmvoid 3862232153Smmarchive_mstring_copy(struct archive_mstring *dest, struct archive_mstring *src) 3863232153Smm{ 3864232153Smm dest->aes_set = src->aes_set; 3865232153Smm archive_string_copy(&(dest->aes_mbs), &(src->aes_mbs)); 3866232153Smm archive_string_copy(&(dest->aes_utf8), &(src->aes_utf8)); 3867232153Smm archive_wstring_copy(&(dest->aes_wcs), &(src->aes_wcs)); 3868232153Smm} 3869232153Smm 3870232153Smmint 3871232153Smmarchive_mstring_get_utf8(struct archive *a, struct archive_mstring *aes, 3872232153Smm const char **p) 3873232153Smm{ 3874232153Smm struct archive_string_conv *sc; 3875232153Smm int r; 3876232153Smm 3877232153Smm /* If we already have a UTF8 form, return that immediately. */ 3878232153Smm if (aes->aes_set & AES_SET_UTF8) { 3879232153Smm *p = aes->aes_utf8.s; 3880232153Smm return (0); 3881232153Smm } 3882232153Smm 3883232153Smm *p = NULL; 3884368707Smm /* Try converting WCS to MBS first if MBS does not exist yet. */ 3885368707Smm if ((aes->aes_set & AES_SET_MBS) == 0) { 3886368707Smm const char *pm; /* unused */ 3887368707Smm archive_mstring_get_mbs(a, aes, &pm); /* ignore errors, we'll handle it later */ 3888368707Smm } 3889232153Smm if (aes->aes_set & AES_SET_MBS) { 3890232153Smm sc = archive_string_conversion_to_charset(a, "UTF-8", 1); 3891232153Smm if (sc == NULL) 3892232153Smm return (-1);/* Couldn't allocate memory for sc. */ 3893299529Smm r = archive_strncpy_l(&(aes->aes_utf8), aes->aes_mbs.s, 3894232153Smm aes->aes_mbs.length, sc); 3895232153Smm if (a == NULL) 3896232153Smm free_sconv_object(sc); 3897232153Smm if (r == 0) { 3898232153Smm aes->aes_set |= AES_SET_UTF8; 3899232153Smm *p = aes->aes_utf8.s; 3900232153Smm return (0);/* success. */ 3901232153Smm } else 3902232153Smm return (-1);/* failure. */ 3903232153Smm } 3904232153Smm return (0);/* success. */ 3905232153Smm} 3906232153Smm 3907232153Smmint 3908232153Smmarchive_mstring_get_mbs(struct archive *a, struct archive_mstring *aes, 3909232153Smm const char **p) 3910232153Smm{ 3911368707Smm struct archive_string_conv *sc; 3912232153Smm int r, ret = 0; 3913232153Smm 3914232153Smm /* If we already have an MBS form, return that immediately. */ 3915232153Smm if (aes->aes_set & AES_SET_MBS) { 3916232153Smm *p = aes->aes_mbs.s; 3917232153Smm return (ret); 3918232153Smm } 3919232153Smm 3920232153Smm *p = NULL; 3921232153Smm /* If there's a WCS form, try converting with the native locale. */ 3922232153Smm if (aes->aes_set & AES_SET_WCS) { 3923232153Smm archive_string_empty(&(aes->aes_mbs)); 3924232153Smm r = archive_string_append_from_wcs(&(aes->aes_mbs), 3925232153Smm aes->aes_wcs.s, aes->aes_wcs.length); 3926232153Smm *p = aes->aes_mbs.s; 3927232153Smm if (r == 0) { 3928232153Smm aes->aes_set |= AES_SET_MBS; 3929232153Smm return (ret); 3930232153Smm } else 3931232153Smm ret = -1; 3932232153Smm } 3933232153Smm 3934368707Smm /* If there's a UTF-8 form, try converting with the native locale. */ 3935368707Smm if (aes->aes_set & AES_SET_UTF8) { 3936368707Smm archive_string_empty(&(aes->aes_mbs)); 3937368707Smm sc = archive_string_conversion_from_charset(a, "UTF-8", 1); 3938368707Smm if (sc == NULL) 3939368707Smm return (-1);/* Couldn't allocate memory for sc. */ 3940368707Smm r = archive_strncpy_l(&(aes->aes_mbs), 3941368707Smm aes->aes_utf8.s, aes->aes_utf8.length, sc); 3942368707Smm if (a == NULL) 3943368707Smm free_sconv_object(sc); 3944368707Smm *p = aes->aes_mbs.s; 3945368707Smm if (r == 0) { 3946368707Smm aes->aes_set |= AES_SET_MBS; 3947368707Smm ret = 0;/* success; overwrite previous error. */ 3948368707Smm } else 3949368707Smm ret = -1;/* failure. */ 3950368707Smm } 3951232153Smm return (ret); 3952232153Smm} 3953232153Smm 3954232153Smmint 3955232153Smmarchive_mstring_get_wcs(struct archive *a, struct archive_mstring *aes, 3956232153Smm const wchar_t **wp) 3957232153Smm{ 3958232153Smm int r, ret = 0; 3959232153Smm 3960232153Smm (void)a;/* UNUSED */ 3961232153Smm /* Return WCS form if we already have it. */ 3962232153Smm if (aes->aes_set & AES_SET_WCS) { 3963232153Smm *wp = aes->aes_wcs.s; 3964232153Smm return (ret); 3965232153Smm } 3966232153Smm 3967232153Smm *wp = NULL; 3968368707Smm /* Try converting UTF8 to MBS first if MBS does not exist yet. */ 3969368707Smm if ((aes->aes_set & AES_SET_MBS) == 0) { 3970368707Smm const char *p; /* unused */ 3971368707Smm archive_mstring_get_mbs(a, aes, &p); /* ignore errors, we'll handle it later */ 3972368707Smm } 3973232153Smm /* Try converting MBS to WCS using native locale. */ 3974232153Smm if (aes->aes_set & AES_SET_MBS) { 3975232153Smm archive_wstring_empty(&(aes->aes_wcs)); 3976232153Smm r = archive_wstring_append_from_mbs(&(aes->aes_wcs), 3977232153Smm aes->aes_mbs.s, aes->aes_mbs.length); 3978232153Smm if (r == 0) { 3979232153Smm aes->aes_set |= AES_SET_WCS; 3980232153Smm *wp = aes->aes_wcs.s; 3981232153Smm } else 3982232153Smm ret = -1;/* failure. */ 3983232153Smm } 3984232153Smm return (ret); 3985232153Smm} 3986232153Smm 3987232153Smmint 3988368707Smmarchive_mstring_get_mbs_l(struct archive *a, struct archive_mstring *aes, 3989232153Smm const char **p, size_t *length, struct archive_string_conv *sc) 3990232153Smm{ 3991232153Smm int r, ret = 0; 3992232153Smm 3993368707Smm (void)r; /* UNUSED */ 3994232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 3995232153Smm /* 3996313570Smm * Internationalization programming on Windows must use Wide 3997232153Smm * characters because Windows platform cannot make locale UTF-8. 3998232153Smm */ 3999232153Smm if (sc != NULL && (aes->aes_set & AES_SET_WCS) != 0) { 4000232153Smm archive_string_empty(&(aes->aes_mbs_in_locale)); 4001232153Smm r = archive_string_append_from_wcs_in_codepage( 4002232153Smm &(aes->aes_mbs_in_locale), aes->aes_wcs.s, 4003232153Smm aes->aes_wcs.length, sc); 4004232153Smm if (r == 0) { 4005232153Smm *p = aes->aes_mbs_in_locale.s; 4006232153Smm if (length != NULL) 4007232153Smm *length = aes->aes_mbs_in_locale.length; 4008232153Smm return (0); 4009232153Smm } else if (errno == ENOMEM) 4010232153Smm return (-1); 4011232153Smm else 4012232153Smm ret = -1; 4013232153Smm } 4014228753Smm#endif 4015232153Smm 4016368707Smm /* If there is not an MBS form but there is a WCS or UTF8 form, try converting 4017232153Smm * with the native locale to be used for translating it to specified 4018232153Smm * character-set. */ 4019368707Smm if ((aes->aes_set & AES_SET_MBS) == 0) { 4020368707Smm const char *pm; /* unused */ 4021368707Smm archive_mstring_get_mbs(a, aes, &pm); /* ignore errors, we'll handle it later */ 4022232153Smm } 4023232153Smm /* If we already have an MBS form, use it to be translated to 4024232153Smm * specified character-set. */ 4025232153Smm if (aes->aes_set & AES_SET_MBS) { 4026232153Smm if (sc == NULL) { 4027232153Smm /* Conversion is unneeded. */ 4028232153Smm *p = aes->aes_mbs.s; 4029232153Smm if (length != NULL) 4030232153Smm *length = aes->aes_mbs.length; 4031232153Smm return (0); 4032232153Smm } 4033238856Smm ret = archive_strncpy_l(&(aes->aes_mbs_in_locale), 4034232153Smm aes->aes_mbs.s, aes->aes_mbs.length, sc); 4035232153Smm *p = aes->aes_mbs_in_locale.s; 4036232153Smm if (length != NULL) 4037232153Smm *length = aes->aes_mbs_in_locale.length; 4038232153Smm } else { 4039232153Smm *p = NULL; 4040232153Smm if (length != NULL) 4041232153Smm *length = 0; 4042232153Smm } 4043232153Smm return (ret); 4044228753Smm} 4045228753Smm 4046232153Smmint 4047232153Smmarchive_mstring_copy_mbs(struct archive_mstring *aes, const char *mbs) 4048232153Smm{ 4049232153Smm if (mbs == NULL) { 4050232153Smm aes->aes_set = 0; 4051232153Smm return (0); 4052232153Smm } 4053232153Smm return (archive_mstring_copy_mbs_len(aes, mbs, strlen(mbs))); 4054232153Smm} 4055232153Smm 4056232153Smmint 4057232153Smmarchive_mstring_copy_mbs_len(struct archive_mstring *aes, const char *mbs, 4058232153Smm size_t len) 4059232153Smm{ 4060232153Smm if (mbs == NULL) { 4061232153Smm aes->aes_set = 0; 4062232153Smm return (0); 4063232153Smm } 4064232153Smm aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ 4065232153Smm archive_strncpy(&(aes->aes_mbs), mbs, len); 4066232153Smm archive_string_empty(&(aes->aes_utf8)); 4067232153Smm archive_wstring_empty(&(aes->aes_wcs)); 4068232153Smm return (0); 4069232153Smm} 4070232153Smm 4071232153Smmint 4072232153Smmarchive_mstring_copy_wcs(struct archive_mstring *aes, const wchar_t *wcs) 4073232153Smm{ 4074238856Smm return archive_mstring_copy_wcs_len(aes, wcs, 4075238856Smm wcs == NULL ? 0 : wcslen(wcs)); 4076232153Smm} 4077232153Smm 4078232153Smmint 4079299529Smmarchive_mstring_copy_utf8(struct archive_mstring *aes, const char *utf8) 4080299529Smm{ 4081299529Smm if (utf8 == NULL) { 4082299529Smm aes->aes_set = 0; 4083346104Smm return (0); 4084299529Smm } 4085299529Smm aes->aes_set = AES_SET_UTF8; 4086299529Smm archive_string_empty(&(aes->aes_mbs)); 4087299529Smm archive_string_empty(&(aes->aes_wcs)); 4088299529Smm archive_strncpy(&(aes->aes_utf8), utf8, strlen(utf8)); 4089299529Smm return (int)strlen(utf8); 4090299529Smm} 4091299529Smm 4092299529Smmint 4093232153Smmarchive_mstring_copy_wcs_len(struct archive_mstring *aes, const wchar_t *wcs, 4094232153Smm size_t len) 4095232153Smm{ 4096232153Smm if (wcs == NULL) { 4097232153Smm aes->aes_set = 0; 4098346104Smm return (0); 4099232153Smm } 4100232153Smm aes->aes_set = AES_SET_WCS; /* Only WCS form set. */ 4101232153Smm archive_string_empty(&(aes->aes_mbs)); 4102232153Smm archive_string_empty(&(aes->aes_utf8)); 4103232153Smm archive_wstrncpy(&(aes->aes_wcs), wcs, len); 4104232153Smm return (0); 4105232153Smm} 4106232153Smm 4107232153Smmint 4108232153Smmarchive_mstring_copy_mbs_len_l(struct archive_mstring *aes, 4109232153Smm const char *mbs, size_t len, struct archive_string_conv *sc) 4110232153Smm{ 4111232153Smm int r; 4112232153Smm 4113232153Smm if (mbs == NULL) { 4114232153Smm aes->aes_set = 0; 4115232153Smm return (0); 4116232153Smm } 4117232153Smm archive_string_empty(&(aes->aes_mbs)); 4118232153Smm archive_wstring_empty(&(aes->aes_wcs)); 4119232153Smm archive_string_empty(&(aes->aes_utf8)); 4120232153Smm#if defined(_WIN32) && !defined(__CYGWIN__) 4121232153Smm /* 4122313570Smm * Internationalization programming on Windows must use Wide 4123232153Smm * characters because Windows platform cannot make locale UTF-8. 4124232153Smm */ 4125232153Smm if (sc == NULL) { 4126232153Smm if (archive_string_append(&(aes->aes_mbs), 4127232153Smm mbs, mbsnbytes(mbs, len)) == NULL) { 4128232153Smm aes->aes_set = 0; 4129232153Smm r = -1; 4130232153Smm } else { 4131232153Smm aes->aes_set = AES_SET_MBS; 4132232153Smm r = 0; 4133232153Smm } 4134232153Smm#if defined(HAVE_ICONV) 4135232153Smm } else if (sc != NULL && sc->cd_w != (iconv_t)-1) { 4136232153Smm /* 4137232153Smm * This case happens only when MultiByteToWideChar() cannot 4138232153Smm * handle sc->from_cp, and we have to iconv in order to 4139232153Smm * translate character-set to wchar_t,UTF-16. 4140232153Smm */ 4141232153Smm iconv_t cd = sc->cd; 4142232153Smm unsigned from_cp; 4143232153Smm int flag; 4144232153Smm 4145232153Smm /* 4146232153Smm * Translate multi-bytes from some character-set to UTF-8. 4147232153Smm */ 4148232153Smm sc->cd = sc->cd_w; 4149238856Smm r = archive_strncpy_l(&(aes->aes_utf8), mbs, len, sc); 4150232153Smm sc->cd = cd; 4151232153Smm if (r != 0) { 4152232153Smm aes->aes_set = 0; 4153232153Smm return (r); 4154232153Smm } 4155232153Smm aes->aes_set = AES_SET_UTF8; 4156232153Smm 4157232153Smm /* 4158232153Smm * Append the UTF-8 string into wstring. 4159232153Smm */ 4160232153Smm flag = sc->flag; 4161232153Smm sc->flag &= ~(SCONV_NORMALIZATION_C 4162232153Smm | SCONV_TO_UTF16| SCONV_FROM_UTF16); 4163232153Smm from_cp = sc->from_cp; 4164232153Smm sc->from_cp = CP_UTF8; 4165232153Smm r = archive_wstring_append_from_mbs_in_codepage(&(aes->aes_wcs), 4166232153Smm aes->aes_utf8.s, aes->aes_utf8.length, sc); 4167232153Smm sc->flag = flag; 4168232153Smm sc->from_cp = from_cp; 4169232153Smm if (r == 0) 4170232153Smm aes->aes_set |= AES_SET_WCS; 4171232153Smm#endif 4172232153Smm } else { 4173232153Smm r = archive_wstring_append_from_mbs_in_codepage( 4174232153Smm &(aes->aes_wcs), mbs, len, sc); 4175232153Smm if (r == 0) 4176232153Smm aes->aes_set = AES_SET_WCS; 4177232153Smm else 4178232153Smm aes->aes_set = 0; 4179232153Smm } 4180232153Smm#else 4181238856Smm r = archive_strncpy_l(&(aes->aes_mbs), mbs, len, sc); 4182232153Smm if (r == 0) 4183232153Smm aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ 4184232153Smm else 4185232153Smm aes->aes_set = 0; 4186232153Smm#endif 4187232153Smm return (r); 4188232153Smm} 4189232153Smm 4190232153Smm/* 4191232153Smm * The 'update' form tries to proactively update all forms of 4192232153Smm * this string (WCS and MBS) and returns an error if any of 4193232153Smm * them fail. This is used by the 'pax' handler, for instance, 4194232153Smm * to detect and report character-conversion failures early while 4195232153Smm * still allowing clients to get potentially useful values from 4196232153Smm * the more tolerant lazy conversions. (get_mbs and get_wcs will 4197232153Smm * strive to give the user something useful, so you can get hopefully 4198232153Smm * usable values even if some of the character conversions are failing.) 4199232153Smm */ 4200232153Smmint 4201232153Smmarchive_mstring_update_utf8(struct archive *a, struct archive_mstring *aes, 4202232153Smm const char *utf8) 4203232153Smm{ 4204232153Smm struct archive_string_conv *sc; 4205232153Smm int r; 4206232153Smm 4207232153Smm if (utf8 == NULL) { 4208232153Smm aes->aes_set = 0; 4209232153Smm return (0); /* Succeeded in clearing everything. */ 4210232153Smm } 4211232153Smm 4212232153Smm /* Save the UTF8 string. */ 4213232153Smm archive_strcpy(&(aes->aes_utf8), utf8); 4214232153Smm 4215232153Smm /* Empty the mbs and wcs strings. */ 4216232153Smm archive_string_empty(&(aes->aes_mbs)); 4217232153Smm archive_wstring_empty(&(aes->aes_wcs)); 4218232153Smm 4219232153Smm aes->aes_set = AES_SET_UTF8; /* Only UTF8 is set now. */ 4220232153Smm 4221232153Smm /* Try converting UTF-8 to MBS, return false on failure. */ 4222232153Smm sc = archive_string_conversion_from_charset(a, "UTF-8", 1); 4223232153Smm if (sc == NULL) 4224232153Smm return (-1);/* Couldn't allocate memory for sc. */ 4225238856Smm r = archive_strcpy_l(&(aes->aes_mbs), utf8, sc); 4226232153Smm if (a == NULL) 4227232153Smm free_sconv_object(sc); 4228232153Smm if (r != 0) 4229232153Smm return (-1); 4230232153Smm aes->aes_set = AES_SET_UTF8 | AES_SET_MBS; /* Both UTF8 and MBS set. */ 4231232153Smm 4232232153Smm /* Try converting MBS to WCS, return false on failure. */ 4233232153Smm if (archive_wstring_append_from_mbs(&(aes->aes_wcs), aes->aes_mbs.s, 4234232153Smm aes->aes_mbs.length)) 4235232153Smm return (-1); 4236232153Smm aes->aes_set = AES_SET_UTF8 | AES_SET_WCS | AES_SET_MBS; 4237232153Smm 4238232153Smm /* All conversions succeeded. */ 4239232153Smm return (0); 4240232153Smm} 4241