/* ------------------------------------------------------------------------ */
#include "lha.h"
+#if !STRCHR_8BIT_CLEAN
+/* should use 8 bit clean version */
+#undef strchr
+#undef strrchr
+#define strchr xstrchr
+#define strrchr xstrrchr
+#endif
+
/* ------------------------------------------------------------------------ */
static char *get_ptr;
char *optional_system_delim = NULL;
int optional_filename_case = NONE;
-#if defined(__CYGWIN__) || defined(__MINGW32__) || defined(__hpux)
-/* Cygwin, HP-UX and other UNIX are able to use SJIS as native code. */
-int default_system_kanji_code = CODE_SJIS;
+#ifdef MULTIBYTE_FILENAME
+int default_system_kanji_code = MULTIBYTE_FILENAME;
#else
-int default_system_kanji_code = CODE_EUC;
+int default_system_kanji_code = NONE;
#endif
+
/* ------------------------------------------------------------------------ */
int
calc_sum(p, len)
put_byte(v >> 24);
}
+#if 0 /* no use */
/* ------------------------------------------------------------------------ */
static void
msdos_to_unix_filename(name, len)
name[i] = toupper(name[i]);
}
}
+#endif /* 0 */
/* added by Koji Arai */
static void
-filename_conv(name, len, size,
- from_code, to_code,
- from_delim, to_delim,
- case_to)
+convert_filename(name, len, size,
+ from_code, to_code,
+ from_delim, to_delim,
+ case_to)
register char *name;
register int len;
register int size;
{
register int i;
+#ifdef MULTIBYTE_FILENAME
+ char tmp[256]; /* 256 is sizeof(LzHeader.name) */
+
+ if (from_code == CODE_SJIS && to_code == CODE_UTF8) {
+ for (i = 0; i < len; i++)
+ if ((unsigned char)name[i] == LHA_PATHSEP) name[i] = '/';
+ sjis_to_utf8(tmp, name, sizeof(tmp));
+ strncpy(name, tmp, size);
+ name[size-1] = 0;
+ len = strlen(name);
+ for (i = 0; i < len; i++)
+ if (name[i] == '/') name[i] = LHA_PATHSEP;
+ from_code = CODE_UTF8;
+ }
+ else if (from_code == CODE_UTF8 && to_code == CODE_SJIS) {
+ for (i = 0; i < len; i++)
+ if ((unsigned char)name[i] == LHA_PATHSEP) name[i] = '/';
+ utf8_to_sjis(tmp, name, sizeof(tmp));
+ strncpy(name, tmp, size);
+ name[size-1] = 0;
+ len = strlen(name);
+ for (i = 0; i < len; i++)
+ if (name[i] == '/') name[i] = LHA_PATHSEP;
+ from_code = CODE_SJIS;
+ }
+#endif
for (i = 0; i < len; i ++) {
#ifdef MULTIBYTE_FILENAME
}
#endif
#endif /* defined(HAVE_FTIME) || defined(HAVE_GETTIMEOFDAY) ||
- * defined(HAVE_TZSET) */
+ * defined(HAVE_TZSET) */
/* ------------------------------------------------------------------------ */
#ifdef NOT_USED
char *system_delim = "";
int filename_case = NONE;
- bzero(hdr, sizeof(LzHeader));
+ memset(hdr, 0, sizeof(LzHeader));
if (((header_size = getc(fp)) == EOF) || (header_size == 0)) {
return FALSE; /* finish */
}
if (hdr->header_level >= 3) {
- fatal_error("Unknown level header");
+ fatal_error("Unknown level header (level %d)", hdr->header_level);
return FALSE;
}
} else {
hdr->header_size = header_size;
}
- bcopy(data + I_METHOD, hdr->method, METHOD_TYPE_STRAGE);
+ memcpy(hdr->method, data + I_METHOD, METHOD_TYPE_STRAGE);
setup_get(data + I_PACKED_SIZE);
hdr->packed_size = get_longword();
hdr->original_size = get_longword();
if ((hdr->header_level = get_byte()) != 2) {
if (calc_sum(data + I_METHOD, header_size) != checksum)
- warning("Checksum error (LHarc file?)", "");
+ warning("Checksum error (LHarc file?)");
name_length = get_byte();
for (i = 0; i < name_length; i++)
hdr->name[i] = (char) get_byte();
hdr->extend_type = EXTEND_GENERIC;
hdr->has_crc = FALSE;
} else {
- fatal_error("Unkonwn header (lha file?)");
- return FALSE;
+ error("Unkonwn header (lha file?)");
+ exit(1);
}
} else {
hdr->has_crc = TRUE;
}
while (extend_size-- > 0)
dmy = get_byte();
- if (hdr->extend_type == EXTEND_UNIX)
- return TRUE;
} else if (hdr->header_level == 1) {
hdr->has_crc = TRUE;
extend_size = header_size - name_length-25;
switch (hdr->extend_type) {
case EXTEND_MSDOS:
- archive_delim = "\xff\\";
+ archive_delim = "\377\\";
+ /* `\' is for level 0 header and broken archive. */
system_delim = "//";
filename_case = noconvertcase ? NONE : TO_LOWER;
case EXTEND_XOSK:
#endif
case EXTEND_UNIX:
- archive_delim = "\xff";
+ archive_delim = "\377\\";
+ /* `\' is for level 0 header and broken archive. */
system_delim = "//";
filename_case = NONE;
break;
case EXTEND_MACOS:
- archive_delim = "\xff/:";
- system_delim = "/:/";
+ archive_delim = "\377/:\\";
+ /* `\' is for level 0 header and broken archive. */
+ system_delim = "/://";
filename_case = NONE;
hdr->unix_last_modified_stamp =
break;
default:
- archive_delim = "\xff\\";
+ archive_delim = "\377\\";
+ /* `\' is for level 0 header and broken archive. */
system_delim = "//";
filename_case = noconvertcase ? NONE : TO_LOWER;
- /* pending: if small letter is included in filename,
+ /* FIXME: if small letter is included in filename,
the generic_to_unix_filename() do not case conversion,
but this code does not consider it. */
name_length += dir_length;
}
- filename_conv(hdr->name, name_length, sizeof(hdr->name),
- archive_kanji_code,
- system_kanji_code,
- archive_delim, system_delim, filename_case);
+ convert_filename(hdr->name, name_length, sizeof(hdr->name),
+ archive_kanji_code,
+ system_kanji_code,
+ archive_delim, system_delim, filename_case);
return TRUE;
}
{
int len;
- int system_kanji_code = default_system_kanji_code;
- char *archive_delim = "";
- char *system_delim = "";
- int filename_case = NONE;
-
memset(hdr, 0, sizeof(LzHeader));
- if (optional_system_kanji_code)
- system_kanji_code = optional_system_kanji_code;
-
if (compress_method == LZHUFF5_METHOD_NUM) /* Changed N.Watazaki */
- bcopy(LZHUFF5_METHOD, hdr->method, METHOD_TYPE_STRAGE);
+ memcpy(hdr->method, LZHUFF5_METHOD, METHOD_TYPE_STRAGE);
else if (compress_method)
- bcopy(LZHUFF1_METHOD, hdr->method, METHOD_TYPE_STRAGE);
+ memcpy(hdr->method, LZHUFF1_METHOD, METHOD_TYPE_STRAGE);
else
- bcopy(LZHUFF0_METHOD, hdr->method, METHOD_TYPE_STRAGE);
+ memcpy(hdr->method, LZHUFF0_METHOD, METHOD_TYPE_STRAGE);
hdr->packed_size = 0;
hdr->original_size = v_stat->st_size;
#endif
#endif /* INCLUDE_OWNER_NAME_IN_HEADER */
if (is_directory(v_stat)) {
- bcopy(LZHDIRS_METHOD, hdr->method, METHOD_TYPE_STRAGE);
+ memcpy(hdr->method, LZHDIRS_METHOD, METHOD_TYPE_STRAGE);
hdr->attribute = GENERIC_DIRECTORY_ATTRIBUTE;
hdr->original_size = 0;
if (len > 0 && hdr->name[len - 1] != '/')
#ifdef S_IFLNK
if (is_symlink(v_stat)) {
- char lkname[257];
+ char lkname[256]; /* FIXME: no enough space */
int len;
- bcopy(LZHDIRS_METHOD, hdr->method, METHOD_TYPE_STRAGE);
+ memcpy(hdr->method, LZHDIRS_METHOD, METHOD_TYPE_STRAGE);
hdr->attribute = GENERIC_DIRECTORY_ATTRIBUTE;
hdr->original_size = 0;
- len = readlink(name, lkname, 256);
- lkname[len] = (char)'\0';
- sprintf(hdr->name, "%s|%s", hdr->name, lkname);
+ len = readlink(name, lkname, sizeof(lkname));
+ if (xsnprintf(hdr->name, sizeof(hdr->name),
+ "%s|%.*s", hdr->name, len, lkname) == -1)
+ error("file name is too long (%s -> %.*s)", hdr->name, len, lkname);
}
#endif
-
- if (generic_format) {
- filename_case = TO_UPPER;
- archive_delim = "\\";
- }
-
- filename_conv(hdr->name, len, sizeof(hdr->name),
- system_kanji_code,
- system_kanji_code, /* no change code */
- system_delim, archive_delim, filename_case);
}
/* ------------------------------------------------------------------------ */
char data[LZHEADER_STRAGE];
char *p;
char *headercrc_ptr;
+
int archive_kanji_code = CODE_SJIS;
int system_kanji_code = default_system_kanji_code;
+ char *archive_delim = "\377";
+ char *system_delim = "/";
+ int filename_case = NONE;
+ char lzname[256];
if (optional_archive_kanji_code)
archive_kanji_code = optional_archive_kanji_code;
if (optional_system_kanji_code)
system_kanji_code = optional_system_kanji_code;
- bzero(data, LZHEADER_STRAGE);
- bcopy(hdr->method, data + I_METHOD, METHOD_TYPE_STRAGE);
+ memset(data, 0, LZHEADER_STRAGE);
+ memcpy(data + I_METHOD, hdr->method, METHOD_TYPE_STRAGE);
setup_put(data + I_PACKED_SIZE);
put_longword(hdr->packed_size);
put_longword(hdr->original_size);
put_byte(hdr->header_level);
- filename_conv(hdr->name, strlen(hdr->name), sizeof(hdr->name),
- system_kanji_code,
- archive_kanji_code, /* no change code */
- "\xff\\/", "\xff\xff\xff", NONE);
+ if (generic_format)
+ filename_case = TO_UPPER;
+
+ if (hdr->header_level == HEADER_LEVEL0) {
+ archive_delim = "\\";
+ }
+
+ strncpy(lzname, hdr->name, sizeof(lzname));
+ convert_filename(lzname, strlen(lzname), sizeof(lzname),
+ system_kanji_code,
+ archive_kanji_code,
+ system_delim, archive_delim, filename_case);
if (hdr->header_level != HEADER_LEVEL2) {
- if (p = (char *) strrchr(hdr->name, DELIM2))
- name_length = strlen(++p);
- else
- name_length = strlen(hdr->name);
+ if (hdr->header_level == HEADER_LEVEL0 ||
+ (p = strchr(lzname, LHA_PATHSEP)) == 0)
+ p = lzname;
+ else
+ ++p;
+ /* level 0 header: write pathname (contain the directory part) */
+ /* level 1 header: write filename (basename only) */
+ name_length = strlen(p);
put_byte(name_length);
- bcopy(p ? p : hdr->name, data + I_NAME, name_length);
+ memcpy(data + I_NAME, p, name_length);
setup_put(data + I_NAME + name_length);
}
put_byte(hdr->user[i]);
}
- if (p = (char *) strrchr(hdr->name, DELIM2)) {
+ if (p = strrchr(lzname, LHA_PATHSEP)) {
int i;
- name_length = p - hdr->name + 1;
+ name_length = p - lzname + 1;
put_word(name_length + 3);
put_byte(2); /* dirname */
for (i = 0; i < name_length; i++)
- put_byte(hdr->name[i]);
+ put_byte(lzname[i]);
}
} /* if generic .. */
data[I_HEADER_CHECKSUM] = calc_sum(data + I_METHOD, header_size);
} else { /* header level 2 */
int i;
- if (p = (char *) strrchr(hdr->name, DELIM2))
+ if (p = strrchr(lzname, LHA_PATHSEP))
name_length = strlen(++p);
else {
- p = hdr->name;
- name_length = strlen(hdr->name);
+ p = lzname;
+ name_length = strlen(lzname);
}
put_word(name_length + 3);
put_byte(1); /* filename */
put_word(hcrc);
}
- if (fwrite(data, sizeof(char), header_size + 2, nafp) == 0)
+ if (fwrite(data, header_size + 2, 1, nafp) == 0)
fatal_error("Cannot write to temporary file");
+}
- filename_conv(hdr->name, strlen(hdr->name), sizeof(hdr->name),
- archive_kanji_code,
- system_kanji_code,
- "\xff\\/", "///", NONE);
+#if MULTIBYTE_FILENAME
+
+#if defined(__APPLE__)
+
+#include <CoreFoundation/CFString.h>
+#include <CoreFoundation/CFStringEncodingExt.h>
+
+/* this is not need for Mac OS X v 10.2 later */
+enum {
+ kCFStringEncodingAllowLossyConversion = 1,
+ kCFStringEncodingBasicDirectionLeftToRight = (1 << 1),
+ kCFStringEncodingBasicDirectionRightToLeft = (1 << 2),
+ kCFStringEncodingSubstituteCombinings = (1 << 3),
+ kCFStringEncodingComposeCombinings = (1 << 4),
+ kCFStringEncodingIgnoreCombinings = (1 << 5),
+ kCFStringEncodingUseCanonical = (1 << 6),
+ kCFStringEncodingUseHFSPlusCanonical = (1 << 7),
+ kCFStringEncodingPrependBOM = (1 << 8),
+ kCFStringEncodingDisableCorporateArea = (1 << 9),
+ kCFStringEncodingASCIICompatibleConversion = (1 << 10),
+};
+
+static int
+ConvertEncodingToUTF8(const char* inCStr,
+ char* outUTF8Buffer,
+ int outUTF8BufferLength,
+ unsigned long scriptEncoding,
+ unsigned long flags)
+{
+ unsigned long unicodeChars;
+ unsigned long srcCharsUsed;
+ unsigned long usedByteLen = 0;
+ UniChar uniStr[512];
+ unsigned long cfResult;
+
+ cfResult = CFStringEncodingBytesToUnicode(scriptEncoding,
+ flags,
+ (char *)inCStr,
+ strlen(inCStr),
+ &srcCharsUsed,
+ uniStr,
+ 512,
+ &unicodeChars);
+ if (cfResult == 0) {
+ cfResult = CFStringEncodingUnicodeToBytes(kCFStringEncodingUTF8,
+ flags,
+ uniStr,
+ unicodeChars,
+ &srcCharsUsed,
+ (char*)outUTF8Buffer,
+ outUTF8BufferLength - 1,
+ &usedByteLen);
+ outUTF8Buffer[usedByteLen] = '\0';
+ }
+
+ return cfResult;
+}
+
+static int
+ConvertUTF8ToEncoding(const char* inUTF8Buf,
+ int inUTF8BufLength,
+ char* outCStrBuffer,
+ int outCStrBufferLength,
+ unsigned long scriptEncoding,
+ unsigned long flags)
+{
+ unsigned long unicodeChars;
+ unsigned long srcCharsUsed;
+ unsigned long usedByteLen = 0;
+ UniChar uniStr[256];
+ unsigned long cfResult;
+
+ cfResult = CFStringEncodingBytesToUnicode(kCFStringEncodingUTF8,
+ flags,
+ (char*)inUTF8Buf,
+ inUTF8BufLength,
+ &srcCharsUsed,
+ uniStr,
+ 255,
+ &unicodeChars);
+ if (cfResult == 0) {
+ cfResult = CFStringEncodingUnicodeToBytes(scriptEncoding,
+ flags,
+ uniStr,
+ unicodeChars,
+ &srcCharsUsed,
+ (char*)outCStrBuffer,
+ outCStrBufferLength - 1,
+ &usedByteLen);
+ outCStrBuffer[usedByteLen] = '\0';
+ }
+
+ return cfResult;
+}
+
+#elif HAVE_ICONV
+#include <iconv.h>
+
+static int
+ConvertEncodingByIconv(const char *src, char *dst, int dstsize,
+ const char *srcEnc, const char *dstEnc)
+{
+ iconv_t ic;
+ static char szTmpBuf[2048];
+ char *src_p;
+ char *dst_p;
+ size_t sLen;
+ size_t iLen;
+
+ dst_p = &szTmpBuf[0];
+ iLen = (size_t)sizeof(szTmpBuf)-1;
+ src_p = (char *)src;
+ sLen = (size_t)strlen(src);
+ memset(szTmpBuf, 0, sizeof(szTmpBuf));
+ memset(dst, 0, dstsize);
+
+ ic = iconv_open(dstEnc, srcEnc);
+ if (ic == (iconv_t)-1) {
+ error("iconv_open() failure");
+ return -1;
+ }
+
+ if (iconv(ic, &src_p, &sLen, &dst_p, &iLen) == (size_t)-1) {
+ error("iconv() failure");
+ iconv_close(ic);
+ return -1;
+ }
+
+ strncpy(dst, szTmpBuf, dstsize);
+
+ iconv_close(ic);
+
+ return 0;
+}
+#endif /* defined(__APPLE__) */
+
+char *
+sjis_to_utf8(char *dst, const char *src, size_t dstsize)
+{
+#if defined(__APPLE__)
+ dst[0] = '\0';
+ if (ConvertEncodingToUTF8(src, dst, dstsize,
+ kCFStringEncodingDOSJapanese,
+ kCFStringEncodingUseHFSPlusCanonical) == 0)
+ return dst;
+#elif HAVE_ICONV
+ if (ConvertEncodingByIconv(src, dst, dstsize, "SJIS", "UTF-8") != -1)
+ return dst;
+#else
+ error("not support utf-8 conversion");
+#endif
+
+ /* not supported */
+ if (dstsize < 1) return dst;
+ dst[dstsize-1] = 0;
+ return strncpy(dst, src, dstsize-1);
+}
+
+char *
+utf8_to_sjis(char *dst, const char *src, size_t dstsize)
+{
+#if defined(__APPLE__)
+ int srclen;
+
+ dst[0] = '\0';
+ srclen = strlen(src);
+ if (ConvertUTF8ToEncoding(src, srclen, dst, dstsize,
+ kCFStringEncodingDOSJapanese,
+ kCFStringEncodingUseHFSPlusCanonical) == 0)
+ return dst;
+#elif HAVE_ICONV
+ if (ConvertEncodingByIconv(src, dst, dstsize, "UTF-8", "SJIS") != -1)
+ return dst;
+#else
+ error("not support utf-8 conversion");
+#endif
+
+ /* not supported */
+ if (dstsize < 1) return dst;
+ dst[dstsize-1] = 0;
+ return strncpy(dst, src, dstsize-1);
}
/*
*p1 |= 0x80;
*p2 |= 0x80;
}
+#endif /* MULTIBYTE_FILENAME */
/* Local Variables: */
/* mode:c */