diff --git a/common/utf8conv.c b/common/utf8conv.c index a912b82ee..8a2cf8a20 100644 --- a/common/utf8conv.c +++ b/common/utf8conv.c @@ -1,785 +1,830 @@ /* utf8conf.c - UTF8 character set conversion * Copyright (C) 1994, 1998, 1999, 2000, 2001, 2003, 2006, * 2008, 2010 Free Software Foundation, Inc. * * This file is part of GnuPG. * * GnuPG is free software; you can redistribute it and/or modify it * under the terms of either * * - the GNU Lesser General Public License as published by the Free * Software Foundation; either version 3 of the License, or (at * your option) any later version. * * or * * - the GNU General Public License as published by the Free * Software Foundation; either version 2 of the License, or (at * your option) any later version. * * or both in parallel, as here. * * GnuPG is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * General Public License for more details. * * You should have received a copies of the GNU General Public License * and the GNU Lesser General Public License along with this program; * if not, see . */ #include #include #include #include #include #ifdef HAVE_LANGINFO_CODESET #include #endif #include #ifndef HAVE_ANDROID_SYSTEM # include #endif #include "util.h" #include "common-defs.h" #include "i18n.h" #include "stringhelp.h" #include "utf8conv.h" #ifndef MB_LEN_MAX #define MB_LEN_MAX 16 #endif static const char *active_charset_name = "iso-8859-1"; static int no_translation; /* Set to true if we let simply pass through. */ static int use_iconv; /* iconv conversion functions required. */ #ifdef HAVE_ANDROID_SYSTEM /* Fake stuff to get things building. */ typedef void *iconv_t; #define ICONV_CONST static iconv_t iconv_open (const char *tocode, const char *fromcode) { (void)tocode; (void)fromcode; return (iconv_t)(-1); } static size_t iconv (iconv_t cd, char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft) { (void)cd; (void)inbuf; (void)inbytesleft; (void)outbuf; (void)outbytesleft; return (size_t)(0); } static int iconv_close (iconv_t cd) { (void)cd; return 0; } #endif /*HAVE_ANDROID_SYSTEM*/ /* Error handler for iconv failures. This is needed to not clutter the output with repeated diagnostics about a missing conversion. */ static void handle_iconv_error (const char *to, const char *from, int use_fallback) { if (errno == EINVAL) { static int shown1, shown2; int x; if (to && !strcmp (to, "utf-8")) { x = shown1; shown1 = 1; } else { x = shown2; shown2 = 1; } if (!x) log_info (_("conversion from '%s' to '%s' not available\n"), from, to); } else { static int shown; if (!shown) log_info (_("iconv_open failed: %s\n"), strerror (errno)); shown = 1; } if (use_fallback) { /* To avoid further error messages we fallback to UTF-8 for the native encoding. Nowadays this seems to be the best bet in case of errors from iconv or nl_langinfo. */ active_charset_name = "utf-8"; no_translation = 0; use_iconv = 0; } } int set_native_charset (const char *newset) { const char *full_newset; if (!newset) { #ifdef HAVE_ANDROID_SYSTEM newset = "utf-8"; #elif defined HAVE_W32_SYSTEM static char codepage[30]; unsigned int cpno; const char *aliases; /* We are a console program thus we need to use the GetConsoleOutputCP function and not the the GetACP which would give the codepage for a GUI program. Note this is not a bulletproof detection because GetConsoleCP might return a different one for console input. Not sure how to cope with that. If the console Code page is not known we fall back to the system code page. */ #ifndef HAVE_W32CE_SYSTEM cpno = GetConsoleOutputCP (); if (!cpno) #endif cpno = GetACP (); sprintf (codepage, "CP%u", cpno ); /* Resolve alias. We use a long string string and not the usual array to optimize if the code is taken to a DSO. Taken from libiconv 1.9.2. */ newset = codepage; for (aliases = ("CP936" "\0" "GBK" "\0" "CP1361" "\0" "JOHAB" "\0" "CP20127" "\0" "ASCII" "\0" "CP20866" "\0" "KOI8-R" "\0" "CP21866" "\0" "KOI8-RU" "\0" "CP28591" "\0" "ISO-8859-1" "\0" "CP28592" "\0" "ISO-8859-2" "\0" "CP28593" "\0" "ISO-8859-3" "\0" "CP28594" "\0" "ISO-8859-4" "\0" "CP28595" "\0" "ISO-8859-5" "\0" "CP28596" "\0" "ISO-8859-6" "\0" "CP28597" "\0" "ISO-8859-7" "\0" "CP28598" "\0" "ISO-8859-8" "\0" "CP28599" "\0" "ISO-8859-9" "\0" "CP28605" "\0" "ISO-8859-15" "\0" "CP65001" "\0" "UTF-8" "\0"); *aliases; aliases += strlen (aliases) + 1, aliases += strlen (aliases) + 1) { if (!strcmp (codepage, aliases) ||(*aliases == '*' && !aliases[1])) { newset = aliases + strlen (aliases) + 1; break; } } #else /*!HAVE_W32_SYSTEM && !HAVE_ANDROID_SYSTEM*/ #ifdef HAVE_LANGINFO_CODESET newset = nl_langinfo (CODESET); #else /*!HAVE_LANGINFO_CODESET*/ /* Try to get the used charset from environment variables. */ static char codepage[30]; const char *lc, *dot, *mod; strcpy (codepage, "iso-8859-1"); lc = getenv ("LC_ALL"); if (!lc || !*lc) { lc = getenv ("LC_CTYPE"); if (!lc || !*lc) lc = getenv ("LANG"); } if (lc && *lc) { dot = strchr (lc, '.'); if (dot) { mod = strchr (++dot, '@'); if (!mod) mod = dot + strlen (dot); if (mod - dot < sizeof codepage && dot != mod) { memcpy (codepage, dot, mod - dot); codepage [mod - dot] = 0; } } } newset = codepage; #endif /*!HAVE_LANGINFO_CODESET*/ #endif /*!HAVE_W32_SYSTEM && !HAVE_ANDROID_SYSTEM*/ } full_newset = newset; if (strlen (newset) > 3 && !ascii_memcasecmp (newset, "iso", 3)) { newset += 3; if (*newset == '-' || *newset == '_') newset++; } /* Note that we silently assume that plain ASCII is actually meant as Latin-1. This makes sense because many Unix system don't have their locale set up properly and thus would get annoying error messages and we have to handle all the "bug" reports. Latin-1 has always been the character set used for 8 bit characters on Unix systems. */ if ( !*newset || !ascii_strcasecmp (newset, "8859-1" ) || !ascii_strcasecmp (newset, "646" ) || !ascii_strcasecmp (newset, "ASCII" ) || !ascii_strcasecmp (newset, "ANSI_X3.4-1968" ) ) { active_charset_name = "iso-8859-1"; no_translation = 0; use_iconv = 0; } else if ( !ascii_strcasecmp (newset, "utf8" ) || !ascii_strcasecmp(newset, "utf-8") ) { active_charset_name = "utf-8"; no_translation = 1; use_iconv = 0; } else { iconv_t cd; cd = iconv_open (full_newset, "utf-8"); if (cd == (iconv_t)-1) { handle_iconv_error (full_newset, "utf-8", 0); return -1; } iconv_close (cd); cd = iconv_open ("utf-8", full_newset); if (cd == (iconv_t)-1) { handle_iconv_error ("utf-8", full_newset, 0); return -1; } iconv_close (cd); active_charset_name = full_newset; no_translation = 0; use_iconv = 1; } return 0; } const char * get_native_charset () { return active_charset_name; } /* Return true if the native charset is utf-8. */ int is_native_utf8 (void) { return no_translation; } /* Convert string, which is in native encoding to UTF8 and return a new allocated UTF-8 string. This function terminates the process on memory shortage. */ char * native_to_utf8 (const char *orig_string) { const unsigned char *string = (const unsigned char *)orig_string; const unsigned char *s; char *buffer; unsigned char *p; size_t length = 0; if (no_translation) { /* Already utf-8 encoded. */ buffer = xstrdup (orig_string); } else if (!use_iconv) { /* For Latin-1 we can avoid the iconv overhead. */ for (s = string; *s; s++) { length++; if (*s & 0x80) length++; } buffer = xmalloc (length + 1); for (p = (unsigned char *)buffer, s = string; *s; s++) { if ( (*s & 0x80 )) { *p++ = 0xc0 | ((*s >> 6) & 3); *p++ = 0x80 | (*s & 0x3f); } else *p++ = *s; } *p = 0; } else { /* Need to use iconv. */ iconv_t cd; const char *inptr; char *outptr; size_t inbytes, outbytes; cd = iconv_open ("utf-8", active_charset_name); if (cd == (iconv_t)-1) { handle_iconv_error ("utf-8", active_charset_name, 1); return native_to_utf8 (string); } for (s=string; *s; s++ ) { length++; if ((*s & 0x80)) length += 5; /* We may need up to 6 bytes for the utf8 output. */ } buffer = xmalloc (length + 1); inptr = string; inbytes = strlen (string); outptr = buffer; outbytes = length; if ( iconv (cd, (ICONV_CONST char **)&inptr, &inbytes, &outptr, &outbytes) == (size_t)-1) { static int shown; if (!shown) log_info (_("conversion from '%s' to '%s' failed: %s\n"), active_charset_name, "utf-8", strerror (errno)); shown = 1; /* We don't do any conversion at all but use the strings as is. */ strcpy (buffer, string); } else /* Success. */ { *outptr = 0; /* We could realloc the buffer now but I doubt that it makes much sense given that it will get freed anyway soon after. */ } iconv_close (cd); } return buffer; } static char * do_utf8_to_native (const char *string, size_t length, int delim, int with_iconv) { int nleft; int i; unsigned char encbuf[8]; int encidx; const unsigned char *s; size_t n; char *buffer = NULL; char *p = NULL; unsigned long val = 0; size_t slen; int resync = 0; /* First pass (p==NULL): count the extended utf-8 characters. */ /* Second pass (p!=NULL): create string. */ for (;;) { for (slen = length, nleft = encidx = 0, n = 0, s = (const unsigned char *)string; slen; s++, slen--) { if (resync) { if (!(*s < 128 || (*s >= 0xc0 && *s <= 0xfd))) { /* Still invalid. */ if (p) { sprintf (p, "\\x%02x", *s); p += 4; } n += 4; continue; } resync = 0; } if (!nleft) { if (!(*s & 0x80)) { /* Plain ascii. */ if ( delim != -1 && (*s < 0x20 || *s == 0x7f || *s == delim || (delim && *s == '\\'))) { n++; if (p) *p++ = '\\'; switch (*s) { case '\n': n++; if ( p ) *p++ = 'n'; break; case '\r': n++; if ( p ) *p++ = 'r'; break; case '\f': n++; if ( p ) *p++ = 'f'; break; case '\v': n++; if ( p ) *p++ = 'v'; break; case '\b': n++; if ( p ) *p++ = 'b'; break; case 0: n++; if ( p ) *p++ = '0'; break; default: n += 3; if (p) { sprintf (p, "x%02x", *s); p += 3; } break; } } else { if (p) *p++ = *s; n++; } } else if ((*s & 0xe0) == 0xc0) /* 110x xxxx */ { val = *s & 0x1f; nleft = 1; encidx = 0; encbuf[encidx++] = *s; } else if ((*s & 0xf0) == 0xe0) /* 1110 xxxx */ { val = *s & 0x0f; nleft = 2; encidx = 0; encbuf[encidx++] = *s; } else if ((*s & 0xf8) == 0xf0) /* 1111 0xxx */ { val = *s & 0x07; nleft = 3; encidx = 0; encbuf[encidx++] = *s; } else if ((*s & 0xfc) == 0xf8) /* 1111 10xx */ { val = *s & 0x03; nleft = 4; encidx = 0; encbuf[encidx++] = *s; } else if ((*s & 0xfe) == 0xfc) /* 1111 110x */ { val = *s & 0x01; nleft = 5; encidx = 0; encbuf[encidx++] = *s; } else /* Invalid encoding: print as \xNN. */ { if (p) { sprintf (p, "\\x%02x", *s); p += 4; } n += 4; resync = 1; } } else if (*s < 0x80 || *s >= 0xc0) /* Invalid utf-8 */ { if (p) { for (i = 0; i < encidx; i++) { sprintf (p, "\\x%02x", encbuf[i]); p += 4; } sprintf (p, "\\x%02x", *s); p += 4; } n += 4 + 4 * encidx; nleft = 0; encidx = 0; resync = 1; } else { encbuf[encidx++] = *s; val <<= 6; val |= *s & 0x3f; if (!--nleft) /* Ready. */ { if (no_translation) { if (p) { for (i = 0; i < encidx; i++) *p++ = encbuf[i]; } n += encidx; encidx = 0; } else if (with_iconv) { /* Our strategy for using iconv is a bit strange but it better keeps compatibility with previous versions in regard to how invalid encodings are displayed. What we do is to keep the utf-8 as is and have the real translation step then at the end. Yes, I know that this is ugly. However we are short of the 1.4 release and for this branch we should not mess too much around with iconv things. One reason for this is that we don't know enough about non-GNU iconv implementation and want to minimize the risk of breaking the code on too many platforms. */ if ( p ) { for (i=0; i < encidx; i++ ) *p++ = encbuf[i]; } n += encidx; encidx = 0; } else /* Latin-1 case. */ { if (val >= 0x80 && val < 256) { /* We can simply print this character */ n++; if (p) *p++ = val; } else { /* We do not have a translation: print utf8. */ if (p) { for (i = 0; i < encidx; i++) { sprintf (p, "\\x%02x", encbuf[i]); p += 4; } } n += encidx * 4; encidx = 0; } } } } } if (!buffer) { /* Allocate the buffer after the first pass. */ buffer = p = xmalloc (n + 1); } else if (with_iconv) { /* Note: See above for comments. */ iconv_t cd; const char *inptr; char *outbuf, *outptr; size_t inbytes, outbytes; *p = 0; /* Terminate the buffer. */ cd = iconv_open (active_charset_name, "utf-8"); if (cd == (iconv_t)-1) { handle_iconv_error (active_charset_name, "utf-8", 1); xfree (buffer); return utf8_to_native (string, length, delim); } /* Allocate a new buffer large enough to hold all possible encodings. */ n = p - buffer + 1; inbytes = n - 1;; inptr = buffer; outbytes = n * MB_LEN_MAX; if (outbytes / MB_LEN_MAX != n) BUG (); /* Actually an overflow. */ outbuf = outptr = xmalloc (outbytes); if ( iconv (cd, (ICONV_CONST char **)&inptr, &inbytes, &outptr, &outbytes) == (size_t)-1) { static int shown; if (!shown) log_info (_("conversion from '%s' to '%s' failed: %s\n"), "utf-8", active_charset_name, strerror (errno)); shown = 1; /* Didn't worked out. Try again but without iconv. */ xfree (buffer); buffer = NULL; xfree (outbuf); outbuf = do_utf8_to_native (string, length, delim, 0); } else /* Success. */ { *outptr = 0; /* Make sure it is a string. */ /* We could realloc the buffer now but I doubt that it makes much sense given that it will get freed anyway soon after. */ xfree (buffer); } iconv_close (cd); return outbuf; } else /* Not using iconv. */ { *p = 0; /* Make sure it is a string. */ return buffer; } } } /* Convert string, which is in UTF-8 to native encoding. Replace illegal encodings by some "\xnn" and quote all control characters. A character with value DELIM will always be quoted, it must be a vanilla ASCII character. A DELIM value of -1 is special: it disables all quoting of control characters. This function terminates the process on memory shortage. */ char * utf8_to_native (const char *string, size_t length, int delim) { return do_utf8_to_native (string, length, delim, use_iconv); } /* Wrapper function for iconv_open, required for W32 as we dlopen that library on that system. */ jnlib_iconv_t jnlib_iconv_open (const char *tocode, const char *fromcode) { return (jnlib_iconv_t)iconv_open (tocode, fromcode); } /* Wrapper function for iconv, required for W32 as we dlopen that library on that system. */ size_t jnlib_iconv (jnlib_iconv_t cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft) { return iconv ((iconv_t)cd, (char**)inbuf, inbytesleft, outbuf, outbytesleft); } /* Wrapper function for iconv_close, required for W32 as we dlopen that library on that system. */ int jnlib_iconv_close (jnlib_iconv_t cd) { return iconv_close ((iconv_t)cd); } #ifdef HAVE_W32_SYSTEM -/* Return a malloced string encoded in UTF-8 from the wide char input +/* Return a malloced string encoded for CODEPAGE from the wide char input string STRING. Caller must free this value. Returns NULL and sets ERRNO on failure. Calling this function with STRING set to NULL is not defined. */ -char * -wchar_to_utf8 (const wchar_t *string) +static char * +wchar_to_cp (const wchar_t *string, unsigned int codepage) { int n; char *result; - n = WideCharToMultiByte (CP_UTF8, 0, string, -1, NULL, 0, NULL, NULL); + n = WideCharToMultiByte (codepage, 0, string, -1, NULL, 0, NULL, NULL); if (n < 0) { gpg_err_set_errno (EINVAL); return NULL; } result = xtrymalloc (n+1); if (!result) return NULL; - n = WideCharToMultiByte (CP_UTF8, 0, string, -1, result, n, NULL, NULL); + n = WideCharToMultiByte (codepage, 0, string, -1, result, n, NULL, NULL); if (n < 0) { xfree (result); gpg_err_set_errno (EINVAL); result = NULL; } return result; } -/* Return a malloced wide char string from an UTF-8 encoded input +/* Return a malloced wide char string from a CODEPAGE encoded input string STRING. Caller must free this value. Returns NULL and sets ERRNO on failure. Calling this function with STRING set to NULL is not defined. */ -wchar_t * -utf8_to_wchar (const char *string) +static wchar_t * +cp_to_wchar (const char *string, unsigned int codepage) { int n; size_t nbytes; wchar_t *result; - n = MultiByteToWideChar (CP_UTF8, 0, string, -1, NULL, 0); + n = MultiByteToWideChar (codepage, 0, string, -1, NULL, 0); if (n < 0) { gpg_err_set_errno (EINVAL); return NULL; } nbytes = (size_t)(n+1) * sizeof(*result); if (nbytes / sizeof(*result) != (n+1)) { gpg_err_set_errno (ENOMEM); return NULL; } result = xtrymalloc (nbytes); if (!result) return NULL; - n = MultiByteToWideChar (CP_UTF8, 0, string, -1, result, n); + n = MultiByteToWideChar (codepage, 0, string, -1, result, n); if (n < 0) { xfree (result); gpg_err_set_errno (EINVAL); result = NULL; } return result; } + + +/* Return a malloced string encoded in the active code page from the + * wide char input string STRING. Caller must free this value. + * Returns NULL and sets ERRNO on failure. Calling this function with + * STRING set to NULL is not defined. */ +char * +wchar_to_native (const wchar_t *string) +{ + return wchar_to_cp (string, CP_ACP); +} + + +/* Return a malloced wide char string from an UTF-8 encoded input + * string STRING. Caller must free this value. Returns NULL and sets + * ERRNO on failure. Calling this function with STRING set to NULL is + * not defined. */ +wchar_t * +native_to_wchar (const char *string) +{ + return cp_to_wchar (string, CP_ACP); +} + + +/* Return a malloced string encoded in UTF-8 from the wide char input + * string STRING. Caller must free this value. Returns NULL and sets + * ERRNO on failure. Calling this function with STRING set to NULL is + * not defined. */ +char * +wchar_to_utf8 (const wchar_t *string) +{ + return wchar_to_cp (string, CP_UTF8); +} + + +/* Return a malloced wide char string from an UTF-8 encoded input + * string STRING. Caller must free this value. Returns NULL and sets + * ERRNO on failure. Calling this function with STRING set to NULL is + * not defined. */ +wchar_t * +utf8_to_wchar (const char *string) +{ + return cp_to_wchar (string, CP_UTF8); +} + #endif /*HAVE_W32_SYSTEM*/ diff --git a/common/utf8conv.h b/common/utf8conv.h index ad7dbe977..def35def0 100644 --- a/common/utf8conv.h +++ b/common/utf8conv.h @@ -1,56 +1,58 @@ /* utf8conf.h * Copyright (C) 2003, 2006 Free Software Foundation, Inc. * * This file is part of GnuPG. * * GnuPG is free software; you can redistribute it and/or modify it * under the terms of either * * - the GNU Lesser General Public License as published by the Free * Software Foundation; either version 3 of the License, or (at * your option) any later version. * * or * * - the GNU General Public License as published by the Free * Software Foundation; either version 2 of the License, or (at * your option) any later version. * * or both in parallel, as here. * * GnuPG is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * General Public License for more details. * * You should have received a copies of the GNU General Public License * and the GNU Lesser General Public License along with this program; * if not, see . */ #ifndef GNUPG_COMMON_UTF8CONF_H #define GNUPG_COMMON_UTF8CONF_H int set_native_charset (const char *newset); const char *get_native_charset (void); int is_native_utf8 (void); char *native_to_utf8 (const char *string); char *utf8_to_native (const char *string, size_t length, int delim); /* Silly wrappers, required for W32 portability. */ typedef void *jnlib_iconv_t; jnlib_iconv_t jnlib_iconv_open (const char *tocode, const char *fromcode); size_t jnlib_iconv (jnlib_iconv_t cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft); int jnlib_iconv_close (jnlib_iconv_t cd); #ifdef HAVE_W32_SYSTEM +char *wchar_to_native (const wchar_t *string); +wchar_t *native_to_wchar (const char *string); char *wchar_to_utf8 (const wchar_t *string); wchar_t *utf8_to_wchar (const char *string); #endif /*HAVE_W32_SYSTEM*/ #endif /*GNUPG_COMMON_UTF8CONF_H*/ diff --git a/tools/gpgtar-create.c b/tools/gpgtar-create.c index 2fd70190c..968dca695 100644 --- a/tools/gpgtar-create.c +++ b/tools/gpgtar-create.c @@ -1,982 +1,982 @@ /* gpgtar-create.c - Create a TAR archive * Copyright (C) 2010 Free Software Foundation, Inc. * * This file is part of GnuPG. * * GnuPG is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 3 of the License, or * (at your option) any later version. * * GnuPG is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, see . */ #include #include #include #include #include #include #include #include #ifdef HAVE_W32_SYSTEM # define WIN32_LEAN_AND_MEAN # include #else /*!HAVE_W32_SYSTEM*/ # include # include # include #endif /*!HAVE_W32_SYSTEM*/ #include #include "i18n.h" #include "../common/exectool.h" #include "../common/sysutils.h" #include "gpgtar.h" #ifndef HAVE_LSTAT #define lstat(a,b) stat ((a), (b)) #endif /* Object to control the file scanning. */ struct scanctrl_s; typedef struct scanctrl_s *scanctrl_t; struct scanctrl_s { tar_header_t flist; tar_header_t *flist_tail; int nestlevel; }; /* Given a fresh header object HDR with only the name field set, try to gather all available info. This is the W32 version. */ #ifdef HAVE_W32_SYSTEM static gpg_error_t fillup_entry_w32 (tar_header_t hdr) { char *p; wchar_t *wfname; WIN32_FILE_ATTRIBUTE_DATA fad; DWORD attr; for (p=hdr->name; *p; p++) if (*p == '/') *p = '\\'; - wfname = utf8_to_wchar (hdr->name); + wfname = native_to_wchar (hdr->name); for (p=hdr->name; *p; p++) if (*p == '\\') *p = '/'; if (!wfname) { - log_error ("error utf8-ing '%s': %s\n", hdr->name, w32_strerror (-1)); + log_error ("error converting '%s': %s\n", hdr->name, w32_strerror (-1)); return gpg_error_from_syserror (); } if (!GetFileAttributesExW (wfname, GetFileExInfoStandard, &fad)) { log_error ("error stat-ing '%s': %s\n", hdr->name, w32_strerror (-1)); xfree (wfname); return gpg_error_from_syserror (); } xfree (wfname); attr = fad.dwFileAttributes; if ((attr & FILE_ATTRIBUTE_NORMAL)) hdr->typeflag = TF_REGULAR; else if ((attr & FILE_ATTRIBUTE_DIRECTORY)) hdr->typeflag = TF_DIRECTORY; else if ((attr & FILE_ATTRIBUTE_DEVICE)) hdr->typeflag = TF_NOTSUP; else if ((attr & (FILE_ATTRIBUTE_OFFLINE | FILE_ATTRIBUTE_TEMPORARY))) hdr->typeflag = TF_NOTSUP; else hdr->typeflag = TF_REGULAR; /* Map some attributes to USTAR defined mode bits. */ hdr->mode = 0640; /* User may read and write, group only read. */ if ((attr & FILE_ATTRIBUTE_DIRECTORY)) hdr->mode |= 0110; /* Dirs are user and group executable. */ if ((attr & FILE_ATTRIBUTE_READONLY)) hdr->mode &= ~0200; /* Clear the user write bit. */ if ((attr & FILE_ATTRIBUTE_HIDDEN)) hdr->mode &= ~0707; /* Clear all user and other bits. */ if ((attr & FILE_ATTRIBUTE_SYSTEM)) hdr->mode |= 0004; /* Make it readable by other. */ /* Only set the size for a regular file. */ if (hdr->typeflag == TF_REGULAR) hdr->size = (fad.nFileSizeHigh * (unsigned long long)(MAXDWORD+1) + fad.nFileSizeLow); hdr->mtime = (((unsigned long long)fad.ftLastWriteTime.dwHighDateTime << 32) | fad.ftLastWriteTime.dwLowDateTime); if (!hdr->mtime) hdr->mtime = (((unsigned long long)fad.ftCreationTime.dwHighDateTime << 32) | fad.ftCreationTime.dwLowDateTime); hdr->mtime -= 116444736000000000ULL; /* The filetime epoch is 1601-01-01. */ hdr->mtime /= 10000000; /* Convert from 0.1us to seconds. */ return 0; } #endif /*HAVE_W32_SYSTEM*/ /* Given a fresh header object HDR with only the name field set, try to gather all available info. This is the POSIX version. */ #ifndef HAVE_W32_SYSTEM static gpg_error_t fillup_entry_posix (tar_header_t hdr) { gpg_error_t err; struct stat sbuf; if (lstat (hdr->name, &sbuf)) { err = gpg_error_from_syserror (); log_error ("error stat-ing '%s': %s\n", hdr->name, gpg_strerror (err)); return err; } if (S_ISREG (sbuf.st_mode)) hdr->typeflag = TF_REGULAR; else if (S_ISDIR (sbuf.st_mode)) hdr->typeflag = TF_DIRECTORY; else if (S_ISCHR (sbuf.st_mode)) hdr->typeflag = TF_CHARDEV; else if (S_ISBLK (sbuf.st_mode)) hdr->typeflag = TF_BLOCKDEV; else if (S_ISFIFO (sbuf.st_mode)) hdr->typeflag = TF_FIFO; else if (S_ISLNK (sbuf.st_mode)) hdr->typeflag = TF_SYMLINK; else hdr->typeflag = TF_NOTSUP; /* FIXME: Save DEV and INO? */ /* Set the USTAR defined mode bits using the system macros. */ if (sbuf.st_mode & S_IRUSR) hdr->mode |= 0400; if (sbuf.st_mode & S_IWUSR) hdr->mode |= 0200; if (sbuf.st_mode & S_IXUSR) hdr->mode |= 0100; if (sbuf.st_mode & S_IRGRP) hdr->mode |= 0040; if (sbuf.st_mode & S_IWGRP) hdr->mode |= 0020; if (sbuf.st_mode & S_IXGRP) hdr->mode |= 0010; if (sbuf.st_mode & S_IROTH) hdr->mode |= 0004; if (sbuf.st_mode & S_IWOTH) hdr->mode |= 0002; if (sbuf.st_mode & S_IXOTH) hdr->mode |= 0001; #ifdef S_IXUID if (sbuf.st_mode & S_IXUID) hdr->mode |= 04000; #endif #ifdef S_IXGID if (sbuf.st_mode & S_IXGID) hdr->mode |= 02000; #endif #ifdef S_ISVTX if (sbuf.st_mode & S_ISVTX) hdr->mode |= 01000; #endif hdr->nlink = sbuf.st_nlink; hdr->uid = sbuf.st_uid; hdr->gid = sbuf.st_gid; /* Only set the size for a regular file. */ if (hdr->typeflag == TF_REGULAR) hdr->size = sbuf.st_size; hdr->mtime = sbuf.st_mtime; return 0; } #endif /*!HAVE_W32_SYSTEM*/ /* Add a new entry. The name of a director entry is ENTRYNAME; if that is NULL, DNAME is the name of the directory itself. Under Windows ENTRYNAME shall have backslashes replaced by standard slashes. */ static gpg_error_t add_entry (const char *dname, const char *entryname, scanctrl_t scanctrl) { gpg_error_t err; tar_header_t hdr; char *p; size_t dnamelen = strlen (dname); assert (dnamelen); hdr = xtrycalloc (1, sizeof *hdr + dnamelen + 1 + (entryname? strlen (entryname) : 0) + 1); if (!hdr) return gpg_error_from_syserror (); p = stpcpy (hdr->name, dname); if (entryname) { if (dname[dnamelen-1] != '/') *p++ = '/'; strcpy (p, entryname); } else { if (hdr->name[dnamelen-1] == '/') hdr->name[dnamelen-1] = 0; } #ifdef HAVE_DOSISH_SYSTEM err = fillup_entry_w32 (hdr); #else err = fillup_entry_posix (hdr); #endif if (err) xfree (hdr); else { if (opt.verbose) gpgtar_print_header (hdr, log_get_stream ()); *scanctrl->flist_tail = hdr; scanctrl->flist_tail = &hdr->next; } return 0; } static gpg_error_t scan_directory (const char *dname, scanctrl_t scanctrl) { gpg_error_t err = 0; #ifdef HAVE_W32_SYSTEM WIN32_FIND_DATAW fi; HANDLE hd = INVALID_HANDLE_VALUE; char *p; if (!*dname) return 0; /* An empty directory name has no entries. */ { char *fname; wchar_t *wfname; fname = xtrymalloc (strlen (dname) + 2 + 2 + 1); if (!fname) { err = gpg_error_from_syserror (); goto leave; } if (!strcmp (dname, "/")) strcpy (fname, "/*"); /* Trailing slash is not allowed. */ else if (!strcmp (dname, ".")) strcpy (fname, "*"); else if (*dname && dname[strlen (dname)-1] == '/') strcpy (stpcpy (fname, dname), "*"); else if (*dname && dname[strlen (dname)-1] != '*') strcpy (stpcpy (fname, dname), "/*"); else strcpy (fname, dname); for (p=fname; *p; p++) if (*p == '/') *p = '\\'; - wfname = utf8_to_wchar (fname); + wfname = native_to_wchar (fname); xfree (fname); if (!wfname) { err = gpg_error_from_syserror (); log_error (_("error reading directory '%s': %s\n"), dname, gpg_strerror (err)); goto leave; } hd = FindFirstFileW (wfname, &fi); if (hd == INVALID_HANDLE_VALUE) { err = gpg_error_from_syserror (); log_error (_("error reading directory '%s': %s\n"), dname, w32_strerror (-1)); xfree (wfname); goto leave; } xfree (wfname); } do { - char *fname = wchar_to_utf8 (fi.cFileName); + char *fname = wchar_to_native (fi.cFileName); if (!fname) { err = gpg_error_from_syserror (); - log_error ("error utf8-ing filename: %s\n", w32_strerror (-1)); + log_error ("error converting filename: %s\n", w32_strerror (-1)); break; } for (p=fname; *p; p++) if (*p == '\\') *p = '/'; if (!strcmp (fname, "." ) || !strcmp (fname, "..")) err = 0; /* Skip self and parent dir entry. */ else if (!strncmp (dname, "./", 2) && dname[2]) err = add_entry (dname+2, fname, scanctrl); else err = add_entry (dname, fname, scanctrl); xfree (fname); } while (!err && FindNextFileW (hd, &fi)); if (err) ; else if (GetLastError () == ERROR_NO_MORE_FILES) err = 0; else { err = gpg_error_from_syserror (); log_error (_("error reading directory '%s': %s\n"), dname, w32_strerror (-1)); } leave: if (hd != INVALID_HANDLE_VALUE) FindClose (hd); #else /*!HAVE_W32_SYSTEM*/ DIR *dir; struct dirent *de; if (!*dname) return 0; /* An empty directory name has no entries. */ dir = opendir (dname); if (!dir) { err = gpg_error_from_syserror (); log_error (_("error reading directory '%s': %s\n"), dname, gpg_strerror (err)); return err; } while ((de = readdir (dir))) { if (!strcmp (de->d_name, "." ) || !strcmp (de->d_name, "..")) continue; /* Skip self and parent dir entry. */ err = add_entry (dname, de->d_name, scanctrl); if (err) goto leave; } leave: closedir (dir); #endif /*!HAVE_W32_SYSTEM*/ return err; } static gpg_error_t scan_recursive (const char *dname, scanctrl_t scanctrl) { gpg_error_t err = 0; tar_header_t hdr, *start_tail, *stop_tail; if (scanctrl->nestlevel > 200) { log_error ("directories too deeply nested\n"); return gpg_error (GPG_ERR_RESOURCE_LIMIT); } scanctrl->nestlevel++; assert (scanctrl->flist_tail); start_tail = scanctrl->flist_tail; scan_directory (dname, scanctrl); stop_tail = scanctrl->flist_tail; hdr = *start_tail; for (; hdr && hdr != *stop_tail; hdr = hdr->next) if (hdr->typeflag == TF_DIRECTORY) { if (opt.verbose > 1) log_info ("scanning directory '%s'\n", hdr->name); scan_recursive (hdr->name, scanctrl); } scanctrl->nestlevel--; return err; } /* Returns true if PATTERN is acceptable. */ static int pattern_valid_p (const char *pattern) { if (!*pattern) return 0; if (*pattern == '.' && pattern[1] == '.') return 0; if (*pattern == '/' || *pattern == DIRSEP_C) return 0; /* Absolute filenames are not supported. */ #ifdef HAVE_DRIVE_LETTERS if (((*pattern >= 'a' && *pattern <= 'z') || (*pattern >= 'A' && *pattern <= 'Z')) && pattern[1] == ':') return 0; /* Drive letter are not allowed either. */ #endif /*HAVE_DRIVE_LETTERS*/ return 1; /* Okay. */ } static void store_xoctal (char *buffer, size_t length, unsigned long long value) { char *p, *pend; size_t n; unsigned long long v; assert (length > 1); v = value; n = length; p = pend = buffer + length; *--p = 0; /* Nul byte. */ n--; do { *--p = '0' + (v % 8); v /= 8; n--; } while (v && n); if (!v) { /* Pad. */ for ( ; n; n--) *--p = '0'; } else /* Does not fit into the field. Store as binary number. */ { v = value; n = length; p = pend = buffer + length; do { *--p = v; v /= 256; n--; } while (v && n); if (!v) { /* Pad. */ for ( ; n; n--) *--p = 0; if (*p & 0x80) BUG (); *p |= 0x80; /* Set binary flag. */ } else BUG (); } } static void store_uname (char *buffer, size_t length, unsigned long uid) { static int initialized; static unsigned long lastuid; static char lastuname[32]; if (!initialized || uid != lastuid) { #ifdef HAVE_W32_SYSTEM mem2str (lastuname, uid? "user":"root", sizeof lastuname); #else struct passwd *pw = getpwuid (uid); lastuid = uid; initialized = 1; if (pw) mem2str (lastuname, pw->pw_name, sizeof lastuname); else { log_info ("failed to get name for uid %lu\n", uid); *lastuname = 0; } #endif } mem2str (buffer, lastuname, length); } static void store_gname (char *buffer, size_t length, unsigned long gid) { static int initialized; static unsigned long lastgid; static char lastgname[32]; if (!initialized || gid != lastgid) { #ifdef HAVE_W32_SYSTEM mem2str (lastgname, gid? "users":"root", sizeof lastgname); #else struct group *gr = getgrgid (gid); lastgid = gid; initialized = 1; if (gr) mem2str (lastgname, gr->gr_name, sizeof lastgname); else { log_info ("failed to get name for gid %lu\n", gid); *lastgname = 0; } #endif } mem2str (buffer, lastgname, length); } static gpg_error_t build_header (void *record, tar_header_t hdr) { gpg_error_t err; struct ustar_raw_header *raw = record; size_t namelen, n; unsigned long chksum; unsigned char *p; memset (record, 0, RECORDSIZE); /* Store name and prefix. */ namelen = strlen (hdr->name); if (namelen < sizeof raw->name) memcpy (raw->name, hdr->name, namelen); else { n = (namelen < sizeof raw->prefix)? namelen : sizeof raw->prefix; for (n--; n ; n--) if (hdr->name[n] == '/') break; if (namelen - n < sizeof raw->name) { /* Note that the N is < sizeof prefix and that the delimiting slash is not stored. */ memcpy (raw->prefix, hdr->name, n); memcpy (raw->name, hdr->name+n+1, namelen - n); } else { err = gpg_error (GPG_ERR_TOO_LARGE); log_error ("error storing file '%s': %s\n", hdr->name, gpg_strerror (err)); return err; } } store_xoctal (raw->mode, sizeof raw->mode, hdr->mode); store_xoctal (raw->uid, sizeof raw->uid, hdr->uid); store_xoctal (raw->gid, sizeof raw->gid, hdr->gid); store_xoctal (raw->size, sizeof raw->size, hdr->size); store_xoctal (raw->mtime, sizeof raw->mtime, hdr->mtime); switch (hdr->typeflag) { case TF_REGULAR: raw->typeflag[0] = '0'; break; case TF_HARDLINK: raw->typeflag[0] = '1'; break; case TF_SYMLINK: raw->typeflag[0] = '2'; break; case TF_CHARDEV: raw->typeflag[0] = '3'; break; case TF_BLOCKDEV: raw->typeflag[0] = '4'; break; case TF_DIRECTORY: raw->typeflag[0] = '5'; break; case TF_FIFO: raw->typeflag[0] = '6'; break; default: return gpg_error (GPG_ERR_NOT_SUPPORTED); } memcpy (raw->magic, "ustar", 6); raw->version[0] = '0'; raw->version[1] = '0'; store_uname (raw->uname, sizeof raw->uname, hdr->uid); store_gname (raw->gname, sizeof raw->gname, hdr->gid); #ifndef HAVE_W32_SYSTEM if (hdr->typeflag == TF_SYMLINK) { int nread; nread = readlink (hdr->name, raw->linkname, sizeof raw->linkname -1); if (nread < 0) { err = gpg_error_from_syserror (); log_error ("error reading symlink '%s': %s\n", hdr->name, gpg_strerror (err)); return err; } raw->linkname[nread] = 0; } #endif /*HAVE_W32_SYSTEM*/ /* Compute the checksum. */ memset (raw->checksum, ' ', sizeof raw->checksum); chksum = 0; p = record; for (n=0; n < RECORDSIZE; n++) chksum += *p++; store_xoctal (raw->checksum, sizeof raw->checksum - 1, chksum); raw->checksum[7] = ' '; return 0; } static gpg_error_t write_file (estream_t stream, tar_header_t hdr) { gpg_error_t err; char record[RECORDSIZE]; estream_t infp; size_t nread, nbytes; int any; err = build_header (record, hdr); if (err) { if (gpg_err_code (err) == GPG_ERR_NOT_SUPPORTED) { log_info ("skipping unsupported file '%s'\n", hdr->name); err = 0; } return err; } if (hdr->typeflag == TF_REGULAR) { infp = es_fopen (hdr->name, "rb"); if (!infp) { err = gpg_error_from_syserror (); log_error ("can't open '%s': %s - skipped\n", hdr->name, gpg_strerror (err)); return err; } } else infp = NULL; err = write_record (stream, record); if (err) goto leave; if (hdr->typeflag == TF_REGULAR) { hdr->nrecords = (hdr->size + RECORDSIZE-1)/RECORDSIZE; any = 0; while (hdr->nrecords--) { nbytes = hdr->nrecords? RECORDSIZE : (hdr->size % RECORDSIZE); if (!nbytes) nbytes = RECORDSIZE; nread = es_fread (record, 1, nbytes, infp); if (nread != nbytes) { err = gpg_error_from_syserror (); log_error ("error reading file '%s': %s%s\n", hdr->name, gpg_strerror (err), any? " (file shrunk?)":""); goto leave; } any = 1; err = write_record (stream, record); if (err) goto leave; } nread = es_fread (record, 1, 1, infp); if (nread) log_info ("note: file '%s' has grown\n", hdr->name); } leave: if (err) es_fclose (infp); else if ((err = es_fclose (infp))) log_error ("error closing file '%s': %s\n", hdr->name, gpg_strerror (err)); return err; } static gpg_error_t write_eof_mark (estream_t stream) { gpg_error_t err; char record[RECORDSIZE]; memset (record, 0, sizeof record); err = write_record (stream, record); if (!err) err = write_record (stream, record); return err; } /* Create a new tarball using the names in the array INPATTERN. If INPATTERN is NULL take the pattern as null terminated strings from stdin. */ gpg_error_t gpgtar_create (char **inpattern, int encrypt, int sign) { gpg_error_t err = 0; struct scanctrl_s scanctrl_buffer; scanctrl_t scanctrl = &scanctrl_buffer; tar_header_t hdr, *start_tail; estream_t outstream = NULL; estream_t cipher_stream = NULL; int eof_seen = 0; if (!inpattern) es_set_binary (es_stdin); memset (scanctrl, 0, sizeof *scanctrl); scanctrl->flist_tail = &scanctrl->flist; while (!eof_seen) { char *pat, *p; int skip_this = 0; if (inpattern) { const char *pattern = *inpattern; if (!pattern) break; /* End of array. */ inpattern++; if (!*pattern) continue; pat = xtrystrdup (pattern); } else /* Read null delimited pattern from stdin. */ { int c; char namebuf[4096]; size_t n = 0; for (;;) { if ((c = es_getc (es_stdin)) == EOF) { if (es_ferror (es_stdin)) { err = gpg_error_from_syserror (); log_error ("error reading '%s': %s\n", "[stdin]", strerror (errno)); goto leave; } /* Note: The Nul is a delimiter and not a terminator. */ c = 0; eof_seen = 1; } if (n >= sizeof namebuf - 1) { if (!skip_this) { skip_this = 1; log_error ("error reading '%s': %s\n", "[stdin]", "filename too long"); } } else namebuf[n++] = c; if (!c) { namebuf[n] = 0; break; } } if (skip_this || n < 2) continue; pat = xtrystrdup (namebuf); } if (!pat) { err = gpg_error_from_syserror (); log_error ("memory allocation problem: %s\n", gpg_strerror (err)); goto leave; } for (p=pat; *p; p++) if (*p == '\\') *p = '/'; if (opt.verbose > 1) log_info ("scanning '%s'\n", pat); start_tail = scanctrl->flist_tail; if (skip_this || !pattern_valid_p (pat)) log_error ("skipping invalid name '%s'\n", pat); else if (!add_entry (pat, NULL, scanctrl) && *start_tail && ((*start_tail)->typeflag & TF_DIRECTORY)) scan_recursive (pat, scanctrl); xfree (pat); } if (opt.outfile) { if (!strcmp (opt.outfile, "-")) outstream = es_stdout; else outstream = es_fopen (opt.outfile, "wb"); if (!outstream) { err = gpg_error_from_syserror (); log_error (_("can't create '%s': %s\n"), opt.outfile, gpg_strerror (err)); goto leave; } } else { outstream = es_stdout; } if (outstream == es_stdout) es_set_binary (es_stdout); if (encrypt || sign) { cipher_stream = outstream; outstream = es_fopenmem (0, "rwb"); if (! outstream) { err = gpg_error_from_syserror (); goto leave; } } for (hdr = scanctrl->flist; hdr; hdr = hdr->next) { err = write_file (outstream, hdr); if (err) goto leave; } err = write_eof_mark (outstream); if (err) goto leave; if (encrypt || sign) { int i; strlist_t arg; const char **argv; err = es_fseek (outstream, 0, SEEK_SET); if (err) goto leave; /* '--encrypt' may be combined with '--symmetric', but 'encrypt' is set either way. Clear it if no recipients are specified. XXX: Fix command handling. */ if (opt.symmetric && opt.recipients == NULL) encrypt = 0; argv = xtrycalloc (strlist_length (opt.gpg_arguments) + 2 * strlist_length (opt.recipients) + 1 + !!encrypt + !!sign + 2 * !!opt.user + !!opt.symmetric, sizeof *argv); if (argv == NULL) { err = gpg_error_from_syserror (); goto leave; } i = 0; if (encrypt) argv[i++] = "--encrypt"; if (sign) argv[i++] = "--sign"; if (opt.user) { argv[i++] = "--local-user"; argv[i++] = opt.user; } if (opt.symmetric) argv[i++] = "--symmetric"; for (arg = opt.recipients; arg; arg = arg->next) { argv[i++] = "--recipient"; argv[i++] = arg->d; } for (arg = opt.gpg_arguments; arg; arg = arg->next) argv[i++] = arg->d; argv[i++] = NULL; assert (i == strlist_length (opt.gpg_arguments) + 2 * strlist_length (opt.recipients) + 1 + !!encrypt + !!sign + 2 * !!opt.user + !!opt.symmetric); err = gnupg_exec_tool_stream (opt.gpg_program, argv, outstream, cipher_stream); xfree (argv); if (err) goto leave; } leave: if (!err) { gpg_error_t first_err; if (outstream != es_stdout) first_err = es_fclose (outstream); else first_err = es_fflush (outstream); outstream = NULL; if (cipher_stream != es_stdout) err = es_fclose (cipher_stream); else err = es_fflush (cipher_stream); cipher_stream = NULL; if (! err) err = first_err; } if (err) { log_error ("creating tarball '%s' failed: %s\n", es_fname_get (outstream), gpg_strerror (err)); if (outstream && outstream != es_stdout) es_fclose (outstream); if (cipher_stream && cipher_stream != es_stdout) es_fclose (cipher_stream); if (opt.outfile) gnupg_remove (opt.outfile); } scanctrl->flist_tail = NULL; while ( (hdr = scanctrl->flist) ) { scanctrl->flist = hdr->next; xfree (hdr); } return err; }