Skip to main content :: ::

KDE API Reference

:: Skip to Link Menu

/ API Reference / kio / kio

kmimemagic.cpp /* This file is part of the KDE libraries Copyright (C) 2000 Fritz Elfert <fritz@kde.org> Copyright (C) 2004 Allan Sandfeld Jensen <kde@carewolf.com> This library is free software; you can redistribute it and/or modify it under the terms of the GNU Library General Public License version 2 as published by the Free Software Foundation. This library 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 Library General Public License for more details. You should have received a copy of the GNU Library General Public License along with this library; see the file COPYING.LIB. If not, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ #include "kmimemagic.h" #include <kdebug.h> #include <kapplication.h> #include <qfile.h> #include <ksimpleconfig.h> #include <kstandarddirs.h> #include <kstaticdeleter.h> #include <klargefile.h> #include <assert.h> static int fsmagic(struct config_rec* conf, const char *fn, KDE_struct_stat *sb); static void process(struct config_rec* conf, const QString &); static int ascmagic(struct config_rec* conf, unsigned char *buf, int nbytes); static int tagmagic(unsigned char *buf, int nbytes); static int textmagic(struct config_rec* conf, unsigned char *, int); static void tryit(struct config_rec* conf, unsigned char *buf, int nb); static int match(struct config_rec* conf, unsigned char *, int); KMimeMagic* KMimeMagic::s_pSelf; static KStaticDeleter<KMimeMagic> kmimemagicsd; KMimeMagic* KMimeMagic::self() { if( !s_pSelf ) initStatic(); return s_pSelf; } void KMimeMagic::initStatic() { s_pSelf = kmimemagicsd.setObject( s_pSelf, new KMimeMagic() ); s_pSelf->setFollowLinks( true ); } #include <stdio.h> #include <unistd.h> #include <stdlib.h> #include <sys/wait.h> #include <sys/types.h> #include <sys/stat.h> #include <fcntl.h> #include <errno.h> #include <ctype.h> #include <time.h> #include <utime.h> #include <stdarg.h> #include <qregexp.h> #include <qstring.h> //#define MIME_MAGIC_DEBUG_TABLE // untested // Uncomment to debug the config-file parsing phase //#define DEBUG_APPRENTICE // Uncomment to debug the matching phase //#define DEBUG_MIMEMAGIC #if (defined DEBUG_MIMEMAGIC || defined DEBUG_APPRENTICE) #define DEBUG_LINENUMBERS #endif /* * Buitltin Mime types */ #define MIME_BINARY_UNKNOWN "application/octet-stream" #define MIME_BINARY_UNREADABLE "application/x-unreadable" #define MIME_BINARY_ZEROSIZE "application/x-zerosize" #define MIME_TEXT_UNKNOWN "text/plain" #define MIME_TEXT_PLAIN "text/plain" #define MIME_INODE_DIR "inode/directory" #define MIME_INODE_CDEV "inode/chardevice" #define MIME_INODE_BDEV "inode/blockdevice" #define MIME_INODE_FIFO "inode/fifo" #define MIME_INODE_LINK "inode/link" #define MIME_INODE_SOCK "inode/socket" // Following should go in magic-file - Fritz #define MIME_APPL_TROFF "application/x-troff" #define MIME_APPL_TAR "application/x-tar" #define MIME_TEXT_FORTRAN "text/x-fortran" #define MAXMIMESTRING 256 #define HOWMANY 4000 /* big enough to recognize most WWW files, and skip GPL-headers */ #define MAXDESC 50 /* max leng of text description */ #define MAXstring 64 /* max leng of "string" types */ typedef union VALUETYPE { unsigned char b; unsigned short h; unsigned long l; char s[MAXstring]; unsigned char hs[2]; /* 2 bytes of a fixed-endian "short" */ unsigned char hl[4]; /* 2 bytes of a fixed-endian "long" */ } VALUETYPE; struct magic { struct magic *next; /* link to next entry */ #ifdef DEBUG_LINENUMBERS int lineno; /* line number from magic file - doesn't say from which one ;) */ #endif short flag; #define INDIR 1 /* if '>(...)' appears, */ #define UNSIGNED 2 /* comparison is unsigned */ short cont_level; /* level of ">" */ struct { char type; /* byte short long */ long offset; /* offset from indirection */ } in; long offset; /* offset to magic number */ unsigned char reln; /* relation (0=eq, '>'=gt, etc) */ char type; /* int, short, long or string. */ char vallen; /* length of string value, if any */ #define BYTE 1 #define SHORT 2 #define LONG 4 #define STRING 5 #define DATE 6 #define BESHORT 7 #define BELONG 8 #define BEDATE 9 #define LESHORT 10 #define LELONG 11 #define LEDATE 12 VALUETYPE value; /* either number or string */ unsigned long mask; /* mask before comparison with value */ char nospflag; /* suppress space character */ /* NOTE: this string is suspected of overrunning - find it! */ char desc[MAXDESC]; /* description */ }; /* * data structures for tar file recognition * -------------------------------------------------------------------------- * Header file for public domain tar (tape archive) program. * * @(#)tar.h 1.20 86/10/29 Public Domain. Created 25 August 1985 by John * Gilmore, ihnp4!hoptoad!gnu. * * Header block on tape. * * I'm going to use traditional DP naming conventions here. A "block" is a big * chunk of stuff that we do I/O on. A "record" is a piece of info that we * care about. Typically many "record"s fit into a "block". */ #define RECORDSIZE 512 #define NAMSIZ 100 #define TUNMLEN 32 #define TGNMLEN 32 union record { char charptr[RECORDSIZE]; struct header { char name[NAMSIZ]; char mode[8]; char uid[8]; char gid[8]; char size[12]; char mtime[12]; char chksum[8]; char linkflag; char linkname[NAMSIZ]; char magic[8]; char uname[TUNMLEN]; char gname[TGNMLEN]; char devmajor[8]; char devminor[8]; } header; }; /* The magic field is filled with this if uname and gname are valid. */ #define TMAGIC "ustar " /* 7 chars and a null */ /* * file-function prototypes */ static int is_tar(unsigned char *, int); static unsigned long signextend(struct magic *, unsigned long); static int getvalue(struct magic *, char **); static int hextoint(int); static char *getstr(char *, char *, int, int *); static int mget(union VALUETYPE *, unsigned char *, struct magic *, int); static int mcheck(union VALUETYPE *, struct magic *); static int mconvert(union VALUETYPE *, struct magic *); static long from_oct(int, char *); /* * includes for ASCII substring recognition formerly "names.h" in file * command * * Original notes: names and types used by ascmagic in file(1). * These tokens are * here because they can appear anywhere in the first HOWMANY bytes, while * tokens in /etc/magic must appear at fixed offsets into the file. Don't * make HOWMANY too high unless you have a very fast CPU. */ /* these types are used calculate index to 'types': keep em in sync! */ /* HTML inserted in first because this is a web server module now */ /* ENG removed because stupid */ #define L_HTML 0x001 /* HTML */ #define L_C 0x002 /* first and foremost on UNIX */ #define L_MAKE 0x004 /* Makefiles */ #define L_PLI 0x008 /* PL/1 */ #define L_MACH 0x010 /* some kinda assembler */ #define L_PAS 0x020 /* Pascal */ #define L_JAVA 0x040 /* Java source */ #define L_CPP 0x080 /* C++ */ #define L_MAIL 0x100 /* Electronic mail */ #define L_NEWS 0x200 /* Usenet Netnews */ #define L_DIFF 0x400 /* Output of diff */ #define L_OBJC 0x800 /* Objective C */ #define P_HTML 0 /* HTML */ #define P_C 1 /* first and foremost on UNIX */ #define P_MAKE 2 /* Makefiles */ #define P_PLI 3 /* PL/1 */ #define P_MACH 4 /* some kinda assembler */ #define P_PAS 5 /* Pascal */ #define P_JAVA 6 /* Java source */ #define P_CPP 7 /* C++ */ #define P_MAIL 8 /* Electronic mail */ #define P_NEWS 9 /* Usenet Netnews */ #define P_DIFF 10 /* Output of diff */ #define P_OBJC 11 /* Objective C */ typedef struct asc_type { const char *type; int kwords; double weight; } asc_type; static const asc_type types[] = { { "text/html", 19, 2 }, // 10 items but 10 different words only { "text/x-c", 13, 1 }, { "text/x-makefile", 4, 1.9 }, { "text/x-pli", 1, 3 }, { "text/x-assembler", 6, 2.1 }, { "text/x-pascal", 1, 1 }, { "text/x-java", 12, 1 }, { "text/x-c++", 19, 1 }, { "message/rfc822", 4, 1.9 }, { "message/news", 3, 2 }, { "text/x-diff", 4, 2 }, { "text/x-objc", 10, 1 } }; #define NTYPES (sizeof(types)/sizeof(asc_type)) static struct names { const char *name; short type; } const names[] = { { "<html", L_HTML }, { "<HTML", L_HTML }, { "<head", L_HTML }, { "<HEAD", L_HTML }, { "<body", L_HTML }, { "<BODY", L_HTML }, { "<title", L_HTML }, { "<TITLE", L_HTML }, { "<h1", L_HTML }, { "<H1", L_HTML }, { "<a", L_HTML }, { "<A", L_HTML }, { "<img", L_HTML }, { "<IMG", L_HTML }, { "<!--", L_HTML }, { "<!doctype", L_HTML }, { "<!DOCTYPE", L_HTML }, { "<div", L_HTML }, { "<DIV", L_HTML }, { "<frame", L_HTML }, { "<FRAME", L_HTML }, { "<frameset", L_HTML }, { "<FRAMESET", L_HTML }, { "<script", L_HTML }, { "<SCRIPT", L_HTML }, { "/*", L_C|L_CPP|L_JAVA|L_OBJC }, { "//", L_C|L_CPP|L_JAVA|L_OBJC }, { "#include", L_C|L_CPP }, { "#ifdef", L_C|L_CPP }, { "#ifndef", L_C|L_CPP }, { "bool", L_C|L_CPP }, { "char", L_C|L_CPP|L_JAVA|L_OBJC }, { "int", L_C|L_CPP|L_JAVA|L_OBJC }, { "float", L_C|L_CPP|L_JAVA|L_OBJC }, { "void", L_C|L_CPP|L_JAVA|L_OBJC }, { "extern", L_C|L_CPP }, { "struct", L_C|L_CPP }, { "union", L_C|L_CPP }, { "implements", L_JAVA }, { "super", L_JAVA }, { "import", L_JAVA }, { "class", L_CPP|L_JAVA }, { "public", L_CPP|L_JAVA }, { "private", L_CPP|L_JAVA }, { "explicit", L_CPP }, { "virtual", L_CPP }, { "namespace", L_CPP }, { "#import", L_OBJC }, { "@interface", L_OBJC }, { "@implementation", L_OBJC }, { "@protocol", L_OBJC }, { "CFLAGS", L_MAKE }, { "LDFLAGS", L_MAKE }, { "all:", L_MAKE }, { ".PHONY:", L_MAKE }, { "srcdir", L_MAKE }, { "exec_prefix", L_MAKE }, /* * Too many files of text have these words in them. Find another way * to recognize Fortrash. */ { ".ascii", L_MACH }, { ".asciiz", L_MACH }, { ".byte", L_MACH }, { ".even", L_MACH }, { ".globl", L_MACH }, { "clr", L_MACH }, { "(input", L_PAS }, { "dcl", L_PLI }, { "Received:", L_MAIL }, /* we now stop at '>' for tokens, so this one won't work { ">From", L_MAIL },*/ { "Return-Path:", L_MAIL }, { "Cc:", L_MAIL }, { "Newsgroups:", L_NEWS }, { "Path:", L_NEWS }, { "Organization:", L_NEWS }, { "---", L_DIFF }, { "+++", L_DIFF }, { "***", L_DIFF }, { "@@", L_DIFF }, { NULL, 0 } }; class KMimeMagicUtimeConf { public: KMimeMagicUtimeConf() { tmpDirs << QString::fromLatin1("/tmp"); // default value // The trick is that we also don't want the user to override globally set // directories. So we have to misuse KStandardDirs :} QStringList confDirs = KGlobal::dirs()->resourceDirs( "config" ); if ( !confDirs.isEmpty() ) { QString globalConf = confDirs.last() + "kmimemagicrc"; if ( QFile::exists( globalConf ) ) { KSimpleConfig cfg( globalConf ); cfg.setGroup( "Settings" ); tmpDirs = cfg.readListEntry( "atimeDirs" ); } if ( confDirs.count() > 1 ) { QString localConf = confDirs.first() + "kmimemagicrc"; if ( QFile::exists( localConf ) ) { KSimpleConfig cfg( localConf ); cfg.setGroup( "Settings" ); tmpDirs += cfg.readListEntry( "atimeDirs" ); } } for ( QStringList::Iterator it = tmpDirs.begin() ; it != tmpDirs.end() ; ++it ) { QString dir = *it; if ( !dir.isEmpty() && dir[ dir.length()-1 ] != '/' ) (*it) += '/'; } } #if 0 // debug code for ( QStringList::Iterator it = tmpDirs.begin() ; it != tmpDirs.end() ; ++it ) kdDebug(7018) << " atimeDir: " << *it << endl; #endif } bool restoreAccessTime( const QString & file ) const { QString dir = file.left( file.findRev( '/' ) ); bool res = tmpDirs.contains( dir ); //kdDebug(7018) << "restoreAccessTime " << file << " dir=" << dir << " result=" << res << endl; return res; } QStringList tmpDirs; }; /* current config */ struct config_rec { bool followLinks; QString resultBuf; int accuracy; struct magic *magic, /* head of magic config list */ *last; KMimeMagicUtimeConf * utimeConf; }; #ifdef MIME_MAGIC_DEBUG_TABLE static void test_table() { struct magic *m; struct magic *prevm = NULL; kdDebug(7018) << "test_table : started" << endl; for (m = conf->magic; m; m = m->next) { if (isprint((((unsigned long) m) >> 24) & 255) && isprint((((unsigned long) m) >> 16) & 255) && isprint((((unsigned long) m) >> 8) & 255) && isprint(((unsigned long) m) & 255)) { //debug("test_table: POINTER CLOBBERED! " //"m=\"%c%c%c%c\" line=%d", (((unsigned long) m) >> 24) & 255, (((unsigned long) m) >> 16) & 255, (((unsigned long) m) >> 8) & 255, ((unsigned long) m) & 255, prevm ? prevm->lineno : -1); break; } prevm = m; } } #endif #define EATAB {while (isascii((unsigned char) *l) && \ isspace((unsigned char) *l)) ++l;} int KMimeMagic::parse_line(char *line, int *rule, int lineno) { int ws_offset; /* delete newline */ if (line[0]) { line[strlen(line) - 1] = '\0'; } /* skip leading whitespace */ ws_offset = 0; while (line[ws_offset] && isspace(line[ws_offset])) { ws_offset++; } /* skip blank lines */ if (line[ws_offset] == 0) { return 0; } /* comment, do not parse */ if (line[ws_offset] == '#') return 0; /* if we get here, we're going to use it so count it */ (*rule)++; /* parse it */ return (parse(line + ws_offset, lineno) != 0); } /* * apprentice - load configuration from the magic file. */ int KMimeMagic::apprentice( const QString& magicfile ) { FILE *f; char line[BUFSIZ + 1]; int errs = 0; int lineno; int rule = 0; QCString fname; if (magicfile.isEmpty()) return -1; fname = QFile::encodeName(magicfile); f = fopen(fname, "r"); if (f == NULL) { kdError(7018) << "can't read magic file " << fname.data() << ": " << strerror(errno) << endl; return -1; } /* parse it */ for (lineno = 1; fgets(line, BUFSIZ, f) != NULL; lineno++) if (parse_line(line, &rule, lineno)) errs++; fclose(f); #ifdef DEBUG_APPRENTICE kdDebug(7018) << "apprentice: conf=" << conf << " file=" << magicfile << " m=" << (conf->magic ? "set" : "NULL") << " m->next=" << ((conf->magic && conf->magic->next) ? "set" : "NULL") << " last=" << (conf->last ? "set" : "NULL") << endl; kdDebug(7018) << "apprentice: read " << lineno << " lines, " << rule << " rules, " << errs << " errors" << endl; #endif #ifdef MIME_MAGIC_DEBUG_TABLE test_table(); #endif return (errs ? -1 : 0); } int KMimeMagic::buff_apprentice(char *buff) { char line[BUFSIZ + 2]; int errs = 0; int lineno = 1; char *start = buff; char *end; int count = 0; int rule = 0; int len = strlen(buff) + 1; /* parse it */ do { count = (len > BUFSIZ-1)?BUFSIZ-1:len; strncpy(line, start, count); line[count] = '\0'; if ((end = strchr(line, '\n'))) { *(++end) = '\0'; count = strlen(line); } else strcat(line, "\n"); start += count; len -= count; if (parse_line(line, &rule, lineno)) errs++; lineno++; } while (len > 0); #ifdef DEBUG_APPRENTICE kdDebug(7018) << "buff_apprentice: conf=" << conf << " m=" << (conf->magic ? "set" : "NULL") << " m->next=" << ((conf->magic && conf->magic->next) ? "set" : "NULL") << " last=" << (conf->last ? "set" : "NULL") << endl; kdDebug(7018) << "buff_apprentice: read " << lineno << " lines, " << rule << " rules, " << errs << " errors" << endl; #endif #ifdef MIME_MAGIC_DEBUG_TABLE test_table(); #endif return (errs ? -1 : 0); } /* * extend the sign bit if the comparison is to be signed */ static unsigned long signextend(struct magic *m, unsigned long v) { if (!(m->flag & UNSIGNED)) switch (m->type) { /* * Do not remove the casts below. They are vital. * When later compared with the data, the sign * extension must have happened. */ case BYTE: v = (char) v; break; case SHORT: case BESHORT: case LESHORT: v = (short) v; break; case DATE: case BEDATE: case LEDATE: case LONG: case BELONG: case LELONG: v = (long) v; break; case STRING: break; default: kdError(7018) << "" << "signextend" << ": can't happen: m->type=" << m->type << endl; return 998; //good value } return v; } /* * parse one line from magic file, put into magic[index++] if valid */ int KMimeMagic::parse(char *l, int #ifdef DEBUG_LINENUMBERS lineno #endif ) { int i = 0; struct magic *m; char *t, *s; /* allocate magic structure entry */ if ((m = (struct magic *) calloc(1, sizeof(struct magic))) == NULL) { kdError(7018) << "parse: Out of memory." << endl; return -1; } /* append to linked list */ m->next = NULL; if (!conf->magic || !conf->last) { conf->magic = conf->last = m; } else { conf->last->next = m; conf->last = m; } /* set values in magic structure */ m->flag = 0; m->cont_level = 0; #ifdef DEBUG_LINENUMBERS m->lineno = lineno; #endif while (*l == '>') { ++l; /* step over */ m->cont_level++; } if (m->cont_level != 0 && *l == '(') { ++l; /* step over */ m->flag |= INDIR; } /* get offset, then skip over it */ m->offset = (int) strtol(l, &t, 0); if (l == t) { kdError(7018) << "parse: offset " << l << " invalid" << endl; } l = t; if (m->flag & INDIR) { m->in.type = LONG; m->in.offset = 0; /* * read [.lbs][+-]nnnnn) */ if (*l == '.') { switch (*++l) { case 'l': m->in.type = LONG; break; case 's': m->in.type = SHORT; break; case 'b': m->in.type = BYTE; break; default: kdError(7018) << "parse: indirect offset type " << *l << " invalid" << endl; break; } l++; } s = l; if (*l == '+' || *l == '-') l++; if (isdigit((unsigned char) *l)) { m->in.offset = strtol(l, &t, 0); if (*s == '-') m->in.offset = -m->in.offset; } else t = l; if (*t++ != ')') { kdError(7018) << "parse: missing ')' in indirect offset" << endl; } l = t; } while (isascii((unsigned char) *l) && isdigit((unsigned char) *l)) ++l; EATAB; #define NBYTE 4 #define NSHORT 5 #define NLONG 4 #define NSTRING 6 #define NDATE 4 #define NBESHORT 7 #define NBELONG 6 #define NBEDATE 6 #define NLESHORT 7 #define NLELONG 6 #define NLEDATE 6 if (*l == 'u') { ++l; m->flag |= UNSIGNED; } /* get type, skip it */ if (strncmp(l, "byte", NBYTE) == 0) { m->type = BYTE; l += NBYTE; } else if (strncmp(l, "short", NSHORT) == 0) { m->type = SHORT; l += NSHORT; } else if (strncmp(l, "long", NLONG) == 0) { m->type = LONG; l += NLONG; } else if (strncmp(l, "string", NSTRING) == 0) { m->type = STRING; l += NSTRING; } else if (strncmp(l, "date", NDATE) == 0) { m->type = DATE; l += NDATE; } else if (strncmp(l, "beshort", NBESHORT) == 0) { m->type = BESHORT; l += NBESHORT; } else if (strncmp(l, "belong", NBELONG) == 0) { m->type = BELONG; l += NBELONG; } else if (strncmp(l, "bedate", NBEDATE) == 0) { m->type = BEDATE; l += NBEDATE; } else if (strncmp(l, "leshort", NLESHORT) == 0) { m->type = LESHORT; l += NLESHORT; } else if (strncmp(l, "lelong", NLELONG) == 0) { m->type = LELONG; l += NLELONG; } else if (strncmp(l, "ledate", NLEDATE) == 0) { m->type = LEDATE; l += NLEDATE; } else { kdError(7018) << "parse: type " << l << " invalid" << endl; return -1; } /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */ if (*l == '&') { ++l; m->mask = signextend(m, strtol(l, &l, 0)); } else m->mask = (unsigned long) ~0L; EATAB; switch (*l) { case '>': case '<': /* Old-style anding: "0 byte &0x80 dynamically linked" */ case '&': case '^': case '=': m->reln = *l; ++l; break; case '!': if (m->type != STRING) { m->reln = *l; ++l; break; } /* FALL THROUGH */ default: if (*l == 'x' && isascii((unsigned char) l[1]) && isspace((unsigned char) l[1])) { m->reln = *l; ++l; goto GetDesc; /* Bill The Cat */ } m->reln = '='; break; } EATAB; if (getvalue(m, &l)) return -1; /* * now get last part - the description */ GetDesc: EATAB; if (l[0] == '\b') { ++l; m->nospflag = 1; } else if ((l[0] == '\\') && (l[1] == 'b')) { ++l; ++l; m->nospflag = 1; } else m->nospflag = 0; // Copy description - until EOL or '#' (for comments) while (*l != '\0' && *l != '#' && i < MAXDESC-1) m->desc[i++] = *l++; m->desc[i] = '\0'; // Remove trailing spaces while (--i>0 && isspace( m->desc[i] )) m->desc[i] = '\0'; // old code //while ((m->desc[i++] = *l++) != '\0' && i < MAXDESC) /* NULLBODY */ ; #ifdef DEBUG_APPRENTICE kdDebug(7018) << "parse: line=" << lineno << " m=" << m << " next=" << m->next << " cont=" << m->cont_level << " desc=" << (m->desc ? m->desc : "NULL") << endl; #endif return 0; } /* * Read a numeric value from a pointer, into the value union of a magic * pointer, according to the magic type. Update the string pointer to point * just after the number read. Return 0 for success, non-zero for failure. */ static int getvalue(struct magic *m, char **p) { int slen; if (m->type == STRING) { *p = getstr(*p, m->value.s, sizeof(m->value.s), &slen); m->vallen = slen; } else if (m->reln != 'x') m->value.l = signextend(m, strtol(*p, p, 0)); return 0; } /* * Convert a string containing C character escapes. Stop at an unescaped * space or tab. Copy the converted version to "p", returning its length in * *slen. Return updated scan pointer as function result. */ static char * getstr(register char *s, register char *p, int plen, int *slen) { char *origs = s, *origp = p; char *pmax = p + plen - 1; register int c; register int val; while ((c = *s++) != '\0') { if (isspace((unsigned char) c)) break; if (p >= pmax) { kdError(7018) << "String too long: " << origs << endl; break; } if (c == '\\') { switch (c = *s++) { case '\0': goto out; default: *p++ = (char) c; break; case 'n': *p++ = '\n'; break; case 'r': *p++ = '\r'; break; case 'b': *p++ = '\b'; break; case 't': *p++ = '\t'; break; case 'f': *p++ = '\f'; break; case 'v': *p++ = '\v'; break; /* \ and up to 3 octal digits */ case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': val = c - '0'; c = *s++; /* try for 2 */ if (c >= '0' && c <= '7') { val = (val << 3) | (c - '0'); c = *s++; /* try for 3 */ if (c >= '0' && c <= '7') val = (val << 3) | (c - '0'); else --s; } else --s; *p++ = (char) val; break; /* \x and up to 3 hex digits */ case 'x': val = 'x'; /* Default if no digits */ c = hextoint(*s++); /* Get next char */ if (c >= 0) { val = c; c = hextoint(*s++); if (c >= 0) { val = (val << 4) + c; c = hextoint(*s++); if (c >= 0) { val = (val << 4) + c; } else --s; } else --s; } else --s; *p++ = (char) val; break; } } else *p++ = (char) c; } out: *p = '\0'; *slen = p - origp; //for ( char* foo = origp; foo < p ; ++foo ) // kdDebug(7018) << " " << *foo << endl; return s; } /* Single hex char to int; -1 if not a hex char. */ static int hextoint(int c) { if (!isascii((unsigned char) c)) return -1; if (isdigit((unsigned char) c)) return c - '0'; if ((c >= 'a') && (c <= 'f')) return c + 10 - 'a'; if ((c >= 'A') && (c <= 'F')) return c + 10 - 'A'; return -1; } /* * Convert the byte order of the data we are looking at */ static int mconvert(union VALUETYPE *p, struct magic *m) { switch (m->type) { case BYTE: return 1; case STRING: /* Null terminate */ p->s[sizeof(p->s) - 1] = '\0'; return 1; #ifndef WORDS_BIGENDIAN case SHORT: #endif case BESHORT: p->h = (short) ((p->hs[0] << 8) | (p->hs[1])); return 1; #ifndef WORDS_BIGENDIAN case LONG: case DATE: #endif case BELONG: case BEDATE: p->l = (long) ((p->hl[0] << 24) | (p->hl[1] << 16) | (p->hl[2] << 8) | (p->hl[3])); return 1; #ifdef WORDS_BIGENDIAN case SHORT: #endif case LESHORT: p->h = (short) ((p->hs[1] << 8) | (p->hs[0])); return 1; #ifdef WORDS_BIGENDIAN case LONG: case DATE: #endif case LELONG: case LEDATE: p->l = (long) ((p->hl[3] << 24) | (p->hl[2] << 16) | (p->hl[1] << 8) | (p->hl[0])); return 1; default: kdError(7018) << "mconvert: invalid type " << m->type << endl; return 0; } } static int mget(union VALUETYPE *p, unsigned char *s, struct magic *m, int nbytes) { long offset = m->offset; switch ( m->type ) { case BYTE: if ( offset + 1 > nbytes-1 ) // nbytes = (size of file) + 1 return 0; break; case SHORT: case BESHORT: case LESHORT: if ( offset + 2 > nbytes-1 ) return 0; break; case LONG: case BELONG: case LELONG: case DATE: case BEDATE: case LEDATE: if ( offset + 4 > nbytes-1 ) return 0; break; case STRING: break; } // The file length might be < sizeof(union VALUETYPE) (David) // -> pad with zeros (the 'file' command does it this way) // Thanks to Stan Covington <stan@calderasystems.com> for detailed report if (offset + (int)sizeof(union VALUETYPE) > nbytes) { int have = nbytes - offset; memset(p, 0, sizeof(union VALUETYPE)); if (have > 0) memcpy(p, s + offset, have); } else memcpy(p, s + offset, sizeof(union VALUETYPE)); if (!mconvert(p, m)) return 0; if (m->flag & INDIR) { switch (m->in.type) { case BYTE: offset = p->b + m->in.offset; break; case SHORT: offset = p->h + m->in.offset; break; case LONG: offset = p->l + m->in.offset; break; } if (offset + (int)sizeof(union VALUETYPE) > nbytes) return 0; memcpy(p, s + offset, sizeof(union VALUETYPE)); if (!mconvert(p, m)) return 0; } return 1; } static int mcheck(union VALUETYPE *p, struct magic *m) { register unsigned long l = m->value.l; register unsigned long v; int matched; if ((m->value.s[0] == 'x') && (m->value.s[1] == '\0')) { kdError(7018) << "BOINK" << endl; return 1; } switch (m->type) { case BYTE: v = p->b; break; case SHORT: case BESHORT: case LESHORT: v = p->h; break; case LONG: case BELONG: case LELONG: case DATE: case BEDATE: case LEDATE: v = p->l; break; case STRING: l = 0; /* * What we want here is: v = strncmp(m->value.s, p->s, * m->vallen); but ignoring any nulls. bcmp doesn't give * -/+/0 and isn't universally available anyway. */ v = 0; { register unsigned char *a = (unsigned char *) m->value.s; register unsigned char *b = (unsigned char *) p->s; register int len = m->vallen; Q_ASSERT(len); while (--len >= 0) if ((v = *b++ - *a++) != 0) break; } break; default: kdError(7018) << "mcheck: invalid type " << m->type << endl; return 0; /* NOTREACHED */ } #if 0 qDebug("Before signextend %08x", v); #endif v = signextend(m, v) & m->mask; #if 0 qDebug("After signextend %08x", v); #endif switch (m->reln) { case 'x': matched = 1; break; case '!': matched = v != l; break; case '=': matched = v == l; break; case '>': if (m->flag & UNSIGNED) matched = v > l; else matched = (long) v > (long) l; break; case '<': if (m->flag & UNSIGNED) matched = v < l; else matched = (long) v < (long) l; break; case '&': matched = (v & l) == l; break; case '^': matched = (v & l) != l; break; default: matched = 0; kdError(7018) << "mcheck: can't happen: invalid relation " << m->reln << "." << endl; break; /* NOTREACHED */ } return matched; } /* * magic_process - process input file fn. Opens the file and reads a * fixed-size buffer to begin processing the contents. */ void process(struct config_rec* conf, const QString & fn) { int fd = 0; unsigned char buf[HOWMANY + 1]; /* one extra for terminating '\0' */ KDE_struct_stat sb; int nbytes = 0; /* number of bytes read from a datafile */ int tagbytes = 0; /* size of prefixed tag */ QCString fileName = QFile::encodeName( fn ); /* * first try judging the file based on its filesystem status */ if (fsmagic(conf, fileName, &sb) != 0) { //resultBuf += "\n"; return; } if ((fd = KDE_open(fileName, O_RDONLY)) < 0) { /* We can't open it, but we were able to stat it. */ /* * if (sb.st_mode & 0002) addResult("writable, "); * if (sb.st_mode & 0111) addResult("executable, "); */ //kdDebug(7018) << "can't read `" << fn << "' (" << strerror(errno) << ")." << endl; conf->resultBuf = MIME_BINARY_UNREADABLE; return; } /* * try looking at the first HOWMANY bytes */ if ((nbytes = read(fd, (char *) buf, HOWMANY)) == -1) { kdError(7018) << "" << fn << " read failed (" << strerror(errno) << ")." << endl; conf->resultBuf = MIME_BINARY_UNREADABLE; (void)close(fd); return; } if ((tagbytes = tagmagic(buf, nbytes))) { // Read buffer at new position lseek(fd, tagbytes, SEEK_SET); nbytes = read(fd, (char*)buf, HOWMANY); if (nbytes < 0) { conf->resultBuf = MIME_BINARY_UNREADABLE; (void)close(fd); return; } } if (nbytes == 0) { conf->resultBuf = MIME_BINARY_ZEROSIZE; } else { buf[nbytes++] = '\0'; /* null-terminate it */ tryit(conf, buf, nbytes); } if ( conf->utimeConf && conf->utimeConf->restoreAccessTime( fn ) ) { /* * Try to restore access, modification times if read it. * This changes the "change" time (ctime), but we can't do anything * about that. */ struct utimbuf utbuf; utbuf.actime = sb.st_atime; utbuf.modtime = sb.st_mtime; (void) utime(fileName, &utbuf); } (void) close(fd); } static void tryit(struct config_rec* conf, unsigned char *buf, int nb) { /* try tests in /etc/magic (or surrogate magic file) */ if (match(conf, buf, nb)) return; /* try known keywords, check for ascii-ness too. */ if (ascmagic(conf, buf, nb) == 1) return; /* see if it's plain text */ if (textmagic(conf, buf, nb)) return; /* abandon hope, all ye who remain here */ conf->resultBuf = MIME_BINARY_UNKNOWN; conf->accuracy = 0; } static int fsmagic(struct config_rec* conf, const char *fn, KDE_struct_stat *sb) { int ret = 0; /* * Fstat is cheaper but fails for files you don't have read perms on. * On 4.2BSD and similar systems, use lstat() to identify symlinks. */ ret = KDE_lstat(fn, sb); /* don't merge into if; see "ret =" above */ if (ret) { return 1; } /* * if (sb->st_mode & S_ISUID) resultBuf += "setuid "; * if (sb->st_mode & S_ISGID) resultBuf += "setgid "; * if (sb->st_mode & S_ISVTX) resultBuf += "sticky "; */ switch (sb->st_mode & S_IFMT) { case S_IFDIR: conf->resultBuf = MIME_INODE_DIR; return 1; case S_IFCHR: conf->resultBuf = MIME_INODE_CDEV; return 1; case S_IFBLK: conf->resultBuf = MIME_INODE_BDEV; return 1; /* TODO add code to handle V7 MUX and Blit MUX files */ #ifdef S_IFIFO case S_IFIFO: conf->resultBuf = MIME_INODE_FIFO; return 1; #endif #ifdef S_IFLNK case S_IFLNK: { char buf[BUFSIZ + BUFSIZ + 4]; register int nch; KDE_struct_stat tstatbuf; if ((nch = readlink(fn, buf, BUFSIZ - 1)) <= 0) { conf->resultBuf = MIME_INODE_LINK; //conf->resultBuf += "\nunreadable"; return 1; }