apprentice.c File Reference

#include <rz_userconf.h>
#include <rz_util.h>
#include <ctype.h>
#include "file.h"
#include "patchlevel.h"
#include "mdump.c"

Go to the source code of this file.

Classes
struct	rz_magic_entry
struct	type_tbl_s

Macros
#define	QUICK   0
#define	EATAB
#define	LOWCASE(l)   (isupper((ut8)(l)) ? tolower((ut8)(l)) : (l))
#define	MAP_FILE   0
#define	XX(s)   s, (sizeof(s) - 1)
#define	XX_NULL   "", 0
#define	MULT   10
#define	SYMLEN   4 /* strlen("text") */
#define	ALLOC_CHUNK   (size_t)10
#define	ALLOC_INCR   (size_t)200
#define	RET   1

Functions
static int	getvalue (RzMagic *ms, struct rz_magic *, const char **, int)
static int	hextoint (int)
static const char *	getstr (RzMagic *, const char *, char *, int, int *, int)
static int	parse (RzMagic *, struct rz_magic_entry **, ut32 *, const char *, size_t, int)
static int	parse_mime (RzMagic *, struct rz_magic_entry **, ut32 *, const char *)
static void	eatsize (const char **)
static int	apprentice_1 (RzMagic *, const char *, int, struct mlist *)
static size_t	apprentice_r_magic_strength (const struct rz_magic *)
static int	apprentice_sort (const void *, const void *)
static int	apprentice_load (RzMagic *, struct rz_magic **, ut32 *, const char *, int)
static void	byteswap (struct rz_magic *, ut32)
static void	bs1 (struct rz_magic *)
static ut16	swap2 (ut16)
static ut32	swap4 (ut32)
static ut64	swap8 (ut64)
static char *	mkdbname (const char *, int)
static int	apprentice_map (RzMagic *, struct rz_magic **, ut32 *, const char *)
static int	apprentice_compile (RzMagic *, struct rz_magic **, ut32 *, const char *)
static int	check_format_type (const char *, int)
static int	check_format (RzMagic *, struct rz_magic *)
static int	get_op (char)
static int	get_type (const char *l, const char **t)
static void	init_file_tables (void)
void	file_delmagic (struct rz_magic *p, int type, size_t entries)
struct mlist *	file_apprentice (RzMagic *ms, const char *fn, int action)
static void	set_test_type (struct rz_magic *mstart, struct rz_magic *m)
static const char *	bgets (char *line, size_t line_sz, const char *data)
static void	load_b (RzMagic *ms, int action, const char *data, int *errs, struct rz_magic_entry **marray, ut32 *marraycount)
static void	load_1 (RzMagic *ms, int action, const char *file, int *errs, struct rz_magic_entry **marray, ut32 *marraycount)
ut64	file_signextend (RzMagic *ms, struct rz_magic *m, ut64 v)
static int	string_modifier_check (RzMagic *ms, struct rz_magic *m)
static int	get_cond (const char *l, const char **t)
static int	check_cond (RzMagic *ms, int cond, ut32 cont_level)
void	file_showstr (FILE *fp, const char *s, size_t len)

Variables
static int	magic_file_formats [FILE_NAMES_SIZE]
static const size_t	file_nformats = FILE_NAMES_SIZE
static const char *	magic_file_names [FILE_NAMES_SIZE]
static const size_t	file_nnames = FILE_NAMES_SIZE
static size_t	maxmagic = 0
static size_t	magicsize = sizeof(struct rz_magic)
static const char	usg_hdr [] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"
static const char	mime_marker [] = "!:mime"
static const size_t	mime_marker_len = sizeof(mime_marker) - 1
static const struct type_tbl_s	type_tbl []
static const ut32	ar []
static const char	ext [] = ".mgc"

Macro Definition Documentation

ALLOC_CHUNK

#define ALLOC_CHUNK   (size_t)10

ALLOC_INCR

#define ALLOC_INCR   (size_t)200

EATAB

#define EATAB

Value:

    { \
        while (isascii((ut8)*l) && isspace((ut8)*l)) { \
            l++; \
        } \
    }

Definition at line 57 of file apprentice.c.

LOWCASE

#define LOWCASE

(

l

)

(isupper((ut8)(l)) ? tolower((ut8)(l)) : (l))

Definition at line 63 of file apprentice.c.

MAP_FILE

#define MAP_FILE   0

Definition at line 66 of file apprentice.c.

MULT

#define MULT   10

QUICK

#define QUICK   0

Definition at line 46 of file apprentice.c.

RET

#define RET   1

SYMLEN

#define SYMLEN   4 /* strlen("text") */

XX

#define XX

(

s

)

s, (sizeof(s) - 1)

XX_NULL

#define XX_NULL   "", 0

Function Documentation

apprentice_1()

static int apprentice_1 ( RzMagic *  ms, const char *  fn, int  action, struct mlist *  mlist  )

static

Definition at line 193 of file apprentice.c.

                                                                                      {
    struct rz_magic *magic = NULL;
    ut32 nmagic = 0;
    struct mlist *ml;
    int rv = -1;
    int mapped;
 
    if (!ms) {
        return -1;
    }
    ms->haderr = 0;
    if (magicsize != FILE_MAGICSIZE) {
        file_error(ms, 0, "magic element size %lu != %lu",
            (unsigned long)(size_t)sizeof(*magic),
            (unsigned long)FILE_MAGICSIZE);
        return -1;
    }
 
    ms->file = fn; // fix use of ms->file before being initialized
    if (action == FILE_COMPILE) {
        rv = apprentice_load(ms, &magic, &nmagic, fn, action);
        if (rv != 0) {
            return -1;
        }
        rv = apprentice_compile(ms, &magic, &nmagic, fn);
        free(magic);
        return rv;
    }
 
    if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) {
        // if (ms->flags & RZ_MAGIC_CHECK)
        //  file_magwarn(ms, "using regular magic file `%s'", fn);
        rv = apprentice_load(ms, &magic, &nmagic, fn, action);
        if (rv != 0) {
            return -1;
        }
    }
 
    mapped = rv;
 
    if (!magic) {
        file_delmagic(magic, mapped, nmagic);
        return -1;
    }
 
    if (!(ml = malloc(sizeof(*ml)))) {
        file_delmagic(magic, mapped, nmagic);
        file_oomem(ms, sizeof(*ml));
        return -1;
    }
 
    ml->magic = magic;
    ml->nmagic = nmagic;
    ml->mapped = mapped;
 
    mlist->prev->next = ml;
    ml->prev = mlist->prev;
    ml->next = mlist;
    mlist->prev = ml;
    return 0;
}

References test-lz4-speed::action, apprentice_compile(), apprentice_load(), apprentice_map(), file_delmagic(), file_error(), file_oomem(), fn, free(), magicsize, malloc(), and NULL.

Referenced by file_apprentice().

apprentice_compile()

static int apprentice_compile ( RzMagic *  ms, struct rz_magic **  magicp, ut32 *  nmagicp, const char *  fn  )

static

Definition at line 1942 of file apprentice.c.

                                                                                                    {
    int fd;
    char *dbname;
    int rv = -1;
 
    dbname = mkdbname(fn, 1);
 
    if (!dbname) {
        goto out;
    }
 
    if ((fd = rz_sys_open(dbname, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0644)) == -1) {
        file_error(ms, errno, "cannot open `%s'", dbname);
        goto out;
    }
 
    if (write(fd, ar, sizeof(ar)) != (int)sizeof(ar)) {
        file_error(ms, errno, "error writing `%s'", dbname);
        goto beach;
    }
 
    if (lseek(fd, (off_t)sizeof(struct rz_magic), SEEK_SET) != sizeof(struct rz_magic)) {
        file_error(ms, errno, "error seeking `%s'", dbname);
        goto beach;
    }
 
    if (write(fd, *magicp, (sizeof(struct rz_magic) * *nmagicp)) != (int)(sizeof(struct rz_magic) * *nmagicp)) {
        file_error(ms, errno, "error writing `%s'", dbname);
        goto beach;
    }
 
    rv = 0;
beach:
    (void)close(fd);
out:
    free(dbname);
    return rv;
}

References ar, close, fd, file_error(), fn, free(), lseek, mkdbname(), O_BINARY, O_CREAT, O_TRUNC, O_WRONLY, out, rz_sys_open(), SEEK_SET, and write.

Referenced by apprentice_1().

apprentice_load()

static int apprentice_load ( RzMagic *  ms, struct rz_magic **  magicp, ut32 *  nmagicp, const char *  fn, int  action  )

static

Definition at line 593 of file apprentice.c.

                                                                                                             {
    ut32 marraycount, i, mentrycount = 0, starttest;
    struct rz_magic_entry *marray;
    struct stat st;
    int errs = 0;
#if __WINDOWS__
    HANDLE hdir;
    WIN32_FIND_DATAW entry;
    wchar_t dir[MAX_PATH];
    wchar_t *wcpath;
    char *cfname;
    char subfn[1024];
#else
    DIR *dir;
    struct dirent *d;
    char subfn[MAXPATHLEN];
#endif
    ms->flags |= RZ_MAGIC_CHECK; /* Enable checks for parsed files */
 
    maxmagic = MAXMAGIS;
    if (!(marray = calloc(maxmagic, sizeof(*marray)))) {
        file_oomem(ms, maxmagic * sizeof(*marray));
        return -1;
    }
    marraycount = 0;
 
    /* print silly verbose header for USG compat. */
    if (action == FILE_CHECK) {
        eprintf("%s\n", usg_hdr);
    }
 
    /* load directory or file */
    if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
#if __WINDOWS__ && !defined(__CYGWIN__)
        if ((wcpath = rz_utf8_to_utf16(fn))) {
            swprintf(dir, _countof(dir), L"%ls\\*.*", wcpath);
            hdir = FindFirstFileW(dir, &entry);
            if (!(hdir == INVALID_HANDLE_VALUE)) {
                do {
                    if (wcsncmp(entry.cFileName, L".", 1) == 0)
                        continue;
                    if ((cfname = rz_utf16_to_utf8(entry.cFileName))) {
                        snprintf(subfn, sizeof(subfn), "%s/%s", fn, cfname);
                        if (stat(subfn, &st) == 0 && S_ISREG(st.st_mode)) {
                            load_1(ms, action, subfn, &errs, &marray, &marraycount);
                        }
                        free(cfname);
                    }
                } while (FindNextFileW(hdir, &entry));
                FindClose(hdir);
            }
            free(wcpath);
        } else {
            errs++;
        }
#else
        dir = opendir(fn);
        if (dir) {
            while ((d = readdir(dir))) {
                if (*d->d_name == '.') {
                    continue;
                }
                snprintf(subfn, sizeof(subfn), "%s/%s", fn, d->d_name);
                if (stat(subfn, &st) == 0 && S_ISREG(st.st_mode)) {
                    load_1(ms, action, subfn, &errs, &marray, &marraycount);
                }
                // else perror (subfn);
            }
            closedir(dir);
        } else {
            errs++;
        }
#endif
    } else {
        load_1(ms, action, fn, &errs, &marray, &marraycount);
    }
    if (errs) {
        goto out;
    }
 
    /* Set types of tests */
    for (i = 0; i < marraycount;) {
        if (marray[i].mp->cont_level != 0) {
            i++;
            continue;
        }
 
        starttest = i;
        do {
            set_test_type(marray[starttest].mp, marray[i].mp);
            if (ms->flags & RZ_MAGIC_DEBUG) {
                (void)fprintf(stderr, "%s%s%s: %s\n",
                    marray[i].mp->mimetype,
                    marray[i].mp->mimetype[0] == '\0' ? "" : "; ",
                    marray[i].mp->desc[0] ? marray[i].mp->desc : "(no description)",
                    marray[i].mp->flag & BINTEST ? "binary" : "text");
                if (marray[i].mp->flag & BINTEST) {
#define SYMLEN 4 /* strlen("text") */
                    char *p = strstr(marray[i].mp->desc, "text");
                    if (p && (p == marray[i].mp->desc || isspace((unsigned char)p[-1])) &&
                        (p + SYMLEN - marray[i].mp->desc == MAXstring ||
                            (p[SYMLEN] == '\0' || isspace((unsigned char)p[SYMLEN])))) {
                        (void)fprintf(stderr,
                            "*** Possible binary test for text type\n");
                    }
                }
            }
        } while (++i < marraycount && marray[i].mp->cont_level != 0);
    }
 
    qsort(marray, marraycount, sizeof(*marray), apprentice_sort);
 
    /*
     * Make sure that any level 0 "default" line is last (if one exists).
     */
    for (i = 0; i < marraycount; i++) {
        if (marray[i].mp->cont_level == 0 &&
            marray[i].mp->type == FILE_DEFAULT) {
            while (++i < marraycount) {
                if (marray[i].mp->cont_level == 0) {
                    break;
                }
            }
            if (i != marraycount) {
                ms->line = marray[i].mp->lineno; /* XXX - Ugh! */
                file_magwarn(ms, "level 0 \"default\" did not sort last");
            }
            break;
        }
    }
 
    for (i = 0; i < marraycount; i++) {
        mentrycount += marray[i].cont_count;
    }
 
    if (!(*magicp = malloc(1 + (sizeof(**magicp) * mentrycount)))) {
        file_oomem(ms, sizeof(**magicp) * mentrycount);
        errs++;
        goto out;
    }
 
    mentrycount = 0;
    for (i = 0; i < marraycount; i++) {
        (void)memcpy(*magicp + mentrycount, marray[i].mp,
            marray[i].cont_count * sizeof(**magicp));
        mentrycount += marray[i].cont_count;
    }
out:
    for (i = 0; i < marraycount; i++) {
        free(marray[i].mp);
    }
    free(marray);
    if (errs) {
        *magicp = NULL;
        *nmagicp = 0;
        return errs;
    }
    *nmagicp = mentrycount;
    return 0;
}

References test-lz4-speed::action, apprentice_sort(), calloc(), rz_magic_entry::cont_count, d, eprintf, file_magwarn(), file_oomem(), fn, free(), HANDLE, i, INVALID_HANDLE_VALUE, isspace, L, load_1(), malloc(), maxmagic, memcpy(), rz_magic_entry::mp, NULL, out, p, qsort(), readdir, S_ISDIR, S_ISREG, set_test_type(), snprintf, stat, SYMLEN, and usg_hdr.

Referenced by apprentice_1().

apprentice_map()

static int apprentice_map ( RzMagic *  ms, struct rz_magic **  magicp, ut32 *  nmagicp, const char *  fn  )

static

Definition at line 1836 of file apprentice.c.

                                                                                                {
    int fd;
    struct stat st;
    ut32 *ptr;
    ut32 version;
    int needsbyteswap;
    char *dbname = NULL;
    void *mm = NULL;
 
    dbname = mkdbname(fn, 0);
    if (!dbname) {
        goto error2;
    }
 
    if ((fd = rz_sys_open(dbname, O_RDONLY | O_BINARY, 0)) == -1) {
        goto error2;
    }
 
    if (fstat(fd, &st) == -1) {
        file_error(ms, errno, "cannot stat `%s'", dbname);
        goto error1;
    }
    if (st.st_size < 8) {
        file_error(ms, 0, "file `%s' is too small", dbname);
        goto error1;
    }
 
#if QUICK
    if ((mm = mmap(0, (size_t)st.st_size, PROT_READ, // OPENBSDBUG  |PROT_WRITE,
             MAP_PRIVATE | MAP_FILE, fd, (off_t)0)) == MAP_FAILED) {
        file_error(ms, errno, "cannot map `%s'", dbname);
        goto error1;
    }
#define RET 2
#else
    if (!(mm = malloc((size_t)st.st_size))) {
        file_oomem(ms, (size_t)st.st_size);
        goto error1;
    }
    if (read(fd, mm, (size_t)st.st_size) != (size_t)st.st_size) {
        file_badread(ms);
        goto error1;
    }
#define RET 1
#endif
    *magicp = mm;
    (void)close(fd);
    fd = -1;
    ptr = (ut32 *)(void *)*magicp;
 
    if (*ptr != MAGICNO) {
        if (swap4(*ptr) != MAGICNO) {
            // OPENBSDBUG file_error(ms, 0, "bad magic in `%s'");
            file_error(ms, 0, "bad magic in `%s'", dbname);
            goto error1;
        }
        needsbyteswap = 1;
    } else {
        needsbyteswap = 0;
    }
 
    version = needsbyteswap ? swap4(ptr[1]) : ptr[1];
    if (version != VERSIONNO) {
        file_error(ms, 0, "File %d.%d supports only %d version magic "
                  "files. `%s' is version %d",
            FILE_VERSION_MAJOR, patchlevel,
            VERSIONNO, dbname, version);
        goto error1;
    }
    *nmagicp = (ut32)(st.st_size / sizeof(struct rz_magic));
    if (*nmagicp > 0) {
        (*nmagicp)--;
    }
    (*magicp)++;
    if (needsbyteswap) {
        byteswap(*magicp, *nmagicp);
    }
    free(dbname);
    return RET;
 
error1:
    if (fd != -1) {
        (void)close(fd);
    }
    if (mm) {
#if QUICK
        (void)munmap((void *)mm, (size_t)st.st_size);
#else
        free(mm);
#endif
    } else {
        *magicp = NULL;
        *nmagicp = 0;
    }
error2:
    free(dbname);
    return -1;
}

References byteswap(), close, fd, file_badread(), file_error(), file_oomem(), FILE_VERSION_MAJOR, fn, free(), fstat, malloc(), MAP_FILE, MAP_PRIVATE, mkdbname(), mmap, munmap, NULL, O_BINARY, O_RDONLY, patchlevel, PROT_READ, read(), RET, rz_sys_open(), swap4(), and version.

Referenced by apprentice_1().

apprentice_r_magic_strength()

static size_t apprentice_r_magic_strength ( const struct rz_magic *  m )

static

Definition at line 325 of file apprentice.c.

                                                                    {
#define MULT 10
    size_t val = 2 * MULT; /* baseline strength */
 
    switch (m->type) {
    case FILE_DEFAULT: /* make sure this sorts last */
        return 0;
    case FILE_BYTE:
        val += 1 * MULT;
        break;
    case FILE_SHORT:
    case FILE_LESHORT:
    case FILE_BESHORT:
        val += 2 * MULT;
        break;
    case FILE_LONG:
    case FILE_LELONG:
    case FILE_BELONG:
    case FILE_MELONG:
        val += 4 * MULT;
        break;
    case FILE_PSTRING:
    case FILE_STRING:
        val += m->vallen * MULT;
        break;
    case FILE_BESTRING16:
    case FILE_LESTRING16:
        val += m->vallen * MULT / 2;
        break;
    case FILE_SEARCH:
    case FILE_REGEX:
        val += m->vallen * RZ_MAX(MULT / m->vallen, 1);
        break;
    case FILE_DATE:
    case FILE_LEDATE:
    case FILE_BEDATE:
    case FILE_MEDATE:
    case FILE_LDATE:
    case FILE_LELDATE:
    case FILE_BELDATE:
    case FILE_MELDATE:
    case FILE_FLOAT:
    case FILE_BEFLOAT:
    case FILE_LEFLOAT:
        val += 4 * MULT;
        break;
    case FILE_QUAD:
    case FILE_BEQUAD:
    case FILE_LEQUAD:
    case FILE_QDATE:
    case FILE_LEQDATE:
    case FILE_BEQDATE:
    case FILE_QLDATE:
    case FILE_LEQLDATE:
    case FILE_BEQLDATE:
    case FILE_DOUBLE:
    case FILE_BEDOUBLE:
    case FILE_LEDOUBLE:
        val += 8 * MULT;
        break;
    default:
        eprintf("Bad type %d\n", m->type);
        abort();
    }
 
    switch (m->reln) {
    case 'x': /* matches anything penalize */
    case '!': /* matches almost anything penalize */
        val = 0;
        break;
    case '=': /* Exact match, prefer */
        val += MULT;
        break;
    case '>':
    case '<': /* comparison match reduce strength */
        val -= 2 * MULT;
        break;
    case '^':
    case '&': /* masking bits, we could count them too */
        val -= MULT;
        break;
    default:
        eprintf("Bad relation %c\n", m->reln);
        abort();
    }
    return val ? val : 1; /* ensure we only return 0 for FILE_DEFAULT */
}

References eprintf, regress::m, MULT, RZ_MAX, and val.

Referenced by apprentice_sort().

apprentice_sort()

static int apprentice_sort ( const void *  a, const void *  b  )

static

Definition at line 416 of file apprentice.c.

                                                         {
    const struct rz_magic_entry *ma = a;
    const struct rz_magic_entry *mb = b;
    size_t sa = apprentice_r_magic_strength(ma->mp);
    size_t sb = apprentice_r_magic_strength(mb->mp);
    if (sa == sb) {
        return 0;
    }
    if (sa > sb) {
        return -1;
    }
    return 1;
}

References a, apprentice_r_magic_strength(), b, rz_magic_entry::mp, and sb.

Referenced by apprentice_load().

bgets()

static const char* bgets ( char *  line, size_t  line_sz, const char *  data  )

static

Definition at line 491 of file apprentice.c.

                                                                       {
    if (!*data) {
        return NULL;
    }
    const char *nl = strchr(data, '\n');
    const int nlsz = nl
        ? nl - data + 1
        : RZ_MIN(line_sz, strlen(data));
    rz_str_ncpy(line, data, nlsz);
    if (!data[nlsz]) {
        return NULL;
    }
    return data + nlsz;
}

References setup::line, NULL, RZ_MIN, and rz_str_ncpy().

Referenced by load_b().

bs1()

static void bs1 ( struct rz_magic *  m )

static

Definition at line 2076 of file apprentice.c.

                                    {
    m->cont_level = swap2(m->cont_level);
    m->offset = swap4((ut32)m->offset);
    m->in_offset = swap4((ut32)m->in_offset);
    m->lineno = swap4((ut32)m->lineno);
    if (MAGIC_IS_STRING(m->type)) {
        m->str_range = swap4(m->str_range);
        m->str_flags = swap4(m->str_flags);
    } else {
        m->value.q = swap8(m->value.q);
        m->num_mask = swap8(m->num_mask);
    }
}

References regress::m, swap2(), swap4(), and swap8().

Referenced by byteswap().

byteswap()

static void byteswap ( struct rz_magic *  magic, ut32  nmagic  )

static

Definition at line 2011 of file apprentice.c.

                                                          {
    ut32 i;
    for (i = 0; i < nmagic; i++) {
        bs1(&magic[i]);
    }
}

References bs1(), and i.

Referenced by apprentice_map().

check_cond()

static int check_cond ( RzMagic *  ms, int  cond, ut32  cont_level  )

static

Definition at line 914 of file apprentice.c.

                                                              {
    int last_cond;
    last_cond = ms->c.li[cont_level].last_cond;
 
    switch (cond) {
    case COND_IF:
        if (last_cond != COND_NONE && last_cond != COND_ELIF) {
            if (ms->flags & RZ_MAGIC_CHECK) {
                file_magwarn(ms, "syntax error: `if'");
            }
            return -1;
        }
        last_cond = COND_IF;
        break;
    case COND_ELIF:
        if (last_cond != COND_IF && last_cond != COND_ELIF) {
            if (ms->flags & RZ_MAGIC_CHECK) {
                file_magwarn(ms, "syntax error: `elif'");
            }
            return -1;
        }
        last_cond = COND_ELIF;
        break;
    case COND_ELSE:
        if (last_cond != COND_IF && last_cond != COND_ELIF) {
            if (ms->flags & RZ_MAGIC_CHECK) {
                file_magwarn(ms, "syntax error: `else'");
            }
            return -1;
        }
        last_cond = COND_NONE;
        break;
    case COND_NONE:
        last_cond = COND_NONE;
        break;
    }
 
    ms->c.li[cont_level].last_cond = last_cond;
    return 0;
}

References cond, and file_magwarn().

Referenced by parse().

check_format()

static int check_format ( RzMagic *  ms, struct rz_magic *  m  )

static

Definition at line 1515 of file apprentice.c.

                                                         {
    char *ptr;
 
    for (ptr = m->desc; *ptr; ptr++) {
        if (*ptr == '%') {
            break;
        }
    }
    if (*ptr == '\0') {
        /* No format string; ok */
        return 1;
    }
 
    if (file_nformats != file_nnames) {
        return -1;
    }
 
    if (m->type >= file_nformats) {
        file_magwarn(ms, "Internal error inconsistency between "
                 "m->type and format strings");
        return -1;
    }
    if (magic_file_formats[m->type] == FILE_FMT_NONE) {
        file_magwarn(ms, "No format string for `%s' with description "
                 "`%s'",
            m->desc, magic_file_names[m->type]);
        return -1;
    }
 
    ptr++;
    if (ptr && check_format_type(ptr, magic_file_formats[m->type]) == -1) {
        /*
         * TODO: this error message is unhelpful if the format
         * string is not one character long
         */
        file_magwarn(ms, "Printf format `%c' is not valid for type "
                 "`%s' in description `%s'",
            ptr && *ptr ? *ptr : '?',
            magic_file_names[m->type], m->desc);
        return -1;
    }
 
    for (; *ptr; ptr++) {
        if (*ptr == '%') {
            file_magwarn(ms,
                "Too many format strings (should have at most one) "
                "for `%s' with description `%s'",
                magic_file_names[m->type], m->desc);
            return -1;
        }
    }
    return 0;
}

References check_format_type(), file_magwarn(), file_nformats, file_nnames, regress::m, magic_file_formats, and magic_file_names.

Referenced by parse().

check_format_type()

static int check_format_type ( const char *  ptr, int  type  )

static

Definition at line 1380 of file apprentice.c.

                                                        {
    int quad = 0;
    if (*ptr == '\0') {
        /* Missing format string; bad */
        return -1;
    }
 
    switch (type) {
    case FILE_FMT_QUAD:
        quad = 1;
        /*FALLTHROUGH*/
    case FILE_FMT_NUM:
        if (*ptr == '-') {
            ptr++;
        }
        if (*ptr == '.') {
            ptr++;
        }
        while (isdigit((ut8)*ptr)) {
            ptr++;
        }
        if (*ptr == '.') {
            ptr++;
        }
        while (isdigit((ut8)*ptr)) {
            ptr++;
        }
        if (quad) {
            if (*ptr++ != 'l') {
                return -1;
            }
            if (*ptr++ != 'l') {
                return -1;
            }
        }
        switch (*ptr++) {
        case 'l':
            switch (*ptr++) {
            case 'i':
            case 'd':
            case 'u':
            case 'x':
            case 'X':
                return 0;
            default:
                return -1;
            }
        case 'h':
            switch (*ptr++) {
            case 'h':
                switch (*ptr++) {
                case 'i':
                case 'd':
                case 'u':
                case 'x':
                case 'X':
                    return 0;
                default:
                    return -1;
                }
            case 'd':
                return 0;
            default:
                return -1;
            }
        case 'i':
        case 'c':
        case 'd':
        case 'u':
        case 'x':
        case 'X':
            return 0;
        default:
            return -1;
        }
    case FILE_FMT_FLOAT:
    case FILE_FMT_DOUBLE:
        if (*ptr == '-') {
            ptr++;
        }
        if (*ptr == '.') {
            ptr++;
        }
        while (isdigit((ut8)*ptr)) {
            ptr++;
        }
        if (*ptr == '.') {
            ptr++;
        }
        while (isdigit((ut8)*ptr)) {
            ptr++;
        }
 
        switch (*ptr++) {
        case 'e':
        case 'E':
        case 'f':
        case 'F':
        case 'g':
        case 'G':
            return 0;
 
        default:
            return -1;
        }
    case FILE_FMT_STR:
        if (*ptr == '-') {
            ptr++;
        }
        while (isdigit((ut8)*ptr)) {
            ptr++;
        }
        if (*ptr == '.') {
            ptr++;
            while (isdigit((ut8)*ptr)) {
                ptr++;
            }
        }
 
        if (*ptr++ == 's') {
            return 0;
        }
        return -1;
    default:
        /* internal error */
        abort();
    }
    /*NOTREACHED*/
    return -1;
}

References isdigit, and type.

Referenced by check_format().

eatsize()

static void eatsize ( const char **  p )

static

Definition at line 1811 of file apprentice.c.

                                    {
    const char *l = *p;
 
    if (LOWCASE(*l) == 'u') {
        l++;
    }
 
    switch (LOWCASE(*l)) {
    case 'l': /* long */
    case 's': /* short */
    case 'h': /* short */
    case 'b': /* char/byte */
    case 'c': /* char/byte */
        l++;
        /*FALLTHROUGH*/
    default:
        break;
    }
 
    *p = l;
}

References LOWCASE, and p.

Referenced by getvalue(), and parse().

file_apprentice()

struct mlist* file_apprentice	(	RzMagic *	ms,
		const char *	fn,
		int	action
	)

Definition at line 277 of file apprentice.c.

                                                                       {
    char *p, *mfn;
    int file_err, errs = -1;
    struct mlist *mlist;
 
    init_file_tables();
    if (!fn) {
        return NULL;
    }
 
    if (!(mfn = strdup(fn))) {
        file_oomem(ms, strlen(fn));
        return NULL;
    }
    fn = mfn;
 
    if (!(mlist = malloc(sizeof(*mlist)))) {
        free(mfn);
        file_oomem(ms, sizeof(*mlist));
        return NULL;
    }
    mlist->next = mlist->prev = mlist;
 
    while (fn) {
        p = strchr(fn, PATHSEP);
        if (p) {
            *p++ = '\0';
        }
        if (*fn == '\0') {
            break;
        }
        file_err = apprentice_1(ms, fn, action, mlist);
        errs = RZ_MAX(errs, file_err);
        fn = p;
    }
    if (errs == -1) {
        free(mfn);
        RZ_FREE(mlist);
        file_error(ms, 0, "could not find any magic files!");
        return NULL;
    }
    free(mfn);
    return mlist;
}

References test-lz4-speed::action, apprentice_1(), file_error(), file_oomem(), fn, free(), init_file_tables(), malloc(), NULL, p, RZ_FREE, RZ_MAX, and strdup().

Referenced by rz_magic_check(), rz_magic_compile(), rz_magic_load(), and rz_magic_load_buffer().

file_delmagic()

void file_delmagic	(	struct rz_magic *	p,
		int	type,
		size_t	entries
	)

Definition at line 255 of file apprentice.c.

                                                                 {
    if (p) {
        switch (type) {
#if QUICK
        case 2:
            p--;
            (void)munmap((void *)p, sizeof(*p) * (entries + 1));
            break;
#endif
        case 1:
            p--;
            /*FALLTHROUGH*/
        case 0:
            RZ_FREE(p);
            break;
        default:
            abort(); // do not abort, ever XXX this is a lib, so it shouldn just report an error
        }
    }
}

References entries(), munmap, p, RZ_FREE, and type.

Referenced by apprentice_1(), and free_mlist().

file_showstr()

void file_showstr	(	FILE *	fp,
		const char *	s,
		size_t	len
	)

Definition at line 1778 of file apprentice.c.

                                                       {
    for (;;) {
        char c = *s++;
        if (len == ~0U) {
            if (c == '\0') {
                break;
            }
        } else {
            if (len-- == 0) {
                break;
            }
        }
        if (c >= 040 && c <= 0176) { /* TODO isprint && !iscntrl */
            fputc(c, fp);
        } else {
            fputc('\\', fp);
            switch (c) {
            case '\a': fputc('a', fp); break;
            case '\b': fputc('b', fp); break;
            case '\f': fputc('f', fp); break;
            case '\n': fputc('n', fp); break;
            case '\r': fputc('r', fp); break;
            case '\t': fputc('t', fp); break;
            case '\v': fputc('v', fp); break;
            default: fprintf(fp, "%.3o", c & 0377); break;
            }
        }
    }
}

References c, len, and s.

Referenced by file_mdump(), and mdebug().

file_signextend()

ut64 file_signextend	(	RzMagic *	ms,
		struct rz_magic *	m,
		ut64	v
	)

Definition at line 757 of file apprentice.c.

                                                              {
    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 FILE_BYTE:
            v = (char)v;
            break;
        case FILE_SHORT:
        case FILE_BESHORT:
        case FILE_LESHORT:
            v = (short)v;
            break;
        case FILE_DATE:
        case FILE_BEDATE:
        case FILE_LEDATE:
        case FILE_MEDATE:
        case FILE_LDATE:
        case FILE_BELDATE:
        case FILE_LELDATE:
        case FILE_MELDATE:
        case FILE_LONG:
        case FILE_BELONG:
        case FILE_LELONG:
        case FILE_MELONG:
        case FILE_FLOAT:
        case FILE_BEFLOAT:
        case FILE_LEFLOAT:
            v = (int32_t)v;
            break;
        case FILE_QUAD:
        case FILE_BEQUAD:
        case FILE_LEQUAD:
        case FILE_QDATE:
        case FILE_QLDATE:
        case FILE_BEQDATE:
        case FILE_BEQLDATE:
        case FILE_LEQDATE:
        case FILE_LEQLDATE:
        case FILE_DOUBLE:
        case FILE_BEDOUBLE:
        case FILE_LEDOUBLE:
            v = (int64_t)v;
            break;
        case FILE_STRING:
        case FILE_PSTRING:
        case FILE_BESTRING16:
        case FILE_LESTRING16:
        case FILE_REGEX:
        case FILE_SEARCH:
        case FILE_DEFAULT:
            break;
        default:
            if (ms->flags & RZ_MAGIC_CHECK) {
                file_magwarn(ms, "cannot happen: m->type=%d\n",
                    m->type);
            }
            return ~0U;
        }
    }
    return v;
}

References file_magwarn(), regress::m, UNSIGNED, and v.

Referenced by getvalue(), magiccheck(), mprint(), and parse().

get_cond()

static int get_cond ( const char *  l, const char **  t  )

static

Definition at line 889 of file apprentice.c.

                                                   {
    const struct cond_tbl_s *p;
    static const struct cond_tbl_s {
        char name[8];
        size_t len;
        int cond;
    } cond_tbl[] = {
        { "if", 2, COND_IF },
        { "elif", 4, COND_ELIF },
        { "else", 4, COND_ELSE },
        { "", 0, COND_NONE },
    };
 
    for (p = cond_tbl; p->len; p++) {
        if (strncmp(l, p->name, p->len) == 0 &&
            isspace((ut8)l[p->len])) {
            if (t) {
                *t = l + p->len;
            }
            break;
        }
    }
    return p->cond;
}

References cond, isspace, len, and p.

Referenced by parse().

get_op()

static int get_op ( char  c )

static

Definition at line 875 of file apprentice.c.

                          {
    switch (c) {
    case '&': return FILE_OPAND;
    case '|': return FILE_OPOR;
    case '^': return FILE_OPXOR;
    case '+': return FILE_OPADD;
    case '-': return FILE_OPMINUS;
    case '*': return FILE_OPMULTIPLY;
    case '/': return FILE_OPDIVIDE;
    case '%': return FILE_OPMODULO;
    default: return -1;
    }
}

References c.

Referenced by parse().

get_type()

static int get_type ( const char *  l, const char **  t  )

static

Definition at line 160 of file apprentice.c.

                                                   {
    const struct type_tbl_s *p;
 
    for (p = type_tbl; p->len; p++) {
        if (strncmp(l, p->name, p->len) == 0) {
            if (t) {
                *t = l + p->len;
            }
            break;
        }
    }
    return p->type;
}

References p, and type_tbl.

Referenced by parse().

getstr()

static const char * getstr ( RzMagic *  ms, const char *  s, char *  p, int  plen, int *  slen, int  action  )

static

Definition at line 1637 of file apprentice.c.

                                                                                                {
    const char *origs = s;
    char *origp = p;
    char *pmax = p + plen - 1;
    int c, val;
 
    while ((c = *s++) != '\0') {
        if (isspace((ut8)c)) {
            break;
        }
        if (p >= pmax) {
            file_error(ms, 0, "string too long: `%s'", origs);
            return NULL;
        }
        if (c == '\\') {
            switch ((c = *s++)) {
            case '\0':
                if (action == FILE_COMPILE) {
                    file_magwarn(ms, "incomplete escape");
                }
                goto out;
            case '\t':
                if (action == FILE_COMPILE) {
                    file_magwarn(ms,
                        "escaped tab found, use \\t instead");
                    action++;
                }
                /*FALLTHROUGH*/
            default:
                if (action == FILE_COMPILE) {
                    if (isprint((ut8)c)) {
                        file_magwarn(ms,
                            "no need to escape `%c'", c);
                    } else {
                        file_magwarn(ms,
                            "unknown escape sequence: \\%03o", c);
                    }
                }
                /*FALLTHROUGH*/
            /* space, perhaps force people to use \040? */
            case ' ':
#if 0
            /*
             * Other things people escape, but shouldn't need to,
             * so we disallow them
             */
            case '\'':
            case '"':
            case '?':
#endif
            /* Relations */
            case '>':
            case '<':
            case '&':
            case '^':
            case '=':
            case '!':
            /* and baskslash itself */
            case '\\':
                *p++ = (char)c;
                break;
            case 'a': *p++ = '\a'; break;
            case 'b': *p++ = '\b'; break;
            case 'f': *p++ = '\f'; break;
            case 'n': *p++ = '\n'; break;
            case 'r': *p++ = '\r'; break;
            case 't': *p++ = '\t'; 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 2 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;
                    } else {
                        --s;
                    }
                } else {
                    --s;
                }
                *p++ = (char)val;
                break;
            }
        } else {
            *p++ = (char)c;
        }
    }
out:
    *p = '\0';
    *slen = p - origp;
    return s;
}

References test-lz4-speed::action, c, file_error(), file_magwarn(), hextoint(), isprint, isspace, NULL, out, p, s, and val.

Referenced by getvalue().

getvalue()

static int getvalue ( RzMagic *  ms, struct rz_magic *  m, const char **  p, int  action  )

static

Definition at line 1574 of file apprentice.c.

                                                                                 {
    int slen;
 
    switch (m->type) {
    case FILE_BESTRING16:
    case FILE_LESTRING16:
    case FILE_STRING:
    case FILE_PSTRING:
    case FILE_REGEX:
    case FILE_SEARCH:
        *p = getstr(ms, *p, m->value.s, sizeof(m->value.s), &slen, action);
        if (!*p) {
            if (ms->flags & RZ_MAGIC_CHECK) {
                file_magwarn(ms, "cannot get string from `%s'",
                    m->value.s);
            }
            return -1;
        }
        m->vallen = slen;
        if (m->type == FILE_PSTRING) {
            m->vallen++;
        }
        return 0;
    case FILE_FLOAT:
    case FILE_BEFLOAT:
    case FILE_LEFLOAT:
        if (m->reln != 'x') {
            char *ep;
#ifdef HAVE_STRTOF
            m->value.f = strtof(*p, &ep);
#else
            m->value.f = (float)strtod(*p, &ep);
#endif
            *p = ep;
        }
        return 0;
    case FILE_DOUBLE:
    case FILE_BEDOUBLE:
    case FILE_LEDOUBLE:
        if (m->reln != 'x') {
            char *ep;
            m->value.d = strtod(*p, &ep);
            *p = ep;
        }
        return 0;
    default:
        if (m->reln != 'x') {
            char *ep;
            m->value.q = file_signextend(ms, m,
                (ut64)strtoull(*p, &ep, 0));
            *p = ep;
            eatsize(p);
        }
        return 0;
    }
}

References test-lz4-speed::action, eatsize(), file_magwarn(), file_signextend(), test-lz4-speed::float, getstr(), regress::m, p, and ut64().

Referenced by parse().

hextoint()

static int hextoint ( int  c )

static

Definition at line 1759 of file apprentice.c.

                           {
    if (!isascii((ut8)c)) {
        return -1;
    }
    if (isdigit((ut8)c)) {
        return c - '0';
    }
    if ((c >= 'a') && (c <= 'f')) {
        return c + 10 - 'a';
    }
    if ((c >= 'A') && (c <= 'F')) {
        return c + 10 - 'A';
    }
    return -1;
}

References c, and isdigit.

Referenced by getstr().

init_file_tables()

static void init_file_tables ( void  )

static

Definition at line 174 of file apprentice.c.

                                   {
    static bool done = false;
    const struct type_tbl_s *p;
    if (done) {
        return;
    }
    done = true;
    for (p = type_tbl; p->len; p++) {
        if (p->type >= FILE_NAMES_SIZE) {
            continue;
        }
        magic_file_names[p->type] = p->name;
        magic_file_formats[p->type] = p->format;
    }
}

References done, magic_file_formats, magic_file_names, p, and type_tbl.

Referenced by file_apprentice().

load_1()

static void load_1 ( RzMagic *  ms, int  action, const char *  file, int *  errs, struct rz_magic_entry **  marray, ut32 *  marraycount  )

static

Definition at line 542 of file apprentice.c.

                                                                                                                            {
    if (*file == '#') {
        load_b(ms, action, file, errs, marray, marraycount);
        return;
    }
    char line[BUFSIZ];
    size_t lineno = 0;
    FILE *f = rz_sys_fopen(ms->file = file, "r");
    if (!f) {
        if (errno != ENOENT) {
            file_error(ms, errno, "cannot read magic file `%s'", file);
        }
        (*errs)++;
        return;
    }
    /* read and parse this file */
    for (ms->line = 1; fgets(line, sizeof(line), f) != NULL; ms->line++) {
        size_t len = strlen(line);
        if (len == 0) { /* null line, garbage, etc */
            continue;
        }
        if (line[len - 1] == '\n') {
            lineno++;
            line[len - 1] = '\0'; /* delete newline */
        }
        if (line[0] == '\0') { /* empty, do not parse */
            continue;
        }
        if (line[0] == '#') { /* comment, do not parse */
            continue;
        }
        if (len > mime_marker_len &&
            memcmp(line, mime_marker, mime_marker_len) == 0) {
            /* MIME type */
            if (parse_mime(ms, marray, marraycount,
                    line + mime_marker_len) != 0) {
                (*errs)++;
            }
            continue;
        }
        if (parse(ms, marray, marraycount, line, lineno, action) != 0) {
            (*errs)++;
        }
    }
    fclose(f);
}

References test-lz4-speed::action, ENOENT, f, benchmark::FILE, file_error(), len, setup::line, load_b(), mime_marker, mime_marker_len, NULL, parse(), parse_mime(), and rz_sys_fopen().

Referenced by apprentice_load().

load_b()

static void load_b ( RzMagic *  ms, int  action, const char *  data, int *  errs, struct rz_magic_entry **  marray, ut32 *  marraycount  )

static

Definition at line 506 of file apprentice.c.

                                                                                                                            {
    char line[BUFSIZ];
    size_t lineno = 0;
    /* read and parse this file */
    for (ms->line = 1; (data = bgets(line, sizeof(line), data)) != NULL; ms->line++) {
        size_t len = strlen(line);
        if (len == 0) { /* null line, garbage, etc */
            continue;
        }
        if (line[len - 1] == '\n') {
            lineno++;
            line[len - 1] = '\0'; /* delete newline */
        }
        if (line[0] == '\0') { /* empty, do not parse */
            continue;
        }
        if (line[0] == '#') { /* comment, do not parse */
            continue;
        }
        if (len > mime_marker_len && !memcmp(line, mime_marker, mime_marker_len)) {
            /* MIME type */
            if (parse_mime(ms, marray, marraycount,
                    line + mime_marker_len) != 0) {
                (*errs)++;
            }
            continue;
        }
        if (parse(ms, marray, marraycount, line, lineno, action) != 0) {
            (*errs)++;
        }
    }
}

References test-lz4-speed::action, bgets(), len, setup::line, mime_marker, mime_marker_len, NULL, parse(), and parse_mime().

Referenced by load_1().

mkdbname()

static char * mkdbname ( const char *  fn, int  strip  )

static

Definition at line 1985 of file apprentice.c.

                                                 {
    char *buf = NULL;
    int fnlen, extlen;
    if (strip) {
        const char *p;
        if ((p = strrchr(fn, '/')) != NULL) {
            fn = ++p;
        }
    }
    fnlen = strlen(fn);
    extlen = strlen(ext);
    if (fnlen + extlen + 1 > MAXPATHLEN) {
        return NULL;
    }
    buf = malloc(fnlen + extlen + 1);
    if (buf) {
        memcpy(buf, fn, fnlen);
        memcpy(buf + fnlen, ext, extlen);
        buf[fnlen + extlen] = 0;
    }
    return buf;
}

References ext, fn, malloc(), memcpy(), NULL, p, and cmd_descs_generate::strip().

Referenced by apprentice_compile(), and apprentice_map().

parse()

static int parse ( RzMagic *  ms, struct rz_magic_entry **  mentryp, ut32 *  nmentryp, const char *  line, size_t  lineno, int  action  )

static

Definition at line 958 of file apprentice.c.

                                                                                                                            {
    static ut32 last_cont_level = 0;
    size_t i;
    struct rz_magic_entry *me;
    struct rz_magic *m;
    const char *l = line;
    char *t;
    int op;
    ut32 cont_level = 0;
 
    for (; *l == '>'; l++, cont_level++) {
        ;
    }
    if (cont_level == 0 || cont_level > last_cont_level) {
        if (file_check_mem(ms, cont_level) == -1) {
            return -1;
        }
    }
    last_cont_level = cont_level;
#define ALLOC_CHUNK (size_t)10
#define ALLOC_INCR  (size_t)200
    if (cont_level != 0) {
        if (*nmentryp == 0) {
            file_error(ms, 0, "No current entry for continuation");
            return -1;
        }
        me = &(*mentryp)[*nmentryp - 1];
        if (me->cont_count == me->max_count) {
            struct rz_magic *nm;
            size_t cnt = me->max_count + ALLOC_CHUNK;
            if (!(nm = realloc(me->mp, sizeof(*nm) * cnt))) {
                file_oomem(ms, sizeof(*nm) * cnt);
                return -1;
            }
            me->mp = nm;
            me->max_count = cnt;
        }
        m = &me->mp[me->cont_count++];
        (void)memset(m, 0, sizeof(*m));
        m->cont_level = cont_level;
    } else {
        if (*nmentryp == maxmagic) {
            struct rz_magic_entry *mp;
 
            maxmagic += ALLOC_INCR;
            if (!(mp = realloc(*mentryp, sizeof(*mp) * maxmagic))) {
                file_oomem(ms, sizeof(*mp) * maxmagic);
                return -1;
            }
            (void)memset(&mp[*nmentryp], 0, sizeof(*mp) * ALLOC_INCR);
            *mentryp = mp;
        }
        me = &(*mentryp)[*nmentryp];
        if (!me->mp) {
            if (!(m = malloc(sizeof(*m) * ALLOC_CHUNK))) {
                file_oomem(ms, sizeof(*m) * ALLOC_CHUNK);
                return -1;
            }
            me->mp = m;
            me->max_count = ALLOC_CHUNK;
        } else {
            m = me->mp;
        }
        (void)memset(m, 0, sizeof(*m));
        m->cont_level = 0;
        me->cont_count = 1;
    }
    m->lineno = lineno;
 
    if (*l == '&') { /* m->cont_level == 0 checked below. */
        ++l; /* step over */
        m->flag |= OFFADD;
    }
    if (*l == '(') {
        ++l; /* step over */
        m->flag |= INDIR;
        if (m->flag & OFFADD) {
            m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
        }
 
        if (*l == '&') { /* m->cont_level == 0 checked below */
            ++l; /* step over */
            m->flag |= OFFADD;
        }
    }
    /* Indirect offsets are not valid at level 0. */
    if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) {
        if (ms->flags & RZ_MAGIC_CHECK) {
            file_magwarn(ms, "relative offset at level 0");
        }
    }
 
    /* get offset, then skip over it */
    m->offset = (ut32)strtoul(l, &t, 0);
    if ((l == t) && (ms->flags & RZ_MAGIC_CHECK)) {
        file_magwarn(ms, "offset `%s' invalid", l);
    }
    l = t;
 
    if (m->flag & INDIR) {
        m->in_type = FILE_LONG;
        m->in_offset = 0;
        /*
         * read [.lbs][+-]nnnnn)
         */
        if (*l == '.') {
            l++;
            switch (*l) {
            case 'l':
                m->in_type = FILE_LELONG;
                break;
            case 'L':
                m->in_type = FILE_BELONG;
                break;
            case 'm':
                m->in_type = FILE_MELONG;
                break;
            case 'h':
            case 's':
                m->in_type = FILE_LESHORT;
                break;
            case 'H':
            case 'S':
                m->in_type = FILE_BESHORT;
                break;
            case 'c':
            case 'b':
            case 'C':
            case 'B':
                m->in_type = FILE_BYTE;
                break;
            case 'e':
            case 'f':
            case 'g':
                m->in_type = FILE_LEDOUBLE;
                break;
            case 'E':
            case 'F':
            case 'G':
                m->in_type = FILE_BEDOUBLE;
                break;
            default:
                if (ms->flags & RZ_MAGIC_CHECK) {
                    file_magwarn(ms,
                        "indirect offset type `%c' invalid",
                        *l);
                }
                break;
            }
            l++;
        }
 
        m->in_op = 0;
        if (*l == '~') {
            m->in_op |= FILE_OPINVERSE;
            l++;
        }
        if ((op = get_op(*l)) != -1) {
            m->in_op |= op;
            l++;
        }
        if (*l == '(') {
            m->in_op |= FILE_OPINDIRECT;
            l++;
        }
        if (isdigit((ut8)*l) || *l == '-') {
            m->in_offset = (int32_t)strtol(l, &t, 0);
            if (l == t) {
                if (ms->flags & RZ_MAGIC_CHECK) {
                    file_magwarn(ms,
                        "in_offset `%s' invalid", l);
                }
            }
            l = t;
        }
        if (*l++ != ')' ||
            ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) {
            if (ms->flags & RZ_MAGIC_CHECK) {
                file_magwarn(ms,
                    "missing ')' in indirect offset");
            }
        }
    }
    EATAB;
 
    m->cond = get_cond(l, &l);
    if (check_cond(ms, m->cond, cont_level) == -1) {
        return -1;
    }
    EATAB;
 
    if (*l == 'u') {
        ++l;
        m->flag |= UNSIGNED;
    }
 
    m->type = get_type(l, &l);
    if (m->type == FILE_INVALID) {
        if (ms->flags & RZ_MAGIC_CHECK) {
            file_magwarn(ms, "type `%s' invalid", l);
        }
        return -1;
    }
 
    /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
    /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
 
    m->mask_op = 0;
    if (*l == '~') {
        if (!MAGIC_IS_STRING(m->type)) {
            m->mask_op |= FILE_OPINVERSE;
        } else if (ms->flags & RZ_MAGIC_CHECK) {
            file_magwarn(ms, "'~' invalid for string types");
        }
        ++l;
    }
    m->str_range = 0;
    m->str_flags = 0;
    m->num_mask = 0;
    if ((op = get_op(*l)) != -1) {
        if (!MAGIC_IS_STRING(m->type)) {
            ut64 val;
            ++l;
            m->mask_op |= op;
            val = (ut64)strtoull(l, &t, 0);
            l = t;
            m->num_mask = file_signextend(ms, m, val);
            eatsize(&l);
        } else if (op == FILE_OPDIVIDE) {
            int have_range = 0;
            while (!isspace((ut8) * ++l)) {
                switch (*l) {
                case '0':
                case '1':
                case '2':
                case '3':
                case '4':
                case '5':
                case '6':
                case '7':
                case '8':
                case '9':
                    if (have_range &&
                        (ms->flags & RZ_MAGIC_CHECK)) {
                        file_magwarn(ms,
                            "multiple ranges");
                    }
                    have_range = 1;
                    m->str_range = strtoul(l, &t, 0);
                    if (m->str_range == 0) {
                        file_magwarn(ms,
                            "zero range");
                    }
                    l = t - 1;
                    break;
                case CHAR_COMPACT_BLANK:
                    m->str_flags |= STRING_COMPACT_BLANK;
                    break;
                case CHAR_COMPACT_OPTIONAL_BLANK:
                    m->str_flags |=
                        STRING_COMPACT_OPTIONAL_BLANK;
                    break;
                case CHAR_IGNORE_LOWERCASE:
                    m->str_flags |= STRING_IGNORE_LOWERCASE;
                    break;
                case CHAR_IGNORE_UPPERCASE:
                    m->str_flags |= STRING_IGNORE_UPPERCASE;
                    break;
                case CHAR_REGEX_OFFSET_START:
                    m->str_flags |= REGEX_OFFSET_START;
                    break;
                default:
                    if (ms->flags & RZ_MAGIC_CHECK) {
                        file_magwarn(ms,
                            "string extension `%c' invalid",
                            *l);
                    }
                    return -1;
                }
                /* allow multiple '/' for readability */
                if (l[1] == '/' && !isspace((ut8)l[2])) {
                    l++;
                }
            }
            if (string_modifier_check(ms, m) == -1) {
                return -1;
            }
        } else {
            if (ms->flags & RZ_MAGIC_CHECK) {
                file_magwarn(ms, "invalid string op: %c", *t);
            }
            return -1;
        }
    }
    /*
     * We used to set mask to all 1's here, instead let's just not do
     * anything if mask = 0 (unless you have a better idea)
     */
    EATAB;
 
    switch (*l) {
    case '>':
    case '<':
    /* Old-style anding: "0 byte &0x80 dynamically linked" */
    case '&':
    case '^':
    case '=':
        m->reln = *l;
        ++l;
        if (*l == '=') {
            /* HP compat: ignore &= etc. */
            ++l;
        }
        break;
    case '!':
        m->reln = *l;
        ++l;
        break;
    default:
        m->reln = '='; /* the default relation */
        if (*l == 'x' && ((isascii((ut8)l[1]) && isspace((ut8)l[1])) || !l[1])) {
            m->reln = *l;
            ++l;
        }
        break;
    }
    /*
     * Grab the value part, except for an 'x' reln.
     */
    if (m->reln != 'x' && getvalue(ms, m, &l, action)) {
        return -1;
    }
 
    /*
     * TODO finish this macro and start using it!
     * #define offsetcheck {if (offset > HOWMANY-1)
     *  magwarn("offset too big"); }
     */
 
    /*
     * Now get last part - the description
     */
    EATAB;
    if (l[0] == '\b') {
        ++l;
        m->flag |= NOSPACE;
    } else if ((l[0] == '\\') && (l[1] == 'b')) {
        ++l;
        ++l;
        m->flag |= NOSPACE;
    }
    for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc);) {
    }
    if (i == sizeof(m->desc)) {
        m->desc[sizeof(m->desc) - 1] = '\0';
        if (ms->flags & RZ_MAGIC_CHECK) {
            file_magwarn(ms, "description `%s' truncated", m->desc);
        }
    }
 
    /*
     * We only do this check while compiling, or if any of the magic
     * files were not compiled.
     */
    if (ms->flags & RZ_MAGIC_CHECK) {
        if (check_format(ms, m) == -1) {
            return -1;
        }
    }
    if (action == FILE_CHECK) {
        file_mdump(m);
    }
    m->mimetype[0] = '\0'; /* initialise MIME type to none */
    if (m->cont_level == 0) {
        ++(*nmentryp); /* make room for next */
    }
    return 0;
}

References test-lz4-speed::action, ALLOC_CHUNK, ALLOC_INCR, check_cond(), check_format(), rz_magic_entry::cont_count, EATAB, eatsize(), file_check_mem(), file_error(), file_magwarn(), file_mdump(), file_oomem(), file_signextend(), get_cond(), get_op(), get_type(), getvalue(), i, isdigit, isspace, setup::line, regress::m, malloc(), rz_magic_entry::max_count, maxmagic, memset(), rz_magic_entry::mp, op, realloc(), string_modifier_check(), UNSIGNED, ut64(), and val.

Referenced by load_1(), and load_b().

parse_mime()

static int parse_mime ( RzMagic *  ms, struct rz_magic_entry **  mentryp, ut32 *  nmentryp, const char *  line  )

static

Definition at line 1341 of file apprentice.c.

                                                                                                      {
    size_t i;
    const char *l = line;
    struct rz_magic *m;
    struct rz_magic_entry *me;
 
    if (*nmentryp == 0) {
        file_error(ms, 0, "No current entry for MIME type");
        return -1;
    }
 
    me = &(*mentryp)[*nmentryp - 1];
    m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
 
    if (m->mimetype[0] != '\0') {
        file_error(ms, 0, "Current entry already has a MIME type: %s\n"
                  "Description: %s\nNew type: %s",
            m->mimetype, m->desc, l);
        return -1;
    }
 
    EATAB;
    for (i = 0;
        *l && ((isascii((ut8)*l) && isalnum((ut8)*l)) || strchr("-+/.", *l)) && i < sizeof(m->mimetype);
        m->mimetype[i++] = *l++) {
    }
    if (i == sizeof(m->mimetype)) {
        m->desc[sizeof(m->mimetype) - 1] = '\0';
        if (ms->flags & RZ_MAGIC_CHECK) {
            file_magwarn(ms, "MIME type `%s' truncated %zu",
                m->mimetype, i);
        }
    } else {
        m->mimetype[i] = '\0';
    }
 
    return (i > 0) ? 0 : -1;
}

References rz_magic_entry::cont_count, EATAB, file_error(), file_magwarn(), i, isalnum, setup::line, regress::m, and rz_magic_entry::mp.

Referenced by load_1(), and load_b().

set_test_type()

static void set_test_type ( struct rz_magic *  mstart, struct rz_magic *  m  )

static

Definition at line 430 of file apprentice.c.

                                                                       {
    switch (m->type) {
    case FILE_BYTE:
    case FILE_SHORT:
    case FILE_LONG:
    case FILE_DATE:
    case FILE_BESHORT:
    case FILE_BELONG:
    case FILE_BEDATE:
    case FILE_LESHORT:
    case FILE_LELONG:
    case FILE_LEDATE:
    case FILE_LDATE:
    case FILE_BELDATE:
    case FILE_LELDATE:
    case FILE_MEDATE:
    case FILE_MELDATE:
    case FILE_MELONG:
    case FILE_QUAD:
    case FILE_LEQUAD:
    case FILE_BEQUAD:
    case FILE_QDATE:
    case FILE_LEQDATE:
    case FILE_BEQDATE:
    case FILE_QLDATE:
    case FILE_LEQLDATE:
    case FILE_BEQLDATE:
    case FILE_FLOAT:
    case FILE_BEFLOAT:
    case FILE_LEFLOAT:
    case FILE_DOUBLE:
    case FILE_BEDOUBLE:
    case FILE_LEDOUBLE:
    case FILE_STRING:
    case FILE_PSTRING:
    case FILE_BESTRING16:
    case FILE_LESTRING16:
        /* binary test, set flag */
        mstart->flag |= BINTEST;
        break;
    case FILE_REGEX:
    case FILE_SEARCH:
        /* binary test if pattern is not text */
        if (file_looks_utf8((const ut8 *)m->value.s, m->vallen, NULL, NULL) == 0) {
            mstart->flag |= BINTEST;
        }
        break;
    case FILE_DEFAULT:
        /* can't deduce anything; we shouldn't see this at the
           top level anyway */
        break;
    case FILE_INVALID:
    default:
        /* invalid search type, but no need to complain here */
        break;
    }
}

References file_looks_utf8(), regress::m, and NULL.

Referenced by apprentice_load().

string_modifier_check()

static int string_modifier_check ( RzMagic *  ms, struct rz_magic *  m  )

static

Definition at line 823 of file apprentice.c.

                                                                  {
    if ((ms->flags & RZ_MAGIC_CHECK) == 0) {
        return 0;
    }
 
    switch (m->type) {
    case FILE_BESTRING16:
    case FILE_LESTRING16:
        if (m->str_flags != 0) {
            file_magwarn(ms,
                "no modifiers allowed for 16-bit strings\n");
            return -1;
        }
        break;
    case FILE_STRING:
    case FILE_PSTRING:
        if ((m->str_flags & REGEX_OFFSET_START) != 0) {
            file_magwarn(ms,
                "'/%c' only allowed on regex and search\n",
                CHAR_REGEX_OFFSET_START);
            return -1;
        }
        break;
    case FILE_SEARCH:
        if (m->str_range == 0) {
            file_magwarn(ms,
                "missing range; defaulting to %d\n",
                STRING_DEFAULT_RANGE);
            m->str_range = STRING_DEFAULT_RANGE;
            return -1;
        }
        break;
    case FILE_REGEX:
        if ((m->str_flags & STRING_COMPACT_BLANK) != 0) {
            file_magwarn(ms, "'/%c' not allowed on regex\n",
                CHAR_COMPACT_BLANK);
            return -1;
        }
        if ((m->str_flags & STRING_COMPACT_OPTIONAL_BLANK) != 0) {
            file_magwarn(ms, "'/%c' not allowed on regex\n",
                CHAR_COMPACT_OPTIONAL_BLANK);
            return -1;
        }
        break;
    default:
        file_magwarn(ms, "coding error: m->type=%d\n",
            m->type);
        return -1;
    }
    return 0;
}

References file_magwarn(), and regress::m.

Referenced by parse().

swap2()

static ut16 swap2 ( ut16  sv )

static

Definition at line 2021 of file apprentice.c.

                           {
    ut16 rv;
    ut8 *s = (ut8 *)(void *)&sv;
    ut8 *d = (ut8 *)(void *)&rv;
    d[0] = s[1];
    d[1] = s[0];
    return rv;
}

References d, and s.

Referenced by bs1().

swap4()

static ut32 swap4 ( ut32  sv )

static

Definition at line 2033 of file apprentice.c.

                           {
    ut32 rv;
    ut8 *s = (ut8 *)(void *)&sv;
    ut8 *d = (ut8 *)(void *)&rv;
    d[0] = s[3];
    d[1] = s[2];
    d[2] = s[1];
    d[3] = s[0];
    return rv;
}

References d, and s.

Referenced by apprentice_map(), and bs1().

swap8()

static ut64 swap8 ( ut64  sv )

static

Definition at line 2047 of file apprentice.c.

                           {
    ut64 rv;
    ut8 *s = (ut8 *)(void *)&sv;
    ut8 *d = (ut8 *)(void *)&rv;
#if 0
    d[0] = s[3];
    d[1] = s[2];
    d[2] = s[1];
    d[3] = s[0];
    d[4] = s[7];
    d[5] = s[6];
    d[6] = s[5];
    d[7] = s[4];
#else
    d[0] = s[7];
    d[1] = s[6];
    d[2] = s[5];
    d[3] = s[4];
    d[4] = s[3];
    d[5] = s[2];
    d[6] = s[1];
    d[7] = s[0];
#endif
    return rv;
}

References d, s, and ut64().

Referenced by bs1().

Variable Documentation

ar

const ut32 ar[]

static

Initial value:

= {
    MAGICNO, VERSIONNO
}

Definition at line 1935 of file apprentice.c.

Referenced by apprentice_compile(), jemalloc_get_bins(), jemalloc_get_chunks(), jemalloc_print_narenas(), reloc_cmp(), and reloc_target_cmp().

ext

const char ext[] = ".mgc"

static

Definition at line 1981 of file apprentice.c.

Referenced by __isScriptFilename(), aarch64_wide_constant_p(), arc_sprintf(), build_3bit_d(), build_3bit_ea(), build_absolute_jump_with_immediate(), build_bitfield_ins(), build_cas2(), build_chk2_cmp2(), build_cpush_cinv(), build_d(), build_d_d_ea(), build_dbxx(), build_ea(), build_ea_a(), build_ea_ea(), build_er_gen_1(), build_imm(), build_imm_ea(), build_imm_special_reg(), build_init_op(), build_link(), build_mm(), build_move16(), build_movem_er(), build_movem_re(), build_movep_er(), build_movep_re(), build_moves(), build_pi_pi(), build_r(), build_re_gen_1(), build_relative_branch(), build_rr(), convert_dot_to_image(), d68000_exg_aa(), d68000_exg_da(), d68000_jmp(), d68000_jsr(), d68000_move_fr_sr(), d68000_move_fr_usp(), d68000_move_to_ccr(), d68000_move_to_sr(), d68000_move_to_usp(), d68000_moveq(), d68000_scc(), d68000_unlk(), d68010_move_fr_ccr(), d68010_movec(), d68020_bfins(), d68020_cpbcc_16(), d68020_cpbcc_32(), d68020_cpdbcc(), d68020_cpgen(), d68020_cprestore(), d68020_cpsave(), d68020_cpscc(), d68020_cptrapcc_16(), d68020_cptrapcc_32(), d68020_divl(), d68020_mull(), d68020_rtm(), extend(), fmove_fpcr(), fmovem(), load_buffer(), log_mark(), m68k_disassemble(), M68K_printInst(), mkdbname(), my_sprintf(), rz_core_run_script(), rz_file_is_c(), rz_lang_get_by_extension(), rz_x509_parse_extension(), rz_x509_parse_extensions(), and search_path_join_test().

file_nformats

const size_t file_nformats = FILE_NAMES_SIZE

static

Definition at line 76 of file apprentice.c.

Referenced by check_format().

file_nnames

const size_t file_nnames = FILE_NAMES_SIZE

static

Definition at line 78 of file apprentice.c.

Referenced by check_format(), and file_mdump().

magic_file_formats

int magic_file_formats[FILE_NAMES_SIZE]

static

Definition at line 75 of file apprentice.c.

Referenced by check_format(), and init_file_tables().

magic_file_names

const char* magic_file_names[FILE_NAMES_SIZE]

static

Definition at line 77 of file apprentice.c.

Referenced by check_format(), file_mdump(), and init_file_tables().

magicsize

size_t magicsize = sizeof(struct rz_magic)

static

Definition at line 103 of file apprentice.c.

Referenced by apprentice_1().

maxmagic

size_t maxmagic = 0

static

Definition at line 102 of file apprentice.c.

Referenced by apprentice_load(), and parse().

mime_marker

const char mime_marker[] = "!:mime"

static

Definition at line 106 of file apprentice.c.

Referenced by load_1(), and load_b().

mime_marker_len

const size_t mime_marker_len = sizeof(mime_marker) - 1

static

Definition at line 107 of file apprentice.c.

Referenced by load_1(), and load_b().

type_tbl

const struct type_tbl_s type_tbl[]

static

Referenced by get_type(), and init_file_tables().

usg_hdr

const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"

static

Definition at line 105 of file apprentice.c.

Referenced by apprentice_load().