#include "file.h"
#ifndef lint
FILE_RCSID("@(#)$File: ascmagic.c,v 1.92 2015/04/09 20:01:41 christos Exp $")
#endif
#include "magic.h"
#include <string.h>
#include <memory.h>
#include <ctype.h>
#include <stdlib.h>
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
#define MAXLINELEN 300
#define ISSPC(x) ((x) == ' ' || (x) == '\t' || (x) == '\r' || (x) == '\n' \
|| (x) == 0x85 || (x) == '\f')
private unsigned char *encode_utf8(unsigned char *, size_t, unichar *, size_t);
private size_t trim_nuls(const unsigned char *, size_t);
private size_t
trim_nuls(const unsigned char *buf, size_t nbytes)
{
while (nbytes > 1 && buf[nbytes - 1] == '\0')
nbytes--;
return nbytes;
}
protected int
file_ascmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes,
int text)
{
unichar *ubuf = NULL;
size_t ulen = 0;
int rv = 1;
const char *code = NULL;
const char *code_mime = NULL;
const char *type = NULL;
if (ms->flags & (MAGIC_APPLE|MAGIC_EXTENSION))
return 0;
nbytes = trim_nuls(buf, nbytes);
if (file_encoding(ms, buf, nbytes, &ubuf, &ulen, &code, &code_mime,
&type) == 0) {
rv = 0;
} else {
rv = file_ascmagic_with_encoding(ms, buf, nbytes, ubuf, ulen, code,
type, text);
}
free(ubuf);
return rv;
}
protected int
file_ascmagic_with_encoding(struct magic_set *ms, const unsigned char *buf,
size_t nbytes, unichar *ubuf, size_t ulen, const char *code,
const char *type, int text)
{
unsigned char *utf8_buf = NULL, *utf8_end;
size_t mlen, i;
int rv = -1;
int mime = ms->flags & MAGIC_MIME;
const char *subtype = NULL;
const char *subtype_mime = NULL;
int has_escapes = 0;
int has_backspace = 0;
int seen_cr = 0;
int n_crlf = 0;
int n_lf = 0;
int n_cr = 0;
int n_nel = 0;
int executable = 0;
size_t last_line_end = (size_t)-1;
int has_long_lines = 0;
if (ms->flags & (MAGIC_APPLE|MAGIC_EXTENSION))
return 0;
nbytes = trim_nuls(buf, nbytes);
if (nbytes <= 1) {
rv = 0;
goto done;
}
if (ulen > 0 && (ms->flags & MAGIC_NO_CHECK_SOFT) == 0) {
mlen = ulen * 6;
if ((utf8_buf = CAST(unsigned char *, malloc(mlen))) == NULL) {
file_oomem(ms, mlen);
goto done;
}
if ((utf8_end = encode_utf8(utf8_buf, mlen, ubuf, ulen))
== NULL)
goto done;
if ((rv = file_softmagic(ms, utf8_buf,
(size_t)(utf8_end - utf8_buf), 0, NULL,
TEXTTEST, text)) == 0)
rv = -1;
}
for (i = 0; i < ulen; i++) {
if (ubuf[i] == '\n') {
if (seen_cr)
n_crlf++;
else
n_lf++;
last_line_end = i;
} else if (seen_cr)
n_cr++;
seen_cr = (ubuf[i] == '\r');
if (seen_cr)
last_line_end = i;
if (ubuf[i] == 0x85) {
n_nel++;
last_line_end = i;
}
if (i > last_line_end + MAXLINELEN)
has_long_lines = 1;
if (ubuf[i] == '\033')
has_escapes = 1;
if (ubuf[i] == '\b')
has_backspace = 1;
}
if (seen_cr && nbytes < HOWMANY)
n_cr++;
if (strcmp(type, "binary") == 0) {
rv = 0;
goto done;
}
if (mime) {
if (!file_printedlen(ms) && (mime & MAGIC_MIME_TYPE) != 0) {
if (subtype_mime) {
if (file_printf(ms, "%s", subtype_mime) == -1)
goto done;
} else {
if (file_printf(ms, "text/plain") == -1)
goto done;
}
}
} else {
if (file_printedlen(ms)) {
#ifdef __APPLE__
if (file_printf(ms, ", ") == -1)
goto done;
#else
switch (file_replace(ms, " text$", ", ")) {
case 0:
switch (file_replace(ms, " text executable$",
", ")) {
case 0:
if (file_printf(ms, ", ") == -1)
goto done;
break;
case -1:
goto done;
default:
executable = 1;
break;
}
break;
case -1:
goto done;
default:
break;
}
#endif
}
if (file_printf(ms, "%s", code) == -1)
goto done;
if (subtype) {
if (file_printf(ms, " %s", subtype) == -1)
goto done;
}
if (file_printf(ms, " %s", type) == -1)
goto done;
if (executable)
if (file_printf(ms, " executable") == -1)
goto done;
if (has_long_lines)
if (file_printf(ms, ", with very long lines") == -1)
goto done;
if ((n_crlf == 0 && n_cr == 0 && n_nel == 0 && n_lf == 0) ||
(n_crlf != 0 || n_cr != 0 || n_nel != 0)) {
if (file_printf(ms, ", with") == -1)
goto done;
if (n_crlf == 0 && n_cr == 0 && n_nel == 0 && n_lf == 0) {
if (file_printf(ms, " no") == -1)
goto done;
} else {
if (n_crlf) {
if (file_printf(ms, " CRLF") == -1)
goto done;
if (n_cr || n_lf || n_nel)
if (file_printf(ms, ",") == -1)
goto done;
}
if (n_cr) {
if (file_printf(ms, " CR") == -1)
goto done;
if (n_lf || n_nel)
if (file_printf(ms, ",") == -1)
goto done;
}
if (n_lf) {
if (file_printf(ms, " LF") == -1)
goto done;
if (n_nel)
if (file_printf(ms, ",") == -1)
goto done;
}
if (n_nel)
if (file_printf(ms, " NEL") == -1)
goto done;
}
if (file_printf(ms, " line terminators") == -1)
goto done;
}
if (has_escapes)
if (file_printf(ms, ", with escape sequences") == -1)
goto done;
if (has_backspace)
if (file_printf(ms, ", with overstriking") == -1)
goto done;
}
rv = 1;
done:
free(utf8_buf);
return rv;
}
private unsigned char *
encode_utf8(unsigned char *buf, size_t len, unichar *ubuf, size_t ulen)
{
size_t i;
unsigned char *end = buf + len;
for (i = 0; i < ulen; i++) {
if (ubuf[i] <= 0x7f) {
if (end - buf < 1)
return NULL;
*buf++ = (unsigned char)ubuf[i];
} else if (ubuf[i] <= 0x7ff) {
if (end - buf < 2)
return NULL;
*buf++ = (unsigned char)((ubuf[i] >> 6) + 0xc0);
*buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
} else if (ubuf[i] <= 0xffff) {
if (end - buf < 3)
return NULL;
*buf++ = (unsigned char)((ubuf[i] >> 12) + 0xe0);
*buf++ = (unsigned char)(((ubuf[i] >> 6) & 0x3f) + 0x80);
*buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
} else if (ubuf[i] <= 0x1fffff) {
if (end - buf < 4)
return NULL;
*buf++ = (unsigned char)((ubuf[i] >> 18) + 0xf0);
*buf++ = (unsigned char)(((ubuf[i] >> 12) & 0x3f) + 0x80);
*buf++ = (unsigned char)(((ubuf[i] >> 6) & 0x3f) + 0x80);
*buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
} else if (ubuf[i] <= 0x3ffffff) {
if (end - buf < 5)
return NULL;
*buf++ = (unsigned char)((ubuf[i] >> 24) + 0xf8);
*buf++ = (unsigned char)(((ubuf[i] >> 18) & 0x3f) + 0x80);
*buf++ = (unsigned char)(((ubuf[i] >> 12) & 0x3f) + 0x80);
*buf++ = (unsigned char)(((ubuf[i] >> 6) & 0x3f) + 0x80);
*buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
} else if (ubuf[i] <= 0x7fffffff) {
if (end - buf < 6)
return NULL;
*buf++ = (unsigned char)((ubuf[i] >> 30) + 0xfc);
*buf++ = (unsigned char)(((ubuf[i] >> 24) & 0x3f) + 0x80);
*buf++ = (unsigned char)(((ubuf[i] >> 18) & 0x3f) + 0x80);
*buf++ = (unsigned char)(((ubuf[i] >> 12) & 0x3f) + 0x80);
*buf++ = (unsigned char)(((ubuf[i] >> 6) & 0x3f) + 0x80);
*buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
} else
return NULL;
}
return buf;
}