diff options
Diffstat (limited to 'ascmagic.c')
| -rw-r--r-- | ascmagic.c | 84 |
1 files changed, 62 insertions, 22 deletions
diff --git a/ascmagic.c b/ascmagic.c index 9236fb4a27a3..279a51a8cfd7 100644 --- a/ascmagic.c +++ b/ascmagic.c @@ -26,8 +26,7 @@ * SUCH DAMAGE. */ /* - * ASCII magic -- file types that we know based on keywords - * that can appear anywhere in the file. + * ASCII magic -- try to detect text encoding. * * Extensively modified by Eric Fischer <enf@pobox.com> in July, 2000, * to handle character codes other than ASCII on a unified basis. @@ -36,7 +35,7 @@ #include "file.h" #ifndef lint -FILE_RCSID("@(#)$File: ascmagic.c,v 1.75 2009/02/03 20:27:51 christos Exp $") +FILE_RCSID("@(#)$File: ascmagic.c,v 1.81 2011/03/15 22:16:29 christos Exp $") #endif /* lint */ #include "magic.h" @@ -93,7 +92,7 @@ file_ascmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes) goto done; } - rv = file_ascmagic_with_encoding(ms, buf, nbytes, ubuf, ulen, code, + rv = file_ascmagic_with_encoding(ms, buf, nbytes, ubuf, ulen, code, type); done: @@ -125,6 +124,7 @@ file_ascmagic_with_encoding(struct magic_set *ms, const unsigned char *buf, int n_lf = 0; int n_cr = 0; int n_nel = 0; + int score, curtype, executable = 0; size_t last_line_end = (size_t)-1; int has_long_lines = 0; @@ -140,27 +140,32 @@ file_ascmagic_with_encoding(struct magic_set *ms, const unsigned char *buf, goto done; } - /* Convert ubuf to UTF-8 and try text soft magic */ - /* malloc size is a conservative overestimate; could be - improved, or at least realloced after conversion. */ - mlen = ulen * 6; - if ((utf8_buf = CAST(unsigned char *, malloc(mlen))) == NULL) { - file_oomem(ms, mlen); - goto done; + if ((ms->flags & MAGIC_NO_CHECK_SOFT) == 0) { + /* Convert ubuf to UTF-8 and try text soft magic */ + /* malloc size is a conservative overestimate; could be + improved, or at least realloced after conversion. */ + 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), TEXTTEST)) != 0) + goto subtype_identified; + else + rv = -1; } - 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), - TEXTTEST)) != 0) - goto done; - else - rv = -1; /* look for tokens from names.h - this is expensive! */ if ((ms->flags & MAGIC_NO_CHECK_TOKENS) != 0) goto subtype_identified; i = 0; + score = 0; + curtype = -1; while (i < ulen) { size_t end; @@ -179,9 +184,18 @@ file_ascmagic_with_encoding(struct magic_set *ms, const unsigned char *buf, for (p = names; p < names + NNAMES; p++) { if (ascmatch((const unsigned char *)p->name, ubuf + i, end - i)) { - subtype = types[p->type].human; - subtype_mime = types[p->type].mime; - goto subtype_identified; + if (curtype == -1) + curtype = p->type; + else if (curtype != p->type) { + score = p->score; + curtype = p->type; + } else + score += p->score; + if (score > 1) { + subtype = types[p->type].human; + subtype_mime = types[p->type].mime; + goto subtype_identified; + } } } @@ -232,7 +246,7 @@ subtype_identified: goto done; } if (mime) { - if ((mime & MAGIC_MIME_TYPE) != 0) { + if (!file_printedlen(ms) && (mime & MAGIC_MIME_TYPE) != 0) { if (subtype_mime) { if (file_printf(ms, "%s", subtype_mime) == -1) goto done; @@ -242,6 +256,28 @@ subtype_identified: } } } else { + if (file_printedlen(ms)) { + switch (file_replace(ms, " text$", ", ")) { + case 0: + switch (file_replace(ms, " text executable$", + ", ")) { + case 0: + if (file_printf(ms, ", ") == -1) + goto done; + case -1: + goto done; + default: + executable = 1; + break; + } + break; + case -1: + goto done; + default: + break; + } + } + if (file_printf(ms, "%s", code) == -1) goto done; @@ -253,6 +289,10 @@ subtype_identified: 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; |