From: Christos Zoulas <christos@zoulas.com>
Date: Fri, 7 Nov 2008 18:57:28 +0000 (+0000)
Subject: handle encoding better.
X-Git-Tag: FILE5_00~30
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=fe3c433f57adb5f684eda1ec62f0e387aa03a503;p=file

handle encoding better.
---

diff --git a/src/ascmagic.c b/src/ascmagic.c
index 1a36829e..b58f1e53 100644
--- a/src/ascmagic.c
+++ b/src/ascmagic.c
@@ -36,7 +36,7 @@
 #include "file.h"
 
 #ifndef	lint
-FILE_RCSID("@(#)$File: ascmagic.c,v 1.71 2008/11/06 21:36:44 rrt Exp $")
+FILE_RCSID("@(#)$File: ascmagic.c,v 1.72 2008/11/07 17:26:17 christos Exp $")
 #endif	/* lint */
 
 #include "magic.h"
@@ -88,12 +88,14 @@ file_ascmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes)
 	nbytes = trim_nuls(buf, nbytes);
 
 	/* If file doesn't look like any sort of text, give up. */
-	if (file_encoding(ms, buf, nbytes, &ubuf, &ulen, &code, &code_mime, &type) == 0) {
+	if (file_encoding(ms, buf, nbytes, &ubuf, &ulen, &code, &code_mime,
+	    &type) == 0) {
 		rv = 0;
 		goto done;
 	}
 
-	rv = file_ascmagic_with_encoding(ms, buf, nbytes, ubuf, ulen, code, code_mime, type);
+	rv = file_ascmagic_with_encoding(ms, buf, nbytes, ubuf, ulen, code, 
+	    type);
 
  done:
 	if (ubuf)
@@ -103,7 +105,9 @@ file_ascmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes)
 }
 
 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 *code_mime, const char *type)
+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)
 {
 	unsigned char *utf8_buf = NULL, *utf8_end;
 	size_t mlen, i;
@@ -225,8 +229,7 @@ subtype_identified:
 		n_cr++;
 
 	if (mime) {
-		if ((mime & MAGIC_MIME_TYPE) &&
-		    !(ms->event_flags & EVENT_WROTE_MIME_TYPE)) {
+		if ((mime & MAGIC_MIME_TYPE) != 0) {
 			if (subtype_mime) {
 				if (file_printf(ms, "%s", subtype_mime) == -1)
 					goto done;
@@ -235,18 +238,6 @@ subtype_identified:
 					goto done;
 			}
 		}
-
-		if ((mime == 0 || mime == MAGIC_MIME) && code_mime) {
-			if ((mime & MAGIC_MIME_TYPE) &&
-			    file_printf(ms, "; charset=") == -1)
-				goto done;
-			if (file_printf(ms, "%s", code_mime) == -1)
-				goto done;
-		}
-
-		if (mime == MAGIC_MIME_ENCODING)
-			if (file_printf(ms, "%s", code_mime) == -1)
-				goto done;
 	} else {
 		if (file_printf(ms, "%s", code) == -1)
 			goto done;
diff --git a/src/file.h b/src/file.h
index 911c453c..897ed569 100644
--- a/src/file.h
+++ b/src/file.h
@@ -27,7 +27,7 @@
  */
 /*
  * file.h - definitions for file(1) program
- * @(#)$File: file.h,v 1.114 2008/11/06 21:17:45 rrt Exp $
+ * @(#)$File: file.h,v 1.115 2008/11/06 23:22:54 christos Exp $
  */
 
 #ifndef __file_h__
@@ -323,7 +323,6 @@ struct magic_set {
 	int flags;			/* Control magic tests. */
 	int event_flags;		/* Note things that happened. */
 #define 		EVENT_HAD_ERR		0x01
-#define 		EVENT_WROTE_MIME_TYPE	0x02
 	const char *file;
 	size_t line;			/* current magic line number */
 
@@ -360,10 +359,14 @@ protected int file_trycdf(struct magic_set *, int, const unsigned char *,
 protected int file_zmagic(struct magic_set *, int, const char *,
     const unsigned char *, size_t);
 protected int file_ascmagic(struct magic_set *, const unsigned char *, size_t);
-protected int file_ascmagic_with_encoding(struct magic_set *, const unsigned char *, size_t, unichar *, size_t, const char *, const char *, const char *);
-protected int file_encoding(struct magic_set *, const unsigned char *, size_t, unichar **, size_t *, const char **, const char **, const char **);
+protected int file_ascmagic_with_encoding(struct magic_set *,
+    const unsigned char *, size_t, unichar *, size_t, const char *,
+    const char *);
+protected int file_encoding(struct magic_set *, const unsigned char *, size_t,
+    unichar **, size_t *, const char **, const char **, const char **);
 protected int file_is_tar(struct magic_set *, const unsigned char *, size_t);
-protected int file_softmagic(struct magic_set *, const unsigned char *, size_t, int);
+protected int file_softmagic(struct magic_set *, const unsigned char *, size_t,
+    int);
 protected struct mlist *file_apprentice(struct magic_set *, const char *, int);
 protected uint64_t file_signextend(struct magic_set *, struct magic *,
     uint64_t);
diff --git a/src/fsmagic.c b/src/fsmagic.c
index e52496c3..30dbf5bd 100644
--- a/src/fsmagic.c
+++ b/src/fsmagic.c
@@ -32,7 +32,7 @@
 #include "file.h"
 
 #ifndef	lint
-FILE_RCSID("@(#)$File: fsmagic.c,v 1.55 2008/11/04 16:38:28 christos Exp $")
+FILE_RCSID("@(#)$File: fsmagic.c,v 1.56 2008/11/07 17:25:59 christos Exp $")
 #endif	/* lint */
 
 #include "magic.h"
@@ -92,7 +92,7 @@ handle_mime(struct magic_set *ms, int mime, const char *str)
 		if (file_printf(ms, "application/%s", str) == -1)
 			return -1;
 		if ((mime & MAGIC_MIME_ENCODING) && file_printf(ms,
-		    "; encoding=") == -1)
+		    "; charset=") == -1)
 			return -1;
 	}
 	if ((mime & MAGIC_MIME_ENCODING) && file_printf(ms, "binary") == -1)
diff --git a/src/funcs.c b/src/funcs.c
index 5a8706bb..b9a06cad 100644
--- a/src/funcs.c
+++ b/src/funcs.c
@@ -27,7 +27,7 @@
 #include "file.h"
 
 #ifndef	lint
-FILE_RCSID("@(#)$File: funcs.c,v 1.49 2008/11/06 22:49:08 rrt Exp $")
+FILE_RCSID("@(#)$File: funcs.c,v 1.50 2008/11/07 17:27:22 christos Exp $")
 #endif	/* lint */
 
 #include "magic.h"
@@ -158,11 +158,15 @@ protected int
 file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
     size_t nb)
 {
-	int m = 0, rv = 0;
+	int m = 0, rv = 0, looks_text = 0;
 	int mime = ms->flags & MAGIC_MIME;
 	const unsigned char *ubuf = CAST(const unsigned char *, buf);
 	unichar *u8buf = NULL;
 	size_t ulen;
+	const char *code = NULL;
+	const char *code_mime = "binary";
+	const char *type = NULL;
+
 
 
 	if (nb == 0) {
@@ -179,6 +183,11 @@ file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
 		return 1;
 	}
 
+	if ((ms->flags & MAGIC_NO_CHECK_ENCODING) == 0) {
+		looks_text = file_encoding(ms, ubuf, nb, &u8buf, &ulen,
+		    &code, &code_mime, &type);
+	}
+
 #ifdef __EMX__
 	if ((ms->flags & MAGIC_NO_CHECK_APPTYPE) == 0 && inname) {
 		switch (file_os2_apptype(ms, inname, buf, nb)) {
@@ -253,17 +262,9 @@ file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
 
 		/* try to discover text encoding */
 		if ((ms->flags & MAGIC_NO_CHECK_ENCODING) == 0) {
-			const char *code = NULL;
-			const char *code_mime = NULL;
-			const char *type = NULL;
-			int looks_text = 0;
-
-			looks_text = file_encoding(ms, ubuf, nb, &u8buf, &ulen,
-			    &code, &code_mime, &type);
 			if (looks_text == 0)
 				if ((m = file_ascmagic_with_encoding( ms, ubuf,
-				    nb, u8buf, ulen, code, code_mime, type))
-				    != 0) {
+				    nb, u8buf, ulen, code, type)) != 0) {
 					if ((ms->flags & MAGIC_DEBUG) != 0)
 						(void)fprintf(stderr,
 						    "ascmagic/enc %d\n", m);
@@ -279,6 +280,13 @@ file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
 	    rv = -1;
 	}
  done:
+	if ((ms->flags & MAGIC_MIME_ENCODING) != 0) {
+		if (ms->flags & MAGIC_MIME_TYPE)
+			if (file_printf(ms, "; charset=") == -1)
+				rv = -1;
+		if (file_printf(ms, "%s", code_mime) == -1)
+			rv = -1;
+	}
 	if (u8buf)
 		free(u8buf);
 	if (rv)
diff --git a/src/is_tar.c b/src/is_tar.c
index 907a265a..0da0308f 100644
--- a/src/is_tar.c
+++ b/src/is_tar.c
@@ -40,7 +40,7 @@
 #include "file.h"
 
 #ifndef lint
-FILE_RCSID("@(#)$File: is_tar.c,v 1.33 2008/11/03 06:18:34 rrt Exp $")
+FILE_RCSID("@(#)$File: is_tar.c,v 1.34 2008/11/04 16:38:28 christos Exp $")
 #endif
 
 #include "magic.h"
@@ -70,7 +70,7 @@ file_is_tar(struct magic_set *ms, const unsigned char *buf, size_t nbytes)
 	int tar;
 	int mime = ms->flags & MAGIC_MIME;
 
-	if ((ms->flags & MAGIC_APPLE) != 0 || mime == MAGIC_MIME_ENCODING)
+	if ((ms->flags & MAGIC_APPLE) != 0)
 		return 0;
 
 	tar = is_tar(buf, nbytes);
diff --git a/src/softmagic.c b/src/softmagic.c
index a32b7afa..ee8661ec 100644
--- a/src/softmagic.c
+++ b/src/softmagic.c
@@ -32,7 +32,7 @@
 #include "file.h"
 
 #ifndef	lint
-FILE_RCSID("@(#)$File: softmagic.c,v 1.130 2008/11/06 23:22:54 christos Exp $")
+FILE_RCSID("@(#)$File: softmagic.c,v 1.131 2008/11/07 17:26:44 christos Exp $")
 #endif	/* lint */
 
 #include "magic.h"
@@ -166,7 +166,7 @@ match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
 		if (*m->desc) {
 			need_separator = 1;
 			printed_something = 1;
-			if ((e = handle_annotation(ms, m)) != -2)
+			if ((e = handle_annotation(ms, m)) != 0)
 				return e;
 			if (print_sep(ms, firstline) == -1)
 				return -1;
@@ -234,7 +234,7 @@ match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
 				 */
 				if (*m->desc) {
 					printed_something = 1;
-					if ((e = handle_annotation(ms, m)) != -2)
+					if ((e = handle_annotation(ms, m)) != 0)
 						return e;
 					if (print_sep(ms, firstline) == -1)
 						return -1;
@@ -249,7 +249,9 @@ match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
 				if (need_separator
 				    && ((m->flag & NOSPACE) == 0)
 				    && *m->desc) {
-					if (file_printf(ms, " ") == -1)
+					if ((ms->flags &
+					    (MAGIC_MIME|MAGIC_APPLE)) == 0 &&
+					    file_printf(ms, " ") == -1)
 						return -1;
 					need_separator = 0;
 				}
@@ -273,9 +275,13 @@ match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
 			returnval = 1;
 		}
 		if ((ms->flags & MAGIC_CONTINUE) == 0 && printed_something) {
+			if (ms->flags & MAGIC_MIME)
+				return 0;
 			return 1; /* don't keep searching */
 		}
 	}
+	if (ms->flags & MAGIC_MIME)
+		return 0;
 	return returnval;  /* This is hit if -k is set or there is no match */
 }
 
@@ -330,6 +336,7 @@ mprint(struct magic_set *ms, struct magic *m)
 	int64_t t = 0;
  	char buf[128];
 	union VALUETYPE *p = &ms->ms_value;
+	int np = (ms->flags & MAGIC_MIME) == 0;
 
   	switch (m->type) {
   	case FILE_BYTE:
@@ -340,11 +347,12 @@ mprint(struct magic_set *ms, struct magic *m)
 		case 1:
 			(void)snprintf(buf, sizeof(buf), "%c",
 			    (unsigned char)v);
-			if (file_printf(ms, m->desc, buf) == -1)
+			if (np && file_printf(ms, m->desc, buf) == -1)
 				return -1;
 			break;
 		default:
-			if (file_printf(ms, m->desc, (unsigned char) v) == -1)
+			if (np &&
+			    file_printf(ms, m->desc, (unsigned char) v) == -1)
 				return -1;
 			break;
 		}
@@ -361,11 +369,12 @@ mprint(struct magic_set *ms, struct magic *m)
 		case 1:
 			(void)snprintf(buf, sizeof(buf), "%hu",
 			    (unsigned short)v);
-			if (file_printf(ms, m->desc, buf) == -1)
+			if (np && file_printf(ms, m->desc, buf) == -1)
 				return -1;
 			break;
 		default:
-			if (file_printf(ms, m->desc, (unsigned short) v) == -1)
+			if (np && 
+			    file_printf(ms, m->desc, (unsigned short) v) == -1)
 				return -1;
 			break;
 		}
@@ -382,11 +391,11 @@ mprint(struct magic_set *ms, struct magic *m)
 			return -1;
 		case 1:
 			(void)snprintf(buf, sizeof(buf), "%u", (uint32_t)v);
-			if (file_printf(ms, m->desc, buf) == -1)
+			if (np && file_printf(ms, m->desc, buf) == -1)
 				return -1;
 			break;
 		default:
-			if (file_printf(ms, m->desc, (uint32_t) v) == -1)
+			if (np && file_printf(ms, m->desc, (uint32_t) v) == -1)
 				return -1;
 			break;
 		}
@@ -397,7 +406,7 @@ mprint(struct magic_set *ms, struct magic *m)
   	case FILE_BEQUAD:
   	case FILE_LEQUAD:
 		v = file_signextend(ms, m, p->q);
-		if (file_printf(ms, m->desc, (uint64_t) v) == -1)
+		if (np && file_printf(ms, m->desc, (uint64_t) v) == -1)
 			return -1;
 		t = ms->offset + sizeof(int64_t);
   		break;
@@ -407,14 +416,14 @@ mprint(struct magic_set *ms, struct magic *m)
   	case FILE_BESTRING16:
   	case FILE_LESTRING16:
 		if (m->reln == '=' || m->reln == '!') {
-			if (file_printf(ms, m->desc, m->value.s) == -1)
+			if (np && file_printf(ms, m->desc, m->value.s) == -1)
 				return -1;
 			t = ms->offset + m->vallen;
 		}
 		else {
 			if (*m->value.s == '\0')
 				p->s[strcspn(p->s, "\n")] = '\0';
-			if (file_printf(ms, m->desc, p->s) == -1)
+			if (np && file_printf(ms, m->desc, p->s) == -1)
 				return -1;
 			t = ms->offset + strlen(p->s);
 			if (m->type == FILE_PSTRING)
@@ -426,7 +435,7 @@ mprint(struct magic_set *ms, struct magic *m)
 	case FILE_BEDATE:
 	case FILE_LEDATE:
 	case FILE_MEDATE:
-		if (file_printf(ms, m->desc, file_fmttime(p->l, 1)) == -1)
+		if (np && file_printf(ms, m->desc, file_fmttime(p->l, 1)) == -1)
 			return -1;
 		t = ms->offset + sizeof(time_t);
 		break;
@@ -435,7 +444,7 @@ mprint(struct magic_set *ms, struct magic *m)
 	case FILE_BELDATE:
 	case FILE_LELDATE:
 	case FILE_MELDATE:
-		if (file_printf(ms, m->desc, file_fmttime(p->l, 0)) == -1)
+		if (np && file_printf(ms, m->desc, file_fmttime(p->l, 0)) == -1)
 			return -1;
 		t = ms->offset + sizeof(time_t);
 		break;
@@ -443,8 +452,8 @@ mprint(struct magic_set *ms, struct magic *m)
 	case FILE_QDATE:
 	case FILE_BEQDATE:
 	case FILE_LEQDATE:
-		if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q, 1))
-		    == -1)
+		if (np && file_printf(ms, m->desc, file_fmttime((uint32_t)p->q,
+		    1)) == -1)
 			return -1;
 		t = ms->offset + sizeof(uint64_t);
 		break;
@@ -452,8 +461,8 @@ mprint(struct magic_set *ms, struct magic *m)
 	case FILE_QLDATE:
 	case FILE_BEQLDATE:
 	case FILE_LEQLDATE:
-		if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q, 0))
-		    == -1)
+		if (np && file_printf(ms, m->desc, file_fmttime((uint32_t)p->q,
+		    0)) == -1)
 			return -1;
 		t = ms->offset + sizeof(uint64_t);
 		break;
@@ -467,11 +476,11 @@ mprint(struct magic_set *ms, struct magic *m)
 			return -1;
 		case 1:
 			(void)snprintf(buf, sizeof(buf), "%g", vf);
-			if (file_printf(ms, m->desc, buf) == -1)
+			if (np && file_printf(ms, m->desc, buf) == -1)
 				return -1;
 			break;
 		default:
-			if (file_printf(ms, m->desc, vf) == -1)
+			if (np && file_printf(ms, m->desc, vf) == -1)
 				return -1;
 			break;
 		}
@@ -487,11 +496,11 @@ mprint(struct magic_set *ms, struct magic *m)
 			return -1;
 		case 1:
 			(void)snprintf(buf, sizeof(buf), "%g", vd);
-			if (file_printf(ms, m->desc, buf) == -1)
+			if (np && file_printf(ms, m->desc, buf) == -1)
 				return -1;
 			break;
 		default:
-			if (file_printf(ms, m->desc, vd) == -1)
+			if (np && file_printf(ms, m->desc, vd) == -1)
 				return -1;
 			break;
 		}
@@ -507,7 +516,10 @@ mprint(struct magic_set *ms, struct magic *m)
 			file_oomem(ms, ms->search.rm_len);
 			return -1;
 		}
-		rval = file_printf(ms, m->desc, cp);
+		if (np)
+			rval = file_printf(ms, m->desc, cp);
+		else
+			rval = 0;
 		free(cp);
 
 		if (rval == -1)
@@ -521,7 +533,7 @@ mprint(struct magic_set *ms, struct magic *m)
 	}
 
 	case FILE_SEARCH:
-	  	if (file_printf(ms, m->desc, m->value.s) == -1)
+	  	if (np && file_printf(ms, m->desc, m->value.s) == -1)
 			return -1;
 		if ((m->str_flags & REGEX_OFFSET_START))
 			t = ms->search.offset;
@@ -530,7 +542,7 @@ mprint(struct magic_set *ms, struct magic *m)
 		break;
 
 	case FILE_DEFAULT:
-	  	if (file_printf(ms, m->desc, m->value.s) == -1)
+	  	if (np && file_printf(ms, m->desc, m->value.s) == -1)
 			return -1;
 		t = ms->offset;
 		break;
@@ -1454,7 +1466,8 @@ mget(struct magic_set *ms, const unsigned char *s,
 		break;
 
 	case FILE_INDIRECT:
-	  	if (file_printf(ms, m->desc) == -1)
+	  	if ((ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0 &&
+		    file_printf(ms, m->desc) == -1)
 			return -1;
 		if (nbytes < offset)
 			return 0;
@@ -1858,25 +1871,19 @@ handle_annotation(struct magic_set *ms, struct magic *m)
 			return -1;
 		return 1;
 	}
-	if (ms->flags & MAGIC_MIME) {
-		if ((ms->flags & MAGIC_MIME_TYPE) && m->mimetype[0]) {
-			ms->event_flags |= EVENT_WROTE_MIME_TYPE;
-			if (file_printf(ms, "%s", m->mimetype) == -1)
-				return -1;
-		}
-		/* If we want an encoding, let ascmagic find it. */
-		if ((ms->flags & MAGIC_MIME_ENCODING))
-			return 0;
-		/* If we didn't write a MIME type, and we want one,
-		   allow ascmagic to run.*/
-		return m->mimetype[0] != '\0';
+	if ((ms->flags & MAGIC_MIME_TYPE) && m->mimetype[0]) {
+		if (file_printf(ms, "%s", m->mimetype) == -1)
+			return -1;
+		return 1;
 	}
-	return -2;
+	return 0;
 }
 
 private int
 print_sep(struct magic_set *ms, int firstline)
 {
+	if (ms->flags & MAGIC_MIME)
+		return 0;
 	if (firstline)
 		return 0;
 	/*