From: Michihiro NAKAJIMA <ggcueroad@gmail.com>
Date: Wed, 9 Mar 2011 17:12:29 +0000 (-0500)
Subject: Add the ability to be able to recognize mtree format without a signature "#mtree"
X-Git-Tag: v3.0.0a~670
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=cbc36660ea2b5b2f891b4b0cad1bf055641978a0;p=thirdparty%2Flibarchive.git

Add the ability to be able to recognize mtree format without a signature "#mtree"
since mtree utility still does not record the signature.

SVN-Revision: 2999
---

diff --git a/Makefile.am b/Makefile.am
index 49ddc5f51..6b28b5460 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -440,6 +440,7 @@ libarchive_test_EXTRA_DIST=\
 	libarchive/test/test_read_format_lha_lh7.lzh.uu			\
 	libarchive/test/test_read_format_lha_withjunk.lzh.uu		\
 	libarchive/test/test_read_format_mtree.mtree.uu			\
+	libarchive/test/test_read_format_mtree_nomagic.mtree.uu		\
 	libarchive/test/test_read_format_raw.data.Z.uu			\
 	libarchive/test/test_read_format_raw.data.uu			\
 	libarchive/test/test_read_format_tar_empty_filename.tar.uu	\
diff --git a/libarchive/archive_read_support_format_mtree.c b/libarchive/archive_read_support_format_mtree.c
index e24aa8692..27dadface 100644
--- a/libarchive/archive_read_support_format_mtree.c
+++ b/libarchive/archive_read_support_format_mtree.c
@@ -1,6 +1,7 @@
 /*-
  * Copyright (c) 2003-2007 Tim Kientzle
  * Copyright (c) 2008 Joerg Sonnenberger
+ * Copyright (c) 2011 Michihiro NAKAJIMA
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -186,20 +187,377 @@ cleanup(struct archive_read *a)
 	return (ARCHIVE_OK);
 }
 
+static ssize_t
+get_line_size(const char *b, ssize_t avail, ssize_t *nlsize)
+{
+	ssize_t len;
+
+	len = 0;
+	while (len < avail) {
+		switch (*b) {
+		case '\0':/* Non-ascii character or control character. */
+			if (nlsize != NULL)
+				*nlsize = 0;
+			return (-1);
+		case '\r':
+			if (avail-len > 1 && b[1] == '\n') {
+				if (nlsize != NULL)
+					*nlsize = 2;
+				return (len+2);
+			}
+			/* FALL THROUGH */
+		case '\n':
+			if (nlsize != NULL)
+				*nlsize = 1;
+			return (len+1);
+		default:
+			b++;
+			len++;
+			break;
+		}
+	}
+	if (nlsize != NULL)
+		*nlsize = 0;
+	return (avail);
+}
+
+static ssize_t
+next_line(struct archive_read *a,
+    const char **b, ssize_t *avail, ssize_t *ravail, ssize_t *nl)
+{
+	ssize_t len;
+	int quit;
+	
+	quit = 0;
+	if (*avail == 0) {
+		*nl = 0;
+		len = 0;
+	} else
+		len = get_line_size(*b, *avail, nl);
+	/*
+	 * Read bytes more while it does not reach the end of line.
+	 */
+	while (*nl == 0 && len == *avail && !quit) {
+		ssize_t diff = *ravail - *avail;
+
+		*b = __archive_read_ahead(a, 160 + *ravail, avail);
+		if (*b == NULL) {
+			if (*ravail >= *avail)
+				return (0);
+			/* Reading bytes reaches the end of file. */
+			*b = __archive_read_ahead(a, *avail, avail);
+			quit = 1;
+		}
+		*ravail = *avail;
+		*b += diff;
+		*avail -= diff;
+		len = get_line_size(*b, *avail, nl);
+	}
+	return (len);
+}
+
+/*
+ * Compare characters with a mtree keyword.
+ * Returns the length of a mtree keyword if matched.
+ * Returns 0 if not matched.
+ */
+int
+bid_keycmp(const char *p, const char *key, ssize_t len)
+{
+	int match_len = 0;
+
+	while (len > 0 && *p && *key) {
+		if (*p == *key) {
+			--len;
+			++p;
+			++key;
+			++match_len;
+			continue;
+		}
+		return (0);/* Not match */
+	}
+	if (*key != '\0')
+		return (0);/* Not match */
+
+	/* A following character should be specified characters */
+	if (p[0] == '=' || p[0] == ' ' || p[0] == '\t' ||
+	    p[0] == '\n' || p[0] == '\r' ||
+	   (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r')))
+		return (match_len);
+	return (0);/* Not match */
+}
+
+/*
+ * Test whether the characters 'p' has is mtree keyword.
+ * Returns the length of a detected keyword.
+ * Returns 0 if any keywords were not found.
+ */
+static ssize_t
+bid_keyword(const char *p,  ssize_t len)
+{
+	static const char *keys_c[] = {
+		"content", "contents", "cksum", NULL
+	};
+	static const char *keys_df[] = {
+		"device", "flags", NULL
+	};
+	static const char *keys_g[] = {
+		"gid", "gname", NULL
+	};
+	static const char *keys_il[] = {
+		"ignore", "link", NULL
+	};
+	static const char *keys_m[] = {
+		"md5", "md5digest", "mode", NULL
+	};
+	static const char *keys_no[] = {
+		"nlink", "optional", NULL
+	};
+	static const char *keys_r[] = {
+		"rmd160", "rmd160digest", NULL
+	};
+	static const char *keys_s[] = {
+		"sha1", "sha1digest",
+		"sha256", "sha256digest",
+		"sha384", "sha384digest",
+		"sha512", "sha512digest",
+		"size", NULL
+	};
+	static const char *keys_t[] = {
+		"tags", "time", "type", NULL
+	};
+	static const char *keys_u[] = {
+		"uid", "uname",	NULL
+	};
+	const char **keys;
+	int i;
+
+	switch (*p) {
+	case 'c': keys = keys_c; break;
+	case 'd': case 'f': keys = keys_df; break;
+	case 'g': keys = keys_g; break;
+	case 'i': case 'l': keys = keys_il; break;
+	case 'm': keys = keys_m; break;
+	case 'n': case 'o': keys = keys_no; break;
+	case 'r': keys = keys_r; break;
+	case 's': keys = keys_s; break;
+	case 't': keys = keys_t; break;
+	case 'u': keys = keys_u; break;
+	default: return (0);/* Unknown key */
+	}
+
+	for (i = 0; keys[i] != NULL; i++) {
+		int l = bid_keycmp(p, keys[i], len);
+		if (l > 0)
+			return (l);
+	}
+	return (0);/* Unkown key */
+}
+
+/*
+ * Test whether there is a set of mtree keywords.
+ * Returns the number of keyword.
+ * Returns -1 if we got incorrect sequence.
+ * This function expects a set of "<space characters>keyword=value".
+ * When "unset" is specified, expects a set of "<space characters>keyword".
+ */
+static int
+bid_keyword_list(const char *p,  ssize_t len, int unset)
+{
+	int l;
+	int keycnt = 0;
+
+	while (len > 0 && *p) {
+		int blank = 0;
+
+		/* Test whether there are blank characters in the line. */
+		while (len >0 && (*p == ' ' || *p == '\t')) {
+			++p;
+			--len;
+			blank = 1;
+		}
+		if (*p == '\n' || *p == '\r')
+			break;
+		if (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r'))
+			break;
+		if (!blank) /* No blank character. */
+			return (-1);
+
+		if (unset) {
+			l = bid_keycmp(p, "all", len);
+			if (l > 0)
+				return (1);
+		}
+		/* Test whether there is a correct key in the line. */
+		l = bid_keyword(p, len);
+		if (l == 0)
+			return (-1);/* Unknown keyword was found. */
+		p += l;
+		len -= l;
+		keycnt++;
+
+		/* Skip value */
+		if (*p == '=') {
+			int value = 0;
+			++p;
+			--len;
+			while (len > 0 && *p != ' ' && *p != '\t') {
+				++p;
+				--len;
+				value = 1;
+			}
+			/* A keyword should have a its value unless
+			 * "/unset" operation. */ 
+			if (!unset && value == 0)
+				return (-1);
+		}
+	}
+	return (keycnt);
+}
+
+static int
+bid_entry(const char *p, ssize_t len)
+{
+	int f = 0;
+	static const unsigned char safe_char[256] = {
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */
+		/* !"$%&'()*+,-./  EXCLUSION:( )(#) */
+		0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */
+		/* 0123456789:;<>?  EXCLUSION:(=) */
+		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */
+		/* @ABCDEFGHIJKLMNO */
+		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */
+		/* PQRSTUVWXYZ[\]^_  */
+		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */
+		/* `abcdefghijklmno */
+		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */
+		/* pqrstuvwxyz{|}~ */
+		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */
+	};
+
+	/*
+	 * Skip the path-name which is quoted.
+	 */
+	while (len > 0 && *p != ' ' && *p != '\t') {
+		if (!safe_char[*(const unsigned char *)p])
+			return (-1);
+		++p;
+		--len;
+		++f;
+	}
+	/* If a path-name was not found, returns error. */
+	if (f == 0)
+		return (-1);
+
+	return (bid_keyword_list(p, len, 0));
+}
+
+#define MAX_BID_ENTRY	3
 
 static int
 mtree_bid(struct archive_read *a)
 {
 	const char *signature = "#mtree";
 	const char *p;
+	ssize_t avail, ravail;
+	ssize_t len, nl;
+	int detected_bytes = 0, entry_cnt = 0, multiline = 0;
 
 	/* Now let's look at the actual header and see if it matches. */
-	p = __archive_read_ahead(a, strlen(signature), NULL);
+	p = __archive_read_ahead(a, strlen(signature), &avail);
 	if (p == NULL)
 		return (-1);
 
 	if (strncmp(p, signature, strlen(signature)) == 0)
 		return (8 * (int)strlen(signature));
+
+	/*
+	 * There is not a mtree signature. Let's try to detect mtree format.
+	 */
+	ravail = avail;
+	for (;;) {
+		len = next_line(a, &p, &avail, &ravail, &nl);
+		/* The terminal character of the line should be
+		 * a new line character, '\r\n' or '\n'. */
+		if (len <= 0 || nl == 0)
+			break;
+		if (!multiline) {
+			/* Leading whitespace is never significant,
+			 * ignore it. */
+			while (len > 0 && (*p == ' ' || *p == '\t')) {
+				++p;
+				--avail;
+				--len;
+			}
+			/* Skip comment or empty line. */ 
+			if (p[0] == '#' || p[0] == '\n' || p[0] == '\r') {
+				p += len;
+				avail -= len;
+				continue;
+			}
+		} else {
+			/* A continuance line; the terminal
+			 * character of previous line was '\' character. */
+			if (bid_keyword_list(p, len, 0) <= 0)
+				break;
+			if (multiline == 1)
+				detected_bytes += len;
+			if (p[len-nl-1] != '\\') {
+				if (multiline == 1 &&
+				    ++entry_cnt >= MAX_BID_ENTRY)
+					break;
+				multiline = 0;
+			}
+			p += len;
+			avail -= len;
+			continue;
+		}
+		if (p[0] != '/') {
+			if (bid_entry(p, len) >= 0) {
+				detected_bytes += len;
+				if (p[len-nl-1] == '\\')
+					/* This line continues. */
+					multiline = 1;
+				else {
+					/* We've got plenty of correct lines
+					 * to assume that this file is a mtree
+					 * format. */
+					if (++entry_cnt >= MAX_BID_ENTRY)
+						break;
+				}
+			} else
+				break;
+		} else if (strncmp(p, "/set", 4) == 0) {
+			if (bid_keyword_list(p+4, len-4, 0) <= 0)
+				break;
+			/* This line continues. */
+			if (p[len-nl-1] == '\\')
+				multiline = 2;
+		} else if (strncmp(p, "/unset", 6) == 0) {
+			if (bid_keyword_list(p+6, len-6, 1) <= 0)
+				break;
+			/* This line continues. */
+			if (p[len-nl-1] == '\\')
+				multiline = 2;
+		} else
+			break;
+
+		/* Test next line. */
+		p += len;
+		avail -= len;
+	}
+	if (entry_cnt >= MAX_BID_ENTRY || (entry_cnt > 0 && len == 0))
+		return (32);
+
 	return (0);
 }
 
diff --git a/libarchive/test/test_read_format_mtree.c b/libarchive/test/test_read_format_mtree.c
index 1ea45ed1e..02c4be36e 100644
--- a/libarchive/test/test_read_format_mtree.c
+++ b/libarchive/test/test_read_format_mtree.c
@@ -177,10 +177,91 @@ test_read_format_mtree3(void)
 }
 
 
+static void
+test_read_format_mtree4(void)
+{
+	const char reffile[] = "test_read_format_mtree_nomagic.mtree";
+	char buff[16];
+	struct archive_entry *ae;
+	struct archive *a;
+	FILE *f;
+
+	assertMakeDir("mtree4", 0777);
+	assertChdir("mtree4");
+
+	extract_reference_file(reffile);
+
+	assert((a = archive_read_new()) != NULL);
+	assertEqualIntA(a, ARCHIVE_OK,
+	    archive_read_support_compression_all(a));
+	assertEqualIntA(a, ARCHIVE_OK,
+	    archive_read_support_format_all(a));
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_open_file(a, reffile, 11));
+
+	/*
+	 * Read "file", whose data is available on disk.
+	 */
+	f = fopen("file", "wb");
+	assert(f != NULL);
+	assertEqualInt(3, fwrite("hi\n", 1, 3, f));
+	fclose(f);
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualInt(archive_format(a), ARCHIVE_FORMAT_MTREE);
+	assertEqualString(archive_entry_pathname(ae), "file");
+	assertEqualInt(archive_entry_uid(ae), 18);
+	assertEqualInt(AE_IFREG, archive_entry_filetype(ae));
+	assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0123);
+	assertEqualInt(archive_entry_size(ae), 3);
+	assertEqualInt(3, archive_read_data(a, buff, 3));
+	assertEqualMem(buff, "hi\n", 3);
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir");
+	assertEqualInt(AE_IFDIR, archive_entry_filetype(ae));
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir/file with space");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "file with space");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir2");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir2/dir3a");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir2/dir3a/indir3a");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir2/fullindir2");
+	assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644);
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir2/indir2");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir2/dir3b");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "dir2/dir3b/indir3b");
+
+	assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+	assertEqualString(archive_entry_pathname(ae), "notindir");
+
+	assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae));
+	assertEqualInt(ARCHIVE_OK, archive_read_close(a));
+	assertEqualInt(ARCHIVE_OK, archive_read_free(a));
+
+	assertChdir("..");
+}
+
 
 DEFINE_TEST(test_read_format_mtree)
 {
 	test_read_format_mtree1();
 	test_read_format_mtree2();
 	test_read_format_mtree3();
+	test_read_format_mtree4();
 }
diff --git a/libarchive/test/test_read_format_mtree_nomagic.mtree.uu b/libarchive/test/test_read_format_mtree_nomagic.mtree.uu
new file mode 100644
index 000000000..be089103d
--- /dev/null
+++ b/libarchive/test/test_read_format_mtree_nomagic.mtree.uu
@@ -0,0 +1,11 @@
+begin 644 test_read_format_mtree_nomagic.mtree
+M9FEL92!T>7!E/69I;&4@=6ED/3$X(&UO9&4],#$R,R!S:7IE/3,*9&ER('1Y
+M<&4]9&ER"B!F:6QE7#`T,'=I=&A<,#0P<W!A8V4@='EP93UF:6QE('5I9#TQ
+M.`H@+BX*9FEL95PP-#!W:71H7#`T,'-P86-E('1Y<&4]9FEL90ID:7(R('1Y
+M<&4]9&ER"B!D:7(S82!T>7!E/61I<@H@(&EN9&ER,V$@='EP93UF:6QE"F1I
+M<C(O9G5L;&EN9&ER,B!T>7!E/69I;&4@;6]D93TP-S<W"B`@+BX*(&EN9&ER
+M,B!T>7!E/69I;&4*(&1I<C-B('1Y<&4]9&ER"B`@:6YD:7(S8B!T>7!E/69I
+M;&4*("`N+@H@+BX*;F]T:6YD:7(@='EP93UF:6QE"F1I<C(O9G5L;&EN9&ER
+,,B!M;V1E/3`V-#0*
+`
+end