unzip: handle "central directory"

 needed for OpenOffice, gmail attachment .zips etc
 conditional on CONFIG_DESKTOP

function                                             old     new   delta
unzip_main                                          1643    1939    +296
find_cds_offset                                        -     173    +173
unzip_skip                                            11      16      +5
------------------------------------------------------------------------------
(add/remove: 1/0 grow/shrink: 2/0 up/down: 474/0)             Total: 474 bytes

1 files changed, 196 insertions, 47 deletions

diff --git a/archival/unzip.c b/archival/unzip.c
index 1c9bc24..e468ff4 100644
--- a/archival/unzip.c
+++ b/archival/unzip.c
@@ -16,12 +16,7 @@
  */
 
 /* TODO
- * Endian issues
  * Zip64 + other methods
- * Improve handling of zip format, ie.
- * - deferred CRC, comp. & uncomp. lengths (zip header flags bit 3)
- * - unix file permissions, etc.
- * - central directory
  */
 
 #include "libbb.h"
@@ -31,12 +26,12 @@ enum {
 #if BB_BIG_ENDIAN
 	ZIP_FILEHEADER_MAGIC = 0x504b0304,
 	ZIP_CDS_MAGIC        = 0x504b0102,
-	ZIP_CDS_END_MAGIC    = 0x504b0506,
+	ZIP_CDE_MAGIC        = 0x504b0506,
 	ZIP_DD_MAGIC         = 0x504b0708,
 #else
 	ZIP_FILEHEADER_MAGIC = 0x04034b50,
 	ZIP_CDS_MAGIC        = 0x02014b50,
-	ZIP_CDS_END_MAGIC    = 0x06054b50,
+	ZIP_CDE_MAGIC        = 0x06054b50,
 	ZIP_DD_MAGIC         = 0x08074b50,
 #endif
 };
@@ -46,16 +41,16 @@ enum {
 typedef union {
 	uint8_t raw[ZIP_HEADER_LEN];
 	struct {
-		uint16_t version;                       /* 0-1 */
-		uint16_t flags;                         /* 2-3 */
-		uint16_t method;                        /* 4-5 */
-		uint16_t modtime;                       /* 6-7 */
-		uint16_t moddate;                       /* 8-9 */
-		uint32_t crc32 PACKED;        /* 10-13 */
-		uint32_t cmpsize PACKED;      /* 14-17 */
-		uint32_t ucmpsize PACKED;     /* 18-21 */
-		uint16_t filename_len;                  /* 22-23 */
-		uint16_t extra_len;                     /* 24-25 */
+		uint16_t version;               /* 0-1 */
+		uint16_t flags;                 /* 2-3 */
+		uint16_t method;                /* 4-5 */
+		uint16_t modtime;               /* 6-7 */
+		uint16_t moddate;               /* 8-9 */
+		uint32_t crc32 PACKED;          /* 10-13 */
+		uint32_t cmpsize PACKED;        /* 14-17 */
+		uint32_t ucmpsize PACKED;       /* 18-21 */
+		uint16_t filename_len;          /* 22-23 */
+		uint16_t extra_len;             /* 24-25 */
 	} formatted PACKED;
 } zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
 
@@ -65,11 +60,11 @@ typedef union {
  */
 struct BUG_zip_header_must_be_26_bytes {
 	char BUG_zip_header_must_be_26_bytes[
-		offsetof(zip_header_t, formatted.extra_len) + 2 ==
-			ZIP_HEADER_LEN ? 1 : -1];
+		offsetof(zip_header_t, formatted.extra_len) + 2
+			== ZIP_HEADER_LEN ? 1 : -1];
 };
 
-#define FIX_ENDIANNESS(zip_header) do { \
+#define FIX_ENDIANNESS_ZIP(zip_header) do { \
 	(zip_header).formatted.version      = SWAP_LE16((zip_header).formatted.version     ); \
 	(zip_header).formatted.flags        = SWAP_LE16((zip_header).formatted.flags       ); \
 	(zip_header).formatted.method       = SWAP_LE16((zip_header).formatted.method      ); \
@@ -82,9 +77,138 @@ struct BUG_zip_header_must_be_26_bytes {
 	(zip_header).formatted.extra_len    = SWAP_LE16((zip_header).formatted.extra_len   ); \
 } while (0)
 
-static void unzip_skip(int fd, off_t skip)
+#define CDS_HEADER_LEN 42
+
+typedef union {
+	uint8_t raw[CDS_HEADER_LEN];
+	struct {
+		/* uint32_t signature; 50 4b 01 02 */
+		uint16_t version_made_by;       /* 0-1 */
+		uint16_t version_needed;        /* 2-3 */
+		uint16_t cds_flags;             /* 4-5 */
+		uint16_t method;                /* 6-7 */
+		uint16_t mtime;                 /* 8-9 */
+		uint16_t mdate;                 /* 10-11 */
+		uint32_t crc32;                 /* 12-15 */
+		uint32_t cmpsize;               /* 16-19 */
+		uint32_t ucmpsize;              /* 20-23 */
+		uint16_t file_name_length;      /* 24-25 */
+		uint16_t extra_field_length;    /* 26-27 */
+		uint16_t file_comment_length;   /* 28-29 */
+		uint16_t disk_number_start;     /* 30-31 */
+		uint16_t internal_file_attributes; /* 32-33 */
+		uint32_t external_file_attributes PACKED; /* 34-37 */
+		uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
+	} formatted PACKED;
+} cds_header_t;
+
+struct BUG_cds_header_must_be_42_bytes {
+	char BUG_cds_header_must_be_42_bytes[
+		offsetof(cds_header_t, formatted.relative_offset_of_local_header) + 4
+			== CDS_HEADER_LEN ? 1 : -1];
+};
+
+#define FIX_ENDIANNESS_CDS(cds_header) do { \
+	(cds_header).formatted.crc32        = SWAP_LE32((cds_header).formatted.crc32       ); \
+	(cds_header).formatted.cmpsize      = SWAP_LE32((cds_header).formatted.cmpsize     ); \
+	(cds_header).formatted.ucmpsize     = SWAP_LE32((cds_header).formatted.ucmpsize    ); \
+	(cds_header).formatted.file_name_length = SWAP_LE16((cds_header).formatted.file_name_length); \
+	(cds_header).formatted.extra_field_length = SWAP_LE16((cds_header).formatted.extra_field_length); \
+	(cds_header).formatted.file_comment_length = SWAP_LE16((cds_header).formatted.file_comment_length); \
+} while (0)
+
+#define CDE_HEADER_LEN 16
+
+typedef union {
+	uint8_t raw[CDE_HEADER_LEN];
+	struct {
+		/* uint32_t signature; 50 4b 05 06 */
+		uint16_t this_disk_no;
+		uint16_t disk_with_cds_no;
+		uint16_t cds_entries_on_this_disk;
+		uint16_t cds_entries_total;
+		uint32_t cds_size;
+		uint32_t cds_offset;
+		/* uint16_t file_comment_length; */
+		/* .ZIP file comment (variable size) */
+	} formatted PACKED;
+} cde_header_t;
+
+struct BUG_cde_header_must_be_16_bytes {
+	char BUG_cde_header_must_be_16_bytes[
+		sizeof(cde_header_t) == CDE_HEADER_LEN ? 1 : -1];
+};
+
+#define FIX_ENDIANNESS_CDE(cde_header) do { \
+	(cde_header).formatted.cds_offset = SWAP_LE16((cde_header).formatted.cds_offset); \
+} while (0)
+
+enum { zip_fd = 3 };
+
+
+#if ENABLE_DESKTOP
+/* NB: does not preserve file position! */
+static uint32_t find_cds_offset(void)
+{
+	unsigned char buf[1024];
+	cde_header_t cde_header;
+	unsigned char *p;
+	off_t end;
+
+	end = xlseek(zip_fd, 0, SEEK_END);
+	if (end < 1024)
+		end = 1024;
+	end -= 1024;
+	xlseek(zip_fd, end, SEEK_SET);
+	full_read(zip_fd, buf, 1024);
+
+	p = buf;
+	while (p <= buf + 1024 - CDE_HEADER_LEN - 4) {
+		if (*p != 'P') {
+			p++;
+			continue;
+		}
+		if (*++p != 'K')
+			continue;
+		if (*++p != 5)
+			continue;
+		if (*++p != 6)
+			continue;
+		/* we found CDE! */
+		memcpy(cde_header.raw, p + 1, CDE_HEADER_LEN);
+		FIX_ENDIANNESS_CDE(cde_header);
+		return cde_header.formatted.cds_offset;
+	}
+	bb_error_msg_and_die("can't find file table");
+};
+
+static uint32_t read_next_cds(int count_m1, uint32_t cds_offset, cds_header_t *cds_ptr)
 {
-	bb_copyfd_exact_size(fd, -1, skip);
+	off_t org;
+
+	org = xlseek(zip_fd, 0, SEEK_CUR);
+
+	if (!cds_offset)
+		cds_offset = find_cds_offset();
+
+	while (count_m1-- >= 0) {
+		xlseek(zip_fd, cds_offset + 4, SEEK_SET);
+		xread(zip_fd, cds_ptr->raw, CDS_HEADER_LEN);
+		FIX_ENDIANNESS_CDS(*cds_ptr);
+		cds_offset += 4 + CDS_HEADER_LEN
+			+ cds_ptr->formatted.file_name_length
+			+ cds_ptr->formatted.extra_field_length
+			+ cds_ptr->formatted.file_comment_length;
+	}
+
+	xlseek(zip_fd, org, SEEK_SET);
+	return cds_offset;
+};
+#endif
+
+static void unzip_skip(off_t skip)
+{
+	bb_copyfd_exact_size(zip_fd, -1, skip);
 }
 
 static void unzip_create_leading_dirs(const char *fn)
@@ -97,17 +221,17 @@ static void unzip_create_leading_dirs(const char *fn)
 	free(name);
 }
 
-static void unzip_extract(zip_header_t *zip_header, int src_fd, int dst_fd)
+static void unzip_extract(zip_header_t *zip_header, int dst_fd)
 {
 	if (zip_header->formatted.method == 0) {
 		/* Method 0 - stored (not compressed) */
 		off_t size = zip_header->formatted.ucmpsize;
 		if (size)
-			bb_copyfd_exact_size(src_fd, dst_fd, size);
+			bb_copyfd_exact_size(zip_fd, dst_fd, size);
 	} else {
 		/* Method 8 - inflate */
 		inflate_unzip_result res;
-		if (inflate_unzip(&res, zip_header->formatted.cmpsize, src_fd, dst_fd) < 0)
+		if (inflate_unzip(&res, zip_header->formatted.cmpsize, zip_fd, dst_fd) < 0)
 			bb_error_msg_and_die("inflate error");
 		/* Validate decompression - crc */
 		if (zip_header->formatted.crc32 != (res.crc ^ 0xffffffffL)) {
@@ -131,9 +255,12 @@ int unzip_main(int argc, char **argv)
 	smallint verbose = 1;
 	smallint listing = 0;
 	smallint overwrite = O_PROMPT;
+#if ENABLE_DESKTOP
+	uint32_t cds_offset;
+	unsigned cds_entries;
+#endif
 	unsigned total_size;
 	unsigned total_entries;
-	int src_fd = -1;
 	int dst_fd = -1;
 	char *src_fn = NULL;
 	char *dst_fn = NULL;
@@ -221,13 +348,14 @@ int unzip_main(int argc, char **argv)
 
 	/* Open input file */
 	if (LONE_DASH(src_fn)) {
-		src_fd = STDIN_FILENO;
+		xdup2(STDIN_FILENO, zip_fd);
 		/* Cannot use prompt mode since zip data is arriving on STDIN */
 		if (overwrite == O_PROMPT)
 			overwrite = O_NEVER;
 	} else {
 		static const char extn[][5] = {"", ".zip", ".ZIP"};
 		int orig_src_fn_len = strlen(src_fn);
+		int src_fd = -1;
 
 		for (i = 0; (i < 3) && (src_fd == -1); i++) {
 			strcpy(src_fn + orig_src_fn_len, extn[i]);
@@ -237,6 +365,7 @@ int unzip_main(int argc, char **argv)
 			src_fn[orig_src_fn_len] = '\0';
 			bb_error_msg_and_die("can't open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
 		}
+		xmove_fd(src_fd, zip_fd);
 	}
 
 	/* Change dir if necessary */
@@ -253,37 +382,63 @@ int unzip_main(int argc, char **argv)
 
 	total_size = 0;
 	total_entries = 0;
+#if ENABLE_DESKTOP
+	cds_entries = 0;
+	cds_offset = 0;
+#endif
 	while (1) {
 		uint32_t magic;
 
 		/* Check magic number */
-		xread(src_fd, &magic, 4);
+		xread(zip_fd, &magic, 4);
 		/* Central directory? It's at the end, so exit */
 		if (magic == ZIP_CDS_MAGIC)
 			break;
+#if ENABLE_DESKTOP
+		/* Data descriptor? It was a streaming file, go on */
+		if (magic == ZIP_DD_MAGIC) {
+			/* skip over duplicate crc32, cmpsize and ucmpsize */
+			unzip_skip(3 * 4);
+			continue;
+		}
+#endif
 		if (magic != ZIP_FILEHEADER_MAGIC)
 			bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 
 		/* Read the file header */
-		xread(src_fd, zip_header.raw, ZIP_HEADER_LEN);
-		FIX_ENDIANNESS(zip_header);
+		xread(zip_fd, zip_header.raw, ZIP_HEADER_LEN);
+		FIX_ENDIANNESS_ZIP(zip_header);
 		if ((zip_header.formatted.method != 0) && (zip_header.formatted.method != 8)) {
 			bb_error_msg_and_die("unsupported method %d", zip_header.formatted.method);
 		}
-		if (zip_header.formatted.flags & (0x0008|0x0001)) {
+#if !ENABLE_DESKTOP
+		if (zip_header.formatted.flags & 0x0009) {
+			bb_error_msg_and_die("zip flags 1 and 8 are not supported");
+		}
+#else
+		if (zip_header.formatted.flags & 0x0001) {
 			/* 0x0001 - encrypted */
+			bb_error_msg_and_die("zip flag 1 (encryption) is not supported");
+		}
+		if (zip_header.formatted.flags & 0x0008) {
+			cds_header_t cds_header;
 			/* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 			 * only from Central Directory. See unzip_doc.txt */
-			bb_error_msg_and_die("zip flags 8 and 1 are not supported");
+			cds_offset = read_next_cds(total_entries - cds_entries, cds_offset, &cds_header);
+			cds_entries = total_entries + 1;
+			zip_header.formatted.crc32    = cds_header.formatted.crc32;
+			zip_header.formatted.cmpsize  = cds_header.formatted.cmpsize;
+			zip_header.formatted.ucmpsize = cds_header.formatted.ucmpsize;
 		}
+#endif
 
 		/* Read filename */
 		free(dst_fn);
 		dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
-		xread(src_fd, dst_fn, zip_header.formatted.filename_len);
+		xread(zip_fd, dst_fn, zip_header.formatted.filename_len);
 
 		/* Skip extra header bytes */
-		unzip_skip(src_fd, zip_header.formatted.extra_len);
+		unzip_skip(zip_header.formatted.extra_len);
 
 		/* Filter zip entries */
 		if (find_list_entry(zreject, dst_fn)
@@ -304,7 +459,6 @@ int unzip_main(int argc, char **argv)
 					   (dostime & 0x000007e0) >> 5,
 					   dst_fn);
 					total_size += zip_header.formatted.ucmpsize;
-					total_entries++;
 				} else {
 					/* short listing -- filenames only */
 					puts(dst_fn);
@@ -315,7 +469,7 @@ int unzip_main(int argc, char **argv)
 			} else if (last_char_is(dst_fn, '/')) { /* Extract directory */
 				if (stat(dst_fn, &stat_buf) == -1) {
 					if (errno != ENOENT) {
-						bb_perror_msg_and_die("cannot stat '%s'", dst_fn);
+						bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 					}
 					if (verbose) {
 						printf("   creating: %s\n", dst_fn);
@@ -335,7 +489,7 @@ int unzip_main(int argc, char **argv)
  check_file:
 				if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
 					if (errno != ENOENT) {
-						bb_perror_msg_and_die("cannot stat '%s'", dst_fn);
+						bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 					}
 					i = 'y';
 				} else { /* File already exists */
@@ -347,7 +501,7 @@ int unzip_main(int argc, char **argv)
 						} else {
 							printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 							if (!fgets(key_buf, sizeof(key_buf), stdin)) {
-								bb_perror_msg_and_die("cannot read input");
+								bb_perror_msg_and_die("can't read input");
 							}
 							i = key_buf[0];
 						}
@@ -368,7 +522,7 @@ int unzip_main(int argc, char **argv)
 			if (verbose) {
 				printf("  inflating: %s\n", dst_fn);
 			}
-			unzip_extract(&zip_header, src_fd, dst_fd);
+			unzip_extract(&zip_header, dst_fd);
 			if (dst_fd != STDOUT_FILENO) {
 				/* closing STDOUT is potentially bad for future business */
 				close(dst_fd);
@@ -379,14 +533,14 @@ int unzip_main(int argc, char **argv)
 			overwrite = O_NEVER;
 		case 'n':
 			/* Skip entry data */
-			unzip_skip(src_fd, zip_header.formatted.cmpsize);
+			unzip_skip(zip_header.formatted.cmpsize);
 			break;
 
 		case 'r':
 			/* Prompt for new name */
 			printf("new name: ");
 			if (!fgets(key_buf, sizeof(key_buf), stdin)) {
-				bb_perror_msg_and_die("cannot read input");
+				bb_perror_msg_and_die("can't read input");
 			}
 			free(dst_fn);
 			dst_fn = xstrdup(key_buf);
@@ -398,12 +552,7 @@ int unzip_main(int argc, char **argv)
 			goto check_file;
 		}
 
-// Looks like bug (data descriptor cannot be identified this way)
-//		/* Data descriptor section */
-//		if (zip_header.formatted.flags & 4) {
-//			/* skip over duplicate crc, compressed size and uncompressed size */
-//			unzip_skip(src_fd, 12);
-//		}
+		total_entries++;
 	}
 
 	if (listing && verbose) {