/*
* The extra data is stored as a list of
* id1+size1+data1 + id2+size2+data2 ...
* triplets. id and size are 2 bytes each.
*/
static void
process_extra(const void* extra, struct zip* zip)
{
int offset = 0;
const char *p = (const char *)extra;
while (offset < zip->extra_length - 4)
{
unsigned short headerid = archive_le16dec(p + offset);
unsigned short datasize = archive_le16dec(p + offset + 2);
offset += 4;
if (offset + datasize > zip->extra_length)
break;
#ifdef DEBUG
fprintf(stderr, "Header id 0x%04x, length %d\n",
headerid, datasize);
#endif
switch (headerid) {
case 0x0001:
/* Zip64 extended information extra field. */
if (datasize >= 8)
zip->uncompressed_size = archive_le64dec(p + offset);
if (datasize >= 16)
zip->compressed_size = archive_le64dec(p + offset + 8);
break;
case 0x5455:
{
/* Extended time field "UT". */
int flags = p[offset];
offset++;
datasize--;
/* Flag bits indicate which dates are present. */
if (flags & 0x01)
{
#ifdef DEBUG
fprintf(stderr, "mtime: %lld -> %d\n",
(long long)zip->mtime,
archive_le32dec(p + offset));
#endif
if (datasize < 4)
break;
zip->mtime = archive_le32dec(p + offset);
offset += 4;
datasize -= 4;
}
if (flags & 0x02)
{
if (datasize < 4)
break;
zip->atime = archive_le32dec(p + offset);
offset += 4;
datasize -= 4;
}
if (flags & 0x04)
{
if (datasize < 4)
break;
zip->ctime = archive_le32dec(p + offset);
offset += 4;
datasize -= 4;
}
break;
}
case 0x7855:
/* Info-ZIP Unix Extra Field (type 2) "Ux". */
#ifdef DEBUG
fprintf(stderr, "uid %d gid %d\n",
archive_le16dec(p + offset),
archive_le16dec(p + offset + 2));
#endif
if (datasize >= 2)
zip->uid = archive_le16dec(p + offset);
if (datasize >= 4)
zip->gid = archive_le16dec(p + offset + 2);
break;
default:
break;
}
offset += datasize;
}
#ifdef DEBUG
if (offset != zip->extra_length)
{
fprintf(stderr,
"Extra data field contents do not match reported size!");
}
#endif
}
int
zip_read_file_header(struct archive_read *a, struct archive_entry *entry,
struct zip *zip)
{
const struct zip_file_header *p;
const void *h;
if ((p = __archive_read_ahead(a, sizeof *p)) == NULL) {
archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
"Truncated ZIP file header");
return (ARCHIVE_FATAL);
}
zip->version = p->version[0];
zip->system = p->version[1];
zip->flags = archive_le16dec(p->flags);
zip->compression = archive_le16dec(p->compression);
if (zip->compression <
sizeof(compression_names)/sizeof(compression_names[0]))
zip->compression_name = compression_names[zip->compression];
else
zip->compression_name = "??";
zip->mtime = zip_time(p->timedate);
zip->ctime = 0;
zip->atime = 0;
zip->mode = 0;
zip->uid = 0;
zip->gid = 0;
zip->crc32 = archive_le32dec(p->crc32);
zip->filename_length = archive_le16dec(p->filename_length);
zip->extra_length = archive_le16dec(p->extra_length);
zip->uncompressed_size = archive_le32dec(p->uncompressed_size);
zip->compressed_size = archive_le32dec(p->compressed_size);
(a->decompressor->consume)(a, sizeof(struct zip_file_header));
/* Read the filename. */
if ((h = __archive_read_ahead(a, zip->filename_length)) == NULL) {
archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
"Truncated ZIP file header");
return (ARCHIVE_FATAL);
}
if (archive_string_ensure(&zip->pathname, zip->filename_length) == NULL)
__archive_errx(1, "Out of memory");
archive_strncpy(&zip->pathname, h, zip->filename_length);
(a->decompressor->consume)(a, zip->filename_length);
archive_entry_set_pathname(entry, zip->pathname.s);
if (zip->pathname.s[archive_strlen(&zip->pathname) - 1] == '/')
zip->mode = AE_IFDIR | 0777;
else
zip->mode = AE_IFREG | 0777;
/* Read the extra data. */
if ((h = __archive_read_ahead(a, zip->extra_length)) == NULL) {
archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
"Truncated ZIP file header");
return (ARCHIVE_FATAL);
}
process_extra(h, zip);
(a->decompressor->consume)(a, zip->extra_length);
/* Populate some additional entry fields: */
archive_entry_set_mode(entry, zip->mode);
archive_entry_set_uid(entry, zip->uid);
archive_entry_set_gid(entry, zip->gid);
archive_entry_set_mtime(entry, zip->mtime, 0);
archive_entry_set_ctime(entry, zip->ctime, 0);
archive_entry_set_atime(entry, zip->atime, 0);
archive_entry_set_size(entry, zip->uncompressed_size);
zip->entry_bytes_remaining = zip->compressed_size;
zip->entry_offset = 0;
/* If there's no body, force read_data() to return EOF immediately. */
if (0 == (zip->flags & ZIP_LENGTH_AT_END)
&& zip->entry_bytes_remaining < 1)
zip->end_of_entry = 1;
/* Set up a more descriptive format name. */
sprintf(zip->format_name, "ZIP %d.%d (%s)",
zip->version / 10, zip->version % 10,
zip->compression_name);
a->archive.archive_format_name = zip->format_name;
return (ARCHIVE_OK);
}
/*
* TODO: This is a performance sink because it forces the read core to
* drop buffered data from the start of file, which will then have to
* be re-read again if this bidder loses.
*
* We workaround this a little by passing in the best bid so far so
* that later bidders can do nothing if they know they'll never
* outbid. But we can certainly do better...
*/
static int
archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid)
{
struct zip *zip = (struct zip *)a->format->data;
int64_t filesize;
const char *p;
/* If someone has already bid more than 32, then avoid
trashing the look-ahead buffers with a seek. */
if (best_bid > 32)
return (-1);
filesize = __archive_read_seek(a, -22, SEEK_END);
/* If we can't seek, then we can't bid. */
if (filesize <= 0)
return 0;
/* TODO: More robust search for end of central directory record. */
if ((p = __archive_read_ahead(a, 22, NULL)) == NULL)
return 0;
/* First four bytes are signature for end of central directory
record. Four zero bytes ensure this isn't a multi-volume
Zip file (which we don't yet support). */
if (memcmp(p, "PK\005\006\000\000\000\000", 8) != 0) {
int64_t i, tail;
int found;
/*
* If there is a comment in end of central directory
* record, 22 bytes are too short. we have to read more
* to properly detect the record. Hopefully, a length
* of the comment is not longer than 16362 bytes(16K-22).
*/
if (filesize + 22 > 1024 * 16) {
tail = 1024 * 16;
filesize = __archive_read_seek(a, tail * -1, SEEK_END);
} else {
tail = filesize + 22;
filesize = __archive_read_seek(a, 0, SEEK_SET);
}
if (filesize < 0)
return 0;
if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL)
return 0;
for (found = 0, i = 0;!found && i < tail - 22;) {
switch (p[i]) {
case 'P':
if (memcmp(p+i,
"PK\005\006\000\000\000\000", 8) == 0) {
p += i;
filesize += tail -
(22 + archive_le16dec(p+20));
found = 1;
} else
i += 8;
break;
case 'K': i += 7; break;
case 005: i += 6; break;
case 006: i += 5; break;
default: i += 1; break;
}
}
if (!found)
return 0;
}
/* Since we've already done the hard work of finding the
end of central directory record, let's save the important
information. */
zip->central_directory_entries = archive_le16dec(p + 10);
zip->central_directory_size = archive_le32dec(p + 12);
zip->central_directory_offset = archive_le32dec(p + 16);
zip->end_of_central_directory_offset = filesize;
/* Just one volume, so central dir must all be on this volume. */
if (zip->central_directory_entries != archive_le16dec(p + 8))
return 0;
/* Central directory can't extend beyond end of this file. */
if (zip->central_directory_offset +
(int64_t)zip->central_directory_size > filesize)
return 0;
/* This is just a tiny bit higher than the maximum returned by
the streaming Zip bidder. This ensures that the more accurate
seeking Zip parser wins whenever seek is available. */
return 32;
}
static int
slurp_central_directory(struct archive_read *a, struct zip *zip)
{
unsigned i;
int64_t correction;
static const struct archive_rb_tree_ops rb_ops = {
&cmp_node, &cmp_key
};
static const struct archive_rb_tree_ops rb_rsrc_ops = {
&rsrc_cmp_node, &rsrc_cmp_key
};
/*
* Consider the archive file we are reading may be SFX.
* So we have to calculate a SFX header size to revise
* ZIP header offsets.
*/
correction = zip->end_of_central_directory_offset -
(zip->central_directory_offset + zip->central_directory_size);
/* The central directory offset is relative value, and so
* we revise this offset for SFX. */
zip->central_directory_offset += correction;
__archive_read_seek(a, zip->central_directory_offset, SEEK_SET);
zip->offset = zip->central_directory_offset;
__archive_rb_tree_init(&zip->tree, &rb_ops);
__archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops);
zip->zip_entries = calloc(zip->central_directory_entries,
sizeof(struct zip_entry));
for (i = 0; i < zip->central_directory_entries; ++i) {
struct zip_entry *zip_entry = &zip->zip_entries[i];
size_t filename_length, extra_length, comment_length;
uint32_t external_attributes;
const char *name, *p, *r;
if ((p = __archive_read_ahead(a, 46, NULL)) == NULL)
return ARCHIVE_FATAL;
if (memcmp(p, "PK\001\002", 4) != 0) {
archive_set_error(&a->archive,
-1, "Invalid central directory signature");
return ARCHIVE_FATAL;
}
zip->have_central_directory = 1;
/* version = p[4]; */
zip_entry->system = p[5];
/* version_required = archive_le16dec(p + 6); */
zip_entry->flags = archive_le16dec(p + 8);
zip_entry->compression = (char)archive_le16dec(p + 10);
zip_entry->mtime = zip_time(p + 12);
zip_entry->crc32 = archive_le32dec(p + 16);
zip_entry->compressed_size = archive_le32dec(p + 20);
zip_entry->uncompressed_size = archive_le32dec(p + 24);
filename_length = archive_le16dec(p + 28);
extra_length = archive_le16dec(p + 30);
comment_length = archive_le16dec(p + 32);
/* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */
/* internal_attributes = archive_le16dec(p + 36); */ /* text bit */
external_attributes = archive_le32dec(p + 38);
zip_entry->local_header_offset =
archive_le32dec(p + 42) + correction;
/* If we can't guess the mode, leave it zero here;
when we read the local file header we might get
more information. */
zip_entry->mode = 0;
if (zip_entry->system == 3) {
zip_entry->mode = external_attributes >> 16;
}
/*
* Mac resource fork files are stored under the
* "__MACOSX/" directory, so we should check if
* it is.
*/
/* Make sure we have the file name. */
if ((p = __archive_read_ahead(a, 46 + filename_length, NULL))
== NULL)
return ARCHIVE_FATAL;
name = p + 46;
r = rsrc_basename(name, filename_length);
if (filename_length >= 9 &&
strncmp("__MACOSX/", name, 9) == 0) {
/* If this file is not a resource fork nor
* a directory. We should treat it as a non
* resource fork file to expose it. */
if (name[filename_length-1] != '/' &&
(r - name < 3 || r[0] != '.' || r[1] != '_')) {
__archive_rb_tree_insert_node(&zip->tree,
&zip_entry->node);
/* Expose its parent directories. */
expose_parent_dirs(zip, name, filename_length);
} else {
/* This file is a resource fork file or
* a directory. */
archive_strncpy(&(zip_entry->rsrcname), name,
filename_length);
__archive_rb_tree_insert_node(&zip->tree_rsrc,
&zip_entry->node);
}
} else {
/* Generate resource fork name to find its resource
* file at zip->tree_rsrc. */
archive_strcpy(&(zip_entry->rsrcname), "__MACOSX/");
archive_strncat(&(zip_entry->rsrcname), name, r - name);
archive_strcat(&(zip_entry->rsrcname), "._");
archive_strncat(&(zip_entry->rsrcname),
name + (r - name), filename_length - (r - name));
/* Register an entry to RB tree to sort it by
* file offset. */
__archive_rb_tree_insert_node(&zip->tree,
&zip_entry->node);
}
/* We don't read the filename until we get to the
local file header. Reading it here would speed up
table-of-contents operations (removing the need to
find and read local file header to get the
filename) at the cost of requiring a lot of extra
space. */
/* We don't read the extra block here. We assume it
will be duplicated at the local file header. */
__archive_read_consume(a,
46 + filename_length + extra_length + comment_length);
}
static int
archive_write_zip_finish_entry(struct archive_write *a)
{
struct zip *zip = a->format_data;
int ret;
#if HAVE_ZLIB_H
if (zip->entry_compression == COMPRESSION_DEFLATE) {
for (;;) {
size_t remainder;
ret = deflate(&zip->stream, Z_FINISH);
if (ret == Z_STREAM_ERROR)
return (ARCHIVE_FATAL);
remainder = zip->len_buf - zip->stream.avail_out;
ret = __archive_write_output(a, zip->buf, remainder);
if (ret != ARCHIVE_OK)
return (ret);
zip->entry_compressed_written += remainder;
zip->written_bytes += remainder;
zip->stream.next_out = zip->buf;
if (zip->stream.avail_out != 0)
break;
zip->stream.avail_out = (uInt)zip->len_buf;
}
deflateEnd(&zip->stream);
}
#endif
/* Write trailing data descriptor. */
if ((zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) != 0) {
char d[24];
memcpy(d, "PK\007\010", 4);
archive_le32enc(d + 4, zip->entry_crc32);
if (zip->entry_uses_zip64) {
archive_le64enc(d + 8, (uint64_t)zip->entry_compressed_written);
archive_le64enc(d + 16, (uint64_t)zip->entry_uncompressed_written);
ret = __archive_write_output(a, d, 24);
zip->written_bytes += 24;
} else {
archive_le32enc(d + 8, (uint32_t)zip->entry_compressed_written);
archive_le32enc(d + 12, (uint32_t)zip->entry_uncompressed_written);
ret = __archive_write_output(a, d, 16);
zip->written_bytes += 16;
}
if (ret != ARCHIVE_OK)
return (ARCHIVE_FATAL);
}
/* Append Zip64 extra data to central directory information. */
if (zip->entry_compressed_written > ARCHIVE_LITERAL_LL(0xffffffff)
|| zip->entry_uncompressed_written > ARCHIVE_LITERAL_LL(0xffffffff)
|| zip->entry_offset > ARCHIVE_LITERAL_LL(0xffffffff)) {
unsigned char zip64[32];
unsigned char *z = zip64, *zd;
memcpy(z, "\001\000\000\000", 4);
z += 4;
if (zip->entry_uncompressed_written >= ARCHIVE_LITERAL_LL(0xffffffff)) {
archive_le64enc(z, zip->entry_uncompressed_written);
z += 8;
}
if (zip->entry_compressed_written >= ARCHIVE_LITERAL_LL(0xffffffff)) {
archive_le64enc(z, zip->entry_compressed_written);
z += 8;
}
if (zip->entry_offset >= ARCHIVE_LITERAL_LL(0xffffffff)) {
archive_le64enc(z, zip->entry_offset);
z += 8;
}
archive_le16enc(zip64 + 2, z - (zip64 + 4));
zd = cd_alloc(zip, z - zip64);
if (zd == NULL) {
archive_set_error(&a->archive, ENOMEM,
"Can't allocate zip data");
return (ARCHIVE_FATAL);
}
memcpy(zd, zip64, z - zip64);
/* Zip64 means version needs to be set to at least 4.5 */
if (archive_le16dec(zip->file_header + 6) < 45)
archive_le16enc(zip->file_header + 6, 45);
}
/* Fix up central directory file header. */
archive_le32enc(zip->file_header + 16, zip->entry_crc32);
archive_le32enc(zip->file_header + 20,
zipmin(zip->entry_compressed_written, ARCHIVE_LITERAL_LL(0xffffffff)));
archive_le32enc(zip->file_header + 24,
zipmin(zip->entry_uncompressed_written, ARCHIVE_LITERAL_LL(0xffffffff)));
archive_le16enc(zip->file_header + 30,
zip->central_directory_bytes - zip->file_header_extra_offset);
archive_le32enc(zip->file_header + 42,
zipmin(zip->entry_offset, ARCHIVE_LITERAL_LL(0xffffffff)));
return (ARCHIVE_OK);
}
