static void EBMLdate_to_timestamp(i64 ed, struct de_timestamp *ts)
{
i64 t;
// ed is the number of nanoseconds since the beginning of 2001.
t = ed/1000000000;
// Now t is seconds since the beginning of 2001.
// We want seconds since the beginning of 1970.
// So, add the number of seconds in the years from 1970 through 2000. This
// is 31 years.
// There are 86400 seconds in a day.
// There are 8 leap days in this range ('72, 76, 80, 84, 88, 92, 96, 00).
t += 86400LL * (31*365 + 8);
de_unix_time_to_timestamp(t, ts, 0x1);
de_timestamp_set_subsec(ts, ((double)(ed%1000000000))/1000000000.0);
}
static int do_gzip_read_member(deark *c, lctx *d, i64 pos1, i64 *member_size)
{
u8 b0, b1;
i64 pos;
i64 n;
i64 foundpos;
i64 string_len;
i64 cmpr_data_len;
i64 mod_time_unix;
u32 crc_calculated;
de_ucstring *member_name = NULL;
int saved_indent_level;
int ret;
struct member_data *md = NULL;
int retval = 0;
md = de_malloc(c, sizeof(struct member_data));
de_dbg_indent_save(c, &saved_indent_level);
de_dbg(c, "gzip member at %d", (int)pos1);
de_dbg_indent(c, 1);
pos = pos1;
b0 = de_getbyte(pos+0);
b1 = de_getbyte(pos+1);
if(b0!=0x1f || b1!=0x8b) {
de_err(c, "Invalid gzip signature at %d. This is not a valid gzip file.",
(int)pos1);
goto done;
}
md->cmpr_code = de_getbyte(pos+2);
if(md->cmpr_code!=0x08) {
de_err(c, "Unsupported compression type (%d)", (int)md->cmpr_code);
goto done;
}
md->flags = de_getbyte(pos+3);
de_dbg(c, "flags: 0x%02x", (unsigned int)md->flags);
pos += 4;
mod_time_unix = de_getu32le(pos);
de_unix_time_to_timestamp(mod_time_unix, &md->mod_time_ts, 0x1);
if(md->mod_time_ts.is_valid) {
char timestamp_buf[64];
de_timestamp_to_string(&md->mod_time_ts, timestamp_buf, sizeof(timestamp_buf), 0);
de_dbg(c, "mod time: %" I64_FMT " (%s)", mod_time_unix, timestamp_buf);
}
pos += 4;
b0 = de_getbyte(pos++);
de_dbg(c, "extra flags: 0x%02x", (unsigned int)b0);
b0 = de_getbyte(pos++);
de_dbg(c, "OS or filesystem: %d (%s)", (int)b0, get_os_name(b0));
if(md->flags & GZIPFLAG_FEXTRA) {
n = de_getu16le(pos); // XLEN
// TODO: It might be interesting to dissect these extra fields, but it's
// hard to find even a single file that uses them.
de_dbg(c, "[extra fields at %d, dpos=%d, dlen=%d]",
(int)pos, (int)(pos+2), (int)n);
pos += 2;
pos += n;
}
if(md->flags & GZIPFLAG_FNAME) {
ret = dbuf_search_byte(c->infile, 0x00, pos, c->infile->len - pos,
&foundpos);
if(!ret) {
de_err(c, "Invalid NAME field");
goto done;
}
string_len = foundpos - pos;
member_name = ucstring_create(c);
#define DE_GZIP_MAX_FNLEN 300
dbuf_read_to_ucstring_n(c->infile, pos, string_len, DE_GZIP_MAX_FNLEN,
member_name, 0, DE_ENCODING_LATIN1);
de_dbg(c, "file name at %d, len=%d: \"%s\"", (int)pos, (int)string_len,
ucstring_getpsz_d(member_name));
pos = foundpos + 1;
}
if(md->flags & GZIPFLAG_FCOMMENT) {
ret = dbuf_search_byte(c->infile, 0x00, pos, c->infile->len - pos,
&foundpos);
if(!ret) {
de_err(c, "Invalid COMMENT field");
goto done;
}
pos = foundpos + 1;
}
if(md->flags & GZIPFLAG_FHCRC) {
md->crc16_reported = (u32)de_getu16le(pos);
de_dbg(c, "crc16 (reported): 0x%04x", (unsigned int)md->crc16_reported);
pos += 2;
}
de_dbg(c, "compressed blocks at %d", (int)pos);
if(!d->output_file) {
// Although any member can have a name and mod time, this metadata
// is ignored for members after the first one.
de_finfo *fi = NULL;
fi = de_finfo_create(c);
if(member_name && c->filenames_from_file) {
de_finfo_set_name_from_ucstring(c, fi, member_name, 0);
fi->original_filename_flag = 1;
}
if(md->mod_time_ts.is_valid) {
fi->mod_time = md->mod_time_ts;
}
d->output_file = dbuf_create_output_file(c, member_name?NULL:"bin", fi, 0);
de_finfo_destroy(c, fi);
}
d->output_file->writecallback_fn = our_writecallback;
d->output_file->userdata = (void*)md;
md->crco = d->crco;
de_crcobj_reset(md->crco);
ret = de_decompress_deflate(c->infile, pos, c->infile->len - pos, d->output_file,
0, &cmpr_data_len, 0);
crc_calculated = de_crcobj_getval(md->crco);
d->output_file->writecallback_fn = NULL;
d->output_file->userdata = NULL;
if(!ret) goto done;
pos += cmpr_data_len;
de_dbg(c, "crc32 (calculated): 0x%08x", (unsigned int)crc_calculated);
md->crc32_reported = (u32)de_getu32le(pos);
de_dbg(c, "crc32 (reported) : 0x%08x", (unsigned int)md->crc32_reported);
pos += 4;
if(crc_calculated != md->crc32_reported) {
de_warn(c, "CRC check failed: Expected 0x%08x, got 0x%08x",
(unsigned int)md->crc32_reported, (unsigned int)crc_calculated);
}
md->isize = de_getu32le(pos);
de_dbg(c, "uncompressed size (mod 2^32): %u", (unsigned int)md->isize);
pos += 4;
retval = 1;
done:
if(retval)
*member_size = pos - pos1;
else
*member_size = 0;
ucstring_destroy(member_name);
de_free(c, md);
de_dbg_indent_restore(c, saved_indent_level);
return retval;
}
static int do_gzip_read_member(deark *c, lctx *d, de_int64 pos1, de_int64 *member_size)
{
de_byte b0, b1;
de_int64 cmpr_code;
de_int64 pos;
de_int64 n;
de_int64 foundpos;
de_int64 string_len;
de_int64 cmpr_data_len;
de_int64 isize;
de_int64 mod_time_unix;
struct de_timestamp mod_time_ts;
de_uint32 crc32_field;
de_ucstring *member_name = NULL;
de_finfo *fi = NULL;
int saved_indent_level;
int ret;
int retval = 0;
mod_time_ts.is_valid = 0;
de_dbg_indent_save(c, &saved_indent_level);
de_dbg(c, "gzip member at %d\n", (int)pos1);
de_dbg_indent(c, 1);
pos = pos1;
b0 = de_getbyte(pos+0);
b1 = de_getbyte(pos+1);
if(b0!=0x1f || b1!=0x8b) {
de_err(c, "Invalid gzip signature at %d. This is not a valid gzip file.\n",
(int)pos1);
goto done;
}
cmpr_code=de_getbyte(pos+2);
if(cmpr_code!=0x08) {
de_err(c, "Unsupported compression type (%d)\n", (int)cmpr_code);
goto done;
}
d->flags = de_getbyte(pos+3);
de_dbg(c, "flags: 0x%02x\n", (unsigned int)d->flags);
pos += 4;
mod_time_unix = de_getui32le(pos);
de_unix_time_to_timestamp(mod_time_unix, &mod_time_ts);
if(mod_time_ts.is_valid) {
char timestamp_buf[64];
de_timestamp_to_string(&mod_time_ts, timestamp_buf, sizeof(timestamp_buf), 1);
de_dbg(c, "mod time: %" INT64_FMT " (%s)\n", mod_time_unix, timestamp_buf);
}
pos += 4;
b0 = de_getbyte(pos++);
de_dbg(c, "extra flags: 0x%02x\n", (unsigned int)b0);
b0 = de_getbyte(pos++);
de_dbg(c, "OS or filesystem: %d (%s)\n", (int)b0, get_os_name(b0));
if(d->flags & GZIPFLAG_FEXTRA) {
n = de_getui16le(pos); // XLEN
// TODO: It might be interesting to dissect these extra fields, but it's
// hard to find even a single file that uses them.
de_dbg(c, "[extra fields at %d, dpos=%d, dlen=%d]\n",
(int)pos, (int)(pos+2), (int)n);
pos += 2;
pos += n;
}
if(d->flags & GZIPFLAG_FNAME) {
ret = dbuf_search_byte(c->infile, 0x00, pos, c->infile->len - pos,
&foundpos);
if(!ret) {
de_err(c, "Invalid NAME field\n");
goto done;
}
string_len = foundpos - pos;
member_name = ucstring_create(c);
dbuf_read_to_ucstring_n(c->infile, pos, string_len, 300, member_name, 0, DE_ENCODING_LATIN1);
de_dbg(c, "file name at %d, len=%d: \"%s\"\n", (int)pos, (int)string_len,
ucstring_get_printable_sz(member_name));
pos = foundpos + 1;
}
if(d->flags & GZIPFLAG_FCOMMENT) {
ret = dbuf_search_byte(c->infile, 0x00, pos, c->infile->len - pos,
&foundpos);
if(!ret) {
de_err(c, "Invalid COMMENT field\n");
goto done;
}
pos = foundpos + 1;
}
if(d->flags & GZIPFLAG_FHCRC) {
pos += 2;
}
de_dbg(c, "compressed blocks at %d\n", (int)pos);
if(!d->output_file) {
fi = de_finfo_create(c);
if(member_name && c->filenames_from_file) {
de_finfo_set_name_from_ucstring(c, fi, member_name);
fi->original_filename_flag = 1;
}
if(mod_time_ts.is_valid) {
fi->mod_time = mod_time_ts;
}
d->output_file = dbuf_create_output_file(c, member_name?NULL:"bin", fi, 0);
}
ret = de_uncompress_deflate(c->infile, pos, c->infile->len - pos, d->output_file, &cmpr_data_len);
if(!ret) goto done;
pos += cmpr_data_len;
crc32_field = (de_uint32)de_getui32le(pos);
de_dbg(c, "crc32: 0x%08x\n", (unsigned int)crc32_field);
pos += 4;
// TODO: Validate CRCs
isize = de_getui32le(pos);
de_dbg(c, "uncompressed size (mod 2^32): %u\n", (unsigned int)isize);
pos += 4;
retval = 1;
done:
if(retval)
*member_size = pos - pos1;
else
*member_size = 0;
ucstring_destroy(member_name);
de_finfo_destroy(c, fi);
de_dbg_indent_restore(c, saved_indent_level);
return retval;
}
