SCI Specifications: Chapter 2 - Resource files

Resource files

Table of Contents SCI0 resources SCI1 resources Decompression algorithms

with significant contributions from Petr Vyhnak and Vladimir Gneushev

In order to allow games to be both distributeable and playable from several floppy disks, SCI was designed to support multi-volume data. The data itself could therefore be spread into separate files, with some of the more common resources present in more than one of them. The global index for these files was a "resource.map" file, which was read during startup and present on the same disk as the interpreter itself. This file contained a linear lookup table that mapped resource type/number tuples to a set of resource number/ offset tuples, which they could subsequently be read from.

SCI0 resources

SCI0 resource.map

The SCI0 map file format is pretty simple: It consists of 6-byte entries, terminated by the sequence 0xffff ffff ffff. The first 2 bytes, interpreted as little endian 16 bit integer, encode resource type (high 5 bits) and number (low 11 bits). The next 4 bytes are a 32 bit LE integer that contains the resource file number in the high 6 bits, and the absolute offset within the file in the low 26 bits. SCI0 performs a linear search to find the resource; however, multiple entries may match the search, since resources may be present more than once (the inverse mapping is not injective).

SCI0 resource.<nr>

SCI0 resource entries start with a four-tuple of little endian 16 bit words, which we will call (id, comp_size, decomp_size, method). id has the usual SCI0 semantics (high 5 are the resource type, low 11 are its number). comp_size and decomp_size are the size of the compressed and the decompressed resource, respectively. The compressed size actually starts counting at the record position of decomp_size, so it counts four bytes in addition to the actual content. method, finally, is the compression method used to store the data.

SCI1 resources

SCI1 resource.map

The SCI1 resource.map starts with an array of 3-byte structures where the 1st byte is the resource type (0x80 ... 0x91) and next 2 bytes (interpreted as little-endian 16 bit integer) represent the absolute offset of the resource's lookup table (within resource.map). This first array is terminated by a 3-byte entry with has 0xFF as a type and the offset pointing to the first byte after the last resource type's lookup table. SCI1 first goes through this list to find the start of list for the correct resource type and remember this offset and the offset from the next entry to know where it ends. The resulting interval contains a sorted list of 6-byte structures, where the first LE 16 bit integer is the resource number, and the next LE 32 bit integer contains the resource file number in its high 4 bits and the absolute resource offset (in the indicated resource file) in its low 28 bits. Because the list is sorted and its length is known, Sierra SCI can use binary search to locate the resource ID it is looking for.

SCI1 resource.<nr>

Later versions of SCI1 changed the resource file structure slightly: The resource header now contains a byte describing the resource's type, and a four-tuple (res_nr, comp_size, decomp_size, method), where comp_size, decomp_size, and method have the same meanings as before (with the exception of method referring to different algorithms), while res_nr is simply the resource's number.

Decompression algorithms

The decompression algorithms used in SCI are as follows: