Doom WAD file specs

Note: all of the long integers used in the Doom WADs are signed 32-bit numbers, in a lo-hi format. So byte values in a file will be need to be transposed before they can be used.

Sample line:

0000000 50 57 41 44 30 02 00 00 58 3c 45 00 f8 09 e8 f8 [PWAD0...X<E.....]

2 byte format: 58 3c 45 00
4 byte format: 00453c58
Final number: 00453c58, the offset to the directory where the lump information is. A lump is a block of data; some contain sound data, some contain graphics data, some contain level structure data, etc.

Doom WAD Identification & Internals
Bytes	Example	Purpose
00-03	50 57 41 44 [ 44415750 -or- 'PWAD' ]	Identifies whether the WAD is a regular WAD (IWAD), or patch WAD (PWAD)
04-07	30 02 00 00 [ 00000230 ]	Number of lumps in the WAD
08-11	58 3c 45 00 [ 00453c58 ]	Byte offset in the WAD file to the start of the Directory

Doom WAD Directory Entry
Bytes	Example	Purpose
00-03	f4 12 00 00 [ 000012f4 ]	Offset from the beginning of the file to the start of the lump
04-07	8a 58 00 00 [ 0000588a ]	Size of the lump in bytes
08-15	4c 49 4e 45 44 45 46 45 [ 454e494c 53464544 -or- 'LINEDEFS' ]	8 byte string, name of the lump

Note that 'marker' and 'label' lump names like "S_START" and "E1M1" will always be zero length, and have zero offset, as they only serve to mark the beginning or end of a set of related lumps.

Handy Commands:
[eightball][osiris 520] od -t x1z -N 48 OSIRIS.WAD
0000000 50 57 41 44 30 02 00 00 58 3c 45 00 f8 09 e8 f8 >PWAD0...X<E.....<
0000020 87 00 01 00 07 00 80 09 80 eb 0e 01 0e 00 07 00 >................<
0000040 50 06 a8 ee 00 00 dc 07 17 00 10 0a 28 ea 00 00 >P...........(...<

outputs the first 48 bytes of a WAD file, in byte groups

[eightball][osiris 522] od -t x4z -N 48 wadfile.wad
0000000 44415750 00000230 00453c58 f8e809f8 >PWAD0...X<E.....<
0000020 00010087 09800007 010eeb80 0007000e >................<
0000040 eea80650 07dc0000 0a100017 0000ea28 >P...........(...<

outputs the first 48 bytes of the wadfile, in word (4 byte) groups

[eightball][osiris 527] od -t x4z -N 48 -j 0x453c58 OSIRIS.WAD
21236130 00000000 00000000 3050414d 00000031 >........MAP01...<
21236150 0000000c 000012e8 4e494854 00005347 >........THINGS..<
21236170 000012f4 0000588a 454e494c 53464544 >.....X..LINEDEFS<

goes to the offset called for in the -j switch, which is the first level in the WAD, display in word groups

GHex is a good GUI editor to use to view the WAD file

Psuedocode for reading the WAD file and extracting the levels the wad replaces

WAD hash

Name: name of the file, minus the .wad extension - separate directory paths from the base path with something that's allowable in a hash key name (underscore?)
full path
levels array

call the subroutine to read the wadfile

input: reference to the wad hash
input: hash key to examine

open the wad file
read the first 12 bytes, from bytes 00-03, determine if it's a pwad or iwad

subroutine that takes a block of 4 bytes, and re-arranges them into an unsigned long (check to see if there's a hex -> decimal perl function that does this)

if it's a iwad, exit
pull the directory offset out of bytes 08-11 for the first lump
pull the name of the map out of bytes 08-15 of the first lump directory entry
add the name of the map to the referenced-in array
cycle thru the rest of the entries for the rest of that lump, looking for the start of the name of the next map (if there's another map in the file)
if EOF, then there's no more maps in that wad, return control to the caller