Wwise SoundBank(.bnk) and SoundPack(.wpk) metadata reader (JSON output)

read_bnk.py

#!/usr/bin/python3
import json
import sys
import base64

# TODO: CLEANUP (maybe, what an ugly thing)

uint32_size = 4
sectionid_size = 4

filename = str(sys.argv[1])

sbnk = []

file = open(filename, 'rb')

# Is `file` in EOF
def feof(file=file):
	if file.read(1): # Can I read a byte?
		file.seek(-1,1)
		return False  # YES, we're not in the end
	return True       # NO, this is the end

def get_bkhd(bytes):
	# ui32(version)ui32(id)ui32(reserved0)ui32(reserved1)
	return {
	  "version": get_int(bytes[0:4]),
	  "id": get_int(bytes[4:8]),
	  "reserved0": get_int(bytes[8:12]),
	  "reserved1": get_int(bytes[12:16]),
	}

def get_hirc(bytes):
	# ui32(object_count)var(object_array)
	# object_array: 1b(type)ui32(len)ui32(id)[len-ui32](data)
	count = get_int(bytes[0:4])
	bytes = bytes[4:]
	objects = []
	for index in range(count):
		type = get_int(bytes[0:1])
		length = get_int(bytes[1:5]) - uint32_size
		id = get_int(bytes[5:9])
		data = bytes[9:9+length]
		bytes = bytes[9+length:]
		objects.append({
			"type": get_hirc_type(type),
			"length":  length,
			"id": id,
			"data": get_data(data, "hirc:"+get_hirc_type(type)),
		})
	return {"count":count, "objects": objects}

def get_hirc_type(type):
	return {
		0x01: "Settings",
		0x02: "Sound",
		0x03: "Event Action",
		0x04: "Event",
		0x05: "Sequence Container",
		0x06: "Switch Container",
		0x07: "Actor Mixer",
		0x08: "Audio Bus",
		0x09: "Blend Container",
		0x0a: "Music Segment",
		0x0b: "Music Track",
		0x0c: "Music Switch Container",
		0x0d: "Music Playlist Container",
		0x0e: "Attenuation",
		0x0f: "Dialogue Event",
		0x10: "Motion Bus",
		0x11: "Motion FX",
		0x12: "Effect",
		0x13: "Auxiliary Bus",
	}.get(type, "unknown("+str(type)+")")

def get_hirc_sound(bytes):
	sfx={}
	sfx["reserved0"]=get_base64(bytes[0:4]) # UNKNOWN
	# Stream Source (Embedded, Streamed, Streamed+ZeroLatency/Prefetched)
	strmsrc = get_int(bytes[4:8])
	if strmsrc > 0:
		strm = "Streamed"
		if strmsrc == 2:
			strm += "(ZeroLatency)"
	else:
		strm = "Embedded"
	sfx["streamSourceType"] = strm
	# File ID
	sfx["streamFileID"] = get_int(bytes[8:12])
	# Source ID
	sfx["streamSourceID"] = get_int(bytes[12:16])
	# BEGIN Embedded
	if strm == 0:
		sfx["streamFileOffset"] = get_int(bytes[16:20])
		sfx["streamFileLength"] = get_int(bytes[20:24])
	# END Embedded
	sfx["soundType"] = "Voice" if bytes[24:25] else "SFX"
	sfx["soundStructure"] = get_base64(bytes[25:])
	return sfx

def get_base64(bytes):
	return "base64:"+str(base64.b64encode(bytes), 'ascii')

def get_int(bytes):
	return int.from_bytes(bytes, 'little') if type(bytes) is not int else bytes

def get_data(bytes, sectionid):
	return {
		'BKHD': get_bkhd,
		'HIRC': get_hirc,
		'hirc:Sound': get_hirc_sound,
	}.get(sectionid, get_base64)(bytes)

eof = feof()
while not eof:
	# .BNK File structure: 4b(ID)ui32(LENGTH)LENGTH(DATA){repeat}EOF
	section = {}
	section["id"] = str(file.read(sectionid_size), 'ascii')
	section["len"] = get_int(file.read(uint32_size))
	section["data"] = get_data(file.read(section["len"]), section["id"])
	sbnk.append(section)
	eof = feof()

file.close()

print(json.dumps(sbnk, indent=2))

read_wpk.py

#!/usr/bin/python3
import json
import sys
import base64


filename = str(sys.argv[1])
file = open(filename, 'rb')

def get_base64(bytes):
	return "base64:"+str(base64.b64encode(bytes), 'ascii')

def get_int(bytes):
	return int.from_bytes(bytes, 'little') if type(bytes) is not int else bytes

def get_strt(file):
	# BEGIN HEADER
	wpk = {}
	wpk["Magic"] = str(file.read(4), 'ascii')
	wpk["Version"] = get_int(file.read(4))
	wpk["Count"] = get_int(file.read(4))
	wpk["Offsets"] = []
	for index in range(wpk["Count"]):
		wpk["Offsets"].append(get_int(file.read(4)))
	# END HEADER
	wpk["Files"] = []
	for offset in wpk["Offsets"]:
		wem = {}
		file.seek(offset)
		wem["DataOffset"] = get_int(file.read(4))
		wem["DataSize"] = get_int(file.read(4))
		wem["NameSize"] = get_int(file.read(4))
		wem["Name"] = str(file.read(wem["NameSize"]*2), 'utf16')
		wpk["Files"].append(wem)
	return wpk
print(json.dumps(get_strt(file), indent=2))
file.close()