How To Read, Parse The Sbm File Format From Superbible Opengl
Solution 1:
Since you are already using NumPy, i'll give yuo an answer which uses NumPy to read the binary file. NumPy has some had functionality to reinterpret buffer data by different data type and that is exactly what is needed to interpret a binary file.
A binary file can be read toe a byte array by numpy.fromfile
.e.g.:
data = numpy.fromfile(filename, dtype=np.byte)
Some bytes of the array can be interpreted as unsigned integer (unit32
) by numpy.frombuffer
. e.g.:
classSB6M_HEADER:
def__init__(self, data):
int_data = np.frombuffer(np.array(data[:16], dtype=np.byte), dtype=np.uint32)
self.magic, self.size, self.num_chunks, self.flags = int_data
print(self.magic, self.size, self.num_chunks, self.flags)
Porting the source code from sb6mfile.h respectively sb7object.cpp to python and reading and "*.sbm" file:
defSB6M_FOURCC(a,b,c,d):
return ( (ord(a) << 0) | (ord(b) << 8) | (ord(c) << 16) | (ord(d) << 24) )
SB6M_MAGIC = SB6M_FOURCC('S','B','6','M')
SB6M_CHUNK_TYPE_INDEX_DATA = SB6M_FOURCC('I','N','D','X')
SB6M_CHUNK_TYPE_VERTEX_DATA = SB6M_FOURCC('V','R','T','X')
SB6M_CHUNK_TYPE_VERTEX_ATTRIBS = SB6M_FOURCC('A','T','R','B')
SB6M_CHUNK_TYPE_SUB_OBJECT_LIST = SB6M_FOURCC('O','L','S','T')
SB6M_CHUNK_TYPE_COMMENT = SB6M_FOURCC('C','M','N','T')
SB6M_CHUNK_TYPE_DATA = SB6M_FOURCC('D','A','T','A')
classSB6M_HEADER:
def__init__(self, data):
int_data = np.frombuffer(np.array(data[:16], dtype=np.byte), dtype=np.uint32)
self.magic, self.size, self.num_chunks, self.flags = int_data
print(self.magic, self.size, self.num_chunks, self.flags)
classSB6M_CHUNK_HEADER:
def__init__(self, data, offset):
int_data = np.frombuffer(np.array(data[offset:offset+8], dtype=np.byte), dtype=np.uint32)
self.type, self.size = int_data
classSB6M_CHUNK_INDEX_DATA(SB6M_CHUNK_HEADER):
def__init__(self, data, offset):
super().__init__(data, offset)
int_data = np.frombuffer(np.array(data[offset+8:offset+20], dtype=np.byte), dtype=np.uint32)
self.index_type, self.index_count, self.index_data_offset = int_data
classSB6M_CHUNK_VERTEX_DATA(SB6M_CHUNK_HEADER):
def__init__(self, data, offset):
super().__init__(data, offset)
int_data = np.frombuffer(np.array(data[offset+8:offset+20], dtype=np.byte), dtype=np.uint32)
self.data_size, self.data_offset, self.total_vertices = int_data
classSB6M_CHUNK_VERTEX_DATA(SB6M_CHUNK_HEADER):
def__init__(self, data, offset):
super().__init__(data, offset)
int_data = np.frombuffer(np.array(data[offset+8:offset+20], dtype=np.byte), dtype=np.uint32)
self.data_size, self.data_offset, self.total_vertices = int_data
SB6M_VERTEX_ATTRIB_FLAG_NORMALIZED = 0x00000001
SB6M_VERTEX_ATTRIB_FLAG_INTEGER = 0x00000002classSB6M_VERTEX_ATTRIB_DECL:
def__init__(self, data, offset):
self.name = ''.join([chr(n) for n in data[offset:offset+64] if n > 30])
int_data = np.frombuffer(np.array(data[offset+64:offset+84], dtype=np.byte), dtype=np.uint32)
self.size, self.type, self.stride, self.flags, self.data_offset = int_data
classSB6M_VERTEX_ATTRIB_CHUNK(SB6M_CHUNK_HEADER):
def__init__(self, data, offset):
super().__init__(data, offset)
int_data = np.frombuffer(np.array(data[offset+8:offset+12], dtype=np.byte), dtype=np.uint32)
self.attrib_count = int_data[0]
self.attrib_data = []
for i inrange(self.attrib_count):
self.attrib_data.append(SB6M_VERTEX_ATTRIB_DECL(data, offset+12+i*84))
classSB6M_DATA_CHUNK(SB6M_CHUNK_HEADER):
def__init__(self, data, offset):
super().__init__(data, offset)
int_data = np.frombuffer(np.array(data[offset+8:offset+20], dtype=np.byte), dtype=np.uint32)
self.encoding, self.data_offset, self.data_length = int_data
classSB6M_SUB_OBJECT_DECL:
def__init__(self, data, offset):
int_data = np.frombuffer(np.array(data[offset:offset+8], dtype=np.byte), dtype=np.uint32)
self.first, self.count = int_data
classSB6M_CHUNK_SUB_OBJECT_LIST(SB6M_CHUNK_HEADER):
def__init__(self, data, offset):
super().__init__(data, offset)
int_data = np.frombuffer(np.array(data[offset+8:offset+12], dtype=np.byte), dtype=np.uint32)
self.count = int_data[0]
self.sub_object = []
for i inrange(self.count):
self.sub_object.append(SB6M_SUB_OBJECT_DECL(data, offset+12+i*8))
defload(filename):
vertex_attrib_chunk = None
vertex_data_chunk = None
index_data_chunk = None
sub_object_chunk = None
data_chunk = Nonetry:
data = numpy.fromfile(filename, dtype=np.byte)
filesize = data.size
header = SB6M_HEADER(data)
offset = header.size
for i inrange(header.num_chunks):
chunk = SB6M_CHUNK_HEADER(data, offset)
if chunk.type == SB6M_CHUNK_TYPE_VERTEX_ATTRIBS:
vertex_attrib_chunk = SB6M_VERTEX_ATTRIB_CHUNK(data, offset)
elif chunk.type == SB6M_CHUNK_TYPE_VERTEX_DATA:
vertex_data_chunk = SB6M_CHUNK_VERTEX_DATA(data, offset)
elif chunk.type == SB6M_CHUNK_TYPE_INDEX_DATA:
index_data_chunk = SB6M_CHUNK_INDEX_DATA(data, offset)
elif chunk.type == SB6M_CHUNK_TYPE_SUB_OBJECT_LIST:
sub_object_chunk = SB6M_CHUNK_SUB_OBJECT_LIST(data, offset)
elif chunk.type == SB6M_CHUNK_TYPE_DATA:
data_chunk = SB6M_DATA_CHUNK(data, offset)
else:
raise
offset += chunk.size
except:
print("error reading file {}".format(filename))
Finally the floating point vertex data can be read:
if vertex_data_chunk and vertex_attrib_chunk:
start = vertex_data_chunk.data_offset
end = start + vertex_data_chunk.data_size
vertex_data = np.frombuffer(np.array(data[start:end], dtype=np.byte), dtype=np.float)
data_buffer = glGenBuffers(1)
glBindBuffer(GL_ARRAY_BUFFER, data_buffer)
glBufferData(GL_ARRAY_BUFFER, vertex_data, GL_STATIC_DRAW)
vertexcount = vertex_data_chunk.total_vertices
vao = glGenVertexArrays(1)
glBindVertexArray(self.vao)
for attrib_i, attrib inenumerate(vertex_attrib_chunk.attrib_data):
if attrib.name=='position'or attrib.name=='map1':
glVertexAttribPointer(attrib_i,
attrib.size, attrib.type,
GL_TRUE if (attrib.flags & SB6M_VERTEX_ATTRIB_FLAG_NORMALIZED) != 0else GL_FALSE,
attrib.stride, ctypes.c_void_p(int(attrib.data_offset)))
glEnableVertexAttribArray(attrib_i)
Finally draw the mesh:
glBindVertexArray(vao)
glDrawArrays(GL_TRIANGLES, 0, vertexcount)
Solution 2:
The next step has to be similar to what happens in the C code:
SB6M_HEADER * header = (SB6M_HEADER *)ptr;
ptr += header->size;
You need to advance the pointer by a known size.
You have that attribute in your header class. Do you set it correctly?
But I would suggest a different approach: do not use raw Pyhon for this at all.
Instead create a wrapper using the original C code found in sb6mfile.h
and a part of the function to read the file found in sb7object.cpp
Then access this code using Python's CFFI. You pass the C header with the SBM types to the files and can use them in Python.
Once this works you can get the data back and put it in a numpy array. Usually this is a bit of work but once it works it is really robust.
Here are some links to get you started:
buffer_size = np_arr.size*np_arr.dtype.itemsize
c_buffer = ffi.buffer(cffi_arr,buffer_size)
np_arr2 = np.frombuffer(c_buffer, dtype=np_arr.dtype)
(https://ammous88.wordpress.com/2014/12/30/numpy-array-with-cffi-c-function/)
Post a Comment for "How To Read, Parse The Sbm File Format From Superbible Opengl"