AstralTransRocketries/mcedit2
Archived
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Store uncooked models with model variants separate from blockstates

Browse Source 
Uncooked models are stored in quadsByResourcePathVariant {(path, variant): quads}
blockState mapping is stored in blockStatesByResourcePathVariant {(path, variant): [nameAndState]}

This allows us to skip cooking models twice when one model (path, variant) is used for more than one blockState
This commit is contained in:
David Vierra 2015-07-15 14:57:22 -10:00
parent 79963a55a9
commit e65817bb60
2 changed files with 47 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
src/mcedit2/rendering/blockmodels.pxd
View File

@ -19,7 +19,8 @@ cdef class BlockModels:
cdef object blocktypes
cdef object modelBlockJsons
cdef object modelStateJsons
cdef object modelQuads
cdef object quadsByResourcePathVariant
cdef object blockStatesByResourcePathVariant
cdef object _texturePaths
cdef public object firstTextures
cdef object cookedModels

59
src/mcedit2/rendering/blockmodels.pyx
View File

@ -3,6 +3,7 @@
blockmodels
"""
from __future__ import absolute_import, print_function
from collections import defaultdict
import json
import logging
import math
@ -80,7 +81,28 @@ UNKNOWN_MODEL = {
u"all": u"MCEDIT_UNKNOWN"
}
}
"""
Some blocks have blockStates not represented in the world file.
These will have additional blockStates with corresponding resourceVariants that do not
have a unique ID/meta combination.
To make these additional block states renderable, we will need to have a function
`getActualBlockState` to get the state, and another function to get the resourceVariant
for that state. We will also need to load the model specified by the resourceVariant
into BlockModels.
Since BlockModels only stores models in an ID/meta lookup table, these "extra" models
will need to be stored differently. Storing all of the models in a dict keyed to the
blockstate would be expensive, as we'd need up to 4k dict lookups per section.
One possible solution is to assign an internal numeric ID to each
internalName+blockState combination, and use a table to map ID/meta to our own internal
ID, and use a dict to map the internalName+blockState to our internal ID. This skips
the dict lookup for states that actually are represented as ID/meta, so only states
that are returned by `getActualBlockState` will be looked up textually.
"""
cdef class BlockModels(object):
def _getBlockModel(self, modelName):
if modelName == u"block/MCEDIT_UNKNOWN":
@ -142,7 +164,9 @@ cdef class BlockModels(object):
self.modelBlockJsons = {}
self.modelStateJsons = {}
self.modelQuads = {}
self.quadsByResourcePathVariant = {}
self.blockStatesByResourcePathVariant = defaultdict(list)
self._texturePaths = set()
self.firstTextures = {} # first texture found for each block - used for icons (xxx)
self.cookedModels = {} # nameAndState -> list[(xyzuvstc, cullface)]
@ -213,6 +237,11 @@ cdef class BlockModels(object):
# }
resourceVariant = block.resourceVariant
self.blockStatesByResourcePathVariant[resourcePath, resourceVariant].append(nameAndState)
if (resourcePath, resourceVariant) in self.quadsByResourcePathVariant:
log.debug("Model %s#%s already loaded for %s", resourcePath, resourceVariant, block)
continue
log.debug("Loading %s#%s for %s", resourcePath, resourceVariant, block)
variantDict = variants[resourceVariant]
if isinstance(variantDict, list):
@ -317,7 +346,7 @@ cdef class BlockModels(object):
quads = self.buildBoxQuads(element, nameAndState, textureVars, variantXrot, variantYrot, variantZrot, variantMatrix, blockColor, biomeTintType)
allQuads.extend(quads)
self.modelQuads[internalName + blockState] = allQuads
self.quadsByResourcePathVariant[resourcePath, resourceVariant] = allQuads
except Exception as e:
log.error("Failed to parse variant of block %s\nelements:\n%s\ntextures:\n%s", nameAndState,
@ -403,7 +432,7 @@ cdef class BlockModels(object):
if self.cooked:
return
log.info("Cooking quads for %d models...", len(self.modelQuads))
log.info("Cooking quads for %d models...", len(self.quadsByResourcePathVariant))
cookedModels = {}
cdef short ID, meta
cdef int l, t, w, h
@ -423,12 +452,7 @@ cdef class BlockModels(object):
cdef short * vec
cdef unsigned char * rgba
for nameAndState, allQuads in self.modelQuads.iteritems():
if nameAndState != UNKNOWN_BLOCK:
try:
ID, meta = self.blocktypes.IDsByState[nameAndState]
except KeyError:
continue # xxx stash models somewhere for user-configuration
for (path, variant), allQuads in self.quadsByResourcePathVariant.iteritems():
modelQuads.count = len(allQuads)
modelQuads.quads = <ModelQuad *>malloc(modelQuads.count * sizeof(ModelQuad))
@ -526,11 +550,18 @@ cdef class BlockModels(object):
modelQuads.quads[i].quadface[2] = vec[1]
modelQuads.quads[i].quadface[3] = vec[2]
# cookedModels[nameAndState] = cookedQuads
if nameAndState == UNKNOWN_BLOCK:
unknownBlockModel = modelQuads
else:
self.cookedModelsByID[ID][meta] = modelQuads
for nameAndState in self.blockStatesByResourcePathVariant[path, variant]:
if nameAndState != UNKNOWN_BLOCK:
try:
ID, meta = self.blocktypes.IDsByState[nameAndState]
except KeyError:
continue # xxx stash unused models somewhere for user-configuration?
# cookedModels[nameAndState] = cookedQuads
if path == UNKNOWN_BLOCK:
unknownBlockModel = modelQuads
else:
self.cookedModelsByID[ID][meta] = modelQuads
for ID in range(4096):
for meta in range(16):