Source code for madcad.hashing

# This file is part of pymadcad,  distributed under license LGPL v3

'''  This modules provide tools for accessing data using its space location (not for final user).
	
	The complexity and therefore the cost of those operations are most of the time close to the hashmap complexity O(1). It means data is found in time independently of the actual size of the mesh or whatever storage it is.
'''

from .mathutils import *
from . import core
from . import mesh
from functools import reduce
from math import floor, ceil, sqrt, inf


[docs]class PositionMap: ''' Holds objects associated with their location. Every object can be bound to multiple locations, and each location can hold multiple objects. `cellsize` defines the box size for location hashing (the smaller it is, the bigger the memory footprint will be for non-point primitives) Attributes defined here: :cellsize: the boxing parameter (DON'T CHANGE IT IF NON-EMPTY) :dict: the hashmap from box to objects lists ''' __slots__ = 'cellsize', 'dict', 'options' def __init__(self, cellsize, iterable=None): self.options = {} self.cellsize = cellsize self.dict = {} if iterable: self.update(iterable) def keysfor(self, space): ''' Rasterize the primitive, yielding the successive position keys currently allowed primitives are :points: vec3 :segments: (vec3,vec3) :triangles: (vec3,vec3,vec3) ''' cell = self.cellsize # point if isinstance(space, vec3): yield tuple(i64vec3(glm.floor(space/cell))) # segment elif isinstance(space, tuple) and len(space) == 2: # permutation of coordinates to get the direction the closer to Z n = glm.abs(space[1]-space[0]) if max(n) < NUMPREC*max((max(p) for p in space)): return if n[1] >= n[0] and n[1] >= n[2]: order,reorder = (2,0,1),(1,2,0) elif n[0] >= n[1] and n[0] >= n[2]: order,reorder = (1,2,0),(2,0,1) else: order,reorder = (0,1,2),(0,1,2) space = [ vec3(p[order[0]], p[order[1]], p[order[2]]) for p in space] # prepare variables cell2 = cell/2 v = space[1]-space[0] dy = v[1]/v[2] dx = v[0]/v[2] o = space[0] fy = lambda z: o[1] + dy*(z-o[2]) fx = lambda z: o[0] + dx*(z-o[2]) # z selection if v[2] > 0: zmin,zmax = space[0][2], space[1][2] else: zmin,zmax = space[1][2], space[0][2] zmin -= zmin%cell for i in range(max(1,ceil((zmax-zmin)/cell))): z = zmin + cell*i + cell2 # y selection if dy > 0: ymin,ymax = fy(z-cell2), fy(z+cell2) else: ymin,ymax = fy(z+cell2), fy(z-cell2) ymin -= ymin%cell for j in range(max(1,ceil((ymax-ymin)/cell))): y = ymin + j*cell + cell2 # x selection if dx > 0: xmin,xmax = fx(z-cell2), fx(z+cell2) else: xmin,xmax = fx(z+cell2), fx(z-cell2) xmin -= xmin%cell for k in range(max(1,ceil((xmax-xmin)/cell))): x = xmin + k*cell + cell2 p = (x,y,z) yield tuple([floor(p[i]/cell) for i in reorder]) # triangle elif isinstance(space, tuple) and len(space) == 3: # permutation of coordinates to get the normal the closer to Z n = glm.abs(cross(space[1]-space[0], space[2]-space[0])) if max(n) < NUMPREC*max((max(p) for p in space)): return if n[1] >= n[0] and n[1] >= n[2]: order,reorder = (2,0,1),(1,2,0) elif n[0] >= n[1] and n[0] >= n[2]: order,reorder = (1,2,0),(2,0,1) else: order,reorder = (0,1,2),(0,1,2) space = [ vec3(p[order[0]], p[order[1]], p[order[2]]) for p in space] # prepare variables v = [space[i-1]-space[i] for i in range(3)] n = cross(v[0],v[1]) dx = -n[0]/n[2] dy = -n[1]/n[2] o = space[0] fz = lambda x,y: o[2] + dx*(x-o[0]) + dy*(y-o[1]) cell2 = cell/2 pmin = reduce(glm.min, space) pmax = reduce(glm.max, space) xmin,xmax = pmin[0],pmax[0] pmin -= pmin%cell pmax += cell - pmax%cell # x selection xmin -= xmin%cell for i in range(max(1,ceil((xmax-xmin)/cell))): x = xmin + cell*i + cell2 # y selection cand = [] for i in range(3): # NOTE: cet interval ajoute parfois des cases inutiles apres les sommets if (space[i-1][0]-x+cell2)*(space[i][0]-x-cell2) <= 0 or (space[i-1][0]-x-cell2)*(space[i][0]-x+cell2) <= 0: d = v[i][1]/(v[i][0] if v[i][0] else inf) cand.append( space[i][1] + d * (x-cell2-space[i][0]) ) cand.append( space[i][1] + d * (x+cell2-space[i][0]) ) ymin,ymax = max(pmin[1],min(cand)), min(pmax[1],max(cand)) ymin -= ymin%cell for j in range(max(1,ceil((ymax-ymin)/cell))): y = ymin + cell*j + cell2 # z selection cand = [] cand.append( fz(x-cell2, y-cell2) ) cand.append( fz(x+cell2, y-cell2) ) cand.append( fz(x-cell2, y+cell2) ) cand.append( fz(x+cell2, y+cell2) ) zmin,zmax = max(pmin[2],min(cand)), min(pmax[2],max(cand)) zmin -= zmin%cell for k in range(max(1,ceil((zmax-zmin)/cell))): z = zmin + cell*k + cell2 # remove box from corners that goes out of the area p = (x,y,z) if pmin[0]<p[0] and pmin[1]<p[1] and pmin[2]<p[2] and p[0]<pmax[0] and p[1]<pmax[1] and p[2]<pmax[2]: yield tuple([floor(p[i]/cell) for i in reorder]) else: raise TypeError("PositionMap only supports keys of type: points, segments, triangles")
[docs] def keysfor(self, space): ''' Rasterize the primitive, yielding the successive position keys currently allowed primitives are :points: vec3 :segments: (vec3,vec3) :triangles: (vec3,vec3,vec3) ''' # point if isinstance(space, vec3): return tuple(i64vec3(glm.floor(space/cell))), # segment elif isinstance(space, tuple) and len(space) == 2: return core.rasterize_segment(space, self.cellsize) # triangle elif isinstance(space, tuple) and len(space) == 3: return core.rasterize_triangle(space, self.cellsize) else: raise TypeError("PositionMap only supports keys of type: points, segments, triangles")
[docs] def update(self, other): ''' Add the elements from an other PositionMap or from an iterable ''' if isinstance(other, PositionMap): assert self.cellsize == other.cellsize for k,v in other.dict.items(): if k in self.dict: self.dict[k].extend(v) else: self.dict[k] = v elif hasattr(other, '__iter__'): for space,obj in other: self.add(space,obj) else: raise TypeError("update requires a PositionMap or an iterable of couples (space, obj)")
[docs] def add(self, space, obj): ''' Add an object associated with a primitive ''' for k in self.keysfor(space): if k not in self.dict: self.dict[k] = [obj] else: self.dict[k].append(obj)
[docs] def get(self, space): ''' Get the objects associated with the given primitive ''' for k in self.keysfor(space): if k in self.dict: yield from self.dict[k]
def __contains__(self, space): return next(self.get(space), None) is not None _display = ( [ vec3(0,0,0), vec3(1,0,0),vec3(0,1,0),vec3(0,0,1), vec3(0,1,1),vec3(1,0,1),vec3(1,1,0), vec3(1,1,1)], [ (0,1),(0,2),(0,3), (1,6),(2,6), (1,5),(3,5), (2,4),(3,4), (4,7),(5,7),(6,7), ], )
[docs] def display(self, scene): web = mesh.Web() if 'color' in self.options: web.options['color'] = self.options['color'] base = vec3(self.cellsize) for k in self.dict: l = len(web.points) web += mesh.Web([base*(p+k) for p in self._display[0]], self._display[1], groups=[k]) return web.display(scene)
[docs]def meshcellsize(mesh): ''' Returns a good cell size to index primitives of a mesh with a PositionMap See implementation. ''' # the number of key for triangles is propotionate to the surface # points are placed on this surface, so sqrt(len(pts)) is a good hint for the point density on the surface return length(mesh.box().width) / sqrt(len(mesh.points))
[docs]class PointSet: ''' Holds a list of points and hash them. The points are holds using indices, that allows to get the point buffer at any time, or to retrieve only a point index. `cellsize` defines the box size in which two points are considered to be the same Methods are inspired from the builtin type set Attributes defined here: :points: the point buffer (READ-ONLY PURPOSE) :cellsize: the boxing parameter (DON'T CHANGE IT IF NON-EMPTY). mendatory and is the distance at which you want to distinguish points :dict: the hashmap from box to point indices Build parameters: :iterable: use it to build the set by inserting elements :manage: pass a list for inplace use it, only indexing will be built ''' __slots__ = 'points', 'cellsize', 'dict' def __init__(self, cellsize, iterable=None, manage=None): self.cellsize = cellsize self.dict = {} if manage is not None: self.points = manage for i in reversed(range(len(self.points))): self.dict[self.keyfor(self.points[i])] = i else: self.points = typedlist(dtype=vec3) if iterable: self.update(iterable)
[docs] def keyfor(self, pt): ''' Hash key for a point ''' return tuple(i64vec3(glm.floor(pt/self.cellsize)))
def keysfor(self, pt): ''' Iterable of positions at which an equivalent point can be ''' vox = pt/self.cellsize k = i64vec3(glm.floor(vox-0.5+NUMPREC)), i64vec3(glm.floor(vox+0.5-NUMPREC)) return ( (k[0][0], k[0][1], k[0][2]), (k[1][0], k[0][1], k[0][2]), (k[0][0], k[1][1], k[0][2]), (k[1][0], k[1][1], k[0][2]), (k[0][0], k[0][1], k[1][2]), (k[1][0], k[0][1], k[1][2]), (k[0][0], k[1][1], k[1][2]), (k[1][0], k[1][1], k[1][2]), )
[docs] def update(self, iterable): ''' Add the points from an iterable ''' for pt in iterable: self.add(pt)
[docs] def difference_update(self, iterable): ''' Remove the points from an iterable ''' for pt in iterable: self.discard(pt)
[docs] def add(self, pt): ''' Add a point ''' for key in self.keysfor(pt): if key in self.dict: return self.dict[key] self.dict[self.keyfor(pt)] = l = len(self.points) self.points.append(pt) return l
[docs] def remove(self, pt): ''' Remove a point ''' for key in self.keysfor(pt): if key in self.dict: del self.dict[key] return else: raise IndexError("position doesn't exist in set")
[docs] def discard(self, pt): ''' Remove the point at given location if any ''' for key in self.keysfor(pt): if key in self.dict: del self.dict[key]
def __contains__(self, pt): for key in self.keysfor(pt): if key in self.dict: return True return False
[docs] def __getitem__(self, pt): for key in self.keysfor(pt): if key in self.dict: return self.dict[key] raise IndexError("position doesn't exist in set")
__iadd__ = update __isub__ = difference_update
[docs] def __add__(self, iterable): s = PositionSet() s.union(self.points) s.union(iterable) return s
[docs] def __sub__(self, iterable): s = PositionSet() s.union(self.points) s.difference(iterable) return s