def encode(self, word):
val = BMap1D()
val.encode(WordEncoder.cmap, word.lower())
rv = BMap1D(self.nbits)
length = len(val) if len(val) < self.nbits else self.nbits
rv[0:length] = val[0:length] #keep correct size
return rv
def encode(self, value): i = int( math.floor(self.buckets * ((value - self.vmin)/float(self.vrange)) ) ) rv = BMap1D(self.nbits) rv.setall(0) end = i + self.width rv[i : end] = 1 return rv
def get_byidxs(self, idxs):
#print ">>> %s" % (idxs,)
#In case 2D indecies are specified
if isinstance(idxs, tuple):
rows, cols = idxs
if BMap2D.is_empty_slice(rows): #cols only case
return self.get_cols(BMap2D.slice2range(cols, self.ncols))
else:
rows = self.slice2range(rows, self.nrows)
if BMap2D.is_empty_slice(cols): #rows only case
return self.get_rows(BMap2D.slice2range(rows, self.nrows))
else:
cols = self.slice2range(cols, self.ncols)
#they are no longer slices but ranges
bm = self.get_rows(rows)
rv = bm.get_cols(cols)
rv.rotate()
return rv
else: #1D index
if isinstance(idxs, (list, np.ndarray)):
rv = []
for i in idxs:
rv.append(self.bmap[i])
return rv
else:
return BMap1D(self.bmap[idxs])
def rotate(self, deg=90):
tmp = BMap1D(self.len)
for r in xrange(self.ncols):
pos = r * self.nrows
tmp[pos:pos + self.nrows] = self.bmap[r:self.len:self.ncols]
self.bmap = tmp
cols = self.ncols
self.ncols = self.nrows
self.nrows = cols
def encode_bucket(self, value):
if value < self.vmin or value > self.vmax:
warnings.warn("Value '%s' outside of range : [%s <=> %s]" %
(value, self.vmin, self.vmax))
#corner case for vmax
i = self.buckets - 1 if value == self.vmax else self.pos(value)
rv = BMap1D(self.buckets)
rv[i] = 1
return rv
def init_cmap(cls, nbits=26):
log.debug('Building charachter map : %sbit code ...' % nbits)
if nbits > 26:
cls.ce = CategoryEncoder(nbits=nbits, ncats=26)
for i, c in enumerate(string.ascii_lowercase):
if nbits == 5:
cls.cmap[c] = BMap1D(format(
i + 1, 'b').zfill(nbits)) #5bits for 26 chars
continue
if nbits == 26: #26 bits for 26 chars
bmap = BMap1D(nbits)
bmap[i:] = 1
cls.cmap[c] = bmap
continue
if nbits > 26:
cls.cmap[c] = cls.ce.encode(i + 1)
continue
def encode(self, value):
if value < self.vmin or value > self.vmax:
warnings.warn("Value '%s' outside of range : [%s <=> %s]" %
(value, self.vmin, self.vmax))
i = self.pos(value)
rv = BMap1D(self.nbits)
#rv.setall(0)
end = i + self.width
rv[i:end] = 1
return rv
def __init__(self, nrows, nbits=None, val=None, randomize=0, ncols=None): self.nrows = nrows self.ncols = ncols if ncols else nbits assert self.ncols , "When creating BMap2D, please provide ncols|nbits" self.len = self.nrows * self.ncols self.ext = 'bmap' self.bmap = BMap1D(self.len) self.index = 0 self.erase() if randomize :#should we generate random noise many = int(self.len * randomize) idxs = np.random.randint(0,self.len,many) for i in np.nditer(idxs) : self.bmap[i] = 1 self.set(val)
def rand_bmap(self):
return BMap1D(list(np.random.randint(0, 2, (self.nrows * self.ncols))))
def rand_item(self):
return BMap1D(list(np.random.randint(0, 2, self.ncols)))
def mk_item(self):
val = BMap1D(self.ncols)
val.setall(0)
return val
def idxs2bits(idxs, nbits):
bit_ary = BMap1D(nbits)
bit_ary.setall(0)
for i in idxs:
bit_ary[i] = 1
return bit_ary
def np2bits(ary):
b = BMap1D()
b.pack(ary.astype(np.bool).tostring())
return b
def add_rows(self, nrows): bits = nrows * self.ncols ba = BMap1D(bits) self.bmap.extend(ba) self.nrows += nrows
