大橙子网站建设,新征程启航
为企业提供网站建设、域名注册、服务器等服务
在上一篇文章中提出用geohash解决匹配点的性能问题,该算法有个缺陷也就是如果在边界上的时候,该点附近的点可能就跨界了,也就是通过该点的geohash值无法找到对应的点。例如:
图中蓝色点为需要绑路映射的点
下方的红点为绑路计算之后的点
导致偏移的原因是
随州街上的点数据 在geohashcode分区时都属于蓝点上方的分区
在数据库中查询出来的结果显示 和蓝色点在一个分区的路网数据位于下图
所以找到的数据均位于下方的道路上 导致偏移问题
为了解决这个问题提出扩散hashcode的代码查找附近范围的geohashcode代码
代码如下
所以找到的数据均位于下方的道路上 导致偏移问题
为了解决这个问题提出扩散hashcode的代码查找附近范围的geohashcode代码
代码如下
成都创新互联公司长期为成百上千家客户提供的网站建设服务,团队从业经验10年,关注不同地域、不同群体,并针对不同对象提供差异化的产品和服务;打造开放共赢平台,与合作伙伴共同营造健康的互联网生态环境。为阆中企业提供专业的网站制作、成都网站建设,阆中网站改版等技术服务。拥有十年丰富建站经验和众多成功案例,为您定制开发。
"""A simple GeoHash implementation."""`
# Forward and reverse base 32 map
BASESEQUENCE = '0123456789bcdefghjkmnpqrstuvwxyz'
BASE32MAP = dict((k, count) for count, k in enumerate(BASESEQUENCE))
BASE32MAPR = dict((count, k) for count, k in enumerate(BASESEQUENCE))
def _bits_to_float(bits, lower=-90.0, middle=0.0, upper=90.0):
"""Convert GeoHash bits to a float."""
for i in bits:
if i:
lower = middle
else:
upper = middle
middle = (upper + lower) / 2
return middle
def _float_to_bits(value, lower=-90.0, middle=0.0, upper=90.0, length=15):
"""Convert a float to a list of GeoHash bits."""
ret = []
for i in range(length):
if value >= middle:
lower = middle
ret.append(1)
else:
upper = middle
ret.append(0)
middle = (upper + lower) / 2
return ret
def _geohash_to_bits(value):
"""Convert a GeoHash to a list of GeoHash bits."""
b = map(BASE32MAP.get, value)
ret = []
for i in b:
out = []
for z in range(5):
out.append(i & 0b1)
i = i >> 1
ret += out[::-1]
return ret
def _bits_to_geohash(value):
"""Convert a list of GeoHash bits to a GeoHash."""
ret = []
# Get 5 bits at a time
for i in (value[i:i + 5] for i in range(0, len(value), 5)):
# Convert binary to integer
# Note: reverse here, the slice above doesn't work quite right in reverse.
total = sum([(bit * 2 ** count) for count, bit in enumerate(i[::-1])])
ret.append(BASE32MAPR[total])
# Join the string and return
return "".join(ret)
# Public
def decode(value):
"""Decode a geohash. Returns a (lon,lat) pair."""
assert value, "Invalid geohash: %s" % value
# Get the GeoHash bits
bits = _geohash_to_bits(value)
# Unzip the GeoHash bits.
lon = bits[0::2]
lat = bits[1::2]
# Convert to lat/lon
return (
_bits_to_float(lon, lower=-180.0, upper=180.0),
_bits_to_float(lat)
)
def encode(lonlat, length=12):
"""Encode a (lon,lat) pair to a GeoHash."""
assert len(lonlat) == 2, "Invalid lon/lat: %s" % lonlat
# Half the length for each component.
length /= 2
lon = _float_to_bits(lonlat[0], lower=-180.0, upper=180.0, length=length * 5)
lat = _float_to_bits(lonlat[1], lower=-90.0, upper=90.0, length=length * 5)
# Zip the GeoHash bits.
ret = []
for a, b in zip(lon, lat):
ret.append(a)
ret.append(b)
return _bits_to_geohash(ret)
def adjacent(geohash, direction):
"""Return the adjacent geohash for a given direction."""
# Based on an MIT licensed implementation by Chris Veness from:
# http://www.movable-type.co.uk/scripts/geohash.html
assert direction in 'nsew', "Invalid direction: %s" % direction
assert geohash, "Invalid geohash: %s" % geohash
neighbor = {
'n': ['p0r21436x8zb9dcf5h7kjnmqesgutwvy', 'bc01fg45238967deuvhjyznpkmstqrwx'],
's': ['14365h7k9dcfesgujnmqp0r2twvyx8zb', '238967debc01fg45kmstqrwxuvhjyznp'],
'e': ['bc01fg45238967deuvhjyznpkmstqrwx', 'p0r21436x8zb9dcf5h7kjnmqesgutwvy'],
'w': ['238967debc01fg45kmstqrwxuvhjyznp', '14365h7k9dcfesgujnmqp0r2twvyx8zb']
}
border = {
'n': ['prxz', 'bcfguvyz'],
's': ['028b', '0145hjnp'],
'e': ['bcfguvyz', 'prxz'],
'w': ['0145hjnp', '028b']
}
last = geohash[-1]
parent = geohash[0:-1]
t = len(geohash) % 2
# Check for edge cases
if (last in border[direction][t]) and (parent):
parent = adjacent(parent, direction)
return parent + BASESEQUENCE[neighbor[direction][t].index(last)]
def neighbors(geohash):
"""Return all neighboring geohashes."""
return {
'n': adjacent(geohash, 'n'),
'ne': adjacent(adjacent(geohash, 'n'), 'e'),
'e': adjacent(geohash, 'e'),
'se': adjacent(adjacent(geohash, 's'), 'e'),
's': adjacent(geohash, 's'),
'sw': adjacent(adjacent(geohash, 's'), 'w'),
'w': adjacent(geohash, 'w'),
'nw': adjacent(adjacent(geohash, 'n'), 'w'),
'c': geohash
}
def neighborsfit(centroid, points):
centroid = encode(centroid)
points = map(encode, points)
for i in range(1, len(centroid)):
g = centroid[0:i]
n = set(neighbors(g).values())
unbounded = [point for point in points if (point[0:i] not in n)]
if unbounded:
break
return g[0:-1]