tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[fix] some weirdness with the dateline and polygons that have a longitude of exactly 180.0

Browse Source
This commit is contained in:
Al
2016-04-29 19:59:53 -04:00
parent 8cf51378a0
commit 99e634aaba
3 changed files with 11 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
scripts/geodata/math/floats.py Normal file
View File

@@ -0,0 +1,5 @@
FLOAT_EPSILON = 1e-09
def isclose(a, b, rel_tol=FLOAT_EPSILON, abs_tol=0.0):
return abs(a - b) <= max(rel_tol * max(abs(a), abs(b)), abs_tol)

6
scripts/geodata/math/sampling.py
View File

@@ -2,7 +2,7 @@ import bisect
import random
import sys
FLOAT_EPSILON = 1e-09
from geodata.math.floats import isclose, FLOAT_EPSILON
def weighted_choice(values, cdf):
@@ -16,10 +16,6 @@ def weighted_choice(values, cdf):
return values[i]
def isclose(a, b, rel_tol=FLOAT_EPSILON, abs_tol=0.0):
return abs(a - b) <= max(rel_tol * max(abs(a), abs(b)), abs_tol)
def check_probability_distribution(probs):
cumulative = 0.0
for p in probs: