Mercurial > hg > GlobalNeighbors
comparison tests/test_data.py @ 0:5dba84370182
initial commit; half-working prototype
| author | Jeff Hammel <k0scist@gmail.com> |
|---|---|
| date | Sat, 24 Jun 2017 12:03:39 -0700 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:5dba84370182 |
|---|---|
| 1 #!/usr/bin/env python | |
| 2 | |
| 3 """ | |
| 4 test date integrity; ensure we know what | |
| 5 data we're dealing with | |
| 6 """ | |
| 7 | |
| 8 import os | |
| 9 import unittest | |
| 10 from globalneighbors.locations import locations | |
| 11 from globalneighbors.read import read_cities | |
| 12 from globalneighbors.schema import primary_key | |
| 13 | |
| 14 here = os.path.dirname(os.path.abspath(__file__)) | |
| 15 data = os.path.join(here, 'data') | |
| 16 full_tsv_lines = 149092 | |
| 17 | |
| 18 class DataIntegrityTest(unittest.TestCase): | |
| 19 | |
| 20 sample_tsv = os.path.join(data, 'sample.tsv') | |
| 21 assert os.path.exists(sample_tsv) | |
| 22 full_tsv = os.path.join(data, 'cities1000.txt') | |
| 23 assert os.path.exists(full_tsv) | |
| 24 | |
| 25 def test_primary_key(self): | |
| 26 """ensure we have a unique key to identify cities""" | |
| 27 | |
| 28 with open(self.full_tsv) as f: | |
| 29 cities = list(read_cities(f)) | |
| 30 n_cities = len(cities) | |
| 31 assert n_cities == full_tsv_lines | |
| 32 | |
| 33 # show we have duplicate names | |
| 34 nameset = set([city['name'] for city in cities]) | |
| 35 assert len(nameset) != n_cities | |
| 36 asciinameset = set([city['asciiname'] for city in cities]) | |
| 37 assert len(asciinameset) != n_cities | |
| 38 | |
| 39 # show we do have a unique key | |
| 40 geonameids = set([city['geonameid'] for city in cities]) | |
| 41 assert len(geonameids) == n_cities | |
| 42 | |
| 43 # is ('name', 'country code') unique? | |
| 44 nameccset = set([(city['asciiname'], city['country code']) | |
| 45 for city in cities]) | |
| 46 assert len(nameccset) != n_cities # Nope! | |
| 47 | |
| 48 def test_latlon(self): | |
| 49 """ | |
| 50 ensure all latitudes and longitudes are in the | |
| 51 range lat=(-90..90) and lon=(-180..180) | |
| 52 """ | |
| 53 | |
| 54 # read cities | |
| 55 with open(self.full_tsv) as f: | |
| 56 cities = list(read_cities(f)) | |
| 57 n_cities = len(cities) | |
| 58 assert n_cities == full_tsv_lines | |
| 59 | |
| 60 # make a location map | |
| 61 city_locations = locations(cities) | |
| 62 | |
| 63 # ensure our data is value | |
| 64 latrange = (-90., 90.) | |
| 65 lonrange = (-180., 180.) | |
| 66 for geoid, (lat, lon) in city_locations.iteritems(): | |
| 67 assert isinstance(geoid, int) | |
| 68 assert latrange[0] <= lat <= latrange[-1] | |
| 69 assert lonrange[0] <= lon <= lonrange[-1] | |
| 70 | |
| 71 | |
| 72 if __name__ == '__main__': | |
| 73 unittest.main() |