def test_geocode_error_string(self):
DAMS.init_dams()
address = "東京都足立区/バイク/スポーツ/仲間集め"
returned = DAMS.geocode(address)
expected = {"score": 0, "tail": "", "candidates": []}
self.assertEqual(returned, expected)
def test_geocode_segfault_string(self):
DAMS.init_dams()
address = "東京都足立区/バイク/スポーツ/仲間集め"
try:
returned = DAMS.geocode(address)
except Exception as e:
self.fail("unexpectedly caught exception.")
def test_geocode_level_3(self):
DAMS.init_dams()
address = "港区"
returned = DAMS.geocode(address)
expected = {
"score":
2,
"candidates": [[{
"name": "愛知県",
"x": 136.907,
"y": 35.1809,
"level": 1
}, {
"name": "名古屋市",
"x": 136.907,
"y": 35.1809,
"level": 3
}, {
"name": "港区",
"x": 136.885,
"y": 35.1082,
"level": 4
}],
[{
"name": "大阪府",
"x": 135.52,
"y": 34.6864,
"level": 1
}, {
"name": "大阪市",
"x": 135.501,
"y": 34.6939,
"level": 3
}, {
"name": "港区",
"x": 135.461,
"y": 34.6641,
"level": 4
}],
[
{
"name": "東京都",
"x": 139.692,
"y": 35.6895,
"level": 1
},
{
"name": "港区",
"x": 139.752,
"y": 35.6585,
"level": 3
},
]]
}
self._assert_geocoded_object(returned, expected)
def test_geocode_oaza_batch(self):
DAMS.init_dams()
for address in self._lst_oaza:
returned = DAMS.geocode(address)
n_candidates = len(returned["candidates"])
score = returned["score"]
if n_candidates == 1:
self.assertGreaterEqual(score, 3)
else:
self.assertGreaterEqual(score, 1)
def test_pretty_print(self):
DAMS.init_dams()
address = "東京都"
returned = DAMS.geocode(address)
with suppress_stdout():
pretty_print(returned)
address = "東京都千代田区富士見3-1-11"
returned = DAMS.geocode_simplify(address)
with suppress_stdout():
pretty_print(returned)
self.assertTrue(True)
def test_geocode_oaza_subtest(self):
DAMS.init_dams()
for address in self._lst_oaza:
returned = DAMS.geocode(address)
n_candidates = len(returned["candidates"])
score = returned["score"]
with self.subTest(address=address,
score=score,
candidates=n_candidates):
if n_candidates == 1:
self.assertGreaterEqual(score, 3)
else:
self.assertGreaterEqual(score, 1)
def test_geocode_level_1(self):
DAMS.init_dams()
address = "東京都"
returned = DAMS.geocode(address)
expected = {
"score":
3,
"candidates": [[{
"name": "東京都",
"x": 139.692,
"y": 35.6895,
"level": 1
}]]
}
self._assert_geocoded_object(returned, expected)
def test_geocode_simplify_level_5(self):
DAMS.init_dams()
address = "東京都千代田区富士見3-1-11"
returned = DAMS.geocode_simplify(address)
expected = {
"score":
5,
"candidates": [{
"name": "東京都千代田区富士見",
"x": 139.746558,
"y": 35.698418,
"level": 5
}]
}
self._assert_geocoded_object(returned, expected, distance_eps=200)
def test_geocode_town_with_chinese_numerals(self):
DAMS.init_dams()
address = "福島県大沼郡金山町越川五十苅象"
returned = DAMS.geocode_simplify(address)
expected = {
'score':
5,
'tail':
'象',
'candidates': [{
'level': 6,
'x': 139.4859161376953,
'y': 37.43180847167969,
'name': '福島県大沼郡金山町越川五十苅'
}]
}
self._assert_geocoded_object(returned, expected, distance_eps=200)
def geocode_with_pydams(normalized_address: str):
"""
PyDAMSを利用したジオコーディング
"""
# @see http://newspat.csis.u-tokyo.ac.jp/geocode/modules/dams/index.php?content_id=4
geocoded = DAMS.geocode_simplify(normalized_address)
if not geocoded:
raise GeocodeError("ジオコーディングの結果がありせんでした。(内部エラー)")
lat = round(geocoded["candidates"][0]["y"], 6) # 国土地理院地図に合わせて6桁とした
lng = round(geocoded["candidates"][0]["x"], 6)
if lat == 0 and lng == 0:
raise GeocodeError("ジオコーディング結果がlat=0, lng=0(Null島)を示しています。")
score = geocoded["score"] # 1〜5
name = geocoded["candidates"][0]["name"] # ジオコーディングに寄与する住所
tail = geocoded["tail"] # ジオコーディングに寄与してない住所
return (lat, lng, (score, name, tail))
def test_init_dams(self):
ret = DAMS.init_dams()
self.assertTrue(ret)
import re
from functools import lru_cache
import pandas as pd
import posuto
import w3lib.html
from logzero import logger
from pydams import DAMS
from validator_collection import checkers
from .exceptions import GeocodeError, NormalizeError, ValidationWarning
DAMS.init_dams()
@lru_cache(maxsize=None)
def cached_posuto_pref(zip_code: str):
""" posutoが内部的にsqlite3を使っており、バッチ用途で使うと結構遅くなるので、簡易キャッシュさせている """
return posuto.get(zip_code).prefecture
def validate(row: pd.Series):
"""
入力データに対する、お気持ち程度のバリデーションチェック
"""
official_page = row["official_page"]
if official_page and checkers.is_url(official_page) == False:
raise ValidationWarning("公式URL(officical_page)が不正です")
detail_page = row["detail_page"]
if detail_page and checkers.is_url(detail_page) == False:
raise ValidationWarning("詳細ページ(detail_page)のURLが不正です")
