datamarket 0.9.27__py3-none-any.whl → 0.9.28__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.


This version of datamarket might be problematic. Click here for more details.

@@ -1,7 +1,10 @@
1
1
  ########################################################################################################################
2
2
  # IMPORTS
3
3
 
4
+ import gettext
4
5
  import logging
6
+ import pycountry
7
+ from geopy.distance import geodesic
5
8
 
6
9
  import requests
7
10
 
@@ -11,6 +14,8 @@ from ..params.nominatim import POSTCODES
11
14
  # CLASSES
12
15
 
13
16
  logger = logging.getLogger(__name__)
17
+ spanish = gettext.translation("iso3166-1", pycountry.LOCALES_DIR, languages=["es"])
18
+ spanish.install()
14
19
 
15
20
 
16
21
  class GeoNames:
@@ -63,28 +68,102 @@ class Nominatim:
63
68
  return requests.get(f"{self.endpoint}/reverse?lat={lat}&lon={lon}&format=json").json()
64
69
 
65
70
  def reverse_parsed(self, lat, lon):
66
- raw_json = self.reverse(lat, lon).get("address", {})
67
- geoname = self.geonames.reverse(lat, lon)
68
-
69
- postcode = self.geonames.validate_postcode(
70
- str(geoname.get("postal_code", ""))
71
- ) or self.geonames.validate_postcode(str(raw_json.get("postcode")))
72
-
73
- city = self.get_attribute(raw_json, ["city", "town", "village"]) or geoname.get("place_name")
74
-
75
- district, quarter = self.get_district_quarter(raw_json)
76
- return {
77
- "country": raw_json.get("country"),
78
- "country_code": (raw_json.get("country_code") or geoname.get("country_code") or "").lower(),
79
- "state": raw_json.get("state") or geoname.get("community"),
80
- "province": self.geonames.get_province_from_postcode(postcode),
81
- "city": city,
82
- "postcode": postcode,
83
- "district": district,
84
- "quarter": quarter,
85
- "street": raw_json.get("road"),
86
- "number": raw_json.get("house_number"),
87
- }
71
+ nominatim_raw_json = self.reverse(lat, lon)
72
+ geonames_raw_json = self.geonames.reverse(lat, lon)
73
+
74
+ nominatim_res_lat_str = nominatim_raw_json.get("lat")
75
+ nominatim_res_lon_str = nominatim_raw_json.get("lon")
76
+ geonames_res_lat_str = geonames_raw_json.get("lat")
77
+ geonames_res_lon_str = geonames_raw_json.get("lon")
78
+
79
+ dist_nominatim = float("inf")
80
+ dist_geonames = float("inf")
81
+
82
+ try:
83
+ input_coords = (float(lat), float(lon))
84
+ except (ValueError, TypeError):
85
+ logger.error(f"Invalid input coordinates for distance calculation: lat={lat}, lon={lon}")
86
+ else:
87
+ if nominatim_res_lat_str and nominatim_res_lon_str:
88
+ try:
89
+ nominatim_coords = (float(nominatim_res_lat_str), float(nominatim_res_lon_str))
90
+ dist_nominatim = geodesic(input_coords, nominatim_coords).km
91
+ except (ValueError, TypeError):
92
+ logger.warning("Invalid Nominatim coordinates for distance calculation.")
93
+
94
+ if geonames_res_lat_str and geonames_res_lon_str:
95
+ try:
96
+ geonames_coords = (float(geonames_res_lat_str), float(geonames_res_lon_str))
97
+ dist_geonames = geodesic(input_coords, geonames_coords).km
98
+ except (ValueError, TypeError):
99
+ logger.warning("Invalid GeoNames coordinates for distance calculation.")
100
+
101
+ if dist_nominatim <= dist_geonames and nominatim_res_lat_str is not None and nominatim_res_lon_str is not None:
102
+ # Use Nominatim data
103
+ raw_address = nominatim_raw_json.get("address", {})
104
+ postcode_str = str(raw_address.get("postcode", ""))
105
+ postcode = self.geonames.validate_postcode(postcode_str)
106
+ province = self.geonames.get_province_from_postcode(postcode) if postcode else None
107
+ city = self.get_attribute(raw_address, ["city", "town", "village"])
108
+ district, quarter = self.get_district_quarter(raw_address)
109
+
110
+ return {
111
+ "country": raw_address.get("country"),
112
+ "country_code": (raw_address.get("country_code") or "").lower(),
113
+ "state": raw_address.get("state"),
114
+ "province": province,
115
+ "city": city,
116
+ "postcode": postcode,
117
+ "district": district,
118
+ "quarter": quarter,
119
+ "street": raw_address.get("road"),
120
+ "number": raw_address.get("house_number"),
121
+ }
122
+
123
+ elif dist_geonames < dist_nominatim and geonames_res_lat_str is not None and geonames_res_lon_str is not None:
124
+ # Use GeoNames data
125
+ geonames_country_code_str = geonames_raw_json.get("country_code")
126
+ country_name = None
127
+ if geonames_country_code_str:
128
+ try:
129
+ country_obj = pycountry.countries.get(alpha_2=geonames_country_code_str.upper())
130
+ if country_obj:
131
+ country_name = spanish.gettext(country_obj.name)
132
+ except LookupError:
133
+ logger.warning(f"Country name not found for code: {geonames_country_code_str} using pycountry.")
134
+
135
+ postcode_str = str(geonames_raw_json.get("postal_code", ""))
136
+ postcode = self.geonames.validate_postcode(postcode_str)
137
+ province = self.geonames.get_province_from_postcode(postcode) if postcode else None
138
+ city = geonames_raw_json.get("place_name")
139
+
140
+ return {
141
+ "country": country_name,
142
+ "country_code": (geonames_country_code_str or "").lower(),
143
+ "state": geonames_raw_json.get("community"),
144
+ "province": province,
145
+ "city": city,
146
+ "postcode": postcode,
147
+ "district": None,
148
+ "quarter": None,
149
+ "street": None,
150
+ "number": None,
151
+ }
152
+
153
+ else:
154
+ # Neither source provided valid coordinates
155
+ return {
156
+ "country": None,
157
+ "country_code": None,
158
+ "state": None,
159
+ "province": None,
160
+ "city": None,
161
+ "postcode": None,
162
+ "district": None,
163
+ "quarter": None,
164
+ "street": None,
165
+ "number": None,
166
+ }
88
167
 
89
168
  def get_district_quarter(self, raw_json):
90
169
  district = self.get_attribute(raw_json, ["city_district", "suburb", "borough"])
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.3
2
2
  Name: datamarket
3
- Version: 0.9.27
3
+ Version: 0.9.28
4
4
  Summary: Utilities that integrate advanced scraping knowledge into just one library.
5
5
  License: GPL-3.0-or-later
6
6
  Author: DataMarket
@@ -28,7 +28,6 @@ Provides-Extra: duckduckgo-search
28
28
  Provides-Extra: fake-useragent
29
29
  Provides-Extra: geoalchemy2
30
30
  Provides-Extra: geopandas
31
- Provides-Extra: geopy
32
31
  Provides-Extra: google-api-python-client
33
32
  Provides-Extra: google-auth-httplib2
34
33
  Provides-Extra: google-auth-oauthlib
@@ -77,7 +76,7 @@ Requires-Dist: dynaconf (>=3.0.0,<4.0.0)
77
76
  Requires-Dist: fake-useragent (>=2.0.0,<3.0.0) ; extra == "fake-useragent"
78
77
  Requires-Dist: geoalchemy2 (>=0.17.0,<0.18.0) ; extra == "geoalchemy2"
79
78
  Requires-Dist: geopandas (>=1.0.0,<2.0.0) ; extra == "geopandas"
80
- Requires-Dist: geopy (>=2.0.0,<3.0.0) ; extra == "geopy"
79
+ Requires-Dist: geopy (>=2.0.0,<3.0.0)
81
80
  Requires-Dist: google-api-python-client (>=2.0.0,<3.0.0) ; extra == "google-api-python-client"
82
81
  Requires-Dist: google-auth-httplib2 (>=0.2.0,<0.3.0) ; extra == "google-auth-httplib2"
83
82
  Requires-Dist: google-auth-oauthlib (>=1.0.0,<2.0.0) ; extra == "google-auth-oauthlib"
@@ -98,6 +97,7 @@ Requires-Dist: playwright (==1.47.0) ; extra == "playwright"
98
97
  Requires-Dist: pre-commit (>=4.0.0,<5.0.0)
99
98
  Requires-Dist: psycopg2-binary (>=2.0.0,<3.0.0)
100
99
  Requires-Dist: pyarrow (>=19.0.0,<20.0.0) ; extra == "pyarrow"
100
+ Requires-Dist: pycountry (>=24.0.0,<25.0.0)
101
101
  Requires-Dist: pydrive2 (>=1.0.0,<2.0.0) ; extra == "pydrive2" or extra == "drive"
102
102
  Requires-Dist: pymupdf (>=1.0.0,<2.0.0) ; extra == "pymupdf"
103
103
  Requires-Dist: pysocks (>=1.0.0,<2.0.0) ; extra == "pysocks"
@@ -4,7 +4,7 @@ datamarket/interfaces/alchemy.py,sha256=4q_gLKCKPK437VKOpdBKSrCyy42P_yWxIhE7KuvH
4
4
  datamarket/interfaces/aws.py,sha256=7KLUeBxmPN7avEMPsu5HC_KHB1N7W6Anp2X8fo43mlw,2383
5
5
  datamarket/interfaces/drive.py,sha256=shbV5jpQVe_KPE-8Idx6Z9te5Zu1SmVfrvSAyd9ZIgE,2915
6
6
  datamarket/interfaces/ftp.py,sha256=o0KlJxtksbop9OjCiQRzyAa2IeG_ExVXagS6apwrAQo,1881
7
- datamarket/interfaces/nominatim.py,sha256=ysIA2J1GhsZ0TJxD6B8N1_a7dkMEqtZQV6mT4Hayecg,3672
7
+ datamarket/interfaces/nominatim.py,sha256=TD4OhWdIwn53Va41BS1ugogHEZw0ANKxWIfc9G2JWLU,7280
8
8
  datamarket/interfaces/peerdb.py,sha256=cwYwvO740GyaPo9zLAwJsf3UeJDGDiYzjQVM9Q6s-_g,23652
9
9
  datamarket/interfaces/proxy.py,sha256=updoOStKd8-nQBbxWbnD9eOt6HksnYi-5dQ0rEySf5M,3152
10
10
  datamarket/interfaces/tinybird.py,sha256=AYrcRGNOCoCt7ojilkWa27POROee9sTCwZ61GGHEPeM,2698
@@ -19,7 +19,7 @@ datamarket/utils/soda.py,sha256=eZTXFbI1P3WoMd1MM-YjoVTpdjTcDSWuvBb7ViBMhSQ,941
19
19
  datamarket/utils/strings.py,sha256=rEX9NeBG4C7RECgT0EQebgoFoxgZMy9-7EcBSxgBANU,5654
20
20
  datamarket/utils/typer.py,sha256=FDF3l6gh3UlAFPsHCtesnekvct2rKz0oFn3uKARBQvE,814
21
21
  datamarket/utils/types.py,sha256=vxdQZdwdXrfPR4Es52gBgol-tMRIOD6oK9cBo3rB0JQ,74
22
- datamarket-0.9.27.dist-info/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
23
- datamarket-0.9.27.dist-info/METADATA,sha256=ZDopWDfk3f0HeTZSVAKSnAmfOPSBUOJNlos9fqGzKJA,6543
24
- datamarket-0.9.27.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
25
- datamarket-0.9.27.dist-info/RECORD,,
22
+ datamarket-0.9.28.dist-info/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
23
+ datamarket-0.9.28.dist-info/METADATA,sha256=5V61eLQ9HR_8Tl7J0sRKwR_mzKbJctIjvEebNtPtqug,6546
24
+ datamarket-0.9.28.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
25
+ datamarket-0.9.28.dist-info/RECORD,,