|
1
|
|
|
import fsutil |
|
2
|
|
|
from benedict import benedict |
|
3
|
|
|
from slugify import slugify |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
|
def _expect_keys(d, keys): |
|
7
|
|
|
assert all( |
|
8
|
|
|
[key in d for key in keys] |
|
9
|
|
|
), f"Invalid keys {list(d.keys())}, missing one or more expected keys {keys}." |
|
10
|
|
|
|
|
11
|
|
|
|
|
12
|
|
|
def _slugify_names(*names): |
|
13
|
|
|
return sorted(set(filter(bool, [slugify(name) for name in names]))) |
|
14
|
|
|
|
|
15
|
|
|
|
|
16
|
|
|
def _update_countries_data(): |
|
17
|
|
|
# https://www.anagrafenazionale.interno.it/area-tecnica/tabelle-di-decodifica/ |
|
18
|
|
|
data_url = "https://www.anagrafenazionale.interno.it/wp-content/uploads/2022/10/tabella_2_statiesteri.xlsx" |
|
19
|
|
|
data = benedict.from_xls(data_url) |
|
20
|
|
|
data.standardize() |
|
21
|
|
|
# print(data.dump()) |
|
22
|
|
|
|
|
23
|
|
|
def map_item(item): |
|
24
|
|
|
if not item: |
|
25
|
|
|
return None |
|
26
|
|
|
|
|
27
|
|
|
_expect_keys( |
|
28
|
|
|
item, |
|
29
|
|
|
[ |
|
30
|
|
|
"codat" "denominazione", |
|
31
|
|
|
"denominazioneistat", |
|
32
|
|
|
"denominazioneistat_en", |
|
33
|
|
|
"datainiziovalidita", |
|
34
|
|
|
"datafinevalidita", |
|
35
|
|
|
], |
|
36
|
|
|
) |
|
37
|
|
|
|
|
38
|
|
|
code = item.get_str("codat").upper() |
|
39
|
|
|
if not code: |
|
40
|
|
|
return None |
|
41
|
|
|
assert len(code) == 4, f"Invalid code: '{code}'" |
|
42
|
|
|
|
|
43
|
|
|
name = item.get_str("denominazione").title() |
|
44
|
|
|
assert name != "", f"Invalid name: '{name}'" |
|
45
|
|
|
name_alt = item.get_str("denominazioneistat").title() |
|
46
|
|
|
name_alt_en = item.get_str("denominazioneistat_en").title() |
|
47
|
|
|
name_slugs = _slugify_names(name, name_alt, name_alt_en) |
|
48
|
|
|
|
|
49
|
|
|
province = "EE" |
|
50
|
|
|
|
|
51
|
|
|
date_created = item.get_datetime("datainiziovalidita") |
|
52
|
|
|
date_deleted = item.get_datetime("datafinevalidita") |
|
53
|
|
|
date_deleted_raw = item.get_str("datafinevalidita") |
|
54
|
|
|
if "9999" in date_deleted_raw: |
|
55
|
|
|
date_deleted = "" |
|
56
|
|
|
|
|
57
|
|
|
return { |
|
58
|
|
|
"active": False if date_deleted else True, |
|
59
|
|
|
"code": code, |
|
60
|
|
|
"date_created": date_created, |
|
61
|
|
|
"date_deleted": date_deleted, |
|
62
|
|
|
"name": name, |
|
63
|
|
|
"name_alt": name_alt, |
|
64
|
|
|
"name_alt_en": name_alt_en, |
|
65
|
|
|
"name_slugs": name_slugs, |
|
66
|
|
|
"province": province, |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
_write_data_json( |
|
70
|
|
|
filepath="../codicefiscale/data/countries.json", |
|
71
|
|
|
data=[map_item(benedict(item)) for item in data["values"]], |
|
72
|
|
|
) |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
def _update_municipalities_data(): |
|
76
|
|
|
# https://www.anagrafenazionale.interno.it/area-tecnica/tabelle-di-decodifica/ |
|
77
|
|
|
data_url = "https://www.anagrafenazionale.interno.it/wp-content/uploads/2022/12/ANPR_archivio_comuni.csv" |
|
78
|
|
|
data = benedict.from_csv(data_url) |
|
79
|
|
|
data.standardize() |
|
80
|
|
|
|
|
81
|
|
|
def map_item(item): |
|
82
|
|
|
if not item: |
|
83
|
|
|
return None |
|
84
|
|
|
|
|
85
|
|
|
_expect_keys( |
|
86
|
|
|
item, |
|
87
|
|
|
[ |
|
88
|
|
|
"stato", |
|
89
|
|
|
"codcatastale", |
|
90
|
|
|
"denominazione_it", |
|
91
|
|
|
"denomtraslitterata", |
|
92
|
|
|
"altradenominazione", |
|
93
|
|
|
"altradenomtraslitterata", |
|
94
|
|
|
"siglaprovincia", |
|
95
|
|
|
"dataistituzione", |
|
96
|
|
|
"datacessazione", |
|
97
|
|
|
], |
|
98
|
|
|
) |
|
99
|
|
|
|
|
100
|
|
|
status = item.get("stato", "").upper() |
|
101
|
|
|
assert len(status) == 1 and status in ["A", "C"], f"Invalid status: '{status}'" |
|
102
|
|
|
active = status == "A" |
|
103
|
|
|
|
|
104
|
|
|
code = item.get_str("codcatastale").upper() |
|
105
|
|
|
assert code == "ND" or len(code) == 4, f"Invalid code: '{code}'" |
|
106
|
|
|
|
|
107
|
|
|
name = item.get_str("denominazione_it").title() |
|
108
|
|
|
assert name != "", f"Invalid name: {name}" |
|
109
|
|
|
|
|
110
|
|
|
name_trans = item.get_str("denomtraslitterata").title() |
|
111
|
|
|
name_alt = item.get_str("altradenominazione").title() |
|
112
|
|
|
name_alt_trans = item.get_str("altradenomtraslitterata").title() |
|
113
|
|
|
name_slugs = _slugify_names(name, name_trans, name_alt, name_alt_trans) |
|
114
|
|
|
|
|
115
|
|
|
province = item.get("siglaprovincia", "").upper() |
|
116
|
|
|
assert len(province) == 2, f"Invalid province: '{province}'" |
|
117
|
|
|
|
|
118
|
|
|
date_created = item.get_datetime("dataistituzione") |
|
119
|
|
|
date_deleted = item.get_datetime("datacessazione") |
|
120
|
|
|
date_deleted_raw = item.get_str("datacessazione") |
|
121
|
|
|
if "9999" in date_deleted_raw: |
|
122
|
|
|
date_deleted = "" |
|
123
|
|
|
|
|
124
|
|
|
return { |
|
125
|
|
|
"active": active, |
|
126
|
|
|
"code": code, |
|
127
|
|
|
"date_created": date_created, |
|
128
|
|
|
"date_deleted": date_deleted, |
|
129
|
|
|
"name": name, |
|
130
|
|
|
"name_trans": name_trans, |
|
131
|
|
|
"name_alt": name_alt, |
|
132
|
|
|
"name_alt_trans": name_alt_trans, |
|
133
|
|
|
"name_slugs": name_slugs, |
|
134
|
|
|
"province": province, |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
_write_data_json( |
|
138
|
|
|
filepath="../codicefiscale/data/municipalities.json", |
|
139
|
|
|
data=[map_item(benedict(item)) for item in data["values"]], |
|
140
|
|
|
) |
|
141
|
|
|
|
|
142
|
|
|
|
|
143
|
|
|
def _write_data_json(filepath, data): |
|
144
|
|
|
data = list(filter(bool, data)) |
|
145
|
|
|
data = sorted(data, key=lambda item: item["name"]) |
|
146
|
|
|
data_filepath = fsutil.join_path(__file__, filepath) |
|
147
|
|
|
fsutil.write_file_json(data_filepath, data, indent=4, sort_keys=True) |
|
148
|
|
|
|
|
149
|
|
|
|
|
150
|
|
|
def main(): |
|
151
|
|
|
_update_countries_data() |
|
152
|
|
|
_update_municipalities_data() |
|
153
|
|
|
|
|
154
|
|
|
|
|
155
|
|
|
if __name__ == "__main__": |
|
156
|
|
|
main() |
|
157
|
|
|
|
fabiocaccamo /
python-codicefiscale
