bazarr/libs/pycountry/__init__.py

302 lines
10 KiB
Python

"""pycountry"""
import os.path
import unicodedata
from importlib import metadata as importlib_metadata
from typing import Dict, List, Optional, Type
import pycountry.db
# We prioritise importing the backported `importlib_resources`
# because the function we use (`importlib.resources.files`) is only
# available from Python 3.9, but the module itself exists since 3.7.
# We install `importlib_resources` on Python < 3.9.
# TODO: Remove usage of importlib_resources once support for Python 3.8 is dropped
try:
import importlib_resources # type: ignore
except ModuleNotFoundError:
from importlib import resources as importlib_resources
def resource_filename(package_or_requirement: str, resource_name: str) -> str:
return str(
importlib_resources.files(package_or_requirement) / resource_name
)
def get_version(distribution_name: str) -> Optional[str]:
try:
return importlib_metadata.version(distribution_name)
except importlib_metadata.PackageNotFoundError:
return "n/a"
# Variable annotations
LOCALES_DIR: str = resource_filename("pycountry", "locales")
DATABASE_DIR: str = resource_filename("pycountry", "databases")
__version__: Optional[str] = get_version("pycountry")
def remove_accents(input_str: str) -> str:
output_str = input_str
if not input_str.isascii():
# Borrowed from https://stackoverflow.com/a/517974/1509718
nfkd_form = unicodedata.normalize("NFKD", input_str)
output_str = "".join(
[c for c in nfkd_form if not unicodedata.combining(c)]
)
return output_str
class ExistingCountries(pycountry.db.Database):
"""Provides access to an ISO 3166 database (Countries)."""
data_class = pycountry.db.Country
root_key = "3166-1"
def search_fuzzy(self, query: str) -> List[Type["ExistingCountries"]]:
query = remove_accents(query.strip().lower())
# A country-code to points mapping for later sorting countries
# based on the query's matching incidence.
results: dict[str, int] = {}
def add_result(country: "pycountry.db.Country", points: int) -> None:
results.setdefault(country.alpha_2, 0)
results[country.alpha_2] += points
# Prio 1: exact matches on country names
try:
add_result(self.lookup(query), 50)
except LookupError:
pass
# Prio 2: exact matches on subdivision names
match_subdivions = pycountry.Subdivisions.match(
self=subdivisions, query=query
)
for candidate in match_subdivions:
print(candidate)
add_result(candidate.country, 49)
# Prio 3: partial matches on country names
for candidate in self:
# Higher priority for a match on the common name
for v in [
candidate._fields.get("name"),
candidate._fields.get("official_name"),
candidate._fields.get("comment"),
]:
if v is not None:
v = remove_accents(v.lower())
if query in v:
# This prefers countries with a match early in their name
# and also balances against countries with a number of
# partial matches and their name containing 'new' in the
# middle
add_result(
candidate, max([5, 30 - (2 * v.find(query))])
)
break
# Prio 4: partial matches on subdivision names
partial_match_subdivisions = pycountry.Subdivisions.partial_match(
self=subdivisions, query=query
)
for candidate in partial_match_subdivisions:
v = candidate._fields.get("name")
v = remove_accents(v.lower())
if query in v:
add_result(candidate.country, max([1, 5 - v.find(query)]))
if not results:
raise LookupError(query)
sorted_results = [
self.get(alpha_2=x[0])
# sort by points first, by alpha2 code second, and to ensure stable
# results the negative value allows us to sort reversely on the
# points but ascending on the country code.
for x in sorted(results.items(), key=lambda x: (-x[1], x[0]))
]
return sorted_results
class HistoricCountries(ExistingCountries):
"""Provides access to an ISO 3166-3 database
(Countries that have been removed from the standard)."""
data_class = pycountry.db.Country
root_key = "3166-3"
class Scripts(pycountry.db.Database):
"""Provides access to an ISO 15924 database (Scripts)."""
data_class = "Script"
root_key = "15924"
class Currencies(pycountry.db.Database):
"""Provides access to an ISO 4217 database (Currencies)."""
data_class = "Currency"
root_key = "4217"
class Languages(pycountry.db.Database):
"""Provides access to an ISO 639-1/2T/3 database (Languages)."""
no_index = ["status", "scope", "type", "inverted_name", "common_name"]
data_class = "Language"
root_key = "639-3"
class LanguageFamilies(pycountry.db.Database):
"""Provides access to an ISO 639-5 database
(Language Families and Groups)."""
data_class = "LanguageFamily"
root_key = "639-5"
class SubdivisionHierarchy(pycountry.db.Data):
def __init__(self, **kw):
if "parent" in kw:
kw["parent_code"] = kw["parent"]
else:
kw["parent_code"] = None
super().__init__(**kw)
self.country_code = self.code.split("-")[0]
if self.parent_code is not None:
self.parent_code = f"{self.country_code}-{self.parent_code}"
@property
def country(self):
return countries.get(alpha_2=self.country_code)
@property
def parent(self):
if not self.parent_code:
return None
return subdivisions.get(code=self.parent_code)
class Subdivisions(pycountry.db.Database):
# Note: subdivisions can be hierarchical to other subdivisions. The
# parent_code attribute is related to other subdivisions, *not*
# the country!
data_class = SubdivisionHierarchy
no_index = ["name", "parent_code", "parent", "type"]
root_key = "3166-2"
def _load(self, *args, **kw):
super()._load(*args, **kw)
# Add index for the country code.
self.indices["country_code"] = {}
for subdivision in self:
divs = self.indices["country_code"].setdefault(
subdivision.country_code.lower(), set()
)
divs.add(subdivision)
def get(self, **kw):
default = kw.setdefault("default", None)
subdivisions = super().get(**kw)
if subdivisions is default and "country_code" in kw:
# This handles the case where we know about a country but there
# are no subdivisions: we return an empty list in this case
# (sticking to the expected type here) instead of None.
if countries.get(alpha_2=kw["country_code"]) is not None:
return []
return subdivisions
def match(self, query):
query = remove_accents(query.strip().lower())
matching_candidates = []
for candidate in subdivisions:
for v in candidate._fields.values():
if v is not None:
v = remove_accents(v.lower())
# Some names include alternative versions which we want to
# match exactly.
for w in v.split(";"):
if w == query:
matching_candidates.append(candidate)
break
return matching_candidates
def partial_match(self, query):
query = remove_accents(query.strip().lower())
matching_candidates = []
for candidate in subdivisions:
v = candidate._fields.get("name")
v = remove_accents(v.lower())
if query in v:
matching_candidates.append(candidate)
return matching_candidates
def search_fuzzy(self, query: str) -> List[Type["Subdivisions"]]:
query = remove_accents(query.strip().lower())
# A Subdivision's code to points mapping for later sorting subdivisions
# based on the query's matching incidence.
results: dict[str, int] = {}
def add_result(
subdivision: "pycountry.db.Subdivision", points: int
) -> None:
results.setdefault(subdivision.code, 0)
results[subdivision.code] += points
# Prio 1: exact matches on subdivision names
match_subdivisions = self.match(query)
for candidate in match_subdivisions:
add_result(candidate, 50)
# Prio 2: partial matches on subdivision names
partial_match_subdivisions = self.partial_match(query)
for candidate in partial_match_subdivisions:
v = candidate._fields.get("name")
v = remove_accents(v.lower())
if query in v:
add_result(candidate, max([1, 5 - v.find(query)]))
if not results:
raise LookupError(query)
sorted_results = [
self.get(code=x[0])
# sort by points first, by alpha2 code second, and to ensure stable
# results the negative value allows us to sort reversely on the
# points but ascending on the country code.
for x in sorted(results.items(), key=lambda x: (-x[1], x[0]))
]
return sorted_results
# Initialize instances with type hints
countries: ExistingCountries = ExistingCountries(
os.path.join(DATABASE_DIR, "iso3166-1.json")
)
subdivisions: Subdivisions = Subdivisions(
os.path.join(DATABASE_DIR, "iso3166-2.json")
)
historic_countries: HistoricCountries = HistoricCountries(
os.path.join(DATABASE_DIR, "iso3166-3.json")
)
currencies: Currencies = Currencies(os.path.join(DATABASE_DIR, "iso4217.json"))
languages: Languages = Languages(os.path.join(DATABASE_DIR, "iso639-3.json"))
language_families: LanguageFamilies = LanguageFamilies(
os.path.join(DATABASE_DIR, "iso639-5.json")
)
scripts: Scripts = Scripts(os.path.join(DATABASE_DIR, "iso15924.json"))