tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[units] Refactoring unit generator to use base alphanumeric generator

Browse Source
This commit is contained in:
Al
2016-05-05 18:47:24 -04:00
parent ec0779aaea
commit 2db255c55f
1 changed files with 6 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
scripts/geodata/addresses/units.py
View File

@@ -50,43 +50,29 @@ class Unit(NumberedComponent):
@classmethod @classmethod
def random(cls, language, country=None, num_floors=None, num_basements=None): def random(cls, language, country=None, num_floors=None, num_basements=None):
unit_props = address_config.get_property('units.alphanumeric', language, country=country) num_type, num_type_props = cls.choose_alphanumeric_type('units.alphanumeric', language, country=country)
values = []
probs = []
for num_type in ('numeric', 'alpha', 'alpha_plus_numeric', 'numeric_plus_alpha'):
key = '{}_probability'.format(num_type)
prob = unit_props.get(key)
if prob is not None:
values.append(num_type)
probs.append(prob)
probs = cdf(probs)
num_type = weighted_choice(values, probs)
num_type_props = unit_props.get(num_type, {})
if num_floors is not None: if num_floors is not None:
number = cls.for_floor(Floor.sample_positive_floors(num_floors)) number = cls.for_floor(Floor.sample_positive_floors(num_floors))
else: else:
number = weighted_choice(cls.numbered_units, cls.unit_probs_cdf) number = weighted_choice(cls.numbered_units, cls.unit_probs_cdf)
if num_type == 'numeric': if num_type == cls.NUMERIC:
return safe_decode(number) return safe_decode(number)
else: else:
alphabet = address_config.get_property('alphabet', language, country=country, default=latin_alphabet) alphabet = address_config.get_property('alphabet', language, country=country, default=latin_alphabet)
letter = sample_alphabet(alphabet) letter = sample_alphabet(alphabet)
if num_type == 'alpha': if num_type == cls.ALPHA:
return safe_decode(letter) return safe_decode(letter)
else: else:
number = weighted_choice(cls.positive_units_letters, cls.positive_units_letters_cdf) number = weighted_choice(cls.positive_units_letters, cls.positive_units_letters_cdf)
whitespace_probability = unit_props.get('{}_whitespace_probability'.format(num_type)) whitespace_probability = num_type_props.get('{}_whitespace_probability'.format(num_type))
whitespace_phrase = six.u(' ') if whitespace_probability and random.random() < whitespace_probability else six.u('') whitespace_phrase = six.u(' ') if whitespace_probability and random.random() < whitespace_probability else six.u('')
if num_type == 'alpha_plus_numeric': if num_type == cls.ALPHA_PLUS_NUMERIC:
return six.u('{}{}{}').format(letter, whitespace_phrase, number) return six.u('{}{}{}').format(letter, whitespace_phrase, number)
elif num_type == 'numeric_plus_alpha': elif num_type == cls.NUMERIC_PLUS_ALPHA:
return six.u('{}{}{}').format(number, whitespace_phrase, letter) return six.u('{}{}{}').format(number, whitespace_phrase, letter)
@classmethod @classmethod