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

[addresses] Updates to unit directional probabilities and adding quadrant addresses for units like 1RF

Browse Source
This commit is contained in:
Al
2016-05-19 14:44:31 -04:00
parent 8756cbb652
commit e5f81a0f81
5 changed files with 127 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
resources/addresses/en.yaml
View File

@@ -719,8 +719,8 @@ directions:
affix: r
direction: right
whitespace_probability: 0.05
numeric_probability: 0.2
numeric_affix_probability: 0.8
numeric_probability: 0.05
numeric_affix_probability: 0.95
left: &left
canonical: left
abbreviated: l
@@ -732,8 +732,8 @@ directions:
affix: l
direction: right
whitespace_probability: 0.05
numeric_probability: 0.2
numeric_affix_probability: 0.8
numeric_probability: 0.05
numeric_affix_probability: 0.95
rear: &rear
canonical: rear
abbreviated: r
@@ -745,8 +745,8 @@ directions:
affix: r
direction: right
whitespace_probability: 0.05
numeric_probability: 0.2
numeric_affix_probability: 0.8
numeric_probability: 0.05
numeric_affix_probability: 0.95
front: &front
canonical: front
abbreviated: frnt
@@ -758,8 +758,8 @@ directions:
affix: f
direction: right
whitespace_probability: 0.05
numeric_probability: 0.2
numeric_affix_probability: 0.8
numeric_probability: 0.05
numeric_affix_probability: 0.95
alternatives:
- alternative: *right
probability: 0.45
@@ -770,6 +770,22 @@ directions:
- alternative: *rear
probability: 0.05
anteroposterior:
alternatives:
- alternative: *front
probability: 0.5
- alternative: *rear
probability: 0.5
lateral:
alternatives:
- alternative: *left
probability: 0.5
- alternative: *right
probability: 0.5
cardinal_directions:
east: &east
canonical: east
@@ -1184,10 +1200,19 @@ units:
# Separate random probability for adding directions like 2L, 2R, etc.
add_direction: true
add_direction_probability: 0.1
# Add directions only for plain numbers
# Add directions for plain numbers
add_direction_numeric: true
# Add direction only e.g. Unit Left
add_direction_standalone: true
# Separate random probability for adding quadrant units like 2RF 2RR 2LF 2LR
add_quadrant: true
add_quadrant_probability: 0.001
add_quadrant_first_direction: lateral
add_quadrant_numeric: true
# If there are 10 floors, create unit numbers like #301 or #1032
use_floor_probability: 0.35
@@ -1232,17 +1257,6 @@ units:
lot_plus_parcel_probability: 0.02
parcel_plus_lot_probability: 0.02
directional:
modifier:
direction: right # e.g. 1
numeric_probability: 0.1
numeric_affix_probability: 0.9
alternatives:
- alternative: *right
- alternative: *left
- alternative: *rear
- alternative: *front
standalone:
sample: false
default: *penthouse

22
resources/addresses/es.yaml
View File

@@ -442,8 +442,8 @@ directions:
affix: d
direction: right
whitespace_probability: 0.1
numeric_probability: 0.7
numeric_affix_probability: 0.3
numeric_probability: 0.4
numeric_affix_probability: 0.6
left: &izquierda
canonical: izquierda
abbreviated: izq
@@ -457,8 +457,8 @@ directions:
affix: i
direction: right
whitespace_probability: 0.1
numeric_probability: 0.7
numeric_affix_probability: 0.3
numeric_probability: 0.4
numeric_affix_probability: 0.6
rear: &trasera
canonical: trasera
abbreviated: tras
@@ -487,6 +487,20 @@ directions:
- alternative: *frente
probability: 0.05
anteroposterior:
alternatives:
- alternative: *frente
probability: 0.5
- alternative: *trasera
probability: 0.5
lateral:
alternatives:
- alternative: *derecha
probability: 0.5
- alternative: *izquierda
probability: 0.5
cardinal_directions:
east: &este
canonical: este

8
scripts/geodata/addresses/directions.py
View File

@@ -8,6 +8,14 @@ class RelativeDirection(NumericPhrase):
dictionaries = ['unit_directions']
class AnteroposteriorDirection(RelativeDirection):
key = 'directions.anteroposterior'
class LateralDirection(RelativeDirection):
key = 'directions.lateral'
class CardinalDirection(NumericPhrase):
key = 'cardinal_directions'
dictionaries = ['cardinal_directions']

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

@@ -41,8 +41,11 @@ latin_alphabet = [chr(i) for i in range(ord('A'), ord('Z') + 1)]
class NumericPhrase(object):
key = None
NUMERIC = 'numeric'
NUMERIC_AFFIX = 'numeric_affix'
@classmethod
def phrase(cls, number, language, country=None):
def pick_phrase_and_type(cls, number, language, country=None):
values, probs = address_config.alternative_probabilities(cls.key, language, dictionaries=cls.dictionaries, country=country)
if not values:
return safe_decode(number) if number is not None else None
@@ -52,7 +55,7 @@ class NumericPhrase(object):
values = []
probs = []
for num_type in ('numeric', 'numeric_affix'):
for num_type in (cls.NUMERIC, cls.NUMERIC_AFFIX):
key = '{}_probability'.format(num_type)
prob = phrase_props.get(key, None)
if prob is not None:
@@ -62,19 +65,16 @@ class NumericPhrase(object):
probs = cdf(probs)
if len(values) < 2:
num_type = 'numeric'
num_type = cls.NUMERIC
else:
num_type = weighted_choice(values, probs)
props = phrase_props[num_type]
return num_type, phrase, phrase_props[num_type]
if num_type == 'numeric':
# Numeric phrase the default is with whitespace e.g. "No 1"
whitespace_default = True
elif num_type == 'numeric_affix':
@classmethod
def combine_with_number(cls, number, phrase, num_type, props, whitespace_default=False):
if num_type == cls.NUMERIC_AFFIX:
phrase = props['affix']
# Numeric affix default is no whitespace e.g. "#1"
whitespace_default = False
direction = props['direction']
whitespace = props.get('whitespace', whitespace_default)
@@ -96,6 +96,12 @@ class NumericPhrase(object):
else:
return safe_decode(number)
@classmethod
def phrase(cls, number, language, country=None):
num_type, phrase, props = cls.pick_phrase_and_type(number, language, country=country)
whitespace_default = num_type == cls.NUMERIC
return cls.combine_with_number(number, phrase, num_type, props, whitespace_default=whitespace_default)
class Number(NumericPhrase):
key = 'numbers'

53
scripts/geodata/addresses/units.py
View File

@@ -2,7 +2,7 @@ import random
import six
from geodata.addresses.config import address_config
from geodata.addresses.directions import RelativeDirection
from geodata.addresses.directions import RelativeDirection, LateralDirection, AnteroposteriorDirection
from geodata.addresses.floors import Floor
from geodata.addresses.numbering import NumberedComponent, sample_alphabet, latin_alphabet
from geodata.configs.utils import nested_get
@@ -101,14 +101,63 @@ class Unit(NumberedComponent):
if add_direction_standalone:
return RelativeDirection.phrase(None, language, country=country)
@classmethod
def add_quadrant(cls, key, unit, language, country=None):
add_quadrant_probability = address_config.get_property('{}.add_quadrant_probability'.format(key),
language, country=country, default=0.0)
if not random.random() < add_quadrant_probability:
return unit
add_quadrant_numeric = address_config.get_property('{}.add_quadrant_numeric'.format(key),
language, country=country)
try:
unit = int(unit)
integer_unit = True
except (ValueError, TypeError):
integer_unit = False
first_direction = address_config.get_property('{}.add_quadrant_first_direction'.format(key),
language, country=country)
if first_direction == 'lateral':
ordering = (LateralDirection, AnteroposteriorDirection)
elif first_direction == 'anteroposterior':
ordering = (AnteroposteriorDirection, LateralDirection)
else:
return unit
if not integer_unit:
add_quadrant_standalone = address_config.get_property('{}.add_quadrant_standalone'.format(key),
language, country=country)
if add_quadrant_standalone:
unit = None
else:
return None
last_num_type = None
for i, c in enumerate(ordering):
num_type, phrase, props = c.pick_phrase_and_type(unit, language, country=country)
whitespace_default = num_type == c.NUMERIC or last_num_type == c.NUMERIC
unit = c.combine_with_number(unit, phrase, num_type, props, whitespace_default=whitespace_default)
last_num_type = num_type
return unit
@classmethod
def phrase(cls, unit, language, country=None, zone=None):
if unit is not None:
key = 'units.alphanumeric' if zone is None else 'units.zones.{}'.format(zone)
direction_unit = None
add_direction = address_config.get_property('{}.add_direction'.format(key), language, country=country)
if add_direction:
unit = cls.add_direction(key, unit, language, country=country)
direction_unit = cls.add_direction(key, unit, language, country=country)
if direction_unit and direction_unit != unit:
unit = direction_unit
else:
add_quadrant = address_config.get_property('{}.add_quadrant'.format(key), language, country=country)
if add_quadrant:
unit = cls.add_quadrant(key, unit, language, country=country)
return cls.numeric_phrase(key, safe_decode(unit), language,
dictionaries=['unit_types_numbered'], country=country)