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

[addresses] Implementing whitespace_probability and ordinal_suffix probability for Roman numerals

Browse Source
This commit is contained in:
Al
2016-07-04 18:08:55 -04:00
parent 2810e2638d
commit d09e0ca966
3 changed files with 22 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
scripts/geodata/addresses/numbering.py
View File

@@ -125,7 +125,12 @@ class Digits(object):
elif digit_type == cls.SPELLOUT:
return cls.rewrite_spellout(d, lang, num_type, props)
elif digit_type == cls.ROMAN_NUMERAL:
return cls.rewrite_roman_numeral(d)
roman_numeral = cls.rewrite_roman_numeral(d)
if random.random() < props.get('ordinal_suffix_probability', 0.0):
ordinal_suffix = ordinal_expressions.get_suffix(d, lang, gender=props.get('gender', None))
if ordinal_suffix:
roman_numeral = six.u('{}{}').format(roman_numeral, ordinal_suffix)
return roman_numeral
elif digit_type == cls.UNICODE_FULL_WIDTH:
return cls.rewrite_full_width(d)
else:
@@ -175,6 +180,10 @@ class NumericPhrase(object):
direction = props['direction']
whitespace = props.get('whitespace', whitespace_default)
whitespace_probability = props.get('whitespace_probability')
if whitespace_probability is not None:
whitespace = random.random() < whitespace_probability
if props.get('title_case', True):
# Title case unless the config specifies otherwise
phrase = phrase.title()
@@ -391,6 +400,10 @@ class NumberedComponent(object):
direction = props['direction']
whitespace = props.get('whitespace', whitespace_default)
whitespace_probability = props.get('whitespace_probability')
if whitespace_probability is not None:
whitespace = random.random() < whitespace_probability
# Occasionally switch up if direction_probability is specified
if random.random() > props.get('direction_probability', 1.0):
if direction == 'left':

9
scripts/geodata/numbers/ordinals.py
View File

@@ -93,11 +93,16 @@ class OrdinalExpressions(object):
return trie.search_suffix(str(num))
def suffixed_number(self, num, lang, gender=None, category=None):
def get_suffix(self, num, lang, gender=None, category=None):
suffixes = self.get_suffixes(num, lang, gender=gender, category=category)
if not suffixes:
return None
suffix = random.choice(suffixes)
return random.choice(suffixes)
def suffixed_number(self, num, lang, gender=None, category=None):
suffix = self.get_suffix(num, lang, gender=gender, category=category)
if not suffix:
return None
return six.u('{}{}').format(safe_decode(num), safe_decode(suffix))
ordinal_expressions = OrdinalExpressions()