[numex] Ordinal spellout using the numex configs
This commit is contained in:
@@ -1,5 +1,7 @@
|
|||||||
import bisect
|
import bisect
|
||||||
|
import math
|
||||||
import os
|
import os
|
||||||
|
import random
|
||||||
import six
|
import six
|
||||||
import yaml
|
import yaml
|
||||||
|
|
||||||
@@ -59,7 +61,7 @@ class NumericExpressions(object):
|
|||||||
|
|
||||||
self.cardinal_rules_ones = dict(self.cardinal_rules_ones)
|
self.cardinal_rules_ones = dict(self.cardinal_rules_ones)
|
||||||
|
|
||||||
def spellout_cardinal(self, num, lang, gender=None, category=None):
|
def spellout_cardinal(self, num, lang, gender=None, category=None, random_choice_cardinals=False):
|
||||||
num = int(num)
|
num = int(num)
|
||||||
remainder = 0
|
remainder = 0
|
||||||
|
|
||||||
@@ -73,6 +75,17 @@ class NumericExpressions(object):
|
|||||||
|
|
||||||
default_separator = self.default_separators.get(lang, self.default_separator)
|
default_separator = self.default_separators.get(lang, self.default_separator)
|
||||||
|
|
||||||
|
if num == 0:
|
||||||
|
cardinal = rules.get((num, gender, category))
|
||||||
|
if cardinal:
|
||||||
|
if not random_choice_cardinals:
|
||||||
|
cardinal = cardinal[0]
|
||||||
|
else:
|
||||||
|
cardinal = random.choice(cardinal)
|
||||||
|
return cardinal['name']
|
||||||
|
else:
|
||||||
|
return None
|
||||||
|
|
||||||
cardinal_part = []
|
cardinal_part = []
|
||||||
|
|
||||||
last_rule = {}
|
last_rule = {}
|
||||||
@@ -99,7 +112,10 @@ class NumericExpressions(object):
|
|||||||
if multiple > 1:
|
if multiple > 1:
|
||||||
multiple_val = rules.get((multiple, None, None))
|
multiple_val = rules.get((multiple, None, None))
|
||||||
if multiple_val:
|
if multiple_val:
|
||||||
multiple_rule = multiple_val[0]
|
if not random_choice_cardinals:
|
||||||
|
multiple_rule = multiple_val[0]
|
||||||
|
else:
|
||||||
|
multiple_rule = random.choice(multiple_val)
|
||||||
elif multiple == 1 and lang in self.cardinal_rules_ones_sorted:
|
elif multiple == 1 and lang in self.cardinal_rules_ones_sorted:
|
||||||
ones_rules = self.cardinal_rules_ones_sorted[lang]
|
ones_rules = self.cardinal_rules_ones_sorted[lang]
|
||||||
j = bisect.bisect_right(ones_rules, val)
|
j = bisect.bisect_right(ones_rules, val)
|
||||||
@@ -112,7 +128,11 @@ class NumericExpressions(object):
|
|||||||
did_left_multiply = False
|
did_left_multiply = False
|
||||||
|
|
||||||
if not use_multiple:
|
if not use_multiple:
|
||||||
rule = cardinal[0] if cardinal else None
|
rule = None
|
||||||
|
if cardinal and not random_choice_cardinals:
|
||||||
|
rule = cardinal[0]
|
||||||
|
elif cardinal:
|
||||||
|
rule = random.choice(cardinal)
|
||||||
else:
|
else:
|
||||||
for rule in cardinal:
|
for rule in cardinal:
|
||||||
left_multiply = rule.get('left') == 'multiply'
|
left_multiply = rule.get('left') == 'multiply'
|
||||||
@@ -177,4 +197,199 @@ class NumericExpressions(object):
|
|||||||
return None
|
return None
|
||||||
return numeral.upper()
|
return numeral.upper()
|
||||||
|
|
||||||
|
def spellout_ordinal(self, num, lang, gender=None, category=None,
|
||||||
|
random_choice_cardinals=False, random_choice_ordinals=False):
|
||||||
|
num = int(num)
|
||||||
|
remainder = 0
|
||||||
|
|
||||||
|
if lang not in self.cardinal_rules:
|
||||||
|
return None
|
||||||
|
|
||||||
|
rules = self.ordinal_rules.get(lang)
|
||||||
|
cardinal_rules = self.cardinal_rules.get(lang)
|
||||||
|
cardinals = self.cardinal_rules_sorted.get(lang)
|
||||||
|
if not rules or not cardinal_rules or not cardinals:
|
||||||
|
return None
|
||||||
|
|
||||||
|
default_separator = self.default_separators.get(lang, self.default_separator)
|
||||||
|
|
||||||
|
expression = []
|
||||||
|
|
||||||
|
last_rule = {}
|
||||||
|
left_multiply_rules = []
|
||||||
|
|
||||||
|
if num == 0:
|
||||||
|
ordinals = rules.get((num, gender, category))
|
||||||
|
if ordinals:
|
||||||
|
if not random_choice_ordinals:
|
||||||
|
ordinal = ordinals[0]
|
||||||
|
else:
|
||||||
|
ordinal = random.choice(ordinals)
|
||||||
|
return ordinal['name']
|
||||||
|
else:
|
||||||
|
return None
|
||||||
|
|
||||||
|
while num:
|
||||||
|
i = bisect.bisect_left(cardinals, num)
|
||||||
|
if i > len(cardinals) - 1:
|
||||||
|
return None
|
||||||
|
if i > 0 and cardinals[i] > num:
|
||||||
|
val = cardinals[i - 1]
|
||||||
|
else:
|
||||||
|
val = cardinals[i]
|
||||||
|
|
||||||
|
if val == num and not remainder:
|
||||||
|
if last_rule.get('right') == 'add':
|
||||||
|
ordinals = rules.get((num, gender, category))
|
||||||
|
if ordinals:
|
||||||
|
if not random_choice_ordinals:
|
||||||
|
ordinal = ordinals[0]
|
||||||
|
else:
|
||||||
|
ordinal = random.choice(ordinals)
|
||||||
|
right_separator = last_rule.get('right_separator', default_separator)
|
||||||
|
|
||||||
|
return right_separator.join([six.u('').join(expression), ordinal['name']])
|
||||||
|
else:
|
||||||
|
return None
|
||||||
|
elif last_rule.get('left') == 'add':
|
||||||
|
last_num = last_rule['value']
|
||||||
|
ordinals = rules.get((last_num, gender, category))
|
||||||
|
if ordinals:
|
||||||
|
if not random_choice_ordinals:
|
||||||
|
ordinal = ordinals[0]
|
||||||
|
else:
|
||||||
|
ordinal = random.choice(ordinals)
|
||||||
|
|
||||||
|
last_rule = ordinal
|
||||||
|
expression.pop()
|
||||||
|
cardinals = cardinal_rules.get((num, None, None))
|
||||||
|
if cardinals:
|
||||||
|
if not random_choice_cardinals:
|
||||||
|
rule = cardinals[0]
|
||||||
|
else:
|
||||||
|
rule = random.choice(cardinals)
|
||||||
|
expression.append(rule['name'])
|
||||||
|
else:
|
||||||
|
return None
|
||||||
|
last = ordinal['name']
|
||||||
|
left_separator = last_rule.get('left_separator', default_separator)
|
||||||
|
return left_separator.join([six.u('').join(expression), ordinal['name']])
|
||||||
|
else:
|
||||||
|
return None
|
||||||
|
else:
|
||||||
|
return None
|
||||||
|
else:
|
||||||
|
ordinal = rules.get((val, None, None), [])
|
||||||
|
cardinal = cardinal_rules.get((val, None, None), [])
|
||||||
|
|
||||||
|
multiple = num // val
|
||||||
|
|
||||||
|
multiple_rule = None
|
||||||
|
|
||||||
|
if multiple > 1:
|
||||||
|
multiple_val = cardinal_rules.get((multiple, None, None))
|
||||||
|
if multiple_val:
|
||||||
|
if not random_choice_cardinals:
|
||||||
|
multiple_rule = multiple_val[0]
|
||||||
|
else:
|
||||||
|
multiple_rule = random.choice(multiple_val)
|
||||||
|
elif multiple == 1 and lang in self.cardinal_rules_ones_sorted:
|
||||||
|
ones_rules = self.cardinal_rules_ones_sorted[lang]
|
||||||
|
j = bisect.bisect_right(ones_rules, val)
|
||||||
|
if j > 0 and ones_rules[j - 1] <= num:
|
||||||
|
multiple_rule = self.cardinal_rules_ones[lang][ones_rules[j - 1]]
|
||||||
|
|
||||||
|
use_multiple = multiple > 1
|
||||||
|
|
||||||
|
is_left_multiply = False
|
||||||
|
did_left_multiply = False
|
||||||
|
|
||||||
|
if not use_multiple:
|
||||||
|
rule = None
|
||||||
|
if ordinal and not remainder:
|
||||||
|
for rule in ordinal:
|
||||||
|
if rule.get('right') == 'add':
|
||||||
|
break
|
||||||
|
else:
|
||||||
|
rule = None
|
||||||
|
|
||||||
|
if not rule and cardinal and not random_choice_cardinals:
|
||||||
|
rule = cardinal[0]
|
||||||
|
elif not rule and cardinal:
|
||||||
|
rule = random.choice(cardinal)
|
||||||
|
else:
|
||||||
|
rule = None
|
||||||
|
have_ordinal = False
|
||||||
|
if ordinal:
|
||||||
|
for rule in ordinal:
|
||||||
|
left_multiply = rule.get('left') == 'multiply'
|
||||||
|
if left_multiply and rule.get('right') == 'add':
|
||||||
|
if not multiple_rule:
|
||||||
|
left_multiply_rules.append(rule)
|
||||||
|
is_left_multiply = True
|
||||||
|
last_rule = rule
|
||||||
|
rule = None
|
||||||
|
have_ordinal = True
|
||||||
|
break
|
||||||
|
else:
|
||||||
|
rule = None
|
||||||
|
|
||||||
|
if not have_ordinal:
|
||||||
|
for rule in cardinal:
|
||||||
|
left_multiply = rule.get('left') == 'multiply'
|
||||||
|
if left_multiply:
|
||||||
|
if not multiple_rule:
|
||||||
|
left_multiply_rules.append(rule)
|
||||||
|
is_left_multiply = True
|
||||||
|
last_rule = rule
|
||||||
|
rule = None
|
||||||
|
break
|
||||||
|
else:
|
||||||
|
rule = None
|
||||||
|
|
||||||
|
if rule is not None:
|
||||||
|
left_add = last_rule.get('left') == 'add'
|
||||||
|
right_add = last_rule.get('right') == 'add'
|
||||||
|
|
||||||
|
if multiple_rule:
|
||||||
|
if right_add and expression:
|
||||||
|
expression.append(last_rule.get('left_separator', default_separator))
|
||||||
|
expression.append(multiple_rule['name'])
|
||||||
|
expression.append(rule.get('left_separator', default_separator))
|
||||||
|
|
||||||
|
if right_add:
|
||||||
|
if not multiple_rule and expression:
|
||||||
|
right_separator = last_rule.get('right_separator', default_separator)
|
||||||
|
expression.append(right_separator)
|
||||||
|
expression.append(rule['name'])
|
||||||
|
elif left_add and expression:
|
||||||
|
last = expression.pop()
|
||||||
|
expression.append(rule['name'])
|
||||||
|
left_separator = last_rule.get('left_separator', default_separator)
|
||||||
|
expression.append(left_separator)
|
||||||
|
expression.append(last)
|
||||||
|
elif not left_add and not right_add:
|
||||||
|
expression.append(rule['name'])
|
||||||
|
|
||||||
|
last_rule = rule
|
||||||
|
|
||||||
|
if left_multiply_rules and 'right' not in rule and 'left' not in rule:
|
||||||
|
left_multiply_rule = left_multiply_rules.pop()
|
||||||
|
print 'left_multiply_rule', left_multiply_rule
|
||||||
|
left_separator = left_multiply_rule.get('left_separator', default_separator)
|
||||||
|
expression.append(left_separator)
|
||||||
|
expression.append(left_multiply_rule['name'])
|
||||||
|
did_left_multiply = True
|
||||||
|
last_rule = left_multiply_rule
|
||||||
|
|
||||||
|
if not is_left_multiply and not did_left_multiply:
|
||||||
|
num -= (multiple * val)
|
||||||
|
elif not did_left_multiply:
|
||||||
|
remainder = num % val
|
||||||
|
num /= val
|
||||||
|
else:
|
||||||
|
num = remainder
|
||||||
|
remainder = 0
|
||||||
|
did_left_multiply = False
|
||||||
|
|
||||||
numeric_expressions = NumericExpressions()
|
numeric_expressions = NumericExpressions()
|
||||||
|
|||||||
Reference in New Issue
Block a user