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

[addresses] Adding null-phrase/null-phrase-alpha-only handling and zero padding to numbered components in sub-building configs

Browse Source
This commit is contained in:
Al
2016-06-14 21:53:43 +02:00
parent e6ac8062d8
commit f02d393b90
1 changed files with 16 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
scripts/geodata/addresses/numbering.py
View File

@@ -75,6 +75,8 @@ class NumericPhrase(object):
def combine_with_number(cls, number, phrase, num_type, props, whitespace_default=False): def combine_with_number(cls, number, phrase, num_type, props, whitespace_default=False):
if num_type == cls.NUMERIC_AFFIX: if num_type == cls.NUMERIC_AFFIX:
phrase = props['affix'] phrase = props['affix']
if 'zero_pad' in props and number.isdigit():
number = number.rjust(props['zero_pad'], props.get('zero_char', '0'))
direction = props['direction'] direction = props['direction']
whitespace = props.get('whitespace', whitespace_default) whitespace = props.get('whitespace', whitespace_default)
@@ -139,6 +141,7 @@ class NumberedComponent(object):
@classmethod @classmethod
def numeric_phrase(cls, key, num, language, country=None, dictionaries=(), strict_numeric=False, is_alpha=False): def numeric_phrase(cls, key, num, language, country=None, dictionaries=(), strict_numeric=False, is_alpha=False):
has_alpha = False has_alpha = False
has_numeric = True
is_none = False is_none = False
if num is not None: if num is not None:
try: try:
@@ -147,10 +150,16 @@ class NumberedComponent(object):
try: try:
num = float(num) num = float(num)
except ValueError: except ValueError:
if not all((c == token_types.NUMERIC) for t, c in tokenize(safe_decode(num))): tokens = tokenize(safe_decode(num))
if strict_numeric: has_numeric = False
return safe_decode(num) for t, c in tokens:
has_alpha = True if c == token_types.NUMERIC:
has_numeric = True
if t.isalnum():
has_alpha = True
if strict_numeric and has_alpha:
return safe_decode(num)
else: else:
is_none = True is_none = True
@@ -253,11 +262,13 @@ class NumberedComponent(object):
phrase = props['affix'] phrase = props['affix']
if props.get('upper_case', True): if props.get('upper_case', True):
phrase = phrase.upper() phrase = phrase.upper()
if 'zero_pad' in props and num.isdigit():
num = num.rjust(props['zero_pad'], props.get('zero_char', '0'))
whitespace_default = False whitespace_default = False
elif num_type == 'ordinal' and safe_decode(num).isdigit(): elif num_type == 'ordinal' and safe_decode(num).isdigit():
num = ordinal_expressions.suffixed_number(num, language, gender=props.get('gender', None)) num = ordinal_expressions.suffixed_number(num, language, gender=props.get('gender', None))
if (num_type == 'ordinal' or has_alpha) and 'null_phrase_probability' in props: if 'null_phrase_probability' in props and (num_type == 'ordinal' or (has_alpha and (has_numeric or 'null_phrase_alpha_only' in props))):
if random.random() < props['null_phrase_probability']: if random.random() < props['null_phrase_probability']:
return num return num