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

[numex] Adding validation checks for numex JSON

Browse Source
This commit is contained in:
Al
2015-07-24 15:21:52 -04:00
parent 96538469dd
commit 87566bb6a5
1 changed files with 14 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
scripts/geodata/i18n/numex.py
View File

@@ -10,6 +10,9 @@ from geodata.encoding import safe_encode
from unicode_paths import DATA_DIR from unicode_paths import DATA_DIR
class InvalidNumexRuleException(Exception):
pass
NUMEX_DATA_DIR = os.path.join(DATA_DIR, 'numex') NUMEX_DATA_DIR = os.path.join(DATA_DIR, 'numex')
NUMEX_RULES_FILE = os.path.join(os.pardir, os.pardir, os.pardir, 'src', 'numex_data.c') NUMEX_RULES_FILE = os.path.join(os.pardir, os.pardir, os.pardir, 'src', 'numex_data.c')
@@ -30,6 +33,11 @@ gender_map = {
CATEGORY_PLURAL = 'CATEGORY_PLURAL' CATEGORY_PLURAL = 'CATEGORY_PLURAL'
CATEGORY_DEFAULT = 'CATEGORY_DEFAULT' CATEGORY_DEFAULT = 'CATEGORY_DEFAULT'
valid_numex_keys = set(['name', 'value', 'type', 'left', 'right', 'gender', 'category', 'radix'])
valid_ordinal_keys = set(['suffixes', 'gender', 'category'])
category_map = { category_map = {
'plural': CATEGORY_PLURAL, 'plural': CATEGORY_PLURAL,
None: CATEGORY_DEFAULT None: CATEGORY_DEFAULT
@@ -114,6 +122,9 @@ def parse_numex_rules(dirname=NUMEX_DATA_DIR, outfile=NUMEX_RULES_FILE):
rule_index = len(all_rules) rule_index = len(all_rules)
for rule in rules: for rule in rules:
invalid_keys = set(rule.keys()) - valid_numex_keys
if invalid_keys:
raise InvalidNumexRuleException(u'Invalid keys: ({}) for language {}, rule: {}'.format(u','.join(invalid_keys), language, rule))
gender = gender_map[rule.get('gender')] gender = gender_map[rule.get('gender')]
rule_type = rule_type_map[rule['type']] rule_type = rule_type_map[rule['type']]
key = rule['name'] key = rule['name']
@@ -141,8 +152,9 @@ def parse_numex_rules(dirname=NUMEX_DATA_DIR, outfile=NUMEX_RULES_FILE):
for rule in ordinal_indicators: for rule in ordinal_indicators:
gender = gender_map[rule.get('gender')] gender = gender_map[rule.get('gender')]
category = category_map[rule.get('category')] category = category_map[rule.get('category')]
if 'suffixes' not in rule: invalid_ordinal_keys = set(rule.keys()) - valid_ordinal_keys
print rule.keys() if invalid_ordinal_keys:
raise InvalidNumexRuleException(u'Invalid keys ({}) in ordinal rule for language {}, rule: {}'.format(u','.join(invalid_ordinal_keys), language, rule))
for key, suffixes in rule['suffixes'].iteritems(): for key, suffixes in rule['suffixes'].iteritems():
for suffix in suffixes: for suffix in suffixes: