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

[categories] Method to determine which keys to check for category matches given a dict of OSM tags

Browse Source
This commit is contained in:
Al
2016-05-22 05:12:46 -04:00
parent 448f010b22
commit b061c619d3
1 changed files with 6 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
scripts/geodata/categories/config.py
View File

@@ -20,7 +20,7 @@ class CategoryConfig(object):
self.language_categories_singular = {} self.language_categories_singular = {}
self.language_categories_plural = {} self.language_categories_plural = {}
self.property_names = set() self.language_property_names = defaultdict(set)
if not os.path.exists(base_dir): if not os.path.exists(base_dir):
raise RuntimeError('{} does not exist'.format(base_dir)) raise RuntimeError('{} does not exist'.format(base_dir))
@@ -39,7 +39,7 @@ class CategoryConfig(object):
reader.next() # headers reader.next() # headers
for key, value, is_plural, phrase in reader: for key, value, is_plural, phrase in reader:
self.property_names.add(key) self.language_property_names[lang].add(key)
is_plural = bool(int(is_plural)) is_plural = bool(int(is_plural))
if is_plural: if is_plural:
plural_rules[(key, value)].append(phrase) plural_rules[(key, value)].append(phrase)
@@ -55,6 +55,10 @@ class CategoryConfig(object):
self.language_categories_plural = {key: dict(value) for key, value self.language_categories_plural = {key: dict(value) for key, value
in six.iteritems(self.language_categories_plural)} in six.iteritems(self.language_categories_plural)}
def has_keys(self, language, keys):
prop_names = self.language_property_names.get(language, set())
return [k for k in keys if k in prop_names]
def get_phrase(self, language, key, value, is_plural=False): def get_phrase(self, language, key, value, is_plural=False):
config = self.language_categories_singular if not is_plural else self.language_categories_plural config = self.language_categories_singular if not is_plural else self.language_categories_plural
if language not in config: if language not in config: