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

[intersections] Abbreviating street names that are not base names with random probabilities

Browse Source
This commit is contained in:
Al
2016-08-19 23:27:29 -04:00
parent 42808c62e3
commit 7951044d74
1 changed files with 11 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
scripts/geodata/osm/formatter.py
View File

@@ -1138,10 +1138,10 @@ class OSMAddressFormatter(object):
if address_language and address_language not in (UNKNOWN_LANGUAGE, AMBIGUOUS_LANGUAGE): if address_language and address_language not in (UNKNOWN_LANGUAGE, AMBIGUOUS_LANGUAGE):
default_language = address_language default_language = address_language
names[lang].append(way[tag]) names[lang].append((way[tag], False))
if base_name_tag in way and default_language: if base_name_tag in way and default_language:
names[default_language].append(way[base_name_tag]) names[default_language].append((way[base_name_tag], True))
if not names: if not names:
continue continue
@@ -1158,9 +1158,9 @@ class OSMAddressFormatter(object):
language_components[namespaced_language] = address_components language_components[namespaced_language] = address_components
for way1, way2 in itertools.combinations(way_names, 2): for way1, way2 in itertools.combinations(way_names, 2):
formatted_intersections = [] intersection_phrases = []
for language in set(way1.keys()) & set(way2.keys()): for language in set(way1.keys()) & set(way2.keys()):
for w1, w2 in itertools.product(way1[language], way2[language]): for (w1, w1_is_base), (w2, w2_is_base) in itertools.product(way1[language], way2[language]):
address_components = language_components[language] address_components = language_components[language]
intersection_phrase = Intersection.phrase(language, country=country) intersection_phrase = Intersection.phrase(language, country=country)
@@ -1170,7 +1170,14 @@ class OSMAddressFormatter(object):
w1 = self.components.cleaned_name(w1) w1 = self.components.cleaned_name(w1)
w2 = self.components.cleaned_name(w2) w2 = self.components.cleaned_name(w2)
if not w1_is_base:
w1 = self.abbreviated_street(w1, language)
if not w2_is_base:
w2 = self.abbreviated_street(w2, langguage)
intersection = IntersectionQuery(road1=w1, intersection_phrase=intersection_phrase, road2=w2) intersection = IntersectionQuery(road1=w1, intersection_phrase=intersection_phrase, road2=w2)
formatted = self.formatter.format_intersection(intersection, address_components, country, language, tag_components=tag_components) formatted = self.formatter.format_intersection(intersection, address_components, country, language, tag_components=tag_components)
if not formatted or not formatted.strip(): if not formatted or not formatted.strip():