safety-screen/Nominatim
1
0
Fork 0
forked from hans/Nominatim
Code Pull Requests Activity

export more data for the tokenizer name preparation

Browse Source 
Adds class, type, country and rank to the exported information
and removes the rather odd hack for countries. Whether a place
represents a country boundary can now be computed by the tokenizer.
This commit is contained in:
Sarah Hoffmann
2021-09-29 11:54:14 +02:00
parent 231250f2eb
commit be65c8303f
7 changed files with 85 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
test/python/test_indexing.py
View File

@@ -29,6 +29,7 @@ class IndexerTestDB:
indexed_date TIMESTAMP,
partition SMALLINT,
admin_level SMALLINT,
country_code TEXT,
address HSTORE,
token_info JSONB,
geometry_sector INTEGER)""")
@@ -54,15 +55,26 @@ class IndexerTestDB:
END IF;
RETURN NEW;
END; $$ LANGUAGE plpgsql;""")
cur.execute("""CREATE OR REPLACE FUNCTION placex_prepare_update(p placex,
OUT name HSTORE,
OUT address HSTORE,
OUT country_feature VARCHAR,
OUT linked_place_id BIGINT)
cur.execute("DROP TYPE IF EXISTS prepare_update_info CASCADE")
cur.execute("""CREATE TYPE prepare_update_info AS (
name HSTORE,
address HSTORE,
rank_address SMALLINT,
country_code TEXT,
class TEXT,
type TEXT,
linked_place_id BIGINT
)""")
cur.execute("""CREATE OR REPLACE FUNCTION placex_indexing_prepare(p placex,
OUT result prepare_update_info)
AS $$
BEGIN
address := p.address;
name := p.name;
result.address := p.address;
result.name := p.name;
result.class := p.class;
result.type := p.type;
result.country_code := p.country_code;
result.rank_address := p.rank_address;
END;
$$ LANGUAGE plpgsql STABLE;
""")

12
test/python/test_tokenizer_icu.py
View File

@@ -323,10 +323,8 @@ class TestPlaceNames:
assert eval(info['names']) == set((t[2] for t in tokens))
def process_named_place(self, names, country_feature=None):
def process_named_place(self, names):
place = {'name': names}
if country_feature:
place['country_feature'] = country_feature
return self.analyzer.process_place(PlaceInfo(place))
@@ -353,7 +351,13 @@ class TestPlaceNames:
def test_country_name(self, word_table):
info = self.process_named_place({'name': 'Norge'}, country_feature='no')
place = PlaceInfo({'name' : {'name': 'Norge'},
'country_code': 'no',
'rank_address': 4,
'class': 'boundary',
'type': 'administrative'})
info = self.analyzer.process_place(place)
self.expect_name_terms(info, '#norge', 'norge')
assert word_table.get_country() == {('no', 'NORGE')}