]> git.openstreetmap.org Git - nominatim.git/commitdiff
Merge remote-tracking branch 'upstream/master'
authorSarah Hoffmann <lonvia@denofr.de>
Tue, 26 Mar 2024 21:44:04 +0000 (22:44 +0100)
committerSarah Hoffmann <lonvia@denofr.de>
Tue, 26 Mar 2024 21:44:04 +0000 (22:44 +0100)
nominatim/api/search/db_search_builder.py
settings/import-extratags.lua
settings/import-full.lua

index 97e7ac0282a79b40e7015bcd1069c8edbadeb09e..e27a24d61eb54f0d7bb1bc04abedc90703dacc8a 100644 (file)
@@ -166,7 +166,7 @@ class SearchBuilder:
         sdata.lookups = [dbf.FieldLookup('name_vector', [t.token for t in hnrs], lookups.LookupAny)]
         expected_count = sum(t.count for t in hnrs)
 
-        partials = {t.token: t.count for trange in address
+        partials = {t.token: t.addr_count for trange in address
                        for t in self.query.get_partials_list(trange)}
 
         if expected_count < 8000:
@@ -222,6 +222,7 @@ class SearchBuilder:
             yield penalty, exp_count, dbf.lookup_by_names(list(name_partials.keys()), addr_tokens)
             return
 
+        addr_count = min(t.addr_count for t in addr_partials) if addr_partials else 30000
         # Partial term to frequent. Try looking up by rare full names first.
         name_fulls = self.query.get_tokens(name, TokenType.WORD)
         if name_fulls:
@@ -231,14 +232,16 @@ class SearchBuilder:
             if partials_indexed:
                 penalty += 1.2 * sum(t.penalty for t in addr_partials if not t.is_indexed)
 
-            yield penalty,fulls_count / (2**len(addr_tokens)), \
-                  self.get_full_name_ranking(name_fulls, addr_partials,
-                                             fulls_count > 30000 / max(1, len(addr_tokens)))
+            if fulls_count < 50000 or addr_count < 30000:
+                yield penalty,fulls_count / (2**len(addr_tokens)), \
+                    self.get_full_name_ranking(name_fulls, addr_partials,
+                                               fulls_count > 30000 / max(1, len(addr_tokens)))
 
         # To catch remaining results, lookup by name and address
         # We only do this if there is a reasonable number of results expected.
         exp_count = exp_count / (2**len(addr_tokens)) if addr_tokens else exp_count
-        if exp_count < 10000 and all(t.is_indexed for t in name_partials.values()):
+        if exp_count < 10000 and addr_count < 20000\
+           and all(t.is_indexed for t in name_partials.values()):
             penalty += 0.35 * max(1 if name_fulls else 0.1,
                                   5 - len(name_partials) - len(addr_tokens))
             yield penalty, exp_count,\
index 830345c64e7d0749594a26a3db6fec22697f70de..d756cf3cbd2d1c106abc33eb605fa42b7c2f5792 100644 (file)
@@ -23,7 +23,16 @@ flex.set_main_tags{
                secondary_link = 'named',
                tertiary_link = 'named'},
     railway = 'named',
-    man_made = 'always',
+    man_made = {'none',
+                pier = 'always',
+                tower = 'always',
+                bridge = 'always',
+                works = 'named',
+                water_tower = 'always',
+                dyke = 'named',
+                lighthouse = 'always',
+                watermill = 'always',
+                tunnel = 'always'},
     aerialway = 'always',
     boundary = {'named',
                 postal_code = 'always'},
@@ -59,7 +68,6 @@ flex.set_prefilters{delete_keys = {'note', 'note:*', 'source', '*source', 'attri
                                               'noexit', 'crossing', 'give_way', 'stop'},
                                    railway = {'level_crossing', 'no', 'rail', 'switch',
                                               'abandoned', 'signal', 'buffer_stop', 'razed'},
-                                   man_made = {'survey_point', 'cutline'},
                                    aerialway = {'pylon', 'no'},
                                    aeroway = {'no'},
                                    amenity = {'no', 'parking_space', 'parking_entrance'},
index 5b1ab060f1cbb437631472d92e278daddbc4ac80..f8b1e7a73375f82cfbf37b168673b8a89cb63e8a 100644 (file)
@@ -23,7 +23,16 @@ flex.set_main_tags{
                secondary_link = 'named',
                tertiary_link = 'named'},
     railway = 'named',
-    man_made = 'always',
+    man_made = {'none',
+                pier = 'always',
+                tower = 'always',
+                bridge = 'always',
+                works = 'named',
+                water_tower = 'always',
+                dyke = 'named',
+                lighthouse = 'always',
+                watermill = 'always',
+                tunnel = 'always'},
     aerialway = 'always',
     boundary = {'named',
                 postal_code = 'always'},
@@ -63,7 +72,6 @@ flex.set_prefilters{delete_keys = {'note', 'note:*', 'source', '*source', 'attri
                                               'noexit', 'crossing', 'give_way', 'stop'},
                                    railway = {'level_crossing', 'no', 'rail', 'switch',
                                               'abandoned', 'signal', 'buffer_stop', 'razed'},
-                                   man_made = {'survey_point', 'cutline'},
                                    aerialway = {'pylon', 'no'},
                                    aeroway = {'no'},
                                    amenity = {'no', 'parking_space', 'parking_entrance'},