Merge remote-tracking branch 'upstream/master'

[nominatim.git] / nominatim / api / search / db_search_builder.py

diff --git a/nominatim/api/search/db_search_builder.py b/nominatim/api/search/db_search_builder.py
index 72a2dc0b5a60dc1961e485ec77f68a5b7d632049..66e7efaf7f729dec8fdc8f0f889295b6ed8a6f67 100644 (file)
--- a/nominatim/api/search/db_search_builder.py
+++ b/nominatim/api/search/db_search_builder.py
@@ -111,9 +111,11 @@ class SearchBuilder:
             penalty = min(categories.penalties)
             categories.penalties = [p - penalty for p in categories.penalties]
             for search in builder:
-                yield dbs.NearSearch(penalty, categories, search)
+                yield dbs.NearSearch(penalty + assignment.penalty, categories, search)
         else:
-            yield from builder
+            for search in builder:
+                search.penalty += assignment.penalty
+                yield search
 
 
     def build_poi_search(self, sdata: dbf.SearchData) -> Iterator[dbs.AbstractSearch]:
@@ -155,13 +157,21 @@ class SearchBuilder:
         """ Build a simple address search for special entries where the
             housenumber is the main name token.
         """
-        partial_tokens: List[int] = []
-        for trange in address:
-            partial_tokens.extend(t.token for t in self.query.get_partials_list(trange))
+        sdata.lookups = [dbf.FieldLookup('name_vector', [t.token for t in hnrs], 'lookup_any')]
+
+        partials = [t for trange in address
+                       for t in self.query.get_partials_list(trange)]
+
+        if len(partials) != 1 or partials[0].count < 10000:
+            sdata.lookups.append(dbf.FieldLookup('nameaddress_vector',
+                                                 [t.token for t in partials], 'lookup_all'))
+        else:
+            sdata.lookups.append(
+                dbf.FieldLookup('nameaddress_vector',
+                                [t.token for t
+                                 in self.query.get_tokens(address[0], TokenType.WORD)],
+                                'lookup_any'))
 
-        sdata.lookups = [dbf.FieldLookup('name_vector', [t.token for t in hnrs], 'lookup_any'),
-                         dbf.FieldLookup('nameaddress_vector', partial_tokens, 'lookup_all')
-                        ]
         sdata.housenumbers = dbf.WeightedStrings([], [])
         yield dbs.PlaceSearch(0.05, sdata, sum(t.count for t in hnrs))
 
@@ -196,53 +206,33 @@ class SearchBuilder:
 
         partials_indexed = all(t.is_indexed for t in name_partials) \
                            and all(t.is_indexed for t in addr_partials)
-        exp_count = min(t.count for t in name_partials)
+        exp_count = min(t.count for t in name_partials) / (2**(len(name_partials) - 1))
 
-        if (len(name_partials) > 3 or exp_count < 1000) and partials_indexed:
+        if (len(name_partials) > 3 or exp_count < 3000) and partials_indexed:
             yield penalty, exp_count, dbf.lookup_by_names(name_tokens, addr_tokens)
             return
 
-        exp_count = min(exp_count, min(t.count for t in addr_partials)) \
-                    if addr_partials else exp_count
-        if exp_count < 1000 and partials_indexed:
-            # Lookup by address partials and restrict results through name terms.
-            # Give this a small penalty because lookups in the address index are
-            # more expensive
-            yield penalty + exp_count/5000, exp_count,\
-                  dbf.lookup_by_addr(name_tokens, addr_tokens)
-            return
-
         # Partial term to frequent. Try looking up by rare full names first.
         name_fulls = self.query.get_tokens(name, TokenType.WORD)
-        rare_names = list(filter(lambda t: t.count < 10000, name_fulls))
+        fulls_count = sum(t.count for t in name_fulls)
         # At this point drop unindexed partials from the address.
         # This might yield wrong results, nothing we can do about that.
         if not partials_indexed:
             addr_tokens = [t.token for t in addr_partials if t.is_indexed]
             penalty += 1.2 * sum(t.penalty for t in addr_partials if not t.is_indexed)
-        if rare_names:
-            # Any of the full names applies with all of the partials from the address
-            yield penalty, sum(t.count for t in rare_names),\
-                  dbf.lookup_by_any_name([t.token for t in rare_names], addr_tokens)
+        # Any of the full names applies with all of the partials from the address
+        yield penalty, fulls_count / (2**len(addr_partials)),\
+              dbf.lookup_by_any_name([t.token for t in name_fulls], addr_tokens,
+                                     'restrict' if fulls_count < 10000 else 'lookup_all')
 
         # To catch remaining results, lookup by name and address
         # We only do this if there is a reasonable number of results expected.
-        if exp_count < 10000:
-            if all(t.is_indexed for t in name_partials):
-                lookup = [dbf.FieldLookup('name_vector', name_tokens, 'lookup_all')]
-            else:
-                # we don't have the partials, try with the non-rare names
-                non_rare_names = [t.token for t in name_fulls if t.count >= 10000]
-                if not non_rare_names:
-                    return
-                lookup = [dbf.FieldLookup('name_vector', non_rare_names, 'lookup_any')]
+        exp_count = exp_count / (2**len(addr_partials)) if addr_partials else exp_count
+        if exp_count < 10000 and all(t.is_indexed for t in name_partials):
+            lookup = [dbf.FieldLookup('name_vector', name_tokens, 'lookup_all')]
             if addr_tokens:
                 lookup.append(dbf.FieldLookup('nameaddress_vector', addr_tokens, 'lookup_all'))
-            penalty += 0.1 * max(0, 5 - len(name_partials) - len(addr_tokens))
-            if len(rare_names) == len(name_fulls):
-                # if there already was a search for all full tokens,
-                # avoid this if anything has been found
-                penalty += 0.25
+            penalty += 0.35 * max(0, 5 - len(name_partials) - len(addr_tokens))
             yield penalty, exp_count, lookup