Merge pull request #3678 from lonvia/search-tweaks

[nominatim.git] / src / nominatim_api / search / token_assignment.py

diff --git a/src/nominatim_api/search/token_assignment.py b/src/nominatim_api/search/token_assignment.py
index cfd9efe5ec9a44c3958cbb72816eda62af86884d..8d25aa8f10bdc3c751a2788c89c5710b50a4f8fb 100644 (file)
--- a/src/nominatim_api/search/token_assignment.py
+++ b/src/nominatim_api/search/token_assignment.py
@@ -269,10 +269,9 @@ class _TokenSequence:
             # <address>,<postcode> should give preference to address search
             if base.postcode.start == 0:
                 penalty = self.penalty
-                self.direction = -1  # name searches are only possible backwards
             else:
                 penalty = self.penalty + 0.1
-                self.direction = 1  # name searches are only possible forwards
+            penalty += 0.1 * max(0, len(base.address) - 1)
             yield dataclasses.replace(base, penalty=penalty)
 
     def _get_assignments_address_forward(self, base: TokenAssignment,
@@ -282,6 +281,11 @@ class _TokenSequence:
         """
         first = base.address[0]
 
+        # The postcode must come after the name.
+        if base.postcode and base.postcode < first:
+            log().var_dump('skip forward', (base.postcode, first))
+            return
+
         log().comment('first word = name')
         yield dataclasses.replace(base, penalty=self.penalty,
                                   name=first, address=base.address[1:])
@@ -293,7 +297,7 @@ class _TokenSequence:
         #  * the containing phrase is strictly typed
         if (base.housenumber and first.end < base.housenumber.start)\
            or (base.qualifier and base.qualifier > first)\
-           or (query.nodes[first.start].ptype != qmod.PhraseType.NONE):
+           or (query.nodes[first.start].ptype != qmod.PHRASE_ANY):
             return
 
         penalty = self.penalty
@@ -317,7 +321,12 @@ class _TokenSequence:
         """
         last = base.address[-1]
 
-        if self.direction == -1 or len(base.address) > 1:
+        # The postcode must come before the name for backward direction.
+        if base.postcode and base.postcode > last:
+            log().var_dump('skip backward', (base.postcode, last))
+            return
+
+        if self.direction == -1 or len(base.address) > 1 or base.postcode:
             log().comment('last word = name')
             yield dataclasses.replace(base, penalty=self.penalty,
                                       name=last, address=base.address[:-1])
@@ -329,7 +338,7 @@ class _TokenSequence:
         #  * the containing phrase is strictly typed
         if (base.housenumber and last.start > base.housenumber.end)\
            or (base.qualifier and base.qualifier < last)\
-           or (query.nodes[last.start].ptype != qmod.PhraseType.NONE):
+           or (query.nodes[last.start].ptype != qmod.PHRASE_ANY):
             return
 
         penalty = self.penalty
@@ -393,7 +402,7 @@ def yield_token_assignments(query: qmod.QueryStruct) -> Iterator[TokenAssignment
         another. It does not include penalties for transitions within a
         type.
     """
-    todo = [_TokenSequence([], direction=0 if query.source[0].ptype == qmod.PhraseType.NONE else 1)]
+    todo = [_TokenSequence([], direction=0 if query.source[0].ptype == qmod.PHRASE_ANY else 1)]
 
     while todo:
         state = todo.pop()