avoid splitting of first token when a housenumber is present

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

diff --git a/nominatim/api/search/icu_tokenizer.py b/nominatim/api/search/icu_tokenizer.py
index 14698a28867ca7ae0fc783f6b6e11385ffe45d8a..f259995db112bbbe537aaa3855f2d4d78e36f5e2 100644 (file)
--- a/nominatim/api/search/icu_tokenizer.py
+++ b/nominatim/api/search/icu_tokenizer.py
@@ -21,10 +21,7 @@ from nominatim.typing import SaRow
 from nominatim.api.connection import SearchConnection
 from nominatim.api.logging import log
 from nominatim.api.search import query as qmod
 from nominatim.api.connection import SearchConnection
 from nominatim.api.logging import log
 from nominatim.api.search import query as qmod

-
-# XXX: TODO
-class AbstractQueryAnalyzer:
-    pass
+from nominatim.api.search.query_analyzer_factory import AbstractQueryAnalyzer

 
 
 DB_TO_TOKEN_TYPE = {
 
 
 DB_TO_TOKEN_TYPE = {


@@ -156,7 +153,7 @@ class ICUQueryAnalyzer(AbstractQueryAnalyzer):
         """
         log().section('Analyze query (using ICU tokenizer)')
         normalized = list(filter(lambda p: p.text,
         """
         log().section('Analyze query (using ICU tokenizer)')
         normalized = list(filter(lambda p: p.text,

-                                 (qmod.Phrase(p.ptype, self.normalizer.transliterate(p.text))
+                                 (qmod.Phrase(p.ptype, self.normalize_text(p.text))

                                   for p in phrases)))
         query = qmod.QueryStruct(normalized)
         log().var_dump('Normalized query', query.source)
                                   for p in phrases)))
         query = qmod.QueryStruct(normalized)
         log().var_dump('Normalized query', query.source)


@@ -190,6 +187,14 @@ class ICUQueryAnalyzer(AbstractQueryAnalyzer):
         return query
 
 
         return query
 
 

+    def normalize_text(self, text: str) -> str:
+        """ Bring the given text into a normalized form. That is the
+            standardized form search will work with. All information removed
+            at this stage is inevitably lost.
+        """
+        return cast(str, self.normalizer.transliterate(text))
+
+

     def split_query(self, query: qmod.QueryStruct) -> Tuple[QueryParts, WordDict]:
         """ Transliterate the phrases and split them into tokens.
 
     def split_query(self, query: qmod.QueryStruct) -> Tuple[QueryParts, WordDict]:
         """ Transliterate the phrases and split them into tokens.
 


@@ -251,12 +256,11 @@ class ICUQueryAnalyzer(AbstractQueryAnalyzer):
                        and (repl.ttype != qmod.TokenType.HOUSENUMBER
                             or len(tlist.tokens[0].lookup_word) > 4):
                         repl.add_penalty(0.39)
                        and (repl.ttype != qmod.TokenType.HOUSENUMBER
                             or len(tlist.tokens[0].lookup_word) > 4):
                         repl.add_penalty(0.39)

-            elif tlist.ttype == qmod.TokenType.HOUSENUMBER:
+            elif tlist.ttype == qmod.TokenType.HOUSENUMBER \
+                 and len(tlist.tokens[0].lookup_word) <= 3:

                 if any(c.isdigit() for c in tlist.tokens[0].lookup_word):
                     for repl in node.starting:
                 if any(c.isdigit() for c in tlist.tokens[0].lookup_word):
                     for repl in node.starting:

-                        if repl.end == tlist.end and repl.ttype != qmod.TokenType.HOUSENUMBER \
-                           and (repl.ttype != qmod.TokenType.HOUSENUMBER
-                                or len(tlist.tokens[0].lookup_word) <= 3):
+                        if repl.end == tlist.end and repl.ttype != qmod.TokenType.HOUSENUMBER:

                             repl.add_penalty(0.5 - tlist.tokens[0].penalty)
             elif tlist.ttype not in (qmod.TokenType.COUNTRY, qmod.TokenType.PARTIAL):
                 norm = parts[i].normalized
                             repl.add_penalty(0.5 - tlist.tokens[0].penalty)
             elif tlist.ttype not in (qmod.TokenType.COUNTRY, qmod.TokenType.PARTIAL):
                 norm = parts[i].normalized