Merge pull request #3328 from lonvia/word-count-into-new-table

[nominatim.git] / test / python / api / search / test_db_search_builder.py

diff --git a/test/python/api/search/test_db_search_builder.py b/test/python/api/search/test_db_search_builder.py
index d1ad320b137f517b2d06ee74384f70e49944d7f2..d3aea90002740d7660e12a4b210bf4cb41344c60 100644 (file)
--- a/test/python/api/search/test_db_search_builder.py
+++ b/test/python/api/search/test_db_search_builder.py
@@ -21,21 +21,18 @@ class MyToken(Token):
 
 
 def make_query(*args):
-    q = None
+    q = QueryStruct([Phrase(PhraseType.NONE, '')])
 
-    for tlist in args:
-        if q is None:
-            q = QueryStruct([Phrase(PhraseType.NONE, '')])
-        else:
-            q.add_node(BreakType.WORD, PhraseType.NONE)
+    for _ in range(max(inner[0] for tlist in args for inner in tlist)):
+        q.add_node(BreakType.WORD, PhraseType.NONE)
+    q.add_node(BreakType.END, PhraseType.NONE)
 
-        start = len(q.nodes) - 1
+    for start, tlist in enumerate(args):
         for end, ttype, tinfo in tlist:
             for tid, word in tinfo:
                 q.add_token(TokenRange(start, end), ttype,
                             MyToken(0.5 if ttype == TokenType.PARTIAL else 0.0, tid, 1, word, True))
 
-    q.add_node(BreakType.END, PhraseType.NONE)
 
     return q
 
@@ -150,27 +147,27 @@ def test_postcode_with_address_with_full_word():
 
 @pytest.mark.parametrize('kwargs', [{'viewbox': '0,0,1,1', 'bounded_viewbox': True},
                                     {'near': '10,10'}])
-def test_category_only(kwargs):
-    q = make_query([(1, TokenType.CATEGORY, [(2, 'foo')])])
+def test_near_item_only(kwargs):
+    q = make_query([(1, TokenType.NEAR_ITEM, [(2, 'foo')])])
     builder = SearchBuilder(q, SearchDetails.from_kwargs(kwargs))
 
-    searches = list(builder.build(TokenAssignment(category=TokenRange(0, 1))))
+    searches = list(builder.build(TokenAssignment(near_item=TokenRange(0, 1))))
 
     assert len(searches) == 1
 
     search = searches[0]
 
     assert isinstance(search, dbs.PoiSearch)
-    assert search.categories.values == [('this', 'that')]
+    assert search.qualifiers.values == [('this', 'that')]
 
 
 @pytest.mark.parametrize('kwargs', [{'viewbox': '0,0,1,1'},
                                     {}])
-def test_category_skipped(kwargs):
-    q = make_query([(1, TokenType.CATEGORY, [(2, 'foo')])])
+def test_near_item_skipped(kwargs):
+    q = make_query([(1, TokenType.NEAR_ITEM, [(2, 'foo')])])
     builder = SearchBuilder(q, SearchDetails.from_kwargs(kwargs))
 
-    searches = list(builder.build(TokenAssignment(category=TokenRange(0, 1))))
+    searches = list(builder.build(TokenAssignment(near_item=TokenRange(0, 1))))
 
     assert len(searches) == 0
 
@@ -287,13 +284,13 @@ def test_name_and_complex_address():
 
 
 def test_name_only_near_search():
-    q = make_query([(1, TokenType.CATEGORY, [(88, 'g')])],
+    q = make_query([(1, TokenType.NEAR_ITEM, [(88, 'g')])],
                    [(2, TokenType.PARTIAL, [(1, 'a')]),
                     (2, TokenType.WORD, [(100, 'a')])])
     builder = SearchBuilder(q, SearchDetails())
 
     searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
-                                                  category=TokenRange(0, 1))))
+                                                  near_item=TokenRange(0, 1))))
 
     assert len(searches) == 1
     search = searches[0]
@@ -312,10 +309,68 @@ def test_name_only_search_with_category():
     assert len(searches) == 1
     search = searches[0]
 
+    assert isinstance(search, dbs.PlaceSearch)
+    assert search.qualifiers.values == [('foo', 'bar')]
+
+
+def test_name_with_near_item_search_with_category_mismatch():
+    q = make_query([(1, TokenType.NEAR_ITEM, [(88, 'g')])],
+                   [(2, TokenType.PARTIAL, [(1, 'a')]),
+                    (2, TokenType.WORD, [(100, 'a')])])
+    builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar')]}))
+
+    searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
+                                                  near_item=TokenRange(0, 1))))
+
+    assert len(searches) == 0
+
+
+def test_name_with_near_item_search_with_category_match():
+    q = make_query([(1, TokenType.NEAR_ITEM, [(88, 'g')])],
+                   [(2, TokenType.PARTIAL, [(1, 'a')]),
+                    (2, TokenType.WORD, [(100, 'a')])])
+    builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar'),
+                                                                         ('this', 'that')]}))
+
+    searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
+                                                  near_item=TokenRange(0, 1))))
+
+    assert len(searches) == 1
+    search = searches[0]
+
     assert isinstance(search, dbs.NearSearch)
     assert isinstance(search.search, dbs.PlaceSearch)
 
 
+def test_name_with_qualifier_search_with_category_mismatch():
+    q = make_query([(1, TokenType.QUALIFIER, [(88, 'g')])],
+                   [(2, TokenType.PARTIAL, [(1, 'a')]),
+                    (2, TokenType.WORD, [(100, 'a')])])
+    builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar')]}))
+
+    searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
+                                                  qualifier=TokenRange(0, 1))))
+
+    assert len(searches) == 0
+
+
+def test_name_with_qualifier_search_with_category_match():
+    q = make_query([(1, TokenType.QUALIFIER, [(88, 'g')])],
+                   [(2, TokenType.PARTIAL, [(1, 'a')]),
+                    (2, TokenType.WORD, [(100, 'a')])])
+    builder = SearchBuilder(q, SearchDetails.from_kwargs({'categories': [('foo', 'bar'),
+                                                                         ('this', 'that')]}))
+
+    searches = list(builder.build(TokenAssignment(name=TokenRange(1, 2),
+                                                  qualifier=TokenRange(0, 1))))
+
+    assert len(searches) == 1
+    search = searches[0]
+
+    assert isinstance(search, dbs.PlaceSearch)
+    assert search.qualifiers.values == [('this', 'that')]
+
+
 def test_name_only_search_with_countries():
     q = make_query([(1, TokenType.PARTIAL, [(1, 'a')]),
                     (1, TokenType.WORD, [(100, 'a')])])
@@ -365,8 +420,8 @@ def test_infrequent_partials_in_name():
     assert len(search.lookups) == 2
     assert len(search.rankings) == 2
 
-    assert set((l.column, l.lookup_type) for l in search.lookups) == \
-            {('name_vector', 'lookup_all'), ('nameaddress_vector', 'restrict')}
+    assert set((l.column, l.lookup_type.__name__) for l in search.lookups) == \
+            {('name_vector', 'LookupAll'), ('nameaddress_vector', 'Restrict')}
 
 
 def test_frequent_partials_in_name_and_address():
@@ -377,19 +432,19 @@ def test_frequent_partials_in_name_and_address():
     assert all(isinstance(s, dbs.PlaceSearch) for s in searches)
     searches.sort(key=lambda s: s.penalty)
 
-    assert set((l.column, l.lookup_type) for l in searches[0].lookups) == \
-            {('name_vector', 'lookup_any'), ('nameaddress_vector', 'restrict')}
-    assert set((l.column, l.lookup_type) for l in searches[1].lookups) == \
-            {('nameaddress_vector', 'lookup_all'), ('name_vector', 'lookup_all')}
+    assert set((l.column, l.lookup_type.__name__) for l in searches[0].lookups) == \
+            {('name_vector', 'LookupAny'), ('nameaddress_vector', 'Restrict')}
+    assert set((l.column, l.lookup_type.__name__) for l in searches[1].lookups) == \
+            {('nameaddress_vector', 'LookupAll'), ('name_vector', 'LookupAll')}
 
 
 def test_too_frequent_partials_in_name_and_address():
-    searches = make_counted_searches(10000, 1, 10000, 1)
+    searches = make_counted_searches(20000, 1, 10000, 1)
 
     assert len(searches) == 1
 
     assert all(isinstance(s, dbs.PlaceSearch) for s in searches)
     searches.sort(key=lambda s: s.penalty)
 
-    assert set((l.column, l.lookup_type) for l in searches[0].lookups) == \
-            {('name_vector', 'lookup_any'), ('nameaddress_vector', 'restrict')}
+    assert set((l.column, l.lookup_type.__name__) for l in searches[0].lookups) == \
+            {('name_vector', 'LookupAny'), ('nameaddress_vector', 'Restrict')}