Cleaned tests and add database cleaning tests on test_import_from_wiki

[nominatim.git] / test / python / test_tools_import_special_phrases.py

diff --git a/test/python/test_tools_import_special_phrases.py b/test/python/test_tools_import_special_phrases.py
index b77ae10dc1233f0ede3a3535a2924a22b89251db..f8f6f2dd5f7f968fbeb9b65d3cdeb0053c3d5ae2 100644 (file)
--- a/test/python/test_tools_import_special_phrases.py
+++ b/test/python/test_tools_import_special_phrases.py
@@ -2,6 +2,7 @@
     Tests for import special phrases methods
     of the class SpecialPhrasesImporter.
 """
     Tests for import special phrases methods
     of the class SpecialPhrasesImporter.
 """

+from mocks import MockParamCapture

 from nominatim.errors import UsageError
 from pathlib import Path
 import tempfile
 from nominatim.errors import UsageError
 from pathlib import Path
 import tempfile


@@ -11,6 +12,52 @@ from nominatim.tools.special_phrases import SpecialPhrasesImporter
 
 TEST_BASE_DIR = Path(__file__) / '..' / '..'
 
 
 TEST_BASE_DIR = Path(__file__) / '..' / '..'
 

+def test_fetch_existing_words_phrases_basic(special_phrases_importer, word_table,
+                                            temp_db_cursor):
+    """
+        Check for the fetch_existing_words_phrases() method.
+        It should return special phrase term added to the word
+        table.
+    """
+    query ="""
+        INSERT INTO word VALUES(99999, 'lookup_token', 'normalized_word',
+        'class', 'type', null, 0, 'near');
+    """
+    temp_db_cursor.execute(query)
+
+    assert not special_phrases_importer.words_phrases_to_delete
+    special_phrases_importer._fetch_existing_words_phrases()
+    contained_phrase = special_phrases_importer.words_phrases_to_delete.pop()
+    assert contained_phrase == ('normalized_word', 'class', 'type', 'near')
+
+@pytest.mark.parametrize("house_type", ['house', 'postcode'])
+def test_fetch_existing_words_phrases_special_cases(special_phrases_importer, word_table,
+                                                    house_type, temp_db_cursor):
+    """
+        Check for the fetch_existing_words_phrases() method.
+        It should return nothing as the terms added correspond
+        to a housenumber and postcode term.
+    """
+    query ="""
+        INSERT INTO word VALUES(99999, 'lookup_token', 'normalized_word',
+        'place', %s, null, 0, 'near');
+    """
+    temp_db_cursor.execute(query, (house_type,))
+
+    special_phrases_importer._fetch_existing_words_phrases()
+    assert not special_phrases_importer.words_phrases_to_delete
+
+def test_fetch_existing_place_classtype_tables(special_phrases_importer, temp_db_cursor):
+    """
+        Check for the fetch_existing_place_classtype_tables() method.
+        It should return the table just created.
+    """
+    temp_db_cursor.execute('CREATE TABLE place_classtype_testclasstypetable()')
+
+    special_phrases_importer._fetch_existing_place_classtype_tables()
+    contained_table = special_phrases_importer.table_phrases_to_delete.pop()
+    assert contained_table == 'place_classtype_testclasstypetable'
+

 def test_check_sanity_class(special_phrases_importer):
     """
         Check for _check_sanity() method.
 def test_check_sanity_class(special_phrases_importer):
     """
         Check for _check_sanity() method.


@@ -80,7 +127,7 @@ def test_convert_settings_giving_json(special_phrases_importer):
     assert returned == json_file
 
 def test_process_amenity_with_operator(special_phrases_importer, getorcreate_amenityoperator_funcs,
     assert returned == json_file
 
 def test_process_amenity_with_operator(special_phrases_importer, getorcreate_amenityoperator_funcs,

-                                       word_table, temp_db_conn):
+                                       temp_db_conn, word_table):

     """
         Test that _process_amenity() execute well the 
         getorcreate_amenityoperator() SQL function and that
     """
         Test that _process_amenity() execute well the 
         getorcreate_amenityoperator() SQL function and that


@@ -90,13 +137,13 @@ def test_process_amenity_with_operator(special_phrases_importer, getorcreate_ame
     special_phrases_importer._process_amenity('', '', '', '', 'in')
 
     with temp_db_conn.cursor() as temp_db_cursor:
     special_phrases_importer._process_amenity('', '', '', '', 'in')
 
     with temp_db_conn.cursor() as temp_db_cursor:

-        temp_db_cursor.execute("SELECT * FROM temp_with_operator WHERE op='near' OR op='in'")
+        temp_db_cursor.execute("SELECT * FROM word WHERE operator='near' OR operator='in'")

         results = temp_db_cursor.fetchall()
 
     assert len(results) == 2
 
 def test_process_amenity_without_operator(special_phrases_importer, getorcreate_amenity_funcs,
         results = temp_db_cursor.fetchall()
 
     assert len(results) == 2
 
 def test_process_amenity_without_operator(special_phrases_importer, getorcreate_amenity_funcs,

-                                          temp_db_conn):
+                                          temp_db_conn, word_table):

     """
         Test that _process_amenity() execute well the
         getorcreate_amenity() SQL function.
     """
         Test that _process_amenity() execute well the
         getorcreate_amenity() SQL function.


@@ -104,7 +151,7 @@ def test_process_amenity_without_operator(special_phrases_importer, getorcreate_
     special_phrases_importer._process_amenity('', '', '', '', '')
 
     with temp_db_conn.cursor() as temp_db_cursor:
     special_phrases_importer._process_amenity('', '', '', '', '')
 
     with temp_db_conn.cursor() as temp_db_cursor:

-        temp_db_cursor.execute("SELECT * FROM temp_without_operator WHERE op='no_operator'")
+        temp_db_cursor.execute("SELECT * FROM word WHERE operator='no_operator'")

         result = temp_db_cursor.fetchone()
 
     assert result
         result = temp_db_cursor.fetchone()
 
     assert result


@@ -154,8 +201,8 @@ def test_grant_access_to_web_user(temp_db_conn, def_config, special_phrases_impo
     assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, phrase_class, phrase_type)
 
 def test_create_place_classtype_table_and_indexes(
     assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, phrase_class, phrase_type)
 
 def test_create_place_classtype_table_and_indexes(

-        temp_db_conn, def_config, placex_table, getorcreate_amenity_funcs,
-        getorcreate_amenityoperator_funcs, special_phrases_importer):
+        temp_db_conn, def_config, placex_table,
+        special_phrases_importer):

     """
         Test that _create_place_classtype_table_and_indexes()
         create the right place_classtype tables and place_id indexes
     """
         Test that _create_place_classtype_table_and_indexes()
         create the right place_classtype tables and place_id indexes


@@ -171,7 +218,7 @@ def test_create_place_classtype_table_and_indexes(
         assert check_placeid_and_centroid_indexes(temp_db_conn, pair[0], pair[1])
         assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, pair[0], pair[1])
 
         assert check_placeid_and_centroid_indexes(temp_db_conn, pair[0], pair[1])
         assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, pair[0], pair[1])
 

-def test_process_xml_content(temp_db_conn, def_config, special_phrases_importer, 
+def test_process_xml_content(temp_db_conn, def_config, special_phrases_importer, word_table,

                              getorcreate_amenity_funcs, getorcreate_amenityoperator_funcs):
     """
         Test that _process_xml_content() process the given xml content right
                              getorcreate_amenity_funcs, getorcreate_amenityoperator_funcs):
     """
         Test that _process_xml_content() process the given xml content right


@@ -188,13 +235,77 @@ def test_process_xml_content(temp_db_conn, def_config, special_phrases_importer,
     assert check_amenities_without_op(temp_db_conn)
     assert results[class_test] and type_test in results.values()
 
     assert check_amenities_without_op(temp_db_conn)
     assert results[class_test] and type_test in results.values()
 

+def test_remove_non_existent_phrases_from_db(special_phrases_importer, default_phrases,
+                                             temp_db_conn):
+    """
+        Check for the remove_non_existent_phrases_from_db() method.
+
+        It should removed entries from the word table which are contained
+        in the words_phrases_to_delete set and not those also contained
+        in the words_phrases_still_exist set.
+
+        place_classtype tables contained in table_phrases_to_delete should
+        be deleted.
+    """
+    with temp_db_conn.cursor() as temp_db_cursor:
+        to_delete_phrase_tuple = ('normalized_word', 'class', 'type', 'near')
+        to_keep_phrase_tuple = (
+            'normalized_word_exists', 'class_exists', 'type_exists', 'near'
+        )
+        special_phrases_importer.words_phrases_to_delete = {
+            to_delete_phrase_tuple,
+            to_keep_phrase_tuple
+        }
+        special_phrases_importer.words_phrases_still_exist = {
+            to_keep_phrase_tuple
+        }
+        special_phrases_importer.table_phrases_to_delete = {
+            'place_classtype_testclasstypetable_to_delete'
+        }
+
+        query_words = 'SELECT word, class, type, operator FROM word;'
+        query_tables = """
+            SELECT table_name
+            FROM information_schema.tables
+            WHERE table_schema='public'
+            AND table_name like 'place_classtype_%';
+        """
+
+        special_phrases_importer._remove_non_existent_phrases_from_db()
+
+        temp_db_cursor.execute(query_words)
+        words_result = temp_db_cursor.fetchall()
+        temp_db_cursor.execute(query_tables)
+        tables_result = temp_db_cursor.fetchall()
+        assert len(words_result) == 1 and words_result[0] == [
+            'normalized_word_exists', 'class_exists', 'type_exists', 'near'
+        ]
+        assert (len(tables_result) == 1 and
+            tables_result[0][0] == 'place_classtype_testclasstypetable_to_keep'
+        )
+

 def test_import_from_wiki(monkeypatch, temp_db_conn, def_config, special_phrases_importer, placex_table, 
 def test_import_from_wiki(monkeypatch, temp_db_conn, def_config, special_phrases_importer, placex_table, 

-                          getorcreate_amenity_funcs, getorcreate_amenityoperator_funcs):
+                          getorcreate_amenity_funcs, getorcreate_amenityoperator_funcs, word_table):

     """
         Check that the main import_from_wiki() method is well executed.
         It should create the place_classtype table, the place_id and centroid indexes,
         grand access to the web user and executing the SQL functions for amenities.
     """
         Check that the main import_from_wiki() method is well executed.
         It should create the place_classtype table, the place_id and centroid indexes,
         grand access to the web user and executing the SQL functions for amenities.

+        It should also update the database well by deleting or preserving existing entries 
+        of the database.

     """
     """

+    #Add some data to the database before execution in order to test
+    #what is deleted and what is preserved.
+    with temp_db_conn.cursor() as temp_db_cursor:
+        temp_db_cursor.execute("""
+            INSERT INTO word VALUES(99999, ' animal shelter', 'animal shelter',
+            'amenity', 'animal_shelter', null, 0, null);
+
+            INSERT INTO word VALUES(99999, ' wrong_lookup_token', 'wrong_normalized_word',
+            'wrong_class', 'wrong_type', null, 0, 'near');
+
+            CREATE TABLE place_classtype_amenity_animal_shelter();
+            CREATE TABLE place_classtype_wrongclass_wrongtype();""")
+

     monkeypatch.setattr('nominatim.tools.special_phrases.SpecialPhrasesImporter._get_wiki_content', mock_get_wiki_content)
     special_phrases_importer.import_from_wiki(['en'])
 
     monkeypatch.setattr('nominatim.tools.special_phrases.SpecialPhrasesImporter._get_wiki_content', mock_get_wiki_content)
     special_phrases_importer.import_from_wiki(['en'])
 


@@ -206,6 +317,45 @@ def test_import_from_wiki(monkeypatch, temp_db_conn, def_config, special_phrases
     assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, class_test, type_test)
     assert check_amenities_with_op(temp_db_conn)
     assert check_amenities_without_op(temp_db_conn)
     assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, class_test, type_test)
     assert check_amenities_with_op(temp_db_conn)
     assert check_amenities_without_op(temp_db_conn)

+    assert check_table_exist(temp_db_conn, 'amenity', 'animal_shelter')
+    assert not check_table_exist(temp_db_conn, 'wrong_class', 'wrong_type')
+
+    #Format (query, should_return_something_bool) use to easily execute all asserts
+    queries_tests = set()
+
+    #Used to check that the correct phrase already in the word table before is still there.
+    query_correct_word = "SELECT * FROM word WHERE word = 'animal shelter'"
+    queries_tests.add((query_correct_word, True))
+
+    #Used to check if wrong phrase was deleted from the word table of the database.
+    query_wrong_word = "SELECT word FROM word WHERE word = 'wrong_normalized_word'"
+    queries_tests.add((query_wrong_word, False))
+
+    #Used to check that correct place_classtype table already in the datase before is still there.
+    query_existing_table = """
+        SELECT table_name
+        FROM information_schema.tables
+        WHERE table_schema='public'
+        AND table_name = 'place_classtype_amenity_animal_shelter';
+    """
+    queries_tests.add((query_existing_table, True))
+
+    #Used to check that wrong place_classtype table was deleted from the database.
+    query_wrong_table = """
+        SELECT table_name
+        FROM information_schema.tables
+        WHERE table_schema='public'
+        AND table_name = 'place_classtype_wrongclass_wrongtype';
+    """
+    queries_tests.add((query_wrong_table, False))
+
+    with temp_db_conn.cursor() as temp_db_cursor:
+        for query in queries_tests:
+            temp_db_cursor.execute(query[0])
+            if (query[1] == True):
+                assert temp_db_cursor.fetchone()
+            else:
+                assert not temp_db_cursor.fetchone()

 
 def mock_get_wiki_content(lang):
     """
 
 def mock_get_wiki_content(lang):
     """


@@ -271,7 +421,7 @@ def check_amenities_with_op(temp_db_conn):
         contains more than one value (so that the SQL function was call more than one time).
     """
     with temp_db_conn.cursor() as temp_db_cursor:
         contains more than one value (so that the SQL function was call more than one time).
     """
     with temp_db_conn.cursor() as temp_db_cursor:

-        temp_db_cursor.execute("SELECT * FROM temp_with_operator")
+        temp_db_cursor.execute("SELECT * FROM word WHERE operator != 'no_operator'")

         return len(temp_db_cursor.fetchall()) > 1
 
 def check_amenities_without_op(temp_db_conn):
         return len(temp_db_cursor.fetchall()) > 1
 
 def check_amenities_without_op(temp_db_conn):


@@ -280,7 +430,7 @@ def check_amenities_without_op(temp_db_conn):
         contains more than one value (so that the SQL function was call more than one time).
     """
     with temp_db_conn.cursor() as temp_db_cursor:
         contains more than one value (so that the SQL function was call more than one time).
     """
     with temp_db_conn.cursor() as temp_db_cursor:

-        temp_db_cursor.execute("SELECT * FROM temp_without_operator")
+        temp_db_cursor.execute("SELECT * FROM word WHERE operator = 'no_operator'")

         return len(temp_db_cursor.fetchall()) > 1
 
 @pytest.fixture
         return len(temp_db_cursor.fetchall()) > 1
 
 @pytest.fixture


@@ -305,6 +455,18 @@ def temp_phplib_dir_with_migration():
 
         yield Path(phpdir)
 
 
         yield Path(phpdir)
 

+@pytest.fixture
+def default_phrases(word_table, temp_db_cursor):
+    temp_db_cursor.execute("""
+        INSERT INTO word VALUES(99999, 'lookup_token', 'normalized_word',
+        'class', 'type', null, 0, 'near');
+
+        INSERT INTO word VALUES(99999, 'lookup_token', 'normalized_word_exists',
+        'class_exists', 'type_exists', null, 0, 'near');
+
+        CREATE TABLE place_classtype_testclasstypetable_to_delete();
+        CREATE TABLE place_classtype_testclasstypetable_to_keep();""")
+

 @pytest.fixture
 def make_strandard_name_func(temp_db_cursor):
     temp_db_cursor.execute("""
 @pytest.fixture
 def make_strandard_name_func(temp_db_cursor):
     temp_db_cursor.execute("""


@@ -317,13 +479,12 @@ def make_strandard_name_func(temp_db_cursor):
 @pytest.fixture
 def getorcreate_amenity_funcs(temp_db_cursor, make_strandard_name_func):
     temp_db_cursor.execute("""
 @pytest.fixture
 def getorcreate_amenity_funcs(temp_db_cursor, make_strandard_name_func):
     temp_db_cursor.execute("""

-        CREATE TABLE temp_without_operator(op TEXT);
-    

         CREATE OR REPLACE FUNCTION getorcreate_amenity(lookup_word TEXT, normalized_word TEXT,
                                                     lookup_class text, lookup_type text)
         RETURNS void as $$
         BEGIN
         CREATE OR REPLACE FUNCTION getorcreate_amenity(lookup_word TEXT, normalized_word TEXT,
                                                     lookup_class text, lookup_type text)
         RETURNS void as $$
         BEGIN

-            INSERT INTO temp_without_operator VALUES('no_operator');
+            INSERT INTO word VALUES(null, lookup_word, normalized_word,
+            lookup_class, lookup_type, null, 0, 'no_operator');

         END;
         $$ LANGUAGE plpgsql""")
 
         END;
         $$ LANGUAGE plpgsql""")
 


@@ -336,6 +497,7 @@ def getorcreate_amenityoperator_funcs(temp_db_cursor, make_strandard_name_func):
                                                     lookup_class text, lookup_type text, op text)
         RETURNS void as $$
         BEGIN 
                                                     lookup_class text, lookup_type text, op text)
         RETURNS void as $$
         BEGIN 

-            INSERT INTO temp_with_operator VALUES(op);
+            INSERT INTO word VALUES(null, lookup_word, normalized_word,
+            lookup_class, lookup_type, null, 0, op);

         END;
         $$ LANGUAGE plpgsql""")
\ No newline at end of file
         END;
         $$ LANGUAGE plpgsql""")
\ No newline at end of file