Merge pull request #2589 from lonvia/clean-housenumbers

Add command for cleaning up word table

.github/workflows/ci-tests.yml

@@ -309,12 +309,20 @@ jobs:
                   NOMINATIM_REPLICATION_MAX_DIFF=1 nominatim replication --once
               working-directory: /home/nominatim/nominatim-project
 
+            - name: Clean up database
+              run: nominatim refresh --postcodes --word-tokens
+              working-directory: /home/nominatim/nominatim-project
+
             - name: Run reverse-only import
               run : |
                   echo 'NOMINATIM_DATABASE_DSN="pgsql:dbname=reverse"' >> .env
                   nominatim import --osm-file ../test.pbf --reverse-only --no-updates
               working-directory: /home/nominatim/data-env-reverse
 
-            - name: Check reverse import
+            - name: Check reverse-only import
               run: nominatim admin --check-database
               working-directory: /home/nominatim/data-env-reverse
+
+            - name: Clean up database (reverse-only import)
+              run: nominatim refresh --postcodes --word-tokens
+              working-directory: /home/nominatim/nominatim-project

nominatim/clicmd/refresh.py

@@ -39,6 +39,8 @@ class UpdateRefresh:
         group = parser.add_argument_group('Data arguments')
         group.add_argument('--postcodes', action='store_true',
                            help='Update postcode centroid table')
+        group.add_argument('--word-tokens', action='store_true',
+                           help='Clean up search terms')
         group.add_argument('--word-counts', action='store_true',
                            help='Compute frequency of full-word search terms')
         group.add_argument('--address-levels', action='store_true',
@@ -76,6 +78,11 @@ class UpdateRefresh:
                 LOG.error("The place table doesn't exist. "
                           "Postcode updates on a frozen database is not possible.")
 
+        if args.word_tokens:
+            LOG.warning('Updating word tokens')
+            tokenizer = self._get_tokenizer(args.config)
+            tokenizer.update_word_tokens()
+
         if args.word_counts:
             LOG.warning('Recompute word statistics')
             self._get_tokenizer(args.config).update_statistics()

nominatim/tokenizer/base.py

@@ -209,6 +209,13 @@ class AbstractTokenizer(ABC):
         """
 
 
+    @abstractmethod
+    def update_word_tokens(self) -> None:
+        """ Do house-keeping on the tokenizers internal data structures.
+            Remove unused word tokens, resort data etc.
+        """
+
+
     @abstractmethod
     def name_analyzer(self) -> AbstractAnalyzer:
         """ Create a new analyzer for tokenizing names and queries

nominatim/tokenizer/icu_tokenizer.py

@@ -112,6 +112,47 @@ class LegacyICUTokenizer(AbstractTokenizer):
             conn.commit()
 
 
+    def _cleanup_housenumbers(self):
+        """ Remove unused house numbers.
+        """
+        with connect(self.dsn) as conn:
+            if not conn.table_exists('search_name'):
+                return
+            with conn.cursor(name="hnr_counter") as cur:
+                cur.execute("""SELECT word_id, word_token FROM word
+                               WHERE type = 'H'
+                                 AND NOT EXISTS(SELECT * FROM search_name
+                                                WHERE ARRAY[word.word_id] && name_vector)
+                                 AND (char_length(word_token) > 6
+                                      OR word_token not similar to '\\d+')
+                            """)
+                candidates = {token: wid for wid, token in cur}
+            with conn.cursor(name="hnr_counter") as cur:
+                cur.execute("""SELECT housenumber FROM placex
+                               WHERE housenumber is not null
+                                     AND (char_length(housenumber) > 6
+                                          OR housenumber not similar to '\\d+')
+                            """)
+                for row in cur:
+                    for hnr in row[0].split(';'):
+                        candidates.pop(hnr, None)
+            LOG.info("There are %s outdated housenumbers.", len(candidates))
+            if candidates:
+                with conn.cursor() as cur:
+                    cur.execute("""DELETE FROM word WHERE word_id = any(%s)""",
+                                (list(candidates.values()), ))
+                conn.commit()
+
+
+
+    def update_word_tokens(self):
+        """ Remove unused tokens.
+        """
+        LOG.warning("Cleaning up housenumber tokens.")
+        self._cleanup_housenumbers()
+        LOG.warning("Tokenizer house-keeping done.")
+
+
     def name_analyzer(self):
         """ Create a new analyzer for tokenizing names and queries
             using this tokinzer. Analyzers are context managers and should

nominatim/tokenizer/legacy_tokenizer.py

@@ -211,6 +211,13 @@ class LegacyTokenizer(AbstractTokenizer):
                     cur.drop_table("word_frequencies")
             conn.commit()
 
+
+    def update_word_tokens(self):
+        """ No house-keeping implemented for the legacy tokenizer.
+        """
+        LOG.info("No tokenizer clean-up available.")
+
+
     def name_analyzer(self):
         """ Create a new analyzer for tokenizing names and queries
             using this tokinzer. Analyzers are context managers and should

test/python/cli/conftest.py

@@ -30,6 +30,7 @@ class DummyTokenizer:
         self.update_sql_functions_called = False
         self.finalize_import_called = False
         self.update_statistics_called = False
+        self.update_word_tokens_called = False
 
     def update_sql_functions(self, *args):
         self.update_sql_functions_called = True
@@ -40,6 +41,9 @@ class DummyTokenizer:
     def update_statistics(self):
         self.update_statistics_called = True
 
+    def update_word_tokens(self):
+        self.update_word_tokens_called = True
+
 
 @pytest.fixture
 def cli_call(src_dir):

test/python/cli/test_cmd_refresh.py

@@ -39,6 +39,11 @@ class TestRefresh:
         assert self.tokenizer_mock.update_statistics_called
 
 
+    def test_refresh_word_tokens(self):
+        assert self.call_nominatim('refresh', '--word-tokens') == 0
+        assert self.tokenizer_mock.update_word_tokens_called
+
+
     def test_refresh_postcodes(self, mock_func_factory, place_table):
         func_mock = mock_func_factory(nominatim.tools.postcodes, 'update_postcodes')
         idx_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_postcodes')

test/python/mock_icu_word_table.py

@@ -58,6 +58,14 @@ class MockIcuWordTable:
         self.conn.commit()
 
 
+    def add_housenumber(self, word_id, word_token):
+        with self.conn.cursor() as cur:
+            cur.execute("""INSERT INTO word (word_id, word_token, type)
+                              VALUES (%s, %s, 'H')
+                        """, (word_id, word_token))
+        self.conn.commit()
+
+
     def count(self):
         with self.conn.cursor() as cur:
             return cur.scalar("SELECT count(*) FROM word")
@@ -68,6 +76,11 @@ class MockIcuWordTable:
             return cur.scalar("SELECT count(*) FROM word WHERE type = 'S'")
 
 
+    def count_housenumbers(self):
+        with self.conn.cursor() as cur:
+            return cur.scalar("SELECT count(*) FROM word WHERE type = 'H'")
+
+
     def get_special(self):
         with self.conn.cursor() as cur:
             cur.execute("SELECT word_token, info, word FROM word WHERE type = 'S'")

test/python/tokenizer/test_icu.py

@@ -9,6 +9,7 @@ Tests for ICU tokenizer.
 """
 import shutil
 import yaml
+import itertools
 
 import pytest
 
@@ -554,3 +555,69 @@ class TestPlaceAddress:
 
         assert 'addr' not in info
 
+
+class TestUpdateWordTokens:
+
+    @pytest.fixture(autouse=True)
+    def setup(self, tokenizer_factory, table_factory, placex_table, word_table):
+        table_factory('search_name', 'place_id BIGINT, name_vector INT[]')
+        self.tok = tokenizer_factory()
+
+
+    @pytest.fixture
+    def search_entry(self, temp_db_cursor):
+        place_id = itertools.count(1000)
+
+        def _insert(*args):
+            temp_db_cursor.execute("INSERT INTO search_name VALUES (%s, %s)",
+                                   (next(place_id), list(args)))
+
+        return _insert
+
+
+    @pytest.mark.parametrize('hnr', ('1a', '1234567', '34 5'))
+    def test_remove_unused_housenumbers(self, word_table, hnr):
+        word_table.add_housenumber(1000, hnr)
+
+        assert word_table.count_housenumbers() == 1
+        self.tok.update_word_tokens()
+        assert word_table.count_housenumbers() == 0
+
+
+    def test_keep_unused_numeral_housenumbers(self, word_table):
+        word_table.add_housenumber(1000, '5432')
+
+        assert word_table.count_housenumbers() == 1
+        self.tok.update_word_tokens()
+        assert word_table.count_housenumbers() == 1
+
+
+    def test_keep_housenumbers_from_search_name_table(self, word_table, search_entry):
+        word_table.add_housenumber(9999, '5432a')
+        word_table.add_housenumber(9991, '9 a')
+        search_entry(123, 9999, 34)
+
+        assert word_table.count_housenumbers() == 2
+        self.tok.update_word_tokens()
+        assert word_table.count_housenumbers() == 1
+
+
+    def test_keep_housenumbers_from_placex_table(self, word_table, placex_table):
+        word_table.add_housenumber(9999, '5432a')
+        word_table.add_housenumber(9990, '34z')
+        placex_table.add(housenumber='34z')
+        placex_table.add(housenumber='25432a')
+
+        assert word_table.count_housenumbers() == 2
+        self.tok.update_word_tokens()
+        assert word_table.count_housenumbers() == 1
+
+
+    def test_keep_housenumbers_from_placex_table_hnr_list(self, word_table, placex_table):
+        word_table.add_housenumber(9991, '9 b')
+        word_table.add_housenumber(9990, '34z')
+        placex_table.add(housenumber='9 a;9 b;9 c')
+
+        assert word_table.count_housenumbers() == 2
+        self.tok.update_word_tokens()
+        assert word_table.count_housenumbers() == 1

test/python/tokenizer/test_legacy.py

@@ -257,6 +257,13 @@ def test_update_statistics(word_table, table_factory, temp_db_cursor, tokenizer_
                                           search_name_count > 0""") > 0
 
 
+def test_update_word_tokens(tokenizer_factory):
+    tok = tokenizer_factory()
+
+    # This is a noop and should just pass.
+    tok.update_word_tokens()
+
+
 def test_normalize(analyzer):
     assert analyzer.normalize('TEsT') == 'test'