Merge remote-tracking branch 'upstream/master'

[nominatim.git] / nominatim / tokenizer / legacy_tokenizer.py

diff --git a/nominatim/tokenizer/legacy_tokenizer.py b/nominatim/tokenizer/legacy_tokenizer.py
index 24af1c3a1a4013f035cac265bb116f644c53c8a0..d901a68d2e53f77e5c96210c11ede863e7e5e36f 100644 (file)
--- a/nominatim/tokenizer/legacy_tokenizer.py
+++ b/nominatim/tokenizer/legacy_tokenizer.py
@@ -16,6 +16,7 @@ from nominatim.db import properties
 from nominatim.db import utils as db_utils
 from nominatim.db.sql_preprocessor import SQLPreprocessor
 from nominatim.errors import UsageError
+from nominatim.tokenizer.base import AbstractAnalyzer, AbstractTokenizer
 
 DBCFG_NORMALIZATION = "tokenizer_normalization"
 DBCFG_MAXWORDFREQ = "tokenizer_maxwordfreq"
@@ -76,7 +77,7 @@ def _check_module(module_dir, conn):
             raise UsageError("Database module cannot be accessed.") from err
 
 
-class LegacyTokenizer:
+class LegacyTokenizer(AbstractTokenizer):
     """ The legacy tokenizer uses a special PostgreSQL module to normalize
         names and queries. The tokenizer thus implements normalization through
         calls to the database.
@@ -112,7 +113,7 @@ class LegacyTokenizer:
             self._init_db_tables(config)
 
 
-    def init_from_project(self):
+    def init_from_project(self, _):
         """ Initialise the tokenizer from the project directory.
         """
         with connect(self.dsn) as conn:
@@ -141,7 +142,7 @@ class LegacyTokenizer:
                               modulepath=modulepath)
 
 
-    def check_database(self):
+    def check_database(self, _):
         """ Check that the tokenizer is set up correctly.
         """
         hint = """\
@@ -185,6 +186,24 @@ class LegacyTokenizer:
             self._save_config(conn, config)
 
 
+    def update_statistics(self):
+        """ Recompute the frequency of full words.
+        """
+        with connect(self.dsn) as conn:
+            with conn.cursor() as cur:
+                cur.drop_table("word_frequencies")
+                LOG.info("Computing word frequencies")
+                cur.execute("""CREATE TEMP TABLE word_frequencies AS
+                                 SELECT unnest(name_vector) as id, count(*)
+                                 FROM search_name GROUP BY id""")
+                cur.execute("CREATE INDEX ON word_frequencies(id)")
+                LOG.info("Update word table with recomputed frequencies")
+                cur.execute("""UPDATE word SET search_name_count = count
+                               FROM word_frequencies
+                               WHERE word_token like ' %' and word_id = id""")
+                cur.drop_table("word_frequencies")
+            conn.commit()
+
     def name_analyzer(self):
         """ Create a new analyzer for tokenizing names and queries
             using this tokinzer. Analyzers are context managers and should
@@ -238,7 +257,7 @@ class LegacyTokenizer:
         properties.set_property(conn, DBCFG_MAXWORDFREQ, config.MAX_WORD_FREQUENCY)
 
 
-class LegacyNameAnalyzer:
+class LegacyNameAnalyzer(AbstractAnalyzer):
     """ The legacy analyzer uses the special Postgresql module for
         splitting names.
 
@@ -255,14 +274,6 @@ class LegacyNameAnalyzer:
         self._cache = _TokenCache(self.conn)
 
 
-    def __enter__(self):
-        return self
-
-
-    def __exit__(self, exc_type, exc_value, traceback):
-        self.close()
-
-
     def close(self):
         """ Free all resources used by the analyzer.
         """
@@ -370,8 +381,7 @@ class LegacyNameAnalyzer:
             to_delete = existing_phrases - norm_phrases
 
             if to_add:
-                psycopg2.extras.execute_values(
-                    cur,
+                cur.execute_values(
                     """ INSERT INTO word (word_id, word_token, word, class, type,
                                           search_name_count, operator)
                         (SELECT nextval('seq_word'), ' ' || make_standard_name(name), name,
@@ -381,8 +391,7 @@ class LegacyNameAnalyzer:
                     to_add)
 
             if to_delete and should_replace:
-                psycopg2.extras.execute_values(
-                    cur,
+                cur.execute_values(
                     """ DELETE FROM word USING (VALUES %s) as v(name, in_class, in_type, op)
                         WHERE word = name and class = in_class and type = in_type
                               and ((op = '-' and operator is null) or op = operator)""",
@@ -414,16 +423,15 @@ class LegacyNameAnalyzer:
         """
         token_info = _TokenInfo(self._cache)
 
-        names = place.get('name')
+        names = place.name
 
         if names:
             token_info.add_names(self.conn, names)
 
-            country_feature = place.get('country_feature')
-            if country_feature and re.fullmatch(r'[A-Za-z][A-Za-z]', country_feature):
-                self.add_country_names(country_feature.lower(), names)
+            if place.is_country():
+                self.add_country_names(place.country_code, names)
 
-        address = place.get('address')
+        address = place.address
         if address:
             self._process_place_address(token_info, address)
 
@@ -582,7 +590,7 @@ class _TokenCache:
         with conn.cursor() as cur:
             cur.execute("""SELECT i, ARRAY[getorcreate_housenumber_id(i::text)]::text
                            FROM generate_series(1, 100) as i""")
-            self._cached_housenumbers = {str(r[0]) : r[1] for r in cur}
+            self._cached_housenumbers = {str(r[0]): r[1] for r in cur}
 
         # For postcodes remember the ones that have already been added
         self.postcodes = set()