Merge pull request #2312 from lonvia/icu-tokenizer

[nominatim.git] / nominatim / indexer / runners.py

diff --git a/nominatim/indexer/runners.py b/nominatim/indexer/runners.py
index dd6ced3881f1bdf25f0df5fa0df96e5d419b2077..aa607faae3f3d48988ebddd738a49c90ba4bb607 100644 (file)
--- a/nominatim/indexer/runners.py
+++ b/nominatim/indexer/runners.py
@@ -2,35 +2,44 @@
 Mix-ins that provide the actual commands for the indexer for various indexing
 tasks.
 """
+import functools
+
+import psycopg2.extras
+
 # pylint: disable=C0111
 
 class AbstractPlacexRunner:
     """ Returns SQL commands for indexing of the placex table.
     """
-    SELECT_SQL = 'SELECT place_id, (placex_prepare_update(placex)).* FROM placex'
+    SELECT_SQL = 'SELECT place_id FROM placex'
 
-    def __init__(self, rank):
+    def __init__(self, rank, analyzer):
         self.rank = rank
-        self._sql_terms = 0
-        self._cached_index_sql = None
+        self.analyzer = analyzer
+
+
+    @staticmethod
+    @functools.lru_cache(maxsize=1)
+    def _index_sql(num_places):
+        return """ UPDATE placex
+                   SET indexed_status = 0, address = v.addr, token_info = v.ti
+                   FROM (VALUES {}) as v(id, addr, ti)
+                   WHERE place_id = v.id
+               """.format(','.join(["(%s, %s::hstore, %s::jsonb)"]  * num_places))
 
-    def _index_sql(self, num_places):
-        if num_places != self._sql_terms:
-            self._cached_index_sql = \
-                """ UPDATE placex
-                    SET indexed_status = 0, address = v.addr
-                    FROM (VALUES {}) as v(id, addr)
-                    WHERE place_id = v.id
-                """.format(','.join(["(%s, %s::hstore)"]  * num_places))
-            self._sql_terms = num_places
 
-        return self._cached_index_sql
+    @staticmethod
+    def get_place_details(worker, ids):
+        worker.perform("""SELECT place_id, (placex_prepare_update(placex)).*
+                          FROM placex WHERE place_id IN %s""",
+                       (tuple((p[0] for p in ids)), ))
 
 
     def index_places(self, worker, places):
         values = []
         for place in places:
             values.extend((place[x] for x in ('place_id', 'address')))
+            values.append(psycopg2.extras.Json(self.analyzer.process_place(place)))
 
         worker.perform(self._index_sql(len(places)), values)
 
@@ -80,6 +89,10 @@ class InterpolationRunner:
         location_property_osmline.
     """
 
+    def __init__(self, analyzer):
+        self.analyzer = analyzer
+
+
     @staticmethod
     def name():
         return "interpolation lines (location_property_osmline)"
@@ -91,15 +104,37 @@ class InterpolationRunner:
 
     @staticmethod
     def sql_get_objects():
-        return """SELECT place_id FROM location_property_osmline
+        return """SELECT place_id
+                  FROM location_property_osmline
                   WHERE indexed_status > 0
                   ORDER BY geometry_sector"""
 
+
     @staticmethod
-    def index_places(worker, ids):
-        worker.perform(""" UPDATE location_property_osmline
-                           SET indexed_status = 0 WHERE place_id IN ({})
-                       """.format(','.join((str(i[0]) for i in ids))))
+    def get_place_details(worker, ids):
+        worker.perform("""SELECT place_id, get_interpolation_address(address, osm_id) as address
+                          FROM location_property_osmline WHERE place_id IN %s""",
+                       (tuple((p[0] for p in ids)), ))
+
+
+    @staticmethod
+    @functools.lru_cache(maxsize=1)
+    def _index_sql(num_places):
+        return """ UPDATE location_property_osmline
+                   SET indexed_status = 0, address = v.addr, token_info = v.ti
+                   FROM (VALUES {}) as v(id, addr, ti)
+                   WHERE place_id = v.id
+               """.format(','.join(["(%s, %s::hstore, %s::jsonb)"]  * num_places))
+
+
+    def index_places(self, worker, places):
+        values = []
+        for place in places:
+            values.extend((place[x] for x in ('place_id', 'address')))
+            values.append(psycopg2.extras.Json(self.analyzer.process_place(place)))
+
+        worker.perform(self._index_sql(len(places)), values)
+
 
 
 class PostcodeRunner: