Merge remote-tracking branch 'upstream/master'

[nominatim.git] / nominatim / tools / migration.py

diff --git a/nominatim/tools/migration.py b/nominatim/tools/migration.py
index 284bd3166f8648cf18b0554c9368dc0368579ccf..76726e8c59d7ab6d58ca310d77ec41963a04a90f 100644 (file)
--- a/nominatim/tools/migration.py
+++ b/nominatim/tools/migration.py
@@ -227,3 +227,83 @@ def create_interpolation_index_on_place(conn, **_):
         cur.execute("""CREATE INDEX IF NOT EXISTS idx_place_interpolations
                        ON place USING gist(geometry)
                        WHERE osm_type = 'W' and address ? 'interpolation'""")
+
+
+@_migration(4, 0, 99, 2)
+def add_step_column_for_interpolation(conn, **_):
+    """ Add a new column 'step' to the interpolations table.
+
+        Also convers the data into the stricter format which requires that
+        startnumbers comply with the odd/even requirements.
+    """
+    if conn.table_has_column('location_property_osmline', 'step'):
+        return
+
+    with conn.cursor() as cur:
+        # Mark invalid all interpolations with no intermediate numbers.
+        cur.execute("""UPDATE location_property_osmline SET startnumber = null
+                       WHERE endnumber - startnumber <= 1 """)
+        # Align the start numbers where odd/even does not match.
+        cur.execute("""UPDATE location_property_osmline
+                       SET startnumber = startnumber + 1,
+                           linegeo = ST_LineSubString(linegeo,
+                                                      1.0 / (endnumber - startnumber)::float,
+                                                      1)
+                       WHERE (interpolationtype = 'odd' and startnumber % 2 = 0)
+                              or (interpolationtype = 'even' and startnumber % 2 = 1)
+                    """)
+        # Mark invalid odd/even interpolations with no intermediate numbers.
+        cur.execute("""UPDATE location_property_osmline SET startnumber = null
+                       WHERE interpolationtype in ('odd', 'even')
+                             and endnumber - startnumber = 2""")
+        # Finally add the new column and populate it.
+        cur.execute("ALTER TABLE location_property_osmline ADD COLUMN step SMALLINT")
+        cur.execute("""UPDATE location_property_osmline
+                         SET step = CASE WHEN interpolationtype = 'all'
+                                         THEN 1 ELSE 2 END
+                    """)
+
+
+@_migration(4, 0, 99, 3)
+def add_step_column_for_tiger(conn, **_):
+    """ Add a new column 'step' to the tiger data table.
+    """
+    if conn.table_has_column('location_property_tiger', 'step'):
+        return
+
+    with conn.cursor() as cur:
+        cur.execute("ALTER TABLE location_property_tiger ADD COLUMN step SMALLINT")
+        cur.execute("""UPDATE location_property_tiger
+                         SET step = CASE WHEN interpolationtype = 'all'
+                                         THEN 1 ELSE 2 END
+                    """)
+
+
+@_migration(4, 0, 99, 4)
+def add_derived_name_column_for_country_names(conn, **_):
+    """ Add a new column 'derived_name' which in the future takes the
+        country names as imported from OSM data.
+    """
+    if not conn.table_has_column('country_name', 'derived_name'):
+        with conn.cursor() as cur:
+            cur.execute("ALTER TABLE country_name ADD COLUMN derived_name public.HSTORE")
+
+
+@_migration(4, 0, 99, 5)
+def mark_internal_country_names(conn, config, **_):
+    """ Names from the country table should be marked as internal to prevent
+        them from being deleted. Only necessary for ICU tokenizer.
+    """
+    import psycopg2.extras # pylint: disable=import-outside-toplevel
+
+    tokenizer = tokenizer_factory.get_tokenizer_for_db(config)
+    with tokenizer.name_analyzer() as analyzer:
+        with conn.cursor() as cur:
+            psycopg2.extras.register_hstore(cur)
+            cur.execute("SELECT country_code, name FROM country_name")
+
+            for country_code, names in cur:
+                if not names:
+                    names = {}
+                names['countrycode'] = country_code
+                analyzer.add_country_names(country_code, names)