Merge pull request #547 from CartoDB/development

Release python-0.21.2

NEWS.md

@@ -1,3 +1,13 @@
+Mar 4th, 2019
+==============
+* Version `0.21.2` of the python library
+    * Fixed TomTom Qps respondes (#546)
+
+Feb 25th, 2019
+==============
+* Version `0.26.2` of the client extension
+    * Fixed bulk geocoding soft limit (#543)
+
 Dec 31st, 2018
 ==============
 * Version `0.35.1` of the server extension and `0.26.1` of the client

client/cdb_dataservices_client--0.26.1--0.26.2.sql

@@ -0,0 +1,96 @@
+--DO NOT MODIFY THIS FILE, IT IS GENERATED AUTOMATICALLY FROM SOURCES
+-- Complain if script is sourced in psql, rather than via CREATE EXTENSION
+\echo Use "ALTER EXTENSION cdb_dataservices_client UPDATE TO '0.26.2'" to load this file. \quit
+
+-- Make sure we have a sane search path to create/update the extension
+SET search_path = "$user",cartodb,public,cdb_dataservices_client;
+
+CREATE OR REPLACE FUNCTION cdb_dataservices_client.cdb_bulk_geocode_street_point (query text,
+    street_column text, city_column text default null, state_column text default null, country_column text default null, batch_size integer DEFAULT NULL)
+RETURNS SETOF cdb_dataservices_client.geocoding AS $$
+DECLARE
+  query_row_count integer;
+  enough_quota boolean;
+  remaining_quota integer;
+  max_batch_size integer;
+
+  cartodb_id_batch integer;
+  batches_n integer;
+  DEFAULT_BATCH_SIZE CONSTANT numeric := 100;
+  MAX_SAFE_BATCH_SIZE CONSTANT numeric := 5000;
+
+  temp_table_name text;
+  username text;
+  orgname text;
+  apikey_permissions json;
+BEGIN
+  SELECT u, o, p INTO username, orgname, apikey_permissions FROM cdb_dataservices_client._cdb_entity_config() AS (u text, o text, p json);
+  IF apikey_permissions IS NULL OR NOT apikey_permissions::jsonb ? 'geocoding' THEN
+    RAISE EXCEPTION 'Geocoding permission denied' USING ERRCODE = '01007';
+  END IF;
+
+  -- JSON value stored "" is taken as literal
+  IF username IS NULL OR username = '' OR username = '""' THEN
+    RAISE EXCEPTION 'Username is a mandatory argument, check it out';
+  END IF;
+
+  SELECT csqi.monthly_quota - csqi.used_quota AS remaining_quota, csqi.max_batch_size
+  INTO remaining_quota, max_batch_size
+  FROM cdb_dataservices_client.cdb_service_quota_info_batch() csqi
+  WHERE service = 'hires_geocoder';
+  RAISE DEBUG 'remaining_quota: %; max_batch_size: %', remaining_quota, max_batch_size;
+
+  IF batch_size IS NULL THEN
+    batch_size := max_batch_size;
+  ELSIF batch_size > max_batch_size THEN
+    RAISE EXCEPTION 'batch_size must be lower than %', max_batch_size + 1;
+  END IF;
+
+  IF batch_size > MAX_SAFE_BATCH_SIZE THEN
+    batch_size := MAX_SAFE_BATCH_SIZE;
+  END IF;
+
+  EXECUTE format('SELECT count(1), ceil(count(1)::float/%s) FROM (%s) _x', batch_size, query)
+  INTO query_row_count, batches_n;
+
+  RAISE DEBUG 'cdb_bulk_geocode_street_point --> query_row_count: %; query: %; country: %; state: %; city: %; street: %',
+      query_row_count, query, country_column, state_column, city_column, street_column;
+  SELECT cdb_dataservices_client.cdb_enough_quota('hires_geocoder', query_row_count) INTO enough_quota;
+  IF enough_quota IS NULL OR NOT enough_quota THEN
+    RAISE EXCEPTION 'Remaining quota: %. Estimated cost: %', remaining_quota, query_row_count;
+  END IF;
+
+  RAISE DEBUG 'batches_n: %', batches_n;
+
+  temp_table_name := 'bulk_geocode_street_' || md5(random()::text);
+
+  EXECUTE format('CREATE TEMPORARY TABLE %s ' ||
+   '(cartodb_id integer, the_geom geometry(Point,4326), metadata jsonb)',
+   temp_table_name);
+
+  select
+    coalesce(street_column, ''''''), coalesce(city_column, ''''''),
+    coalesce(state_column, ''''''), coalesce(country_column, '''''')
+  into street_column, city_column, state_column, country_column;
+
+  IF batches_n > 0 THEN
+    FOR cartodb_id_batch in 0..(batches_n - 1)
+    LOOP
+      EXECUTE format(
+        'WITH geocoding_data as (' ||
+        '   SELECT ' ||
+        '      json_build_object(''id'', cartodb_id, ''address'', %s, ''city'', %s, ''state'', %s, ''country'', %s) as data , ' ||
+        '      floor((row_number() over () - 1)::float/$1) as batch' ||
+        '   FROM (%s) _x' ||
+        ') ' ||
+        'INSERT INTO %s SELECT (cdb_dataservices_client._cdb_bulk_geocode_street_point(jsonb_agg(data))).* ' ||
+        'FROM geocoding_data ' ||
+        'WHERE batch = $2', street_column, city_column, state_column, country_column, query, temp_table_name)
+      USING batch_size, cartodb_id_batch;
+
+    END LOOP;
+  END IF;
+
+  RETURN QUERY EXECUTE 'SELECT * FROM ' || quote_ident(temp_table_name);
+END;
+$$ LANGUAGE 'plpgsql' SECURITY DEFINER VOLATILE PARALLEL UNSAFE;

client/cdb_dataservices_client--0.26.2--0.26.1.sql

@@ -0,0 +1,96 @@
+--DO NOT MODIFY THIS FILE, IT IS GENERATED AUTOMATICALLY FROM SOURCES
+-- Complain if script is sourced in psql, rather than via CREATE EXTENSION
+\echo Use "ALTER EXTENSION cdb_dataservices_client UPDATE TO '0.26.1'" to load this file. \quit
+
+-- Make sure we have a sane search path to create/update the extension
+SET search_path = "$user",cartodb,public,cdb_dataservices_client;
+
+CREATE OR REPLACE FUNCTION cdb_dataservices_client.cdb_bulk_geocode_street_point (query text,
+    street_column text, city_column text default null, state_column text default null, country_column text default null, batch_size integer DEFAULT NULL)
+RETURNS SETOF cdb_dataservices_client.geocoding AS $$
+DECLARE
+  query_row_count integer;
+  enough_quota boolean;
+  remaining_quota integer;
+  max_batch_size integer;
+
+  cartodb_id_batch integer;
+  batches_n integer;
+  DEFAULT_BATCH_SIZE CONSTANT numeric := 100;
+  MAX_SAFE_BATCH_SIZE CONSTANT numeric := 5000;
+
+  temp_table_name text;
+  username text;
+  orgname text;
+  apikey_permissions json;
+BEGIN
+  SELECT u, o, p INTO username, orgname, apikey_permissions FROM cdb_dataservices_client._cdb_entity_config() AS (u text, o text, p json);
+  IF apikey_permissions IS NULL OR NOT apikey_permissions::jsonb ? 'geocoding' THEN
+    RAISE EXCEPTION 'Geocoding permission denied' USING ERRCODE = '01007';
+  END IF;
+
+  -- JSON value stored "" is taken as literal
+  IF username IS NULL OR username = '' OR username = '""' THEN
+    RAISE EXCEPTION 'Username is a mandatory argument, check it out';
+  END IF;
+
+  SELECT csqi.monthly_quota - csqi.used_quota AS remaining_quota, csqi.max_batch_size
+  INTO remaining_quota, max_batch_size
+  FROM cdb_dataservices_client.cdb_service_quota_info_batch() csqi
+  WHERE service = 'hires_geocoder';
+  RAISE DEBUG 'remaining_quota: %; max_batch_size: %', remaining_quota, max_batch_size;
+
+  IF batch_size IS NULL THEN
+    batch_size := max_batch_size;
+  ELSIF batch_size > max_batch_size THEN
+    RAISE EXCEPTION 'batch_size must be lower than %', max_batch_size + 1;
+  END IF;
+
+  IF batch_size > MAX_SAFE_BATCH_SIZE THEN
+    batch_size := MAX_SAFE_BATCH_SIZE;
+  END IF;
+
+  EXECUTE format('SELECT count(1), ceil(count(1)::float/%s) FROM (%s) _x', batch_size, query)
+  INTO query_row_count, batches_n;
+
+  RAISE DEBUG 'cdb_bulk_geocode_street_point --> query_row_count: %; query: %; country: %; state: %; city: %; street: %',
+      query_row_count, query, country_column, state_column, city_column, street_column;
+  SELECT cdb_dataservices_client.cdb_enough_quota('hires_geocoder', query_row_count) INTO enough_quota;
+  IF remaining_quota < query_row_count THEN
+    RAISE EXCEPTION 'Remaining quota: %. Estimated cost: %', remaining_quota, query_row_count;
+  END IF;
+
+  RAISE DEBUG 'batches_n: %', batches_n;
+
+  temp_table_name := 'bulk_geocode_street_' || md5(random()::text);
+
+  EXECUTE format('CREATE TEMPORARY TABLE %s ' ||
+   '(cartodb_id integer, the_geom geometry(Point,4326), metadata jsonb)',
+   temp_table_name);
+
+  select
+    coalesce(street_column, ''''''), coalesce(city_column, ''''''),
+    coalesce(state_column, ''''''), coalesce(country_column, '''''')
+  into street_column, city_column, state_column, country_column;
+
+  IF batches_n > 0 THEN
+    FOR cartodb_id_batch in 0..(batches_n - 1)
+    LOOP
+      EXECUTE format(
+        'WITH geocoding_data as (' ||
+        '   SELECT ' ||
+        '      json_build_object(''id'', cartodb_id, ''address'', %s, ''city'', %s, ''state'', %s, ''country'', %s) as data , ' ||
+        '      floor((row_number() over () - 1)::float/$1) as batch' ||
+        '   FROM (%s) _x' ||
+        ') ' ||
+        'INSERT INTO %s SELECT (cdb_dataservices_client._cdb_bulk_geocode_street_point(jsonb_agg(data))).* ' ||
+        'FROM geocoding_data ' ||
+        'WHERE batch = $2', street_column, city_column, state_column, country_column, query, temp_table_name)
+      USING batch_size, cartodb_id_batch;
+
+    END LOOP;
+  END IF;
+
+  RETURN QUERY EXECUTE 'SELECT * FROM ' || quote_ident(temp_table_name);
+END;
+$$ LANGUAGE 'plpgsql' SECURITY DEFINER VOLATILE PARALLEL UNSAFE;

client/cdb_dataservices_client.control

@@ -1,5 +1,5 @@
 comment = 'CartoDB dataservices client API extension'
-default_version = '0.26.1'
+default_version = '0.26.2'
 requires = 'plproxy, cartodb'
 superuser = true
 schema = cdb_dataservices_client

client/sql/21_bulk_geocoding_functions.sql

@@ -49,7 +49,7 @@ BEGIN
   RAISE DEBUG 'cdb_bulk_geocode_street_point --> query_row_count: %; query: %; country: %; state: %; city: %; street: %',
       query_row_count, query, country_column, state_column, city_column, street_column;
   SELECT cdb_dataservices_client.cdb_enough_quota('hires_geocoder', query_row_count) INTO enough_quota;
-  IF remaining_quota < query_row_count THEN
+  IF enough_quota IS NULL OR NOT enough_quota THEN
     RAISE EXCEPTION 'Remaining quota: %. Estimated cost: %', remaining_quota, query_row_count;
   END IF;
 

client/test/expected/21_bulk_geocoding_functions_test.out

@@ -9,6 +9,13 @@ CREATE FUNCTION cdb_dataservices_client.cdb_enough_quota (service TEXT ,input_si
 RETURNS BOOLEAN as $$
   SELECT FALSE;
 $$ LANGUAGE SQL;
+ALTER FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point(searches jsonb) RENAME TO _cdb_bulk_geocode_street_point_mocked;
+CREATE FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point(searches jsonb)
+RETURNS SETOF cdb_dataservices_client.geocoding AS $$
+BEGIN
+  RAISE NOTICE 'called with this searches: %', searches;
+END;
+$$ LANGUAGE 'plpgsql' SECURITY DEFINER STABLE PARALLEL UNSAFE;
 -- No permissions granted
 -- Test bulk size not mandatory (it will get the optimal)
 SELECT cdb_dataservices_client.cdb_bulk_geocode_street_point('select 1 as cartodb_id', '''Valladolid, Spain''', null, null, null, null);
@@ -42,6 +49,18 @@ ERROR:  Remaining quota: 0. Estimated cost: 1
 -- Test quota check by mocking quota 0
 SELECT cdb_dataservices_client.cdb_bulk_geocode_street_point('select 1 as cartodb_id', '''Valladolid, Spain''');
 ERROR:  Remaining quota: 0. Estimated cost: 1
+-- Check that when cdb_enough_quota returns true (ie. when soft_limit is set to true, even if not enough quota)
+-- it is able to proceed with the bulk geocode
+CREATE OR REPLACE FUNCTION cdb_dataservices_client.cdb_enough_quota (service TEXT ,input_size NUMERIC)
+RETURNS BOOLEAN as $$
+  SELECT TRUE;
+$$ LANGUAGE SQL;
+SELECT cdb_dataservices_client.cdb_bulk_geocode_street_point('select 1 as cartodb_id', '''Valladolid, Spain''');
+NOTICE:  called with this searches: [{"id": 1, "city": "", "state": "", "address": "Valladolid, Spain", "country": ""}]
+ cdb_bulk_geocode_street_point 
+-------------------------------
+(0 rows)
+
 -- Remove permissions
 SELECT CDB_Conf_RemoveConf('api_keys_postgres');
  cdb_conf_removeconf 
@@ -51,5 +70,7 @@ SELECT CDB_Conf_RemoveConf('api_keys_postgres');
 
 DROP FUNCTION cdb_dataservices_client.cdb_service_quota_info_batch;
 DROP FUNCTION cdb_dataservices_client.cdb_enough_quota;
+DROP FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point;
 ALTER FUNCTION cdb_dataservices_client.cdb_enough_quota_mocked (service TEXT ,input_size NUMERIC) RENAME TO cdb_enough_quota;
 ALTER FUNCTION cdb_dataservices_client.cdb_service_quota_info_batch_mocked() RENAME TO cdb_service_quota_info_batch;
+ALTER FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point_mocked(searches jsonb) RENAME TO _cdb_bulk_geocode_street_point;

client/test/sql/21_bulk_geocoding_functions_test.sql

@@ -12,6 +12,14 @@ RETURNS BOOLEAN as $$
   SELECT FALSE;
 $$ LANGUAGE SQL;
 
+ALTER FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point(searches jsonb) RENAME TO _cdb_bulk_geocode_street_point_mocked;
+CREATE FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point(searches jsonb)
+RETURNS SETOF cdb_dataservices_client.geocoding AS $$
+BEGIN
+  RAISE NOTICE 'called with this searches: %', searches;
+END;
+$$ LANGUAGE 'plpgsql' SECURITY DEFINER STABLE PARALLEL UNSAFE;
+
 -- No permissions granted
 -- Test bulk size not mandatory (it will get the optimal)
 SELECT cdb_dataservices_client.cdb_bulk_geocode_street_point('select 1 as cartodb_id', '''Valladolid, Spain''', null, null, null, null);
@@ -32,12 +40,21 @@ SELECT cdb_dataservices_client.cdb_bulk_geocode_street_point('select 1 as cartod
 -- Test quota check by mocking quota 0
 SELECT cdb_dataservices_client.cdb_bulk_geocode_street_point('select 1 as cartodb_id', '''Valladolid, Spain''');
 
+-- Check that when cdb_enough_quota returns true (ie. when soft_limit is set to true, even if not enough quota)
+-- it is able to proceed with the bulk geocode
+CREATE OR REPLACE FUNCTION cdb_dataservices_client.cdb_enough_quota (service TEXT ,input_size NUMERIC)
+RETURNS BOOLEAN as $$
+  SELECT TRUE;
+$$ LANGUAGE SQL;
+SELECT cdb_dataservices_client.cdb_bulk_geocode_street_point('select 1 as cartodb_id', '''Valladolid, Spain''');
+
 -- Remove permissions
 SELECT CDB_Conf_RemoveConf('api_keys_postgres');
 
 DROP FUNCTION cdb_dataservices_client.cdb_service_quota_info_batch;
 DROP FUNCTION cdb_dataservices_client.cdb_enough_quota;
+DROP FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point;
 
 ALTER FUNCTION cdb_dataservices_client.cdb_enough_quota_mocked (service TEXT ,input_size NUMERIC) RENAME TO cdb_enough_quota;
 ALTER FUNCTION cdb_dataservices_client.cdb_service_quota_info_batch_mocked() RENAME TO cdb_service_quota_info_batch;
-
+ALTER FUNCTION cdb_dataservices_client._cdb_bulk_geocode_street_point_mocked(searches jsonb) RENAME TO _cdb_bulk_geocode_street_point;

server/lib/python/cartodb_services/cartodb_services/tools/qps.py

@@ -2,13 +2,17 @@ import time
 import random
 from datetime import datetime
 from exceptions import TimeoutException
+import re
 
 DEFAULT_RETRY_TIMEOUT = 60
 DEFAULT_QUERIES_PER_SECOND = 10
 
-TOMTOM_403_RATE_LIMIT_HEADER = 'Account Over Queries Per Second Limit'
+TOMTOM_403_RATE_LIMIT_HEADERS = [
+    'Account Over Queries Per Second Limit',
+    'Developer Over Qps'
+]
 TOMTOM_DETAIL_HEADER = 'X-Error-Detail-Header'
-
+TOMTOM_403_RATE_LIMIT_HEADER_PATTERN = re.compile('|'.join(TOMTOM_403_RATE_LIMIT_HEADERS), re.IGNORECASE)
 
 def qps_retry(original_function=None, **options):
     """ Query Per Second retry decorator
@@ -49,9 +53,11 @@ class QPSService:
                 response = getattr(e, 'response', None)
                 if response is not None:
                     if self._provider is not None and self._provider == 'tomtom' and (response.status_code == 403):
-                        if response.headers.get(TOMTOM_DETAIL_HEADER) != TOMTOM_403_RATE_LIMIT_HEADER:
+                        detail_header = response.headers.get(TOMTOM_DETAIL_HEADER)
+                        if detail_header and TOMTOM_403_RATE_LIMIT_HEADER_PATTERN.search(detail_header):
+                            self.retry(start_time, attempt_number)
+                        else:
                             raise e
-                        self.retry(start_time, attempt_number)
                     elif response.status_code == 429:
                         self.retry(start_time, attempt_number)
                     else:

server/lib/python/cartodb_services/setup.py

@@ -10,7 +10,7 @@ from setuptools import setup, find_packages
 setup(
     name='cartodb_services',
 
-    version='0.21.1',
+    version='0.21.2',
 
     description='CartoDB Services API Python Library',