New version 0.19.0

Add CDB_EstimateRowCount function

.travis.yml

@@ -1,14 +1,41 @@
 language: c
 
 addons:
-  postgresql: 9.3
+  postgresql: 9.5
 
 before_install:
+  # Add custom PPAs from cartodb
+  - sudo add-apt-repository -y ppa:cartodb/postgresql-9.5
+  - sudo add-apt-repository -y ppa:cartodb/gis
+  - sudo add-apt-repository -y ppa:cartodb/gis-testing
   - sudo apt-get update
-  #- sudo apt-get install -q postgresql-9.3-postgis-2.1
-  - sudo apt-get update
-  - sudo apt-get install -q postgresql-server-dev-9.3
-  - sudo apt-get install -q postgresql-plpython-9.3
+
+  # Force instalation of libgeos-3.5.0 (presumably needed because of existing version of postgis)
+  - sudo apt-get -y install libgeos-3.5.0=3.5.0-1cdb2
+
+  # Install postgres db and build deps
+  - sudo /etc/init.d/postgresql stop # stop travis default instance
+  - sudo apt-get -y remove --purge postgresql-9.1
+  - sudo apt-get -y remove --purge postgresql-9.2
+  - sudo apt-get -y remove --purge postgresql-9.3
+  - sudo apt-get -y remove --purge postgresql-9.4
+  - sudo apt-get -y remove --purge postgresql-9.5
+  - sudo rm -rf /var/lib/postgresql/
+  - sudo rm -rf /var/log/postgresql/
+  - sudo rm -rf /etc/postgresql/
+  - sudo apt-get -y remove --purge postgis-2.2
+  - sudo apt-get -y autoremove
+
+  - sudo apt-get -y install postgresql-9.5=9.5.2-3cdb2
+  - sudo apt-get -y install postgresql-server-dev-9.5=9.5.2-3cdb2
+  - sudo apt-get -y install postgresql-plpython-9.5=9.5.2-3cdb2
+  - sudo apt-get -y install postgresql-9.5-postgis-scripts=2.2.2.0-cdb2
+  - sudo apt-get -y install postgresql-9.5-postgis-2.2=2.2.2.0-cdb2
+
+  # configure it to accept local connections from postgres
+  - echo -e "# TYPE  DATABASE        USER            ADDRESS                 METHOD \nlocal   all             postgres                                trust\nlocal   all             all                                     trust\nhost    all             all             127.0.0.1/32            trust" \
+    | sudo tee /etc/postgresql/9.5/main/pg_hba.conf
+  - sudo /etc/init.d/postgresql restart 9.5
 
 script:
   - make

CONTRIBUTING.md

@@ -62,3 +62,7 @@ A useful query:
 ```sql
 SELECT * FROM pg_extension_update_paths('cartodb') WHERE path IS NOT NULL AND source = cdb_version();
 ```
+
+## Submitting Contributions
+
+* You will need to sign a Contributor License Agreement (CLA) before making a submission. [Learn more here](https://carto.com/contributions).

Makefile

@@ -1,7 +1,7 @@
 # cartodb/Makefile
 
 EXTENSION = cartodb
-EXTVERSION = 0.18.4
+EXTVERSION = 0.19.0
 
 SED = sed
 
@@ -78,6 +78,8 @@ UPGRADABLE = \
   0.18.2 \
   0.18.3 \
   0.18.4 \
+  0.18.5 \
+  0.19.0 \
   $(EXTVERSION)dev \
   $(EXTVERSION)next \
   $(END)

NEWS.md

@@ -1,3 +1,15 @@
+0.19.0 (2017-04-11)
+
+* Add new function `CDB_EstimateRowCount` #295
+
+0.18.5 (2016-11-30)
+
+* Add to new overview creation strategies #290
+* Fix tests: race condition with publicuser #157
+* Fix: CDB_Stats divisions by zero #181
+* Better implementation of `CDB_EqualIntervalBins` #244
+* New tests for binning functions #249
+
 0.18.4 (2016-11-04)
 
 * No functional changes; fixes the migration from previous versions #288

doc/CDB_EstimateRowCount.md

@@ -0,0 +1,25 @@
+Estimate the number of rows of a query.
+
+
+#### Using the function
+
+```sql
+SELECT CDB_EstimateRowCount($$
+  UPDATE addresses SET the_geom = cdb_geocode_street_point(addr, city, state, 'US');
+$$) AS row_count;
+```
+
+Result:
+
+```
+ row_count
+-----------
+         5
+(1 row)
+```
+
+#### Arguments
+
+CDB_EstimateRowCount(query)
+
+* **query** text: the SQL query to estimate the row count for.

scripts-available/CDB_EqualIntervalBins.sql

@@ -1,8 +1,8 @@
 --
 -- Calculate the equal interval bins for a given column
 --
--- @param in_array A numeric array of numbers to determine the best
---                   to determine the bin boundary
+-- @param in_array An array of numbers to determine the best
+--                   bin boundary
 --
 -- @param breaks The number of bins you want to find.
 --  
@@ -11,27 +11,14 @@
 -- 
 --
 
-CREATE OR REPLACE FUNCTION CDB_EqualIntervalBins ( in_array NUMERIC[], breaks INT ) RETURNS NUMERIC[] as $$
-DECLARE 
-    diff numeric;
-    min_val numeric;
-    max_val numeric;
-    tmp_val numeric;
-    i INT := 1;
-    reply numeric[];
-BEGIN
-    SELECT min(e), max(e) INTO min_val, max_val FROM ( SELECT unnest(in_array) e ) x WHERE e IS NOT NULL;
-    diff = (max_val - min_val) / breaks::numeric;
-    LOOP
-        IF i < breaks THEN
-            tmp_val = min_val + i::numeric * diff;
-            reply = array_append(reply, tmp_val);
-            i := i+1;
-        ELSE
-            reply = array_append(reply, max_val);
-            EXIT;
-        END IF;
-    END LOOP;
-    RETURN reply;
-END;
-$$ language plpgsql IMMUTABLE;
+CREATE OR REPLACE FUNCTION CDB_EqualIntervalBins ( in_array anyarray, breaks INT ) RETURNS anyarray as $$
+WITH stats AS (
+  SELECT min(e), (max(e)-min(e))/breaks AS del
+    FROM (SELECT unnest(in_array) e) AS p)
+SELECT array_agg(bins)
+  FROM (
+    SELECT min + generate_series(1,breaks)*del AS bins
+      FROM stats) q;
+$$ LANGUAGE SQL IMMUTABLE;
+
+DROP FUNCTION IF EXISTS CDB_EqualIntervalBins( numeric[], integer);

scripts-available/CDB_EstimateRowCount.sql

@@ -0,0 +1,31 @@
+-- Internal function to generate stats for a table if they don't exist
+CREATE OR REPLACE FUNCTION _CDB_GenerateStats(reloid REGCLASS)
+RETURNS VOID
+AS $$
+DECLARE
+  has_stats BOOLEAN;
+BEGIN
+  SELECT EXISTS (
+    SELECT * FROM pg_catalog.pg_statistic WHERE starelid = reloid
+  ) INTO has_stats;
+  IF NOT has_stats THEN
+    EXECUTE Format('ANALYZE %s;', reloid);
+  END IF;
+END
+$$ LANGUAGE 'plpgsql' VOLATILE STRICT SECURITY DEFINER;
+
+-- Return a row count estimate of the result of a query using statistics
+CREATE OR REPLACE FUNCTION CDB_EstimateRowCount(query text)
+RETURNS Numeric
+AS $$
+DECLARE
+  plan JSON;
+BEGIN
+  -- Make sure statistics exist for all the tables of the query
+  PERFORM _CDB_GenerateStats(tabname) FROM  unnest(CDB_QueryTablesText(query)) AS tabname;
+
+  -- Use the query planner to obtain an estimate of the number of result rows
+  EXECUTE 'EXPLAIN (FORMAT JSON) ' || query INTO STRICT plan;
+  RETURN plan->0->'Plan'->'Plan Rows';
+END
+$$ LANGUAGE 'plpgsql' VOLATILE STRICT;

scripts-available/CDB_Overviews.sql

@@ -697,6 +697,356 @@ AS $$
   END;
 $$ LANGUAGE PLPGSQL;
 
+
+CREATE OR REPLACE FUNCTION _CDB_GridCluster_Reduce_Strategy(reloid REGCLASS, ref_z INTEGER, overview_z INTEGER, grid_px FLOAT8 DEFAULT NULL)
+RETURNS REGCLASS
+AS $$
+  DECLARE
+    overview_rel TEXT;
+    reduction FLOAT8;
+    base_name TEXT;
+    pixel_m FLOAT8;
+    grid_m FLOAT8;
+    offset_m FLOAT8;
+    offset_x TEXT;
+    offset_y TEXT;
+    cell_x TEXT;
+    cell_y TEXT;
+    aggr_attributes TEXT;
+    attributes TEXT;
+    columns TEXT;
+    gtypes TEXT[];
+    schema_name TEXT;
+    table_name TEXT;
+    point_geom TEXT;
+  BEGIN
+    SELECT _CDB_GeometryTypes(reloid) INTO gtypes;
+    IF gtypes IS NULL OR array_upper(gtypes, 1) <> 1 OR gtypes[1] <> 'ST_Point' THEN
+      -- This strategy only supports datasets with point geomety
+      RETURN NULL;
+    END IF;
+
+    --TODO: check applicability: geometry type, minimum number of points...
+
+    overview_rel := _CDB_Overview_Name(reloid, ref_z, overview_z);
+
+    -- Grid size in pixels at Z level overview_z
+    IF grid_px IS NULL THEN
+      grid_px := 1.0;
+    END IF;
+
+    SELECT * FROM _cdb_split_table_name(reloid) INTO schema_name, table_name;
+
+    -- pixel_m: size of a pixel in webmercator units (meters)
+    SELECT CDB_XYZ_Resolution(overview_z) INTO pixel_m;
+    -- grid size in meters
+    grid_m = grid_px * pixel_m;
+
+    attributes := _CDB_Aggregable_Attributes_Expression(reloid);
+    aggr_attributes := _CDB_Aggregated_Attributes_Expression(reloid);
+    IF attributes <> '' THEN
+      attributes := ', ' || attributes;
+    END IF;
+    IF aggr_attributes <> '' THEN
+      aggr_attributes := aggr_attributes || ', ';
+    END IF;
+
+    -- Center of each cell:
+    cell_x := Format('gx*%1$s + %2$s', grid_m, grid_m/2);
+    cell_y := Format('gy*%1$s + %2$s', grid_m, grid_m/2);
+
+    -- Displacement to the nearest pixel center:
+    IF MOD(grid_px::numeric, 1.0::numeric) = 0 THEN
+      offset_m := pixel_m/2 - MOD((grid_m/2)::numeric, pixel_m::numeric)::float8;
+      offset_x := Format('%s', offset_m);
+      offset_y := Format('%s', offset_m);
+    ELSE
+      offset_x := Format('%2$s/2 - MOD((%1$s)::numeric, (%2$s)::numeric)::float8', cell_x, pixel_m);
+      offset_y := Format('%2$s/2 - MOD((%1$s)::numeric, (%2$s)::numeric)::float8', cell_y, pixel_m);
+    END IF;
+
+    point_geom := Format('ST_SetSRID(ST_MakePoint(%1$s + %3$s, %2$s + %4$s), 3857)', cell_x, cell_y, offset_x, offset_y);
+
+    -- compute the resulting columns in the same order as in the base table
+    WITH cols AS (
+      SELECT
+        CASE c
+        WHEN 'cartodb_id' THEN 'cartodb_id'
+        WHEN 'the_geom' THEN
+          Format('ST_Transform(%s, 4326) AS the_geom', point_geom)
+        WHEN 'the_geom_webmercator' THEN
+           Format('%s AS the_geom_webmercator', point_geom)
+        ELSE c
+        END AS column
+        FROM CDB_ColumnNames(reloid) c
+    )
+    SELECT string_agg(s.column, ',') FROM (
+      SELECT * FROM cols
+    ) AS s INTO columns;
+
+    IF NOT columns LIKE '%_feature_count%' THEN
+      columns := columns || ', n AS _feature_count';
+    END IF;
+
+    EXECUTE Format('DROP TABLE IF EXISTS %I.%I CASCADE;', schema_name, overview_rel);
+
+    -- Now we cluster the data using a grid of size grid_m
+    -- and selecte the centroid (average coordinates) of each cluster.
+    -- If we had a selected numeric attribute of interest we could use it
+    -- as a weight for the average coordinates.
+    EXECUTE Format('
+      CREATE TABLE %7$I.%3$I AS
+         WITH clusters AS (
+           SELECT
+             %5$s
+             count(*) AS n,
+             Floor(ST_X(f.the_geom_webmercator)/%2$s)::int AS gx,
+             Floor(ST_Y(f.the_geom_webmercator)/%2$s)::int AS gy,
+             MIN(cartodb_id) AS cartodb_id
+          FROM %1$s f
+          GROUP BY gx, gy
+         )
+         SELECT %6$s FROM clusters
+    ', reloid::text, grid_m, overview_rel, attributes, aggr_attributes, columns, schema_name);
+
+    RETURN Format('%I.%I', schema_name, overview_rel)::regclass;
+  END;
+$$ LANGUAGE PLPGSQL;
+
+-- This strategy places the aggregation of each cluster at the centroid of the cluster members.
+CREATE OR REPLACE FUNCTION _CDB_GridClusterCentroid_Reduce_Strategy(reloid REGCLASS, ref_z INTEGER, overview_z INTEGER, grid_px FLOAT8 DEFAULT NULL)
+RETURNS REGCLASS
+AS $$
+  DECLARE
+    overview_rel TEXT;
+    reduction FLOAT8;
+    base_name TEXT;
+    pixel_m FLOAT8;
+    grid_m FLOAT8;
+    offset_m FLOAT8;
+    offset_x TEXT;
+    offset_y TEXT;
+    cell_x TEXT;
+    cell_y TEXT;
+    aggr_attributes TEXT;
+    attributes TEXT;
+    columns TEXT;
+    gtypes TEXT[];
+    schema_name TEXT;
+    table_name TEXT;
+    point_geom TEXT;
+  BEGIN
+    SELECT _CDB_GeometryTypes(reloid) INTO gtypes;
+    IF gtypes IS NULL OR array_upper(gtypes, 1) <> 1 OR gtypes[1] <> 'ST_Point' THEN
+      -- This strategy only supports datasets with point geomety
+      RETURN NULL;
+    END IF;
+
+    --TODO: check applicability: geometry type, minimum number of points...
+
+    overview_rel := _CDB_Overview_Name(reloid, ref_z, overview_z);
+
+    -- Grid size in pixels at Z level overview_z
+    IF grid_px IS NULL THEN
+      grid_px := 1.0;
+    END IF;
+
+    SELECT * FROM _cdb_split_table_name(reloid) INTO schema_name, table_name;
+
+    -- pixel_m: size of a pixel in webmercator units (meters)
+    SELECT CDB_XYZ_Resolution(overview_z) INTO pixel_m;
+    -- grid size in meters
+    grid_m = grid_px * pixel_m;
+
+    attributes := _CDB_Aggregable_Attributes_Expression(reloid);
+    aggr_attributes := _CDB_Aggregated_Attributes_Expression(reloid);
+    IF attributes <> '' THEN
+      attributes := ', ' || attributes;
+    END IF;
+    IF aggr_attributes <> '' THEN
+      aggr_attributes := aggr_attributes || ', ';
+    END IF;
+
+    -- Center of each cell:
+    cell_x := Format('gx*%1$s + %2$s', grid_m, grid_m/2);
+    cell_y := Format('gy*%1$s + %2$s', grid_m, grid_m/2);
+
+    -- Displacement to the nearest pixel center:
+    IF MOD(grid_px::numeric, 1.0::numeric) = 0 THEN
+      offset_m := pixel_m/2 - MOD((grid_m/2)::numeric, pixel_m::numeric)::float8;
+      offset_x := Format('%s', offset_m);
+      offset_y := Format('%s', offset_m);
+    ELSE
+      offset_x := Format('%2$s/2 - MOD((%1$s)::numeric, (%2$s)::numeric)::float8', cell_x, pixel_m);
+      offset_y := Format('%2$s/2 - MOD((%1$s)::numeric, (%2$s)::numeric)::float8', cell_y, pixel_m);
+    END IF;
+
+    point_geom := Format('ST_SetSRID(ST_MakePoint(%1$s + %3$s, %2$s + %4$s), 3857)', cell_x, cell_y, offset_x, offset_y);
+
+    -- compute the resulting columns in the same order as in the base table
+    WITH cols AS (
+      SELECT
+        CASE c
+        WHEN 'cartodb_id' THEN 'cartodb_id'
+        WHEN 'the_geom' THEN
+          'ST_Transform(ST_SetSRID(ST_MakePoint(_sum_of_x/n, _sum_of_y/n), 3857), 4326) AS the_geom'
+        WHEN 'the_geom_webmercator' THEN
+          'ST_SetSRID(ST_MakePoint(_sum_of_x/n, _sum_of_y/n), 3857) AS the_geom_webmercator'
+        ELSE c
+        END AS column
+        FROM CDB_ColumnNames(reloid) c
+    )
+    SELECT string_agg(s.column, ',') FROM (
+      SELECT * FROM cols
+    ) AS s INTO columns;
+
+    IF NOT columns LIKE '%_feature_count%' THEN
+      columns := columns || ', n AS _feature_count';
+    END IF;
+
+    EXECUTE Format('DROP TABLE IF EXISTS %I.%I CASCADE;', schema_name, overview_rel);
+
+    -- Now we cluster the data using a grid of size grid_m
+    -- and selecte the centroid (average coordinates) of each cluster.
+    -- If we had a selected numeric attribute of interest we could use it
+    -- as a weight for the average coordinates.
+    EXECUTE Format('
+      CREATE TABLE %7$I.%3$I AS
+         WITH clusters AS (
+           SELECT
+             %5$s
+             count(*) AS n,
+             SUM(ST_X(f.the_geom_webmercator)) AS _sum_of_x,
+             SUM(ST_Y(f.the_geom_webmercator)) AS _sum_of_y,
+             Floor(ST_Y(f.the_geom_webmercator)/%2$s)::int AS gy,
+             Floor(ST_X(f.the_geom_webmercator)/%2$s)::int AS gx,
+             MIN(cartodb_id) AS cartodb_id
+          FROM %1$s f
+          GROUP BY gx, gy
+         )
+         SELECT %6$s FROM clusters
+    ', reloid::text, grid_m, overview_rel, attributes, aggr_attributes, columns, schema_name);
+
+    RETURN Format('%I.%I', schema_name, overview_rel)::regclass;
+  END;
+$$ LANGUAGE PLPGSQL;
+
+-- This strategy places the aggregation of each cluster at the position of one of the cluster members.
+CREATE OR REPLACE FUNCTION _CDB_GridClusterSample_Reduce_Strategy(reloid REGCLASS, ref_z INTEGER, overview_z INTEGER, grid_px FLOAT8 DEFAULT NULL)
+RETURNS REGCLASS
+AS $$
+  DECLARE
+    overview_rel TEXT;
+    reduction FLOAT8;
+    base_name TEXT;
+    pixel_m FLOAT8;
+    grid_m FLOAT8;
+    offset_m FLOAT8;
+    offset_x TEXT;
+    offset_y TEXT;
+    cell_x TEXT;
+    cell_y TEXT;
+    aggr_attributes TEXT;
+    attributes TEXT;
+    columns TEXT;
+    gtypes TEXT[];
+    schema_name TEXT;
+    table_name TEXT;
+    point_geom TEXT;
+  BEGIN
+    SELECT _CDB_GeometryTypes(reloid) INTO gtypes;
+    IF gtypes IS NULL OR array_upper(gtypes, 1) <> 1 OR gtypes[1] <> 'ST_Point' THEN
+      -- This strategy only supports datasets with point geomety
+      RETURN NULL;
+    END IF;
+
+    --TODO: check applicability: geometry type, minimum number of points...
+
+    overview_rel := _CDB_Overview_Name(reloid, ref_z, overview_z);
+
+    -- Grid size in pixels at Z level overview_z
+    IF grid_px IS NULL THEN
+      grid_px := 1.0;
+    END IF;
+
+    SELECT * FROM _cdb_split_table_name(reloid) INTO schema_name, table_name;
+
+    -- pixel_m: size of a pixel in webmercator units (meters)
+    SELECT CDB_XYZ_Resolution(overview_z) INTO pixel_m;
+    -- grid size in meters
+    grid_m = grid_px * pixel_m;
+
+    attributes := _CDB_Aggregable_Attributes_Expression(reloid);
+    aggr_attributes := _CDB_Aggregated_Attributes_Expression(reloid);
+    IF attributes <> '' THEN
+      attributes := ', ' || attributes;
+    END IF;
+    IF aggr_attributes <> '' THEN
+      aggr_attributes := aggr_attributes || ', ';
+    END IF;
+
+    -- Center of each cell:
+    cell_x := Format('gx*%1$s + %2$s', grid_m, grid_m/2);
+    cell_y := Format('gy*%1$s + %2$s', grid_m, grid_m/2);
+
+    -- Displacement to the nearest pixel center:
+    IF MOD(grid_px::numeric, 1.0::numeric) = 0 THEN
+      offset_m := pixel_m/2 - MOD((grid_m/2)::numeric, pixel_m::numeric)::float8;
+      offset_x := Format('%s', offset_m);
+      offset_y := Format('%s', offset_m);
+    ELSE
+      offset_x := Format('%2$s/2 - MOD((%1$s)::numeric, (%2$s)::numeric)::float8', cell_x, pixel_m);
+      offset_y := Format('%2$s/2 - MOD((%1$s)::numeric, (%2$s)::numeric)::float8', cell_y, pixel_m);
+    END IF;
+
+    point_geom := Format('ST_SetSRID(ST_MakePoint(%1$s + %3$s, %2$s + %4$s), 3857)', cell_x, cell_y, offset_x, offset_y);
+
+    -- compute the resulting columns in the same order as in the base table
+    WITH cols AS (
+      SELECT
+        CASE c
+        WHEN 'cartodb_id' THEN 'cartodb_id'
+        ELSE c
+        END AS column
+        FROM CDB_ColumnNames(reloid) c
+    )
+    SELECT string_agg(s.column, ',') FROM (
+      SELECT * FROM cols
+    ) AS s INTO columns;
+
+    IF NOT columns LIKE '%_feature_count%' THEN
+      columns := columns || ', n AS _feature_count';
+    END IF;
+
+    EXECUTE Format('DROP TABLE IF EXISTS %I.%I CASCADE;', schema_name, overview_rel);
+
+    -- Now we cluster the data using a grid of size grid_m
+    -- and select the centroid (average coordinates) of each cluster.
+    -- If we had a selected numeric attribute of interest we could use it
+    -- as a weight for the average coordinates.
+    EXECUTE Format('
+      CREATE TABLE %7$I.%3$I AS
+         WITH clusters AS (
+           SELECT
+             %5$s
+             count(*) AS n,
+             Floor(ST_X(_f.the_geom_webmercator)/%2$s)::int AS gx,
+             Floor(ST_Y(_f.the_geom_webmercator)/%2$s)::int AS gy,
+             MIN(cartodb_id) AS cartodb_id
+          FROM %1$s _f
+          GROUP BY gx, gy
+         ),
+         cluster_geom AS (
+           SELECT the_geom, the_geom_webmercator, clusters.*
+             FROM clusters INNER JOIN %1$s _g ON (clusters.cartodb_id = _g.cartodb_id)
+         )
+         SELECT %6$s FROM cluster_geom
+    ', reloid::text, grid_m, overview_rel, attributes, aggr_attributes, columns, schema_name);
+
+    RETURN Format('%I.%I', schema_name, overview_rel)::regclass;
+  END;
+$$ LANGUAGE PLPGSQL;
+
 -- Create overview tables for a dataset.
 -- Scope: public
 -- Parameters:

scripts-available/CDB_Stats.sql

@@ -4,7 +4,7 @@
 -- @param in_array A numeric array of numbers
 --
 -- Returns: statistical quantity chosen
--- 
+--
 -- References: http://www.itl.nist.gov/div898/handbook/eda/section3/eda35b.htm
 --
 
@@ -13,17 +13,21 @@ CREATE OR REPLACE FUNCTION CDB_Kurtosis ( in_array NUMERIC[] ) RETURNS NUMERIC a
 DECLARE
     a numeric;
     c numeric;
-    s numeric;
     k numeric;
 BEGIN
-    SELECT AVG(e), COUNT(e)::numeric, stddev(e) INTO a, c, s FROM ( SELECT unnest(in_array) e ) x;
+    SELECT AVG(e), COUNT(e)::numeric * power(stddev(e),4) INTO a, c FROM ( SELECT unnest(in_array) e ) x;
 
-    EXECUTE 'SELECT sum(power($1 - e, 4)) / ( $2 * power($3, 4)) - 3
-             FROM (SELECT unnest($4) e ) x'
-    INTO k
-    USING a, c, s, in_array;
+    IF c=0 THEN
+      RETURN 0;
+    ELSE
 
-    RETURN k;
+      EXECUTE 'SELECT sum(power($1 - e, 4)) / ($2 ) - 3
+             FROM (SELECT unnest($3) e ) x'
+      INTO k
+      USING a, c, in_array;
+
+      RETURN k;
+    END IF;
 END;
 $$ language plpgsql IMMUTABLE;
 
@@ -32,16 +36,18 @@ CREATE OR REPLACE FUNCTION CDB_Skewness ( in_array NUMERIC[] ) RETURNS NUMERIC a
 DECLARE
     a numeric;
     c numeric;
-    s numeric;
     sk numeric;
 BEGIN
-    SELECT AVG(e), COUNT(e)::numeric, stddev(e) INTO a, c, s FROM ( SELECT unnest(in_array) e ) x;
+    SELECT AVG(e), COUNT(e)::numeric * power(stddev(e),3) INTO a, c FROM ( SELECT unnest(in_array) e ) x;
+    IF c=0 THEN
+      RETURN 0;
+    ELSE
+      EXECUTE 'SELECT sum(power($1 - e, 3)) / ( $2 )
+             FROM (SELECT unnest($3) e ) x'
+      INTO sk
+      USING a, c, in_array;
 
-    EXECUTE 'SELECT sum(power($1 - e, 3)) / ( $2 * power($3, 3))
-             FROM (SELECT unnest($4) e ) x'
-    INTO sk
-    USING a, c, s, in_array;
-
-    RETURN sk;
+      RETURN sk;
+    END IF;
 END;
 $$ language plpgsql IMMUTABLE;

scripts-enabled/280-CDB_EstimateRowCount.sql

@@ -0,0 +1 @@
+../scripts-available/CDB_EstimateRowCount.sql

test/CDB_DigitSeparatorTest_expect

@@ -1,5 +1,6 @@
 BEGIN
 CREATE TABLE
+COPY 3
 none||
 only_com_dec|.|,
 only_dot_dec|,|.

test/CDB_EqualIntervalBinsTest.sql

@@ -2,4 +2,10 @@ WITH data AS (
     SELECT array_agg(x::numeric) s FROM generate_series(1,300) x 
         WHERE x % 5 != 0 AND x % 7 != 0
     ) 
-SELECT round(unnest(CDB_EqualIntervalBins(s, 7)),7) FROM data
+SELECT round(unnest(CDB_EqualIntervalBins(s, 7)),7) FROM data;
+
+WITH data_nulls AS (
+    SELECT array_agg(CASE WHEN x % 2 != 0 THEN x ELSE NULL END::numeric) s FROM generate_series(1,100) x
+        WHERE x % 5 != 0 AND x % 7 != 0
+    )
+SELECT round(unnest(CDB_EqualIntervalBins(s, 7)),7) FROM data_nulls;

test/CDB_EqualIntervalBinsTest_expect

@@ -5,3 +5,10 @@
 213.8571429
 256.4285714
 299.0000000
+15.0000000
+29.0000000
+43.0000000
+57.0000000
+71.0000000
+85.0000000
+99.0000000

test/CDB_EstimateRowCountTest.sql

@@ -0,0 +1,10 @@
+SET client_min_messages TO error;
+\set VERBOSITY terse
+CREATE TABLE tmptab1(id INT);
+INSERT INTO tmptab1(id) VALUES (1), (2), (3);
+CREATE TABLE tmptab2(id INT, value NUMERIC);
+INSERT INTO tmptab2(id, value) VALUES (1, 10.0), (2, 20.0);
+SELECT CDB_EstimateRowCount('SELECT SUM(value) FROM tmptab1 INNER JOIN tmptab2 ON (tmptab1.id = tmptab2.id);') AS row_count;
+SELECT CDB_EstimateRowCount('UPDATE tmptab2 SET value = 30 WHERE id=2;') AS row_count;
+DROP TABLE tmptab2;
+DROP TABLE tmptab1;

test/CDB_EstimateRowCountTest_expect

@@ -0,0 +1,9 @@
+SET
+CREATE TABLE
+INSERT 0 3
+CREATE TABLE
+INSERT 0 2
+1
+1
+DROP TABLE
+DROP TABLE

test/CDB_HeadsTailsBinsTest.sql

@@ -1,5 +1,11 @@
 WITH data AS (
-    SELECT array_agg(x) x FROM generate_series(1,100) x 
+    SELECT array_agg(x::numeric) s FROM generate_series(1,100) x
         WHERE x % 5 != 0 AND x % 7 != 0
-    ) 
-SELECT round(unnest(CDB_HeadsTailsBins(x, 7)),2) FROM data
+    )
+SELECT round(unnest(CDB_HeadsTailsBins(s, 7)),2) FROM data;
+
+WITH data_nulls AS (
+    SELECT array_agg(CASE WHEN x % 2 != 0 THEN x ELSE NULL END::numeric) s FROM generate_series(1,100) x
+        WHERE x % 5 != 0 AND x % 7 != 0
+    )
+SELECT round(unnest(CDB_HeadsTailsBins(s, 7)),2) FROM data_nulls;

test/CDB_HeadsTailsBinsTest_expect

@@ -5,3 +5,9 @@
 96.50
 98.00
 99.00
+49.76
+74.65
+88.50
+94.50
+98.00
+99.00

test/CDB_JenksBinsTest.sql

@@ -1,5 +1,11 @@
 WITH data AS (
-    SELECT array_agg(x) x FROM generate_series(1,100) x 
+    SELECT array_agg(x::numeric) s FROM generate_series(1,300) x
         WHERE x % 5 != 0 AND x % 7 != 0
     ) 
-SELECT unnest(CDB_JenksBins(x, 7)) FROM data
+SELECT unnest(CDB_JenksBins(s, 7)) FROM data;
+
+WITH data_nulls AS (
+    SELECT array_agg(CASE WHEN x % 2 != 0 THEN x ELSE NULL END::numeric) s FROM generate_series(1,300) x
+        WHERE x % 5 != 0 AND x % 7 != 0
+    )
+SELECT unnest(CDB_JenksBins(s, 7)) FROM data_nulls;

test/CDB_JenksBinsTest_expect

@@ -1,7 +1,14 @@
-13
-29
 43
-57
-71
-83
-99
+86
+129
+172
+213
+257
+299
+37
+51
+97
+157
+213
+241
+

test/CDB_QuantileBinsTest.sql

@@ -1,5 +1,11 @@
 WITH data AS (
-    SELECT array_agg(x) x FROM generate_series(1,100) x 
+    SELECT array_agg(x::numeric) s FROM generate_series(1,100) x
         WHERE x % 5 != 0 AND x % 7 != 0
     ) 
-SELECT unnest(CDB_QuantileBins(x, 7)) FROM data
+SELECT unnest(CDB_QuantileBins(s, 7)) FROM data;
+
+WITH data_nulls AS (
+    SELECT array_agg(CASE WHEN x % 2 != 0 THEN x ELSE NULL END::numeric) s FROM generate_series(1,100) x
+        WHERE x % 5 != 0 AND x % 7 != 0
+    )
+SELECT unnest(CDB_QuantileBins(s, 7)) FROM data_nulls;

test/CDB_QuantileBinsTest_expect

@@ -4,4 +4,11 @@
 57
 71
 86
+99
+29
+57
+87
+
+
+
 99

test/CDB_QueryStatementsTest.sql

@@ -1,3 +1,6 @@
+SET client_min_messages TO error;
+\set VERBOSITY terse
+
 WITH q AS ( SELECT CDB_QueryStatements('
 SELECT * FROM geometry_columns;
 ') as statement )

test/CDB_QueryStatementsTest_expect

@@ -1,3 +1,4 @@
+SET
 1|1|SELECT * FROM geometry_columns
 2|1|SELECT * FROM geometry_columns
 3|1|SELECT * FROM geometry_columns

test/CDB_QueryTablesTest.sql

@@ -1,3 +1,5 @@
+SET client_min_messages TO warning;
+\set VERBOSITY terse
 
 WITH inp AS ( select 'SELECT * FROM geometry_columns'::text as q )
  SELECT q, CDB_QueryTables(q) from inp;

test/CDB_QueryTablesTest_expect

@@ -1,17 +1,14 @@
+SET
 SELECT * FROM geometry_columns|{pg_catalog.pg_attribute,pg_catalog.pg_class,pg_catalog.pg_constraint,pg_catalog.pg_namespace,pg_catalog.pg_type}
 SELECT a.attname FROM pg_class c JOIN pg_attribute a on (a.attrelid = c.oid)|{pg_catalog.pg_attribute,pg_catalog.pg_class}
 CREATE table "my'tab;le" as select 1|{}
 SELECT a.oid, b.oid FROM pg_class a, pg_class b|{pg_catalog.pg_class}
 SELECT 1 as col1; select 2 as col2|{}
 WARNING:  CDB_QueryTables cannot explain query: select 1 from nonexistant (42P01: relation "nonexistant" does not exist)
-CONTEXT:  PL/pgSQL function cdb_querytables(text) line 3 at RETURN
 ERROR:  relation "nonexistant" does not exist
-CONTEXT:  PL/pgSQL function cdb_querytables(text) line 3 at RETURN
 begin; select * from pg_class; commit;|{pg_catalog.pg_class}
 WARNING:  CDB_QueryTables cannot explain query: select * from test (42P01: relation "test" does not exist)
-CONTEXT:  PL/pgSQL function cdb_querytables(text) line 3 at RETURN
 ERROR:  relation "test" does not exist
-CONTEXT:  PL/pgSQL function cdb_querytables(text) line 3 at RETURN
 WITH a AS (select * from pg_class) select * from a|{pg_catalog.pg_class}
 CREATE SCHEMA
 CREATE TABLE

test/extension/test.sh

@@ -228,6 +228,7 @@ function tear_down() {
     sql 'DROP ROLE cdb_testmember_2;'
 
     tear_down_database
+    DATABASE=postgres sql postgres 'DROP ROLE IF EXISTS publicuser';
 }
 
 
@@ -486,6 +487,18 @@ function test_foreign_tables() {
     ${CMD} -d fdw_target -f scripts-available/CDB_QueryTables.sql
     ${CMD} -d fdw_target -f scripts-available/CDB_TableMetadata.sql
 
+    DATABASE=fdw_target sql postgres "DO
+\$\$
+BEGIN
+   IF NOT EXISTS (
+      SELECT *
+      FROM   pg_catalog.pg_user
+      WHERE  usename = 'publicuser') THEN
+
+      CREATE ROLE publicuser LOGIN;
+   END IF;
+END
+\$\$;"
     DATABASE=fdw_target sql postgres 'CREATE SCHEMA test_fdw;'
     DATABASE=fdw_target sql postgres 'CREATE TABLE test_fdw.foo (a int);'
     DATABASE=fdw_target sql postgres 'INSERT INTO test_fdw.foo (a) values (42);'