cartodb4/observatory-extension
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

getmeasure bypassing several older functions, areas not yet implemented

Browse Source
This commit is contained in:
John Krauss
2016-07-13 12:20:01 -04:00
parent 5229279ee9
commit bbc6f9ef36
4 changed files with 77 additions and 161 deletions
Download Patch File Download Diff File Expand all files Collapse all files

97
src/pg/sql/40_observatory_utility.sql
View File

@@ -37,81 +37,6 @@ END;
$$ LANGUAGE plpgsql;
-- A function that gets the column data for multiple columns
-- Old: OBS_GetColumnData
CREATE OR REPLACE FUNCTION cdb_observatory._OBS_GetColumnData(
geometry_id text,
column_ids text[],
timespan text
)
RETURNS SETOF JSON
AS $$
BEGIN
-- figure out highest-weight geometry_id/timespan pair for the first data column
-- TODO this should be done for each data column separately
IF geometry_id IS NULL OR timespan IS NULL THEN
EXECUTE '
SELECT data_t.timespan timespan, geom_c.id boundary_id
FROM observatory.obs_table data_t,
observatory.obs_column_table data_ct,
observatory.obs_column data_c,
observatory.obs_column_table geoid_ct,
observatory.obs_column_to_column c2c,
observatory.obs_column geom_c
WHERE data_c.id = $2
AND data_ct.column_id = data_c.id
AND data_ct.table_id = data_t.id
AND geoid_ct.table_id = data_t.id
AND geoid_ct.column_id = c2c.source_id
AND c2c.reltype = ''geom_ref''
AND geom_c.id = c2c.target_id
AND CASE WHEN $3 IS NULL THEN True ELSE $3 = timespan END
AND CASE WHEN $1 IS NULL THEN True ELSE $1 = geom_c.id END
ORDER BY geom_c.weight DESC,
data_t.timespan DESC
LIMIT 1
' INTO timespan, geometry_id
USING geometry_id, (column_ids)[1], timespan;
END IF;
RETURN QUERY
EXECUTE '
WITH geomref AS (
SELECT ct.table_id id
FROM observatory.OBS_column_to_column c2c,
observatory.OBS_column_table ct
WHERE c2c.reltype = ''geom_ref''
AND c2c.target_id = $1
AND c2c.source_id = ct.column_id
),
column_ids as (
select row_number() over () as no, a.column_id as column_id from (select unnest($2) as column_id) a
)
SELECT row_to_json(a) from (
select colname,
tablename,
aggregate,
name,
type,
c.description,
$1 AS boundary_id
FROM column_ids, observatory.OBS_column c, observatory.OBS_column_table ct, observatory.OBS_table t
WHERE column_ids.column_id = c.id
AND c.id = ct.column_id
AND t.id = ct.table_id
AND t.timespan = $3
AND t.id in (SELECT id FROM geomref)
order by column_ids.no
) a
'
USING geometry_id, column_ids, timespan
RETURN;
END;
$$ LANGUAGE plpgsql;
--Test point cause Stuart always seems to make random points in the water
CREATE OR REPLACE FUNCTION cdb_observatory._TestPoint()
RETURNS geometry(Point, 4326)
@@ -158,28 +83,6 @@ BEGIN
END;
$$ LANGUAGE plpgsql;
CREATE OR REPLACE FUNCTION cdb_observatory._OBS_GetRelatedColumn(columns_ids text[], reltype text )
RETURNS TEXT[]
AS $$
DECLARE
result TEXT[];
BEGIN
EXECUTE '
With ids as (
select row_number() over() as no, id from (select unnest($1) as id) t
)
select array_agg(target_id order by no)
FROM ids
LEFT JOIN observatory.obs_column_to_column
on source_id = id
where reltype = $2 or reltype is null
'
INTO result
using columns_ids, reltype;
return result;
END;
$$ LANGUAGE plpgsql;
-- Function that replaces all non digits or letters with _ trims and lowercases the
-- passed measure name

100
src/pg/sql/41_observatory_augmentation.sql
View File

@@ -345,37 +345,91 @@ CREATE OR REPLACE FUNCTION cdb_observatory.OBS_GetMeasure(
RETURNS NUMERIC
AS $$
DECLARE
map_type TEXT;
numer_aggregate TEXT;
numer_colname TEXT;
numer_geomref_colname TEXT;
numer_tablename TEXT;
denom_colname TEXT;
denom_geomref_colname TEXT;
denom_tablename TEXT;
geom_colname TEXT;
geom_geomref_colname TEXT;
geom_tablename TEXT;
result NUMERIC;
measure_ids TEXT[];
denominator_id TEXT;
vals NUMERIC[];
sql TEXT;
BEGIN
IF normalize ILIKE 'area' THEN
measure_ids := ARRAY[measure_id];
ELSIF normalize ILIKE 'denominator' THEN
EXECUTE 'SELECT (cdb_observatory._OBS_GetRelatedColumn(ARRAY[$1], ''denominator''))[1]
' INTO denominator_id
USING measure_id;
measure_ids := ARRAY[measure_id, denominator_id];
ELSIF normalize ILIKE 'none' THEN
-- TODO we need a switch on obs_get to disable area normalization
RAISE EXCEPTION 'No normalization not yet supported.';
EXECUTE
$query$
SELECT numer_aggregate, numer_colname, numer_geomref_colname, numer_tablename,
denom_colname, denom_geomref_colname, denom_tablename,
geom_colname, geom_geomref_colname, geom_tablename
FROM observatory.obs_meta
WHERE (geom_id = $1 OR ($1 = ''))
AND numer_id = $2
AND (numer_timespan = $3 OR ($3 = ''))
ORDER BY geom_weight DESC, numer_timespan DESC
LIMIT 1
$query$
INTO numer_colname, numer_geomref_colname, numer_table,
denom_colname, denom_geomref_colname, denom_table,
geom_colname, geom_geomref_colname, geom_table
USING COALESCE(boundary_id, ''), measure_id, COALESCE(time_span, '');
IF normalize ILIKE 'area' AND numer_aggregate ILIKE 'sum' THEN
-- area normalized
map_type := 'areaNormalized';
ELSIF normalize ILIKE 'denominator' AND numer_aggregate ILIKE 'sum' THEN
-- denominated
map_type := 'denominated';
ELSE
RAISE EXCEPTION 'Only valid inputs for "normalize" are "area" (default) and "denominator".';
-- predenominated
map_type := 'predenominated';
END IF;
EXECUTE '
SELECT ARRAY_AGG(val) FROM (SELECT (cdb_observatory._OBS_Get($1, $2, $3, $4)->>''value'')::NUMERIC val) b
'
INTO vals
USING geom, measure_ids, time_span, boundary_id;
IF normalize ILIKE 'denominator' THEN
RETURN (vals)[1]/(vals)[2];
IF ST_GeometryType(geom) = 'ST_Point' THEN
IF map_type = 'areaNormalied' THEN
sql = format('SELECT numer.%I / (ST_Area(geom.%I::Geography) * 1000000)
FROM observatory.%I numer, observatory.%I geom
WHERE numer.%I = geom.%I
AND ST_WITHIN(%L, geom.%I)',
numer_colname, geom_colname, numer_tablename,
geom_tablename, numer_geomref_colname,
geom_geomref_colname, geom, geom_colname);
ELSIF map_type = 'denominated' THEN
sql = format('SELECT numer.%I / denom.%I
FROM observatory.%I numer, observatory.%I geom, observatory.%I denom
WHERE numer.%I = geom.%I
AND geom.%I = denom.%I
AND ST_WITHIN(%L, geom.%I)',
numer_colname, denom_colname, numer_tablename,
geom_tablename, denom_tablename,
numer_geomref_colname, geom_geomref_colname,
geom_geomref_colname, denom_geomref_colname,
geom, geom_colname);
ELSIF map_type = 'predenominated' THEN
sql = format('SELECT numer.%I
FROM observatory.%I numer, observatory.%I geom
WHERE numer.%I = geom.%I
AND ST_WITHIN(%L, geom.%I)',
numer_colname, numer_tablename,
geom_tablename, numer_geomref_colname,
geom_geomref_colname, geom, geom_colname);
END IF;
ELSIF ST_GeometryType(geom) IN ('ST_Polygon', 'ST_MultiPolygon') THEN
IF map_type = 'areaNormalied' THEN
ELSIF map_type = 'denominated' THEN
ELSIF map_type = 'predenominated' THEN
END IF;
ELSE
RETURN (vals)[1];
RAISE EXCEPTION 'Invalid geometry type (%), can only handle ''ST_Point'', ''ST_Polygon'', and ''ST_MultiPolygon''',
ST_GeometryType(geom);
END IF;
EXECUTE sql INTO result;
RETURN result;
END;
$$ LANGUAGE plpgsql;

9
src/pg/test/expected/40_observatory_utility_test.out
View File

@@ -9,21 +9,12 @@ t
_obs_geomtable_with_null_response
t
(1 row)
test_get_obs_column_with_geoid_and_census_1|test_get_obs_column_with_geoid_and_census_2
t|t
(1 row)
obs_getcolumndata_missing_measure
t
(1 row)
_obs_buildsnapshotquery_test_1
t
(1 row)
_obs_buildsnapshotquery_test_2
t
(1 row)
_obs_getrelatedcolumn_test
t
(1 row)
_obs_standardizemeasurename_test
t
(1 row)

32
src/pg/test/sql/40_observatory_utility_test.sql
View File

@@ -29,29 +29,6 @@ SELECT
-- 'us.census.tiger.census_tract'
-- );
WITH result as (
SELECT
array_agg(a) expected from cdb_observatory._OBS_GetColumnData(
'us.census.tiger.census_tract',
Array['us.census.spielman_singleton_segments.X55', 'us.census.acs.B01003001'],
'2010 - 2014') a
)
select
(expected)[1]::text = '{"colname":"x55","tablename":"obs_65f29658e096ca1485bf683f65fdbc9f05ec3c5d","aggregate":null,"name":"Spielman-Singleton Segments: 55 Clusters","type":"Text","description":"Sociodemographic classes from Spielman and Singleton 2015, 55 clusters","boundary_id":"us.census.tiger.census_tract"}' as test_get_obs_column_with_geoid_and_census_1,
(expected)[2]::text = '{"colname":"total_pop","tablename":"obs_b393b5b88c6adda634b2071a8005b03c551b609a","aggregate":"sum","name":"Total Population","type":"Numeric","description":"The total number of all people living in a given geographic area. This is a very useful catch-all denominator when calculating rates.","boundary_id":"us.census.tiger.census_tract"}' as test_get_obs_column_with_geoid_and_census_2
from result;
-- should be null-valued
WITH result as (
SELECT
array_agg(a) expected from cdb_observatory._OBS_GetColumnData(
'us.census.tiger.census_tract',
Array['us.census.tiger.baloney'],
'2010 - 2014') a
)
select expected is null as OBS_GetColumnData_missing_measure
from result;
-- OBS_BuildSnapshotQuery
-- Should give back: SELECT vals[1] As total_pop, vals[2] As male_pop, vals[3] As female_pop, vals[4] As median_age
SELECT
@@ -65,15 +42,6 @@ SELECT
Array['mandarin_orange']
) = 'SELECT vals[1] As mandarin_orange' As _OBS_BuildSnapshotQuery_test_2;
SELECT cdb_observatory._OBS_GetRelatedColumn(
Array[
'es.ine.t3_1',
'us.census.acs.B01003001',
'us.census.acs.B01001002'
],
'denominator'
) = '{es.ine.t1_1,NULL,us.census.acs.B01003001}' As _OBS_GetRelatedColumn_test;
-- should give back a standardized measure name
SELECT cdb_observatory._OBS_StandardizeMeasureName('test 343 %% 2 qqq }}{{}}') = 'test_343_2_qqq' As _OBS_StandardizeMeasureName_test;