Going green: handle special float values for histogram when overviews are involved

This commit is contained in:
Daniel García Aubert
2017-06-14 19:00:37 +02:00
parent 06d40e8b1e
commit dee00e6abd

View File

@@ -12,12 +12,27 @@ var columnTypeQueryTpl = dot.template(
var BIN_MIN_NUMBER = 6;
var BIN_MAX_NUMBER = 48;
var filteredQueryTpl = dot.template([
'filtered_source AS (',
' SELECT *',
' FROM ({{=it._query}}) _cdb_filtered_source',
' WHERE',
' {{=it._column}} IS NOT NULL',
' AND',
' {{=it._column}} != \'infinity\'::float',
' AND',
' {{=it._column}} != \'-infinity\'::float',
' AND',
' {{=it._column}} != \'NaN\'::float',
')'
].join(' \n'));
var basicsQueryTpl = dot.template([
'basics AS (',
' SELECT',
' max({{=it._column}}) AS max_val, min({{=it._column}}) AS min_val,',
' sum({{=it._column}}*_feature_count)/sum(_feature_count) AS avg_val, sum(_feature_count) AS total_rows',
' FROM ({{=it._query}}) _cdb_basics',
' FROM filtered_source',
')'
].join(' \n'));
@@ -26,7 +41,7 @@ var overrideBasicsQueryTpl = dot.template([
' SELECT',
' max({{=it._end}}) AS max_val, min({{=it._start}}) AS min_val,',
' sum({{=it._column}}*_feature_count)/sum(_feature_count) AS avg_val, sum(_feature_count) AS total_rows',
' FROM ({{=it._query}}) _cdb_basics',
' FROM filtered_source',
')'
].join('\n'));
@@ -37,7 +52,7 @@ var iqrQueryTpl = dot.template([
' SELECT quartile, max(_cdb_iqr_column) AS quartile_max from (',
' SELECT {{=it._column}} AS _cdb_iqr_column, ntile(4) over (order by {{=it._column}}',
' ) AS quartile',
' FROM ({{=it._query}}) _cdb_rank) _cdb_quartiles',
' FROM filtered_source) _cdb_quartiles',
' WHERE quartile = 1 or quartile = 3',
' GROUP BY quartile',
' ) _cdb_iqr',
@@ -56,7 +71,7 @@ var binsQueryTpl = dot.template([
' )',
' )',
' END AS bins_number',
' FROM basics, iqrange, ({{=it._query}}) _cdb_bins',
' FROM basics, iqrange, filtered_source',
' LIMIT 1',
')'
].join('\n'));
@@ -76,11 +91,34 @@ var nullsQueryTpl = dot.template([
')'
].join('\n'));
var infinitiesQueryTpl = dot.template([
'infinities AS (',
' SELECT',
' count(*) AS infinities_count',
' FROM ({{=it._query}}) _cdb_histogram_infinities',
' WHERE',
' {{=it._column}} = \'infinity\'::float',
' OR',
' {{=it._column}} = \'-infinity\'::float',
')'
].join('\n'));
var nansQueryTpl = dot.template([
'nans AS (',
' SELECT',
' count(*) AS nans_count',
' FROM ({{=it._query}}) _cdb_histogram_infinities',
' WHERE {{=it._column}} = \'NaN\'::float',
')'
].join('\n'));
var histogramQueryTpl = dot.template([
'SELECT',
' (max_val - min_val) / cast(bins_number as float) AS bin_width,',
' bins_number,',
' nulls_count,',
' infinities_count,',
' nans_count,',
' avg_val,',
' CASE WHEN min_val = max_val',
' THEN 0',
@@ -90,9 +128,8 @@ var histogramQueryTpl = dot.template([
' max({{=it._column}})::numeric AS max,',
' sum({{=it._column}}*_feature_count)/sum(_feature_count)::numeric AS avg,',
' sum(_feature_count) AS freq',
'FROM ({{=it._query}}) _cdb_histogram, basics, nulls, bins',
'WHERE {{=it._column}} IS NOT NULL',
'GROUP BY bin, bins_number, bin_width, nulls_count, avg_val',
'FROM filtered_source, basics, nulls, infinities, nans, bins',
'GROUP BY bin, bins_number, bin_width, nulls_count, infinities_count, nans_count, avg_val',
'ORDER BY bin'
].join('\n'));
@@ -156,7 +193,12 @@ Histogram.prototype.sql = function(psql, override, callback) {
var _query = this.rewrittenQuery(this.query);
var basicsQuery, binsQuery;
var filteredQuery, basicsQuery, binsQuery;
filteredQuery = filteredQueryTpl({
_query: _query,
_column: _column
});
if (override && _.has(override, 'start') && _.has(override, 'end') && _.has(override, 'bins')) {
basicsQuery = overrideBasicsQueryTpl({
@@ -202,11 +244,20 @@ Histogram.prototype.sql = function(psql, override, callback) {
var histogramSql = [
"WITH",
[
filteredQuery,
basicsQuery,
binsQuery,
nullsQueryTpl({
_query: _query,
_column: _column
}),
infinitiesQueryTpl({
_query: _query,
_column: _column
}),
nansQueryTpl({
_query: _query,
_column: _column
})
].join(',\n'),
histogramQueryTpl({