Merge branch 'master' into 691-date-histogram

This commit is contained in:
Daniel García Aubert
2017-08-01 16:07:27 +02:00
26 changed files with 1357 additions and 251 deletions
+4
View File
@@ -193,6 +193,10 @@ BaseController.prototype.sendError = function(req, res, err, label) {
var statusCode = findStatusCode(err);
if (err.message === 'Tile does not exist' && req.params.format === 'mvt') {
statusCode = 204;
}
debug('[%s ERROR] -- %d: %s, %s', label, statusCode, err, err.stack);
// If a callback was requested, force status to 200
+105 -35
View File
@@ -5,11 +5,32 @@ var debug = require('debug')('windshaft:widget:aggregation');
var dot = require('dot');
dot.templateSettings.strip = false;
var filteredQueryTpl = dot.template([
'filtered_source AS (',
' SELECT *',
' FROM ({{=it._query}}) _cdb_filtered_source',
' {{?it._aggregationColumn && it._isFloatColumn}}WHERE',
' {{=it._aggregationColumn}} != \'infinity\'::float',
' AND',
' {{=it._aggregationColumn}} != \'-infinity\'::float',
' AND',
' {{=it._aggregationColumn}} != \'NaN\'::float{{?}}',
')'
].join(' \n'));
var summaryQueryTpl = dot.template([
'summary AS (',
' SELECT',
' count(1) AS count,',
' sum(CASE WHEN {{=it._column}} IS NULL THEN 1 ELSE 0 END) AS nulls_count',
' {{?it._isFloatColumn}},sum(',
' CASE',
' WHEN {{=it._aggregationColumn}} = \'infinity\'::float OR {{=it._aggregationColumn}} = \'-infinity\'::float',
' THEN 1',
' ELSE 0',
' END',
' ) AS infinities_count,',
' sum(CASE WHEN {{=it._aggregationColumn}} = \'NaN\'::float THEN 1 ELSE 0 END) AS nans_count{{?}}',
' FROM ({{=it._query}}) _cdb_aggregation_nulls',
')'
].join('\n'));
@@ -18,7 +39,7 @@ var rankedCategoriesQueryTpl = dot.template([
'categories AS(',
' SELECT {{=it._column}} AS category, {{=it._aggregation}} AS value,',
' row_number() OVER (ORDER BY {{=it._aggregation}} desc) as rank',
' FROM ({{=it._query}}) _cdb_aggregation_all',
' FROM filtered_source',
' {{?it._aggregationColumn!==null}}WHERE {{=it._aggregationColumn}} IS NOT NULL{{?}}',
' GROUP BY {{=it._column}}',
' ORDER BY 2 DESC',
@@ -44,22 +65,25 @@ var categoriesSummaryCountQueryTpl = dot.template([
].join('\n'));
var rankedAggregationQueryTpl = dot.template([
'SELECT CAST(category AS text), value, false as agg, nulls_count, min_val, max_val, count, categories_count',
'SELECT CAST(category AS text), value, false as agg, nulls_count, min_val, max_val,',
' count, categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
' FROM categories, summary, categories_summary_min_max, categories_summary_count',
' WHERE rank < {{=it._limit}}',
'UNION ALL',
'SELECT \'Other\' category, {{=it._aggregationFn}}(value) as value, true as agg, nulls_count, min_val, max_val,',
' count, categories_count',
'SELECT \'Other\' category, {{=it._aggregationFn}}(value) as value, true as agg, nulls_count,',
' min_val, max_val, count, categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
' FROM categories, summary, categories_summary_min_max, categories_summary_count',
' WHERE rank >= {{=it._limit}}',
'GROUP BY nulls_count, min_val, max_val, count, categories_count'
'GROUP BY nulls_count, min_val, max_val, count,',
' categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}'
].join('\n'));
var aggregationQueryTpl = dot.template([
'SELECT CAST({{=it._column}} AS text) AS category, {{=it._aggregation}} AS value, false as agg,',
' nulls_count, min_val, max_val, count, categories_count',
' nulls_count, min_val, max_val, count, categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
'FROM ({{=it._query}}) _cdb_aggregation_all, summary, categories_summary_min_max, categories_summary_count',
'GROUP BY category, nulls_count, min_val, max_val, count, categories_count',
'GROUP BY category, nulls_count, min_val, max_val, count,',
' categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
'ORDER BY value DESC'
].join('\n'));
@@ -84,7 +108,7 @@ var TYPE = 'aggregation';
}
}
*/
function Aggregation(query, options) {
function Aggregation(query, options, queries) {
if (!_.isString(options.column)) {
throw new Error('Aggregation expects `column` in widget options');
}
@@ -108,9 +132,11 @@ function Aggregation(query, options) {
BaseWidget.apply(this);
this.query = query;
this.queries = queries;
this.column = options.column;
this.aggregation = options.aggregation;
this.aggregationColumn = options.aggregationColumn;
this._isFloatColumn = null;
}
Aggregation.prototype = new BaseWidget();
@@ -119,19 +145,39 @@ Aggregation.prototype.constructor = Aggregation;
module.exports = Aggregation;
Aggregation.prototype.sql = function(psql, override, callback) {
var self = this;
if (!callback) {
callback = override;
override = {};
}
if (this.aggregationColumn && this._isFloatColumn === null) {
this._isFloatColumn = false;
this.getColumnType(psql, this.aggregationColumn, this.queries.no_filters, function (err, type) {
if (!err && !!type) {
self._isFloatColumn = type.float;
}
self.sql(psql, override, callback);
});
return null;
}
var _query = this.query;
var aggregationSql;
if (!!override.ownFilter) {
aggregationSql = [
this.getCategoriesCTESql(_query, this.column, this.aggregation, this.aggregationColumn),
this.getCategoriesCTESql(
_query,
this.column,
this.aggregation,
this.aggregationColumn,
this._isFloatColumn
),
aggregationQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column,
_aggregation: this.getAggregationSql(),
@@ -140,8 +186,15 @@ Aggregation.prototype.sql = function(psql, override, callback) {
].join('\n');
} else {
aggregationSql = [
this.getCategoriesCTESql(_query, this.column, this.aggregation, this.aggregationColumn),
this.getCategoriesCTESql(
_query,
this.column,
this.aggregation,
this.aggregationColumn,
this._isFloatColumn
),
rankedAggregationQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column,
_aggregationFn: this.aggregation !== 'count' ? this.aggregation : 'sum',
@@ -155,30 +208,38 @@ Aggregation.prototype.sql = function(psql, override, callback) {
return callback(null, aggregationSql);
};
Aggregation.prototype.getCategoriesCTESql = function(query, column, aggregation, aggregationColumn) {
Aggregation.prototype.getCategoriesCTESql = function(query, column, aggregation, aggregationColumn, isFloatColumn) {
return [
"WITH",
[
summaryQueryTpl({
_query: query,
_column: column
}),
rankedCategoriesQueryTpl({
_query: query,
_column: column,
_aggregation: this.getAggregationSql(),
_aggregationColumn: aggregation !== 'count' ? aggregationColumn : null
}),
categoriesSummaryMinMaxQueryTpl({
_query: query,
_column: column
}),
categoriesSummaryCountQueryTpl({
_query: query,
_column: column
})
].join(',\n')
].join('\n');
"WITH",
[
filteredQueryTpl({
_isFloatColumn: isFloatColumn,
_query: this.query,
_column: this.column,
_aggregationColumn: aggregation !== 'count' ? aggregationColumn : null
}),
summaryQueryTpl({
_isFloatColumn: isFloatColumn,
_query: query,
_column: column,
_aggregationColumn: aggregation !== 'count' ? aggregationColumn : null
}),
rankedCategoriesQueryTpl({
_query: query,
_column: column,
_aggregation: this.getAggregationSql(),
_aggregationColumn: aggregation !== 'count' ? aggregationColumn : null
}),
categoriesSummaryMinMaxQueryTpl({
_query: query,
_column: column
}),
categoriesSummaryCountQueryTpl({
_query: query,
_column: column
})
].join(',\n')
].join('\n');
};
var aggregationFnQueryTpl = dot.template('{{=it._aggregationFn}}({{=it._aggregationColumn}})');
@@ -193,6 +254,8 @@ Aggregation.prototype.format = function(result) {
var categories = [];
var count = 0;
var nulls = 0;
var nans = 0;
var infinities = 0;
var minValue = 0;
var maxValue = 0;
var categoriesCount = 0;
@@ -202,12 +265,15 @@ Aggregation.prototype.format = function(result) {
var firstRow = result.rows[0];
count = firstRow.count;
nulls = firstRow.nulls_count;
nans = firstRow.nans_count;
infinities = firstRow.infinities_count;
minValue = firstRow.min_val;
maxValue = firstRow.max_val;
categoriesCount = firstRow.categories_count;
result.rows.forEach(function(row) {
categories.push(_.omit(row, 'count', 'nulls_count', 'min_val', 'max_val', 'categories_count'));
categories.push(_.omit(row, 'count', 'nulls_count', 'min_val',
'max_val', 'categories_count', 'nans_count', 'infinities_count'));
});
}
@@ -215,6 +281,8 @@ Aggregation.prototype.format = function(result) {
aggregation: this.aggregation,
count: count,
nulls: nulls,
nans: nans,
infinities: infinities,
min: minValue,
max: maxValue,
categoriesCount: categoriesCount,
@@ -253,6 +321,8 @@ Aggregation.prototype.search = function(psql, userQuery, callback) {
var self = this;
var _userQuery = psql.escapeLiteral('%' + userQuery + '%');
var _value = this.aggregation !== 'count' && this.aggregationColumn ?
this.aggregation + '(' + this.aggregationColumn + ')' : 'count(1)';
// TODO unfiltered will be wrong as filters are already applied at this point
var query = searchQueryTpl({
@@ -265,7 +335,7 @@ Aggregation.prototype.search = function(psql, userQuery, callback) {
_searchFiltered: filterCategoriesQueryTpl({
_query: this.query,
_column: this.column,
_value: 'count(1)',
_value: _value,
_userQuery: _userQuery
})
});
+42
View File
@@ -1,3 +1,6 @@
var dot = require('dot');
dot.templateSettings.strip = false;
function BaseDataview() {}
module.exports = BaseDataview;
@@ -24,3 +27,42 @@ BaseDataview.prototype.getResult = function(psql, override, callback) {
BaseDataview.prototype.search = function(psql, userQuery, callback) {
return callback(null, this.format({ rows: [] }));
};
var FLOAT_OIDS = {
700: true,
701: true,
1700: true
};
var DATE_OIDS = {
1082: true,
1114: true,
1184: true
};
var columnTypeQueryTpl = dot.template(
'SELECT pg_typeof({{=it.column}})::oid FROM ({{=it.query}}) _cdb_column_type limit 1'
);
BaseDataview.prototype.getColumnType = function (psql, column, query, callback) {
var readOnlyTransaction = true;
var columnTypeQuery = columnTypeQueryTpl({
column: column, query: query
});
psql.query(columnTypeQuery, function(err, result) {
if (err) {
return callback(err);
}
var pgType = result.rows[0].pg_typeof;
callback(null, getPGTypeName(pgType));
}, readOnlyTransaction);
};
function getPGTypeName (pgType) {
return {
float: FLOAT_OIDS.hasOwnProperty(pgType),
date: DATE_OIDS.hasOwnProperty(pgType)
};
}
+36 -7
View File
@@ -7,9 +7,19 @@ dot.templateSettings.strip = false;
var formulaQueryTpl = dot.template([
'SELECT',
'{{=it._operation}}({{=it._column}}) AS result,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
'FROM ({{=it._query}}) _cdb_formula'
' {{=it._operation}}({{=it._column}}) AS result,',
' (SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
' {{?it._isFloatColumn}},(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls',
' WHERE {{=it._column}} = \'infinity\'::float OR {{=it._column}} = \'-infinity\'::float) AS infinities_count',
' ,(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls',
' WHERE {{=it._column}} = \'NaN\'::float) AS nans_count{{?}}',
'FROM ({{=it._query}}) _cdb_formula',
'{{?it._isFloatColumn && it._operation !== \'count\'}}WHERE',
' {{=it._column}} != \'infinity\'::float',
'AND',
' {{=it._column}} != \'-infinity\'::float',
'AND',
' {{=it._column}} != \'NaN\'::float{{?}}'
].join('\n'));
var VALID_OPERATIONS = {
@@ -31,7 +41,7 @@ var TYPE = 'formula';
}
}
*/
function Formula(query, options) {
function Formula(query, options, queries) {
if (!_.isString(options.operation)) {
throw new Error('Formula expects `operation` in widget options');
}
@@ -47,8 +57,10 @@ function Formula(query, options) {
BaseWidget.apply(this);
this.query = query;
this.queries = queries;
this.column = options.column || '1';
this.operation = options.operation;
this._isFloatColumn = null;
}
Formula.prototype = new BaseWidget();
@@ -57,14 +69,27 @@ Formula.prototype.constructor = Formula;
module.exports = Formula;
Formula.prototype.sql = function(psql, override, callback) {
var self = this;
if (!callback) {
callback = override;
override = {};
}
var _query = this.query;
if (this._isFloatColumn === null) {
this._isFloatColumn = false;
this.getColumnType(psql, this.column, this.queries.no_filters, function (err, type) {
if (!err && !!type) {
self._isFloatColumn = type.float;
}
self.sql(psql, override, callback);
});
return null;
}
var formulaSql = formulaQueryTpl({
_query: _query,
_isFloatColumn: this._isFloatColumn,
_query: this.query,
_operation: this.operation,
_column: this.column
});
@@ -78,13 +103,17 @@ Formula.prototype.format = function(result) {
var formattedResult = {
operation: this.operation,
result: 0,
nulls: 0
nulls: 0,
nans: 0,
infinities: 0
};
if (result.rows.length) {
formattedResult.operation = this.operation;
formattedResult.result = result.rows[0].result;
formattedResult.nulls = result.rows[0].nulls_count;
formattedResult.nans = result.rows[0].nans_count;
formattedResult.infinities = result.rows[0].infinities_count;
}
return formattedResult;
+124 -52
View File
@@ -5,11 +5,6 @@ var debug = require('debug')('windshaft:dataview:histogram');
var dot = require('dot');
dot.templateSettings.strip = false;
var columnTypeQueryTpl = dot.template(
'SELECT pg_typeof({{=it.column}})::oid FROM ({{=it.query}}) _cdb_histogram_column_type limit 1'
);
var columnCastTpl = dot.template("date_part('epoch', {{=it.column}})");
var dateIntervalQueryTpl = dot.template([
'WITH',
'dates AS (',
@@ -54,12 +49,27 @@ var MAX_INTERVAL_VALUE = 366;
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',
' {{?it._isFloatColumn}}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,',
' avg({{=it._column}}) AS avg_val, count(1) AS total_rows',
' FROM ({{=it._query}}) _cdb_basics',
' FROM filtered_source',
')'
].join(' \n'));
@@ -68,7 +78,7 @@ var overrideBasicsQueryTpl = dot.template([
' SELECT',
' max({{=it._end}}) AS max_val, min({{=it._start}}) AS min_val,',
' avg({{=it._column}}) AS avg_val, count(1) AS total_rows',
' FROM ({{=it._query}}) _cdb_basics',
' FROM filtered_source',
')'
].join('\n'));
@@ -79,7 +89,7 @@ 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',
@@ -98,7 +108,7 @@ var binsQueryTpl = dot.template([
' )',
' )',
' END AS bins_number',
' FROM basics, iqrange, ({{=it._query}}) _cdb_bins',
' FROM basics, iqrange, filtered_source',
' LIMIT 1',
')'
].join('\n'));
@@ -118,11 +128,34 @@ 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,',
' {{?it._isFloatColumn}}infinities_count,',
' nans_count,{{?}}',
' avg_val,',
' CASE WHEN min_val = max_val',
' THEN 0',
@@ -132,9 +165,9 @@ var histogramQueryTpl = dot.template([
' max({{=it._column}})::numeric AS max,',
' avg({{=it._column}})::numeric AS avg,',
' 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, bins{{?it._isFloatColumn}}, infinities, nans{{?}}',
'GROUP BY bin, bins_number, bin_width, nulls_count,',
' avg_val{{?it._isFloatColumn}}, infinities_count, nans_count{{?}}',
'ORDER BY bin'
].join('\n'));
@@ -268,54 +301,47 @@ Histogram.prototype.constructor = Histogram;
module.exports = Histogram;
var DATE_OIDS = {
1082: true,
1114: true,
1184: true
};
Histogram.prototype.sql = function(psql, override, callback) {
// jshint maxcomplexity: 7
var self = this;
if (!callback) {
callback = override;
override = {};
}
var self = this;
var _column = this.column;
var columnTypeQuery = columnTypeQueryTpl({
column: _column, query: this.queries.no_filters
});
if (this._columnType === null) {
psql.query(columnTypeQuery, function(err, result) {
this.getColumnType(psql, this.column, this.queries.no_filters, function (err, type) {
// assume numeric, will fail later
self._columnType = 'numeric';
if (!err && !!result.rows[0]) {
var pgType = result.rows[0].pg_typeof;
if (DATE_OIDS.hasOwnProperty(pgType)) {
self._columnType = 'date';
}
if (!err && !!type) {
self._columnType = Object.keys(type).find(function (key) {
return type[key];
});
}
self.sql(psql, override, callback);
}, true); // use read-only transaction
return null;
}
this._buildQuery(psql, override, callback);
};
Histogram.prototype._buildQuery = function (psql, override, callback) {
var filteredQuery, basicsQuery, binsQuery;
var _column = this.column;
var _query = this.query;
if (this._columnType === 'date') {
return this._buildDateHistogramQuery(psql, override, callback);
}
if (this._columnType === 'date') {
_column = columnCastTpl({ column: this.column});
}
var _query = this.query;
filteredQuery = filteredQueryTpl({
_isFloatColumn: this._columnType === 'float',
_query: _query,
_column: _column
});
var basicsQuery, binsQuery;
if (override && _.has(override, 'start') && _.has(override, 'end') && _.has(override, 'bins')) {
if (this._shouldOverride(override)) {
debug('overriding with %j', override);
basicsQuery = overrideBasicsQueryTpl({
_query: _query,
@@ -335,7 +361,7 @@ Histogram.prototype.sql = function(psql, override, callback) {
_column: _column
});
if (override && _.has(override, 'bins')) {
if (this._shouldOverrideBins(override)) {
binsQuery = [
overrideBinsQueryTpl({
_bins: override.bins
@@ -356,17 +382,34 @@ Histogram.prototype.sql = function(psql, override, callback) {
}
}
var histogramSql = [
"WITH",
[
basicsQuery,
binsQuery,
nullsQueryTpl({
var cteSql = [
filteredQuery,
basicsQuery,
binsQuery,
nullsQueryTpl({
_query: _query,
_column: _column
})
];
if (this._columnType === 'float') {
cteSql.push(
infinitiesQueryTpl({
_query: _query,
_column: _column
}),
nansQueryTpl({
_query: _query,
_column: _column
})
].join(',\n'),
);
}
var histogramSql = [
"WITH",
cteSql.join(',\n'),
histogramQueryTpl({
_isFloatColumn: this._columnType === 'float',
_query: _query,
_column: _column
})
@@ -377,6 +420,14 @@ Histogram.prototype.sql = function(psql, override, callback) {
return callback(null, histogramSql);
};
Histogram.prototype._shouldOverride = function (override) {
return override && _.has(override, 'start') && _.has(override, 'end') && _.has(override, 'bins');
};
Histogram.prototype._shouldOverrideBins = function (override) {
return override && _.has(override, 'bins');
};
Histogram.prototype._buildDateHistogramQuery = function (psql, override, callback) {
var _column = this.column;
var _query = this.query;
@@ -492,14 +543,16 @@ Histogram.prototype.format = function(result, override) {
override = override || {};
var buckets = [];
var aggregation = getAggregation(override, this.aggregation);
var offset = getOffset(override, this.offset);
var binsCount = getBinsCount(override);
var width = getWidth(override);
var binsStart = getBinStart(override);
var nulls = 0;
var infinities = 0;
var nans = 0;
var avg;
var timestampStart;
var aggregation;
var offset;
if (result.rows.length) {
var firstRow = result.rows[0];
@@ -507,10 +560,27 @@ Histogram.prototype.format = function(result, override) {
width = firstRow.bin_width || width;
avg = firstRow.avg_val;
nulls = firstRow.nulls_count;
binsStart = populateBinStart(override, firstRow);
timestampStart = firstRow.timestamp_start;
infinities = firstRow.infinities_count;
nans = firstRow.nans_count;
binsStart = populateBinStart(override, firstRow);
if (timestampStart) {
aggregation = getAggregation(override, this.aggregation);
offset = getOffset(override, this.offset);
}
buckets = result.rows.map(function(row) {
return _.omit(row, 'bins_number', 'bin_width', 'nulls_count', 'avg_val', 'timestamp_start');
return _.omit(
row,
'bins_number',
'bin_width',
'nulls_count',
'infinities_count',
'nans_count',
'avg_val',
'timestamp_start'
);
});
}
@@ -522,6 +592,8 @@ Histogram.prototype.format = function(result, override) {
bins_count: binsCount,
bins_start: binsStart,
nulls: nulls,
infinities: infinities,
nans: nans,
avg: avg,
bins: buckets
};
@@ -1,14 +1,36 @@
var BaseOverviewsDataview = require('./base');
var BaseDataview = require('../aggregation');
var debug = require('debug')('windshaft:widget:aggregation:overview');
var dot = require('dot');
dot.templateSettings.strip = false;
var filteredQueryTpl = dot.template([
'filtered_source AS (',
' SELECT *',
' FROM ({{=it._query}}) _cdb_filtered_source',
' {{?it._aggregationColumn && it._isFloatColumn}}WHERE',
' {{=it._aggregationColumn}} != \'infinity\'::float',
' AND',
' {{=it._aggregationColumn}} != \'-infinity\'::float',
' AND',
' {{=it._aggregationColumn}} != \'NaN\'::float{{?}}',
')'
].join(' \n'));
var summaryQueryTpl = dot.template([
'summary AS (',
' SELECT',
' sum(_feature_count) AS count,',
' sum(CASE WHEN {{=it._column}} IS NULL THEN 1 ELSE 0 END) AS nulls_count',
' {{?it._isFloatColumn}},sum(',
' CASE',
' WHEN {{=it._aggregationColumn}} = \'infinity\'::float OR {{=it._aggregationColumn}} = \'-infinity\'::float',
' THEN 1',
' ELSE 0',
' END',
' ) AS infinities_count,',
' sum(CASE WHEN {{=it._aggregationColumn}} = \'NaN\'::float THEN 1 ELSE 0 END) AS nans_count{{?}}',
' FROM ({{=it._query}}) _cdb_aggregation_nulls',
')'
].join('\n'));
@@ -17,7 +39,7 @@ var rankedCategoriesQueryTpl = dot.template([
'categories AS(',
' SELECT {{=it._column}} AS category, {{=it._aggregation}} AS value,',
' row_number() OVER (ORDER BY {{=it._aggregation}} desc) as rank',
' FROM ({{=it._query}}) _cdb_aggregation_all',
' FROM filtered_source',
' {{?it._aggregationColumn!==null}}WHERE {{=it._aggregationColumn}} IS NOT NULL{{?}}',
' GROUP BY {{=it._column}}',
' ORDER BY 2 DESC',
@@ -36,40 +58,46 @@ var categoriesSummaryCountQueryTpl = dot.template([
' SELECT count(1) AS categories_count',
' FROM (',
' SELECT {{=it._column}} AS category',
' FROM ({{=it._query}}) _cdb_categories',
' FROM filtered_source',
' GROUP BY {{=it._column}}',
' ) _cdb_categories_count',
')'
].join('\n'));
var rankedAggregationQueryTpl = dot.template([
'SELECT CAST(category AS text), value, false as agg, nulls_count, min_val, max_val, count, categories_count',
'SELECT CAST(category AS text), value, false as agg, nulls_count, min_val, max_val,',
' count, categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
' FROM categories, summary, categories_summary_min_max, categories_summary_count',
' WHERE rank < {{=it._limit}}',
'UNION ALL',
'SELECT \'Other\' category, sum(value), true as agg, nulls_count, min_val, max_val, count, categories_count',
'SELECT \'Other\' category, sum(value), true as agg, nulls_count, min_val, max_val,',
' count, categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
' FROM categories, summary, categories_summary_min_max, categories_summary_count',
' WHERE rank >= {{=it._limit}}',
'GROUP BY nulls_count, min_val, max_val, count, categories_count'
'GROUP BY nulls_count, min_val, max_val, count,',
' categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}'
].join('\n'));
var aggregationQueryTpl = dot.template([
'SELECT CAST({{=it._column}} AS text) AS category, {{=it._aggregation}} AS value, false as agg,',
' nulls_count, min_val, max_val, count, categories_count',
'FROM ({{=it._query}}) _cdb_aggregation_all, summary, categories_summary_min_max, categories_summary_count',
'GROUP BY category, nulls_count, min_val, max_val, count, categories_count',
' nulls_count, min_val, max_val, count, categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
'FROM filtered_source, summary, categories_summary_min_max, categories_summary_count',
'GROUP BY category, nulls_count, min_val, max_val, count,',
' categories_count{{?it._isFloatColumn}}, nans_count, infinities_count{{?}}',
'ORDER BY value DESC'
].join('\n'));
var CATEGORIES_LIMIT = 6;
function Aggregation(query, options, queryRewriter, queryRewriteData, params) {
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params);
function Aggregation(query, options, queryRewriter, queryRewriteData, params, queries) {
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params, queries);
this.query = query;
this.queries = queries;
this.column = options.column;
this.aggregation = options.aggregation;
this.aggregationColumn = options.aggregationColumn;
this._isFloatColumn = null;
}
Aggregation.prototype = Object.create(BaseOverviewsDataview.prototype);
@@ -78,27 +106,49 @@ Aggregation.prototype.constructor = Aggregation;
module.exports = Aggregation;
Aggregation.prototype.sql = function(psql, override, callback) {
var self = this;
if (!callback) {
callback = override;
override = {};
}
var _query = this.rewrittenQuery(this.query);
var _aggregationColumn = this.aggregation !== 'count' ? this.aggregationColumn : null;
if (this.aggregationColumn && this._isFloatColumn === null) {
this._isFloatColumn = false;
this.getColumnType(psql, this.aggregationColumn, this.queries.no_filters, function (err, type) {
if (!err && !!type) {
self._isFloatColumn = type.float;
}
self.sql(psql, override, callback);
});
return null;
}
var aggregationSql;
if (!!override.ownFilter) {
aggregationSql = [
"WITH",
[
summaryQueryTpl({
filteredQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column
_column: this.column,
_aggregationColumn: _aggregationColumn
}),
summaryQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column,
_aggregationColumn: _aggregationColumn
}),
rankedCategoriesQueryTpl({
_query: _query,
_column: this.column,
_aggregation: this.getAggregationSql(),
_aggregationColumn: this.aggregation !== 'count' ? this.aggregationColumn : null
_aggregationColumn: _aggregationColumn
}),
categoriesSummaryMinMaxQueryTpl({
_query: _query,
@@ -110,6 +160,7 @@ Aggregation.prototype.sql = function(psql, override, callback) {
})
].join(',\n'),
aggregationQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column,
_aggregation: this.getAggregationSql(),
@@ -120,15 +171,23 @@ Aggregation.prototype.sql = function(psql, override, callback) {
aggregationSql = [
"WITH",
[
summaryQueryTpl({
filteredQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column
_column: this.column,
_aggregationColumn: _aggregationColumn
}),
summaryQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column,
_aggregationColumn: _aggregationColumn
}),
rankedCategoriesQueryTpl({
_query: _query,
_column: this.column,
_aggregation: this.getAggregationSql(),
_aggregationColumn: this.aggregation !== 'count' ? this.aggregationColumn : null
_aggregationColumn: _aggregationColumn
}),
categoriesSummaryMinMaxQueryTpl({
_query: _query,
@@ -140,6 +199,7 @@ Aggregation.prototype.sql = function(psql, override, callback) {
})
].join(',\n'),
rankedAggregationQueryTpl({
_isFloatColumn: this._isFloatColumn,
_query: _query,
_column: this.column,
_limit: CATEGORIES_LIMIT
@@ -147,6 +207,8 @@ Aggregation.prototype.sql = function(psql, override, callback) {
].join('\n');
}
debug(aggregationSql);
return callback(null, aggregationSql);
};
@@ -1,14 +1,15 @@
var _ = require('underscore');
var BaseDataview = require('../base');
function BaseOverviewsDataview(query, queryOptions, BaseDataview, queryRewriter, queryRewriteData, options) {
function BaseOverviewsDataview(query, queryOptions, BaseDataview, queryRewriter, queryRewriteData, options, queries) {
this.BaseDataview = BaseDataview;
this.query = query;
this.queryOptions = queryOptions;
this.queryRewriter = queryRewriter;
this.queryRewriteData = queryRewriteData;
this.options = options;
this.baseDataview = new this.BaseDataview(this.query, this.queryOptions);
this.queries = queries;
this.baseDataview = new this.BaseDataview(this.query, this.queryOptions, this.queries);
}
module.exports = BaseOverviewsDataview;
@@ -1,34 +1,61 @@
var BaseOverviewsDataview = require('./base');
var BaseDataview = require('../formula');
var debug = require('debug')('windshaft:widget:formula:overview');
var dot = require('dot');
dot.templateSettings.strip = false;
var formulaQueryTpls = {
'count': dot.template([
'SELECT',
'sum(_feature_count) AS result,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
'FROM ({{=it._query}}) _cdb_formula'
].join('\n')),
'sum': dot.template([
'SELECT',
'sum({{=it._column}}*_feature_count) AS result,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
'FROM ({{=it._query}}) _cdb_formula'
].join('\n')),
'avg': dot.template([
'SELECT',
'sum({{=it._column}}*_feature_count)/sum(_feature_count) AS result,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
'FROM ({{=it._query}}) _cdb_formula'
].join('\n')),
'count': dot.template([
'SELECT',
'sum(_feature_count) AS result,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
'{{?it._isFloatColumn}},(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_infinities',
' WHERE {{=it._column}} = \'infinity\'::float OR {{=it._column}} = \'-infinity\'::float) AS infinities_count,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nans',
' WHERE {{=it._column}} = \'NaN\'::float) AS nans_count{{?}}',
'FROM ({{=it._query}}) _cdb_formula'
].join('\n')),
'sum': dot.template([
'SELECT',
'sum({{=it._column}}*_feature_count) AS result,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
'{{?it._isFloatColumn}},(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_infinities',
' WHERE {{=it._column}} = \'infinity\'::float OR {{=it._column}} = \'-infinity\'::float) AS infinities_count',
',(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nans',
' WHERE {{=it._column}} = \'NaN\'::float) AS nans_count{{?}}',
'FROM ({{=it._query}}) _cdb_formula',
'{{?it._isFloatColumn}}WHERE',
' {{=it._column}} != \'infinity\'::float',
'AND',
' {{=it._column}} != \'-infinity\'::float',
'AND',
' {{=it._column}} != \'NaN\'::float{{?}}'
].join('\n')),
'avg': dot.template([
'SELECT',
'sum({{=it._column}}*_feature_count)/sum(_feature_count) AS result,',
'(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nulls WHERE {{=it._column}} IS NULL) AS nulls_count',
'{{?it._isFloatColumn}},(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_infinities',
' WHERE {{=it._column}} = \'infinity\'::float OR {{=it._column}} = \'-infinity\'::float) AS infinities_count',
',(SELECT count(1) FROM ({{=it._query}}) _cdb_formula_nans',
' WHERE {{=it._column}} = \'NaN\'::float) AS nans_count{{?}}',
'FROM ({{=it._query}}) _cdb_formula',
'{{?it._isFloatColumn}}WHERE',
' {{=it._column}} != \'infinity\'::float',
'AND',
' {{=it._column}} != \'-infinity\'::float',
'AND',
' {{=it._column}} != \'NaN\'::float{{?}}'
].join('\n')),
};
function Formula(query, options, queryRewriter, queryRewriteData, params) {
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params);
function Formula(query, options, queryRewriter, queryRewriteData, params, queries) {
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params, queries);
this.column = options.column || '1';
this.operation = options.operation;
this._isFloatColumn = null;
this.queries = queries;
}
Formula.prototype = Object.create(BaseOverviewsDataview.prototype);
@@ -36,21 +63,38 @@ Formula.prototype.constructor = Formula;
module.exports = Formula;
Formula.prototype.sql = function(psql, override, callback) {
Formula.prototype.sql = function (psql, override, callback) {
var self = this;
var formulaQueryTpl = formulaQueryTpls[this.operation];
if ( formulaQueryTpl ) {
if (formulaQueryTpl) {
// supported formula for use with overviews
if (this._isFloatColumn === null) {
this._isFloatColumn = false;
this.getColumnType(psql, this.column, this.queries.no_filters, function (err, type) {
if (!err && !!type) {
self._isFloatColumn = type.float;
}
self.sql(psql, override, callback);
});
return null;
}
var formulaSql = formulaQueryTpl({
_query: this.rewrittenQuery(this.query),
_isFloatColumn: this._isFloatColumn,
_query: this.rewrittenQuery(this.query),
_operation: this.operation,
_column: this.column
_column: this.column
});
callback = callback || override;
debug(formulaSql);
return callback(null, formulaSql);
}
// default behaviour
return this.defaultSql(psql, override, callback);
};
@@ -1,23 +1,35 @@
var _ = require('underscore');
var BaseOverviewsDataview = require('./base');
var BaseDataview = require('../histogram');
var debug = require('debug')('windshaft:dataview:histogram:overview');
var dot = require('dot');
dot.templateSettings.strip = false;
var columnTypeQueryTpl = dot.template(
'SELECT pg_typeof({{=it.column}})::oid FROM ({{=it.query}}) _cdb_histogram_column_type limit 1'
);
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',
' {{?it._isFloatColumn}}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 +38,7 @@ 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 +49,7 @@ 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 +68,7 @@ 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 +88,34 @@ 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,',
' {{?it._isFloatColumn}}infinities_count,',
' nans_count,{{?}}',
' avg_val,',
' CASE WHEN min_val = max_val',
' THEN 0',
@@ -90,14 +125,14 @@ 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',
'FROM filtered_source, basics, nulls, bins{{?it._isFloatColumn}},infinities, nans{{?}}',
'GROUP BY bin, bins_number, bin_width, nulls_count, avg_val',
' {{?it._isFloatColumn}}, infinities_count, nans_count{{?}}',
'ORDER BY bin'
].join('\n'));
function Histogram(query, options, queryRewriter, queryRewriteData, params, queries) {
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params);
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params, queries);
this.query = query;
this.queries = queries;
@@ -112,36 +147,23 @@ Histogram.prototype.constructor = Histogram;
module.exports = Histogram;
var DATE_OIDS = {
1082: true,
1114: true,
1184: true
};
Histogram.prototype.sql = function(psql, override, callback) {
var self = this;
if (!callback) {
callback = override;
override = {};
}
var self = this;
var _column = this.column;
var columnTypeQuery = columnTypeQueryTpl({
column: _column, query: this.rewrittenQuery(this.queries.no_filters)
});
if (this._columnType === null) {
psql.query(columnTypeQuery, function(err, result) {
this.getColumnType(psql, this.column, this.queries.no_filters, function (err, type) {
// assume numeric, will fail later
self._columnType = 'numeric';
if (!err && !!result.rows[0]) {
var pgType = result.rows[0].pg_typeof;
if (DATE_OIDS.hasOwnProperty(pgType)) {
self._columnType = 'date';
}
if (!err && !!type) {
self._columnType = Object.keys(type).find(function (key) {
return type[key];
});
}
self.sql(psql, override, callback);
}, true); // use read-only transaction
@@ -154,11 +176,24 @@ Histogram.prototype.sql = function(psql, override, callback) {
return this.defaultSql(psql, override, callback);
}
var histogramSql = this._buildQuery(override);
return callback(null, histogramSql);
};
Histogram.prototype._buildQuery = function (override) {
var filteredQuery, basicsQuery, binsQuery;
var _column = this.column;
var _query = this.rewrittenQuery(this.query);
var basicsQuery, binsQuery;
filteredQuery = filteredQueryTpl({
_isFloatColumn: this._columnType === 'float',
_query: _query,
_column: _column
});
if (override && _.has(override, 'start') && _.has(override, 'end') && _.has(override, 'bins')) {
if (this._shouldOverride(override)) {
debug('overriding with %j', override);
basicsQuery = overrideBasicsQueryTpl({
_query: _query,
_column: _column,
@@ -177,7 +212,7 @@ Histogram.prototype.sql = function(psql, override, callback) {
_column: _column
});
if (override && _.has(override, 'bins')) {
if (this._shouldOverrideBins(override)) {
binsQuery = [
overrideBinsQueryTpl({
_bins: override.bins
@@ -198,22 +233,50 @@ Histogram.prototype.sql = function(psql, override, callback) {
}
}
var cteSql = [
filteredQuery,
basicsQuery,
binsQuery,
nullsQueryTpl({
_query: _query,
_column: _column
})
];
var histogramSql = [
"WITH",
[
basicsQuery,
binsQuery,
nullsQueryTpl({
if (this._columnType === 'float') {
cteSql.push(
infinitiesQueryTpl({
_query: _query,
_column: _column
}),
nansQueryTpl({
_query: _query,
_column: _column
})
].join(',\n'),
);
}
var histogramSql = [
"WITH",
cteSql.join(',\n'),
histogramQueryTpl({
_isFloatColumn: this._columnType === 'float',
_query: _query,
_column: _column
})
].join('\n');
return callback(null, histogramSql);
debug(histogramSql);
return histogramSql;
};
Histogram.prototype._shouldOverride = function (override) {
return override && _.has(override, 'start') && _.has(override, 'end') && _.has(override, 'bins');
};
Histogram.prototype._shouldOverrideBins = function (override) {
return override && _.has(override, 'bins');
};
@@ -1,8 +1,8 @@
var BaseOverviewsDataview = require('./base');
var BaseDataview = require('../list');
function List(query, options, queryRewriter, queryRewriteData, params) {
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params);
function List(query, options, queryRewriter, queryRewriteData, params, queries) {
BaseOverviewsDataview.call(this, query, options, BaseDataview, queryRewriter, queryRewriteData, params, queries);
}
List.prototype = Object.create(BaseOverviewsDataview.prototype);