Merge pull request #1120 from CartoDB/improve-metadata-sample

Improve efficiency of query samples (esp. for FDW's)
This commit is contained in:
Daniel G. Aubert
2019-09-04 12:37:41 +02:00
committed by GitHub
4 changed files with 76 additions and 61 deletions
+1
View File
@@ -5,6 +5,7 @@ Released 2019-XX-XX
Announcements:
- Improved efficiency of query samples while instatiating a map (#1120).
- Cache control header fine tuning. Set a shorter value for "max-age" directive if there is no way to know when to trigger the invalidation.
- Update deps:
- Update `cartodb-query-tables` to version [`0.5.0`](https://github.com/CartoDB/node-cartodb-query-tables/releases/tag/0.5.0)
@@ -125,22 +125,43 @@ function mergeColumns(results) {
}
}
const SAMPLE_SEED = 0.5;
const DEFAULT_SAMPLE_ROWS = 100;
function _sample(ctx, numRows) {
if (ctx.metaOptions.sample) {
const sampleProb = Math.min(ctx.metaOptions.sample.num_rows / numRows, 1);
// We'll use a safety limit just in case numRows is a bad estimate
const requestedRows = ctx.metaOptions.sample.num_rows || DEFAULT_SAMPLE_ROWS;
const limit = Math.ceil(requestedRows * 1.5);
let columns = ctx.metaOptions.sample.include_columns;
return queryUtils.queryPromise(ctx.dbConnection, _getSQL(
ctx,
sql => queryUtils.getQuerySample(sql, sampleProb, limit, SAMPLE_SEED, columns)
)).then(res => ({ sample: res.rows }));
function _sample(ctx) {
if (!ctx.metaOptions.sample) {
return Promise.resolve();
}
return Promise.resolve();
const limit = ctx.metaOptions.sample.num_rows || DEFAULT_SAMPLE_ROWS;
const columns = ctx.metaOptions.sample.include_columns;
const sqlMaxMin = _getSQL(ctx, sql => queryUtils.getMaxMinSpanColumnQuery(sql));
return queryUtils.queryPromise(ctx.dbConnection, sqlMaxMin)
.then(maxMinRes => {
const { min_id: min, id_span: span } = maxMinRes.rows[0];
if (!min || !span) {
return { rows: {} };
}
const values = _getSampleValuesFromRange(min, span, limit);
const sqlSample = _getSQL(ctx, sql => queryUtils.getSampleFromIdsQuery(sql, values, columns));
return queryUtils.queryPromise(ctx.dbConnection, sqlSample);
})
.then(res => ({ sample: res.rows }));
}
function _getSampleValuesFromRange (min, span, limit) {
const sample = new Set();
limit = limit < span ? limit : span;
while (sample.size < limit) {
sample.add(Math.floor(min + Math.random() * span));
}
return Array.from(sample);
}
function _columnsMetadataRequired(options) {
@@ -294,8 +315,8 @@ function (layer, dbConnection, callback) {
Promise.all([
_estimatedFeatureCount(ctx).then(
({ estimatedFeatureCount }) => _sample(ctx, estimatedFeatureCount)
.then(sampleResults => mergeResults([sampleResults, { estimatedFeatureCount }]))
({ estimatedFeatureCount }) => _sample(ctx)
.then(sampleResults => mergeResults([ sampleResults, { estimatedFeatureCount }] ))
),
_featureCount(ctx),
_aggrFeatureCount(ctx),
+12 -46
View File
@@ -98,58 +98,24 @@ function columnSelector(columns) {
throw new TypeError(`Bad argument type for columns: ${typeof columns}`);
}
module.exports.getQuerySample = function (query, sampleProb, limit = null, randomSeed = 0.5, columns = null) {
const singleTable = simpleQueryTable(query);
if (singleTable) {
return getTableSample(singleTable.table, columns || singleTable.columns, sampleProb, limit, randomSeed);
}
const limitClause = limit ? `LIMIT ${limit}` : '';
module.exports.getMaxMinSpanColumnQuery = function (query, column = 'cartodb_id') {
return `
WITH __cdb_rndseed AS (
SELECT setseed(${randomSeed})
)
SELECT ${columnSelector(columns)}
FROM (${substituteDummyTokens(query)}) AS __cdb_query
WHERE random() < ${sampleProb}
${limitClause}
SELECT
min(${column}) AS min_id,
max(${column}) AS max_id,
(max(${column}) - min(${column})) AS id_span
FROM (${substituteDummyTokens(query)}) _cdb_metadata_max_min_span;
`;
};
function getTableSample(table, columns, sampleProb, limit = null, randomSeed = 0.5) {
const limitClause = limit ? `LIMIT ${limit}` : '';
sampleProb *= 100;
randomSeed *= Math.pow(2, 31) - 1;
module.exports.getSampleFromIdsQuery = function (query, ids, columns, column = 'cartodb_id') {
return `
SELECT ${columnSelector(columns)}
FROM ${table}
TABLESAMPLE BERNOULLI (${sampleProb}) REPEATABLE (${randomSeed}) ${limitClause}
SELECT
${columnSelector(columns)}
FROM (${substituteDummyTokens(query)}) _cdb_metadata_sample
WHERE ${column} IN (${ids.join(',')})
`;
}
function simpleQueryTable(sql) {
const basicQuery =
/\s*SELECT\s+([\*a-z0-9_,\s]+?)\s+FROM\s+((\"[^"]+\"|[a-z0-9_]+)\.)?(\"[^"]+\"|[a-z0-9_]+)\s*;?\s*/i;
const unwrappedQuery = new RegExp('^' + basicQuery.source + '$', 'i');
// queries for named maps are wrapped like this:
var wrappedQuery = new RegExp(
'^\\s*SELECT\\s+\\*\\s+FROM\\s+\\(' +
basicQuery.source +
'\\)\\s+AS\\s+wrapped_query\\s+WHERE\\s+\\d+=1\\s*$',
'i'
);
let match = sql.match(unwrappedQuery);
if (!match) {
match = sql.match(wrappedQuery);
}
if (match) {
const columns = match[1];
const schema = match[3];
const table = match[4];
return { table: schema ? `${schema}.${table}` : table, columns };
}
return false;
}
};
function getQueryLimited(query, limit = 0) {
return `
@@ -583,6 +583,33 @@ describe(`[${desc}] Create mapnik layergroup`, function() {
});
});
it('should not provide a sample when the source table is empty', function (done) {
var testClient = new TestClient({
version: "1.4.0",
layers: [
{
type: "mapnik",
options: {
sql: "SELECT * FROM test_table_100 limit 0",
cartocss_version: "2.3.0",
cartocss: "#layer { line-width:16; }",
metadata: {
sample: {
num_rows: 30
}
}
}
}
]
});
testClient.getLayergroup(function(err, layergroup) {
assert.ifError(err);
assert.deepStrictEqual(layergroup.metadata.layers[0].meta.stats.sample, {});
testClient.drain(done);
});
});
it('can specify sample columns', function(done) {
var testClient = new TestClient({
version: '1.4.0',