Clean up

elastic · Aug 16, 2022 · e4f786e · e4f786e
1 parent 5fbc10b
commit e4f786e
Show file tree

Hide file tree

Showing 8 changed files with 71 additions and 42 deletions.
diff --git a/...ugins/discover/public/application/main/components/field_stats_table/field_stats_table.tsx b/...ugins/discover/public/application/main/components/field_stats_table/field_stats_table.tsx
@@ -29,13 +29,32 @@ export interface DataVisualizerGridEmbeddableInput extends EmbeddableInput {
   query?: Query | AggregateQuery;
   visibleFieldNames?: string[];
   filters?: Filter[];
+  /**
+   * Whether to show the mini chart distributions when table is first rendered
+   */
   showPreviewByDefault?: boolean;
+  /**
+   * Whether to show option directly in table rows to edit the fields in the data view
+   */
+  allowEditDataView?: boolean;
   /**
    * Callback to add a filter to filter bar
    */
   onAddFilter?: (field: DataViewField | string, value: string, type: '+' | '-') => void;
+  /**
+   * Session ID used for Kibana's search to save and restore search sessions
+   */
   sessionId?: string;
+  /**
+   * List of fields to fetch field statistics for
+   * since we might not have fetch all fields at once, but rather all that are available
+   */
   fieldsToFetch?: string[];
+  /**
+   * The preferred mode for sampling data for the field statistics
+   * default as 'autoRandomSampler'
+   */
+  samplingMode?: string;
 }
 export interface DataVisualizerGridEmbeddableOutput extends EmbeddableOutput {
   showDistributions?: boolean;

diff --git a/.../public/application/index_data_visualizer/embeddables/grid_embeddable/grid_embeddable.tsx b/.../public/application/index_data_visualizer/embeddables/grid_embeddable/grid_embeddable.tsx
@@ -43,17 +43,41 @@ export interface DataVisualizerGridInput {
   dataView: DataView;
   savedSearch?: SavedSearch | SavedSearchSavedObject | null;
   query?: Query;
+  /**
+   * List of fields/rows to show in the table (e.g. Discover's 'Selected fields')
+   */
   visibleFieldNames?: string[];
   filters?: Filter[];
+  /**
+   * Whether to show the mini chart distributions when table is first rendered
+   */
   showPreviewByDefault?: boolean;
+  /**
+   * Whether to show option directly in table rows to edit the fields in the data view
+   */
   allowEditDataView?: boolean;
+  /**
+   * Unique embeddable id as there might be multiple instances in Dashboard
+   */
   id?: string;
   /**
    * Callback to add a filter to filter bar
    */
   onAddFilter?: (field: DataViewField | string, value: string, type: '+' | '-') => void;
+  /**
+   * Session ID used for Kibana's search to save and restore search sessions
+   */
   sessionId?: string;
+  /**
+   * List of fields to fetch field statistics for
+   * since we might not have fetch all fields at once, but rather all that are available
+   */
   fieldsToFetch?: string[];
+  /**
+   * The preferred mode for sampling data for the field statistics
+   * default as 'autoRandomSampler'
+   */
+  samplingMode?: string;
 }
 export type DataVisualizerGridEmbeddableInput = EmbeddableInput & DataVisualizerGridInput;
 export type DataVisualizerGridEmbeddableOutput = EmbeddableOutput;
@@ -82,8 +106,15 @@ export const EmbeddableWrapper = ({
     [dataVisualizerListState, onOutputChange]
   );
 
-  const { configs, searchQueryLanguage, searchString, extendedColumns, progress, setLastRefresh } =
-    useDataVisualizerGridData(input, dataVisualizerListState);
+  const {
+    configs,
+    searchQueryLanguage,
+    searchString,
+    extendedColumns,
+    progress,
+    setLastRefresh,
+    overallStats,
+  } = useDataVisualizerGridData(input, dataVisualizerListState);
 
   useEffect(() => {
     setLastRefresh(Date.now());
@@ -141,6 +172,7 @@ export const EmbeddableWrapper = ({
       showPreviewByDefault={input?.showPreviewByDefault}
       onChange={onOutputChange}
       loading={progress < 100}
+      totalCount={overallStats?.documentCountStats?.totalCount ?? 0}
     />
   );
 };

diff --git a/...plugins/data_visualizer/public/application/index_data_visualizer/hooks/use_field_stats.ts b/...plugins/data_visualizer/public/application/index_data_visualizer/hooks/use_field_stats.ts
@@ -179,7 +179,6 @@ export function useFieldStatsSearchStrategy(
       sessionId: searchStrategyParams?.sessionId,
     };
 
-    console.log('samplingProbability', samplingProbability);
     const batches = createBatchedRequests(
       pageOfConfigs.map((config, idx) => ({
         fieldName: config.fieldName,

diff --git a/...ugins/data_visualizer/public/application/index_data_visualizer/hooks/use_overall_stats.ts b/...ugins/data_visualizer/public/application/index_data_visualizer/hooks/use_overall_stats.ts
@@ -211,22 +211,7 @@ export function useOverallStats<TParams extends OverallStatsSearchStrategyParams
 
     const sub = rateLimitingForkJoin<
       AggregatableFieldOverallStats | NonAggregatableFieldOverallStats | undefined
-    >(
-      [
-        // from(
-        //   getDocumentCountStats(
-        //     data.search,
-        //     searchStrategyParams,
-        //     searchOptions,
-        //     browserSessionSeed,
-        //     probability
-        //   )
-        // ),
-        ...aggregatableOverallStatsObs,
-        ...nonAggregatableFieldsObs,
-      ],
-      MAX_CONCURRENT_REQUESTS
-    );
+    >([...aggregatableOverallStatsObs, ...nonAggregatableFieldsObs], MAX_CONCURRENT_REQUESTS);
 
     searchSubscription$.current = sub.subscribe({
       next: (value) => {
@@ -245,7 +230,6 @@ export function useOverallStats<TParams extends OverallStatsSearchStrategyParams
         });
 
         const totalCount = documentCountStats?.totalCount ?? 0;
-        console.log('aggregatableOverallStatsResp', aggregatableOverallStatsResp);
 
         const aggregatableOverallStats = processAggregatableFieldsExistResponse(
           aggregatableOverallStatsResp,

diff --git a/...public/application/index_data_visualizer/search_strategy/requests/get_date_field_stats.ts b/...public/application/index_data_visualizer/search_strategy/requests/get_date_field_stats.ts
@@ -27,7 +27,7 @@ export const getDateFieldsStatsRequest = (
   params: FieldStatsCommonRequestParams,
   fields: Field[]
 ) => {
-  const { index, query, runtimeFieldMap, samplerShardSize } = params;
+  const { index, query, runtimeFieldMap } = params;
 
   const size = 0;
 

diff --git a/...r/public/application/index_data_visualizer/search_strategy/requests/get_document_stats.ts b/...r/public/application/index_data_visualizer/search_strategy/requests/get_document_stats.ts
@@ -125,7 +125,12 @@ export const getDocumentCountStats = async (
     index,
     body: {
       query,
-      aggs: aggregations,
+      ...(Array.isArray(fieldsToFetch) &&
+      timeFieldName !== undefined &&
+      intervalMs !== undefined &&
+      intervalMs > 0
+        ? { aggs: aggregations }
+        : {}),
       ...(isPopulatedObject(runtimeFieldMap) ? { runtime_mappings: runtimeFieldMap } : {}),
     },
     track_total_hits: false,

diff --git a/...lic/application/index_data_visualizer/search_strategy/requests/get_numeric_field_stats.ts b/...lic/application/index_data_visualizer/search_strategy/requests/get_numeric_field_stats.ts
@@ -16,7 +16,7 @@ import {
   ISearchOptions,
 } from '@kbn/data-plugin/common';
 import type { ISearchStart } from '@kbn/data-plugin/public';
-import { buildSamplerAggregation, getSamplerAggregationsResponsePath } from '@kbn/ml-agg-utils';
+import { getSamplerAggregationsResponsePath } from '@kbn/ml-agg-utils';
 import { isPopulatedObject } from '@kbn/ml-is-populated-object';
 import { buildRandomSamplerAggregation } from './build_random_sampler_agg';
 import { MAX_PERCENT, PERCENTILE_SPACING, SAMPLER_TOP_TERMS_THRESHOLD } from './constants';
@@ -35,7 +35,7 @@ export const getNumericFieldsStatsRequest = (
   params: FieldStatsCommonRequestParams,
   fields: Field[]
 ) => {
-  const { index, query, runtimeFieldMap, samplerShardSize } = params;
+  const { index, query, runtimeFieldMap } = params;
 
   const size = 0;
 
@@ -79,14 +79,13 @@ export const getNumericFieldsStatsRequest = (
       } as AggregationsTermsAggregation,
     };
 
-    // If cardinality >= SAMPLE_TOP_TERMS_THRESHOLD, run the top terms aggregation
-    // in a sampler aggregation, even if no sampling has been specified (samplerShardSize < 1).
-    if (samplerShardSize < 1 && field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD) {
-      aggs[`${safeFieldName}_top`] = buildSamplerAggregation(
+    if (field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD) {
+      aggs[`${safeFieldName}_top`] = buildRandomSamplerAggregation(
         {
           top,
         },
-        0.05
+        params.samplingProbability,
+        params.browserSessionSeed
       );
     } else {
       aggs[`${safeFieldName}_top`] = top;
@@ -132,7 +131,6 @@ export const fetchNumericFieldsStats = (
         if (!isIKibanaSearchResponse(resp)) return resp;
 
         const aggregations = resp.rawResponse.aggregations;
-        console.log('NUMERIC aggregations', aggregations);
         const aggsPath = getSamplerAggregationsResponsePath(samplerShardSize);
 
         const batchStats: NumericFieldStats[] = [];
@@ -151,7 +149,7 @@ export const fetchNumericFieldsStats = (
           );
 
           const topAggsPath = [...aggsPath, `${safeFieldName}_top`];
-          if (samplerShardSize < 1 && field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD) {
+          if (field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD) {
             topAggsPath.push('top');
           }
 
@@ -163,8 +161,7 @@ export const fetchNumericFieldsStats = (
             min: get(fieldStatsResp, 'min', 0),
             max: get(fieldStatsResp, 'max', 0),
             avg: get(fieldStatsResp, 'avg', 0),
-            isTopValuesSampled:
-              field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD || samplerShardSize > 0,
+            isTopValuesSampled: field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD,
             topValues,
             topValuesSampleSize: get(aggregations, ['sample', 'doc_count']),
             // @todo: remove
@@ -188,7 +185,6 @@ export const fetchNumericFieldsStats = (
             );
           }
 
-          console.log(stats.fieldName, stats);
           batchStats.push(stats);
         });
 

diff --git a/...blic/application/index_data_visualizer/search_strategy/requests/get_string_field_stats.ts b/...blic/application/index_data_visualizer/search_strategy/requests/get_string_field_stats.ts
@@ -33,7 +33,7 @@ export const getStringFieldStatsRequest = (
   params: FieldStatsCommonRequestParams,
   fields: Field[]
 ) => {
-  const { index, query, runtimeFieldMap, samplerShardSize } = params;
+  const { index, query, runtimeFieldMap } = params;
 
   const size = 0;
 
@@ -50,9 +50,7 @@ export const getStringFieldStatsRequest = (
       } as AggregationsTermsAggregation,
     };
 
-    // If cardinality >= SAMPLE_TOP_TERMS_THRESHOLD, run the top terms aggregation
-    // in a sampler aggregation, even if no sampling has been specified (samplerShardSize < 1).
-    if (samplerShardSize < 1 && field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD) {
+    if (field.cardinality >= SAMPLER_TOP_TERMS_THRESHOLD) {
       aggs[`${safeFieldName}_top`] = {
         sampler: {
           shard_size: SAMPLER_TOP_TERMS_SHARD_SIZE,
@@ -107,9 +105,6 @@ export const fetchStringFieldsStats = (
         const aggsPath = getSamplerAggregationsResponsePath(samplerShardSize);
         const batchStats: StringFieldStats[] = [];
 
-        console.log('fields', fields);
-        console.log('aggregations', aggregations);
-
         fields.forEach((field, i) => {
           const safeFieldName = field.safeFieldName;
 
@@ -128,7 +123,6 @@ export const fetchStringFieldsStats = (
             // @todo: remove
             topValuesSamplerShardSize: get(aggregations, ['sample', 'doc_count']),
           };
-          console.log(stats.fieldName, stats);
 
           batchStats.push(stats);
         });