DataFrame: insert null values along interval (#44622) (#44792)

(cherry picked from commit 3504844ad7)

Co-authored-by: Leon Sorokin <leeoniya@gmail.com>
This commit is contained in:
Grot (@grafanabot)
2022-02-02 14:05:14 -06:00
committed by GitHub
co-authored by Leon Sorokin
parent e031568aa4
commit b4579334d4
8 changed files with 365 additions and 29 deletions
@@ -67,6 +67,12 @@ export interface FieldConfig<TOptions = any> {
min?: number | null;
max?: number | null;
// Interval indicates the expected regular step between values in the series.
// When an interval exists, consumers can identify "missing" values when the expected value is not present.
// The grafana timeseries visualization will render disconnected values when missing values are found it the time field.
// The interval uses the same units as the values. For time.Time, this is defined in milliseconds.
interval?: number | null;
// Convert input values into a display string
mappings?: ValueMapping[];
@@ -0,0 +1,215 @@
import { ArrayVector, FieldType, MutableDataFrame } from '@grafana/data';
import { applyNullInsertThreshold } from './nullInsertThreshold';
function randInt(min: number, max: number) {
return Math.floor(Math.random() * (max - min + 1) + min);
}
function genFrame() {
let fieldCount = 10;
let valueCount = 3000;
let step = 1000;
let skipProb = 0.5;
let skipSteps = [1, 5]; // min, max
let allValues = Array(fieldCount);
allValues[0] = Array(valueCount);
for (let i = 0, curStep = Date.now(); i < valueCount; i++) {
curStep = allValues[0][i] = curStep + step * (Math.random() < skipProb ? randInt(skipSteps[0], skipSteps[1]) : 1);
}
for (let fi = 1; fi < fieldCount; fi++) {
let values = Array(valueCount);
for (let i = 0; i < valueCount; i++) {
values[i] = Math.random() * 100;
}
allValues[fi] = values;
}
return {
length: valueCount,
fields: allValues.map((values, i) => {
return {
name: 'A-' + i,
type: i === 0 ? FieldType.time : FieldType.number,
config: {
interval: i === 0 ? step : null,
},
values: new ArrayVector(values),
};
}),
};
}
describe('nullInsertThreshold Transformer', () => {
test('should insert nulls at +threshold between adjacent > threshold: 1', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, values: [1, 3, 10] },
{ name: 'One', type: FieldType.number, config: { custom: { insertNulls: 1 } }, values: [4, 6, 8] },
{ name: 'Two', type: FieldType.string, config: { custom: { insertNulls: 1 } }, values: ['a', 'b', 'c'] },
],
});
const result = applyNullInsertThreshold(df);
expect(result.fields[0].values.toArray()).toStrictEqual([1, 2, 3, 4, 10]);
expect(result.fields[1].values.toArray()).toStrictEqual([4, null, 6, null, 8]);
expect(result.fields[2].values.toArray()).toStrictEqual(['a', null, 'b', null, 'c']);
});
test('should insert nulls at +threshold between adjacent > threshold: 2', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, values: [5, 7, 11] },
{ name: 'One', type: FieldType.number, config: { custom: { insertNulls: 2 } }, values: [4, 6, 8] },
{ name: 'Two', type: FieldType.string, config: { custom: { insertNulls: 2 } }, values: ['a', 'b', 'c'] },
],
});
const result = applyNullInsertThreshold(df);
expect(result.fields[0].values.toArray()).toStrictEqual([5, 7, 9, 11]);
expect(result.fields[1].values.toArray()).toStrictEqual([4, 6, null, 8]);
expect(result.fields[2].values.toArray()).toStrictEqual(['a', 'b', null, 'c']);
});
test('should insert nulls at +interval between adjacent > interval: 1', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, config: { interval: 1 }, values: [1, 3, 10] },
{ name: 'One', type: FieldType.number, values: [4, 6, 8] },
{ name: 'Two', type: FieldType.string, values: ['a', 'b', 'c'] },
],
});
const result = applyNullInsertThreshold(df);
expect(result.fields[0].values.toArray()).toStrictEqual([1, 2, 3, 4, 10]);
expect(result.fields[1].values.toArray()).toStrictEqual([4, null, 6, null, 8]);
expect(result.fields[2].values.toArray()).toStrictEqual(['a', null, 'b', null, 'c']);
});
// TODO: make this work
test.skip('should insert nulls at +threshold (when defined) instead of +interval', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, config: { interval: 2 }, values: [5, 7, 11] },
{ name: 'One', type: FieldType.number, config: { custom: { insertNulls: 1 } }, values: [4, 6, 8] },
{ name: 'Two', type: FieldType.string, config: { custom: { insertNulls: 1 } }, values: ['a', 'b', 'c'] },
],
});
const result = applyNullInsertThreshold(df);
expect(result.fields[0].values.toArray()).toStrictEqual([5, 6, 7, 8, 11]);
expect(result.fields[1].values.toArray()).toStrictEqual([4, null, 6, null, 8]);
expect(result.fields[2].values.toArray()).toStrictEqual(['a', null, 'b', null, 'c']);
});
test('should insert nulls at midpoints between adjacent > interval: 2', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, config: { interval: 2 }, values: [5, 7, 11] },
{ name: 'One', type: FieldType.number, values: [4, 6, 8] },
{ name: 'Two', type: FieldType.string, values: ['a', 'b', 'c'] },
],
});
const result = applyNullInsertThreshold(df);
expect(result.fields[0].values.toArray()).toStrictEqual([5, 7, 9, 11]);
expect(result.fields[1].values.toArray()).toStrictEqual([4, 6, null, 8]);
expect(result.fields[2].values.toArray()).toStrictEqual(['a', 'b', null, 'c']);
});
test('should noop on fewer than two values', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, config: { interval: 1 }, values: [1] },
{ name: 'Value', type: FieldType.number, values: [1] },
],
});
const result = applyNullInsertThreshold(df);
expect(result).toBe(df);
});
test('should noop on invalid threshold', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, values: [1, 2, 4] },
{ name: 'Value', type: FieldType.number, config: { custom: { insertNulls: -1 } }, values: [1, 1, 1] },
],
});
const result = applyNullInsertThreshold(df);
expect(result).toBe(df);
});
test('should noop on invalid interval', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, config: { interval: -1 }, values: [1, 2, 4] },
{ name: 'Value', type: FieldType.number, values: [1, 1, 1] },
],
});
const result = applyNullInsertThreshold(df);
expect(result).toBe(df);
});
test('should noop when no missing steps', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, config: { interval: 1 }, values: [1, 2, 3] },
{ name: 'Value', type: FieldType.number, values: [1, 1, 1] },
],
});
const result = applyNullInsertThreshold(df);
expect(result).toBe(df);
});
test('should noop when refFieldName not found', () => {
const df = new MutableDataFrame({
refId: 'A',
fields: [
{ name: 'Time', type: FieldType.time, config: { interval: 1 }, values: [1, 2, 5] },
{ name: 'Value', type: FieldType.number, values: [1, 1, 1] },
],
});
const result = applyNullInsertThreshold(df, 'Time2');
expect(result).toBe(df);
});
test('perf stress test should be <= 10ms', () => {
// 10 fields x 3,000 values with 50% skip (output = 10 fields x 6,000 values)
let bigFrameA = genFrame();
// eslint-disable-next-line no-console
console.time('insertValues-10x3k');
applyNullInsertThreshold(bigFrameA);
// eslint-disable-next-line no-console
console.timeEnd('insertValues-10x3k');
});
});
@@ -0,0 +1,115 @@
import { ArrayVector, DataFrame, FieldType } from '@grafana/data';
type InsertMode = (prev: number, next: number, threshold: number) => number;
const INSERT_MODES = {
threshold: (prev: number, next: number, threshold: number) => prev + threshold,
midpoint: (prev: number, next: number, threshold: number) => (prev + next) / 2,
// previous time + 1ms to prevent StateTimeline from forward-interpolating prior state
plusone: (prev: number, next: number, threshold: number) => prev + 1,
};
export function applyNullInsertThreshold(
frame: DataFrame,
refFieldName?: string | null,
insertMode: InsertMode = INSERT_MODES.threshold
): DataFrame {
if (frame.length < 2) {
return frame;
}
const refField = frame.fields.find((field) => {
// note: getFieldDisplayName() would require full DF[]
return refFieldName != null ? field.name === refFieldName : field.type === FieldType.time;
});
if (refField == null) {
return frame;
}
const thresholds = frame.fields.map((field) => field.config.custom?.insertNulls ?? refField.config.interval ?? null);
const uniqueThresholds = new Set<number>(thresholds);
uniqueThresholds.delete(null as any);
if (uniqueThresholds.size === 0) {
return frame;
}
if (uniqueThresholds.size === 1) {
const threshold = uniqueThresholds.values().next().value;
if (threshold <= 0) {
return frame;
}
const refValues = refField.values.toArray();
const frameValues = frame.fields.map((field) => field.values.toArray());
const filledFieldValues = nullInsertThreshold(refValues, frameValues, threshold, insertMode);
if (filledFieldValues === frameValues) {
return frame;
}
return {
...frame,
length: filledFieldValues[0].length,
fields: frame.fields.map((field, i) => ({
...field,
values: new ArrayVector(filledFieldValues[i]),
})),
};
}
// TODO: unique threshold-per-field (via overrides) is unimplemented
// should be done by processing each (refField + thresholdA-field1 + thresholdA-field2...)
// as a separate nullInsertThreshold() dataset, then re-join into single dataset via join()
return frame;
}
function nullInsertThreshold(refValues: number[], frameValues: any[][], threshold: number, getInsertValue: InsertMode) {
const len = refValues.length;
let prevValue: number = refValues[0];
const refValuesNew: number[] = [prevValue];
for (let i = 1; i < len; i++) {
const curValue = refValues[i];
if (curValue - prevValue > threshold) {
refValuesNew.push(getInsertValue(prevValue, curValue, threshold));
}
refValuesNew.push(curValue);
prevValue = curValue;
}
const filledLen = refValuesNew.length;
if (filledLen === len) {
return frameValues;
}
const filledFieldValues: any[][] = [];
for (let fieldValues of frameValues) {
let filledValues;
if (fieldValues !== refValues) {
filledValues = Array(filledLen);
for (let i = 0, j = 0; i < filledLen; i++) {
filledValues[i] = refValues[j] === refValuesNew[i] ? fieldValues[j++] : null;
}
} else {
filledValues = refValuesNew;
}
filledFieldValues.push(filledValues);
}
return filledFieldValues;
}
@@ -1,6 +1,7 @@
import { XYFieldMatchers } from './types';
import { ArrayVector, DataFrame, FieldConfig, FieldType, outerJoinDataFrames } from '@grafana/data';
import { nullToUndefThreshold } from './nullToUndefThreshold';
import { applyNullInsertThreshold } from './nullInsertThreshold';
import { AxisPlacement, GraphFieldConfig, ScaleDistribution, ScaleDistributionConfig } from '@grafana/schema';
import { FIXED_UNIT } from './GraphNG';
@@ -32,7 +33,7 @@ function applySpanNullsThresholds(frame: DataFrame) {
export function preparePlotFrame(frames: DataFrame[], dimFields: XYFieldMatchers) {
let alignedFrame = outerJoinDataFrames({
frames: frames,
frames: frames.map((frame) => applyNullInsertThreshold(frame)),
joinBy: dimFields.x,
keep: dimFields.y,
keepOriginIndices: true,
@@ -1,5 +1,6 @@
import { DataFrame, FieldConfig, FieldSparkline, IndexVector } from '@grafana/data';
import { GraphFieldConfig } from '@grafana/schema';
import { applyNullInsertThreshold } from '../GraphNG/nullInsertThreshold';
/** @internal
* Given a sparkline config returns a DataFrame ready to be turned into Plot data set
@@ -11,7 +12,7 @@ export function preparePlotFrame(sparkline: FieldSparkline, config?: FieldConfig
...config,
};
return {
return applyNullInsertThreshold({
refId: 'sparkline',
fields: [
sparkline.x ?? IndexVector.newField(length),
@@ -21,5 +22,5 @@ export function preparePlotFrame(sparkline: FieldSparkline, config?: FieldConfig
},
],
length,
};
});
}