HeatmapNG: support y min/max config (#50900)

Co-authored-by: Ryan McKinley <ryantxu@gmail.com>

public/app/plugins/panel/heatmap-new/module.tsx

@@ -1,14 +1,16 @@
 import React from 'react';
 
-import { FieldConfigProperty, FieldType, identityOverrideProcessor, PanelPlugin } from '@grafana/data';
+import { FieldConfigProperty, FieldType, identityOverrideProcessor, PanelData, PanelPlugin } from '@grafana/data';
 import { config } from '@grafana/runtime';
 import { AxisPlacement, GraphFieldConfig, ScaleDistribution, ScaleDistributionConfig } from '@grafana/schema';
 import { addHideFrom, ScaleDistributionEditor } from '@grafana/ui/src/options/builder';
 import { ColorScale } from 'app/core/components/ColorScale/ColorScale';
 import { addHeatmapCalculationOptions } from 'app/features/transformers/calculateHeatmap/editor/helper';
+import { readHeatmapScanlinesCustomMeta } from 'app/features/transformers/calculateHeatmap/heatmap';
 import { HeatmapCellLayout } from 'app/features/transformers/calculateHeatmap/models.gen';
 
 import { HeatmapPanel } from './HeatmapPanel';
+import { prepareHeatmapData } from './fields';
 import { heatmapChangedHandler, heatmapMigrationHandler } from './migrations';
 import { PanelOptions, defaultPanelOptions, HeatmapColorMode, HeatmapColorScale } from './models.gen';
 import { colorSchemes, quantizeScheme } from './palettes';
@@ -46,6 +48,13 @@ export const plugin = new PanelPlugin<PanelOptions, GraphFieldConfig>(HeatmapPan
   .setPanelOptions((builder, context) => {
     const opts = context.options ?? defaultPanelOptions;
 
+    let isOrdinalY = false;
+
+    try {
+      const v = prepareHeatmapData({ series: context.data } as PanelData, opts, config.theme2);
+      isOrdinalY = readHeatmapScanlinesCustomMeta(v.heatmap).yOrdinalDisplay != null;
+    } catch {}
+
     let category = ['Heatmap'];
 
     builder.addRadio({
@@ -81,8 +90,8 @@ export const plugin = new PanelPlugin<PanelOptions, GraphFieldConfig>(HeatmapPan
       },
     });
 
-    // TODO: support clamping the min/max range when there is a real axis
-    if (false && opts.calculate) {
+    if (!isOrdinalY) {
+      // if undefined, then show the min+max
       builder
         .addNumberInput({
           path: 'yAxis.min',

public/app/plugins/panel/heatmap-new/utils.ts

@@ -260,6 +260,7 @@ export function prepConfig(opts: PrepConfigOpts) {
   const yScale = yFieldConfig?.scaleDistribution ?? { type: ScaleDistribution.Linear };
   const yAxisReverse = Boolean(yAxisConfig.reverse);
   const shouldUseLogScale = yScale.type !== ScaleDistribution.Linear || heatmapType === DataFrameType.HeatmapSparse;
+  const isOrdianalY = readHeatmapScanlinesCustomMeta(dataRef.current?.heatmap).yOrdinalDisplay != null;
 
   // random to prevent syncing y in other heatmaps
   // TODO: try to match TimeSeries y keygen algo to sync with TimeSeries panels (when not isOrdianalY)
@@ -275,11 +276,34 @@ export function prepConfig(opts: PrepConfigOpts) {
     distribution: shouldUseLogScale ? ScaleDistribution.Log : ScaleDistribution.Linear,
     log: yScale.log ?? 2,
     range:
-      // sparse already accounts for le/ge by explicit yMin & yMax cell bounds, so use default log ranging
+      // sparse already accounts for le/ge by explicit yMin & yMax cell bounds, so no need to expand y range
       heatmapType === DataFrameType.HeatmapSparse
-        ? undefined
+        ? (u, dataMin, dataMax) => {
+            let scaleMin: number | null, scaleMax: number | null;
+
+            [scaleMin, scaleMax] = shouldUseLogScale
+              ? uPlot.rangeLog(dataMin, dataMax, (yScale.log ?? 2) as unknown as uPlot.Scale.LogBase, true)
+              : [dataMin, dataMax];
+
+            if (shouldUseLogScale && !isOrdianalY) {
+              let { min: explicitMin, max: explicitMax } = yAxisConfig;
+
+              // guard against <= 0
+              if (explicitMin != null && explicitMin > 0) {
+                scaleMin = explicitMin;
+              }
+
+              if (explicitMax != null && explicitMax > 0) {
+                scaleMax = explicitMax;
+              }
+            }
+
+            return [scaleMin, scaleMax];
+          }
         : // dense and ordinal only have one of yMin|yMax|y, so expand range by one cell in the direction of le/ge/unknown
           (u, dataMin, dataMax) => {
+            let { min: explicitMin, max: explicitMax } = yAxisConfig;
+
             // logarithmic expansion
             if (shouldUseLogScale) {
               let yExp = u.scales[yScaleKey].log!;
@@ -316,6 +340,17 @@ export function prepConfig(opts: PrepConfigOpts) {
                 dataMin /= yExp / 2;
                 dataMax *= yExp / 2;
               }
+
+              if (!isOrdianalY) {
+                // guard against <= 0
+                if (explicitMin != null && explicitMin > 0) {
+                  dataMin = explicitMin;
+                }
+
+                if (explicitMax != null && explicitMax > 0) {
+                  dataMax = explicitMax;
+                }
+              }
             }
             // linear expansion
             else {
@@ -337,12 +372,17 @@ export function prepConfig(opts: PrepConfigOpts) {
               } else {
                 // how to expand scale range if inferred non-regular or log buckets?
               }
+
+              if (!isOrdianalY) {
+                dataMin = explicitMin ?? dataMin;
+                dataMax = explicitMax ?? dataMax;
+              }
             }
+
             return [dataMin, dataMax];
           },
   });
 
-  const isOrdianalY = readHeatmapScanlinesCustomMeta(dataRef.current?.heatmap).yOrdinalDisplay != null;
   const disp = dataRef.current?.heatmap?.fields[1].display ?? getValueFormat('short');
 
   builder.addAxis({