fix: validate band information is consistent (#1017)

#### Motivation

When tiffs are retiled they need to have consistent banding information
or GDAL will not be able to join them together easily

#### Modification

When a retile is requested validate that all tiffs in the retile have
the same consistent band information.

#### Checklist

_If not applicable, provide explanation of why._

- [ ] Tests updated
- [ ] Docs updated
- [ ] Issue linked in Title

Author: blacha

src/commands/tileindex-validate/__test__/tileindex.validate.data.ts

@@ -1,4 +1,4 @@
-import { Size, Source, Tiff, TiffImage } from '@cogeotiff/core';
+import { SampleFormat, Size, Source, Tiff, TiffImage, TiffTag } from '@cogeotiff/core';
 
 import { MapSheet } from '../../../utils/mapsheet.js';
 
@@ -29,7 +29,18 @@ export class FakeCogTiff extends Tiff {
     image: Partial<{ origin: number[]; epsg: number; resolution: number[]; size: Size; isGeoLocated: boolean }>,
   ) {
     super({ url: new URL(uri) } as Source);
-    this.images = [{ ...structuredClone(DefaultTiffImage), valueGeo, ...image } as any];
+    this.images = [
+      {
+        ...structuredClone(DefaultTiffImage),
+        valueGeo,
+        ...image,
+        fetch(tag: TiffTag) {
+          if (tag === TiffTag.BitsPerSample) return [8, 8, 8];
+          if (tag === TiffTag.SampleFormat) return [SampleFormat.Uint, SampleFormat.Uint, SampleFormat.Uint];
+          return null;
+        },
+      } as any,
+    ];
   }
 
   static fromTileName(tileName: string): FakeCogTiff {

src/commands/tileindex-validate/__test__/tileindex.validate.test.ts

@@ -53,16 +53,18 @@ describe('getTileName', () => {
     assert.equal(convertTileName('CK08_50000_0101', 50000), 'CK08');
   });
 
-  for (const sheet of MapSheetData) {
-    it('should get the top left 1:50k, 1:10k, 1:5k, 1:1k, and 1:500 for ' + sheet.code, () => {
+  it('should get the top left 1:50k, 1:10k, 1:5k, 1:1k, and 1:500 for all sheets', () => {
+    for (const sheet of MapSheetData) {
       assert.equal(getTileName(sheet.origin.x, sheet.origin.y, 50000), sheet.code);
       assert.equal(getTileName(sheet.origin.x, sheet.origin.y, 10000), sheet.code + '_10000_0101');
       assert.equal(getTileName(sheet.origin.x, sheet.origin.y, 5000), sheet.code + '_5000_0101');
       assert.equal(getTileName(sheet.origin.x, sheet.origin.y, 1000), sheet.code + '_1000_0101');
       assert.equal(getTileName(sheet.origin.x, sheet.origin.y, 500), sheet.code + '_500_001001');
-    });
+    }
+  });
 
-    it('should get the bottom right 1:50k, 1:10k, 1:5k, 1:1k for ' + sheet.code, () => {
+  it('should get the bottom right 1:50k, 1:10k, 1:5k, 1:1k for all sheets', () => {
+    for (const sheet of MapSheetData) {
       // for each scale calculate the bottom right tile then find the mid point of it
       // then look up the tile name from the midpoint and ensure it is the same
       for (const scale of [10_000, 5_000, 1_000, 500] as const) {
@@ -74,8 +76,8 @@ describe('getTileName', () => {
         const midPointY = ret.origin.y - ret.height / 2;
         assert.equal(getTileName(midPointX, midPointY, scale), sheetName);
       }
-    });
-  }
+    }
+  });
 });
 
 describe('tiffLocation', () => {

src/commands/tileindex-validate/tileindex.validate.ts

@@ -6,6 +6,7 @@ import { boolean, command, flag, number, option, optional, restPositionals, stri
 import { CliInfo } from '../../cli.info.js';
 import { logger } from '../../log.js';
 import { isArgo } from '../../utils/argo.js';
+import { extractBandInformation } from '../../utils/band.js';
 import { FileFilter, getFiles } from '../../utils/chunk.js';
 import { findBoundingBox } from '../../utils/geotiff.js';
 import { GridSize, GridSizes, MapSheet, MapSheetTileGridSize, SheetRanges } from '../../utils/mapsheet.js';
@@ -276,7 +277,11 @@ export const commandTileIndexValidate = command({
     for (const val of outputs.values()) {
       if (val.length < 2) continue;
       if (args.retile) {
-        logger.info({ tileName: val[0]?.tileName, uris: val.map((v) => v.source) }, 'TileIndex:Retile');
+        const bandType = validateConsistentBands(val);
+        logger.info(
+          { tileName: val[0]?.tileName, uris: val.map((v) => v.source), bands: bandType },
+          'TileIndex:Retile',
+        );
       } else {
         retileNeeded = true;
         logger.error({ tileName: val[0]?.tileName, uris: val.map((v) => v.source) }, 'TileIndex:Duplicate');
@@ -300,6 +305,31 @@ export const commandTileIndexValidate = command({
   },
 });
 
+/**
+ * Validate all tiffs have consistent band information
+ * @returns list of bands in the first image if consistent with the other images
+ * @throws if one image does not have consistent band information
+ */
+function validateConsistentBands(locs: TiffLocation[]): string[] {
+  const firstBands = locs[0]?.bands ?? [];
+  const firstBand = firstBands.join(',');
+
+  for (let i = 1; i < locs.length; i++) {
+    const currentBands = locs[i]?.bands.join(',');
+
+    // If the current image doesn't have the same band information gdalbuildvrt will fail
+    if (currentBands !== firstBand) {
+      // Dump all the imagery and their band types into logs so it can be debugged later
+      for (const v of locs) {
+        logger.error({ path: v.source, bands: v.bands.join(',') }, 'TileIndex:Bands:Heterogenous');
+      }
+
+      throw new Error(`heterogenous bands: ${currentBands} vs ${firstBand} from: ${locs[0]?.source}`);
+    }
+  }
+  return firstBands;
+}
+
 export function groupByTileName(tiffs: TiffLocation[]): Map<string, TiffLocation[]> {
   const duplicates: Map<string, TiffLocation[]> = new Map();
   for (const loc of tiffs) {
@@ -319,6 +349,12 @@ export interface TiffLocation {
   epsg?: number | null;
   /** Output tile name */
   tileName: string;
+  /**
+   * List of bands inside the tiff in the format `uint8` `uint16`
+   *
+   * @see {@link extractBandInformation} for more information on bad types
+   */
+  bands: string[];
 }
 
 /**
@@ -368,7 +404,13 @@ export async function extractTiffLocations(
         //   // Also need to allow for ~1.5cm of error between bounding boxes.
         //   // assert bbox == MapSheet.getMapTileIndex(tileName).bbox
         // }
-        return { bbox, source: tiff.source.url.href, tileName, epsg: tiff.images[0]?.epsg };
+        return {
+          bbox,
+          source: tiff.source.url.href,
+          tileName,
+          epsg: tiff.images[0]?.epsg,
+          bands: await extractBandInformation(tiff),
+        };
       } catch (e) {
         logger.error({ reason: e, source: tiff.source }, 'ExtractTiffLocation:Failed');
         return null;
@@ -386,6 +428,7 @@ export async function extractTiffLocations(
 
   return output;
 }
+
 export function getSize(extent: [number, number, number, number]): Size {
   return { width: extent[2] - extent[0], height: extent[3] - extent[1] };
 }

src/utils/__test__/band.test.ts

@@ -0,0 +1,19 @@
+import assert from 'node:assert';
+import { describe, it } from 'node:test';
+
+import { createTiff } from '../../commands/common.js';
+import { extractBandInformation } from '../band.js';
+
+describe('extractBandInformation', () => {
+  it('should extract basic band information (8-bit)', async () => {
+    const testTiff = await createTiff('./src/commands/tileindex-validate/__test__/data/8b.tiff');
+    const bands = await extractBandInformation(testTiff);
+    assert.equal(bands.join(','), 'uint8,uint8,uint8');
+  });
+
+  it('should extract basic band information (16-bit)', async () => {
+    const testTiff = await createTiff('./src/commands/tileindex-validate/__test__/data/16b.tiff');
+    const bands = await extractBandInformation(testTiff);
+    assert.equal(bands.join(','), 'uint16,uint16,uint16');
+  });
+});

src/utils/band.ts

@@ -0,0 +1,56 @@
+import { SampleFormat, Tiff, TiffImage, TiffTag } from '@cogeotiff/core';
+
+function getDataType(i: SampleFormat): string {
+  switch (i) {
+    case SampleFormat.Uint:
+      return 'uint';
+    case SampleFormat.Int:
+      return 'int';
+    case SampleFormat.Float:
+      return 'float';
+    case SampleFormat.Void:
+      return 'void';
+    case SampleFormat.ComplexFloat:
+      return 'cfloat';
+    case SampleFormat.ComplexInt:
+      return 'cint';
+    default:
+      return 'unknown';
+  }
+}
+/**
+ * Load the band information from a tiff and return it as a array of human friendly names
+ *
+ * @example
+ * `[uint16, uint16, uint16]` 3 band uint16
+ *
+ * @param tiff Tiff to extract band information from
+ * @returns list of band information
+ * @throws {Error} if cannot extract band information
+ */
+export async function extractBandInformation(tiff: Tiff): Promise<string[]> {
+  const firstImage = tiff.images[0] as TiffImage;
+
+  const [dataType, bitsPerSample] = await Promise.all([
+    /** firstImage.fetch(TiffTag.Photometric), **/ // TODO enable RGB detection
+    firstImage.fetch(TiffTag.SampleFormat),
+    firstImage.fetch(TiffTag.BitsPerSample),
+  ]);
+
+  if (bitsPerSample == null) {
+    throw new Error(`Failed to extract band information from: ${tiff.source.url.href}`);
+  }
+
+  if (dataType && dataType.length !== bitsPerSample.length) {
+    throw new Error(`Datatype and bits per sample miss match: ${tiff.source.url.href}`);
+  }
+
+  const imageBands: string[] = [];
+  for (let i = 0; i < bitsPerSample.length; i++) {
+    const type = getDataType(dataType ? (dataType[i] as SampleFormat) : SampleFormat.Uint);
+    const bits = bitsPerSample[i];
+    imageBands.push(`${type}${bits}`);
+  }
+
+  return imageBands;
+}