fix(config): do not use "rgb" or projection codes for imagery names (#…

…2908)

#### Description

Most of LINZs imagery is currently stored in S3 with the format
`'s3://linz-imagery/:rgion/:name/:projection/:imageryType/` this
structure means basmeaps guesses the imagery name as `:imageryType` so
all imagery is called "rgb"

This makes a ignore list of common folder structures LINZ uses which do
not make sense as imagery names, eg "rgb", "dem_1m" or "3857"

Instead it traverses up the imagery path to see if there is a better
name,

before this change a imagery path of
`'s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/2193/rgb/`
would be called "rgb" after this change it is called
"auckland_sn5600_1979_0.375m"


#### Checklist
*If not applicable, provide explanation of why.*
- [x] Tests updated
- [ ] Docs updated
- [ ] Issue linked in Title

packages/config/src/json/__tests__/config.loader.test.ts

@@ -3,7 +3,7 @@ import { FsMemory, SourceMemory } from '@chunkd/source-memory';
 import { fileURLToPath } from 'node:url';
 import o from 'ospec';
 import { ConfigProviderMemory } from '../../memory/memory.config.js';
-import { initConfigFromUrls } from '../tiff.config.js';
+import { getImageryName, initConfigFromUrls } from '../tiff.config.js';
 
 const simpleTiff = new URL('../../../../__tests__/static/rgba8_tiled.tiff', import.meta.url);
 
@@ -57,4 +57,39 @@ o.spec('config import', () => {
     o(ret.tileSet.layers[1][3857]).equals(ret.imagery[1].id);
     o(ret.tileSet.layers[1].name).equals(ret.imagery[1].name);
   });
+
+  o('should load tiff from filesystem with projection and imagery type in name', async () => {
+    const buf = await fsa.read(fileURLToPath(simpleTiff));
+    await fsa.write('memory://tiffs/tile-tiff-name/2193/rgb/tiff-a.tiff', buf);
+
+    const cfg = new ConfigProviderMemory();
+    const ret = await initConfigFromUrls(cfg, [new URL('memory://tiffs/tile-tiff-name/2193/rgb/')]);
+
+    o(ret.imagery.length).equals(1);
+    const imagery = ret.imagery[0];
+    o(imagery.name).equals('tile-tiff-name');
+    o(imagery.files).deepEquals([{ name: 'tiff-a.tiff', x: 0, y: -64, width: 64, height: 64 }]);
+  });
+
+  o('should ignore 2193 and rgb from imagery names', () => {
+    o(getImageryName(new URL('s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/2193/rgb/'))).equals(
+      'auckland_sn5600_1979_0.375m',
+    );
+    o(getImageryName(new URL('s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/2193/rgbi/'))).equals(
+      'auckland_sn5600_1979_0.375m',
+    );
+    o(getImageryName(new URL('s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/2193/dem_1m/'))).equals(
+      'auckland_sn5600_1979_0.375m',
+    );
+    o(getImageryName(new URL('s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/2193/dsm_1m/'))).equals(
+      'auckland_sn5600_1979_0.375m',
+    );
+    o(getImageryName(new URL('s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/3857/dsm_1m/'))).equals(
+      'auckland_sn5600_1979_0.375m',
+    );
+
+    o(getImageryName(new URL('s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/3857/DSM_1m/'))).equals(
+      'auckland_sn5600_1979_0.375m',
+    );
+  });
 });

packages/config/src/json/tiff.config.ts

@@ -120,6 +120,54 @@ async function loadStacFromURL(target: URL): Promise<StacCollection | null> {
   }
 }
 
+/** When attempting to guess a folder name, try and ignore common imagery types and projection */
+const IgnoredTitles = new Set([
+  // Imagery
+  'rgb',
+  'rgbi',
+  // Elevation
+  'dem_1m',
+  'dsm_1m',
+  //Projections
+  '2193',
+  '3857',
+]);
+
+/**
+ * Guess a better imagery name from a target URL
+ *
+ * A lot of our source paths include the type of imagery eg "rgb", "rgbi" or "dem_1m",
+ * these names are not super helpful and often there are better names further up the pathname
+ *
+ * @example
+ * ```typescript
+ * getImageryName('s3://linz-imagery/auckland/auckland_sn5600_1979_0.375m/2193/rgb/')
+ * // auckland_sn5600_1979_0.375m
+ * ```
+ *
+ * The list of paths ignored are from
+ *
+ * @see {IgnoredTitles}
+ *
+ * For common imagery paths see:
+ *
+ * @see {@link https://github.com/linz/imagery}
+ * @see {@link https://github.com/linz/elevation}
+ */
+export function getImageryName(target: URL): string {
+  const parts = target.pathname.split('/'); // URL folders are always "/"
+
+  for (let i = parts.length - 1; i >= 0; i--) {
+    const part = parts[i];
+    if (part === '') continue;
+    if (IgnoredTitles.has(part.toLowerCase())) continue;
+    return part;
+  }
+
+  // Everything is ignored just use basename
+  return basename(target.pathname);
+}
+
 /**
  * Attempt to load all imagery inside of a path and create a configuration from it
  *
@@ -144,14 +192,14 @@ export async function initImageryFromTiffUrl(
     if (stac == null) log?.warn({ target: targetPath }, 'Tiff:StacNotFound');
     const params = computeTiffSummary(target, tiffs);
 
-    const folderName = basename(targetPath);
-    const title = stac?.title ?? folderName;
+    const imageryName = getImageryName(target);
+    const title = stac?.title ?? imageryName;
     const tileMatrix =
       params.projection === EpsgCode.Nztm2000 ? Nztm2000QuadTms : TileMatrixSets.tryGet(params.projection);
 
     const imagery: ConfigImageryTiff = {
       id: provider.Imagery.id(sha256base58(target.href)),
-      name: folderName,
+      name: imageryName,
       title,
       updatedAt: Date.now(),
       projection: params.projection,
@@ -164,7 +212,7 @@ export async function initImageryFromTiffUrl(
       collection: stac ?? undefined,
     };
     imagery.overviews = await ConfigJson.findImageryOverviews(imagery);
-    log?.info({ title, files: imagery.files.length }, 'Tiff:Loaded');
+    log?.info({ title, imageryName, files: imagery.files.length }, 'Tiff:Loaded');
     provider.put(imagery);
 
     return imagery;
@@ -178,23 +226,23 @@ export async function initImageryFromTiffUrl(
  *
  * Given the following folder structure
  *
- * ```
- * /imagery/invercargill_2022_0.05m/*.tiff
- * /imagery/wellington_2022_0.05/*.tiff
+ * ```typescript
+ * "/imagery/invercargill_2022_0.05m/*.tiff"
+ * "/imagery/wellington_2022_0.05/*.tiff"
  * ```
  *
  * A. A single imagery datasets
  *
- * ```
+ * ```typescript
  * target = ["/imagery/invercargill_2022_0.05m/"]
  * ```
  *
  * will load all tiffs from the resulting folder into a single tile set `aerial`
  *
  * B: A tree of imagery datasets
  *
- * ```
- * target = ["/imagery/invercargill_2022_0.05m/", "/imagery/wellington_2022_0.05/*.tiff"]
+ * ```typescript
+ * target = ["/imagery/invercargill_2022_0.05m/*.tiff", "/imagery/wellington_2022_0.05/*.tiff"]
  * ```
  *
  * will load all tiffs from all folders targets into a single tile set "aerial",