fix(pt): resolve Philips PET private SUV bulkdata before scaling (#6096)

* fix(pt): resolve Philips PET private SUV bulkdata before scaling

When a DICOMweb server delivers the Philips PET private tags SUVScaleFactor
(7053,1000) / ActivityConcentrationScaleFactor (7053,1009) as bulkdata, dcmjs
naturalization leaves them as { BulkDataURI } objects. These were fed verbatim
to calculate-suv, which treats the object as a valid value and silently
corrupts the SUV scaling factors.

Resolve these scalar private tags to numbers during ingestion - in both the
lazy (async) and non-lazy (sync) DICOMweb metadata paths, before INSTANCES_ADDED
fires - by decoding the bulkdata (VR-aware: DS/IS text or little-endian FL/FD).
Harden getPTImageIdInstanceMetadata to coerce values to finite numbers (reusing
@ohif/core utils.toNumber) and reject unresolved bulkdata objects so they can
never reach calculate-suv. Share the bulkdata-attach helper between both
metadata paths. Adds unit tests for the bulkdata decoder/resolver and for
getPTImageIdInstanceMetadata.

* refactor(bulkdata): move PET bulkdata resolution to a generic core tag registry

Generalize resolvePETPrivateScalarBulkData into a datasource-agnostic
utils.resolveBulkDataTags in @ohif/core, backed by a static tag registry
seeded with the Philips PET SUV/activity-concentration scalar tags and
extensible via registerResolvedBulkDataTags.

* fix(bulkdata): strip NUL padding and refresh qido auth before resolution

Address review feedback:
- decodeText now strips NUL (0x00) padding, which String.trim() leaves
  intact, so NUL-padded DS/IS values no longer decode as NaN. Adds a
  regression test.
- refresh qidoDicomWebClient.headers before resolveBulkDataTags in both
  series-metadata paths; retrieveBulkData is bound to qidoDicomWebClient,
  matching every other qido op in this file.

* fix(dicomweb): await deferred metadata storage

* fix(dicomweb): support single-part bulkdata responses

platform/core/src/utils/index.ts

@@ -50,6 +50,12 @@ import areAllImageOrientationsEqual from './areAllImageOrientationsEqual';
 import { structuredCloneWithFunctions } from './structuredCloneWithFunctions';
 import { buildButtonCommands } from './buildButtonCommands';
 import { thumbnailNoImageModalities } from './thumbnailNoImageModalities';
+import {
+  resolveBulkDataTags,
+  registerResolvedBulkDataTags,
+  getResolvedBulkDataTags,
+  decodeNumericBulkData,
+} from './resolveBulkDataTags';
 
 import { downloadBlob, downloadUrl, downloadCsv, downloadDicom } from './downloadBlob';
 
@@ -111,6 +117,10 @@ const utils = {
   downloadUrl,
   downloadCsv,
   downloadDicom,
+  resolveBulkDataTags,
+  registerResolvedBulkDataTags,
+  getResolvedBulkDataTags,
+  decodeNumericBulkData,
 };
 
 export {
@@ -155,6 +165,10 @@ export {
   downloadUrl,
   downloadCsv,
   downloadDicom,
+  resolveBulkDataTags,
+  registerResolvedBulkDataTags,
+  getResolvedBulkDataTags,
+  decodeNumericBulkData,
 };
 
 export default utils;

platform/core/src/utils/resolveBulkDataTags.test.ts

@@ -0,0 +1,201 @@
+import { TextEncoder, TextDecoder } from 'util';
+import {
+  decodeNumericBulkData,
+  resolveBulkDataTags,
+  registerResolvedBulkDataTags,
+  getResolvedBulkDataTags,
+} from './resolveBulkDataTags';
+
+// jsdom does not expose TextEncoder/TextDecoder; the Node util implementations
+// are spec-compatible and match what browsers provide at runtime.
+Object.assign(globalThis, { TextEncoder, TextDecoder });
+
+const textBuffer = (s: string): ArrayBuffer => new TextEncoder().encode(s).buffer;
+const float32Buffer = (n: number): ArrayBuffer => new Float32Array([n]).buffer;
+const float64Buffer = (n: number): ArrayBuffer => new Float64Array([n]).buffer;
+
+describe('decodeNumericBulkData', () => {
+  it('decodes a space-padded DS string (Philips SUVScaleFactor)', () => {
+    // The actual bytes returned by Orthanc for (7053,1000): "0.00038 "
+    expect(decodeNumericBulkData(textBuffer('0.00038 '))).toBeCloseTo(0.00038, 8);
+  });
+
+  it('decodes a plain DS string', () => {
+    expect(decodeNumericBulkData(textBuffer('1.881732'))).toBeCloseTo(1.881732, 6);
+  });
+
+  it('decodes a NUL-padded DS string', () => {
+    // Some servers pad DS/IS values with NUL (0x00) rather than space; trim()
+    // does not strip NUL, so this guards the explicit NUL handling in decodeText.
+    expect(decodeNumericBulkData(textBuffer('0.00038\0'))).toBeCloseTo(0.00038, 8);
+  });
+
+  it('decodes scientific notation', () => {
+    expect(decodeNumericBulkData(textBuffer('3.8e-4'))).toBeCloseTo(0.00038, 8);
+  });
+
+  it('takes the first value of a multi-valued DS string', () => {
+    expect(decodeNumericBulkData(textBuffer('1.5\\2.5'))).toBe(1.5);
+  });
+
+  it('decodes a 4-byte little-endian FL value', () => {
+    expect(decodeNumericBulkData(float32Buffer(0.00038))).toBeCloseTo(0.00038, 7);
+  });
+
+  it('decodes an 8-byte little-endian FD value', () => {
+    expect(decodeNumericBulkData(float64Buffer(0.00038))).toBeCloseTo(0.00038, 12);
+  });
+
+  it('accepts a typed-array view, not just an ArrayBuffer', () => {
+    expect(decodeNumericBulkData(new Uint8Array(textBuffer('2.0')))).toBe(2);
+  });
+
+  it('returns undefined for an empty buffer', () => {
+    expect(decodeNumericBulkData(new ArrayBuffer(0))).toBeUndefined();
+  });
+
+  it('returns undefined for non-numeric text', () => {
+    expect(decodeNumericBulkData(textBuffer('not-a-number'))).toBeUndefined();
+  });
+
+  it('returns undefined for null / undefined / non-buffer input', () => {
+    expect(decodeNumericBulkData(null)).toBeUndefined();
+    expect(decodeNumericBulkData(undefined)).toBeUndefined();
+    expect(decodeNumericBulkData({ BulkDataURI: 'http://x' })).toBeUndefined();
+  });
+});
+
+describe('resolveBulkDataTags', () => {
+  const SUV_TAG = '70531000';
+  const AC_TAG = '70531009';
+
+  it('registers the Philips PET scalar tags by default', () => {
+    expect(getResolvedBulkDataTags()).toEqual(expect.arrayContaining([SUV_TAG, AC_TAG]));
+  });
+
+  it('resolves a Philips bulkdata tag to a number via retrieveBulkData', async () => {
+    const instance: Record<string, unknown> = {
+      Modality: 'PT',
+      [SUV_TAG]: {
+        BulkDataURI: 'http://x/bulk/70531000',
+        retrieveBulkData: jest.fn().mockResolvedValue(textBuffer('0.00038 ')),
+      },
+    };
+
+    await resolveBulkDataTags([instance]);
+
+    expect(instance[SUV_TAG]).toBeCloseTo(0.00038, 8);
+  });
+
+  it('resolves both Philips scalar tags', async () => {
+    const instance: Record<string, unknown> = {
+      Modality: 'PT',
+      [SUV_TAG]: {
+        BulkDataURI: 'http://x/1',
+        retrieveBulkData: jest.fn().mockResolvedValue(textBuffer('0.00038')),
+      },
+      [AC_TAG]: {
+        BulkDataURI: 'http://x/2',
+        retrieveBulkData: jest.fn().mockResolvedValue(textBuffer('1.881732')),
+      },
+    };
+
+    await resolveBulkDataTags([instance]);
+
+    expect(instance[SUV_TAG]).toBeCloseTo(0.00038, 8);
+    expect(instance[AC_TAG]).toBeCloseTo(1.881732, 6);
+  });
+
+  it('resolves registered tags regardless of modality', async () => {
+    const instance: Record<string, unknown> = {
+      Modality: 'CT',
+      [SUV_TAG]: {
+        BulkDataURI: 'http://x',
+        retrieveBulkData: jest.fn().mockResolvedValue(textBuffer('0.5')),
+      },
+    };
+
+    await resolveBulkDataTags([instance]);
+
+    expect(instance[SUV_TAG]).toBe(0.5);
+  });
+
+  it('resolves additionally registered tags', async () => {
+    const CUSTOM_TAG = '00091001';
+    registerResolvedBulkDataTags(CUSTOM_TAG);
+    expect(getResolvedBulkDataTags()).toContain(CUSTOM_TAG);
+
+    const instance: Record<string, unknown> = {
+      [CUSTOM_TAG]: {
+        BulkDataURI: 'http://x/custom',
+        retrieveBulkData: jest.fn().mockResolvedValue(textBuffer('42.5')),
+      },
+    };
+
+    await resolveBulkDataTags([instance]);
+
+    expect(instance[CUSTOM_TAG]).toBe(42.5);
+  });
+
+  it('registers arrays of tags and normalizes casing', async () => {
+    registerResolvedBulkDataTags(['0019100a']);
+    expect(getResolvedBulkDataTags()).toContain('0019100A');
+
+    // The naturalized dataset may key the tag in either casing.
+    const instance: Record<string, unknown> = {
+      '0019100a': {
+        BulkDataURI: 'http://x',
+        retrieveBulkData: jest.fn().mockResolvedValue(textBuffer('7')),
+      },
+    };
+
+    await resolveBulkDataTags([instance]);
+
+    expect(instance['0019100a']).toBe(7);
+  });
+
+  it('uses the cached value.Value without fetching again', async () => {
+    const retrieveBulkData = jest.fn();
+    const instance: Record<string, unknown> = {
+      [SUV_TAG]: { BulkDataURI: 'http://x', Value: textBuffer('0.5'), retrieveBulkData },
+    };
+
+    await resolveBulkDataTags([instance]);
+
+    expect(instance[SUV_TAG]).toBe(0.5);
+    expect(retrieveBulkData).not.toHaveBeenCalled();
+  });
+
+  it('leaves an already-numeric value untouched', async () => {
+    const instance: Record<string, unknown> = { [SUV_TAG]: 0.00038 };
+    await resolveBulkDataTags([instance]);
+    expect(instance[SUV_TAG]).toBe(0.00038);
+  });
+
+  it('leaves the value untouched when bulkdata cannot be fetched (no retrieveBulkData)', async () => {
+    const value = { BulkDataURI: 'http://x' };
+    const instance: Record<string, unknown> = { [SUV_TAG]: value };
+
+    await resolveBulkDataTags([instance]);
+
+    expect(instance[SUV_TAG]).toBe(value);
+  });
+
+  it('does not throw and leaves the value when retrieveBulkData rejects', async () => {
+    const warn = jest.spyOn(console, 'warn').mockImplementation(() => undefined);
+    const value = {
+      BulkDataURI: 'http://x',
+      retrieveBulkData: jest.fn().mockRejectedValue(new Error('request failed')),
+    };
+    const instance: Record<string, unknown> = { [SUV_TAG]: value };
+
+    await expect(resolveBulkDataTags([instance])).resolves.toBeUndefined();
+    expect(instance[SUV_TAG]).toBe(value);
+    warn.mockRestore();
+  });
+
+  it('is a no-op for empty / non-array input', async () => {
+    await expect(resolveBulkDataTags([])).resolves.toBeUndefined();
+    await expect(resolveBulkDataTags(undefined as unknown as unknown[])).resolves.toBeUndefined();
+  });
+});

platform/core/src/utils/resolveBulkDataTags.ts

@@ -0,0 +1,194 @@
+/**
+ * A static registry of DICOM tags whose values must be resolved from bulkdata
+ * into plain numbers during metadata ingestion, before INSTANCES_ADDED fires.
+ *
+ * Background: some servers return small scalar values - notably the Philips
+ * SUV Scale Factor (7053,1000) and Activity Concentration Scale Factor
+ * (7053,1009) - as bulkdata rather than inline. dcmjs `naturalizeDataset` then
+ * leaves them as `{ BulkDataURI: '...' }` objects under their raw hex key.
+ * Downstream consumers (e.g. SUV scaling via calculate-suv) expect numbers and
+ * silently corrupt when handed an object, so data sources resolve the
+ * registered tags eagerly so that every subscriber reads a fully-resolved
+ * number.
+ *
+ * The registry is intentionally not tied to any data source: it is a static
+ * list that any data source can consume via `resolveBulkDataTags`, and that
+ * extensions can extend via `registerResolvedBulkDataTags`.
+ *
+ * Resolution reuses the `retrieveBulkData` method that data sources bind onto
+ * each bulkdata value; when it is absent the value is left untouched and
+ * consumers fall back gracefully.
+ */
+
+// Tags that may arrive as bulkdata and must be resolved to numbers, keyed by
+// the naturalized (comma-less) hex tag. Seeded with the Philips PET Private
+// Group scalar tags; both are VR DS in the standard Philips definition, but
+// the VR is auto-detected (see decode below) because servers occasionally
+// encode them as FL/FD.
+const resolvedBulkDataTags = new Set<string>([
+  '70531000', // Philips SUV Scale Factor
+  '70531009', // Philips Activity Concentration Scale Factor
+]);
+
+/**
+ * Registers additional tags (naturalized comma-less hex form, e.g.
+ * '70531000') to be resolved from bulkdata during metadata ingestion.
+ */
+export function registerResolvedBulkDataTags(tags: string | string[]): void {
+  const list = Array.isArray(tags) ? tags : [tags];
+  for (const tag of list) {
+    if (typeof tag === 'string' && tag.length) {
+      resolvedBulkDataTags.add(tag.toUpperCase());
+    }
+  }
+}
+
+/**
+ * Returns the tags currently registered for eager bulkdata resolution.
+ */
+export function getResolvedBulkDataTags(): string[] {
+  return [...resolvedBulkDataTags];
+}
+
+function toUint8(raw: unknown): Uint8Array | undefined {
+  // Check views first: ArrayBuffer.isView is realm-agnostic.
+  if (ArrayBuffer.isView(raw)) {
+    const view = raw as ArrayBufferView;
+    return new Uint8Array(view.buffer, view.byteOffset, view.byteLength);
+  }
+  // `instanceof ArrayBuffer` is realm-specific, so fall back to a tag check so
+  // that buffers created in another realm (workers, tests) are still handled.
+  if (
+    raw instanceof ArrayBuffer ||
+    Object.prototype.toString.call(raw) === '[object ArrayBuffer]'
+  ) {
+    return new Uint8Array(raw as ArrayBuffer);
+  }
+  return undefined;
+}
+
+// A DS/IS value is printable ASCII (digits, sign, decimal point, exponent,
+// backslash separator, spaces, null padding); raw FL/FD bytes generally are
+// not. This lets us auto-detect the encoding without trusting a VR field, which
+// dcmjs drops when a value is delivered as bulkdata.
+function isPrintableNumeric(bytes: Uint8Array): boolean {
+  if (bytes.length === 0) {
+    return false;
+  }
+  for (let i = 0; i < bytes.length; i++) {
+    const b = bytes[i];
+    const isDigit = b >= 0x30 && b <= 0x39; // 0-9
+    const isAllowed =
+      b === 0x2b || // +
+      b === 0x2d || // -
+      b === 0x2e || // .
+      b === 0x45 || // E
+      b === 0x65 || // e
+      b === 0x5c || // backslash (multi-value separator)
+      b === 0x20 || // space (padding)
+      b === 0x00; // null (padding)
+    if (!isDigit && !isAllowed) {
+      return false;
+    }
+  }
+  return true;
+}
+
+function decodeText(bytes: Uint8Array): number | undefined {
+  // `String.trim()` strips ASCII/Unicode whitespace but NOT NUL (0x00), which
+  // `isPrintableNumeric` treats as valid padding, so strip NULs explicitly -
+  // otherwise a NUL-padded value like "0.00038\0" survives as NaN.
+  const text = new TextDecoder().decode(bytes).replace(/\0+/g, '').trim();
+  // DS/IS may be multi-valued (backslash-delimited); take the first value.
+  const first = text.split('\\')[0].trim();
+  if (!first) {
+    return undefined;
+  }
+  const n = Number(first);
+  return Number.isFinite(n) ? n : undefined;
+}
+
+function decodeBinaryFloat(bytes: Uint8Array): number | undefined {
+  const dv = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);
+  let n: number | undefined;
+  if (bytes.byteLength === 4) {
+    n = dv.getFloat32(0, /* littleEndian */ true);
+  } else if (bytes.byteLength === 8) {
+    n = dv.getFloat64(0, /* littleEndian */ true);
+  }
+  return n !== undefined && Number.isFinite(n) ? n : undefined;
+}
+
+/**
+ * Decodes a bulkdata buffer into a single number. The VR is not available on a
+ * naturalized bulkdata value (dcmjs drops it), so the encoding is auto-detected:
+ * printable-ASCII payloads are decoded as DS/IS text, otherwise the bytes are
+ * read as little-endian IEEE-754 (FL = 4 bytes, FD = 8 bytes).
+ */
+export function decodeNumericBulkData(raw: unknown): number | undefined {
+  const bytes = toUint8(raw);
+  if (!bytes || bytes.byteLength === 0) {
+    return undefined;
+  }
+  if (isPrintableNumeric(bytes)) {
+    return decodeText(bytes);
+  }
+  return decodeBinaryFloat(bytes) ?? decodeText(bytes);
+}
+
+async function resolveValueToNumber(value): Promise<number | undefined> {
+  // retrieveBulkData caches the resolved buffer on value.Value, so prefer it.
+  let buffer = value.Value;
+  if (buffer == null && typeof value.retrieveBulkData === 'function') {
+    buffer = await value.retrieveBulkData();
+  }
+  if (buffer == null) {
+    return undefined;
+  }
+  return decodeNumericBulkData(buffer);
+}
+
+/**
+ * Resolves, in place, the registered bulkdata tags on a single naturalized
+ * instance. No-op for values that are already numbers or that have no
+ * resolvable bulkdata.
+ */
+async function resolveInstance(instance): Promise<void> {
+  if (!instance) {
+    return;
+  }
+
+  await Promise.all(
+    [...resolvedBulkDataTags].map(async tag => {
+      // Registered tags are normalized to uppercase hex; naturalized datasets
+      // key unknown private tags by their hex tag, so check both casings.
+      const key = tag in instance ? tag : tag.toLowerCase();
+      const value = instance[key];
+      // Inline (already a number) or absent: nothing to resolve.
+      if (value == null || typeof value !== 'object') {
+        return;
+      }
+      try {
+        const num = await resolveValueToNumber(value);
+        if (num !== undefined) {
+          instance[key] = num;
+        }
+      } catch (error) {
+        console.warn(`resolveBulkDataTags: failed to resolve tag ${tag}`, error);
+      }
+    })
+  );
+}
+
+/**
+ * Resolves the registered bulkdata tags across a set of naturalized
+ * instances, mutating them in place.
+ */
+export async function resolveBulkDataTags(instances): Promise<void> {
+  if (!Array.isArray(instances) || !instances.length) {
+    return;
+  }
+  await Promise.all(instances.map(resolveInstance));
+}
+
+export default resolveBulkDataTags;