dicom-curate 0.27.0 → 0.28.0

package/dist/esm/applyMappingsWorker.js

@@ -79688,17 +79688,20 @@ async function fetchWithRetry(...args) {
 // src/hash.ts
 var import_md5 = __toESM(require_md5(), 1);
 var import_js_crc = __toESM(require_crc(), 1);
-async function hash(buffer, hashMethod) {
+var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+async function hash(buffer, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
       return await sha256Hex(buffer);
     case "crc32":
       return crc32Hex(buffer);
+    case "crc64":
+      return crc64Hex(buffer);
+    case "aws-s3-etag-2025":
+      return awsS3Etag(buffer, hashPartSize ?? DEFAULT_HASH_PART_SIZE);
     case "md5":
     default:
       return md5Hex(buffer);
-    case "crc64":
-      return crc64Hex(buffer);
   }
 }
 async function sha256Hex(buffer) {
@@ -79709,6 +79712,27 @@ async function sha256Hex(buffer) {
 function md5Hex(buffer) {
   return (0, import_md5.default)(new Uint8Array(buffer));
 }
+function awsS3Etag(buffer, partSize) {
+  if (buffer.byteLength <= partSize) {
+    return md5Hex(buffer);
+  }
+  return multipartMd5(buffer, partSize);
+}
+function multipartMd5(buffer, partSize) {
+  const totalSize = buffer.byteLength;
+  const partCount = Math.ceil(totalSize / partSize);
+  const rawDigests = new Uint8Array(partCount * 16);
+  for (let i4 = 0; i4 < partCount; i4++) {
+    const start = i4 * partSize;
+    const end = Math.min(start + partSize, totalSize);
+    const partBuffer = buffer.slice(start, end);
+    const hex = (0, import_md5.default)(new Uint8Array(partBuffer));
+    for (let j4 = 0; j4 < 16; j4++) {
+      rawDigests[i4 * 16 + j4] = parseInt(hex.slice(j4 * 2, j4 * 2 + 2), 16);
+    }
+  }
+  return `${(0, import_md5.default)(rawDigests)}-${partCount}`;
+}
 function crc32Hex(input) {
   let bytes;
   if (input instanceof Uint8Array) {
@@ -79785,6 +79809,7 @@ async function curateOne({
   outputTarget,
   mappingOptions,
   hashMethod,
+  hashPartSize,
   previousSourceFileInfo,
   previousMappedFileInfo
 }) {
@@ -79884,7 +79909,11 @@ async function curateOne({
   let canSkip = false;
   if (previousSourceFileInfo?.preMappedHash !== void 0) {
     try {
-      preMappedHash = await hash(fileArrayBuffer, hashMethod ?? "md5");
+      preMappedHash = await hash(
+        fileArrayBuffer,
+        hashMethod ?? "md5",
+        hashPartSize
+      );
     } catch (e4) {
       console.warn(`Failed to compute preMappedHash for ${fileInfo.name}`, e4);
     }
@@ -79989,7 +80018,11 @@ async function curateOne({
   }
   if (!preMappedHash) {
     try {
-      preMappedHash = await hash(fileArrayBuffer, hashMethod ?? "md5");
+      preMappedHash = await hash(
+        fileArrayBuffer,
+        hashMethod ?? "md5",
+        hashPartSize
+      );
     } catch (e4) {
       console.warn(`Failed to compute preMappedHash for ${fileInfo.name}`, e4);
     }
@@ -80000,7 +80033,11 @@ async function curateOne({
     const modifiedArrayBuffer = mappedDicomData.write({
       allowInvalidVRLength: true
     });
-    postMappedHash = await hash(modifiedArrayBuffer, hashMethod ?? "md5");
+    postMappedHash = await hash(
+      modifiedArrayBuffer,
+      hashMethod ?? "md5",
+      hashPartSize
+    );
     fileArrayBuffer = null;
     const previousPostMappedHash = previousMappedFileInfo ? previousMappedFileInfo(clonedMapResults.outputFilePath)?.postMappedHash : void 0;
     if (previousPostMappedHash !== void 0 && previousPostMappedHash === postMappedHash) {
@@ -85832,6 +85869,7 @@ fixupNodeWorkerEnvironment().then(() => {
               fileInfo,
               outputTarget: event.data.outputTarget ?? {},
               hashMethod: event.data.hashMethod,
+              hashPartSize: event.data.hashPartSize,
               mappingOptions,
               previousSourceFileInfo: event.data.previousFileInfo,
               previousMappedFileInfo: (targetName) => {

package/dist/esm/curateOne.js

@@ -73397,17 +73397,20 @@ async function fetchWithRetry(...args) {
 // src/hash.ts
 var import_md5 = __toESM(require_md5(), 1);
 var import_js_crc = __toESM(require_crc(), 1);
-async function hash(buffer, hashMethod) {
+var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+async function hash(buffer, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
       return await sha256Hex(buffer);
     case "crc32":
       return crc32Hex(buffer);
+    case "crc64":
+      return crc64Hex(buffer);
+    case "aws-s3-etag-2025":
+      return awsS3Etag(buffer, hashPartSize ?? DEFAULT_HASH_PART_SIZE);
     case "md5":
     default:
       return md5Hex(buffer);
-    case "crc64":
-      return crc64Hex(buffer);
   }
 }
 async function sha256Hex(buffer) {
@@ -73418,6 +73421,27 @@ async function sha256Hex(buffer) {
 function md5Hex(buffer) {
   return (0, import_md5.default)(new Uint8Array(buffer));
 }
+function awsS3Etag(buffer, partSize) {
+  if (buffer.byteLength <= partSize) {
+    return md5Hex(buffer);
+  }
+  return multipartMd5(buffer, partSize);
+}
+function multipartMd5(buffer, partSize) {
+  const totalSize = buffer.byteLength;
+  const partCount = Math.ceil(totalSize / partSize);
+  const rawDigests = new Uint8Array(partCount * 16);
+  for (let i4 = 0; i4 < partCount; i4++) {
+    const start = i4 * partSize;
+    const end = Math.min(start + partSize, totalSize);
+    const partBuffer = buffer.slice(start, end);
+    const hex = (0, import_md5.default)(new Uint8Array(partBuffer));
+    for (let j4 = 0; j4 < 16; j4++) {
+      rawDigests[i4 * 16 + j4] = parseInt(hex.slice(j4 * 2, j4 * 2 + 2), 16);
+    }
+  }
+  return `${(0, import_md5.default)(rawDigests)}-${partCount}`;
+}
 function crc32Hex(input) {
   let bytes;
   if (input instanceof Uint8Array) {
@@ -73494,6 +73518,7 @@ async function curateOne({
   outputTarget,
   mappingOptions,
   hashMethod,
+  hashPartSize,
   previousSourceFileInfo,
   previousMappedFileInfo
 }) {
@@ -73593,7 +73618,11 @@ async function curateOne({
   let canSkip = false;
   if (previousSourceFileInfo?.preMappedHash !== void 0) {
     try {
-      preMappedHash = await hash(fileArrayBuffer, hashMethod ?? "md5");
+      preMappedHash = await hash(
+        fileArrayBuffer,
+        hashMethod ?? "md5",
+        hashPartSize
+      );
     } catch (e4) {
       console.warn(`Failed to compute preMappedHash for ${fileInfo.name}`, e4);
     }
@@ -73698,7 +73727,11 @@ async function curateOne({
   }
   if (!preMappedHash) {
     try {
-      preMappedHash = await hash(fileArrayBuffer, hashMethod ?? "md5");
+      preMappedHash = await hash(
+        fileArrayBuffer,
+        hashMethod ?? "md5",
+        hashPartSize
+      );
     } catch (e4) {
       console.warn(`Failed to compute preMappedHash for ${fileInfo.name}`, e4);
     }
@@ -73709,7 +73742,11 @@ async function curateOne({
     const modifiedArrayBuffer = mappedDicomData.write({
       allowInvalidVRLength: true
     });
-    postMappedHash = await hash(modifiedArrayBuffer, hashMethod ?? "md5");
+    postMappedHash = await hash(
+      modifiedArrayBuffer,
+      hashMethod ?? "md5",
+      hashPartSize
+    );
     fileArrayBuffer = null;
     const previousPostMappedHash = previousMappedFileInfo ? previousMappedFileInfo(clonedMapResults.outputFilePath)?.postMappedHash : void 0;
     if (previousPostMappedHash !== void 0 && previousPostMappedHash === postMappedHash) {

package/dist/esm/hash.js

@@ -640,17 +640,20 @@ var require_crc = __commonJS({
 // src/hash.ts
 var import_md5 = __toESM(require_md5(), 1);
 var import_js_crc = __toESM(require_crc(), 1);
-async function hash(buffer, hashMethod) {
+var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+async function hash(buffer, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
       return await sha256Hex(buffer);
     case "crc32":
       return crc32Hex(buffer);
+    case "crc64":
+      return crc64Hex(buffer);
+    case "aws-s3-etag-2025":
+      return awsS3Etag(buffer, hashPartSize ?? DEFAULT_HASH_PART_SIZE);
     case "md5":
     default:
       return md5Hex(buffer);
-    case "crc64":
-      return crc64Hex(buffer);
   }
 }
 async function sha256Hex(buffer) {
@@ -661,6 +664,27 @@ async function sha256Hex(buffer) {
 function md5Hex(buffer) {
   return (0, import_md5.default)(new Uint8Array(buffer));
 }
+function awsS3Etag(buffer, partSize) {
+  if (buffer.byteLength <= partSize) {
+    return md5Hex(buffer);
+  }
+  return multipartMd5(buffer, partSize);
+}
+function multipartMd5(buffer, partSize) {
+  const totalSize = buffer.byteLength;
+  const partCount = Math.ceil(totalSize / partSize);
+  const rawDigests = new Uint8Array(partCount * 16);
+  for (let i = 0; i < partCount; i++) {
+    const start = i * partSize;
+    const end = Math.min(start + partSize, totalSize);
+    const partBuffer = buffer.slice(start, end);
+    const hex = (0, import_md5.default)(new Uint8Array(partBuffer));
+    for (let j = 0; j < 16; j++) {
+      rawDigests[i * 16 + j] = parseInt(hex.slice(j * 2, j * 2 + 2), 16);
+    }
+  }
+  return `${(0, import_md5.default)(rawDigests)}-${partCount}`;
+}
 function crc32Hex(input) {
   let bytes;
   if (input instanceof Uint8Array) {

package/dist/esm/index.js

@@ -81248,17 +81248,20 @@ async function fetchWithRetry(...args) {
 // src/hash.ts
 var import_md5 = __toESM(require_md5(), 1);
 var import_js_crc = __toESM(require_crc(), 1);
-async function hash(buffer, hashMethod) {
+var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+async function hash(buffer, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
       return await sha256Hex(buffer);
     case "crc32":
       return crc32Hex(buffer);
+    case "crc64":
+      return crc64Hex(buffer);
+    case "aws-s3-etag-2025":
+      return awsS3Etag(buffer, hashPartSize ?? DEFAULT_HASH_PART_SIZE);
     case "md5":
     default:
       return md5Hex(buffer);
-    case "crc64":
-      return crc64Hex(buffer);
   }
 }
 async function sha256Hex(buffer) {
@@ -81269,6 +81272,27 @@ async function sha256Hex(buffer) {
 function md5Hex(buffer) {
   return (0, import_md5.default)(new Uint8Array(buffer));
 }
+function awsS3Etag(buffer, partSize) {
+  if (buffer.byteLength <= partSize) {
+    return md5Hex(buffer);
+  }
+  return multipartMd5(buffer, partSize);
+}
+function multipartMd5(buffer, partSize) {
+  const totalSize = buffer.byteLength;
+  const partCount = Math.ceil(totalSize / partSize);
+  const rawDigests = new Uint8Array(partCount * 16);
+  for (let i4 = 0; i4 < partCount; i4++) {
+    const start = i4 * partSize;
+    const end = Math.min(start + partSize, totalSize);
+    const partBuffer = buffer.slice(start, end);
+    const hex = (0, import_md5.default)(new Uint8Array(partBuffer));
+    for (let j4 = 0; j4 < 16; j4++) {
+      rawDigests[i4 * 16 + j4] = parseInt(hex.slice(j4 * 2, j4 * 2 + 2), 16);
+    }
+  }
+  return `${(0, import_md5.default)(rawDigests)}-${partCount}`;
+}
 function crc32Hex(input) {
   let bytes;
   if (input instanceof Uint8Array) {
@@ -81345,6 +81369,7 @@ async function curateOne({
   outputTarget,
   mappingOptions,
   hashMethod,
+  hashPartSize,
   previousSourceFileInfo,
   previousMappedFileInfo
 }) {
@@ -81444,7 +81469,11 @@ async function curateOne({
   let canSkip = false;
   if (previousSourceFileInfo?.preMappedHash !== void 0) {
     try {
-      preMappedHash = await hash(fileArrayBuffer, hashMethod ?? "md5");
+      preMappedHash = await hash(
+        fileArrayBuffer,
+        hashMethod ?? "md5",
+        hashPartSize
+      );
     } catch (e4) {
       console.warn(`Failed to compute preMappedHash for ${fileInfo.name}`, e4);
     }
@@ -81549,7 +81578,11 @@ async function curateOne({
   }
   if (!preMappedHash) {
     try {
-      preMappedHash = await hash(fileArrayBuffer, hashMethod ?? "md5");
+      preMappedHash = await hash(
+        fileArrayBuffer,
+        hashMethod ?? "md5",
+        hashPartSize
+      );
     } catch (e4) {
       console.warn(`Failed to compute preMappedHash for ${fileInfo.name}`, e4);
     }
@@ -81560,7 +81593,11 @@ async function curateOne({
     const modifiedArrayBuffer = mappedDicomData.write({
       allowInvalidVRLength: true
     });
-    postMappedHash = await hash(modifiedArrayBuffer, hashMethod ?? "md5");
+    postMappedHash = await hash(
+      modifiedArrayBuffer,
+      hashMethod ?? "md5",
+      hashPartSize
+    );
     fileArrayBuffer = null;
     const previousPostMappedHash = previousMappedFileInfo ? previousMappedFileInfo(clonedMapResults.outputFilePath)?.postMappedHash : void 0;
     if (previousPostMappedHash !== void 0 && previousPostMappedHash === postMappedHash) {
@@ -87487,7 +87524,7 @@ async function dispatchMappingJobs() {
     const { fileInfo, previousFileInfo } = filesToProcess.pop();
     const mappingWorker = availableMappingWorkers.pop();
     workerCurrentFile.set(mappingWorker, fileInfo);
-    const { outputTarget, hashMethod, ...mappingOptions } = (
+    const { outputTarget, hashMethod, hashPartSize, ...mappingOptions } = (
       // Not partial anymore.
       mappingWorkerOptions
     );
@@ -87497,6 +87534,7 @@ async function dispatchMappingJobs() {
       outputTarget: await getHttpOutputHeaders(outputTarget),
       previousFileInfo,
       hashMethod,
+      hashPartSize,
       serializedMappingOptions: serializeMappingOptions(mappingOptions)
     });
     workersActive += 1;
@@ -87777,6 +87815,7 @@ async function collectMappingOptions(organizeOptions) {
   const skipModifications = organizeOptions.skipModifications ?? false;
   const skipValidation = organizeOptions.skipValidation ?? false;
   const hashMethod = organizeOptions.hashMethod;
+  const hashPartSize = organizeOptions.hashPartSize;
   const dateOffset = organizeOptions.dateOffset;
   if (requiresDateOffset(deIdOpts) && !dateOffset?.match(iso8601)) {
     throw new Error(
@@ -87791,7 +87830,8 @@ async function collectMappingOptions(organizeOptions) {
     skipModifications,
     skipValidation,
     dateOffset,
-    hashMethod
+    hashMethod,
+    hashPartSize
   };
 }
 function queueFilesForMapping(organizeOptions) {

package/dist/esm/mappingWorkerPool.js

@@ -12102,7 +12102,7 @@ async function dispatchMappingJobs() {
     const { fileInfo, previousFileInfo } = filesToProcess.pop();
     const mappingWorker = availableMappingWorkers.pop();
     workerCurrentFile.set(mappingWorker, fileInfo);
-    const { outputTarget, hashMethod, ...mappingOptions } = (
+    const { outputTarget, hashMethod, hashPartSize, ...mappingOptions } = (
       // Not partial anymore.
       mappingWorkerOptions
     );
@@ -12112,6 +12112,7 @@ async function dispatchMappingJobs() {
       outputTarget: await getHttpOutputHeaders(outputTarget),
       previousFileInfo,
       hashMethod,
+      hashPartSize,
       serializedMappingOptions: serializeMappingOptions(mappingOptions)
     });
     workersActive += 1;

package/dist/types/applyMappingsWorker.d.ts

@@ -9,6 +9,7 @@ export type MappingRequest = {
         preMappedHash?: string;
     };
     hashMethod?: THashMethod;
+    hashPartSize?: number;
     serializedMappingOptions: TSerializedMappingOptions;
 } | {
     request: 'fileInfoIndex';

package/dist/types/curateOne.d.ts

@@ -4,6 +4,7 @@ export type TCurateOneArgs = {
     outputTarget: TOutputTarget;
     mappingOptions: TMappingOptions;
     hashMethod?: THashMethod;
+    hashPartSize?: number;
     previousSourceFileInfo?: {
         size?: number;
         mtime?: string;
@@ -13,6 +14,6 @@ export type TCurateOneArgs = {
         postMappedHash?: string;
     } | undefined;
 };
-export declare function curateOne({ fileInfo, outputTarget, mappingOptions, hashMethod, previousSourceFileInfo, previousMappedFileInfo, }: TCurateOneArgs): Promise<Omit<Partial<TMapResults>, 'anomalies'> & {
+export declare function curateOne({ fileInfo, outputTarget, mappingOptions, hashMethod, hashPartSize, previousSourceFileInfo, previousMappedFileInfo, }: TCurateOneArgs): Promise<Omit<Partial<TMapResults>, 'anomalies'> & {
     anomalies: TMapResults['anomalies'];
 }>;

package/dist/types/hash.d.ts

@@ -1,2 +1,2 @@
 import { THashMethod } from './types';
-export declare function hash(buffer: ArrayBuffer, hashMethod: THashMethod): Promise<string>;
+export declare function hash(buffer: ArrayBuffer, hashMethod: THashMethod, hashPartSize?: number): Promise<string>;

package/dist/types/mappingWorkerPool.d.ts

@@ -9,6 +9,7 @@ import type { TMappingOptions, TFileInfo, TProgressMessage, TOutputTarget, TFile
 export type TMappingWorkerOptions = TMappingOptions & {
     outputTarget?: TOutputTarget;
     hashMethod?: THashMethod;
+    hashPartSize?: number;
 };
 export type ProgressCallback = (message: TProgressMessage) => void;
 export declare const availableMappingWorkers: Worker[];

package/dist/types/types.d.ts

@@ -29,6 +29,7 @@ export type OrganizeOptions = {
     dateOffset?: Iso8601Duration;
     skipCollectingMappings?: boolean;
     hashMethod?: THashMethod;
+    hashPartSize?: number;
     fileInfoIndex?: TFileInfoIndex;
     excludedPathGlobs?: string[];
     workerCount?: number;
@@ -49,7 +50,7 @@ export type OrganizeOptions = {
     inputType: 's3';
     inputS3Bucket: TS3BucketOptions;
 });
-export type THashMethod = 'crc64' | 'crc32' | 'sha256' | 'md5';
+export type THashMethod = 'crc64' | 'crc32' | 'sha256' | 'md5' | 'aws-s3-etag-2025';
 export type THTTPHeaderProvider = () => Promise<Record<string, string>> | Record<string, string>;
 export type THTTPOptions = {
     url: string;