dicom-curate 0.40.2 → 0.40.3

package/dist/esm/applyMappingsWorker.js

@@ -86462,6 +86462,9 @@ var md5 = /* @__PURE__ */ createHasher(() => new _MD5());
 // src/hash.ts
 var import_js_crc = __toESM(require_crc(), 1);
 var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+function phiSafeToken(input) {
+  return bytesToHex(sha256(new TextEncoder().encode(input))).slice(0, 16);
+}
 async function hashStream(stream, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
@@ -86830,7 +86833,7 @@ async function curateOne({
         // fileInfo, keeping it in the private log. See #283.
         anomalies: [`Could not parse file as DICOM data`],
         errors: [`File is not a valid DICOM file or is corrupted`],
-        sourceInstanceUID: `invalid_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, "_")}`,
+        sourceInstanceUID: `invalid_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
         fileInfo: {
           name: fileInfo.name,
           size: fileInfo.size,

package/dist/esm/curateOne.js

@@ -80171,6 +80171,9 @@ var md5 = /* @__PURE__ */ createHasher(() => new _MD5());
 // src/hash.ts
 var import_js_crc = __toESM(require_crc(), 1);
 var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+function phiSafeToken(input) {
+  return bytesToHex(sha256(new TextEncoder().encode(input))).slice(0, 16);
+}
 async function hashStream(stream, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
@@ -80539,7 +80542,7 @@ async function curateOne({
         // fileInfo, keeping it in the private log. See #283.
         anomalies: [`Could not parse file as DICOM data`],
         errors: [`File is not a valid DICOM file or is corrupted`],
-        sourceInstanceUID: `invalid_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, "_")}`,
+        sourceInstanceUID: `invalid_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
         fileInfo: {
           name: fileInfo.name,
           size: fileInfo.size,

package/dist/esm/hash.js

@@ -800,6 +800,9 @@ var sha256 = /* @__PURE__ */ createHasher(
 // src/hash.ts
 var import_js_crc = __toESM(require_crc(), 1);
 var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+function phiSafeToken(input) {
+  return bytesToHex(sha256(new TextEncoder().encode(input))).slice(0, 16);
+}
 async function hash(buffer, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
@@ -975,7 +978,8 @@ function crc64Hex(input) {
 }
 export {
   hash,
-  hashStream
+  hashStream,
+  phiSafeToken
 };
 /*! Bundled license information:
 

package/dist/esm/index.js

@@ -88025,6 +88025,9 @@ var md5 = /* @__PURE__ */ createHasher(() => new _MD5());
 // src/hash.ts
 var import_js_crc = __toESM(require_crc(), 1);
 var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+function phiSafeToken(input) {
+  return bytesToHex(sha256(new TextEncoder().encode(input))).slice(0, 16);
+}
 async function hash(buffer, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
@@ -88433,7 +88436,7 @@ async function curateOne({
         // fileInfo, keeping it in the private log. See #283.
         anomalies: [`Could not parse file as DICOM data`],
         errors: [`File is not a valid DICOM file or is corrupted`],
-        sourceInstanceUID: `invalid_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, "_")}`,
+        sourceInstanceUID: `invalid_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
         fileInfo: {
           name: fileInfo.name,
           size: fileInfo.size,
@@ -94710,7 +94713,9 @@ async function dispatchMappingJobs() {
       const hasReadErrors = !!errors && errors.length > 0;
       if (hasReadErrors) {
         const scanErrorResult = {
-          sourceInstanceUID: `scan_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, "_")}`,
+          // phiSafeToken rather than the sanitised name: this result reaches
+          // the server-bound log, so the UID must not encode the filename.
+          sourceInstanceUID: `scan_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
           // Intentionally no outputFilePath: an unread file has no
           // de-identified output path, and we must not place the raw input
           // path in the server-bound channel.

package/dist/esm/mappingWorkerPool.js

@@ -86462,6 +86462,9 @@ var md5 = /* @__PURE__ */ createHasher(() => new _MD5());
 // src/hash.ts
 var import_js_crc = __toESM(require_crc(), 1);
 var DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024;
+function phiSafeToken(input) {
+  return bytesToHex(sha256(new TextEncoder().encode(input))).slice(0, 16);
+}
 async function hashStream(stream, hashMethod, hashPartSize) {
   switch (hashMethod) {
     case "sha256":
@@ -86830,7 +86833,7 @@ async function curateOne({
         // fileInfo, keeping it in the private log. See #283.
         anomalies: [`Could not parse file as DICOM data`],
         errors: [`File is not a valid DICOM file or is corrupted`],
-        sourceInstanceUID: `invalid_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, "_")}`,
+        sourceInstanceUID: `invalid_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
         fileInfo: {
           name: fileInfo.name,
           size: fileInfo.size,
@@ -93110,7 +93113,9 @@ async function dispatchMappingJobs() {
       const hasReadErrors = !!errors && errors.length > 0;
       if (hasReadErrors) {
         const scanErrorResult = {
-          sourceInstanceUID: `scan_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, "_")}`,
+          // phiSafeToken rather than the sanitised name: this result reaches
+          // the server-bound log, so the UID must not encode the filename.
+          sourceInstanceUID: `scan_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
           // Intentionally no outputFilePath: an unread file has no
           // de-identified output path, and we must not place the raw input
           // path in the server-bound channel.

package/dist/esm/scanDirectoryWorker.js

@@ -37137,7 +37137,15 @@ var DEFAULT_EXCLUDED_FILETYPES = [
   ".ds_store"
 ];
 function safeReadErrorMessage(error2) {
-  const detail = error2 instanceof Error ? `${error2.name}: ${error2.message}` : String(error2);
+  let detail = "unknown error";
+  if (error2 && typeof error2 === "object") {
+    const { code, name } = error2;
+    if (typeof code === "string" && code) {
+      detail = code;
+    } else if (typeof name === "string" && name) {
+      detail = name;
+    }
+  }
   return `Unable to read file (filesystem error): ${detail}`;
 }
 var keepScanning = true;

package/dist/types/hash.d.ts

@@ -1,4 +1,12 @@
 import type { THashMethod } from './types';
+/**
+ * Derive a short, stable, PHI-safe token from a string (e.g. a file path).
+ * Used to build synthetic sourceInstanceUIDs for files that failed to read or
+ * parse, where embedding the (sanitised) filename would leak PHI into the
+ * server-bound log. Same input always yields the same token, so results stay
+ * correlatable across passes and runs.
+ */
+export declare function phiSafeToken(input: string): string;
 export declare function hash(buffer: ArrayBuffer, hashMethod: THashMethod, hashPartSize?: number): Promise<string>;
 /**
  * Computes a hash over a stream of data chunks without materializing the full

package/dist/umd/dicom-curate.umd.js

@@ -54699,6 +54699,16 @@
       var crcExports = requireCrc();
 
       const DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024; // 5 MB — matches @aws-sdk/lib-storage default
+      /**
+       * Derive a short, stable, PHI-safe token from a string (e.g. a file path).
+       * Used to build synthetic sourceInstanceUIDs for files that failed to read or
+       * parse, where embedding the (sanitised) filename would leak PHI into the
+       * server-bound log. Same input always yields the same token, so results stay
+       * correlatable across passes and runs.
+       */
+      function phiSafeToken(input) {
+          return bytesToHex(sha256(new TextEncoder().encode(input))).slice(0, 16);
+      }
       /**
        * Computes a hash over a stream of data chunks without materializing the full
        * data into memory at once.
@@ -55137,7 +55147,7 @@
                       // fileInfo, keeping it in the private log. See #283.
                       anomalies: [`Could not parse file as DICOM data`],
                       errors: [`File is not a valid DICOM file or is corrupted`],
-                      sourceInstanceUID: `invalid_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, '_')}`,
+                      sourceInstanceUID: `invalid_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
                       fileInfo: {
                           name: fileInfo.name,
                           size: fileInfo.size,
@@ -68864,13 +68874,24 @@
         ];
         /**
          * Build a PHI-safe error message for a file that could not be read during
-         * scanning. Intentionally omits the filename/path so the string is safe to
-         * place in `errors` (which is shared between the private and server-bound
-         * logs). The raw path/name is carried separately in `fileInfo` so it appears
-         * only in the private (input) log.
+         * scanning. Only the error code/name is included — never `error.message`,
+         * because node fs errors embed the full raw path in the message (e.g.
+         * "ENOENT: no such file or directory, stat '/path/to/file.dcm'") and this
+         * string goes into `errors`, which is shared between the private and
+         * server-bound logs. The raw path/name is carried separately in `fileInfo`
+         * so it appears only in the private (input) log.
          */
         function safeReadErrorMessage(error) {
-            const detail = error instanceof Error ? `${error.name}: ${error.message}` : String(error);
+            let detail = 'unknown error';
+            if (error && typeof error === 'object') {
+                const { code, name } = error;
+                if (typeof code === 'string' && code) {
+                    detail = code;
+                }
+                else if (typeof name === 'string' && name) {
+                    detail = name;
+                }
+            }
             return `Unable to read file (filesystem error): ${detail}`;
         }
         let keepScanning = true;
@@ -126358,6 +126379,16 @@
     var crcExports = requireCrc();
 
     const DEFAULT_HASH_PART_SIZE = 5 * 1024 * 1024; // 5 MB — matches @aws-sdk/lib-storage default
+    /**
+     * Derive a short, stable, PHI-safe token from a string (e.g. a file path).
+     * Used to build synthetic sourceInstanceUIDs for files that failed to read or
+     * parse, where embedding the (sanitised) filename would leak PHI into the
+     * server-bound log. Same input always yields the same token, so results stay
+     * correlatable across passes and runs.
+     */
+    function phiSafeToken(input) {
+        return bytesToHex(sha256(new TextEncoder().encode(input))).slice(0, 16);
+    }
     async function hash(buffer, hashMethod, hashPartSize) {
         switch (hashMethod) {
             case 'sha256':
@@ -126857,7 +126888,7 @@
                     // fileInfo, keeping it in the private log. See #283.
                     anomalies: [`Could not parse file as DICOM data`],
                     errors: [`File is not a valid DICOM file or is corrupted`],
-                    sourceInstanceUID: `invalid_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, '_')}`,
+                    sourceInstanceUID: `invalid_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
                     fileInfo: {
                         name: fileInfo.name,
                         size: fileInfo.size,
@@ -140883,7 +140914,9 @@
                 const hasReadErrors = !!errors && errors.length > 0;
                 if (hasReadErrors) {
                     const scanErrorResult = {
-                        sourceInstanceUID: `scan_${fileInfo.name.replace(/[^a-zA-Z0-9]/g, '_')}`,
+                        // phiSafeToken rather than the sanitised name: this result reaches
+                        // the server-bound log, so the UID must not encode the filename.
+                        sourceInstanceUID: `scan_${phiSafeToken(`${fileInfo.path}/${fileInfo.name}`)}`,
                         // Intentionally no outputFilePath: an unread file has no
                         // de-identified output path, and we must not place the raw input
                         // path in the server-bound channel.