dicom-curate 0.40.7 → 0.41.0

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

@@ -53787,7 +53787,7 @@
 
       const FILEBASENAME = Symbol('fileBasename');
       const FILENAME = Symbol('filename');
-      function getParser(inputPathPattern, inputFilePath, naturalData, dicomPS315EOptions, columnMappings, additionalData) {
+      function getParser(inputPathPattern, inputFilePath, naturalData, dicomPS315EOptions, columnMappings, additionalData, naturalMetaData) {
           function protectUid$1(uid) {
               let protectedUid = uid;
               if (dicomPS315EOptions !== 'Off') {
@@ -53847,6 +53847,24 @@
                           return getCsvMapping(columnMappings, mapping, key, value);
                       };
                   })();
+          // Reads the file meta information group (group 0002), which getDicom cannot
+          // see because it operates on the dataset alone. Needed to identify a file by
+          // MediaStorageSOPClassUID (e.g. for a DICOMDIR disguised as a .DCM).
+          //
+          // Values are always the ORIGINAL ones, even in postExclude where getDicom has
+          // already switched to de-identified values — the meta group is captured once
+          // and never re-read. Only string values are returned; a non-string meta value
+          // (e.g. the OB FileMetaInformationVersion) reads as undefined.
+          function getMetaDicom(attrName) {
+              if (!naturalMetaData)
+                  return undefined;
+              if (attrName in data$1.DicomMetaDictionary.dictionary) {
+                  // if in hex like "(0002,0002)", convert to text key
+                  attrName = data$1.DicomMetaDictionary.dictionary[attrName].name;
+              }
+              const value = lodashExports.get(naturalMetaData, attrName);
+              return typeof value === 'string' ? value : undefined;
+          }
           function missingDicom(attrName) {
               const value = getDicom(attrName);
               return typeof value === 'undefined' || value === '';
@@ -53856,6 +53874,7 @@
               getFilePathComp,
               getMapping,
               getDicom,
+              getMetaDicom,
               missingDicom,
               protectUid: protectUid$1,
               // TODO: Phase this out in favor of ISO8601 duration handling.
@@ -53902,15 +53921,31 @@
           }
           // Make make the naturalized data so parser code operates on with tags not hex
           const naturalData = data$1.DicomMetaDictionary.naturalizeDataset(dicomData.dict);
+          // File meta group (0002), kept separate from the dataset so preExclude can
+          // reach MediaStorageSOPClassUID. Absent on a file that was not read from
+          // Part 10 bytes, so guard rather than assume.
+          const naturalMetaData = dicomData.meta
+              ? data$1.DicomMetaDictionary.naturalizeDataset(dicomData.meta)
+              : undefined;
           mapResults.sourceInstanceUID = naturalData.SOPInstanceUID;
           const finalSpec = composeSpecs(mappingOptions.curationSpec());
           // create a parser object to be used in the eval'ed mappingFunctions
-          const parser = getParser(finalSpec.inputPathPattern, inputFilePath, naturalData, finalSpec.dicomPS315EOptions, mappingOptions.columnMappings, finalSpec.additionalData);
-          // preExclude: original DICOM tags visible via parser.getDicom
+          const parser = getParser(finalSpec.inputPathPattern, inputFilePath, naturalData, finalSpec.dicomPS315EOptions, mappingOptions.columnMappings, finalSpec.additionalData, naturalMetaData);
+          // preExclude: original DICOM tags visible via parser.getDicom / parser.getMetaDicom
           let preExcludeError;
           try {
               if (finalSpec.preExclude?.(parser)) {
                   mapResults.excluded = 'pre';
+                  // Write pass only: collectMappings runs on both the form-generation pass
+                  // (skipWrite) and the write pass, so emitting on both would double-count
+                  // the file for anomaly consumers. Matches the benign scan-exclusion
+                  // anomalies, which are likewise write-pass only.
+                  //
+                  // Name only, never the full path: anomalies are shared with the
+                  // server-bound log, and the raw path is carried in fileInfo instead.
+                  if (!mappingOptions.skipWrite) {
+                      mapResults.anomalies.push(`Skipped pre-excluded file: ${parser.getFilePathComp(parser.FILENAME)}`);
+                  }
                   return [naturalData, mapResults];
               }
           }
@@ -54002,13 +54037,18 @@
                   mapResults.outputFilePath = parts.join('/');
               }
           }
-          // postExclude: output path is finalised; parser.getDicom() returns de-identified values
+          // postExclude: output path is finalised; parser.getDicom() returns de-identified
+          // values — but parser.getMetaDicom() still returns the original meta group.
           try {
               if (finalSpec.postExclude?.({
                   ...parser,
                   outputFilePath: mapResults.outputFilePath,
               })) {
                   mapResults.excluded = 'post';
+                  // Write pass only — see the preExclude note above.
+                  if (!mappingOptions.skipWrite) {
+                      mapResults.anomalies.push(`Skipped post-excluded file: ${parser.getFilePathComp(parser.FILENAME)}`);
+                  }
                   return [naturalData, mapResults];
               }
           }
@@ -68839,7 +68879,7 @@
     /* eslint-enable */
 
     var WorkerFactory = /*#__PURE__*/createInlineWorkerFactory(/* rollup-plugin-web-worker-loader */function () {
-    (function () {
+    ((function (exports) {
         '__worker_loader_strict__';
 
         var _documentCurrentScript = typeof document !== 'undefined' ? document.currentScript : null;
@@ -68900,6 +68940,23 @@
             'thumbs.db',
             '.ds_store',
         ];
+        /**
+         * Name-based exclusion decision for an S3 object key.
+         *
+         * Exported for testing: the S3 listing path needs a live bucket, and the worker
+         * runs in its own process where the S3 client cannot be mocked.
+         *
+         * Unlike the filesystem paths, an S3 key carries its whole prefix while the
+         * exclusion list holds bare filenames — and the defaults apply here too.
+         */
+        function isS3KeyExcludedByName(key, extraExcludedFiletypes, includeDefaults) {
+            const objectName = key.slice(key.lastIndexOf('/') + 1);
+            const allExcludedFiletypes = [
+                ...(includeDefaults ? DEFAULT_EXCLUDED_FILETYPES : []),
+                ...extraExcludedFiletypes,
+            ];
+            return allExcludedFiletypes.some((excluded) => objectName.toLowerCase() === excluded.toLowerCase());
+        }
         /**
          * Build a PHI-safe error message for a file that could not be read during
          * scanning. Only the error code/name is included — never `error.message`,
@@ -69018,7 +69075,7 @@
                     return false;
                 }
                 // Check if the file is in the list of excluded files
-                if (excludedFiletypes.some((excluded) => s3Item.Key.toLowerCase() === excluded.toLowerCase())) {
+                if (isS3KeyExcludedByName(s3Item.Key, excludedFiletypes, !noDefaultExclusions)) {
                     fileAnomalies.push(`Skipped excluded file: ${s3Item.Key}`);
                     return false;
                 }
@@ -69564,7 +69621,11 @@
             __proto__: null
         });
 
-    })();
+        exports.isS3KeyExcludedByName = isS3KeyExcludedByName;
+
+        return exports;
+
+    }))({});
 
     });
     /* eslint-enable */
@@ -125511,7 +125572,7 @@
 
     const FILEBASENAME = Symbol('fileBasename');
     const FILENAME = Symbol('filename');
-    function getParser(inputPathPattern, inputFilePath, naturalData, dicomPS315EOptions, columnMappings, additionalData) {
+    function getParser(inputPathPattern, inputFilePath, naturalData, dicomPS315EOptions, columnMappings, additionalData, naturalMetaData) {
         function protectUid$1(uid) {
             let protectedUid = uid;
             if (dicomPS315EOptions !== 'Off') {
@@ -125571,6 +125632,24 @@
                         return getCsvMapping(columnMappings, mapping, key, value);
                     };
                 })();
+        // Reads the file meta information group (group 0002), which getDicom cannot
+        // see because it operates on the dataset alone. Needed to identify a file by
+        // MediaStorageSOPClassUID (e.g. for a DICOMDIR disguised as a .DCM).
+        //
+        // Values are always the ORIGINAL ones, even in postExclude where getDicom has
+        // already switched to de-identified values — the meta group is captured once
+        // and never re-read. Only string values are returned; a non-string meta value
+        // (e.g. the OB FileMetaInformationVersion) reads as undefined.
+        function getMetaDicom(attrName) {
+            if (!naturalMetaData)
+                return undefined;
+            if (attrName in data$1.DicomMetaDictionary.dictionary) {
+                // if in hex like "(0002,0002)", convert to text key
+                attrName = data$1.DicomMetaDictionary.dictionary[attrName].name;
+            }
+            const value = lodashExports.get(naturalMetaData, attrName);
+            return typeof value === 'string' ? value : undefined;
+        }
         function missingDicom(attrName) {
             const value = getDicom(attrName);
             return typeof value === 'undefined' || value === '';
@@ -125580,6 +125659,7 @@
             getFilePathComp,
             getMapping,
             getDicom,
+            getMetaDicom,
             missingDicom,
             protectUid: protectUid$1,
             // TODO: Phase this out in favor of ISO8601 duration handling.
@@ -125626,15 +125706,31 @@
         }
         // Make make the naturalized data so parser code operates on with tags not hex
         const naturalData = data$1.DicomMetaDictionary.naturalizeDataset(dicomData.dict);
+        // File meta group (0002), kept separate from the dataset so preExclude can
+        // reach MediaStorageSOPClassUID. Absent on a file that was not read from
+        // Part 10 bytes, so guard rather than assume.
+        const naturalMetaData = dicomData.meta
+            ? data$1.DicomMetaDictionary.naturalizeDataset(dicomData.meta)
+            : undefined;
         mapResults.sourceInstanceUID = naturalData.SOPInstanceUID;
         const finalSpec = composeSpecs(mappingOptions.curationSpec());
         // create a parser object to be used in the eval'ed mappingFunctions
-        const parser = getParser(finalSpec.inputPathPattern, inputFilePath, naturalData, finalSpec.dicomPS315EOptions, mappingOptions.columnMappings, finalSpec.additionalData);
-        // preExclude: original DICOM tags visible via parser.getDicom
+        const parser = getParser(finalSpec.inputPathPattern, inputFilePath, naturalData, finalSpec.dicomPS315EOptions, mappingOptions.columnMappings, finalSpec.additionalData, naturalMetaData);
+        // preExclude: original DICOM tags visible via parser.getDicom / parser.getMetaDicom
         let preExcludeError;
         try {
             if (finalSpec.preExclude?.(parser)) {
                 mapResults.excluded = 'pre';
+                // Write pass only: collectMappings runs on both the form-generation pass
+                // (skipWrite) and the write pass, so emitting on both would double-count
+                // the file for anomaly consumers. Matches the benign scan-exclusion
+                // anomalies, which are likewise write-pass only.
+                //
+                // Name only, never the full path: anomalies are shared with the
+                // server-bound log, and the raw path is carried in fileInfo instead.
+                if (!mappingOptions.skipWrite) {
+                    mapResults.anomalies.push(`Skipped pre-excluded file: ${parser.getFilePathComp(parser.FILENAME)}`);
+                }
                 return [naturalData, mapResults];
             }
         }
@@ -125726,13 +125822,18 @@
                 mapResults.outputFilePath = parts.join('/');
             }
         }
-        // postExclude: output path is finalised; parser.getDicom() returns de-identified values
+        // postExclude: output path is finalised; parser.getDicom() returns de-identified
+        // values — but parser.getMetaDicom() still returns the original meta group.
         try {
             if (finalSpec.postExclude?.({
                 ...parser,
                 outputFilePath: mapResults.outputFilePath,
             })) {
                 mapResults.excluded = 'post';
+                // Write pass only — see the preExclude note above.
+                if (!mappingOptions.skipWrite) {
+                    mapResults.anomalies.push(`Skipped post-excluded file: ${parser.getFilePathComp(parser.FILENAME)}`);
+                }
                 return [naturalData, mapResults];
             }
         }