dicom-curate 0.20.1 → 0.21.0

package/dist/esm/scanDirectoryWorker.js

@@ -37199,6 +37199,45 @@ async function shouldProcessFileItem(s3Item, fileAnomalies) {
     return true;
   }
 }
+async function shouldProcessFileNode(filePath, fileName, fileSize, fileAnomalies) {
+  const allExcludedFiletypes = [
+    ...DEFAULT_EXCLUDED_FILETYPES,
+    ...excludedFiletypes
+  ];
+  try {
+    if (allExcludedFiletypes.some(
+      (excluded) => fileName.toLowerCase() === excluded.toLowerCase()
+    )) {
+      fileAnomalies.push(`Skipped excluded file: ${fileName}`);
+      return false;
+    }
+    if (fileSize < 132) {
+      fileAnomalies.push(
+        `Skipped very small file: ${fileName} (${fileSize} bytes)`
+      );
+      return false;
+    }
+    const fs = await import("fs/promises");
+    const fh = await fs.open(filePath, "r");
+    try {
+      const buffer = Buffer.alloc(4);
+      await fh.read(buffer, 0, 4, 128);
+      const dicomSignature = buffer.toString("ascii");
+      if (dicomSignature === "DICM") {
+        return true;
+      }
+    } finally {
+      await fh.close();
+    }
+    fileAnomalies.push(`Skipped file without DICOM signature: ${fileName}`);
+    return false;
+  } catch (error2) {
+    fileAnomalies.push(
+      `Unable to determine file validity - processing anyway: ${fileName} - ${error2}`
+    );
+    return true;
+  }
+}
 fixupNodeWorkerEnvironment().then(() => {
   globalThis.addEventListener("message", (event) => {
     switch (event.data.request) {
@@ -37376,12 +37415,13 @@ async function scanDirectoryNode(dirPath) {
       if (entry.isFile() && keepScanning) {
         const filePath = path.join(currentPath, entry.name);
         const stats = await fs.stat(filePath);
-        const fileBuffer = await fs.readFile(filePath);
-        const file = new File([new Uint8Array(fileBuffer)], entry.name, {
-          type: "application/dicom"
-        });
         const fileAnomalies = [];
-        if (await shouldProcessFile(file, fileAnomalies)) {
+        if (await shouldProcessFileNode(
+          filePath,
+          entry.name,
+          stats.size,
+          fileAnomalies
+        )) {
           globalThis.postMessage({
             response: "file",
             fileIndex: fileIndex++,

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

@@ -159,6 +159,50 @@
                 return true;
             }
         }
+        /**
+         * Node-specific file validation that reads only the bytes needed
+         * instead of loading the entire file into memory.
+         */
+        async function shouldProcessFileNode(filePath, fileName, fileSize, fileAnomalies) {
+            const allExcludedFiletypes = [
+                ...DEFAULT_EXCLUDED_FILETYPES,
+                ...excludedFiletypes,
+            ];
+            try {
+                // Check if the file is in the list of excluded files
+                if (allExcludedFiletypes.some((excluded) => fileName.toLowerCase() === excluded.toLowerCase())) {
+                    fileAnomalies.push(`Skipped excluded file: ${fileName}`);
+                    return false;
+                }
+                // Check filesize - (valid) DICOM files are at least 132 bytes (128-byte preamble + 4-byte signature)
+                if (fileSize < 132) {
+                    fileAnomalies.push(`Skipped very small file: ${fileName} (${fileSize} bytes)`);
+                    return false;
+                }
+                // Check for DICOM signature "DICM" at offset 128 by reading only 4 bytes
+                const fs = await import('fs/promises');
+                const fh = await fs.open(filePath, 'r');
+                try {
+                    const buffer = Buffer.alloc(4);
+                    await fh.read(buffer, 0, 4, 128);
+                    const dicomSignature = buffer.toString('ascii');
+                    if (dicomSignature === 'DICM') {
+                        return true;
+                    }
+                }
+                finally {
+                    await fh.close();
+                }
+                // Don't parse file without DICOM signature
+                fileAnomalies.push(`Skipped file without DICOM signature: ${fileName}`);
+                return false;
+            }
+            catch (error) {
+                fileAnomalies.push(`Unable to determine file validity - processing anyway: ${fileName} - ${error}`);
+                // If vetting process fails, let the parser decide
+                return true;
+            }
+        }
         fixupNodeWorkerEnvironment().then(() => {
             globalThis.addEventListener('message', (event) => {
                 switch (event.data.request) {
@@ -359,12 +403,8 @@
                     if (entry.isFile() && keepScanning) {
                         const filePath = path.join(currentPath, entry.name);
                         const stats = await fs.stat(filePath);
-                        const fileBuffer = await fs.readFile(filePath);
-                        const file = new File([new Uint8Array(fileBuffer)], entry.name, {
-                            type: 'application/dicom',
-                        });
                         const fileAnomalies = [];
-                        if (await shouldProcessFile(file, fileAnomalies)) {
+                        if (await shouldProcessFileNode(filePath, entry.name, stats.size, fileAnomalies)) {
                             // Send file to processing pipeline
                             globalThis.postMessage({
                                 response: 'file',