@ubercode/dcmtk 0.12.0 → 0.13.0

package/dist/tools.d.cts

@@ -78,6 +78,8 @@ interface Dcm2jsonOptions extends ToolBaseOptions {
     readonly directOnly?: boolean | undefined;
     /** Assume the specified character set when SpecificCharacterSet (0008,0005) is absent. Passed to dcm2xml as `+Ca`. Only effective on the XML path (dcm2json binary does not support this flag). */
     readonly charsetAssume?: string | undefined;
+    /** Fallback charset to retry with when UTF-8 conversion fails. On charset error, the XML path retries with `+Ca <fallback>`. `'Latin1'` recommended — maps every byte to a valid character. */
+    readonly charsetFallback?: string | undefined;
     /** Verbosity level for diagnostic output. `'verbose'` maps to `-v`, `'debug'` maps to `-d`. */
     readonly verbosity?: 'verbose' | 'debug' | undefined;
 }

package/dist/tools.d.ts

@@ -78,6 +78,8 @@ interface Dcm2jsonOptions extends ToolBaseOptions {
     readonly directOnly?: boolean | undefined;
     /** Assume the specified character set when SpecificCharacterSet (0008,0005) is absent. Passed to dcm2xml as `+Ca`. Only effective on the XML path (dcm2json binary does not support this flag). */
     readonly charsetAssume?: string | undefined;
+    /** Fallback charset to retry with when UTF-8 conversion fails. On charset error, the XML path retries with `+Ca <fallback>`. `'Latin1'` recommended — maps every byte to a valid character. */
+    readonly charsetFallback?: string | undefined;
     /** Verbosity level for diagnostic output. `'verbose'` maps to `-v`, `'debug'` maps to `-d`. */
     readonly verbosity?: 'verbose' | 'debug' | undefined;
 }

package/dist/tools.js

@@ -470,7 +470,12 @@ var parser = new XMLParser({
   ignoreAttributes: false,
   attributeNamePrefix: "@_",
   parseTagValue: false,
-  isArray: (name) => ARRAY_TAG_NAMES.has(name)
+  isArray: (name) => ARRAY_TAG_NAMES.has(name),
+  // DICOM XML does not use XML entities. Disable entity processing to avoid
+  // the default expansion limit (1000) which rejects files with >1000 tags.
+  // Without this, large studies fall through to dcm2json which hangs on
+  // compressed pixel data (DCMTK bug).
+  processEntities: false
 });
 function xmlToJson(xml) {
   try {
@@ -531,6 +536,7 @@ var Dcm2jsonOptionsSchema = z.object({
   signal: z.instanceof(AbortSignal).optional(),
   directOnly: z.boolean().optional(),
   charsetAssume: z.string().min(1).optional(),
+  charsetFallback: z.string().min(1).optional(),
   verbosity: z.enum(["verbose", "debug"]).optional()
 }).strict().optional();
 var VERBOSITY_FLAGS2 = { verbose: "-v", debug: "-d" };
@@ -544,27 +550,34 @@ function buildXmlOpts(options) {
   const result = {};
   if (options?.verbosity !== void 0) result["verbosity"] = options.verbosity;
   if (options?.charsetAssume !== void 0) result["charsetAssume"] = options.charsetAssume;
+  if (options?.charsetFallback !== void 0) result["charsetFallback"] = options.charsetFallback;
   return result;
 }
+function isCharsetError(stderrOutput) {
+  return stderrOutput.includes("convert character encoding") || stderrOutput.includes("Illegal byte sequence");
+}
+async function runXmlAndParse(binary, args, execOpts) {
+  const xmlResult = await execCommand(binary, args, execOpts);
+  if (!xmlResult.ok) return err(xmlResult.error);
+  if (xmlResult.value.exitCode !== 0) {
+    return err(createToolError("dcm2xml", args, xmlResult.value.exitCode, xmlResult.value.stderr));
+  }
+  const jsonResult = xmlToJson(xmlResult.value.stdout);
+  if (!jsonResult.ok) return err(jsonResult.error);
+  return ok({ data: jsonResult.value, source: "xml" });
+}
 async function tryXmlPath(inputPath, timeoutMs, signal, opts) {
   const xmlBinary = resolveBinary("dcm2xml");
-  if (!xmlBinary.ok) {
-    return err(xmlBinary.error);
-  }
+  if (!xmlBinary.ok) return err(xmlBinary.error);
   const charsetArgs = opts?.charsetAssume !== void 0 ? ["+Ca", opts.charsetAssume] : [];
   const xmlArgs = [...buildVerbosityArgs(opts?.verbosity), ...charsetArgs, "-nat", inputPath];
-  const xmlResult = await execCommand(xmlBinary.value, xmlArgs, { timeoutMs, signal });
-  if (!xmlResult.ok) {
-    return err(xmlResult.error);
+  const execOpts = signal !== void 0 ? { timeoutMs, signal } : { timeoutMs };
+  const result = await runXmlAndParse(xmlBinary.value, xmlArgs, execOpts);
+  if (!result.ok && opts?.charsetFallback !== void 0 && isCharsetError(result.error.message)) {
+    const fallbackArgs = [...buildVerbosityArgs(opts.verbosity), "+Ca", opts.charsetFallback, "-nat", inputPath];
+    return runXmlAndParse(xmlBinary.value, fallbackArgs, execOpts);
   }
-  if (xmlResult.value.exitCode !== 0) {
-    return err(createToolError("dcm2xml", xmlArgs, xmlResult.value.exitCode, xmlResult.value.stderr));
-  }
-  const jsonResult = xmlToJson(xmlResult.value.stdout);
-  if (!jsonResult.ok) {
-    return err(jsonResult.error);
-  }
-  return ok({ data: jsonResult.value, source: "xml" });
+  return result;
 }
 async function tryDirectPath(inputPath, timeoutMs, signal, verbosity) {
   const jsonBinary = resolveBinary("dcm2json");