@cosyte/dicom 0.1.5 → 0.1.7

package/CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelog
 
+## 0.1.7
+
+### Patch Changes
+
+- d04ae1a: `serializeDicom()` now refuses a Private Creator or File Meta element whose value is over 65,534 bytes with `DicomSerializeError` code `VALUE_TOO_LONG_FOR_VR`, instead of throwing a `RangeError`.
+
+  PS3.5 section 6.2.2 has a value that long, in a VR with a 16-bit Value Length Field, written as `UN`, which the writer has done since 0.1.5, but forbids `UN` for a Private Creator Data Element (gggg,0010-00FF) and for any File Meta Information element (0002,xxxx). Such a value has no encoding, and the writer still put its length through the 16-bit write. Callers now see a `DicomSerializeError` they can narrow on `err.code === "VALUE_TOO_LONG_FOR_VR"`, with a fixed message that carries no tag, length or byte of the value, and nothing is written. It applies to a Private Creator under the Explicit VR syntaxes (Implicit VR Little Endian has a 32-bit length and writes it) and to a File Meta element under every syntax, since the File Meta group is always Explicit VR Little Endian. `deidentify()` throws the same error when it re-encodes a Sequence Item that keeps such a creator, which only a `RetainSafePrivate` run whose `Profile` names a creator that long does. `SERIALIZE_ERROR_CODES` gains the new code, so an exhaustive `switch` over `SerializeErrorCode` needs one more case.
+
+## 0.1.6
+
+### Patch Changes
+
+- 39ef9b3: `deidentify()` and `Dataset.directory` now strip a value's padding in linear time, so a long run of NULL or SPACE bytes in a file no longer stalls them for seconds.
+
+  Seven trims of file-controlled text used patterns that backtracked quadratically on a run of pad bytes ending in any other byte: the NULL trim of every UID `deidentify()` remaps, its reads of a Private Creator, a Block Identifying Information Status and the Burned In Annotation flag, and the DICOMDIR's reads of its `CS` values. A SOP Instance UID of 50,000 NULL bytes and one letter took `deidentify()` close to four seconds, and the time grew with the square of the run. Each trim now scans the value once from its end, or from each end, and returns the same result as before.
+
 ## 0.1.5
 
 ### Patch Changes

package/dist/index.cjs

@@ -11,7 +11,7 @@ var __export = (target, all) => {
 };
 
 // src/version.ts
-var VERSION = "0.1.5";
+var VERSION = "0.1.7";
 
 // src/dictionary/index.ts
 var dictionary_exports = {};
@@ -11141,8 +11141,24 @@ function directoryRecords(source) {
 }
 function csValues(el) {
   if (el === void 0) return void 0;
-  const text = el.rawBytes.toString("latin1").replace(/[\0 ]+$/u, "");
-  return Object.freeze(text === "" ? [] : text.split("\\").map((v) => v.replace(/^ +| +$/gu, "")));
+  const text = stripTrailingPad(el.rawBytes.toString("latin1"));
+  return Object.freeze(text === "" ? [] : text.split("\\").map(trimSpaces));
+}
+function stripTrailingPad(text) {
+  let end = text.length;
+  while (end > 0) {
+    const code = text.charCodeAt(end - 1);
+    if (code !== 0 && code !== 32) break;
+    end--;
+  }
+  return text.slice(0, end);
+}
+function trimSpaces(value) {
+  let start = 0;
+  let end = value.length;
+  while (start < end && value.charCodeAt(start) === 32) start++;
+  while (end > start && value.charCodeAt(end - 1) === 32) end--;
+  return value.slice(start, end);
 }
 function analyzeDirectory(source) {
   if (!isDicomdir(source.fileMeta)) return void 0;
@@ -13801,7 +13817,7 @@ function stripPad(bytes, tag, vr, position, warnings) {
   }
   return bytes.subarray(0, end);
 }
-function trimSpaces(value) {
+function trimSpaces2(value) {
   let start = 0;
   let end = value.length;
   while (start < end && value.charCodeAt(start) === SPACE2) start++;
@@ -13860,7 +13876,7 @@ function decodeElementValue(element) {
   }
   if (vr === "LO" || vr === "SH" || vr === "UC") {
     const text = decodeCharsetString(trimmed, tag, vr, charset, position, warnings);
-    const values = text.split("\\").map(trimSpaces);
+    const values = text.split("\\").map(trimSpaces2);
     return attachWarnings({ kind: "strings", values }, warnings);
   }
   if (vr === "LT" || vr === "ST" || vr === "UT") {
@@ -13873,7 +13889,7 @@ function decodeElementValue(element) {
   }
   if (vr === "AE" || vr === "CS" || vr === "AS" || vr === "UI") {
     const text = decodeAsciiString(trimmed, tag, vr, position, warnings);
-    const values = text.split("\\").map(trimSpaces);
+    const values = text.split("\\").map(trimSpaces2);
     return attachWarnings({ kind: "strings", values }, warnings);
   }
   if (vr === "DS") {
@@ -14942,8 +14958,43 @@ function toDate(value, options) {
   const instant = new Date(utc.getTime() - offsetMinutes * 6e4);
   return Number.isNaN(instant.getTime()) ? void 0 : instant;
 }
+
+// src/serialize/errors.ts
+var SERIALIZE_ERROR_CODES = {
+  MISSING_TRANSFER_SYNTAX: "MISSING_TRANSFER_SYNTAX",
+  UNSUPPORTED_TRANSFER_SYNTAX: "UNSUPPORTED_TRANSFER_SYNTAX",
+  INVALID_ENCAPSULATED_PIXEL_DATA: "INVALID_ENCAPSULATED_PIXEL_DATA",
+  DIRECTORY_OFFSET_UNRESOLVED: "DIRECTORY_OFFSET_UNRESOLVED",
+  DIRECTORY_OFFSET_DEFLATED: "DIRECTORY_OFFSET_DEFLATED",
+  VALUE_TOO_LONG_FOR_VR: "VALUE_TOO_LONG_FOR_VR"
+};
+var DicomSerializeError = class extends Error {
+  code;
+  /**
+   * Construct a new `DicomSerializeError`. The `message` MUST be built only
+   * from structural facts (never a decoded attribute value) so the error is
+   * always safe to log.
+   *
+   * @internal
+   */
+  constructor(code, message) {
+    super(`[${code}] ${message}`);
+    this.name = "DicomSerializeError";
+    this.code = code;
+  }
+};
+
+// src/serialize/element.ts
 var UNDEFINED_LENGTH4 = 4294967295;
 var MAX_SHORT_FORM_LENGTH = 65534;
+var PRIVATE_CREATOR_TOO_LONG_MESSAGE = "A Private Creator Data Element carries a value longer than 65,534 bytes under an Explicit VR Transfer Syntax, in a VR whose Value Length Field is 16 bits. PS3.5 section 6.2.2 has such a value written with VR UN, but says UN shall not be used for a Private Creator Data Element, so the value has no encoding and nothing is written.";
+var FILE_META_TOO_LONG_MESSAGE = "A File Meta Information element carries a value longer than 65,534 bytes in a VR whose Explicit VR Value Length Field is 16 bits. PS3.5 section 6.2.2 has such a value written with VR UN, but says UN shall not be used for a File Meta Information element, so the value has no encoding and nothing is written.";
+function refuseTooLongForVr(kind) {
+  throw new DicomSerializeError(
+    SERIALIZE_ERROR_CODES.VALUE_TOO_LONG_FOR_VR,
+    kind === "privateCreator" ? PRIVATE_CREATOR_TOO_LONG_MESSAGE : FILE_META_TOO_LONG_MESSAGE
+  );
+}
 var NULL_PAD_VRS = /* @__PURE__ */ new Set(["UI", "OB", "OW", "OF", "OD", "OL", "OV", "UN"]);
 function padValue(value, vr) {
   if (value.length % 2 === 0) return buffer.Buffer.from(value);
@@ -15006,7 +15057,9 @@ function encodeDatasetElement(el, encoding, rawBytes = el.rawBytes) {
       value
     ]);
   }
-  if (value.length > MAX_SHORT_FORM_LENGTH && !isPrivateCreator(group, element)) {
+  if (value.length > MAX_SHORT_FORM_LENGTH) {
+    if (isPrivateCreator(group, element)) refuseTooLongForVr("privateCreator");
+    if (group === 2) refuseTooLongForVr("fileMeta");
     return buffer.Buffer.concat([
       uint16(group, littleEndian),
       uint16(element, littleEndian),
@@ -15025,30 +15078,6 @@ function encodeDatasetElement(el, encoding, rawBytes = el.rawBytes) {
     value
   ]);
 }
-
-// src/serialize/errors.ts
-var SERIALIZE_ERROR_CODES = {
-  MISSING_TRANSFER_SYNTAX: "MISSING_TRANSFER_SYNTAX",
-  UNSUPPORTED_TRANSFER_SYNTAX: "UNSUPPORTED_TRANSFER_SYNTAX",
-  INVALID_ENCAPSULATED_PIXEL_DATA: "INVALID_ENCAPSULATED_PIXEL_DATA",
-  DIRECTORY_OFFSET_UNRESOLVED: "DIRECTORY_OFFSET_UNRESOLVED",
-  DIRECTORY_OFFSET_DEFLATED: "DIRECTORY_OFFSET_DEFLATED"
-};
-var DicomSerializeError = class extends Error {
-  code;
-  /**
-   * Construct a new `DicomSerializeError`. The `message` MUST be built only
-   * from structural facts (never a decoded attribute value) so the error is
-   * always safe to log.
-   *
-   * @internal
-   */
-  constructor(code, message) {
-    super(`[${code}] ${message}`);
-    this.name = "DicomSerializeError";
-    this.code = code;
-  }
-};
 var UNDEFINED_LENGTH5 = 4294967295;
 var ITEM2 = 4294893568;
 var ITEM_DELIMITATION = 4294893581;
@@ -15493,6 +15522,7 @@ function encodeMetaElement(tag, vr, rawValue) {
       value
     ]);
   }
+  if (value.length > MAX_SHORT_FORM_LENGTH) refuseTooLongForVr("fileMeta");
   const lengthBuf = buffer.Buffer.alloc(2);
   lengthBuf.writeUInt16LE(value.length, 0);
   return buffer.Buffer.concat([groupBuf, elementBuf, vrBuf, lengthBuf, value]);
@@ -16766,14 +16796,19 @@ function dummyBytes(vr) {
   const buf = buffer.Buffer.from(text, "latin1");
   return buf.length % 2 === 0 ? buf : buffer.Buffer.concat([buf, buffer.Buffer.from([32])]);
 }
+function stripTrailingNulls(text) {
+  let end = text.length;
+  while (end > 0 && text.charCodeAt(end - 1) === 0) end--;
+  return text.slice(0, end);
+}
 function remapUidBytes(rawBytes, remap) {
-  const raw = rawBytes.toString("latin1").replace(/\0+$/, "");
+  const raw = stripTrailingNulls(rawBytes.toString("latin1"));
   const remapped = raw.split("\\").map((uid2) => remap(uid2.trim())).join("\\");
   const buf = buffer.Buffer.from(remapped, "latin1");
   return buf.length % 2 === 0 ? buf : buffer.Buffer.concat([buf, buffer.Buffer.from([0])]);
 }
 function uidValueMultiplicity(rawBytes) {
-  const raw = rawBytes.toString("latin1").replace(/\0+$/, "");
+  const raw = stripTrailingNulls(rawBytes.toString("latin1"));
   if (raw.length === 0) return 0;
   return raw.split("\\").length;
 }
@@ -17175,7 +17210,7 @@ function insertedSequence(tag, items, encoding) {
   });
 }
 function decodeCreator(el) {
-  return el.rawBytes.toString("latin1").replace(/[\0 ]+$/, "").trim();
+  return trimTrailingPad(el.rawBytes).toString("latin1").trim();
 }
 function isPrivateCreatorElement(tag) {
   const { element } = splitTag(tag);
@@ -17303,7 +17338,7 @@ function declarationKey(group, creator) {
   return `${String(group)}:${creator}`;
 }
 function decodeEnumerated(el) {
-  return el.rawBytes.toString("latin1").replace(/[\0 ]+$/, "").trim();
+  return trimTrailingPad(el.rawBytes).toString("latin1").trim();
 }
 function readUnsignedShorts(el) {
   const values = [];
@@ -17740,7 +17775,7 @@ function hasUncleanedBurnedIn(ds) {
   if (!ds.has(TAG_PIXEL_DATA)) return false;
   const flag = ds.get(TAG_BURNED_IN_ANNOTATION);
   if (flag === void 0) return true;
-  return flag.rawBytes.toString("latin1").replace(/[\0 ]+$/, "").trim().toUpperCase() !== "NO";
+  return trimTrailingPad(flag.rawBytes).toString("latin1").trim().toUpperCase() !== "NO";
 }
 function deidentify(ds, options = {}) {
   if (isRefusedTransferSyntax(ds.fileMeta?.transferSyntaxUID)) {