@cosyte/dicom 0.1.4 → 0.1.6

package/CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelog
 
+## 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
+
+- c827734: `serializeDicom()` and `deidentify()` now write a value over 65,534 bytes in a VR with a 16-bit length field as `UN`, as PS3.5 section 6.2.2 requires, instead of throwing a `RangeError`.
+
+  Under the Explicit VR transfer syntaxes, VRs such as `CS`, `DS`, `LO`, `PN`, `UI` and `US` carry a 16-bit length field, and the element encoder put every length through it unchecked. A value the reader tolerated at the odd length FFFFH, padded to 65,536 bytes on write, made `serializeDicom()` throw a `RangeError` it does not document, and so did a multi-valued UID that `deidentify()` remapped past 65,534 bytes; inside a Sequence, `deidentify()` threw it itself while rebuilding the Item. Such a value is now written in the long form with VR `UN`, Little Endian under Explicit VR Big Endian too, and reads back byte for byte. A Private Creator, for which the same section forbids `UN`, is not rewritten.
+
 ## 0.1.4
 
 ### Patch Changes

package/dist/index.cjs

@@ -11,7 +11,7 @@ var __export = (target, all) => {
 };
 
 // src/version.ts
-var VERSION = "0.1.4";
+var VERSION = "0.1.6";
 
 // 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") {
@@ -14943,6 +14959,7 @@ function toDate(value, options) {
   return Number.isNaN(instant.getTime()) ? void 0 : instant;
 }
 var UNDEFINED_LENGTH4 = 4294967295;
+var MAX_SHORT_FORM_LENGTH = 65534;
 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);
@@ -14961,6 +14978,19 @@ function uint32(n, littleEndian) {
   else buf.writeUInt32BE(n, 0);
   return buf;
 }
+function littleEndianValue(value, vr, littleEndian) {
+  const stride = BE_VR_STRIDE[vr];
+  if (littleEndian || stride === 0) return value;
+  const swapped = buffer.Buffer.from(value);
+  const whole = swapped.subarray(0, swapped.length - swapped.length % stride);
+  if (stride === 2) whole.swap16();
+  else if (stride === 4) whole.swap32();
+  else whole.swap64();
+  return swapped;
+}
+function isPrivateCreator(group, element) {
+  return group % 2 === 1 && element >= 16 && element <= 255;
+}
 function isFullSpanElement(el, encoding) {
   if (el.length === UNDEFINED_LENGTH4) return true;
   return el.vr === "SQ" && encoding !== "implicit";
@@ -14992,6 +15022,17 @@ function encodeDatasetElement(el, encoding, rawBytes = el.rawBytes) {
       value
     ]);
   }
+  if (value.length > MAX_SHORT_FORM_LENGTH && !isPrivateCreator(group, element)) {
+    return buffer.Buffer.concat([
+      uint16(group, littleEndian),
+      uint16(element, littleEndian),
+      buffer.Buffer.from("UN", "ascii"),
+      buffer.Buffer.from([0, 0]),
+      // reserved (PS3.5 §7.1.2)
+      uint32(value.length, littleEndian),
+      littleEndianValue(value, el.vr, littleEndian)
+    ]);
+  }
   return buffer.Buffer.concat([
     uint16(group, littleEndian),
     uint16(element, littleEndian),
@@ -16741,14 +16782,19 @@ 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;
 }
@@ -17150,7 +17196,7 @@ function insertedSequence(tag, items, encoding) {
   });
 }
 function decodeCreator(el) {
-  return el.rawBytes.toString("latin1").replace(/[\0 ]+$/, "").trim();
+  return trimTrailingPad(el.rawBytes).toString("latin1").trim();
 }
 function isPrivateCreatorElement(tag) {
   const { element } = splitTag(tag);
@@ -17278,7 +17324,7 @@ function declarationKey(group, creator) {
   return `${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 = [];
@@ -17715,7 +17761,7 @@ 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)) {