@cosyte/dicom 0.1.3 → 0.1.5

package/CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelog
 
+## 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
+
+- d21f159: Decoding a string value now takes time linear in its length, so a long run of spaces, digits, or pad bytes no longer stalls `Element.value` or `parseSpecificCharacterSet()` for seconds.
+
+  Three checks on file-controlled text used patterns that backtracked quadratically on such a run: the trim of each value of a multi-valued string VR, the DS number grammar, and the trim of each Specific Character Set term. Each now scans the value once from each end and returns the same result as before. De-identification also strips every asterisk from the conditional action branch it takes, not only the first; no row of the Annex E table puts one there today, so no output changes.
+
 ## 0.1.3
 
 ### Patch Changes

package/dist/index.cjs

@@ -11,7 +11,7 @@ var __export = (target, all) => {
 };
 
 // src/version.ts
-var VERSION = "0.1.3";
+var VERSION = "0.1.5";
 
 // src/dictionary/index.ts
 var dictionary_exports = {};
@@ -13223,7 +13223,18 @@ var MULTIBYTE_LABELS = /* @__PURE__ */ new Set([
 ]);
 function parseSpecificCharacterSet(bytes) {
   const raw = bytes.toString("latin1");
-  return raw.split("\\").map((t) => t.replace(/[ ]+$/u, "").replace(/^[ ]+/u, "")).map((t) => t === "" || TERM_TO_LABEL.has(t) ? t : WITHHELD);
+  return raw.split("\\").map(trimTermPad).map((t) => t === "" || TERM_TO_LABEL.has(t) ? t : WITHHELD);
+}
+function trimTermPad(term) {
+  let end = term.length;
+  while (end > 0) {
+    const code = term.charCodeAt(end - 1);
+    if (code !== 0 && code !== 32) break;
+    end--;
+  }
+  let start = 0;
+  while (start < end && term.charCodeAt(start) === 32) start++;
+  return term.slice(start, end);
 }
 function isKnownCharsetTerm(term) {
   return TERM_TO_LABEL.has(term);
@@ -13790,6 +13801,14 @@ function stripPad(bytes, tag, vr, position, warnings) {
   }
   return bytes.subarray(0, end);
 }
+function trimSpaces(value) {
+  let start = 0;
+  let end = value.length;
+  while (start < end && value.charCodeAt(start) === SPACE2) start++;
+  while (end > start && value.charCodeAt(end - 1) === SPACE2) end--;
+  return value.slice(start, end);
+}
+var DECIMAL_STRING = /^[+-]?(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][+-]?\d+)?$/u;
 function attachWarnings(value, warnings) {
   return warnings.length > 0 ? { ...value, warnings } : value;
 }
@@ -13841,7 +13860,7 @@ function decodeElementValue(element) {
   }
   if (vr === "LO" || vr === "SH" || vr === "UC") {
     const text = decodeCharsetString(trimmed, tag, vr, charset, position, warnings);
-    const values = text.split("\\").map((v) => v.replace(/^ +| +$/gu, ""));
+    const values = text.split("\\").map(trimSpaces);
     return attachWarnings({ kind: "strings", values }, warnings);
   }
   if (vr === "LT" || vr === "ST" || vr === "UT") {
@@ -13854,7 +13873,7 @@ function decodeElementValue(element) {
   }
   if (vr === "AE" || vr === "CS" || vr === "AS" || vr === "UI") {
     const text = decodeAsciiString(trimmed, tag, vr, position, warnings);
-    const values = text.split("\\").map((v) => v.replace(/^ +| +$/gu, ""));
+    const values = text.split("\\").map(trimSpaces);
     return attachWarnings({ kind: "strings", values }, warnings);
   }
   if (vr === "DS") {
@@ -13862,7 +13881,7 @@ function decodeElementValue(element) {
     const values = text.split("\\").map((v) => {
       const t = v.trim();
       if (t.length === 0) return null;
-      if (!/^[+-]?(\d+\.?\d*|\.\d+)([eE][+-]?\d+)?$/u.test(t)) return null;
+      if (!DECIMAL_STRING.test(t)) return null;
       const n = Number(t);
       return Number.isFinite(n) ? n : null;
     });
@@ -14924,6 +14943,7 @@ 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);
@@ -14942,6 +14962,19 @@ 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";
@@ -14973,6 +15006,17 @@ 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),
@@ -16695,7 +16739,7 @@ function effectiveCode(action, active) {
   return action.basicProfile;
 }
 function resolveAction(code) {
-  const first = code.split("/", 1)[0]?.replace("*", "") ?? "K";
+  const first = code.split("/", 1)[0]?.replaceAll("*", "") ?? "K";
   return first;
 }
 var TEXT_DUMMIES = {