dicom-synth 1.10.0 → 1.12.0

package/dist/esm/index.js

@@ -154,6 +154,13 @@ function validateSeed(value, path) {
     );
   }
 }
+function validateUidSalt(value, path) {
+  if (typeof value !== "string" || value.length === 0) {
+    throw new Error(
+      `${path}: must be a non-empty string; got ${JSON.stringify(value)}`
+    );
+  }
+}
 function validateSizeKbLimit(kb, path) {
   if (kb * 1024 > MAX_PIXEL_BYTES) {
     throw new Error(`${path}: exceeds the 512 MB limit`);
@@ -288,26 +295,100 @@ function validateEntry(entry, path) {
   return e;
 }
 var MODALITY_TAG = "00080060";
+function validateTagValue(key, value, path) {
+  const isModalityKey = key === "Modality" || key.toUpperCase() === MODALITY_TAG;
+  if (isModalityKey && typeof value === "string" && Object.hasOwn(VALID_MODALITIES, value)) {
+    throw new Error(
+      `${path}.${key}: set a preset modality (${Object.keys(VALID_MODALITIES).join("/")}) via the "modality" field, not tags \u2014 it is coupled to SOPClassUID and type-1 attributes`
+    );
+  }
+  if (typeof value === "string") {
+    for (const name of findTemplateTokens(value)) {
+      if (!TEMPLATE_VOCAB.includes(name)) {
+        throw new Error(
+          `${path}.${key}: unknown placeholder "{${name}}" \u2014 allowed: ${TEMPLATE_VOCAB.map((v) => `{${v}}`).join(", ")}`
+        );
+      }
+    }
+  }
+}
 function validateTags(tags, path) {
   if (typeof tags !== "object" || tags === null || Array.isArray(tags)) {
     throw new Error(`${path}: must be an object`);
   }
   for (const [key, value] of Object.entries(tags)) {
     validateTagKey(key, path);
-    const isModalityKey = key === "Modality" || key.toUpperCase() === MODALITY_TAG;
-    if (isModalityKey && typeof value === "string" && Object.hasOwn(VALID_MODALITIES, value)) {
-      throw new Error(
-        `${path}.${key}: set a preset modality (${Object.keys(VALID_MODALITIES).join("/")}) via the "modality" field, not tags \u2014 it is coupled to SOPClassUID and type-1 attributes`
+    validateTagValue(key, value, path);
+  }
+}
+function validateArrayOfLength(value, count, path) {
+  if (!Array.isArray(value)) {
+    throw new Error(`${path}: must be an array`);
+  }
+  if (value.length !== count) {
+    throw new Error(
+      `${path}: length (${value.length}) must equal count (${count})`
+    );
+  }
+  return value;
+}
+function validateInstances(instances, path) {
+  if (typeof instances !== "object" || instances === null || Array.isArray(instances)) {
+    throw new Error(`${path}: must be an object`);
+  }
+  const inst = instances;
+  validatePositiveInt(inst.count, `${path}.count`);
+  const count = inst.count;
+  const hasKb = "targetSizeKb" in inst;
+  const hasBytes = "targetBytes" in inst;
+  if (hasKb === hasBytes) {
+    throw new Error(
+      `${path}: exactly one of "targetSizeKb" or "targetBytes" is required`
+    );
+  }
+  if (hasKb) {
+    const sizes = validateArrayOfLength(
+      inst.targetSizeKb,
+      count,
+      `${path}.targetSizeKb`
+    );
+    sizes.forEach((kb, i) => {
+      validatePositiveInt(kb, `${path}.targetSizeKb[${i}]`);
+      validateSizeKbLimit(kb, `${path}.targetSizeKb[${i}]`);
+    });
+  } else {
+    const sizes = validateArrayOfLength(
+      inst.targetBytes,
+      count,
+      `${path}.targetBytes`
+    );
+    sizes.forEach((b, i) => {
+      validateLargeTargetBytes(b, `${path}.targetBytes[${i}]`);
+    });
+  }
+  if ("modality" in inst) {
+    if (Array.isArray(inst.modality)) {
+      validateArrayOfLength(inst.modality, count, `${path}.modality`).forEach(
+        (m, i) => {
+          validateEnum(m, VALID_MODALITIES, `${path}.modality[${i}]`);
+        }
       );
+    } else {
+      validateEnum(inst.modality, VALID_MODALITIES, `${path}.modality`);
+    }
+  }
+  if ("tags" in inst) {
+    const tags = inst.tags;
+    if (typeof tags !== "object" || tags === null || Array.isArray(tags)) {
+      throw new Error(`${path}.tags: must be an object`);
     }
-    if (typeof value === "string") {
-      for (const name of findTemplateTokens(value)) {
-        if (!TEMPLATE_VOCAB.includes(name)) {
-          throw new Error(
-            `${path}.${key}: unknown placeholder "{${name}}" \u2014 allowed: ${TEMPLATE_VOCAB.map((v) => `{${v}}`).join(", ")}`
-          );
+    for (const [key, arr] of Object.entries(tags)) {
+      validateTagKey(key, `${path}.tags`);
+      validateArrayOfLength(arr, count, `${path}.tags.${key}`).forEach(
+        (value, i) => {
+          validateTagValue(key, value, `${path}.tags[${i}]`);
         }
-      }
+      );
     }
   }
 }
@@ -316,16 +397,27 @@ function validateSeries(series, path) {
     throw new Error(`${path}: must be an object`);
   }
   const s = series;
-  if (!Array.isArray(s.entries) || s.entries.length === 0) {
-    throw new Error(`${path}.entries: must be a non-empty array`);
+  const hasEntries = "entries" in s;
+  const hasInstances = "instances" in s;
+  if (hasEntries === hasInstances) {
+    throw new Error(
+      `${path}: must have exactly one of "entries" or "instances"`
+    );
   }
-  for (let i = 0; i < s.entries.length; i++) {
-    const e = validateEntry(s.entries[i], `${path}.entries[${i}]`);
-    if (!TYPE_CAPABILITIES[e.type].image) {
-      throw new Error(
-        `${path}.entries[${i}].type: "${e.type}" cannot be used inside a series \u2014 only image-generating types can be grouped`
-      );
+  if (hasEntries) {
+    if (!Array.isArray(s.entries) || s.entries.length === 0) {
+      throw new Error(`${path}.entries: must be a non-empty array`);
     }
+    for (let i = 0; i < s.entries.length; i++) {
+      const e = validateEntry(s.entries[i], `${path}.entries[${i}]`);
+      if (!TYPE_CAPABILITIES[e.type].image) {
+        throw new Error(
+          `${path}.entries[${i}].type: "${e.type}" cannot be used inside a series \u2014 only image-generating types can be grouped`
+        );
+      }
+    }
+  } else {
+    validateInstances(s.instances, `${path}.instances`);
   }
   if ("tags" in s) validateTags(s.tags, `${path}.tags`);
   return s;
@@ -370,6 +462,7 @@ function validateDatasetSpec(raw) {
     (study, i) => validateStudy(study, `studies[${i}]`)
   );
   if ("seed" in r) validateSeed(r.seed, "DatasetSpec.seed");
+  if ("uidSalt" in r) validateUidSalt(r.uidSalt, "DatasetSpec.uidSalt");
   if ("layout" in r) {
     validateEnum(r.layout, VALID_LAYOUTS, "DatasetSpec.layout");
   }
@@ -385,6 +478,7 @@ function validateDatasetSpec(raw) {
     ...entries.length > 0 ? { entries } : {},
     ...studies.length > 0 ? { studies } : {},
     ...r.seed !== void 0 ? { seed: r.seed } : {},
+    ...r.uidSalt !== void 0 ? { uidSalt: r.uidSalt } : {},
     ...r.layout !== void 0 ? { layout: r.layout } : {},
     ...r.pathQuirks !== void 0 ? { pathQuirks: r.pathQuirks } : {}
   };
@@ -500,6 +594,7 @@ function validateParametricSpec(raw) {
     }
   }
   if ("seed" in r) validateSeed(r.seed, "ParametricSpec.seed");
+  if ("uidSalt" in r) validateUidSalt(r.uidSalt, "ParametricSpec.uidSalt");
   if ("layout" in r) {
     validateEnum(r.layout, VALID_LAYOUTS, "ParametricSpec.layout");
   }
@@ -507,6 +602,7 @@ function validateParametricSpec(raw) {
     studies,
     ...edgeCases.length > 0 ? { edgeCases } : {},
     ...r.seed !== void 0 ? { seed: r.seed } : {},
+    ...r.uidSalt !== void 0 ? { uidSalt: r.uidSalt } : {},
     ...r.layout !== void 0 ? { layout: r.layout } : {}
   };
 }
@@ -694,8 +790,19 @@ import { closeSync, mkdirSync as mkdirSync2, openSync, writeSync } from "node:fs
 import { resolve as resolve2 } from "node:path";
 
 // src/syntheticFixtures/uid.ts
-import { randomBytes } from "node:crypto";
-var ROLE_CODE = { study: 1, series: 2, sop: 3 };
+import { createHash, randomBytes } from "node:crypto";
+function hashUid(salt, key) {
+  const digest = createHash("sha256").update(salt).update(" ").update(key).digest();
+  const n = BigInt(`0x${digest.subarray(0, 16).toString("hex")}`);
+  return `2.25.${n.toString()}`;
+}
+function randomUid() {
+  const n = BigInt(`0x${randomBytes(16).toString("hex")}`);
+  return `2.25.${n.toString()}`;
+}
+function seedToSalt(seed) {
+  return seed === void 0 ? void 0 : `seed:${seed}`;
+}
 function lcg(seed) {
   let s = seed >>> 0;
   return () => {
@@ -703,60 +810,28 @@ function lcg(seed) {
     return s;
   };
 }
-function formatUid(a, b, role) {
-  return `2.25.${a}.${b}.${ROLE_CODE[role]}`;
-}
-function randomUid(role) {
-  const buf = randomBytes(8);
-  return formatUid(buf.readUInt32BE(0), buf.readUInt32BE(4), role);
-}
-function seededUid(next, role) {
-  return formatUid(next(), next(), role);
-}
-var GROUP_STREAM_OFFSET = 2654435769;
 var PARAMETRIC_STREAM_OFFSET = 2246822507;
 function seededStream(seed, offset, a, b) {
   return lcg(seed * 999983 + offset + a * 1000003 + b * 7919 >>> 0);
 }
-function makeGroupUidGenerator(seed) {
-  if (seed === void 0) {
-    return {
-      study: () => randomUid("study"),
-      series: () => randomUid("series")
-    };
+function makeGroupUidGenerator(salt) {
+  if (salt === void 0) {
+    return { study: () => randomUid(), series: () => randomUid() };
   }
   return {
-    study: (studyIndex) => seededUid(
-      seededStream(seed, GROUP_STREAM_OFFSET, studyIndex + 1, 0),
-      "study"
-    ),
-    series: (studyIndex, seriesIndex) => seededUid(
-      seededStream(
-        seed,
-        GROUP_STREAM_OFFSET,
-        studyIndex + 1,
-        seriesIndex + 1
-      ),
-      "series"
-    )
+    study: (studyIndex) => hashUid(salt, `study:${studyIndex}`),
+    series: (studyIndex, seriesIndex) => hashUid(salt, `series:${studyIndex}:${seriesIndex}`)
   };
 }
-function makeUidGenerator(seed) {
-  if (seed === void 0) {
-    return () => ({
-      study: randomUid("study"),
-      series: randomUid("series"),
-      sop: randomUid("sop")
-    });
+function makeUidGenerator(salt) {
+  if (salt === void 0) {
+    return () => ({ study: randomUid(), series: randomUid(), sop: randomUid() });
   }
-  return (fileIndex) => {
-    const next = seededStream(seed, 0, fileIndex, 0);
-    return {
-      study: seededUid(next, "study"),
-      series: seededUid(next, "series"),
-      sop: seededUid(next, "sop")
-    };
-  };
+  return (fileIndex) => ({
+    study: hashUid(salt, `study:flat:${fileIndex}`),
+    series: hashUid(salt, `series:flat:${fileIndex}`),
+    sop: hashUid(salt, `sop:${fileIndex}`)
+  });
 }
 
 // src/syntheticFixtures/streamWrite.ts
@@ -869,7 +944,7 @@ function streamLargeImage(outPath, options) {
     fragmentBytes = FRAGMENT_MAX_BYTES
   } = options;
   const modality = options.modality ?? "CT";
-  const uid = options.uid ?? makeUidGenerator(options.seed)(0);
+  const uid = options.uid ?? makeUidGenerator(options.uidSalt ?? seedToSalt(options.seed))(0);
   const identity = { modality, uid, tags: options.tags };
   mkdirSync2(resolve2(outPath, ".."), { recursive: true });
   const zeroChunk = Buffer.alloc(Math.max(1, chunkBytes));
@@ -921,6 +996,7 @@ function writeLargeImageFile(spec, outPath, options = {}) {
   return streamLargeImage(outPath, {
     targetBytes,
     modality: spec.modality,
+    uidSalt: spec.uidSalt,
     seed: spec.seed,
     chunkBytes: options.chunkBytes
   });
@@ -1033,19 +1109,51 @@ function filename(type, index, padWidth) {
 function entryCount(entries) {
   return entries.reduce((sum, e) => sum + (e.count ?? 1), 0);
 }
+function seriesFileCount(series) {
+  return series.instances ? series.instances.count : entryCount(series.entries ?? []);
+}
 function studyFileCount(studies) {
   return studies.reduce(
-    (sum, study) => sum + (study.count ?? 1) * study.series.reduce((s, series) => s + entryCount(series.entries), 0),
+    (sum, study) => sum + (study.count ?? 1) * study.series.reduce((s, series) => s + seriesFileCount(series), 0),
     0
   );
 }
-function resolveUid(seed, index, override) {
-  return { ...makeUidGenerator(seed)(index), ...override };
+function* seriesFiles(series) {
+  if (series.instances) {
+    const inst = series.instances;
+    for (let i = 0; i < inst.count; i++) {
+      const modality = Array.isArray(inst.modality) ? inst.modality[i] : inst.modality;
+      const baseSpec = {
+        ...inst.targetBytes ? { type: "large-image", targetBytes: inst.targetBytes[i] } : { type: "valid-image", targetSizeKb: inst.targetSizeKb?.[i] },
+        ...modality !== void 0 ? { modality } : {}
+      };
+      const instanceTags = {};
+      if (inst.tags) {
+        for (const [key, values] of Object.entries(inst.tags)) {
+          instanceTags[key] = values[i];
+        }
+      }
+      yield { baseSpec, instanceTags };
+    }
+    return;
+  }
+  for (const entry of series.entries ?? []) {
+    const { count = 1, tags, ...baseSpec } = entry;
+    for (let i = 0; i < count; i++) {
+      yield { baseSpec, instanceTags: tags ?? {} };
+    }
+  }
+}
+function resolveUid(salt, index, override) {
+  return { ...makeUidGenerator(salt)(index), ...override };
+}
+function effectiveSalt(spec) {
+  return spec.uidSalt ?? seedToSalt(spec.seed);
 }
 async function generateFile(spec, options) {
   const index = options?.index ?? 0;
   const padWidth = options?.padWidth ?? 3;
-  const uid = resolveUid(options?.seed, index, options?.uid);
+  const uid = resolveUid(effectiveSalt(options ?? {}), index, options?.uid);
   const resolvedSpec = "tags" in spec && spec.tags ? {
     ...spec,
     tags: resolveTagTemplates(spec.tags, { index, ...options?.context })
@@ -1117,6 +1225,9 @@ function hierarchicalName(studyOrdinal, seriesIndex, instanceNumber) {
 function withResolvedSeed(spec) {
   return spec.seed === void 0 ? { ...spec, seed: randomBytes2(4).readUInt32BE(0) } : spec;
 }
+function withResolvedSalt(spec) {
+  return spec.uidSalt === void 0 ? { ...spec, uidSalt: randomBytes2(16).toString("hex") } : spec;
+}
 function computeNames(type, index, padWidth, grouped, quirks) {
   let names;
   if (grouped) {
@@ -1157,7 +1268,7 @@ function* planCollection(spec) {
       globalIndex++;
     }
   }
-  const groupUids = makeGroupUidGenerator(spec.seed);
+  const groupUids = makeGroupUidGenerator(effectiveSalt(spec));
   let studyOrdinal = 0;
   for (const study of studies) {
     const copies = study.count ?? 1;
@@ -1166,57 +1277,56 @@ function* planCollection(spec) {
       for (const [seriesIndex, series] of study.series.entries()) {
         const seriesUid = groupUids.series(studyOrdinal, seriesIndex);
         let instanceNumber = 0;
-        for (const entry of series.entries) {
-          const { count = 1, ...fileSpec } = entry;
-          for (let i = 0; i < count; i++) {
-            instanceNumber++;
-            const tags = {
-              InstanceNumber: instanceNumber,
-              ...study.tags,
-              ...series.tags,
-              ...fileSpec.tags
-            };
-            yield {
-              fileSpec: { ...fileSpec, tags },
+        for (const { baseSpec, instanceTags } of seriesFiles(series)) {
+          instanceNumber++;
+          const tags = {
+            InstanceNumber: instanceNumber,
+            ...study.tags,
+            ...series.tags,
+            ...instanceTags
+          };
+          yield {
+            fileSpec: { ...baseSpec, tags },
+            index: globalIndex,
+            uidOverride: { study: studyUid, series: seriesUid },
+            context: {
               index: globalIndex,
-              uidOverride: { study: studyUid, series: seriesUid },
-              context: {
-                index: globalIndex,
-                studyIndex: studyOrdinal,
-                seriesIndex,
-                instanceNumber
-              },
-              ...computeNames(
-                fileSpec.type,
-                globalIndex,
-                padWidth,
-                hierarchical ? { studyOrdinal, seriesIndex, instanceNumber } : void 0,
-                quirks
-              )
-            };
-            globalIndex++;
-          }
+              studyIndex: studyOrdinal,
+              seriesIndex,
+              instanceNumber
+            },
+            ...computeNames(
+              baseSpec.type,
+              globalIndex,
+              padWidth,
+              hierarchical ? { studyOrdinal, seriesIndex, instanceNumber } : void 0,
+              quirks
+            )
+          };
+          globalIndex++;
         }
       }
       studyOrdinal++;
     }
   }
 }
-async function materialisePlan(plan, seed) {
+async function materialisePlan(plan, salt) {
   const file = await generateFile(plan.fileSpec, {
     index: plan.index,
-    seed,
+    uidSalt: salt,
     uid: plan.uidOverride,
     context: plan.context
   });
   return { ...file, filename: plan.filename, relativePath: plan.relativePath };
 }
 async function* generateCollectionFromSpec(spec) {
+  const salt = effectiveSalt(spec);
   for (const plan of planCollection(spec)) {
-    yield await materialisePlan(plan, spec.seed);
+    yield await materialisePlan(plan, salt);
   }
 }
 async function* previewCollection(spec) {
+  const salt = effectiveSalt(spec);
   for (const plan of planCollection(spec)) {
     if (plan.fileSpec.type === "large-image") {
       yield {
@@ -1226,7 +1336,7 @@ async function* previewCollection(spec) {
         approxBytes: plan.fileSpec.targetBytes
       };
     } else {
-      const file = await materialisePlan(plan, spec.seed);
+      const file = await materialisePlan(plan, salt);
       yield {
         relativePath: plan.relativePath,
         type: plan.fileSpec.type,
@@ -1240,6 +1350,7 @@ async function writeCollectionFromSpec(spec, outDir) {
   const root = resolve3(outDir);
   mkdirSync3(root, { recursive: true });
   const manifest = [];
+  const salt = effectiveSalt(spec);
   const createdDirs = /* @__PURE__ */ new Set([root]);
   const ensureDir = (filePath) => {
     const dir = dirname(filePath);
@@ -1258,7 +1369,7 @@ async function writeCollectionFromSpec(spec, outDir) {
           `large-image targetBytes (${large.targetBytes}) exceeds the 50 GB limit`
         );
       }
-      const uid = resolveUid(spec.seed, plan.index, plan.uidOverride);
+      const uid = resolveUid(salt, plan.index, plan.uidOverride);
       streamLargeImage(filePath, {
         targetBytes: large.targetBytes,
         modality: large.modality,
@@ -1266,7 +1377,7 @@ async function writeCollectionFromSpec(spec, outDir) {
         tags: resolveTagTemplates(large.tags, plan.context)
       });
     } else {
-      const file = await materialisePlan(plan, spec.seed);
+      const file = await materialisePlan(plan, salt);
       await writeFile(filePath, file.buffer);
     }
     manifest.push({
@@ -1444,8 +1555,117 @@ function extractTags(record, keepKeywords) {
   }
   return found ? tags : void 0;
 }
+function sizeKbOf(bytes) {
+  return Math.max(1, Math.round(bytes / 1024));
+}
+function bucketBytes(bytes, tolerance) {
+  if (tolerance <= 0 || bytes < 1) return bytes;
+  const factor = 1 + tolerance;
+  const bucket = Math.floor(Math.log(bytes) / Math.log(factor));
+  return Math.max(1, Math.round(factor ** (bucket + 0.5)));
+}
+function collapseKeyOf(r) {
+  return JSON.stringify([
+    r.large,
+    r.modality ?? null,
+    r.large ? r.bytes : sizeKbOf(r.bytes),
+    r.tags ?? null
+  ]);
+}
+function accumulate(byKey, r) {
+  const key = collapseKeyOf(r);
+  const existing = byKey.get(key);
+  if (existing) {
+    existing.count++;
+  } else {
+    byKey.set(key, {
+      modality: r.modality,
+      large: r.large,
+      bytes: r.bytes,
+      tags: r.tags,
+      count: 1
+    });
+  }
+}
+function entryOf(acc) {
+  return {
+    ...acc.modality !== void 0 ? { modality: acc.modality } : {},
+    ...acc.large ? { type: "large-image", targetBytes: acc.bytes } : { type: "valid-image", targetSizeKb: sizeKbOf(acc.bytes) },
+    ...acc.tags !== void 0 ? { tags: acc.tags } : {},
+    ...acc.count > 1 ? { count: acc.count } : {}
+  };
+}
+function collapseEntries(records) {
+  const byKey = /* @__PURE__ */ new Map();
+  for (const r of records) accumulate(byKey, r);
+  return [...byKey.values()].map(entryOf);
+}
+function tagValuesEqual(a, b) {
+  if (a === b) return true;
+  if (typeof a === "object" && a !== null && typeof b === "object" && b !== null)
+    return JSON.stringify(a) === JSON.stringify(b);
+  return false;
+}
+function tryInstancesForm(records) {
+  const count = records.length;
+  if (count === 0) return null;
+  const large = records[0]?.large ?? false;
+  const modality = records[0]?.modality;
+  const columns = /* @__PURE__ */ new Map();
+  for (let i = 0; i < count; i++) {
+    const r = records[i];
+    if (r.large !== large || r.modality !== modality) return null;
+    if (!r.tags) continue;
+    for (const [k, v] of Object.entries(r.tags)) {
+      let col = columns.get(k);
+      if (!col) {
+        col = { values: new Array(count), present: 0 };
+        columns.set(k, col);
+      }
+      col.values[i] = v;
+      col.present++;
+    }
+  }
+  if (columns.size === 0) return null;
+  const sharedTags = {};
+  const perInstanceTags = {};
+  for (const [k, col] of columns) {
+    if (col.present !== count) return null;
+    const first = col.values[0];
+    if (col.values.every((v) => tagValuesEqual(v, first))) {
+      sharedTags[k] = first;
+    } else {
+      perInstanceTags[k] = col.values;
+    }
+  }
+  if (Object.keys(perInstanceTags).length === 0) return null;
+  const instances = {
+    count,
+    ...large ? { targetBytes: records.map((r) => r.bytes) } : { targetSizeKb: records.map((r) => sizeKbOf(r.bytes)) },
+    ...modality !== void 0 ? { modality } : {},
+    // perInstanceTags is non-empty here (we returned null above otherwise).
+    tags: perInstanceTags
+  };
+  return {
+    instances,
+    ...Object.keys(sharedTags).length ? { tags: sharedTags } : {}
+  };
+}
+function buildSeriesSpec(records) {
+  return tryInstancesForm(records) ?? { entries: collapseEntries(records) };
+}
+function seriesSpecFromAccum(accum) {
+  return Array.isArray(accum) ? buildSeriesSpec(accum) : { entries: [...accum.values()].map(entryOf) };
+}
 function describeDirectory(dir, options = {}) {
   const keepKeywords = options.keepTags ? resolveKeepTags(options.keepTags) : [];
+  const keepTags = keepKeywords.length > 0;
+  const tolerance = options.sizeTolerance ?? 0;
+  if (!Number.isFinite(tolerance) || tolerance < 0) {
+    throw new Error(
+      `describe: sizeTolerance must be a non-negative fraction; got ${JSON.stringify(options.sizeTolerance)}`
+    );
+  }
   const studies = /* @__PURE__ */ new Map();
   const stats = { dicomFiles: 0, skipped: 0, unknownModality: 0 };
   for (const path of [...walkFiles(dir)].sort()) {
@@ -1477,8 +1697,9 @@ function describeDirectory(dir, options = {}) {
         stats.unknownModality++;
       }
     }
-    const sizeKb = Math.max(1, Math.round(size / 1024));
-    const tags = keepKeywords.length ? extractTags(record, keepKeywords) : void 0;
+    const tags = keepTags ? extractTags(record, keepKeywords) : void 0;
+    const bytes = large ? size : Math.min(bucketBytes(size, tolerance), MAX_PIXEL_BYTES);
+    const fileRecord = { large, bytes, modality, tags };
     let study = studies.get(studyUid);
     if (!study) {
       study = /* @__PURE__ */ new Map();
@@ -1486,37 +1707,18 @@ function describeDirectory(dir, options = {}) {
     }
     let series = study.get(seriesUid);
     if (!series) {
-      series = /* @__PURE__ */ new Map();
+      series = keepTags ? [] : /* @__PURE__ */ new Map();
       study.set(seriesUid, series);
     }
-    const collapseKey = JSON.stringify([
-      large,
-      modality ?? null,
-      large ? size : sizeKb,
-      tags ?? null
-    ]);
-    const existing = series.get(collapseKey);
-    if (existing) {
-      existing.count++;
+    if (Array.isArray(series)) {
+      series.push(fileRecord);
     } else {
-      series.set(collapseKey, { modality, large, bytes: size, tags, count: 1 });
+      accumulate(series, fileRecord);
     }
   }
-  const studySpecs = [...studies.values()].map((study) => {
-    const seriesSpecs = [...study.values()].map((series) => {
-      const entries = [...series.values()].map((acc) => ({
-        ...acc.modality !== void 0 ? { modality: acc.modality } : {},
-        ...acc.large ? { type: "large-image", targetBytes: acc.bytes } : {
-          type: "valid-image",
-          targetSizeKb: Math.max(1, Math.round(acc.bytes / 1024))
-        },
-        ...acc.tags !== void 0 ? { tags: acc.tags } : {},
-        ...acc.count > 1 ? { count: acc.count } : {}
-      }));
-      return { entries };
-    });
-    return { series: seriesSpecs };
-  });
+  const studySpecs = [...studies.values()].map((study) => ({
+    series: [...study.values()].map(seriesSpecFromAccum)
+  }));
   if (studySpecs.length === 0) {
     throw new Error(`describe: no DICOM series found in ${dir}`);
   }
@@ -1557,7 +1759,7 @@ function loadCaseById(casesJsonPath, id) {
 }
 
 // src/public-fixtures/fetch.ts
-import { createHash } from "node:crypto";
+import { createHash as createHash2 } from "node:crypto";
 import { existsSync, mkdirSync as mkdirSync4, readFileSync as readFileSync3, writeFileSync as writeFileSync2 } from "node:fs";
 import { homedir } from "node:os";
 import { join as join3 } from "node:path";
@@ -1566,7 +1768,7 @@ function caseCachePath(sha256, root = DEFAULT_CACHE_ROOT) {
   return join3(root, sha256, "file.dcm");
 }
 function verifySha256(buffer, expected) {
-  const h = createHash("sha256").update(buffer).digest("hex");
+  const h = createHash2("sha256").update(buffer).digest("hex");
   if (h !== expected) {
     throw new Error(
       `SHA256 mismatch: expected ${expected}, got ${h}. Upstream may have changed; bump metadata after review.`
@@ -1698,6 +1900,9 @@ function resolveParametricSpec(spec) {
     ...entries.length > 0 ? { entries } : {},
     studies,
     seed,
+    // Pass an explicit salt through; when absent, UIDs derive from `seed` so the
+    // resolved spec stays deterministic ("same seed → identical resolved spec").
+    ...validated.uidSalt !== void 0 ? { uidSalt: validated.uidSalt } : {},
     ...validated.layout !== void 0 ? { layout: validated.layout } : {}
   };
 }
@@ -1711,16 +1916,19 @@ export {
   fetchPublicCaseToCache,
   generateCollectionFromSpec,
   generateFile,
+  hashUid,
   loadCaseById,
   loadCasesFromJson,
   loadDefaultCases,
   previewCollection,
   resolveParametricSpec,
   resolveTagTemplates,
+  seedToSalt,
   streamLargeImage,
   validateDatasetSpec,
   validateParametricSpec,
   verifySha256,
+  withResolvedSalt,
   withResolvedSeed,
   writeCollectionFromSpec,
   writeLargeImageFile