dicom-synth 1.9.0 → 1.11.0

package/dist/esm/describe/describe.js

@@ -32,6 +32,25 @@ var dcmjs = loadDcmjs();
 var MAX_PIXEL_BYTES = 512 * 1024 * 1024;
 var MAX_STREAM_BYTES = 50 * 1024 * 1024 * 1024;
 
+// src/syntheticFixtures/tagTemplate.ts
+var TEMPLATE_VOCAB = [
+  "index",
+  "studyIndex",
+  "seriesIndex",
+  "instanceNumber"
+];
+var TEMPLATE_PART_RE = /\{\{|\}\}|\{([a-zA-Z]+)\}/g;
+function escapeTagTemplate(value) {
+  return value.replace(/\{/g, "{{").replace(/\}/g, "}}");
+}
+function findTemplateTokens(value) {
+  const names = [];
+  for (const m of value.matchAll(TEMPLATE_PART_RE)) {
+    if (m[1] !== void 0) names.push(m[1]);
+  }
+  return names;
+}
+
 // src/schema/validate.ts
 var TYPE_CAPABILITIES = {
   "valid-image": { image: true },
@@ -219,12 +238,102 @@ 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(`${path}: must be an object`);
   }
-  for (const key of Object.keys(tags)) {
+  for (const [key, value] of Object.entries(tags)) {
     validateTagKey(key, path);
+    validateTagValue(key, value, path);
+  }
+}
+function validateArrayOfLength(value, count, path) {
+  if (!Array.isArray(value)) {
+    throw new Error(`${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(`${path}.tags: must be an object`);
+    }
+    for (const [key, arr] of Object.entries(tags)) {
+      validateTagKey(key, `${path}.tags`);
+      validateArrayOfLength(arr, count, `${path}.tags.${key}`).forEach(
+        (value, i) => {
+          validateTagValue(key, value, `${path}.tags[${i}]`);
+        }
+      );
+    }
   }
 }
 function validateSeries(series, path) {
@@ -232,16 +341,27 @@ 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;
@@ -310,6 +430,13 @@ function validateDatasetSpec(raw) {
 var dcmjsAny = dcmjs;
 var HEX_TAG_RE2 = /^[0-9a-fA-F]{8}$/;
 var SUPPORTED_MODALITIES = /* @__PURE__ */ new Set(["CT", "PT", "MR", "CR"]);
+var RESERVED_KEEP_KEYWORDS = /* @__PURE__ */ new Set([
+  "Units",
+  "DecayCorrection",
+  "CorrectedImage",
+  "ScanningSequence",
+  "SequenceVariant"
+]);
 function buildHexToKeyword() {
   const map = /* @__PURE__ */ new Map();
   const nm = dcmjsAny.data.DicomMetaDictionary.nameMap;
@@ -325,20 +452,37 @@ function buildHexToKeyword() {
 function resolveKeepTags(keepTags) {
   const nameMap = dcmjsAny.data.DicomMetaDictionary.nameMap;
   let hexToKeyword;
-  return keepTags.map((tag) => {
+  const kept = [];
+  for (const tag of keepTags) {
+    let keyword;
     if (HEX_TAG_RE2.test(tag)) {
       hexToKeyword ?? (hexToKeyword = buildHexToKeyword());
-      const keyword = hexToKeyword.get(tag.toUpperCase());
-      if (!keyword) {
+      const mapped = hexToKeyword.get(tag.toUpperCase());
+      if (!mapped) {
         throw new Error(`describe: unknown hex tag "${tag}"`);
       }
-      return keyword;
+      keyword = mapped;
+    } else {
+      if (nameMap && !Object.hasOwn(nameMap, tag)) {
+        throw new Error(`describe: unknown tag keyword "${tag}"`);
+      }
+      keyword = tag;
+    }
+    if (RESERVED_KEEP_KEYWORDS.has(keyword)) {
+      console.warn(
+        `describe: ignoring keep-tag "${keyword}" \u2014 modality-coupled tags are reproduced from the modality field`
+      );
+      continue;
     }
-    if (nameMap && !Object.hasOwn(nameMap, tag)) {
-      throw new Error(`describe: unknown tag keyword "${tag}"`);
+    if (nameMap?.[keyword]?.vr === "UI") {
+      console.warn(
+        `describe: ignoring keep-tag "${keyword}" \u2014 UID-valued tags are not kept verbatim`
+      );
+      continue;
     }
-    return tag;
-  });
+    kept.push(keyword);
+  }
+  return kept;
 }
 function* walkFiles(dir) {
   for (const entry of readdirSync(dir, { withFileTypes: true })) {
@@ -430,15 +574,126 @@ function extractTags(record, keepKeywords) {
   let found = false;
   for (const keyword of keepKeywords) {
     const value = record[keyword];
-    if (value !== void 0) {
-      tags[keyword] = value;
-      found = true;
+    if (value === void 0) continue;
+    if (keyword === "Modality" && typeof value === "string" && SUPPORTED_MODALITIES.has(value)) {
+      continue;
     }
+    tags[keyword] = typeof value === "string" ? escapeTagTemplate(value) : value;
+    found = true;
   }
   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()) {
@@ -470,8 +725,9 @@ 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();
@@ -479,37 +735,18 @@ 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}`);
   }
@@ -519,5 +756,6 @@ function describeDirectory(dir, options = {}) {
 }
 export {
   describeDirectory,
-  readHeaderOnly
+  readHeaderOnly,
+  tagValuesEqual
 };