@delali/narsil 0.3.1-next.69 → 0.3.1-next.70

package/dist/index.browser.mjs

@@ -1904,23 +1904,14 @@ function generateId() {
   return `${hex.slice(0, 8)}-${hex.slice(8, 12)}-${hex.slice(12, 16)}-${hex.slice(16, 20)}-${hex.slice(20)}`;
 }
 
+// src/schema/constants.ts
+var MAX_NESTING_DEPTH = 4;
+var MAX_DOCUMENT_NESTING_DEPTH = 32;
+var DEFAULT_PATTERN_VALUE_LIMIT = 8192;
+var MIN_PATTERN_VALUE_LIMIT = 1;
+var MAX_PATTERN_VALUE_LIMIT = 65536;
+
 // src/schema/validator/shared.ts
-var SORTABLE_TEXT_FIELD_TYPE = "string:sortable";
-var SCALAR_FIELD_TYPES = /* @__PURE__ */ new Set([
-  "string",
-  SORTABLE_TEXT_FIELD_TYPE,
-  "number",
-  "boolean",
-  "enum",
-  "geopoint",
-  "string[]",
-  "number[]",
-  "boolean[]",
-  "enum[]"
-]);
-function isTextFieldType(fieldType) {
-  return fieldType === "string" || fieldType === SORTABLE_TEXT_FIELD_TYPE;
-}
 var VECTOR_PATTERN = /^vector\[(\d+)]$/;
 function isGeopointOrVectorType(fieldType) {
   return fieldType === "geopoint" || fieldType !== void 0 && VECTOR_PATTERN.test(fieldType);
@@ -1932,6 +1923,144 @@ function isPlainObject(value) {
   return typeof value === "object" && value !== null && !Array.isArray(value);
 }
 
+// src/schema/validator/field-type.ts
+var FIELD_OPTION_ORDER = ["sortable", "pattern", "partial"];
+var OPTIONS_BY_BASE = {
+  string: FIELD_OPTION_ORDER,
+  "string[]": FIELD_OPTION_ORDER,
+  verbatim: ["sortable"],
+  "verbatim[]": ["sortable"],
+  number: [],
+  "number[]": [],
+  boolean: [],
+  "boolean[]": [],
+  enum: [],
+  "enum[]": [],
+  geopoint: [],
+  vector: []
+};
+var TYPE_SEPARATOR = ":";
+function isNamedBase(name) {
+  return name !== "vector" && Object.hasOwn(OPTIONS_BY_BASE, name);
+}
+function describeAllowedOptions(baseName, allowed) {
+  if (allowed.length === 0) return `the "${baseName}" type has no options`;
+  const quoted = allowed.map((option) => `"${option}"`);
+  if (quoted.length === 1) return `the only option of the "${baseName}" type is ${quoted[0]}`;
+  return `the options of the "${baseName}" type are ${quoted.slice(0, -1).join(", ")}, and ${quoted[quoted.length - 1]}`;
+}
+function invalidType(path, type, message) {
+  throw new NarsilError(ErrorCodes.SCHEMA_INVALID_TYPE, message, { field: path, type });
+}
+function parseBase(baseName, path, type) {
+  if (isNamedBase(baseName)) return { base: baseName, dimension: 0 };
+  const vectorMatch = VECTOR_PATTERN.exec(baseName);
+  if (vectorMatch === null) {
+    invalidType(path, type, `Field "${path}" has an unsupported type: "${type}"`);
+  }
+  const dimension = Number.parseInt(vectorMatch[1], 10);
+  if (dimension <= 0) {
+    throw new NarsilError(
+      ErrorCodes.SCHEMA_INVALID_VECTOR_DIMENSION,
+      `Vector field "${path}" must have a positive dimension, got ${dimension}`,
+      { field: path, dimension }
+    );
+  }
+  return { base: "vector", dimension };
+}
+function parseFieldType(type, path) {
+  const separatorIndex = type.indexOf(TYPE_SEPARATOR);
+  const baseName = separatorIndex === -1 ? type : type.slice(0, separatorIndex);
+  const { base, dimension } = parseBase(baseName, path, type);
+  const allowed = OPTIONS_BY_BASE[base];
+  const options = /* @__PURE__ */ new Set();
+  if (separatorIndex !== -1) {
+    for (const option of type.slice(separatorIndex + 1).split(TYPE_SEPARATOR)) {
+      if (!allowed.includes(option)) {
+        invalidType(
+          path,
+          type,
+          `Field "${path}" has type "${type}" with the option "${option}", but ${describeAllowedOptions(baseName, allowed)}`
+        );
+      }
+      if (options.has(option)) {
+        invalidType(path, type, `Field "${path}" has the "${option}" option twice in type "${type}"`);
+      }
+      options.add(option);
+    }
+  }
+  const ordered = FIELD_OPTION_ORDER.filter((option) => options.has(option));
+  const verbatim = base === "verbatim" || base === "verbatim[]";
+  return {
+    type: [baseName, ...ordered].join(TYPE_SEPARATOR),
+    base,
+    sortable: options.has("sortable"),
+    pattern: verbatim || options.has("pattern"),
+    partial: options.has("partial"),
+    dimension
+  };
+}
+function parsedTypeOf(type) {
+  if (type === void 0) return void 0;
+  try {
+    return parseFieldType(type, "");
+  } catch {
+    return void 0;
+  }
+}
+function withSortableOption(parsed) {
+  if (parsed.sortable) return parsed.type;
+  const separatorIndex = parsed.type.indexOf(TYPE_SEPARATOR);
+  const baseName = separatorIndex === -1 ? parsed.type : parsed.type.slice(0, separatorIndex);
+  return parseFieldType(`${baseName}${TYPE_SEPARATOR}sortable${parsed.type.slice(baseName.length)}`, "").type;
+}
+function isListBase(base) {
+  return base.endsWith("[]");
+}
+function isWordIndexedBase(base) {
+  return base === "string" || base === "string[]";
+}
+function isSingleStringFieldType(fieldType) {
+  return parsedTypeOf(fieldType)?.base === "string";
+}
+function isWordIndexedFieldType(fieldType) {
+  const parsed = parsedTypeOf(fieldType);
+  return parsed !== void 0 && isWordIndexedBase(parsed.base);
+}
+function isStringValuedBase(base) {
+  return base === "string" || base === "string[]" || base === "verbatim" || base === "verbatim[]";
+}
+
+// src/core/ordering/code-points.ts
+function exceedsCodePoints(value, maximum) {
+  if (value.length <= maximum) return false;
+  let counted = 0;
+  let i = 0;
+  while (i < value.length) {
+    const unit = value.charCodeAt(i);
+    if (unit >= 55296 && unit < 56320 && i + 1 < value.length) {
+      const low = value.charCodeAt(i + 1);
+      if (low >= 56320 && low < 57344) i++;
+    }
+    i++;
+    counted++;
+    if (counted > maximum) return true;
+  }
+  return false;
+}
+function adjustUtf16Unit(unit) {
+  if (unit < 55296) return unit;
+  if (unit < 57344) return unit + 8192;
+  return unit - 2048;
+}
+function compareCodePoints(a, b) {
+  const shorter = a.length < b.length ? a.length : b.length;
+  let i = 0;
+  while (i < shorter && a.charCodeAt(i) === b.charCodeAt(i)) i++;
+  if (i === shorter) return a.length - b.length;
+  return adjustUtf16Unit(a.charCodeAt(i)) - adjustUtf16Unit(b.charCodeAt(i));
+}
+
 // src/schema/validator/field-values.ts
 function validateGeopoint(path, value) {
   if (!isPlainObject(value)) {
@@ -2024,10 +2153,19 @@ function validateTypedArray(path, value, elementType) {
     }
   }
 }
-function validateFieldValue(path, value, type) {
-  switch (type) {
+function validatePatternValue(path, value, patternValueLimit, index) {
+  if (!exceedsCodePoints(value, patternValueLimit)) return;
+  const subject = index === void 0 ? `The value of field "${path}"` : `Element ${index} of field "${path}"`;
+  throw new NarsilError(
+    ErrorCodes.DOC_VALIDATION_FAILED,
+    `${subject} is longer than ${patternValueLimit} code points, which is the pattern value limit of this index`,
+    { field: path, limit: patternValueLimit, ...index === void 0 ? {} : { index } }
+  );
+}
+function validateFieldValue(path, value, type, patternValueLimit) {
+  switch (type.base) {
     case "string":
-    case SORTABLE_TEXT_FIELD_TYPE:
+    case "verbatim":
       if (typeof value !== "string") {
         throw new NarsilError(
           ErrorCodes.DOC_VALIDATION_FAILED,
@@ -2039,6 +2177,7 @@ function validateFieldValue(path, value, type) {
           }
         );
       }
+      if (type.pattern) validatePatternValue(path, value, patternValueLimit);
       return;
     case "number":
       if (typeof value !== "number" || Number.isNaN(value)) {
@@ -2079,7 +2218,12 @@ function validateFieldValue(path, value, type) {
       validateGeopoint(path, value);
       return;
     case "string[]":
+    case "verbatim[]":
       validateTypedArray(path, value, "string");
+      if (type.pattern) {
+        const elements = value;
+        for (let i = 0; i < elements.length; i++) validatePatternValue(path, elements[i], patternValueLimit, i);
+      }
       return;
     case "number[]":
       validateTypedArray(path, value, "number");
@@ -2090,19 +2234,12 @@ function validateFieldValue(path, value, type) {
     case "enum[]":
       validateTypedArray(path, value, "string");
       return;
-    default: {
-      const vectorMatch = VECTOR_PATTERN.exec(type);
-      if (vectorMatch) {
-        validateVector(path, value, Number.parseInt(vectorMatch[1], 10));
-      }
-    }
+    case "vector":
+      validateVector(path, value, type.dimension);
+      return;
   }
 }
 
-// src/schema/constants.ts
-var MAX_NESTING_DEPTH = 4;
-var MAX_DOCUMENT_NESTING_DEPTH = 32;
-
 // src/schema/validator/storable.ts
 var RESERVED_OBJECT_KEYS = /* @__PURE__ */ new Set(["__proto__", "constructor", "prototype"]);
 function reject(path, reason) {
@@ -2180,32 +2317,44 @@ function vectorFieldPaths(schema) {
   vectorPathCache.set(schema, paths);
   return paths;
 }
-function validateDocumentFields(doc, schema, prefix) {
-  for (const [field, type] of Object.entries(schema)) {
+var parsedLevelCache = /* @__PURE__ */ new WeakMap();
+function parsedLevel(schema, prefix) {
+  let entries = parsedLevelCache.get(schema);
+  if (entries !== void 0) return entries;
+  entries = Object.entries(schema).map(([field, type]) => {
+    if (isPlainObject(type)) return { field, nested: type, type: null };
     const path = prefix ? `${prefix}.${field}` : field;
+    return { field, nested: null, type: parseFieldType(type, path) };
+  });
+  parsedLevelCache.set(schema, entries);
+  return entries;
+}
+function validateDocumentFields(doc, schema, prefix, patternValueLimit) {
+  for (const { field, nested, type } of parsedLevel(schema, prefix)) {
     const value = doc[field];
     if (value === void 0 || value === null) continue;
-    if (isPlainObject(type)) {
+    const path = prefix ? `${prefix}.${field}` : field;
+    if (nested !== null) {
       if (!isPlainObject(value)) {
         throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, `Field "${path}" expected an object`, {
           field: path,
           received: Array.isArray(value) ? "array" : typeof value
         });
       }
-      validateDocumentFields(value, type, path);
+      validateDocumentFields(value, nested, path, patternValueLimit);
       continue;
     }
-    validateFieldValue(path, value, type);
+    if (type !== null) validateFieldValue(path, value, type, patternValueLimit);
   }
 }
-function validateDocument(document2, schema) {
+function validateDocument(document2, schema, patternValueLimit = DEFAULT_PATTERN_VALUE_LIMIT) {
   if (!isPlainObject(document2)) {
     throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, "Document must be a plain object", {
       received: typeof document2
     });
   }
   assertStorableDocument(document2, vectorFieldPaths(schema));
-  validateDocumentFields(document2, schema, "");
+  validateDocumentFields(document2, schema, "", patternValueLimit);
 }
 function collectExtraFields(doc, schema, prefix, extras) {
   for (const key of Object.keys(doc)) {
@@ -2263,6 +2412,11 @@ function validateSchemaFields(schema, depth, prefix) {
       { path: prefix, maxDepth: MAX_NESTING_DEPTH }
     );
   }
+  let canonical = null;
+  const replace = (field, value) => {
+    if (canonical === null) canonical = { ...schema };
+    canonical[field] = value;
+  };
   for (const [field, type] of Object.entries(schema)) {
     const path = prefix ? `${prefix}.${field}` : field;
     if (PROTOTYPE_POLLUTION_KEYS.has(field)) {
@@ -2286,7 +2440,8 @@ function validateSchemaFields(schema, depth, prefix) {
       );
     }
     if (isPlainObject(type)) {
-      validateSchemaFields(type, depth + 1, path);
+      const nested = validateSchemaFields(type, depth + 1, path);
+      if (nested !== type) replace(field, nested);
       continue;
     }
     if (typeof type !== "string") {
@@ -2299,24 +2454,10 @@ function validateSchemaFields(schema, depth, prefix) {
         }
       );
     }
-    if (SCALAR_FIELD_TYPES.has(type)) continue;
-    const vectorMatch = VECTOR_PATTERN.exec(type);
-    if (vectorMatch) {
-      const dimension = Number.parseInt(vectorMatch[1], 10);
-      if (dimension <= 0) {
-        throw new NarsilError(
-          ErrorCodes.SCHEMA_INVALID_VECTOR_DIMENSION,
-          `Vector field "${path}" must have a positive dimension, got ${dimension}`,
-          { field: path, dimension }
-        );
-      }
-      continue;
-    }
-    throw new NarsilError(ErrorCodes.SCHEMA_INVALID_TYPE, `Field "${path}" has an unsupported type: "${type}"`, {
-      field: path,
-      type
-    });
+    const parsed = parseFieldType(type, path);
+    if (parsed.type !== type) replace(field, parsed.type);
   }
+  return canonical ?? schema;
 }
 function validateSchema(schema) {
   if (!isPlainObject(schema)) {
@@ -2327,7 +2468,19 @@ function validateSchema(schema) {
   if (Object.keys(schema).length === 0) {
     throw new NarsilError(ErrorCodes.SCHEMA_INVALID_TYPE, "Schema must define at least one field");
   }
-  validateSchemaFields(schema, 1, "");
+  return validateSchemaFields(schema, 1, "");
+}
+function requireValidFieldTypes(flatSchema) {
+  for (const [path, type] of Object.entries(flatSchema)) {
+    if (typeof type !== "string") {
+      throw new NarsilError(
+        ErrorCodes.SCHEMA_INVALID_TYPE,
+        `Field "${path}" has an invalid type definition: ${String(type)}`,
+        { field: path, type: String(type) }
+      );
+    }
+    parseFieldType(type, path);
+  }
 }
 function flattenRecursive(schema, prefix, result) {
   for (const [field, type] of Object.entries(schema)) {
@@ -2348,14 +2501,22 @@ function flattenSchema(schema) {
   flattenedSchemas.set(schema, Object.freeze(result));
   return result;
 }
+var parsedSchemas = /* @__PURE__ */ new WeakMap();
+function schemaFieldsOf(schema) {
+  const cached = parsedSchemas.get(schema);
+  if (cached !== void 0) return cached;
+  const fields = [];
+  for (const [path, type] of Object.entries(flattenSchema(schema))) {
+    fields.push({ ...parseFieldType(type, path), path });
+  }
+  const frozen = Object.freeze(fields);
+  parsedSchemas.set(schema, frozen);
+  return frozen;
+}
 function extractVectorFieldsFromSchema(schema) {
-  const flat = flattenSchema(schema);
   const result = /* @__PURE__ */ new Map();
-  for (const [fieldPath, fieldType] of Object.entries(flat)) {
-    const match = VECTOR_PATTERN.exec(fieldType);
-    if (match) {
-      result.set(fieldPath, Number.parseInt(match[1], 10));
-    }
+  for (const field of schemaFieldsOf(schema)) {
+    if (field.base === "vector") result.set(field.path, field.dimension);
   }
   return result;
 }
@@ -2494,6 +2655,9 @@ function metadataToWire(meta) {
       ...promotion.storage !== void 0 ? { storage: promotion.storage } : {}
     };
   }
+  if (meta.patternValueLimit !== void 0) {
+    wire.pattern_value_limit = meta.patternValueLimit;
+  }
   if (meta.indexUuid !== void 0) {
     wire.index_uuid = meta.indexUuid;
   }
@@ -2615,6 +2779,16 @@ function wireToMetadata(raw) {
       meta.vectorPromotion = restored;
     }
   }
+  if (raw.pattern_value_limit !== void 0) {
+    if (!isPositiveInteger2(raw.pattern_value_limit)) {
+      throw new NarsilError(
+        ErrorCodes.CONFIG_INVALID,
+        `The pattern value limit in the index metadata is ${String(raw.pattern_value_limit)}, but the limit must be a whole number of code points`,
+        { patternValueLimit: String(raw.pattern_value_limit) }
+      );
+    }
+    meta.patternValueLimit = raw.pattern_value_limit;
+  }
   if (typeof raw.index_uuid === "string" && raw.index_uuid.length > 0) {
     meta.indexUuid = raw.index_uuid;
   }
@@ -2709,6 +2883,7 @@ function wireToPartition(raw) {
       fieldLengths: doc.field_lengths ?? {}
     };
   }
+  requireValidFieldTypes(raw.schema ?? {});
   dropStoredVectorValues(documents, raw.schema ?? {});
   const invertedIndex = {};
   for (const [token, list] of Object.entries(raw.inverted_index ?? {})) {
@@ -2793,6 +2968,7 @@ function validateHnswMetric2(value) {
   return void 0;
 }
 function storedDocumentsOf(raw) {
+  requireValidFieldTypes(raw.schema ?? {});
   const documents = {};
   for (const [docId, doc] of Object.entries(raw.documents ?? {})) {
     documents[docId] = {
@@ -2883,20 +3059,6 @@ function deserializePayloadV2(data) {
   };
 }
 
-// src/core/ordering/code-points.ts
-function adjustUtf16Unit(unit) {
-  if (unit < 55296) return unit;
-  if (unit < 57344) return unit + 8192;
-  return unit - 2048;
-}
-function compareCodePoints(a, b) {
-  const shorter = a.length < b.length ? a.length : b.length;
-  let i = 0;
-  while (i < shorter && a.charCodeAt(i) === b.charCodeAt(i)) i++;
-  if (i === shorter) return a.length - b.length;
-  return adjustUtf16Unit(a.charCodeAt(i)) - adjustUtf16Unit(b.charCodeAt(i));
-}
-
 // src/core/ordering/fold-table.ts
 var SINGLE_FOLDS = "1t,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;2j,lj;b,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;2,w;1,w;1,w;1,w;1,w;1,w;1,w;y,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;4,1;2,1;2,1;3,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;3,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,-3d;1,1;2,1;2,1;2,-7g;2,5u;1,1;2,1;2,5q;1,1;2,5p;1,5p;1,1;3,27;1,5m;1,5n;1,1;2,5p;1,5r;2,5v;1,5t;1,1;4,5v;1,5x;2,5y;1,1;2,1;2,1;2,62;1,1;2,62;3,1;2,62;1,1;2,61;1,61;1,1;2,1;2,63;1,1;4,1;8,2;1,1;2,2;1,1;2,2;1,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;3,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;3,2;1,1;2,1;2,-2p;1,-1k;1,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,-3m;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;8,8bv;1,1;2,-4j;1,8bs;3,1;2,-5f;1,1x;1,1z;1,1;2,1;2,1;2,1;2,1;6v,38;17,1;2,1;4,1;9,38;7,12;2,11;1,11;1,11;2,1s;2,1r;1,1r;2,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;2,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;n,1;d,8;1,-u;1,-p;4,-f;1,-m;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,-1i;1,-1c;3,-1o;1,-1s;2,1;2,-7;1,1;3,-3m;1,-3m;1,-3m;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,28;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1d,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;a,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,f;1,1;2,1;2,1;2,1;2,1;2,1;2,1;3,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;3,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;28a,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;1,5ls;2,5ls;6,5ls;mj,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1oj,-4su;1,-4st;1,-4sk;1,-4si;1,-4si;1,-4sj;1,-4sc;1,-4ro;1,r7n;1,1;7,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;1,-2bk;3,-2bk;1,-2bk;1,-2bk;8x,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;7,-1m;5,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;a,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;9,-8;1,-8;1,-8;1,-8;1,-8;1,-8;b,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;9,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;9,-8;1,-8;1,-8;1,-8;1,-8;1,-8;c,-8;2,-8;2,-8;2,-8;9,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;1,-8;21,-8;1,-8;1,-22;1,-22;3,-5j9;a,-2e;1,-2e;1,-2e;1,-2e;d,-8;1,-8;1,-2s;1,-2s;d,-8;1,-8;1,-34;1,-34;1,-7;c,-3k;1,-3k;1,-3i;1,-3i;8b,-5st;4,-6gv;1,-6di;7,s;1a,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;1,g;k,1;mr,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1,q;1f5,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1,1c;1d,1;2,-8af;1,-2xy;1,-89z;3,1;2,1;2,1;2,-8bg;1,-8al;1,-8bj;1,-8bi;2,1;3,1;9,-8cf;1,-8cf;1,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;9,1;2,1;5,1;nym,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;k,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;3s,1;2,1;2,1;2,1;2,1;2,1;2,1;4,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;b,1;2,1;2,-r9g;1,1;2,1;2,1;2,1;2,1;5,1;2,-wmg;3,1;2,1;4,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,-wn8;1,-wnj;1,-wnf;1,-wn5;1,-wn8;2,-wlu;1,-wmi;1,-wlx;1,ps;1,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,-1c;1,-wn7;1,-raw;1,1;2,1;2,-wo7;1,1;2,1;2,1;2,1;2,1;2,1;2,1;2,1;2,-wu9;p,1;or,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;1,-tzk;ggy,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;xy,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;3t,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;1,14;4d,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;2,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;1,13;2,13;1,13;1,13;1,13;1,13;1,13;1,13;2,13;1,13;1d7,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;1,1s;4e,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;27v,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;gw1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1,w;1t,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;1,r;o5k,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;1,y;";
 var MULTI_FOLDS = "67,37,37;29,2x,lj;p,jg,32;4n,2y,lo;bk,qh,lk,ld;w,qt,lk,ld;d3,12d,136;4y7,2w,mp;1,38,lk;1,3b,lm;1,3d,lm;1,2p,ji;4,37,37;4y,qt,lv;2,qt,lv,lc;2,qt,lv,ld;2,qt,lv,n6;16,64g,qh;1,64h,qh;1,64i,qh;1,64j,qh;1,64k,qh;1,64l,qh;1,64m,qh;1,64n,qh;1,64g,qh;1,64h,qh;1,64i,qh;1,64j,qh;1,64k,qh;1,64l,qh;1,64m,qh;1,64n,qh;1,65c,qh;1,65d,qh;1,65e,qh;1,65f,qh;1,65g,qh;1,65h,qh;1,65i,qh;1,65j,qh;1,65c,qh;1,65d,qh;1,65e,qh;1,65f,qh;1,65g,qh;1,65h,qh;1,65i,qh;1,65j,qh;1,674,qh;1,675,qh;1,676,qh;1,677,qh;1,678,qh;1,679,qh;1,67a,qh;1,67b,qh;1,674,qh;1,675,qh;1,676,qh;1,677,qh;1,678,qh;1,679,qh;1,67a,qh;1,67b,qh;3,67k,qh;1,q9,qh;1,q4,qh;2,q9,n6;1,q9,n6,qh;5,q9,qh;6,67o,qh;1,qf,qh;1,q6,qh;2,qf,n6;1,qf,n6,qh;5,qf,qh;6,qh,lk,lc;1,qh,lk,ld;3,qh,n6;1,qh,lk,n6;b,qt,lk,lc;1,qt,lk,ld;1,qp,lv;2,qt,n6;1,qt,lk,n6;b,67w,qh;1,qx,qh;1,r2,qh;2,qx,n6;1,qx,n6,qh;5,qx,qh;179g,2u,2u;1,2u,2x;1,2u,30;1,2u,2u,2x;1,2u,2u,30;1,37,38;1,37,38;d,12s,12u;1,12s,12d;1,12s,12j;1,132,12u;1,12s,12l;";
@@ -4185,12 +4347,12 @@ function stringArraysEqual(a, b) {
   }
   return true;
 }
-function textFieldsChanged(oldDoc, newDoc, flatSchema) {
-  for (const [path, fieldType] of Object.entries(flatSchema)) {
-    if (!isTextFieldType(fieldType) && fieldType !== "string[]") continue;
+function textFieldsChanged(oldDoc, newDoc, fields) {
+  for (const { path, base } of fields) {
+    if (!isWordIndexedBase(base)) continue;
     const oldVal = getNestedValue(oldDoc, path);
     const newVal = getNestedValue(newDoc, path);
-    if (fieldType !== "string[]") {
+    if (base === "string") {
       if (oldVal !== newVal) return true;
     } else {
       if (!stringArraysEqual(oldVal, newVal)) return true;
@@ -4919,14 +5081,14 @@ function createNumericIndex() {
 }
 
 // src/core/partition/indexing.ts
-function ensureFieldIndex(state, fieldPath, fieldType) {
-  if ((fieldType === "number" || fieldType === "number[]") && !state.numericIndexes.has(fieldPath)) {
+function ensureFieldIndex(state, fieldPath, base) {
+  if ((base === "number" || base === "number[]") && !state.numericIndexes.has(fieldPath)) {
     state.numericIndexes.set(fieldPath, createNumericIndex());
-  } else if ((fieldType === "boolean" || fieldType === "boolean[]") && !state.booleanIndexes.has(fieldPath)) {
+  } else if ((base === "boolean" || base === "boolean[]") && !state.booleanIndexes.has(fieldPath)) {
     state.booleanIndexes.set(fieldPath, createBooleanIndex());
-  } else if ((fieldType === "enum" || fieldType === "enum[]") && !state.enumIndexes.has(fieldPath)) {
+  } else if ((base === "enum" || base === "enum[]") && !state.enumIndexes.has(fieldPath)) {
     state.enumIndexes.set(fieldPath, createEnumIndex());
-  } else if (fieldType === "geopoint" && !state.geoIndexes.has(fieldPath)) {
+  } else if (base === "geopoint" && !state.geoIndexes.has(fieldPath)) {
     state.geoIndexes.set(fieldPath, createGeoIndex());
   }
 }
@@ -5052,26 +5214,26 @@ function indexStringArrayField(state, internalId, fieldPath, arr, language, opti
   applySurfaceCounts(state, surfaceCounts, 1);
   tokensByField[fieldPath] = fieldTokenList;
 }
-function indexDocument(state, docId, document2, flatSchema, language, options) {
+function indexDocument(state, docId, document2, fields, language, options) {
   const fieldLengths = {};
   const tokensByField = {};
   const internalId = resolveInternalId(state, docId);
-  for (const [fieldPath, fieldType] of Object.entries(flatSchema)) {
+  for (const { path: fieldPath, base } of fields) {
     const value = getNestedValue(document2, fieldPath);
     if (value === void 0 || value === null) continue;
-    ensureFieldIndex(state, fieldPath, fieldType);
-    if (isTextFieldType(fieldType)) {
+    ensureFieldIndex(state, fieldPath, base);
+    if (base === "string") {
       indexStringField(state, internalId, fieldPath, value, language, options, fieldLengths, tokensByField);
-    } else if (fieldType === "number") {
+    } else if (base === "number") {
       getOrCreateNumericIndex(state, fieldPath).insert(internalId, value);
-    } else if (fieldType === "boolean") {
+    } else if (base === "boolean") {
       getOrCreateBooleanIndex(state, fieldPath).insert(internalId, value);
-    } else if (fieldType === "enum") {
+    } else if (base === "enum") {
       state.enumIndexes.get(fieldPath)?.insert(internalId, value);
-    } else if (fieldType === "geopoint") {
+    } else if (base === "geopoint") {
       const geo = value;
       state.geoIndexes.get(fieldPath)?.insert(internalId, geo.lat, geo.lon);
-    } else if (fieldType === "string[]") {
+    } else if (base === "string[]") {
       indexStringArrayField(
         state,
         internalId,
@@ -5082,13 +5244,13 @@ function indexDocument(state, docId, document2, flatSchema, language, options) {
         fieldLengths,
         tokensByField
       );
-    } else if (fieldType === "number[]") {
+    } else if (base === "number[]") {
       const numIdx = getOrCreateNumericIndex(state, fieldPath);
       for (const num of value) numIdx.insert(internalId, num);
-    } else if (fieldType === "boolean[]") {
+    } else if (base === "boolean[]") {
       const boolIdx = getOrCreateBooleanIndex(state, fieldPath);
       for (const b of value) boolIdx.insert(internalId, b);
-    } else if (fieldType === "enum[]") {
+    } else if (base === "enum[]") {
       const enumIdx = state.enumIndexes.get(fieldPath);
       if (enumIdx) {
         for (const e of value) enumIdx.insert(internalId, e);
@@ -5097,16 +5259,16 @@ function indexDocument(state, docId, document2, flatSchema, language, options) {
   }
   return { fieldLengths, tokensByField };
 }
-function removeFromIndexes(state, docId, storedDoc, flatSchema, language, options) {
+function removeFromIndexes(state, docId, storedDoc, schemaFields, language, options) {
   const fieldLengths = {};
   const tokensByField = {};
   const fields = storedDoc.fields;
   const internalId = resolveInternalId(state, docId);
   const opts = tokenizeOptions(options);
-  for (const [fieldPath, fieldType] of Object.entries(flatSchema)) {
+  for (const { path: fieldPath, base } of schemaFields) {
     const value = getNestedValue(fields, fieldPath);
     if (value === void 0 || value === null) continue;
-    if (isTextFieldType(fieldType)) {
+    if (base === "string") {
       const result = tokenize(value, language, opts);
       fieldLengths[fieldPath] = result.tokens.length;
       const uniqueTokens = /* @__PURE__ */ new Set();
@@ -5121,15 +5283,15 @@ function removeFromIndexes(state, docId, storedDoc, flatSchema, language, option
       for (const token of uniqueTokens) state.invertedIdx.remove(token, internalId);
       applySurfaceCounts(state, surfaceCounts, -1);
       tokensByField[fieldPath] = fieldTokenList;
-    } else if (fieldType === "number") {
+    } else if (base === "number") {
       state.numericIndexes.get(fieldPath)?.remove(internalId, value);
-    } else if (fieldType === "boolean") {
+    } else if (base === "boolean") {
       state.booleanIndexes.get(fieldPath)?.remove(internalId, value);
-    } else if (fieldType === "enum") {
+    } else if (base === "enum") {
       state.enumIndexes.get(fieldPath)?.remove(internalId, value);
-    } else if (fieldType === "geopoint") {
+    } else if (base === "geopoint") {
       state.geoIndexes.get(fieldPath)?.remove(internalId);
-    } else if (fieldType === "string[]") {
+    } else if (base === "string[]") {
       const arr = value;
       const uniqueTokens = /* @__PURE__ */ new Set();
       const fieldTokenList = [];
@@ -5147,17 +5309,17 @@ function removeFromIndexes(state, docId, storedDoc, flatSchema, language, option
       for (const token of uniqueTokens) state.invertedIdx.remove(token, internalId);
       applySurfaceCounts(state, surfaceCounts, -1);
       tokensByField[fieldPath] = fieldTokenList;
-    } else if (fieldType === "number[]") {
+    } else if (base === "number[]") {
       const numIdx = state.numericIndexes.get(fieldPath);
       if (numIdx) {
         for (const num of value) numIdx.remove(internalId, num);
       }
-    } else if (fieldType === "boolean[]") {
+    } else if (base === "boolean[]") {
       const boolIdx = state.booleanIndexes.get(fieldPath);
       if (boolIdx) {
         for (const b of value) boolIdx.remove(internalId, b);
       }
-    } else if (fieldType === "enum[]") {
+    } else if (base === "enum[]") {
       const enumIdx = state.enumIndexes.get(fieldPath);
       if (enumIdx) {
         for (const e of value) enumIdx.remove(internalId, e);
@@ -5175,10 +5337,10 @@ function resolveInternalId2(state, docId) {
   }
   return internalId;
 }
-function updateFieldIndexOnly(state, docId, oldFields, newDoc, flatSchema) {
+function updateFieldIndexOnly(state, docId, oldFields, newDoc, fields) {
   const internalId = resolveInternalId2(state, docId);
-  for (const [fieldPath, fieldType] of Object.entries(flatSchema)) {
-    if (isTextFieldType(fieldType) || fieldType === "string[]") continue;
+  for (const { path: fieldPath, base: fieldType } of fields) {
+    if (isWordIndexedBase(fieldType)) continue;
     const oldVal = getNestedValue(oldFields, fieldPath);
     const newVal = getNestedValue(newDoc, fieldPath);
     if (oldVal === newVal) continue;
@@ -5333,6 +5495,16 @@ function requireFieldKind(fieldPath, operator, type, accepted) {
     operator
   );
 }
+function requireTextRangeField(fieldPath, operator, type) {
+  const parsed = parsedTypeOf(type);
+  if (parsed?.sortable && (parsed.base === "string" || parsed.base === "verbatim")) return;
+  const declared = type === void 0 ? `the schema has no field "${fieldPath}"` : `field "${fieldPath}" has type "${type}"`;
+  invalid(
+    `The engine applies "${operator}" with a text bound to a "string" or "verbatim" field whose type includes "sortable", but ${declared}`,
+    fieldPath,
+    operator
+  );
+}
 function requireRadius(fieldPath, radius) {
   const circle = radius;
   if (circle === null || typeof circle !== "object") {
@@ -5393,22 +5565,32 @@ function requireValidFilter(expression, fieldTypes, declaredOnly = false) {
 }
 function requireValidOperands(fieldPath, filter, type) {
   for (const operator of RANGE_OPERATORS) {
-    if (filter[operator] === void 0) continue;
-    if (!isFiniteNumber(filter[operator])) {
-      invalid(`"${operator}" on field "${fieldPath}" must be a finite number`, fieldPath, operator);
+    const bound = filter[operator];
+    if (bound === void 0) continue;
+    if (typeof bound === "string") {
+      requireTextRangeField(fieldPath, operator, type);
+      continue;
+    }
+    if (!isFiniteNumber(bound)) {
+      invalid(`"${operator}" on field "${fieldPath}" must be a finite number or a string`, fieldPath, operator);
     }
     requireFieldKind(fieldPath, operator, type, ["number"]);
   }
   if (filter.between !== void 0) {
     const bounds = filter.between;
-    if (!Array.isArray(bounds) || bounds.length !== 2 || !isFiniteNumber(bounds[0]) || !isFiniteNumber(bounds[1])) {
-      invalid(
-        `"between" on field "${fieldPath}" must be two finite numbers, the lower bound first`,
-        fieldPath,
-        "between"
-      );
+    const isPair = Array.isArray(bounds) && bounds.length === 2;
+    if (isPair && typeof bounds[0] === "string" && typeof bounds[1] === "string") {
+      requireTextRangeField(fieldPath, "between", type);
+    } else {
+      if (!isPair || !isFiniteNumber(bounds[0]) || !isFiniteNumber(bounds[1])) {
+        invalid(
+          `"between" on field "${fieldPath}" must be two finite numbers or two strings, the lower bound first`,
+          fieldPath,
+          "between"
+        );
+      }
+      requireFieldKind(fieldPath, "between", type, ["number"]);
     }
-    requireFieldKind(fieldPath, "between", type, ["number"]);
   }
   for (const operator of TEXT_OPERATORS) {
     if (filter[operator] === void 0) continue;
@@ -5668,6 +5850,14 @@ function applyGeoPolygonBitset(filter, geoIndex, capacity) {
 }
 
 // src/filters/evaluator.ts
+var UNBOUNDED = { lower: null, upper: null };
+function textBoundsOf(operator, bound) {
+  if (operator === "gt" || operator === "gte") {
+    return { ...UNBOUNDED, lower: bound, lowerInclusive: operator === "gte", upperInclusive: false };
+  }
+  return { ...UNBOUNDED, upper: bound, upperInclusive: operator === "lte", lowerInclusive: false };
+}
+var TEXT_RANGE_OPERATORS = ["gt", "gte", "lt", "lte"];
 var EXPRESSION_KEYS = new Set(FILTER_EXPRESSION_KEYS);
 var OPERATOR_KEYS = new Set(FIELD_FILTER_OPERATORS);
 function requireKnownKeys2(value, known, describe) {
@@ -5739,20 +5929,33 @@ function evaluateFieldFilter(fieldPath, filter, context) {
   if ("ne" in f && f.ne !== void 0) {
     bitsets.push(applyNeBitset(f.ne, getAllDocsBitset, capacity, getValue, fieldIndex));
   }
-  if ("gt" in f && f.gt !== void 0) {
+  for (const operator of TEXT_RANGE_OPERATORS) {
+    const bound = f[operator];
+    if (typeof bound === "string") bitsets.push(context.textRangeBitset(fieldPath, textBoundsOf(operator, bound)));
+  }
+  if (typeof f.gt === "number") {
     bitsets.push(applyGtBitset(f.gt, getAllDocsBitset, capacity, getValue, fieldIndex));
   }
-  if ("lt" in f && f.lt !== void 0) {
+  if (typeof f.lt === "number") {
     bitsets.push(applyLtBitset(f.lt, getAllDocsBitset, capacity, getValue, fieldIndex));
   }
-  if ("gte" in f && f.gte !== void 0) {
+  if (typeof f.gte === "number") {
     bitsets.push(applyGteBitset(f.gte, getAllDocsBitset, capacity, getValue, fieldIndex));
   }
-  if ("lte" in f && f.lte !== void 0) {
+  if (typeof f.lte === "number") {
     bitsets.push(applyLteBitset(f.lte, getAllDocsBitset, capacity, getValue, fieldIndex));
   }
-  if ("between" in f && f.between !== void 0) {
-    bitsets.push(applyBetweenBitset(f.between, getAllDocsBitset, capacity, getValue, fieldIndex));
+  if (Array.isArray(f.between)) {
+    const [low, high] = f.between;
+    if (typeof low === "string" && typeof high === "string") {
+      bitsets.push(
+        context.textRangeBitset(fieldPath, { lower: low, lowerInclusive: true, upper: high, upperInclusive: true })
+      );
+    } else {
+      bitsets.push(
+        applyBetweenBitset([low, high], getAllDocsBitset, capacity, getValue, fieldIndex)
+      );
+    }
   }
   if ("in" in f && f.in !== void 0) {
     bitsets.push(applyInBitset(f.in, getAllDocsBitset, capacity, getValue, fieldIndex));
@@ -5814,266 +6017,32 @@ function resolveGeoIndex(fieldPath, fieldIndex) {
   return fieldIndex.index;
 }
 
-// src/core/partition/filters.ts
-var EMPTY_GEO_INDEX = {
-  radiusQuery: () => /* @__PURE__ */ new Set(),
-  polygonQuery: () => /* @__PURE__ */ new Set()
-};
-function buildFilterContext(state, schema) {
-  const flat = getFlatSchema(state, schema);
-  const fieldIndexes = {};
-  const capacity = state.docStore.internalIdCapacity();
-  for (const [fieldPath, fieldType] of Object.entries(flat)) {
-    if (fieldType === "number" || fieldType === "number[]") {
-      const numIdx = state.numericIndexes.get(fieldPath);
-      if (numIdx) {
-        fieldIndexes[fieldPath] = {
-          type: "numeric",
-          index: {
-            eq: (v) => numIdx.queryEq(v),
-            gt: (v) => numIdx.queryGt(v),
-            gte: (v) => numIdx.queryGte(v),
-            lt: (v) => numIdx.queryLt(v),
-            lte: (v) => numIdx.queryLte(v),
-            between: (min, max) => numIdx.queryBetween(min, max),
-            allDocIds: () => numIdx.getAllDocIds(),
-            eqBitset: (v, cap) => numIdx.queryEqBitset(v, cap),
-            gtBitset: (v, cap) => numIdx.queryGtBitset(v, cap),
-            gteBitset: (v, cap) => numIdx.queryGteBitset(v, cap),
-            ltBitset: (v, cap) => numIdx.queryLtBitset(v, cap),
-            lteBitset: (v, cap) => numIdx.queryLteBitset(v, cap),
-            betweenBitset: (min, max, cap) => numIdx.queryBetweenBitset(min, max, cap),
-            allDocIdsBitset: (cap) => numIdx.getAllDocIdsBitset(cap)
-          }
-        };
-      }
-    } else if (fieldType === "boolean" || fieldType === "boolean[]") {
-      const boolIdx = state.booleanIndexes.get(fieldPath);
-      if (boolIdx) {
-        fieldIndexes[fieldPath] = {
-          type: "boolean",
-          index: {
-            getTrue: () => boolIdx.queryEq(true),
-            getFalse: () => boolIdx.queryEq(false),
-            allDocIds: () => boolIdx.getAllDocIds(),
-            getTrueBitset: (cap) => boolIdx.queryEqBitset(true, cap),
-            getFalseBitset: (cap) => boolIdx.queryEqBitset(false, cap),
-            allDocIdsBitset: (cap) => boolIdx.getAllDocIdsBitset(cap)
-          }
-        };
-      }
-    } else if (fieldType === "enum" || fieldType === "enum[]") {
-      const enumIdx = state.enumIndexes.get(fieldPath);
-      if (enumIdx) {
-        fieldIndexes[fieldPath] = {
-          type: "enum",
-          index: {
-            getDocIds: (v) => enumIdx.queryEq(v),
-            allDocIds: () => enumIdx.getAllDocIds(),
-            getDocIdsBitset: (v, cap) => enumIdx.queryEqBitset(v, cap),
-            getDocIdsInBitset: (values, cap) => enumIdx.queryInBitset(values, cap),
-            allDocIdsBitset: (cap) => enumIdx.getAllDocIdsBitset(cap)
-          }
-        };
-      }
-    } else if (fieldType === "geopoint") {
-      const geoIdx = state.geoIndexes.get(fieldPath);
-      fieldIndexes[fieldPath] = {
-        type: "geopoint",
-        index: geoIdx === void 0 ? EMPTY_GEO_INDEX : geoIdx
-      };
+// src/core/partition/constants.ts
+var TARGET_BLOCK_ENTRIES = 128;
+var PRUNING_REJECTION_SLACK = 1e-12;
+var SCORE_BUFFER_INITIAL_TOUCHED_CAPACITY = 1024;
+var SORT_COLUMN_MINIMUM_REBUILD_THRESHOLD = 1024;
+var SORT_COLUMN_REBUILD_FRACTION_SHIFT = 2;
+
+// src/core/partition/sort-columns/order.ts
+var MISSING_RANK = -1;
+var BYTES_PER_RANK = 4;
+var BYTES_PER_ORDERED_ENTRY = 4;
+function compareAscending(a, b) {
+  return compareComparableValues(a, b, "asc");
+}
+function buildOrder(store, liveIds, capacity) {
+  const counts = /* @__PURE__ */ new Map();
+  const present = [];
+  const missing2 = [];
+  for (const internalId of liveIds) {
+    const value = store.get(internalId);
+    if (value === null) {
+      missing2.push(internalId);
+      continue;
     }
-  }
-  let cachedAllDocIds = null;
-  let cachedAllDocIdsBitset = null;
-  return {
-    fieldTypes: flat,
-    fieldIndexes,
-    getFieldValue: (internalId, fieldPath) => getFieldValueByInternalId(state.docStore, internalId, fieldPath),
-    get allDocIds() {
-      if (!cachedAllDocIds) {
-        cachedAllDocIds = getAllInternalDocIds(state.docStore);
-      }
-      return cachedAllDocIds;
-    },
-    capacity,
-    get allDocIdsBitset() {
-      if (!cachedAllDocIdsBitset) {
-        cachedAllDocIdsBitset = bitsetFromSet(this.allDocIds, capacity);
-      }
-      return cachedAllDocIdsBitset;
-    }
-  };
-}
-function applyPartitionFilters(state, filters, schema) {
-  const context = buildFilterContext(state, schema);
-  const resultBitset = evaluateFilters(filters, context);
-  const resolver = state.docStore.resolver();
-  const externalResult = /* @__PURE__ */ new Set();
-  for (let wi = 0; wi < resultBitset.length; wi++) {
-    let word = resultBitset[wi];
-    if (word === 0) continue;
-    const base = wi << 5;
-    while (word !== 0) {
-      const tz = Math.clz32(word & -word) ^ 31;
-      const internalId = base + tz;
-      const externalId = resolver.toExternal(internalId);
-      if (externalId !== void 0) {
-        externalResult.add(externalId);
-      }
-      word &= word - 1;
-    }
-  }
-  return externalResult;
-}
-function applyPartitionFiltersBitset(state, filters, schema) {
-  const context = buildFilterContext(state, schema);
-  return evaluateFilters(filters, context);
-}
-function partitionFilterMatches(state, filters, schema) {
-  const accepted = applyPartitionFiltersBitset(state, filters, schema);
-  const docStore = state.docStore;
-  const resolver = docStore.resolver();
-  let count = 0;
-  for (let wordIndex = 0; wordIndex < accepted.length; wordIndex++) {
-    let word = accepted[wordIndex];
-    if (word === 0) continue;
-    const base = wordIndex << 5;
-    while (word !== 0) {
-      const trailingZeros = Math.clz32(word & -word) ^ 31;
-      if (resolver.toExternal(base + trailingZeros) !== void 0) count++;
-      word &= word - 1;
-    }
-  }
-  return {
-    count,
-    hasExternal(docId) {
-      const internalId = docStore.getInternalId(docId);
-      return internalId !== void 0 && bitsetHas(accepted, internalId);
-    },
-    hasInternal(internalId) {
-      return bitsetHas(accepted, internalId);
-    }
-  };
-}
-
-// src/core/partition/posting-columns.ts
-function postingColumns(list) {
-  const deletedDocs = list.deletedDocs;
-  return {
-    docIds: list.docIds,
-    termFrequencies: list.termFrequencies,
-    fieldNameIndices: list.fieldNameIndices,
-    deletedDocs,
-    hasDeleted: deletedDocs.size > 0,
-    count: list.length
-  };
-}
-
-// src/core/partition/matches.ts
-function markPostingList(matched, list, fieldNames, fields, filterBitset) {
-  const { docIds, fieldNameIndices, deletedDocs, hasDeleted, count } = postingColumns(list);
-  for (let pi = 0; pi < count; pi++) {
-    const internalId = docIds[pi];
-    if (hasDeleted && deletedDocs.has(internalId)) continue;
-    if (filterBitset && !bitsetHas(filterBitset, internalId)) continue;
-    if (fields && !fields.includes(fieldNames[fieldNameIndices[pi]])) continue;
-    bitsetSet(matched, internalId);
-  }
-}
-function searchFulltextMatches(state, params) {
-  const { queryTokens, prefixExpansion, fields, tolerance = 0, prefixLength = 2, exact = false, filterBitset } = params;
-  const fieldNames = state.fieldNameTable.names;
-  const matched = createBitSet(state.docStore.internalIdCapacity());
-  for (const qt of queryTokens) {
-    if (prefixExpansion && qt.token === prefixExpansion.token) {
-      const seen = /* @__PURE__ */ new Set();
-      for (const term of [qt.token, ...prefixExpansion.terms]) {
-        if (seen.has(term)) continue;
-        seen.add(term);
-        const postingList = state.invertedIdx.lookup(term);
-        if (postingList) markPostingList(matched, postingList, fieldNames, fields, filterBitset);
-      }
-      continue;
-    }
-    if (exact || tolerance === 0) {
-      const postingList = state.invertedIdx.lookup(qt.token);
-      if (postingList) markPostingList(matched, postingList, fieldNames, fields, filterBitset);
-      continue;
-    }
-    for (const match of state.invertedIdx.fuzzyLookup(qt.token, tolerance, prefixLength)) {
-      markPostingList(matched, match.postingList, fieldNames, fields, filterBitset);
-    }
-  }
-  const docStore = state.docStore;
-  const resolver = docStore.resolver();
-  let count = 0;
-  for (let wordIndex = 0; wordIndex < matched.length; wordIndex++) {
-    let word = matched[wordIndex];
-    if (word === 0) continue;
-    const base = wordIndex << 5;
-    while (word !== 0) {
-      const trailingZeros = Math.clz32(word & -word) ^ 31;
-      if (resolver.toExternal(base + trailingZeros) !== void 0) count++;
-      word &= word - 1;
-    }
-  }
-  return {
-    count,
-    hasExternal(docId) {
-      const internalId = docStore.getInternalId(docId);
-      return internalId !== void 0 && bitsetHas(matched, internalId);
-    },
-    hasInternal(internalId) {
-      return bitsetHas(matched, internalId);
-    },
-    matchedDocIds() {
-      const external = /* @__PURE__ */ new Set();
-      for (let wordIndex = 0; wordIndex < matched.length; wordIndex++) {
-        let word = matched[wordIndex];
-        if (word === 0) continue;
-        const base = wordIndex << 5;
-        while (word !== 0) {
-          const trailingZeros = Math.clz32(word & -word) ^ 31;
-          const externalId = resolver.toExternal(base + trailingZeros);
-          if (externalId !== void 0) external.add(externalId);
-          word &= word - 1;
-        }
-      }
-      return external;
-    },
-    ordinalBitset() {
-      return matched;
-    }
-  };
-}
-
-// src/core/partition/constants.ts
-var TARGET_BLOCK_ENTRIES = 128;
-var PRUNING_REJECTION_SLACK = 1e-12;
-var SCORE_BUFFER_INITIAL_TOUCHED_CAPACITY = 1024;
-var SORT_COLUMN_MINIMUM_REBUILD_THRESHOLD = 1024;
-var SORT_COLUMN_REBUILD_FRACTION_SHIFT = 2;
-
-// src/core/partition/sort-columns/order.ts
-var MISSING_RANK = -1;
-var BYTES_PER_RANK = 4;
-var BYTES_PER_ORDERED_ENTRY = 4;
-function compareAscending(a, b) {
-  return compareComparableValues(a, b, "asc");
-}
-function buildOrder(store, liveIds, capacity) {
-  const counts = /* @__PURE__ */ new Map();
-  const present = [];
-  const missing2 = [];
-  for (const internalId of liveIds) {
-    const value = store.get(internalId);
-    if (value === null) {
-      missing2.push(internalId);
-      continue;
-    }
-    present.push(internalId);
-    counts.set(value, (counts.get(value) ?? 0) + 1);
+    present.push(internalId);
+    counts.set(value, (counts.get(value) ?? 0) + 1);
   }
   const values = Array.from(counts.keys());
   values.sort(compareAscending);
@@ -6156,8 +6125,9 @@ function grow(length, wanted) {
   return next;
 }
 function kindForFieldType(fieldType) {
-  if (fieldType === "number") return "number";
-  if (fieldType === "boolean") return "boolean";
+  const base = parsedTypeOf(fieldType)?.base;
+  if (base === "number") return "number";
+  if (base === "boolean") return "boolean";
   return "mixed";
 }
 function createValueStore(initialKind) {
@@ -6280,7 +6250,8 @@ function createValueStore(initialKind) {
 
 // src/core/partition/sort-columns/index.ts
 function mayHoldList(fieldType) {
-  return fieldType === void 0 || fieldType.endsWith("[]");
+  const parsed = parsedTypeOf(fieldType);
+  return parsed === void 0 || isListBase(parsed.base);
 }
 function columnKeyOf(field, fieldType, mode) {
   return mayHoldList(fieldType) ? `${mode}:${field}` : `value:${field}`;
@@ -6416,6 +6387,48 @@ function createSortColumnSet(docStore) {
   };
 }
 
+// src/core/partition/sort-columns/range.ts
+function withinBounds(value, bounds) {
+  if (value === null) return false;
+  if (bounds.lower !== null) {
+    const comparison = compareComparableValues(value, bounds.lower, "asc");
+    if (comparison < 0 || comparison === 0 && !bounds.lowerInclusive) return false;
+  }
+  if (bounds.upper !== null) {
+    const comparison = compareComparableValues(value, bounds.upper, "asc");
+    if (comparison > 0 || comparison === 0 && !bounds.upperInclusive) return false;
+  }
+  return true;
+}
+function selectTextRange(column, requested, capacity, isLive2) {
+  const bounds = {
+    ...requested,
+    lower: requested.lower === null ? null : truncateSortString(requested.lower),
+    upper: requested.upper === null ? null : truncateSortString(requested.upper)
+  };
+  const order = column.order;
+  const ordered = order.ordered;
+  const result = createBitSet(capacity);
+  let start = 0;
+  if (bounds.lower !== null) {
+    const lowerRank = rankOfValue(order, bounds.lower);
+    start = column.seek(bounds.lowerInclusive ? lowerRank : lowerRank + 1, "asc");
+  }
+  let end = ordered.length;
+  if (bounds.upper !== null) {
+    const upperRank = rankOfValue(order, bounds.upper);
+    end = column.seek(bounds.upperInclusive ? upperRank + 1 : upperRank, "asc");
+  }
+  for (let position = start; position < end; position++) {
+    const internalId = ordered[position];
+    if (!column.isDirty(internalId) && isLive2(internalId)) bitsetSet(result, internalId);
+  }
+  for (const internalId of column.dirtyStream().present) {
+    if (withinBounds(column.valueOf(internalId), bounds)) bitsetSet(result, internalId);
+  }
+  return result;
+}
+
 // src/core/partition/sort-columns/select.ts
 function rankOrdersAfter(rank, other, direction) {
   if (rank === MISSING_RANK) return other !== MISSING_RANK;
@@ -6563,58 +6576,299 @@ function selectSortedPage(state, request) {
     if (page.length === limit && rankOrdersAfter(leadingRank, page[limit - 1].leadingRank, leadingDirection)) {
       break;
     }
-    const key = new Array(columns.length);
-    for (let i = 0; i < columns.length; i++) key[i] = columns[i].valueOf(internalId);
-    const candidate = { internalId, externalId, leadingRank, key };
-    if (anchorKey !== null && anchorId !== null) {
-      if (compareWithAnchor(columns, directions, candidate, anchorKey, anchorId) <= 0) continue;
+    const key = new Array(columns.length);
+    for (let i = 0; i < columns.length; i++) key[i] = columns[i].valueOf(internalId);
+    const candidate = { internalId, externalId, leadingRank, key };
+    if (anchorKey !== null && anchorId !== null) {
+      if (compareWithAnchor(columns, directions, candidate, anchorKey, anchorId) <= 0) continue;
+    }
+    if (page.length === limit) {
+      if (compareCandidates(columns, directions, candidate, page[limit - 1]) >= 0) continue;
+      page.pop();
+    }
+    insertOrdered(page, candidate, columns, directions);
+  }
+  return page.map((entry) => ({ id: entry.externalId, key: entry.key }));
+}
+
+// src/core/partition/sorting.ts
+function ensureColumns(state) {
+  if (state.sortColumns === null) {
+    state.sortColumns = createSortColumnSet(state.docStore);
+  }
+  return state.sortColumns;
+}
+function textRangeBitsetOf(state, fieldPath, fieldType, bounds, capacity) {
+  const column = ensureColumns(state).column(fieldPath, fieldType, "min");
+  const resolver = state.docStore.resolver();
+  return selectTextRange(column, bounds, capacity, (internalId) => resolver.toExternal(internalId) !== void 0);
+}
+function sortedPageOf(state, request) {
+  ensureColumns(state);
+  return selectSortedPage(state, request);
+}
+function sortValuesOf(state, docId, fields, fieldTypes, modes) {
+  const set = state.sortColumns;
+  const internalId = state.docStore.getInternalId(docId);
+  const stored = state.docStore.get(docId);
+  const key = new Array(fields.length);
+  for (let i = 0; i < fields.length; i++) {
+    const mode = modes[i] ?? "min";
+    if (set !== null && internalId !== void 0 && set.holds(fields[i], fieldTypes[i], mode)) {
+      key[i] = set.column(fields[i], fieldTypes[i], mode).valueOf(internalId);
+      continue;
+    }
+    key[i] = stored === void 0 ? null : toReducedSortValue(readSortField(stored.fields, fields[i]), mode);
+  }
+  return key;
+}
+function recordSortValues(state, internalId, document2) {
+  state.sortColumns?.record(internalId, document2);
+}
+function forgetSortValues(state, internalId) {
+  if (internalId === void 0) return;
+  state.sortColumns?.forget(internalId);
+}
+function refreshSortColumns(state) {
+  state.sortColumns?.refresh();
+}
+function sortColumnBytes(state) {
+  return state.sortColumns === null ? 0 : state.sortColumns.estimateBytes();
+}
+
+// src/core/partition/filters.ts
+var EMPTY_GEO_INDEX = {
+  radiusQuery: () => /* @__PURE__ */ new Set(),
+  polygonQuery: () => /* @__PURE__ */ new Set()
+};
+function buildFilterContext(state, schema) {
+  const flat = getFlatSchema(state, schema);
+  const fieldIndexes = {};
+  const capacity = state.docStore.internalIdCapacity();
+  for (const { path: fieldPath, base: fieldType } of schemaFieldsOf(schema)) {
+    if (fieldType === "number" || fieldType === "number[]") {
+      const numIdx = state.numericIndexes.get(fieldPath);
+      if (numIdx) {
+        fieldIndexes[fieldPath] = {
+          type: "numeric",
+          index: {
+            eq: (v) => numIdx.queryEq(v),
+            gt: (v) => numIdx.queryGt(v),
+            gte: (v) => numIdx.queryGte(v),
+            lt: (v) => numIdx.queryLt(v),
+            lte: (v) => numIdx.queryLte(v),
+            between: (min, max) => numIdx.queryBetween(min, max),
+            allDocIds: () => numIdx.getAllDocIds(),
+            eqBitset: (v, cap) => numIdx.queryEqBitset(v, cap),
+            gtBitset: (v, cap) => numIdx.queryGtBitset(v, cap),
+            gteBitset: (v, cap) => numIdx.queryGteBitset(v, cap),
+            ltBitset: (v, cap) => numIdx.queryLtBitset(v, cap),
+            lteBitset: (v, cap) => numIdx.queryLteBitset(v, cap),
+            betweenBitset: (min, max, cap) => numIdx.queryBetweenBitset(min, max, cap),
+            allDocIdsBitset: (cap) => numIdx.getAllDocIdsBitset(cap)
+          }
+        };
+      }
+    } else if (fieldType === "boolean" || fieldType === "boolean[]") {
+      const boolIdx = state.booleanIndexes.get(fieldPath);
+      if (boolIdx) {
+        fieldIndexes[fieldPath] = {
+          type: "boolean",
+          index: {
+            getTrue: () => boolIdx.queryEq(true),
+            getFalse: () => boolIdx.queryEq(false),
+            allDocIds: () => boolIdx.getAllDocIds(),
+            getTrueBitset: (cap) => boolIdx.queryEqBitset(true, cap),
+            getFalseBitset: (cap) => boolIdx.queryEqBitset(false, cap),
+            allDocIdsBitset: (cap) => boolIdx.getAllDocIdsBitset(cap)
+          }
+        };
+      }
+    } else if (fieldType === "enum" || fieldType === "enum[]") {
+      const enumIdx = state.enumIndexes.get(fieldPath);
+      if (enumIdx) {
+        fieldIndexes[fieldPath] = {
+          type: "enum",
+          index: {
+            getDocIds: (v) => enumIdx.queryEq(v),
+            allDocIds: () => enumIdx.getAllDocIds(),
+            getDocIdsBitset: (v, cap) => enumIdx.queryEqBitset(v, cap),
+            getDocIdsInBitset: (values, cap) => enumIdx.queryInBitset(values, cap),
+            allDocIdsBitset: (cap) => enumIdx.getAllDocIdsBitset(cap)
+          }
+        };
+      }
+    } else if (fieldType === "geopoint") {
+      const geoIdx = state.geoIndexes.get(fieldPath);
+      fieldIndexes[fieldPath] = {
+        type: "geopoint",
+        index: geoIdx === void 0 ? EMPTY_GEO_INDEX : geoIdx
+      };
+    }
+  }
+  let cachedAllDocIds = null;
+  let cachedAllDocIdsBitset = null;
+  return {
+    fieldTypes: flat,
+    fieldIndexes,
+    getFieldValue: (internalId, fieldPath) => getFieldValueByInternalId(state.docStore, internalId, fieldPath),
+    textRangeBitset: (fieldPath, bounds) => textRangeBitsetOf(state, fieldPath, flat[fieldPath], bounds, capacity),
+    get allDocIds() {
+      if (!cachedAllDocIds) {
+        cachedAllDocIds = getAllInternalDocIds(state.docStore);
+      }
+      return cachedAllDocIds;
+    },
+    capacity,
+    get allDocIdsBitset() {
+      if (!cachedAllDocIdsBitset) {
+        cachedAllDocIdsBitset = bitsetFromSet(this.allDocIds, capacity);
+      }
+      return cachedAllDocIdsBitset;
+    }
+  };
+}
+function applyPartitionFilters(state, filters, schema) {
+  const context = buildFilterContext(state, schema);
+  const resultBitset = evaluateFilters(filters, context);
+  const resolver = state.docStore.resolver();
+  const externalResult = /* @__PURE__ */ new Set();
+  for (let wi = 0; wi < resultBitset.length; wi++) {
+    let word = resultBitset[wi];
+    if (word === 0) continue;
+    const base = wi << 5;
+    while (word !== 0) {
+      const tz = Math.clz32(word & -word) ^ 31;
+      const internalId = base + tz;
+      const externalId = resolver.toExternal(internalId);
+      if (externalId !== void 0) {
+        externalResult.add(externalId);
+      }
+      word &= word - 1;
+    }
+  }
+  return externalResult;
+}
+function applyPartitionFiltersBitset(state, filters, schema) {
+  const context = buildFilterContext(state, schema);
+  return evaluateFilters(filters, context);
+}
+function partitionFilterMatches(state, filters, schema) {
+  const accepted = applyPartitionFiltersBitset(state, filters, schema);
+  const docStore = state.docStore;
+  const resolver = docStore.resolver();
+  let count = 0;
+  for (let wordIndex = 0; wordIndex < accepted.length; wordIndex++) {
+    let word = accepted[wordIndex];
+    if (word === 0) continue;
+    const base = wordIndex << 5;
+    while (word !== 0) {
+      const trailingZeros = Math.clz32(word & -word) ^ 31;
+      if (resolver.toExternal(base + trailingZeros) !== void 0) count++;
+      word &= word - 1;
+    }
+  }
+  return {
+    count,
+    hasExternal(docId) {
+      const internalId = docStore.getInternalId(docId);
+      return internalId !== void 0 && bitsetHas(accepted, internalId);
+    },
+    hasInternal(internalId) {
+      return bitsetHas(accepted, internalId);
+    }
+  };
+}
+
+// src/core/partition/posting-columns.ts
+function postingColumns(list) {
+  const deletedDocs = list.deletedDocs;
+  return {
+    docIds: list.docIds,
+    termFrequencies: list.termFrequencies,
+    fieldNameIndices: list.fieldNameIndices,
+    deletedDocs,
+    hasDeleted: deletedDocs.size > 0,
+    count: list.length
+  };
+}
+
+// src/core/partition/matches.ts
+function markPostingList(matched, list, fieldNames, fields, filterBitset) {
+  const { docIds, fieldNameIndices, deletedDocs, hasDeleted, count } = postingColumns(list);
+  for (let pi = 0; pi < count; pi++) {
+    const internalId = docIds[pi];
+    if (hasDeleted && deletedDocs.has(internalId)) continue;
+    if (filterBitset && !bitsetHas(filterBitset, internalId)) continue;
+    if (fields && !fields.includes(fieldNames[fieldNameIndices[pi]])) continue;
+    bitsetSet(matched, internalId);
+  }
+}
+function searchFulltextMatches(state, params) {
+  const { queryTokens, prefixExpansion, fields, tolerance = 0, prefixLength = 2, exact = false, filterBitset } = params;
+  const fieldNames = state.fieldNameTable.names;
+  const matched = createBitSet(state.docStore.internalIdCapacity());
+  for (const qt of queryTokens) {
+    if (prefixExpansion && qt.token === prefixExpansion.token) {
+      const seen = /* @__PURE__ */ new Set();
+      for (const term of [qt.token, ...prefixExpansion.terms]) {
+        if (seen.has(term)) continue;
+        seen.add(term);
+        const postingList = state.invertedIdx.lookup(term);
+        if (postingList) markPostingList(matched, postingList, fieldNames, fields, filterBitset);
+      }
+      continue;
+    }
+    if (exact || tolerance === 0) {
+      const postingList = state.invertedIdx.lookup(qt.token);
+      if (postingList) markPostingList(matched, postingList, fieldNames, fields, filterBitset);
+      continue;
     }
-    if (page.length === limit) {
-      if (compareCandidates(columns, directions, candidate, page[limit - 1]) >= 0) continue;
-      page.pop();
+    for (const match of state.invertedIdx.fuzzyLookup(qt.token, tolerance, prefixLength)) {
+      markPostingList(matched, match.postingList, fieldNames, fields, filterBitset);
     }
-    insertOrdered(page, candidate, columns, directions);
-  }
-  return page.map((entry) => ({ id: entry.externalId, key: entry.key }));
-}
-
-// src/core/partition/sorting.ts
-function ensureColumns(state) {
-  if (state.sortColumns === null) {
-    state.sortColumns = createSortColumnSet(state.docStore);
   }
-}
-function sortedPageOf(state, request) {
-  ensureColumns(state);
-  return selectSortedPage(state, request);
-}
-function sortValuesOf(state, docId, fields, fieldTypes, modes) {
-  const set = state.sortColumns;
-  const internalId = state.docStore.getInternalId(docId);
-  const stored = state.docStore.get(docId);
-  const key = new Array(fields.length);
-  for (let i = 0; i < fields.length; i++) {
-    const mode = modes[i] ?? "min";
-    if (set !== null && internalId !== void 0 && set.holds(fields[i], fieldTypes[i], mode)) {
-      key[i] = set.column(fields[i], fieldTypes[i], mode).valueOf(internalId);
-      continue;
+  const docStore = state.docStore;
+  const resolver = docStore.resolver();
+  let count = 0;
+  for (let wordIndex = 0; wordIndex < matched.length; wordIndex++) {
+    let word = matched[wordIndex];
+    if (word === 0) continue;
+    const base = wordIndex << 5;
+    while (word !== 0) {
+      const trailingZeros = Math.clz32(word & -word) ^ 31;
+      if (resolver.toExternal(base + trailingZeros) !== void 0) count++;
+      word &= word - 1;
     }
-    key[i] = stored === void 0 ? null : toReducedSortValue(readSortField(stored.fields, fields[i]), mode);
   }
-  return key;
-}
-function recordSortValues(state, internalId, document2) {
-  state.sortColumns?.record(internalId, document2);
-}
-function forgetSortValues(state, internalId) {
-  if (internalId === void 0) return;
-  state.sortColumns?.forget(internalId);
-}
-function refreshSortColumns(state) {
-  state.sortColumns?.refresh();
-}
-function sortColumnBytes(state) {
-  return state.sortColumns === null ? 0 : state.sortColumns.estimateBytes();
+  return {
+    count,
+    hasExternal(docId) {
+      const internalId = docStore.getInternalId(docId);
+      return internalId !== void 0 && bitsetHas(matched, internalId);
+    },
+    hasInternal(internalId) {
+      return bitsetHas(matched, internalId);
+    },
+    matchedDocIds() {
+      const external = /* @__PURE__ */ new Set();
+      for (let wordIndex = 0; wordIndex < matched.length; wordIndex++) {
+        let word = matched[wordIndex];
+        if (word === 0) continue;
+        const base = wordIndex << 5;
+        while (word !== 0) {
+          const trailingZeros = Math.clz32(word & -word) ^ 31;
+          const externalId = resolver.toExternal(base + trailingZeros);
+          if (externalId !== void 0) external.add(externalId);
+          word &= word - 1;
+        }
+      }
+      return external;
+    },
+    ordinalBitset() {
+      return matched;
+    }
+  };
 }
 
 // src/core/partition/memory.ts
@@ -6789,11 +7043,11 @@ var EMPTY_STATS = {
   averageFieldLengths: {},
   docFrequencies: {}
 };
-function textFieldsOf(flatSchema) {
-  return Object.entries(flatSchema).filter(([, fieldType]) => isTextFieldType(fieldType) || fieldType === "string[]");
+function textFieldsOf(fields) {
+  return fields.filter((field) => isWordIndexedBase(field.base));
 }
 function rebuildTextIndex(state, schema, language, options) {
-  const textFields = textFieldsOf(getFlatSchema(state, schema));
+  const textFields = textFieldsOf(schemaFieldsOf(schema));
   state.invertedIdx.clear();
   state.surfaceRegistry.clear();
   state.stats.deserialize(EMPTY_STATS);
@@ -6805,10 +7059,10 @@ function rebuildTextIndex(state, schema, language, options) {
       const fields = stored.fields;
       const fieldLengths = {};
       const tokensByField = {};
-      for (const [fieldPath, fieldType] of textFields) {
+      for (const { path: fieldPath, base } of textFields) {
         const value = getNestedValue(fields, fieldPath);
         if (value === void 0 || value === null) continue;
-        if (isTextFieldType(fieldType)) {
+        if (base === "string") {
           indexStringField(
             state,
             internalId,
@@ -8485,18 +8739,17 @@ function createPartitionIndex(partitionId, trackPositions = true) {
         );
       }
       if (options?.validate !== false) {
-        validateDocument(document2, schema);
+        validateDocument(document2, schema, options?.patternValueLimit);
         if (options?.strict) {
           validateDocumentStrict(document2, schema);
         }
       }
       const internalId = state.docStore.ensureInternalId(docId);
-      const flatSchema = getFlatSchema(state, schema);
       const { fieldLengths, tokensByField } = indexDocument(
         state,
         docId,
         document2,
-        flatSchema,
+        schemaFieldsOf(schema),
         language,
         options
       );
@@ -8516,8 +8769,14 @@ function createPartitionIndex(partitionId, trackPositions = true) {
           partitionId
         });
       }
-      const flatSchema = getFlatSchema(state, schema);
-      const { fieldLengths, tokensByField } = removeFromIndexes(state, docId, stored, flatSchema, language, options);
+      const { fieldLengths, tokensByField } = removeFromIndexes(
+        state,
+        docId,
+        stored,
+        schemaFieldsOf(schema),
+        language,
+        options
+      );
       const internalId = state.docStore.getInternalId(docId);
       state.docStore.remove(docId);
       state.stats.removeDocument(fieldLengths, tokensByField);
@@ -8548,19 +8807,19 @@ function createPartitionIndex(partitionId, trackPositions = true) {
         });
       }
       if (options?.validate !== false) {
-        validateDocument(document2, schema);
+        validateDocument(document2, schema, options?.patternValueLimit);
         if (options?.strict) {
           validateDocumentStrict(document2, schema);
         }
       }
-      const flatSchema = getFlatSchema(state, schema);
-      const needsTextReindex = textFieldsChanged(stored.fields, document2, flatSchema);
+      const fields = schemaFieldsOf(schema);
+      const needsTextReindex = textFieldsChanged(stored.fields, document2, fields);
       if (needsTextReindex) {
         const { fieldLengths: oldFieldLengths, tokensByField: oldTokens } = removeFromIndexes(
           state,
           docId,
           stored,
-          flatSchema,
+          fields,
           language,
           options
         );
@@ -8573,7 +8832,7 @@ function createPartitionIndex(partitionId, trackPositions = true) {
           state,
           docId,
           document2,
-          flatSchema,
+          fields,
           language,
           options
         );
@@ -8581,7 +8840,7 @@ function createPartitionIndex(partitionId, trackPositions = true) {
         state.stats.addDocument(newFieldLengths, newTokens);
         recordSortValues(state, internalId, document2);
       } else {
-        updateFieldIndexOnly(state, docId, stored.fields, document2, flatSchema);
+        updateFieldIndexOnly(state, docId, stored.fields, document2, fields);
         state.docStore.store(docId, document2, stored.fieldLengths);
         const internalId = state.docStore.getInternalId(docId);
         if (internalId !== void 0) recordSortValues(state, internalId, document2);
@@ -8973,7 +9232,7 @@ function validateEmbeddingConfig(embeddingConfig, schema, instanceAdapter) {
           { field: source, targetField }
         );
       }
-      if (!isTextFieldType(sourceType)) {
+      if (!isSingleStringFieldType(sourceType)) {
         throw new NarsilError(
           ErrorCodes.EMBEDDING_CONFIG_INVALID,
           `Embedding source field "${source}" must be a string type, got "${sourceType}"`,
@@ -9389,10 +9648,11 @@ function requireSortableFields(sort, schema, strict = false) {
       );
     }
     requireSortMode(entry, fieldType);
-    if (fieldType === "string") {
+    const parsed = parsedTypeOf(fieldType);
+    if (parsed !== void 0 && isStringValuedBase(parsed.base) && !parsed.sortable) {
       throw new NarsilError(
         ErrorCodes.SEARCH_INVALID_FIELD,
-        `The engine sorts by text field "${entry.field}" only where the schema declares it "${SORTABLE_TEXT_FIELD_TYPE}", because ordering text takes far more memory per document than ordering a number`,
+        `The engine sorts by text field "${entry.field}" only where its type includes "sortable", as in "${withSortableOption(parsed)}", because it keeps up to the first ${SORT_VALUE_MAX_CODE_POINTS} code points of every document's value in memory to sort by a text field`,
         { field: entry.field, fieldType }
       );
     }
@@ -9426,22 +9686,6 @@ function invalidCursor(cursor, reason) {
   const shown = cursor.length > 100 ? `${cursor.slice(0, 100)}...` : cursor;
   return new NarsilError(ErrorCodes.SEARCH_INVALID_CURSOR, `Failed to decode cursor: ${reason}`, { cursor: shown });
 }
-function exceedsCodePoints(value, maximum) {
-  if (value.length <= maximum) return false;
-  let counted = 0;
-  let i = 0;
-  while (i < value.length) {
-    const unit = value.charCodeAt(i);
-    if (unit >= 55296 && unit < 56320 && i + 1 < value.length) {
-      const low = value.charCodeAt(i + 1);
-      if (low >= 56320 && low < 57344) i++;
-    }
-    i++;
-    counted++;
-    if (counted > maximum) return true;
-  }
-  return false;
-}
 function sortSignatureOf(sort) {
   const fields = normalizeSort(sort);
   if (fields.length === 0) return null;
@@ -9770,7 +10014,7 @@ function validateSearchFields(fields, flatSchema) {
     if (!fieldType) {
       throw new NarsilError(ErrorCodes.SEARCH_INVALID_FIELD, `Field "${field}" does not exist in the schema`, { field });
     }
-    if (!isTextFieldType(fieldType) && fieldType !== "string[]") {
+    if (!isWordIndexedFieldType(fieldType)) {
       throw new NarsilError(
         ErrorCodes.SEARCH_INVALID_FIELD,
         `Field "${field}" has type "${fieldType}" which cannot be used for full-text search`,
@@ -10638,9 +10882,11 @@ function resolvePartitionInsertOptions(config, analysis, options) {
   const applyStrict = config.strict === true;
   const applyAnalyzer = analysis.stopWords !== void 0 || analysis.customTokenizer !== void 0;
   const applySurfaces = config.surfaceForms !== false;
-  if (!applyStrict && !applyAnalyzer && !applySurfaces) return options;
+  const applyPatternLimit = config.patternValueLimit !== void 0;
+  if (!applyStrict && !applyAnalyzer && !applySurfaces && !applyPatternLimit) return options;
   const resolved = { ...options };
   if (applyStrict) resolved.strict = true;
+  if (applyPatternLimit) resolved.patternValueLimit = config.patternValueLimit;
   if (applyAnalyzer) {
     resolved.stopWordOverride = options?.stopWordOverride ?? analysis.stopWords;
     resolved.customTokenizer = options?.customTokenizer ?? analysis.customTokenizer;
@@ -10649,6 +10895,44 @@ function resolvePartitionInsertOptions(config, analysis, options) {
   return resolved;
 }
 
+// src/partitioning/manager/capacity.ts
+function createCapacityChecks(source) {
+  const { indexName, config } = source;
+  return {
+    assertCapacity(pendingWrites = 0, partitionCountCap) {
+      const maxDocsPerPartition = config.partitions?.maxDocsPerPartition;
+      if (maxDocsPerPartition === void 0) return;
+      const partitionCount = source.partitionCount();
+      const effectivePartitionCount = partitionCountCap === void 0 ? partitionCount : Math.min(partitionCount, partitionCountCap);
+      const totalCapacity = maxDocsPerPartition * effectivePartitionCount;
+      const currentCount = source.documentCount() + pendingWrites;
+      if (currentCount < totalCapacity) return;
+      throw new NarsilError(
+        ErrorCodes.PARTITION_CAPACITY_EXCEEDED,
+        `Index "${indexName}" has reached its capacity of ${totalCapacity} documents (${maxDocsPerPartition} per partition \xD7 ${effectivePartitionCount} partitions)`,
+        {
+          indexName,
+          currentCount,
+          totalCapacity,
+          maxDocsPerPartition,
+          partitionCount: effectivePartitionCount
+        }
+      );
+    },
+    assertPartitionCapacity(partitionId, pendingWrites = 0) {
+      const maxDocsPerPartition = config.partitions?.maxDocsPerPartition;
+      if (maxDocsPerPartition === void 0) return;
+      const currentCount = source.getPartition(partitionId).count() + pendingWrites;
+      if (currentCount < maxDocsPerPartition) return;
+      throw new NarsilError(
+        ErrorCodes.PARTITION_CAPACITY_EXCEEDED,
+        `Partition ${partitionId} of index "${indexName}" has reached its capacity of ${maxDocsPerPartition} documents`,
+        { indexName, partitionId, currentCount, maxDocsPerPartition }
+      );
+    }
+  };
+}
+
 // src/partitioning/manager/nested-values.ts
 function setNestedValue(obj, path, value) {
   if (!path.includes(".")) {
@@ -10795,27 +11079,17 @@ function createPartitionManager(indexName, config, language, router, initialPart
       }
       partitions.length = newCount;
     },
-    assertCapacity(pendingWrites = 0, partitionCountCap) {
-      const currentMaxDocs = config.partitions?.maxDocsPerPartition;
-      if (currentMaxDocs === void 0) return;
-      const effectivePartitionCount = partitionCountCap === void 0 ? partitions.length : Math.min(partitions.length, partitionCountCap);
-      const totalCapacity = currentMaxDocs * effectivePartitionCount;
-      if (manager.countDocuments() + pendingWrites >= totalCapacity) {
-        throw new NarsilError(
-          ErrorCodes.PARTITION_CAPACITY_EXCEEDED,
-          `Index "${indexName}" has reached its capacity of ${totalCapacity} documents (${currentMaxDocs} per partition \xD7 ${effectivePartitionCount} partitions)`,
-          {
-            indexName,
-            currentCount: docPartitionMap.size + pendingWrites,
-            totalCapacity,
-            maxDocsPerPartition: currentMaxDocs,
-            partitionCount: effectivePartitionCount
-          }
-        );
-      }
+    ...createCapacityChecks({
+      indexName,
+      config,
+      partitionCount: () => partitions.length,
+      documentCount: () => manager.countDocuments(),
+      getPartition: (partitionId) => manager.getPartition(partitionId)
+    }),
+    routePartition(docId) {
+      return router.route(docId, partitions.length);
     },
     insert(docId, document2, options) {
-      manager.assertCapacity();
       const pid = router.route(docId, partitions.length);
       const insertOpts = resolveInsertOptions(options);
       partitions[pid].insert(docId, document2, config.schema, language, insertOpts);
@@ -18802,7 +19076,7 @@ function refuseRepeatedDocIds(documents) {
 function validateAsAnInsertDoes(documents, schema, options) {
   if (options?.validate === false) return;
   for (const { document: document2 } of documents) {
-    validateDocument(document2, schema);
+    validateDocument(document2, schema, options?.patternValueLimit);
     if (options?.strict) validateDocumentStrict(document2, schema);
   }
 }
@@ -19012,7 +19286,7 @@ var FieldLengths = class {
 function buildSegmentPayload(documents, schema, language, options, trackPositions) {
   refuseRepeatedDocIds(documents);
   validateAsAnInsertDoes(documents, schema, options);
-  const fields = Object.entries(flattenSchema(schema));
+  const fields = schemaFieldsOf(schema);
   const tokenizing = tokenizeOptions(options);
   const postings = new TextPostings(trackPositions);
   const surfaces = new SurfaceCounts();
@@ -19030,10 +19304,10 @@ function buildSegmentPayload(documents, schema, language, options, trackPosition
   };
   for (let ordinal = 0; ordinal < documents.length; ordinal++) {
     const document2 = documents[ordinal].document;
-    for (const [fieldPath, fieldType] of fields) {
+    for (const { path: fieldPath, base: fieldType } of fields) {
       const value = getNestedValue(document2, fieldPath);
       if (value === void 0 || value === null) continue;
-      if (fieldType !== "string" && fieldType !== "string[]") {
+      if (!isWordIndexedBase(fieldType)) {
         indexFilterValue(filters, ordinal, fieldPath, fieldType, value);
         continue;
       }
@@ -20882,6 +21156,7 @@ async function decodeSnapshotBundle(data) {
   if (!raw.schema || typeof raw.schema !== "object") {
     throw new NarsilError(ErrorCodes.PERSISTENCE_LOAD_FAILED, "Snapshot bundle missing schema");
   }
+  requireValidFieldTypes(raw.schema);
   if (typeof raw.language !== "string") {
     throw new NarsilError(ErrorCodes.PERSISTENCE_LOAD_FAILED, "Snapshot bundle missing language");
   }
@@ -21832,6 +22107,7 @@ function validateManifestCommon(raw) {
   if (typeof raw.language !== "string") {
     throw new NarsilError(ErrorCodes.PERSISTENCE_LOAD_FAILED, "Segment manifest is missing language");
   }
+  requireValidFieldTypes(raw.schema);
   return { schema: raw.schema, language: raw.language };
 }
 function normalizeCheckpoint2(raw) {
@@ -22252,6 +22528,9 @@ function reconstructSchemaFromMetadata(metadata) {
   if (metadata.vectorPromotion !== void 0) {
     config.vectorPromotion = metadata.vectorPromotion;
   }
+  if (metadata.patternValueLimit !== void 0) {
+    config.patternValueLimit = metadata.patternValueLimit;
+  }
   return config;
 }
 function isRecord4(value) {
@@ -24767,6 +25046,9 @@ function buildMetadata(indexName, hooks, documentCount) {
   if (config.vectorPromotion !== void 0) {
     metadata.vectorPromotion = config.vectorPromotion;
   }
+  if (config.patternValueLimit !== void 0) {
+    metadata.patternValueLimit = config.patternValueLimit;
+  }
   if (config.indexUuid !== void 0) {
     metadata.indexUuid = config.indexUuid;
   }
@@ -24880,7 +25162,8 @@ function createDurabilityFromTier(tier, wiring) {
         ...entry.config.trackPositions !== void 0 ? { trackPositions: entry.config.trackPositions } : {},
         ...entry.config.strict !== void 0 ? { strict: entry.config.strict } : {},
         ...entry.config.required !== void 0 ? { required: entry.config.required } : {},
-        ...entry.config.vectorPromotion !== void 0 ? { vectorPromotion: entry.config.vectorPromotion } : {}
+        ...entry.config.vectorPromotion !== void 0 ? { vectorPromotion: entry.config.vectorPromotion } : {},
+        ...entry.config.patternValueLimit !== void 0 ? { patternValueLimit: entry.config.patternValueLimit } : {}
       };
     },
     createIndexFromMetadata: wiring.createIndexFromMetadata,
@@ -24889,12 +25172,172 @@ function createDurabilityFromTier(tier, wiring) {
   });
 }
 
+// src/engine/constants.ts
+var MIN_DOCUMENTS_FOR_SEGMENTS = 64;
+var BATCH_CHUNK_SIZE = 1e3;
+var DEFAULT_LIMIT = DEFAULT_PAGE_SIZE;
+var DEFAULT_OFFSET = 0;
+var INDEX_NAME_PATTERN = /^[a-zA-Z0-9][a-zA-Z0-9._-]*$/;
+var ABSOLUTE_WINDOWS_PATH_PATTERN = /^[a-zA-Z]:[\\/]/;
+var MODULE_URL_SCHEME_PATTERN = /^(file|https?|data|node):/;
+var MAX_INDEX_NAME_LENGTH = 255;
+var MAX_DOC_ID_LENGTH = 512;
+var MIN_SWEEP_INTERVAL_MS = 100;
+var MAX_SWEEP_INTERVAL_MS = 6e4;
+var HEAP_PRESSURE_WARNING_FRACTION = 0.9;
+var HEAP_PRESSURE_REARM_FRACTION = 0.8;
+var STATISTICS_BROADCAST_INTERVAL_MS = 5e3;
+var FOREIGN_STATS_TTL_MS = 6e4;
+var DEFAULT_SUGGEST_LIMIT = 10;
+var MAX_SUGGEST_LIMIT = 100;
+
+// src/engine/validation.ts
+function now() {
+  if (typeof performance !== "undefined" && typeof performance.now === "function") {
+    return performance.now();
+  }
+  return Date.now();
+}
+function validateIndexName(name) {
+  const given = name;
+  if (typeof given !== "string" || name.length === 0) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "Index name must be a string of at least one character", {
+      received: given === null ? "null" : typeof given
+    });
+  }
+  if (name.length > MAX_INDEX_NAME_LENGTH) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, `Index name must not exceed ${MAX_INDEX_NAME_LENGTH} characters`, {
+      indexName: name,
+      length: name.length
+    });
+  }
+  if (!INDEX_NAME_PATTERN.test(name)) {
+    throw new NarsilError(
+      ErrorCodes.CONFIG_INVALID,
+      `Index name "${name}" contains invalid characters; use alphanumeric, dots, hyphens, and underscores only`,
+      { indexName: name }
+    );
+  }
+  if (name.includes("..")) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, `Index name "${name}" must not contain ".."`, { indexName: name });
+  }
+}
+function validateDocId(docId) {
+  const given = docId;
+  if (typeof given !== "string") {
+    throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, "Document ID must be a string", {
+      received: given === null ? "null" : Array.isArray(given) ? "array" : typeof given
+    });
+  }
+  if (docId.length === 0) {
+    throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, "Document ID must not be empty", { docId });
+  }
+  if (docId.length > MAX_DOC_ID_LENGTH) {
+    throw new NarsilError(
+      ErrorCodes.DOC_VALIDATION_FAILED,
+      `Document ID must not exceed ${MAX_DOC_ID_LENGTH} characters`,
+      {
+        docId,
+        length: docId.length
+      }
+    );
+  }
+  if (docId.includes("\0")) {
+    throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, "Document ID must not contain null bytes", { docId });
+  }
+}
+function validatePartitionConfig(partitions) {
+  const { maxDocsPerPartition, maxPartitions, watermark } = partitions;
+  if (maxDocsPerPartition !== void 0 && (!Number.isInteger(maxDocsPerPartition) || maxDocsPerPartition < 1)) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "partitions.maxDocsPerPartition must be a positive integer", {
+      maxDocsPerPartition
+    });
+  }
+  if (maxPartitions !== void 0 && (!Number.isInteger(maxPartitions) || maxPartitions < 1)) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "partitions.maxPartitions must be a positive integer", {
+      maxPartitions
+    });
+  }
+  if (watermark !== void 0 && (typeof watermark !== "number" || !(watermark > 0) || watermark > 1)) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "partitions.watermark must be above 0 and at most 1", {
+      watermark
+    });
+  }
+}
+function validatePatternValueLimit(limit) {
+  if (limit === void 0) return;
+  if (typeof limit !== "number" || !Number.isInteger(limit) || limit < MIN_PATTERN_VALUE_LIMIT || limit > MAX_PATTERN_VALUE_LIMIT) {
+    throw new NarsilError(
+      ErrorCodes.CONFIG_INVALID,
+      `patternValueLimit must be a whole number of code points from ${MIN_PATTERN_VALUE_LIMIT} to ${MAX_PATTERN_VALUE_LIMIT}`,
+      { patternValueLimit: typeof limit === "number" ? limit : String(limit) }
+    );
+  }
+}
+function requirePositiveSafeInteger(field, value) {
+  if (value === void 0) return;
+  if (!Number.isSafeInteger(value) || value < 1) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, `${field} must be a positive safe integer`, { field, value });
+  }
+}
+function validateBootstrapModule(bootstrapModule) {
+  if (bootstrapModule === void 0) return;
+  const absolute = bootstrapModule.startsWith("/") || ABSOLUTE_WINDOWS_PATH_PATTERN.test(bootstrapModule) || MODULE_URL_SCHEME_PATTERN.test(bootstrapModule);
+  if (absolute) return;
+  throw new NarsilError(
+    ErrorCodes.CONFIG_INVALID,
+    `workers.bootstrapModule is "${bootstrapModule}", which a worker would resolve against Narsil's own directory, so pass an absolute path or a URL such as new URL('./bootstrap.js', import.meta.url).href`,
+    { bootstrapModule }
+  );
+}
+function validateWorkerConfig(workers, lifecycle) {
+  if (workers === void 0) return;
+  requirePositiveSafeInteger("workers.count", workers.count);
+  requirePositiveSafeInteger("workers.promotionThreshold", workers.promotionThreshold);
+  requirePositiveSafeInteger("workers.idleTimeoutMs", workers.idleTimeoutMs);
+  if (workers.mainCopyQueries !== void 0 && workers.mainCopyQueries !== "lone" && workers.mainCopyQueries !== "none") {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "workers.mainCopyQueries must be 'lone' or 'none'", {
+      mainCopyQueries: workers.mainCopyQueries
+    });
+  }
+  validateBootstrapModule(workers.bootstrapModule);
+  const closeAfter = lifecycle?.idleTimeoutMs;
+  if (workers.idleTimeoutMs !== void 0 && closeAfter !== void 0 && workers.idleTimeoutMs > closeAfter) {
+    throw new NarsilError(
+      ErrorCodes.CONFIG_INVALID,
+      "workers.idleTimeoutMs must be at most lifecycle.idleTimeoutMs, so that an index drops its copies before it closes",
+      { workersIdleTimeoutMs: workers.idleTimeoutMs, lifecycleIdleTimeoutMs: closeAfter }
+    );
+  }
+}
+function validateBM25Params(bm25) {
+  if (bm25 === void 0) return;
+  const { k1, b } = bm25;
+  if (k1 !== void 0 && (!Number.isFinite(k1) || k1 < 0)) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "bm25.k1 must be a finite number at or above 0", { k1 });
+  }
+  if (b !== void 0 && (!Number.isFinite(b) || b < 0 || b > 1)) {
+    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "bm25.b must be between 0 and 1", { b });
+  }
+}
+function clampLimit(limit) {
+  return clampRowCount(limit, DEFAULT_LIMIT);
+}
+function clampOffset(offset) {
+  return clampRowCount(offset, DEFAULT_OFFSET);
+}
+
 // src/engine/vector-fields.ts
 function getVectorFieldPaths(schema) {
   return new Set(extractVectorFieldsFromSchema(schema).keys());
 }
 
 // src/engine/index-from-metadata.ts
+var RECOVERY_CONFIG_ERROR_CODES = /* @__PURE__ */ new Set([
+  ErrorCodes.CONFIG_INVALID,
+  ErrorCodes.SCHEMA_INVALID_TYPE,
+  ErrorCodes.SCHEMA_INVALID_VECTOR_DIMENSION
+]);
 function resolveEmbeddingAdapter(deps, metadata, schema) {
   const name = metadata.embedding?.adapter ?? null;
   if (!metadata.embedding) return { adapter: null, name };
@@ -24913,6 +25356,15 @@ function resolveEmbeddingAdapter(deps, metadata, schema) {
   }
   return { adapter: candidate, name };
 }
+function recoveryFailure(metadata, err) {
+  if (!(err instanceof NarsilError) || !RECOVERY_CONFIG_ERROR_CODES.has(err.code)) return err;
+  return new NarsilError(err.code, `Recovery of index "${metadata.indexName}" failed: ${err.message}`, {
+    ...err.details,
+    indexName: metadata.indexName,
+    tokenizer: metadata.tokenizer,
+    stopWords: metadata.stopWords
+  });
+}
 async function createIndexFromMetadata(deps, metadata, loadData) {
   const existing = deps.indexRegistry.get(metadata.indexName);
   if (existing !== void 0) {
@@ -24923,19 +25375,18 @@ async function createIndexFromMetadata(deps, metadata, loadData) {
   }
   const indexConfig = reconstructSchemaFromMetadata(metadata);
   const language = getLanguage(indexConfig.language ?? "english");
+  try {
+    indexConfig.schema = validateSchema(indexConfig.schema);
+    validatePatternValueLimit(indexConfig.patternValueLimit);
+  } catch (err) {
+    throw recoveryFailure(metadata, err);
+  }
   const embedding = resolveEmbeddingAdapter(deps, metadata, indexConfig.schema);
   try {
     validateVectorStorage(indexConfig.vectorPromotion, deps.filesystemDurability);
     if (loadData) deps.executor.createIndex(metadata.indexName, indexConfig, language);
   } catch (err) {
-    if (err instanceof NarsilError && err.code === ErrorCodes.CONFIG_INVALID) {
-      throw new NarsilError(err.code, `Recovery of index "${metadata.indexName}" failed: ${err.message}`, {
-        indexName: metadata.indexName,
-        tokenizer: metadata.tokenizer,
-        stopWords: metadata.stopWords
-      });
-    }
-    throw err;
+    throw recoveryFailure(metadata, err);
   }
   deps.indexRegistry.set(metadata.indexName, {
     config: indexConfig,
@@ -24960,25 +25411,6 @@ async function createIndexFromMetadata(deps, metadata, loadData) {
   }
 }
 
-// src/engine/constants.ts
-var MIN_DOCUMENTS_FOR_SEGMENTS = 64;
-var BATCH_CHUNK_SIZE = 1e3;
-var DEFAULT_LIMIT = DEFAULT_PAGE_SIZE;
-var DEFAULT_OFFSET = 0;
-var INDEX_NAME_PATTERN = /^[a-zA-Z0-9][a-zA-Z0-9._-]*$/;
-var ABSOLUTE_WINDOWS_PATH_PATTERN = /^[a-zA-Z]:[\\/]/;
-var MODULE_URL_SCHEME_PATTERN = /^(file|https?|data|node):/;
-var MAX_INDEX_NAME_LENGTH = 255;
-var MAX_DOC_ID_LENGTH = 512;
-var MIN_SWEEP_INTERVAL_MS = 100;
-var MAX_SWEEP_INTERVAL_MS = 6e4;
-var HEAP_PRESSURE_WARNING_FRACTION = 0.9;
-var HEAP_PRESSURE_REARM_FRACTION = 0.8;
-var STATISTICS_BROADCAST_INTERVAL_MS = 5e3;
-var FOREIGN_STATS_TTL_MS = 6e4;
-var DEFAULT_SUGGEST_LIMIT = 10;
-var MAX_SUGGEST_LIMIT = 100;
-
 // src/engine/index-state/constants.ts
 var DEFAULT_MAX_REOPEN_WAITERS = 64;
 var MAX_REOPEN_FAILURES = 5;
@@ -25561,21 +25993,10 @@ function createHeapPressureNotifier(deps) {
 }
 
 // src/engine/watermark.ts
-function createWatermarkNotifier(deps) {
+function createWatermarkNotifier(deps, scope = "index") {
   const latchedCapacity = /* @__PURE__ */ new Map();
-  function check(indexName) {
-    const partitionConfig = deps.getPartitionConfig(indexName);
-    const maxDocs = partitionConfig?.maxDocsPerPartition;
-    const watermark = partitionConfig?.watermark;
-    if (maxDocs === void 0 || watermark === void 0) {
-      latchedCapacity.delete(indexName);
-      return;
-    }
-    const manager = deps.getManager(indexName);
-    if (!manager) {
-      latchedCapacity.delete(indexName);
-      return;
-    }
+  const latchedPartitions = /* @__PURE__ */ new Map();
+  function checkIndex(indexName, manager, maxDocs, watermark) {
     const partitionCount = manager.partitionCount;
     const capacity = maxDocs * partitionCount;
     const documentCount = manager.countDocuments();
@@ -25588,19 +26009,63 @@ function createWatermarkNotifier(deps) {
     latchedCapacity.set(indexName, capacity);
     deps.emit({ indexName, documentCount, capacity, partitionCount });
   }
+  function checkPartitions(indexName, manager, maxDocs, watermark, insertedDocIds) {
+    const threshold = watermark * maxDocs;
+    const latched = latchedPartitions.get(indexName) ?? /* @__PURE__ */ new Set();
+    latchedPartitions.set(indexName, latched);
+    const checked = /* @__PURE__ */ new Set();
+    for (const docId of insertedDocIds) {
+      const partitionId = manager.partitionIdOf(docId) ?? manager.routePartition(docId);
+      if (checked.has(partitionId)) continue;
+      checked.add(partitionId);
+      const documentCount = manager.getPartition(partitionId).count();
+      if (documentCount < threshold) {
+        latched.delete(partitionId);
+        continue;
+      }
+      if (latched.has(partitionId)) continue;
+      latched.add(partitionId);
+      deps.emit({ indexName, documentCount, capacity: maxDocs, partitionCount: manager.partitionCount, partitionId });
+    }
+  }
+  function check(indexName, inserted) {
+    const partitionConfig = deps.getPartitionConfig(indexName);
+    const maxDocs = partitionConfig?.maxDocsPerPartition;
+    const watermark = partitionConfig?.watermark;
+    if (maxDocs === void 0 || watermark === void 0) {
+      forget(indexName);
+      return;
+    }
+    const manager = deps.getManager(indexName);
+    if (manager === void 0) {
+      forget(indexName);
+      return;
+    }
+    if (scope === "index") {
+      checkIndex(indexName, manager, maxDocs, watermark);
+      return;
+    }
+    if (inserted !== void 0) {
+      checkPartitions(indexName, manager, maxDocs, watermark, typeof inserted === "string" ? [inserted] : inserted);
+    }
+  }
   function forget(indexName) {
     latchedCapacity.delete(indexName);
+    latchedPartitions.delete(indexName);
   }
   return { check, forget };
 }
 
 // src/engine/notifiers.ts
 function wireWatermarkNotifier(wiring) {
-  return createWatermarkNotifier({
-    getManager: wiring.getManager,
-    getPartitionConfig: (indexName) => wiring.indexRegistry.get(indexName)?.config.partitions,
-    emit: (payload) => emitEngineEvent(wiring.eventHandlers, "partitionWatermark", payload)
-  });
+  return createWatermarkNotifier(
+    {
+      getManager: wiring.getManager,
+      getPartitionConfig: (indexName) => wiring.indexRegistry.get(indexName)?.config.partitions,
+      emit: (payload) => emitEngineEvent(wiring.eventHandlers, "partitionWatermark", payload)
+    },
+    wiring.capacityScope
+  );
 }
 function warnUnheardHeapPressure(payload) {
   console.warn(
@@ -27575,134 +28040,8 @@ function createWorkerOrchestrator(config, executor, indexRegistry, callbacks, ve
   };
 }
 
-// src/engine/validation.ts
-function now() {
-  if (typeof performance !== "undefined" && typeof performance.now === "function") {
-    return performance.now();
-  }
-  return Date.now();
-}
-function validateIndexName(name) {
-  const given = name;
-  if (typeof given !== "string" || name.length === 0) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "Index name must be a string of at least one character", {
-      received: given === null ? "null" : typeof given
-    });
-  }
-  if (name.length > MAX_INDEX_NAME_LENGTH) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, `Index name must not exceed ${MAX_INDEX_NAME_LENGTH} characters`, {
-      indexName: name,
-      length: name.length
-    });
-  }
-  if (!INDEX_NAME_PATTERN.test(name)) {
-    throw new NarsilError(
-      ErrorCodes.CONFIG_INVALID,
-      `Index name "${name}" contains invalid characters; use alphanumeric, dots, hyphens, and underscores only`,
-      { indexName: name }
-    );
-  }
-  if (name.includes("..")) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, `Index name "${name}" must not contain ".."`, { indexName: name });
-  }
-}
-function validateDocId(docId) {
-  const given = docId;
-  if (typeof given !== "string") {
-    throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, "Document ID must be a string", {
-      received: given === null ? "null" : Array.isArray(given) ? "array" : typeof given
-    });
-  }
-  if (docId.length === 0) {
-    throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, "Document ID must not be empty", { docId });
-  }
-  if (docId.length > MAX_DOC_ID_LENGTH) {
-    throw new NarsilError(
-      ErrorCodes.DOC_VALIDATION_FAILED,
-      `Document ID must not exceed ${MAX_DOC_ID_LENGTH} characters`,
-      {
-        docId,
-        length: docId.length
-      }
-    );
-  }
-  if (docId.includes("\0")) {
-    throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, "Document ID must not contain null bytes", { docId });
-  }
-}
-function validatePartitionConfig(partitions) {
-  const { maxDocsPerPartition, maxPartitions, watermark } = partitions;
-  if (maxDocsPerPartition !== void 0 && (!Number.isInteger(maxDocsPerPartition) || maxDocsPerPartition < 1)) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "partitions.maxDocsPerPartition must be a positive integer", {
-      maxDocsPerPartition
-    });
-  }
-  if (maxPartitions !== void 0 && (!Number.isInteger(maxPartitions) || maxPartitions < 1)) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "partitions.maxPartitions must be a positive integer", {
-      maxPartitions
-    });
-  }
-  if (watermark !== void 0 && (typeof watermark !== "number" || !(watermark > 0) || watermark > 1)) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "partitions.watermark must be above 0 and at most 1", {
-      watermark
-    });
-  }
-}
-function requirePositiveSafeInteger(field, value) {
-  if (value === void 0) return;
-  if (!Number.isSafeInteger(value) || value < 1) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, `${field} must be a positive safe integer`, { field, value });
-  }
-}
-function validateBootstrapModule(bootstrapModule) {
-  if (bootstrapModule === void 0) return;
-  const absolute = bootstrapModule.startsWith("/") || ABSOLUTE_WINDOWS_PATH_PATTERN.test(bootstrapModule) || MODULE_URL_SCHEME_PATTERN.test(bootstrapModule);
-  if (absolute) return;
-  throw new NarsilError(
-    ErrorCodes.CONFIG_INVALID,
-    `workers.bootstrapModule is "${bootstrapModule}", which a worker would resolve against Narsil's own directory, so pass an absolute path or a URL such as new URL('./bootstrap.js', import.meta.url).href`,
-    { bootstrapModule }
-  );
-}
-function validateWorkerConfig(workers, lifecycle) {
-  if (workers === void 0) return;
-  requirePositiveSafeInteger("workers.count", workers.count);
-  requirePositiveSafeInteger("workers.promotionThreshold", workers.promotionThreshold);
-  requirePositiveSafeInteger("workers.idleTimeoutMs", workers.idleTimeoutMs);
-  if (workers.mainCopyQueries !== void 0 && workers.mainCopyQueries !== "lone" && workers.mainCopyQueries !== "none") {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "workers.mainCopyQueries must be 'lone' or 'none'", {
-      mainCopyQueries: workers.mainCopyQueries
-    });
-  }
-  validateBootstrapModule(workers.bootstrapModule);
-  const closeAfter = lifecycle?.idleTimeoutMs;
-  if (workers.idleTimeoutMs !== void 0 && closeAfter !== void 0 && workers.idleTimeoutMs > closeAfter) {
-    throw new NarsilError(
-      ErrorCodes.CONFIG_INVALID,
-      "workers.idleTimeoutMs must be at most lifecycle.idleTimeoutMs, so that an index drops its copies before it closes",
-      { workersIdleTimeoutMs: workers.idleTimeoutMs, lifecycleIdleTimeoutMs: closeAfter }
-    );
-  }
-}
-function validateBM25Params(bm25) {
-  if (bm25 === void 0) return;
-  const { k1, b } = bm25;
-  if (k1 !== void 0 && (!Number.isFinite(k1) || k1 < 0)) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "bm25.k1 must be a finite number at or above 0", { k1 });
-  }
-  if (b !== void 0 && (!Number.isFinite(b) || b < 0 || b > 1)) {
-    throw new NarsilError(ErrorCodes.CONFIG_INVALID, "bm25.b must be between 0 and 1", { b });
-  }
-}
-function clampLimit(limit) {
-  return clampRowCount(limit, DEFAULT_LIMIT);
-}
-function clampOffset(offset) {
-  return clampRowCount(offset, DEFAULT_OFFSET);
-}
-
 // src/engine/core.ts
-function createEngineCore(config, hooks) {
+function createEngineCore(config, hooks, capacityScope = "index") {
   validateWorkerConfig(config?.workers, config?.lifecycle);
   const vectorWorkerCount = splitWorkerBudget(resolveWorkerCount(config?.workers?.count)).vector;
   const vectorCopyPolicy = {
@@ -27850,7 +28189,8 @@ function createEngineCore(config, hooks) {
   const notifierWiring = {
     eventHandlers,
     indexRegistry,
-    getManager: (indexName) => executor.getManager(indexName)
+    getManager: (indexName) => executor.getManager(indexName),
+    capacityScope
   };
   const watermarkNotifier = wireWatermarkNotifier(notifierWiring);
   const heapPressureNotifier = wireHeapPressureNotifier(notifierWiring);
@@ -27894,6 +28234,7 @@ function createEngineCore(config, hooks) {
     pendingRebalanceWrites: (indexName) => waqMap.get(indexName)?.size ?? 0,
     rebalanceTargetPartitionCount: (indexName) => rebalanceTargets.get(indexName),
     bufferedDocState: (indexName, docId) => waqMap.get(indexName)?.bufferedDocState(docId),
+    capacityScope,
     checkWatermark: watermarkNotifier.check,
     checkHeapPressure: heapPressureNotifier.check
   };
@@ -28026,13 +28367,16 @@ async function runLifecycleHook(core, hookName, context) {
 async function announceIndexCreated(core, name, config) {
   await runLifecycleHook(core, "onIndexCreate", { indexName: name, config });
 }
-async function createEngineIndex(core, config, name, indexConfig, indexUuid) {
+async function createEngineIndex(core, config, name, requestedConfig, creation = {}) {
+  const { indexUuid, adapterBinding = "required" } = creation;
   core.guardShutdown();
   validateIndexName(name);
   if (core.indexRegistry.has(name)) {
     throw new NarsilError(ErrorCodes.INDEX_ALREADY_EXISTS, `Index "${name}" already exists`, { indexName: name });
   }
-  validateSchema(indexConfig.schema);
+  const schema = validateSchema(requestedConfig.schema);
+  const indexConfig = schema === requestedConfig.schema ? requestedConfig : { ...requestedConfig, schema };
+  validatePatternValueLimit(indexConfig.patternValueLimit);
   validateVectorPromotion(indexConfig.vectorPromotion);
   validateVectorStorage(indexConfig.vectorPromotion, core.filesystemDurability);
   validateBM25Params(indexConfig.bm25);
@@ -28058,24 +28402,31 @@ async function createEngineIndex(core, config, name, indexConfig, indexUuid) {
   let resolvedEmbeddingAdapter = null;
   let embeddingAdapterName = null;
   if (indexConfig.embedding) {
-    let configuredAdapter = indexConfig.embedding.adapter;
+    const configuredAdapter = indexConfig.embedding.adapter;
+    const fields = indexConfig.embedding.fields;
     if (typeof configuredAdapter === "string") {
       embeddingAdapterName = configuredAdapter;
       const registered = core.embeddingAdapters.get(configuredAdapter);
-      if (!registered) {
+      if (registered !== void 0) {
+        resolvedEmbeddingAdapter = validateEmbeddingConfig(
+          { fields, adapter: registered },
+          indexConfig.schema,
+          config?.embedding
+        );
+      } else if (adapterBinding === "required") {
         throw new NarsilError(
           ErrorCodes.EMBEDDING_CONFIG_INVALID,
           `Embedding adapter "${configuredAdapter}" is not registered on this engine`,
           { adapter: configuredAdapter, available: [...core.embeddingAdapters.keys()] }
         );
       }
-      configuredAdapter = registered;
+    } else if (adapterBinding === "required" || configuredAdapter !== void 0 || config?.embedding !== void 0) {
+      resolvedEmbeddingAdapter = validateEmbeddingConfig(
+        { fields, adapter: configuredAdapter },
+        indexConfig.schema,
+        config?.embedding
+      );
     }
-    resolvedEmbeddingAdapter = validateEmbeddingConfig(
-      { fields: indexConfig.embedding.fields, adapter: configuredAdapter },
-      indexConfig.schema,
-      config?.embedding
-    );
   }
   if (indexConfig.required && indexConfig.required.length > 0) {
     validateRequiredFieldsInSchema(indexConfig.required, indexConfig.schema);
@@ -28088,7 +28439,7 @@ async function createEngineIndex(core, config, name, indexConfig, indexUuid) {
     embeddingAdapter: resolvedEmbeddingAdapter,
     embeddingAdapterName,
     vectorFieldPaths: getVectorFieldPaths(indexConfig.schema),
-    indexUuid: null,
+    indexUuid: indexUuid ?? null,
     heldPartitions: null,
     documentCount: 0,
     partitionCount: indexConfig.partitions?.maxPartitions ?? 1
@@ -28418,15 +28769,43 @@ function providedDocId(document2) {
 function asBatchInsertError(err) {
   return err instanceof NarsilError ? err : new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, String(err));
 }
-function admitInsert(ctx, indexName, manager, docId, pendingAdmitted = 0) {
-  if (!ctx.isRebalancing(indexName)) {
-    manager.assertCapacity(pendingAdmitted);
+function createPendingAdmissions() {
+  return { total: 0, byPartition: /* @__PURE__ */ new Map() };
+}
+function capsEachPartition(ctx, manager) {
+  return ctx.capacityScope === "partition" && manager.config.partitions?.maxDocsPerPartition !== void 0;
+}
+function recordPendingAdmission(ctx, pending, manager, docId) {
+  pending.total += 1;
+  if (!capsEachPartition(ctx, manager)) return;
+  const partitionId = manager.routePartition(docId);
+  pending.byPartition.set(partitionId, (pending.byPartition.get(partitionId) ?? 0) + 1);
+}
+function admitInsert(ctx, indexName, manager, docId, pending) {
+  const rebalancing = ctx.isRebalancing(indexName);
+  if (rebalancing) {
+    const bufferedState = ctx.bufferedDocState(indexName, docId);
+    const exists = bufferedState !== void 0 ? bufferedState === "present" : manager.has(docId);
+    if (exists) {
+      throw new NarsilError(ErrorCodes.DOC_ALREADY_EXISTS, `Document "${docId}" already exists`, { docId });
+    }
+  }
+  assertRoomFor(ctx, indexName, manager, docId, pending, rebalancing);
+}
+function admitBufferedUpsert(ctx, indexName, manager, docId) {
+  assertRoomFor(ctx, indexName, manager, docId, void 0, true);
+}
+function assertRoomFor(ctx, indexName, manager, docId, pending, rebalancing) {
+  if (ctx.capacityScope === "partition") {
+    if (!capsEachPartition(ctx, manager)) return;
+    const partitionId = manager.routePartition(docId);
+    manager.assertPartitionCapacity(partitionId, pending?.byPartition.get(partitionId) ?? 0);
     return;
   }
-  const bufferedState = ctx.bufferedDocState(indexName, docId);
-  const exists = bufferedState !== void 0 ? bufferedState === "present" : manager.has(docId);
-  if (exists) {
-    throw new NarsilError(ErrorCodes.DOC_ALREADY_EXISTS, `Document "${docId}" already exists`, { docId });
+  const pendingAdmitted = pending?.total ?? 0;
+  if (!rebalancing) {
+    manager.assertCapacity(pendingAdmitted);
+    return;
   }
   manager.assertCapacity(
     ctx.pendingRebalanceWrites(indexName) + pendingAdmitted,
@@ -28591,7 +28970,7 @@ async function insertDocument(ctx, indexName, document2, docId, options) {
     await applyInsert();
   }
   if (buffered) {
-    ctx.checkWatermark(indexName);
+    ctx.checkWatermark(indexName, resolvedDocId);
     ctx.checkHeapPressure(indexName);
     if (options?.wait === true) await awaitWriteVisibility(ctx, indexName);
     return resolvedDocId;
@@ -28615,7 +28994,7 @@ async function insertDocument(ctx, indexName, document2, docId, options) {
       vecIndex.scheduleBuild();
     }
   }
-  ctx.checkWatermark(indexName);
+  ctx.checkWatermark(indexName, resolvedDocId);
   ctx.checkHeapPressure(indexName);
   await ctx.orchestrator.scaleOutReadyIndexes();
   if (options?.wait === true) await awaitWriteVisibility(ctx, indexName);
@@ -28726,6 +29105,7 @@ async function admitBatchDocuments(ctx, indexName, documents, failed) {
   const required = entry.config.required;
   const admitted = [];
   const admittedIds = /* @__PURE__ */ new Set();
+  const pending = createPendingAdmissions();
   for (let chunkStart = 0; chunkStart < documents.length; chunkStart += BATCH_CHUNK_SIZE) {
     if (ctx.abortController.signal.aborted) break;
     const chunkEnd = Math.min(chunkStart + BATCH_CHUNK_SIZE, documents.length);
@@ -28758,15 +29138,16 @@ async function admitBatchDocuments(ctx, indexName, documents, failed) {
         if (extractedVectors.size > 0) {
           validateVectorDimensions(extractedVectors, vecIndexes);
         }
-        admitInsert(ctx, indexName, manager, docId, admitted.length);
+        admitInsert(ctx, indexName, manager, docId, pending);
         if (admittedIds.has(docId) || manager.has(docId)) {
           throw new NarsilError(ErrorCodes.DOC_ALREADY_EXISTS, `Document "${docId}" already exists`, { docId });
         }
-        validateDocument(partitionDoc, entry.config.schema);
+        validateDocument(partitionDoc, entry.config.schema, entry.config.patternValueLimit);
         if (strict) {
           validateDocumentStrict(partitionDoc, entry.config.schema);
         }
         admittedIds.add(docId);
+        recordPendingAdmission(ctx, pending, manager, docId);
         admitted.push({ docId, document: documents[i], partitionDoc, extractedVectors });
       } catch (err) {
         failed.push({ docId, error: asBatchInsertError(err) });
@@ -29165,7 +29546,7 @@ async function insertBatchViaSegments(ctx, indexName, documents, options) {
   for (const fieldPath of outcome.touchedVectorFields) {
     vecIndexes.get(fieldPath)?.scheduleBuild();
   }
-  ctx.checkWatermark(indexName);
+  ctx.checkWatermark(indexName, outcome.succeeded);
   ctx.checkHeapPressure(indexName);
   await ctx.orchestrator.scaleOutReadyIndexes();
   if (options?.wait === true) await awaitWriteVisibility(ctx, indexName);
@@ -29298,7 +29679,7 @@ async function insertDocumentBatch(ctx, indexName, documents, options) {
       vecIndex.scheduleBuild();
     }
   }
-  ctx.checkWatermark(indexName);
+  ctx.checkWatermark(indexName, succeeded);
   ctx.checkHeapPressure(indexName);
   await ctx.orchestrator.scaleOutReadyIndexes();
   if (options?.wait === true) await awaitWriteVisibility(ctx, indexName);
@@ -29531,7 +29912,7 @@ function applyOfUpdate(ctx, indexName, prepared) {
       const bufferedState = ctx.bufferedDocState(indexName, docId);
       const exists = bufferedState !== void 0 ? bufferedState === "present" : manager.has(docId);
       if (!exists) {
-        manager.assertCapacity(ctx.pendingRebalanceWrites(indexName), ctx.rebalanceTargetPartitionCount(indexName));
+        admitBufferedUpsert(ctx, indexName, manager, docId);
       }
     }
     if (ctx.bufferIfRebalancing(indexName, { action: "update", docId, document: document2, indexName })) {
@@ -29891,40 +30272,40 @@ async function executeRebalance(manager, indexName, targetPartitionCount, ctx) {
 }
 
 // src/engine/snapshot/restore-config.ts
-function invalid5(message) {
+function rejectSnapshot(message) {
   throw new NarsilError(ErrorCodes.DOC_VALIDATION_FAILED, `Invalid snapshot: ${message}`);
 }
 function isPlainObject3(value) {
   return value !== null && typeof value === "object" && !Array.isArray(value);
 }
-function restoredBm25(raw) {
+function restoredBm25(raw, reject2) {
   if (raw === void 0) {
     return void 0;
   }
   if (!isPlainObject3(raw)) {
-    invalid5("bm25 must be an object");
+    reject2("bm25 must be an object");
   }
   const { k1, b } = raw;
   if (k1 !== void 0 && (typeof k1 !== "number" || !Number.isFinite(k1))) {
-    invalid5("bm25 k1 must be a finite number");
+    reject2("bm25 k1 must be a finite number");
   }
   if (b !== void 0 && (typeof b !== "number" || !Number.isFinite(b))) {
-    invalid5("bm25 b must be a finite number");
+    reject2("bm25 b must be a finite number");
   }
   return {
     ...k1 !== void 0 ? { k1 } : {},
     ...b !== void 0 ? { b } : {}
   };
 }
-function restoredStopWords(name, list) {
+function restoredStopWords(name, list, reject2) {
   if (name !== void 0 && typeof name !== "string") {
-    invalid5("stopWords must be a name");
+    reject2("stopWords must be a name");
   }
   if (list !== void 0 && (!Array.isArray(list) || !list.every((word) => typeof word === "string"))) {
-    invalid5("stopWordList must be a list of words");
+    reject2("stopWordList must be a list of words");
   }
   if (typeof name === "string" && Array.isArray(list)) {
-    invalid5("stopWords and stopWordList cannot both be present");
+    reject2("stopWords and stopWordList cannot both be present");
   }
   if (typeof name === "string") {
     return name;
@@ -29934,22 +30315,22 @@ function restoredStopWords(name, list) {
   }
   return void 0;
 }
-function restoredPartitionConfig(raw) {
+function restoredPartitionConfig(raw, reject2 = rejectSnapshot) {
   if (raw === void 0) {
     return void 0;
   }
   if (!isPlainObject3(raw)) {
-    invalid5("partitionConfig must be an object");
+    reject2("partitionConfig must be an object");
   }
   const { maxDocsPerPartition, maxPartitions, watermark } = raw;
   if (maxDocsPerPartition !== void 0 && (typeof maxDocsPerPartition !== "number" || !Number.isInteger(maxDocsPerPartition) || maxDocsPerPartition < 1)) {
-    invalid5("partitionConfig maxDocsPerPartition must be a positive integer");
+    reject2("partitionConfig maxDocsPerPartition must be a positive integer");
   }
   if (maxPartitions !== void 0 && (typeof maxPartitions !== "number" || !Number.isInteger(maxPartitions) || maxPartitions < 1)) {
-    invalid5("partitionConfig maxPartitions must be a positive integer");
+    reject2("partitionConfig maxPartitions must be a positive integer");
   }
   if (watermark !== void 0 && (typeof watermark !== "number" || !(watermark > 0) || watermark > 1)) {
-    invalid5("partitionConfig watermark must be above 0 and at most 1");
+    reject2("partitionConfig watermark must be above 0 and at most 1");
   }
   return {
     ...typeof maxDocsPerPartition === "number" ? { maxDocsPerPartition } : {},
@@ -29957,61 +30338,61 @@ function restoredPartitionConfig(raw) {
     ...typeof watermark === "number" ? { watermark } : {}
   };
 }
-function restoredScoring(raw) {
+function restoredScoring(raw, reject2) {
   if (raw === void 0) {
     return void 0;
   }
   if (raw !== "local" && raw !== "dfs" && raw !== "broadcast") {
-    invalid5("defaultScoring must be 'local', 'dfs', or 'broadcast'");
+    reject2("defaultScoring must be 'local', 'dfs', or 'broadcast'");
   }
   return raw;
 }
-function restoredBoolean(raw, field) {
+function restoredBoolean(raw, field, reject2) {
   if (raw === void 0) {
     return void 0;
   }
   if (typeof raw !== "boolean") {
-    invalid5(`${field} must be a boolean`);
+    reject2(`${field} must be a boolean`);
   }
   return raw;
 }
-function restoredRequired(raw) {
+function restoredRequired(raw, reject2) {
   if (raw === void 0) {
     return void 0;
   }
   if (!Array.isArray(raw) || !raw.every((field) => typeof field === "string")) {
-    invalid5("required must be a list of field paths");
+    reject2("required must be a list of field paths");
   }
   return raw;
 }
-function restoredVectorPromotion(raw) {
+function restoredVectorPromotion(raw, reject2) {
   if (raw === void 0) {
     return void 0;
   }
   if (!isPlainObject3(raw) || raw.hnswConfig !== void 0 && !isPlainObject3(raw.hnswConfig)) {
-    invalid5("vectorPromotion must be an object");
+    reject2("vectorPromotion must be an object");
   }
   return raw;
 }
-function restoredEmbedding(raw) {
+function restoredEmbedding(raw, reject2 = rejectSnapshot) {
   if (raw === void 0) {
     return void 0;
   }
   if (!isPlainObject3(raw)) {
-    invalid5("embedding must be an object");
+    reject2("embedding must be an object");
   }
   const { fields, adapter } = raw;
   if (adapter !== void 0 && typeof adapter !== "string") {
-    invalid5("embedding adapter must be a name");
+    reject2("embedding adapter must be a name");
   }
   if (!isPlainObject3(fields)) {
-    invalid5("embedding fields must be an object");
+    reject2("embedding fields must be an object");
   }
   for (const value of Object.values(fields)) {
     const isPath = typeof value === "string";
     const isPathList = Array.isArray(value) && value.every((path) => typeof path === "string");
     if (!isPath && !isPathList) {
-      invalid5("embedding fields must map vector fields to source paths");
+      reject2("embedding fields must map vector fields to source paths");
     }
   }
   return {
@@ -30019,30 +30400,40 @@ function restoredEmbedding(raw) {
     ...adapter !== void 0 ? { adapter } : {}
   };
 }
-function restoredConfigFields(envelope) {
-  if (envelope.tokenizer !== void 0 && typeof envelope.tokenizer !== "string") {
-    invalid5("tokenizer must be a name");
+function readSharedConfigFields(source, reject2) {
+  if (source.tokenizer !== void 0 && typeof source.tokenizer !== "string") {
+    reject2("tokenizer must be a name");
+  }
+  const stopWords2 = restoredStopWords(source.stopWords, source.stopWordList, reject2);
+  const bm25 = restoredBm25(source.bm25, reject2);
+  const surfaceForms = restoredBoolean(source.surfaceForms, "surfaceForms", reject2);
+  const defaultScoring = restoredScoring(source.defaultScoring, reject2);
+  const trackPositions = restoredBoolean(source.trackPositions, "trackPositions", reject2);
+  const strict = restoredBoolean(source.strict, "strict", reject2);
+  const required = restoredRequired(source.required, reject2);
+  const vectorPromotion = restoredVectorPromotion(source.vectorPromotion, reject2);
+  if (source.patternValueLimit !== void 0 && typeof source.patternValueLimit !== "number") {
+    reject2("patternValueLimit must be a number");
   }
-  const stopWords2 = restoredStopWords(envelope.stopWords, envelope.stopWordList);
-  const bm25 = restoredBm25(envelope.bm25);
-  const surfaceForms = restoredBoolean(envelope.surfaceForms, "surfaceForms");
-  const partitions = restoredPartitionConfig(envelope.partitionConfig);
-  const defaultScoring = restoredScoring(envelope.defaultScoring);
-  const trackPositions = restoredBoolean(envelope.trackPositions, "trackPositions");
-  const strict = restoredBoolean(envelope.strict, "strict");
-  const required = restoredRequired(envelope.required);
-  const vectorPromotion = restoredVectorPromotion(envelope.vectorPromotion);
   return {
-    ...typeof envelope.tokenizer === "string" ? { tokenizer: envelope.tokenizer } : {},
+    ...typeof source.tokenizer === "string" ? { tokenizer: source.tokenizer } : {},
     ...stopWords2 !== void 0 ? { stopWords: stopWords2 } : {},
     ...bm25 !== void 0 ? { bm25 } : {},
     ...surfaceForms !== void 0 ? { surfaceForms } : {},
-    ...partitions !== void 0 ? { partitions } : {},
     ...defaultScoring !== void 0 ? { defaultScoring } : {},
     ...trackPositions !== void 0 ? { trackPositions } : {},
     ...strict !== void 0 ? { strict } : {},
     ...required !== void 0 ? { required } : {},
-    ...vectorPromotion !== void 0 ? { vectorPromotion } : {}
+    ...vectorPromotion !== void 0 ? { vectorPromotion } : {},
+    ...typeof source.patternValueLimit === "number" ? { patternValueLimit: source.patternValueLimit } : {}
+  };
+}
+function restoredConfigFields(envelope) {
+  const shared2 = readSharedConfigFields(envelope, rejectSnapshot);
+  const partitions = restoredPartitionConfig(envelope.partitionConfig);
+  return {
+    ...shared2,
+    ...partitions !== void 0 ? { partitions } : {}
   };
 }
 
@@ -30101,6 +30492,7 @@ async function createSnapshot(manager, entry) {
     ...config.strict !== void 0 ? { strict: config.strict } : {},
     ...config.required !== void 0 ? { required: config.required } : {},
     ...config.vectorPromotion !== void 0 ? { vectorPromotion: config.vectorPromotion } : {},
+    ...config.patternValueLimit !== void 0 ? { patternValueLimit: config.patternValueLimit } : {},
     ...config.embedding !== void 0 ? {
       embedding: {
         fields: config.embedding.fields,
@@ -30153,8 +30545,8 @@ async function restoreFromSnapshot(indexName, data, deps) {
   const configFields = restoredConfigFields(envelope);
   const embedding = restoredEmbedding(envelope.embedding);
   const language = getLanguage(envelope.language);
-  const schema = envelope.schema;
-  validateSchema(schema);
+  const schema = validateSchema(envelope.schema);
+  validatePatternValueLimit(configFields.patternValueLimit);
   validateVectorPromotion(configFields.vectorPromotion);
   validateVectorStorage(configFields.vectorPromotion, deps.filesystemDurability);
   if (configFields.required !== void 0 && configFields.required.length > 0) {
@@ -30506,7 +30898,7 @@ function listBindingOf(filters) {
 function requireBoostableFields(boost, flatSchema) {
   for (const [field, weight] of Object.entries(boost)) {
     const fieldType = flatSchema[field];
-    if (fieldType === void 0 || !isTextFieldType(fieldType) && fieldType !== "string[]") {
+    if (fieldType === void 0 || !isWordIndexedFieldType(fieldType)) {
       throw new NarsilError(
         ErrorCodes.SEARCH_INVALID_FIELD,
         fieldType === void 0 ? `The engine boosts a text field that the schema declares, and the schema declares no field "${field}"` : `The engine boosts a text field alone, and the schema declares field "${field}" as "${fieldType}"`,