@delali/narsil 0.3.1-next.74 → 0.3.1-next.75

package/dist/{chunk-T5ZWEWVD.mjs → chunk-AIUOAN7Y.mjs}

@@ -1,5 +1,5 @@
-import { tokenizeOptions, getNestedValue, tokenizeIterator, fuzzyTermMatches } from './chunk-I23G3IZC.mjs';
-import { compareCodePoints } from './chunk-LJYTW7GT.mjs';
+import { tokenizeOptions, getNestedValue, tokenizeIterator, fuzzyTermMatches } from './chunk-UOJIV2NG.mjs';
+import { compareCodePoints } from './chunk-IBR6634K.mjs';
 
 // src/core/partition/suggestions.ts
 function verbatimOccurrences(state, token) {

package/dist/{chunk-PO53FV4Q.mjs → chunk-AMGDG4ZJ.mjs}

@@ -1,7 +1,7 @@
 import { OSQ_STEPS, OSQ_GRID, OSQ_REFINE_ROUNDS, OSQ_REFINE_TOLERANCE, OSQ4_NARROW_DIMENSION_LIMIT, DEFAULT_OSQ4_NARROW_OVERSAMPLE, DEFAULT_OVERSAMPLE_BY_BITS, OSQ_QUERY_BITS, OSQ_LAMBDA } from './chunk-ZJ4CSAB2.mjs';
 import { nativeSearchCoreIsEnabled, loadNativeCore, stopUsingTheNativeSearchCore, noteFallback } from './chunk-2E35UM3G.mjs';
 import { COMPACTION_TOMBSTONE_RATIO, COMPACTION_ABSOLUTE_THRESHOLD, MAX_LAYER_CAP, PLACEMENT_ROOM_REQUESTS, INITIAL_LIST_CAPACITY, DEFAULT_EF_SEARCH, INITIAL_UPPER_SLOTS } from './chunk-KEHK2O6P.mjs';
-import { compareCodePoints } from './chunk-LJYTW7GT.mjs';
+import { compareCodePoints } from './chunk-IBR6634K.mjs';
 import { NarsilError, ErrorCodes } from './chunk-I64LXKO3.mjs';
 import { readSync, closeSync, openSync } from 'fs';
 

package/dist/{chunk-LX4JERXI.mjs → chunk-BT62MUM5.mjs}

@@ -1,4 +1,4 @@
-import { requireDocumentWithinSizeLimit, validateRequiredFields } from './chunk-Z3DQPOJQ.mjs';
+import { requireDocumentWithinSizeLimit, validateRequiredFields } from './chunk-356UOTK2.mjs';
 import { ErrorCodes, NarsilError } from './chunk-I64LXKO3.mjs';
 
 // src/narsil/internals.ts

package/dist/{chunk-NOZUQM7J.mjs → chunk-BT6BI4GV.mjs}

@@ -1,4 +1,4 @@
-import { MAX_SORT_FIELDS } from './chunk-3DDUNI6N.mjs';
+import { MAX_FILTER_DEPTH, MAX_SORT_FIELDS } from './chunk-V3B24XR7.mjs';
 import { nativeSearchCoreIsEnabled, loadNativeCore } from './chunk-2E35UM3G.mjs';
 import { NarsilError, ServerErrorCodes, ErrorCodes } from './chunk-I64LXKO3.mjs';
 import { mkdtemp, open, rm } from 'fs/promises';
@@ -423,6 +423,43 @@ function serializeBatchResult(result) {
   };
 }
 
+// src/server/regex-policy.ts
+function isRecord(value) {
+  return value !== null && typeof value === "object" && !Array.isArray(value);
+}
+function regexFieldIn(expression, depth) {
+  if (!isRecord(expression) || depth > MAX_FILTER_DEPTH) return null;
+  if (isRecord(expression.fields)) {
+    for (const [fieldPath, filter] of Object.entries(expression.fields)) {
+      if (isRecord(filter) && Object.hasOwn(filter, "regex")) return fieldPath;
+    }
+  }
+  for (const nested of [expression.and, expression.or]) {
+    if (!Array.isArray(nested)) continue;
+    for (const child of nested) {
+      const found = regexFieldIn(child, depth + 1);
+      if (found !== null) return found;
+    }
+  }
+  return regexFieldIn(expression.not, depth + 1);
+}
+function refusedRegex(filters, allowRegex) {
+  if (allowRegex) return null;
+  const fieldPath = regexFieldIn(filters, 0);
+  if (fieldPath === null) return null;
+  return new NarsilError(
+    ErrorCodes.SEARCH_INVALID_FILTER,
+    `This server refuses every "regex" test, because its operator set the allowRegex server option to false. Test field "${fieldPath}" with "contains" or "wildcard"`,
+    { fieldPath, operator: "regex", option: "allowRegex" }
+  );
+}
+function refuseRegex(ctx, filters, allowRegex) {
+  const refused = refusedRegex(filters, allowRegex);
+  if (refused === null) return false;
+  respondError(ctx, refused);
+  return true;
+}
+
 // src/server/validation.ts
 function isPlainObject(value) {
   return typeof value === "object" && value !== null && !Array.isArray(value);
@@ -616,7 +653,7 @@ function parseTaskListQuery(params, maxPageSize) {
 
 // src/server/handlers/document-reads.ts
 function createDocumentReadHandlers(deps) {
-  const { engine, limits } = deps;
+  const { engine, limits, allowRegex } = deps;
   async function get(ctx) {
     try {
       const document = await engine.get(ctx.params[0], ctx.params[1]);
@@ -668,6 +705,7 @@ function createDocumentReadHandlers(deps) {
       rejectInvalid(ctx, failure);
       return;
     }
+    if (refuseRegex(ctx, params.filters, allowRegex)) return;
     try {
       respondJson(ctx, await engine.listDocuments(ctx.params[0], params));
     } catch (err) {
@@ -1155,4 +1193,4 @@ function childAppOf(app) {
   return candidate;
 }
 
-export { ASYNC_IMPORT_CAPABILITY, DEFAULT_IMPORT_BATCH_SIZE, DEFAULT_MAX_BODY_BYTES, DEFAULT_MAX_CONCURRENT_REQUESTS, DEFAULT_MAX_CONCURRENT_TASKS, DEFAULT_MAX_FETCH_DOCUMENTS, DEFAULT_MAX_IMPORT_BYTES, DEFAULT_MAX_IMPORT_ERRORS, DEFAULT_MAX_LINE_BYTES, DEFAULT_MAX_RESTORE_BYTES, DEFAULT_MAX_RESULT_WINDOW, DEFAULT_MAX_RETAINED_TASKS, DEFAULT_MAX_TASK_PAGE_SIZE, DEFAULT_TASK_PAGE_SIZE, DOCUMENT_EXPORT_CAPABILITY, DOCUMENT_EXPORT_CONTENT_TYPE, IMPORT_FLUSHES_IN_FLIGHT, INDEX_LIFECYCLE_CAPABILITY, INDEX_TOUCH_INTERVALS_PER_IDLE_TIMEOUT, MAIN_THREAD_GATE_SLOT, MAX_INDEX_TOUCH_INTERVAL_MS, MIN_COUNTED_SERVER_LIMIT, MIN_UNBOUNDED_SERVER_LIMIT, REBUILD_ANALYSIS_CAPABILITY, RELAY_STREAM_CHUNKS_IN_FLIGHT, RUNNING_TASK_TTL_MS, SERVER_CAPABILITIES, SNAPSHOT_ARCHIVE_CONTENT_TYPE, TASK_CANCEL_CAPABILITY, TASK_FILTER_CAPABILITY, TASK_PROGRESS_WRITE_INTERVAL_MS, TERMINAL_TASK_TTL_MS, acceptorOf, authorizerFor, badRequest, childAppOf, clearGateSlot, corsHeaderReader, createDocumentReadHandlers, createRouteRunner, createSharedGateBuffer, createSharedRequestGate, createVersionHandler, gateSlotOfWorker, httpStatusForNarsilError, loadUWebSockets, parseJson, parseJsonOptional, parseTaskListQuery, registerServerRoutes, rejectInvalid, resolveCors, respondError, respondJson, sendError, sendRelayed, sendStream, serializeBatchResult, serializeNarsilError, serverRoutes, statusLine, toHttpError, validateBatch, validateCreateIndex, validateDocumentBody, validateQuery, validateRebalance, validateSuggest, versionReport };
+export { ASYNC_IMPORT_CAPABILITY, DEFAULT_IMPORT_BATCH_SIZE, DEFAULT_MAX_BODY_BYTES, DEFAULT_MAX_CONCURRENT_REQUESTS, DEFAULT_MAX_CONCURRENT_TASKS, DEFAULT_MAX_FETCH_DOCUMENTS, DEFAULT_MAX_IMPORT_BYTES, DEFAULT_MAX_IMPORT_ERRORS, DEFAULT_MAX_LINE_BYTES, DEFAULT_MAX_RESTORE_BYTES, DEFAULT_MAX_RESULT_WINDOW, DEFAULT_MAX_RETAINED_TASKS, DEFAULT_MAX_TASK_PAGE_SIZE, DEFAULT_TASK_PAGE_SIZE, DOCUMENT_EXPORT_CAPABILITY, DOCUMENT_EXPORT_CONTENT_TYPE, IMPORT_FLUSHES_IN_FLIGHT, INDEX_LIFECYCLE_CAPABILITY, INDEX_TOUCH_INTERVALS_PER_IDLE_TIMEOUT, MAIN_THREAD_GATE_SLOT, MAX_INDEX_TOUCH_INTERVAL_MS, MIN_COUNTED_SERVER_LIMIT, MIN_UNBOUNDED_SERVER_LIMIT, REBUILD_ANALYSIS_CAPABILITY, RELAY_STREAM_CHUNKS_IN_FLIGHT, RUNNING_TASK_TTL_MS, SERVER_CAPABILITIES, SNAPSHOT_ARCHIVE_CONTENT_TYPE, TASK_CANCEL_CAPABILITY, TASK_FILTER_CAPABILITY, TASK_PROGRESS_WRITE_INTERVAL_MS, TERMINAL_TASK_TTL_MS, acceptorOf, authorizerFor, badRequest, childAppOf, clearGateSlot, corsHeaderReader, createDocumentReadHandlers, createRouteRunner, createSharedGateBuffer, createSharedRequestGate, createVersionHandler, gateSlotOfWorker, httpStatusForNarsilError, loadUWebSockets, parseJson, parseJsonOptional, parseTaskListQuery, refuseRegex, registerServerRoutes, rejectInvalid, resolveCors, respondError, respondJson, sendError, sendRelayed, sendStream, serializeBatchResult, serializeNarsilError, serverRoutes, statusLine, toHttpError, validateBatch, validateCreateIndex, validateDocumentBody, validateQuery, validateRebalance, validateSuggest, versionReport };

package/dist/{chunk-LJYTW7GT.mjs → chunk-IBR6634K.mjs}

@@ -290,4 +290,4 @@ function compareSortValues(aValues, bValues, directions) {
   return 0;
 }
 
-export { SORT_MODES, SORT_VALUE_MAX_CODE_POINTS, appendFoldedCodePoint, codePointCount, compareCodePoints, compareComparableKeys, compareComparableValues, compareSortValues, defaultSortMode, exceedsCodePoints, foldedCodePointCount, holdsUnpairedSurrogate, isSortMode, readSortField, sortModeOf, toComparableSortValue, toReducedSortValue, truncateSortString };
+export { SORT_MODES, SORT_VALUE_MAX_CODE_POINTS, appendFoldedCodePoint, codePointCount, compareCodePoints, compareComparableKeys, compareComparableValues, compareSortValues, defaultSortMode, exceedsCodePoints, foldedCodePointCount, holdsUnpairedSurrogate, isSortMode, readSortField, singleFoldTable, sortModeOf, toComparableSortValue, toReducedSortValue, truncateSortString };

package/dist/{chunk-CUZOX6CN.mjs → chunk-LOARLB4F.mjs}

@@ -1,17 +1,17 @@
-import { validateWorkerConfig, validateIndexName, validatePatternValueLimit, requireSurfaceFormsForPartialFields, validateBM25Params, validatePartitionConfig, executeQuery, queryBindingOf, executePreflight, executeSuggest, executeListDocuments, validateDocId } from './chunk-CTOSGRM7.mjs';
+import { validateWorkerConfig, validateIndexName, validatePatternValueLimit, requireSurfaceFormsForPartialFields, validateBM25Params, validatePartitionConfig, executeQuery, queryBindingOf, executePreflight, executeSuggest, executeListDocuments, validateDocId } from './chunk-WL2GDSCV.mjs';
 import { ReplicationMessageTypes } from './chunk-GCM3JFVA.mjs';
-import { WIRE_BATCH_BUDGET, resolveVectorText, bindEngineCore, restoredConfigFields, restoredEmbedding, collectRequiredFieldFailures, embedChunkDocuments, providedDocId, requireDocumentObject, admitInsert, asBatchInsertError, requireMatchingDocId, embedDocumentFields, assertDocumentCarriesMappedVectors, admitBufferedUpsert, createPendingAdmissions, recordPendingAdmission } from './chunk-LX4JERXI.mjs';
-import { NATIVE_CRC32_MIN_BYTES, generateId, decodeEveryStoredDocument, createSharedFrozenSegment, shareSegmentArrays, splitWorkerBudget, resolveWorkerCount, createDirectExecutor, createPartitionRouter, requirePatternSearchFor, requirePartialWordSearchFor, MAX_FILE_BYTES, resolveWorkerEntry, missingWorkerEntry, readHeapStatistics, nodeThreadId, CHECKSUM_YIELD_CHUNK_BYTES, createPartitionManager, buildSegmentPayload, resolvePartitionInsertOptions, createFrozenSegment, resolveRequestThreadCount, vectorsPerFile, vectorsToBytes, VECTOR_FILE_PAYLOAD_VERSION, deadBitsWhere, NO_LIVE_VECTOR_AT_POSITION, createPartitionIndex, selectionKeeps, decodeMapWithoutFields, isCompositePartition, segmentArraysOf, mergeSegmentContents, survivorDocumentsOf, partitionStateOf, VECTOR_FILE_MAX_VECTORS, deadBytesFor, resolveIndexAnalysis, mergeFrozenSegments, workerResourceLimits, decodeVectorIndexParts, bytesToVectors, freezeSegmentShared, DEFAULT_BACKPRESSURE_LIMIT, DEFAULT_REQUEST_TIMEOUT_MS, SHUTDOWN_TIMEOUT_MS, decodeVectorFilePayload, decodeVectorGraphPartPayload, assembleGraphParts, countDead, encodeSegmentState, CONSECUTIVE_TIMEOUTS_BEFORE_DEATH, VALID_HNSW_METRICS, spawnNodeWorker, decodeVectorGraphPayload, graphPartsOfEarlierPayload, decodeStoredDocument, isDead } from './chunk-BVBIQQKG.mjs';
-import { collectQueryTermStats, mergePartitionStats, oversampledFacetConfig, fnv1a, kWayMerge, mergeFacets, setNestedValue } from './chunk-VKX2BPXD.mjs';
+import { WIRE_BATCH_BUDGET, resolveVectorText, bindEngineCore, restoredConfigFields, restoredEmbedding, collectRequiredFieldFailures, embedChunkDocuments, providedDocId, requireDocumentObject, admitInsert, asBatchInsertError, requireMatchingDocId, embedDocumentFields, assertDocumentCarriesMappedVectors, admitBufferedUpsert, createPendingAdmissions, recordPendingAdmission } from './chunk-BT62MUM5.mjs';
+import { NATIVE_CRC32_MIN_BYTES, generateId, decodeEveryStoredDocument, createSharedFrozenSegment, shareSegmentArrays, splitWorkerBudget, resolveWorkerCount, createDirectExecutor, createPartitionRouter, requirePartialWordSearchFor, MAX_FILE_BYTES, resolveWorkerEntry, missingWorkerEntry, readHeapStatistics, nodeThreadId, CHECKSUM_YIELD_CHUNK_BYTES, createPartitionManager, buildSegmentPayload, resolvePartitionInsertOptions, createFrozenSegment, resolveRequestThreadCount, vectorsPerFile, vectorsToBytes, VECTOR_FILE_PAYLOAD_VERSION, deadBitsWhere, NO_LIVE_VECTOR_AT_POSITION, createPartitionIndex, selectionKeeps, decodeMapWithoutFields, isCompositePartition, segmentArraysOf, mergeSegmentContents, survivorDocumentsOf, partitionStateOf, VECTOR_FILE_MAX_VECTORS, deadBytesFor, resolveIndexAnalysis, mergeFrozenSegments, workerResourceLimits, decodeVectorIndexParts, bytesToVectors, freezeSegmentShared, DEFAULT_BACKPRESSURE_LIMIT, DEFAULT_REQUEST_TIMEOUT_MS, SHUTDOWN_TIMEOUT_MS, decodeVectorFilePayload, decodeVectorGraphPartPayload, assembleGraphParts, countDead, encodeSegmentState, CONSECUTIVE_TIMEOUTS_BEFORE_DEATH, VALID_HNSW_METRICS, spawnNodeWorker, decodeVectorGraphPayload, graphPartsOfEarlierPayload, decodeStoredDocument, isDead } from './chunk-YOVS4EV5.mjs';
+import { requirePatternSearchFor, collectQueryTermStats, mergePartitionStats, oversampledFacetConfig, fnv1a, availablePatternSearch, kWayMerge, mergeFacets, PATTERN_HOST, setNestedValue } from './chunk-NAMGDK45.mjs';
 import { detectRuntime, getLanguage } from './chunk-4MQ5F4IL.mjs';
-import { MAX_INDEX_CONFIGURATION_BYTES, STATISTICS_BROADCAST_INTERVAL_MS, MAX_SWEEP_INTERVAL_MS, MIN_SWEEP_INTERVAL_MS, HEAP_PRESSURE_WARNING_FRACTION, MAX_EXPORT_LINE_BYTES, IMPORT_BATCH_DOCUMENTS, FOREIGN_STATS_TTL_MS, MAX_IMPORT_REFUSALS_REPORTED, BATCH_CHUNK_SIZE, MIN_DOCUMENTS_FOR_SEGMENTS, HEAP_PRESSURE_REARM_FRACTION, EXPORT_CHUNK_BYTES } from './chunk-3DDUNI6N.mjs';
-import { releaseLocksHeldBy } from './chunk-PO53FV4Q.mjs';
+import { MAX_INDEX_CONFIGURATION_BYTES, STATISTICS_BROADCAST_INTERVAL_MS, MAX_SWEEP_INTERVAL_MS, MIN_SWEEP_INTERVAL_MS, HEAP_PRESSURE_WARNING_FRACTION, MAX_EXPORT_LINE_BYTES, IMPORT_BATCH_DOCUMENTS, FOREIGN_STATS_TTL_MS, MAX_IMPORT_REFUSALS_REPORTED, BATCH_CHUNK_SIZE, MIN_DOCUMENTS_FOR_SEGMENTS, HEAP_PRESSURE_REARM_FRACTION, EXPORT_CHUNK_BYTES } from './chunk-V3B24XR7.mjs';
+import { releaseLocksHeldBy } from './chunk-AMGDG4ZJ.mjs';
 import { requireSearchBackendSettings } from './chunk-2E35UM3G.mjs';
-import { DEFAULT_PATTERN_WORK_CAP, MATCHER_STATE_BUDGET_BYTES, MATCHER_SIMULATION_BYTES_PER_POSITION, MATCHER_STATE_OVERHEAD_BYTES, PATTERN_TEST_OPERATORS, MATCHER_CODE_POINT_BYTES } from './chunk-ZKRV5M2P.mjs';
-import { getNestedValue, parseVectorDimension } from './chunk-I23G3IZC.mjs';
-import { extractVectorFieldsFromSchema, messagePackUpperBoundBytes, validateSchema, requireCreatableSchema, validateVectorPromotion, validateVectorStorage, raiseEarlierReleaseGraphSetting, flattenSchema, isSingleStringFieldType, requireValidFieldTypes, encodeDocument, VECTOR_PATTERN, schemaFieldsOf, MAX_DOCUMENT_NESTING_DEPTH, isQuantizationMode, isStorageMode, validateRequiredFields, requireDocumentWithinSizeLimit, validateDocument, validateDocumentStrict } from './chunk-Z3DQPOJQ.mjs';
+import { DEFAULT_PATTERN_WORK_CAP, MATCHER_STATE_BUDGET_BYTES, PATTERN_TEST_OPERATORS, PATTERN_EXPRESSION_OPERATORS, MATCHER_CODE_POINT_BYTES, MATCHER_SIMULATION_BYTES_PER_POSITION, MATCHER_STATE_OVERHEAD_BYTES } from './chunk-7ODNRREZ.mjs';
+import { getNestedValue, parseVectorDimension } from './chunk-UOJIV2NG.mjs';
+import { extractVectorFieldsFromSchema, messagePackUpperBoundBytes, validateSchema, requireCreatableSchema, validateVectorPromotion, validateVectorStorage, raiseEarlierReleaseGraphSetting, flattenSchema, isSingleStringFieldType, requireValidFieldTypes, encodeDocument, VECTOR_PATTERN, schemaFieldsOf, MAX_DOCUMENT_NESTING_DEPTH, isQuantizationMode, isStorageMode, validateRequiredFields, requireDocumentWithinSizeLimit, validateDocument, validateDocumentStrict } from './chunk-356UOTK2.mjs';
 import { MIN_M } from './chunk-KEHK2O6P.mjs';
-import { compareCodePoints, foldedCodePointCount, codePointCount } from './chunk-LJYTW7GT.mjs';
+import { compareCodePoints, foldedCodePointCount, codePointCount } from './chunk-IBR6634K.mjs';
 import { createRequestId, isValidWorkerResponse, threadSlotOfWorker } from './chunk-POWOTUKJ.mjs';
 import { ErrorCodes, NarsilError } from './chunk-I64LXKO3.mjs';
 import { encode, decode } from '@msgpack/msgpack';
@@ -10228,57 +10228,61 @@ var POSITIONS_PER_WORD = 32;
 function positionWords(positions) {
   return Math.max(1, Math.ceil(positions / POSITIONS_PER_WORD));
 }
-function automatonBytes(positions) {
-  return 2 * positionWords(positions) * Uint32Array.BYTES_PER_ELEMENT + positions * MATCHER_CODE_POINT_BYTES;
-}
 function smallestMatcherBytes(positions) {
-  return automatonBytes(positions) + positions * MATCHER_SIMULATION_BYTES_PER_POSITION + MATCHER_STATE_OVERHEAD_BYTES + positions * Uint32Array.BYTES_PER_ELEMENT;
+  return 2 * positionWords(positions) * Uint32Array.BYTES_PER_ELEMENT + positions * MATCHER_CODE_POINT_BYTES + positions * MATCHER_SIMULATION_BYTES_PER_POSITION + MATCHER_STATE_OVERHEAD_BYTES + positions * Uint32Array.BYTES_PER_ELEMENT;
 }
-function widestPatternSplit(positions) {
-  return Math.max(1, Math.floor(MATCHER_STATE_BUDGET_BYTES / smallestMatcherBytes(positions)));
+function widestPatternSplit(footprintBytes) {
+  return Math.max(1, Math.floor(MATCHER_STATE_BUDGET_BYTES / Math.max(1, footprintBytes)));
 }
 
 // src/core/pattern-index/query-positions.ts
 function isRecord4(value) {
   return value !== null && typeof value === "object" && !Array.isArray(value);
 }
-function textPositions(text, caseFold) {
-  if (typeof text !== "string") return 0;
-  return caseFold ? foldedCodePointCount(text) : codePointCount(text);
+function literalFootprint(text, caseFold) {
+  if (typeof text !== "string") return null;
+  return smallestMatcherBytes(caseFold ? foldedCodePointCount(text) : codePointCount(text));
 }
 function larger(left, right) {
   if (left === null) return right;
   if (right === null) return left;
   return Math.max(left, right);
 }
-function largestTestOf(filter) {
+function largestTestOf(filter, module) {
   const caseFold = filter.caseFold === true;
   let largest = null;
   for (const operator of PATTERN_TEST_OPERATORS) {
     const operand = filter[operator];
     if (operand === void 0) continue;
     const texts = Array.isArray(operand) ? operand : [operand];
-    for (const text of texts) largest = larger(largest, textPositions(text, caseFold));
+    for (const text of texts) largest = larger(largest, literalFootprint(text, caseFold));
+  }
+  for (const operator of PATTERN_EXPRESSION_OPERATORS) {
+    const pattern = filter[operator];
+    if (typeof pattern !== "string") continue;
+    largest = larger(largest, module.expressionFootprint(operator, pattern, caseFold, PATTERN_HOST));
   }
   return largest;
 }
-function largestIn(expression, patternFields) {
+function largestIn(expression, patternFields, module) {
   let largest = null;
   if (isRecord4(expression.fields)) {
     for (const [fieldPath, filter] of Object.entries(expression.fields)) {
-      if (patternFields.has(fieldPath) && isRecord4(filter)) largest = larger(largest, largestTestOf(filter));
+      if (patternFields.has(fieldPath) && isRecord4(filter)) largest = larger(largest, largestTestOf(filter, module));
     }
   }
   const nested = [...expression.and ?? [], ...expression.or ?? []];
   if (expression.not !== void 0) nested.push(expression.not);
-  for (const child of nested) largest = larger(largest, largestIn(child, patternFields));
+  for (const child of nested) largest = larger(largest, largestIn(child, patternFields, module));
   return largest;
 }
-function largestPatternTest(expression, schema) {
+function largestPatternFootprint(expression, schema) {
   if (expression === void 0) return null;
+  const module = availablePatternSearch();
+  if (module === null) return null;
   const patternFields = /* @__PURE__ */ new Set();
   for (const { path, pattern } of schemaFieldsOf(schema)) if (pattern) patternFields.add(path);
-  return patternFields.size === 0 ? null : largestIn(expression, patternFields);
+  return patternFields.size === 0 ? null : largestIn(expression, patternFields, module);
 }
 
 // src/engine/orchestration/search.ts
@@ -10345,7 +10349,7 @@ function isWorkCapError(err) {
 function patternSplitWidth(state, indexName, params) {
   const schema = state.indexRegistry.get(indexName)?.config.schema;
   if (schema === void 0) return Number.POSITIVE_INFINITY;
-  const largest = largestPatternTest(params.filters, schema);
+  const largest = largestPatternFootprint(params.filters, schema);
   if (largest === null) return Number.POSITIVE_INFINITY;
   if (typeof SharedArrayBuffer !== "function") return 1;
   return widestPatternSplit(largest);

package/dist/{chunk-VKX2BPXD.mjs → chunk-NAMGDK45.mjs}

@@ -1,10 +1,47 @@
-import { MAX_SORT_FIELDS, MAX_SORT_FIELD_NAME_LENGTH, MAX_CURSOR_LENGTH, MAX_CURSOR_ANCHOR_CODE_POINTS, RESULT_WINDOW, DEFAULT_FUZZY_PREFIX_LENGTH, DEFAULT_PAGE_SIZE, PREFIX_MAX_EXPANSIONS, PARTIAL_WORD_MIN_CODE_POINTS, PARTIAL_WORDS_ADDED_PER_QUERY } from './chunk-3DDUNI6N.mjs';
-import { DEFAULT_PATTERN_WORK_CAP, PATTERN_TEST_OPERATORS, bitsetIsEmpty, MAX_PATTERN_TEXT_CODE_POINTS, MAX_PATTERN_POSITIONS, createBitSet, bitsetSet, PATTERN_WORK_FLUSH_UNITS, MATCHER_STATE_BUDGET_BYTES } from './chunk-ZKRV5M2P.mjs';
-import { getFlatSchema, tokenize, getFieldValueByInternalId, getFieldValueForDoc, fuzzyTermMatches } from './chunk-I23G3IZC.mjs';
-import { MAX_LATITUDE_DEGREES, MAX_LONGITUDE_DEGREES, MIN_POLYGON_POINTS, flattenSchema, parsedTypeOf, isStringValuedBase, withSortableOption, isGeopointOrVectorType, EARTH_RADIUS_METERS, RADIUS_BOX_MARGIN_RATIO, RADIUS_BOX_MARGIN_RADIANS, FULL_TURN_DEGREES, ANTIMERIDIAN_SPAN_DEGREES, PROTOTYPE_POLLUTION_KEYS, DEG_TO_RAD, LOCATION_CODE_BITS_PER_AXIS, isLatitude, isLongitude, isWordIndexedFieldType, partialFieldsOf, MAX_COVERING_CELLS } from './chunk-Z3DQPOJQ.mjs';
-import { compareCodePoints, sortModeOf, defaultSortMode, SORT_VALUE_MAX_CODE_POINTS, toReducedSortValue, readSortField, compareComparableKeys, toComparableSortValue, exceedsCodePoints, holdsUnpairedSurrogate, isSortMode, SORT_MODES, compareSortValues, foldedCodePointCount, codePointCount } from './chunk-LJYTW7GT.mjs';
+import { MAX_SORT_FIELDS, MAX_SORT_FIELD_NAME_LENGTH, MAX_CURSOR_LENGTH, MAX_CURSOR_ANCHOR_CODE_POINTS, RESULT_WINDOW, DEFAULT_FUZZY_PREFIX_LENGTH, MAX_FILTER_DEPTH, DEFAULT_PAGE_SIZE, MAX_FILTER_ARRAY_SIZE, GEO_UNIT_METRES, MAX_FILTER_TESTS, PREFIX_MAX_EXPANSIONS, PARTIAL_WORD_MIN_CODE_POINTS, PARTIAL_WORDS_ADDED_PER_QUERY } from './chunk-V3B24XR7.mjs';
+import { DEFAULT_PATTERN_WORK_CAP, patternSearchModule, PATTERN_TEST_OPERATORS, bitsetIsEmpty, PATTERN_EXPRESSION_OPERATORS, MAX_PATTERN_POSITIONS, createBitSet, bitsetSet, MAX_PATTERN_TEXT_CODE_POINTS, PATTERN_WORK_FLUSH_UNITS, MATCHER_STATE_BUDGET_BYTES } from './chunk-7ODNRREZ.mjs';
+import { getFlatSchema, tokenize, getFieldValueByInternalId, getFieldValueForDoc, fuzzyTermMatches } from './chunk-UOJIV2NG.mjs';
+import { schemaFieldsOf, MAX_LATITUDE_DEGREES, MAX_LONGITUDE_DEGREES, MIN_POLYGON_POINTS, flattenSchema, parsedTypeOf, isStringValuedBase, withSortableOption, isGeopointOrVectorType, EARTH_RADIUS_METERS, RADIUS_BOX_MARGIN_RATIO, RADIUS_BOX_MARGIN_RADIANS, FULL_TURN_DEGREES, ANTIMERIDIAN_SPAN_DEGREES, PROTOTYPE_POLLUTION_KEYS, DEG_TO_RAD, LOCATION_CODE_BITS_PER_AXIS, isLatitude, isLongitude, isWordIndexedFieldType, partialFieldsOf, MAX_COVERING_CELLS } from './chunk-356UOTK2.mjs';
+import { appendFoldedCodePoint, compareCodePoints, sortModeOf, defaultSortMode, SORT_VALUE_MAX_CODE_POINTS, toReducedSortValue, readSortField, compareComparableKeys, toComparableSortValue, exceedsCodePoints, singleFoldTable, holdsUnpairedSurrogate, isSortMode, SORT_MODES, compareSortValues, foldedCodePointCount, codePointCount } from './chunk-IBR6634K.mjs';
 import { NarsilError, ErrorCodes } from './chunk-I64LXKO3.mjs';
 
+// src/core/pattern-index/platform.ts
+function platformPatternSearch() {
+  return patternSearchModule;
+}
+
+// src/core/pattern-index/registry.ts
+var registered = null;
+function isPatternSearchModule(value) {
+  return value !== null && typeof value === "object" && value.name === "pattern" && "createIndex" in value && typeof value.createIndex === "function" && "readArrays" in value && typeof value.readArrays === "function" && "mergeArrays" in value && typeof value.mergeArrays === "function" && "matchInto" in value && typeof value.matchInto === "function" && "checkExpression" in value && typeof value.checkExpression === "function" && "expressionFootprint" in value && typeof value.expressionFootprint === "function";
+}
+function registerPatternSearch(module) {
+  if (!isPatternSearchModule(module)) {
+    throw new NarsilError(
+      ErrorCodes.CONFIG_INVALID,
+      'registerPatternSearch takes the patternSearch object that "@delali/narsil/pattern" exports'
+    );
+  }
+  registered = module;
+}
+function availablePatternSearch() {
+  return registered ?? platformPatternSearch();
+}
+function requirePatternSearchFor(schema) {
+  for (const { path, pattern } of schemaFieldsOf(schema)) {
+    if (pattern) requirePatternSearch(path);
+  }
+}
+function requirePatternSearch(fieldPath) {
+  const module = availablePatternSearch();
+  if (module !== null) return module;
+  throw new NarsilError(
+    ErrorCodes.CONFIG_INVALID,
+    `Field "${fieldPath}" is a pattern field, but this bundle contains no pattern search code. Add import { patternSearch } from '@delali/narsil/pattern', then call registerPatternSearch(patternSearch) before you create or load the index`,
+    { field: fieldPath }
+  );
+}
+
 // src/core/pattern-index/work-meter.ts
 var COUNTER_BYTES = 8;
 function capExceeded(cap) {
@@ -529,12 +566,6 @@ function decodeCursorText(encoded) {
   return new TextDecoder().decode(bytes);
 }
 
-// src/filters/constants.ts
-var MAX_FILTER_DEPTH = 30;
-var MAX_FILTER_TESTS = 1024;
-var MAX_FILTER_ARRAY_SIZE = 65536;
-var GEO_UNIT_METRES = { km: 1e3, mi: 1609.344, m: 1 };
-
 // src/filters/operands/checks.ts
 function operandKindOf(type) {
   if (type === void 0) return void 0;
@@ -955,6 +986,34 @@ function requireSizeOperands(fieldPath, filter, type) {
   }
 }
 
+// src/core/pattern-index/host.ts
+var NO_SOURCES = [];
+var foldSources = null;
+function sourcesByFold() {
+  if (foldSources !== null) return foldSources;
+  const sources = /* @__PURE__ */ new Map();
+  const add = (source, folded) => {
+    const known = sources.get(folded);
+    if (known === void 0) sources.set(folded, [source]);
+    else known.push(source);
+  };
+  for (const [source, folded] of singleFoldTable()) add(source, folded);
+  foldSources = sources;
+  return sources;
+}
+function singleFoldSources(folded) {
+  return sourcesByFold().get(folded) ?? NO_SOURCES;
+}
+var PATTERN_HOST = { appendFoldedCodePoint, singleFoldSources };
+function patternIndexOf(indexes, fieldPath) {
+  let index = indexes.get(fieldPath);
+  if (index === void 0) {
+    index = requirePatternSearch(fieldPath).createIndex(PATTERN_HOST);
+    indexes.set(fieldPath, index);
+  }
+  return index;
+}
+
 // src/filters/operands/pattern.ts
 function requirePatternField(fieldPath, operator, type) {
   if (parsedTypeOf(type)?.pattern === true) return;
@@ -965,14 +1024,16 @@ function requirePatternField(fieldPath, operator, type) {
     operator
   );
 }
+function requireTextLength(fieldPath, operator, text) {
+  if (!exceedsCodePoints(text, MAX_PATTERN_TEXT_CODE_POINTS)) return;
+  invalid(
+    `The text of "${operator}" on pattern field "${fieldPath}" holds more than ${MAX_PATTERN_TEXT_CODE_POINTS} code points`,
+    fieldPath,
+    operator
+  );
+}
 function requirePatternText(fieldPath, operator, text, caseFold) {
-  if (exceedsCodePoints(text, MAX_PATTERN_TEXT_CODE_POINTS)) {
-    invalid(
-      `The text of "${operator}" on pattern field "${fieldPath}" holds more than ${MAX_PATTERN_TEXT_CODE_POINTS} code points`,
-      fieldPath,
-      operator
-    );
-  }
+  requireTextLength(fieldPath, operator, text);
   if (caseFold && foldedCodePointCount(text) > MAX_PATTERN_POSITIONS) {
     invalid(
       `The text of "${operator}" on pattern field "${fieldPath}" has a case fold of more than ${MAX_PATTERN_POSITIONS} code points, while a pattern holds at most ${MAX_PATTERN_POSITIONS} positions`,
@@ -982,6 +1043,23 @@ function requirePatternText(fieldPath, operator, text, caseFold) {
   }
 }
 var ELEMENT_OPERATORS2 = ["matchesAny", "containsAll"];
+function requireExpressions(fieldPath, filter, type) {
+  for (const operator of PATTERN_EXPRESSION_OPERATORS) {
+    const pattern = filter[operator];
+    if (pattern === void 0) continue;
+    if (typeof pattern !== "string")
+      invalid(`"${operator}" on field "${fieldPath}" must be a string`, fieldPath, operator);
+    requirePatternField(fieldPath, operator, type);
+    requireTextLength(fieldPath, operator, pattern);
+    requirePatternSearch(fieldPath).checkExpression(
+      operator,
+      pattern,
+      filter.caseFold === true,
+      fieldPath,
+      PATTERN_HOST
+    );
+  }
+}
 function requirePatternOperands(fieldPath, filter, type) {
   if (filter.contains !== void 0) {
     if (typeof filter.contains !== "string") {
@@ -995,6 +1073,7 @@ function requirePatternOperands(fieldPath, filter, type) {
     }
     requirePatternField(fieldPath, "caseFold", type);
   }
+  requireExpressions(fieldPath, filter, type);
   if (parsedTypeOf(type)?.pattern !== true) return;
   const caseFold = filter.caseFold === true;
   for (const operator of PATTERN_TEST_OPERATORS) {
@@ -2078,4 +2157,4 @@ function setNestedValue(obj, path, value) {
   current[segments[segments.length - 1]] = value;
 }
 
-export { GEO_UNIT_METRES, MAX_FILTER_ARRAY_SIZE, MAX_FILTER_DEPTH, MAX_FILTER_TESTS, applyPagination, applyProjection, applySorting, clampRowCount, cloneProjected, collectQueryTermStats, computeFacets, countsWithoutScores, coveringRanges, createPatternWorkMeter, decodePageCursor, encodePageCursor, everyValueFacetConfig, fanOutMatchCount, fanOutQuery, filterTestCount, fnv1a, fnv1aBytes, fnv1aUint32s, fulltextMatches, joinPatternWorkMeter, kWayMerge, locationCode, mergeFacets, mergePartitionStats, normalizeSort, oversampledFacetConfig, oversampledShardSize, partitionsIn, polygonBoxes, polygonContainment, projectionKeepsField, pruneStatsToQueryTerms, radiusBoxes, readSortValues, requireMatchingCursor, requireSortableFields, requireValidFilter, requireValidOperands, requireWithinResultWindow, resolveProjection, sanitizeGlobalStats, setNestedValue, sortModesOf, sortSignatureEntry, sortSignatureOf, undeclaredFieldOnStrictIndex, undeclaredFilterField };
+export { PATTERN_HOST, applyPagination, applyProjection, applySorting, availablePatternSearch, clampRowCount, cloneProjected, collectQueryTermStats, computeFacets, countsWithoutScores, coveringRanges, createPatternWorkMeter, decodePageCursor, encodePageCursor, everyValueFacetConfig, fanOutMatchCount, fanOutQuery, filterTestCount, fnv1a, fnv1aBytes, fnv1aUint32s, fulltextMatches, joinPatternWorkMeter, kWayMerge, locationCode, mergeFacets, mergePartitionStats, normalizeSort, oversampledFacetConfig, oversampledShardSize, partitionsIn, patternIndexOf, polygonBoxes, polygonContainment, projectionKeepsField, pruneStatsToQueryTerms, radiusBoxes, readSortValues, registerPatternSearch, requireMatchingCursor, requirePatternSearch, requirePatternSearchFor, requireSortableFields, requireValidFilter, requireValidOperands, requireWithinResultWindow, resolveProjection, sanitizeGlobalStats, setNestedValue, sortModesOf, sortSignatureEntry, sortSignatureOf, undeclaredFieldOnStrictIndex, undeclaredFilterField };

package/dist/{chunk-TVFLHM7M.mjs → chunk-PXDY5VRT.mjs}

@@ -1,7 +1,7 @@
-import { readSharedConfigFields, restoredEmbedding, restoredPartitionConfig, MAX_PARTITION_COUNT, MAX_REPLICATION_FACTOR } from './chunk-LX4JERXI.mjs';
+import { readSharedConfigFields, restoredEmbedding, restoredPartitionConfig, MAX_PARTITION_COUNT, MAX_REPLICATION_FACTOR } from './chunk-BT62MUM5.mjs';
 import { getLanguage } from './chunk-4MQ5F4IL.mjs';
-import { MAX_INDEX_NAME_LENGTH, INDEX_NAME_PATTERN } from './chunk-3DDUNI6N.mjs';
-import { compareCodePoints } from './chunk-LJYTW7GT.mjs';
+import { MAX_INDEX_NAME_LENGTH, INDEX_NAME_PATTERN } from './chunk-V3B24XR7.mjs';
+import { compareCodePoints } from './chunk-IBR6634K.mjs';
 import { NarsilError, ErrorCodes } from './chunk-I64LXKO3.mjs';
 import { encode, decode } from '@msgpack/msgpack';
 

package/dist/{chunk-I23G3IZC.mjs → chunk-UOJIV2NG.mjs}

@@ -1,4 +1,4 @@
-import { isWordIndexedBase, flattenSchema } from './chunk-Z3DQPOJQ.mjs';
+import { isWordIndexedBase, flattenSchema } from './chunk-356UOTK2.mjs';
 import { createNarsilError, ErrorCodes } from './chunk-I64LXKO3.mjs';
 
 // src/core/tokenizer/constants.ts

package/dist/{chunk-3DDUNI6N.mjs → chunk-V3B24XR7.mjs}

@@ -1,3 +1,9 @@
+// src/filters/constants.ts
+var MAX_FILTER_DEPTH = 30;
+var MAX_FILTER_TESTS = 1024;
+var MAX_FILTER_ARRAY_SIZE = 65536;
+var GEO_UNIT_METRES = { km: 1e3, mi: 1609.344, m: 1 };
+
 // src/search/constants.ts
 var RESULT_WINDOW = 1e4;
 var DEFAULT_PAGE_SIZE = 10;
@@ -45,4 +51,4 @@ var SUGGEST_OVERSAMPLE_FACTOR = 1.5;
 var SUGGEST_OVERSAMPLE_PADDING = 10;
 var MAX_SUGGEST_SCATTER_LIMIT = Math.ceil(MAX_SUGGEST_LIMIT * SUGGEST_OVERSAMPLE_FACTOR) + SUGGEST_OVERSAMPLE_PADDING;
 
-export { ABSOLUTE_WINDOWS_PATH_PATTERN, BATCH_CHUNK_SIZE, DEFAULT_FUZZY_PREFIX_LENGTH, DEFAULT_LIMIT, DEFAULT_LINEAR_ALPHA, DEFAULT_MAX_PER_GROUP, DEFAULT_OFFSET, DEFAULT_PAGE_SIZE, DEFAULT_RANK_CONSTANT, DEFAULT_SUGGEST_LIMIT, EXPORT_CHUNK_BYTES, FOREIGN_STATS_TTL_MS, HEAP_PRESSURE_REARM_FRACTION, HEAP_PRESSURE_WARNING_FRACTION, IMPORT_BATCH_DOCUMENTS, INDEX_NAME_PATTERN, MAX_CURSOR_ANCHOR_CODE_POINTS, MAX_CURSOR_LENGTH, MAX_DOC_ID_LENGTH, MAX_EXPORT_LINE_BYTES, MAX_IMPORT_REFUSALS_REPORTED, MAX_INDEX_CONFIGURATION_BYTES, MAX_INDEX_NAME_LENGTH, MAX_LINEAR_ALPHA, MAX_PREFIX_LENGTH, MAX_SORT_FIELDS, MAX_SORT_FIELD_NAME_LENGTH, MAX_SUGGEST_LIMIT, MAX_SUGGEST_SCATTER_LIMIT, MAX_SWEEP_INTERVAL_MS, MAX_TOLERANCE, MIN_DOCUMENTS_FOR_SEGMENTS, MIN_LINEAR_ALPHA, MIN_RANK_CONSTANT, MIN_SWEEP_INTERVAL_MS, MODULE_URL_SCHEME_PATTERN, PARTIAL_WORDS_ADDED_PER_QUERY, PARTIAL_WORD_MIN_CODE_POINTS, PREFIX_MAX_EXPANSIONS, RESULT_WINDOW, STATISTICS_BROADCAST_INTERVAL_MS, SUGGEST_OVERSAMPLE_FACTOR, SUGGEST_OVERSAMPLE_PADDING };
+export { ABSOLUTE_WINDOWS_PATH_PATTERN, BATCH_CHUNK_SIZE, DEFAULT_FUZZY_PREFIX_LENGTH, DEFAULT_LIMIT, DEFAULT_LINEAR_ALPHA, DEFAULT_MAX_PER_GROUP, DEFAULT_OFFSET, DEFAULT_PAGE_SIZE, DEFAULT_RANK_CONSTANT, DEFAULT_SUGGEST_LIMIT, EXPORT_CHUNK_BYTES, FOREIGN_STATS_TTL_MS, GEO_UNIT_METRES, HEAP_PRESSURE_REARM_FRACTION, HEAP_PRESSURE_WARNING_FRACTION, IMPORT_BATCH_DOCUMENTS, INDEX_NAME_PATTERN, MAX_CURSOR_ANCHOR_CODE_POINTS, MAX_CURSOR_LENGTH, MAX_DOC_ID_LENGTH, MAX_EXPORT_LINE_BYTES, MAX_FILTER_ARRAY_SIZE, MAX_FILTER_DEPTH, MAX_FILTER_TESTS, MAX_IMPORT_REFUSALS_REPORTED, MAX_INDEX_CONFIGURATION_BYTES, MAX_INDEX_NAME_LENGTH, MAX_LINEAR_ALPHA, MAX_PREFIX_LENGTH, MAX_SORT_FIELDS, MAX_SORT_FIELD_NAME_LENGTH, MAX_SUGGEST_LIMIT, MAX_SUGGEST_SCATTER_LIMIT, MAX_SWEEP_INTERVAL_MS, MAX_TOLERANCE, MIN_DOCUMENTS_FOR_SEGMENTS, MIN_LINEAR_ALPHA, MIN_RANK_CONSTANT, MIN_SWEEP_INTERVAL_MS, MODULE_URL_SCHEME_PATTERN, PARTIAL_WORDS_ADDED_PER_QUERY, PARTIAL_WORD_MIN_CODE_POINTS, PREFIX_MAX_EXPANSIONS, RESULT_WINDOW, STATISTICS_BROADCAST_INTERVAL_MS, SUGGEST_OVERSAMPLE_FACTOR, SUGGEST_OVERSAMPLE_PADDING };

package/dist/{chunk-CTOSGRM7.mjs → chunk-WL2GDSCV.mjs}

@@ -1,9 +1,9 @@
-import { clampRowCount, sortSignatureOf, requireSortableFields, decodePageCursor, requireMatchingCursor, resolveProjection, encodePageCursor, requireWithinResultWindow, normalizeSort, fanOutQuery, sortModesOf, applySorting, applyPagination, projectionKeepsField, countsWithoutScores, fanOutMatchCount, fnv1aBytes, requireValidFilter, createPatternWorkMeter, everyValueFacetConfig, fulltextMatches, mergeFacets, pruneStatsToQueryTerms, partitionsIn, undeclaredFieldOnStrictIndex } from './chunk-VKX2BPXD.mjs';
-import { MAX_INDEX_NAME_LENGTH, INDEX_NAME_PATTERN, MAX_DOC_ID_LENGTH, DEFAULT_LIMIT, DEFAULT_MAX_PER_GROUP, DEFAULT_SUGGEST_LIMIT, MAX_SUGGEST_LIMIT, MAX_SUGGEST_SCATTER_LIMIT, ABSOLUTE_WINDOWS_PATH_PATTERN, MODULE_URL_SCHEME_PATTERN, DEFAULT_LINEAR_ALPHA, MIN_LINEAR_ALPHA, MAX_LINEAR_ALPHA, DEFAULT_RANK_CONSTANT, MIN_RANK_CONSTANT, DEFAULT_OFFSET, DEFAULT_FUZZY_PREFIX_LENGTH, RESULT_WINDOW, PARTIAL_WORD_MIN_CODE_POINTS, PARTIAL_WORDS_ADDED_PER_QUERY, MAX_TOLERANCE, MAX_PREFIX_LENGTH } from './chunk-3DDUNI6N.mjs';
+import { clampRowCount, sortSignatureOf, requireSortableFields, decodePageCursor, requireMatchingCursor, resolveProjection, encodePageCursor, requireWithinResultWindow, normalizeSort, fanOutQuery, sortModesOf, applySorting, applyPagination, projectionKeepsField, countsWithoutScores, fanOutMatchCount, fnv1aBytes, requireValidFilter, createPatternWorkMeter, everyValueFacetConfig, fulltextMatches, mergeFacets, pruneStatsToQueryTerms, partitionsIn, undeclaredFieldOnStrictIndex } from './chunk-NAMGDK45.mjs';
+import { MAX_INDEX_NAME_LENGTH, INDEX_NAME_PATTERN, MAX_DOC_ID_LENGTH, DEFAULT_LIMIT, DEFAULT_MAX_PER_GROUP, DEFAULT_SUGGEST_LIMIT, MAX_SUGGEST_LIMIT, MAX_SUGGEST_SCATTER_LIMIT, ABSOLUTE_WINDOWS_PATH_PATTERN, MODULE_URL_SCHEME_PATTERN, DEFAULT_LINEAR_ALPHA, MIN_LINEAR_ALPHA, MAX_LINEAR_ALPHA, DEFAULT_RANK_CONSTANT, MIN_RANK_CONSTANT, DEFAULT_OFFSET, DEFAULT_FUZZY_PREFIX_LENGTH, RESULT_WINDOW, PARTIAL_WORD_MIN_CODE_POINTS, PARTIAL_WORDS_ADDED_PER_QUERY, MAX_TOLERANCE, MAX_PREFIX_LENGTH } from './chunk-V3B24XR7.mjs';
 import { MIN_OVERSAMPLE } from './chunk-ZJ4CSAB2.mjs';
-import { tokenize, fuzzyTermMatches } from './chunk-I23G3IZC.mjs';
-import { partialFieldsOf, MIN_PATTERN_VALUE_LIMIT, MAX_PATTERN_VALUE_LIMIT, flattenSchema, isWordIndexedFieldType, isGeopointOrVectorType } from './chunk-Z3DQPOJQ.mjs';
-import { holdsUnpairedSurrogate, compareCodePoints, compareComparableKeys, codePointCount } from './chunk-LJYTW7GT.mjs';
+import { tokenize, fuzzyTermMatches } from './chunk-UOJIV2NG.mjs';
+import { partialFieldsOf, MIN_PATTERN_VALUE_LIMIT, MAX_PATTERN_VALUE_LIMIT, flattenSchema, isWordIndexedFieldType, isGeopointOrVectorType } from './chunk-356UOTK2.mjs';
+import { holdsUnpairedSurrogate, compareCodePoints, compareComparableKeys, codePointCount } from './chunk-IBR6634K.mjs';
 import { NarsilError, ErrorCodes } from './chunk-I64LXKO3.mjs';
 
 // src/search/cursor-binding.ts

package/dist/{chunk-BVBIQQKG.mjs → chunk-YOVS4EV5.mjs}

@@ -1,15 +1,14 @@
-import { joinPatternWorkMeter, countsWithoutScores, fanOutMatchCount, fanOutQuery, sanitizeGlobalStats, computeFacets, cloneProjected, projectionKeepsField, setNestedValue, fnv1a, everyValueFacetConfig, mergeFacets, undeclaredFilterField, requireValidOperands, kWayMerge, polygonContainment, polygonBoxes, radiusBoxes, locationCode, coveringRanges, GEO_UNIT_METRES } from './chunk-VKX2BPXD.mjs';
+import { requirePatternSearchFor, patternIndexOf, joinPatternWorkMeter, countsWithoutScores, fanOutMatchCount, fanOutQuery, sanitizeGlobalStats, computeFacets, cloneProjected, projectionKeepsField, setNestedValue, fnv1a, requirePatternSearch, PATTERN_HOST, everyValueFacetConfig, mergeFacets, undeclaredFilterField, requireValidOperands, kWayMerge, polygonContainment, polygonBoxes, radiusBoxes, locationCode, coveringRanges } from './chunk-NAMGDK45.mjs';
 import { getLanguage, detectRuntime } from './chunk-4MQ5F4IL.mjs';
-import { MAX_DOC_ID_LENGTH } from './chunk-3DDUNI6N.mjs';
-import { SCRATCH_SLOTS_PER_THREAD_POOL, osqRecordBytes, openSharedVectorField, osqBitsOf, createVectorStore, DISK_STORAGE_BLOCK_BYTES, createOsqQuantizer, magnitude, createHNSWIndex, nativeStoreScratchBytes, createOrdinalFilter, isTombstoned, addToOrdinalFilter, ordinalFilterValues, nativeScoresOf, toScore, removeFromOrdinalFilter, euclideanDistance, dotProduct, cosineSimilarityWithMagnitudes, ordinalFilterHas, STORE_CALIBRATION_GENERATION, entryForOrd, toDistance, sharedMemoryAvailable, requireWholeNeighbourLists, topLayerInParts, GRAPH_BUILD_CHUNKS_IN_FLIGHT_PER_WORKER, GRAPH_BUILD_DISPATCH_CHUNK, releaseLocksHeldBy, VECTOR_SEARCH_TIMEOUT_MS, VECTOR_SEARCH_LOAD_TIMEOUT_MS, VECTOR_INSERT_TIMEOUT_MS, VECTOR_SCRATCH_SLOTS } from './chunk-PO53FV4Q.mjs';
+import { MAX_DOC_ID_LENGTH, GEO_UNIT_METRES } from './chunk-V3B24XR7.mjs';
+import { SCRATCH_SLOTS_PER_THREAD_POOL, osqRecordBytes, openSharedVectorField, osqBitsOf, createVectorStore, DISK_STORAGE_BLOCK_BYTES, createOsqQuantizer, magnitude, createHNSWIndex, nativeStoreScratchBytes, createOrdinalFilter, isTombstoned, addToOrdinalFilter, ordinalFilterValues, nativeScoresOf, toScore, removeFromOrdinalFilter, euclideanDistance, dotProduct, cosineSimilarityWithMagnitudes, ordinalFilterHas, STORE_CALIBRATION_GENERATION, entryForOrd, toDistance, sharedMemoryAvailable, requireWholeNeighbourLists, topLayerInParts, GRAPH_BUILD_CHUNKS_IN_FLIGHT_PER_WORKER, GRAPH_BUILD_DISPATCH_CHUNK, releaseLocksHeldBy, VECTOR_SEARCH_TIMEOUT_MS, VECTOR_SEARCH_LOAD_TIMEOUT_MS, VECTOR_INSERT_TIMEOUT_MS, VECTOR_SCRATCH_SLOTS } from './chunk-AMGDG4ZJ.mjs';
 import { noteFallback } from './chunk-2E35UM3G.mjs';
-import { GrowableUint32, patternSearchModule } from './chunk-XW25X6AU.mjs';
-import { bitsetHas, createBitSet, bitsetSet, bitsetOrInto, bitsetAndNotInto, bitsetAndInto, PATTERN_RUN_CODE_POINTS, bitsetFromSet, bitsetIsEmpty, bitsetAnd, bitsetNot } from './chunk-ZKRV5M2P.mjs';
-import { encodePartialSpellings, countSpelling, expandTermPrefix, suggestDisplayTerms, spellingCountsOf, codePointOrder, encodeStringBlob, partialWordSearchModule, partitionDocumentFrequency } from './chunk-T5ZWEWVD.mjs';
-import { tokenizeOptions, getNestedValue, visitTokens, textFieldsChanged, fuzzyPrefixOf, boundedLevenshtein, getFlatSchema, tokenize, getOrCreateFieldNameIndex, tokenizeIterator, getFieldValueByInternalId } from './chunk-I23G3IZC.mjs';
-import { schemaFieldsOf, partialFieldsOf, isWordIndexedBase, validateDocument, validateDocumentStrict, encodeDocument, extractVectorFieldsFromSchema, parsedTypeOf, isListBase, VINCENTY_CONVERGENCE_THRESHOLD, EARTH_RADIUS_METERS, VINCENTY_MAX_ITERATIONS, DEG_TO_RAD } from './chunk-Z3DQPOJQ.mjs';
+import { GrowableUint32, bitsetHas, createBitSet, bitsetSet, bitsetOrInto, bitsetAndNotInto, bitsetAndInto, PATTERN_RUN_CODE_POINTS, bitsetFromSet, bitsetIsEmpty, bitsetAnd, bitsetNot } from './chunk-7ODNRREZ.mjs';
+import { encodePartialSpellings, countSpelling, expandTermPrefix, suggestDisplayTerms, spellingCountsOf, codePointOrder, encodeStringBlob, partialWordSearchModule, partitionDocumentFrequency } from './chunk-AIUOAN7Y.mjs';
+import { tokenizeOptions, getNestedValue, visitTokens, textFieldsChanged, fuzzyPrefixOf, boundedLevenshtein, getFlatSchema, tokenize, getOrCreateFieldNameIndex, tokenizeIterator, getFieldValueByInternalId } from './chunk-UOJIV2NG.mjs';
+import { partialFieldsOf, schemaFieldsOf, isWordIndexedBase, validateDocument, validateDocumentStrict, encodeDocument, extractVectorFieldsFromSchema, parsedTypeOf, isListBase, VINCENTY_CONVERGENCE_THRESHOLD, EARTH_RADIUS_METERS, VINCENTY_MAX_ITERATIONS, DEG_TO_RAD } from './chunk-356UOTK2.mjs';
 import { MAX_M, EARLIER_RELEASE_MIN_M, MIN_M } from './chunk-KEHK2O6P.mjs';
-import { compareCodePoints, appendFoldedCodePoint, toReducedSortValue, readSortField, defaultSortMode, compareComparableValues, truncateSortString, SORT_VALUE_MAX_CODE_POINTS } from './chunk-LJYTW7GT.mjs';
+import { compareCodePoints, toReducedSortValue, readSortField, defaultSortMode, compareComparableValues, truncateSortString, SORT_VALUE_MAX_CODE_POINTS } from './chunk-IBR6634K.mjs';
 import { NarsilError, ErrorCodes } from './chunk-I64LXKO3.mjs';
 import { totalmem } from 'os';
 import { getHeapStatistics } from 'v8';
@@ -944,54 +943,6 @@ function createSurfaceRegistry() {
   };
 }
 
-// src/core/pattern-index/platform.ts
-function platformPatternSearch() {
-  return patternSearchModule;
-}
-
-// src/core/pattern-index/registry.ts
-var registered = null;
-function isPatternSearchModule(value) {
-  return value !== null && typeof value === "object" && value.name === "pattern" && "createIndex" in value && typeof value.createIndex === "function" && "readArrays" in value && typeof value.readArrays === "function" && "mergeArrays" in value && typeof value.mergeArrays === "function" && "matchInto" in value && typeof value.matchInto === "function";
-}
-function registerPatternSearch(module) {
-  if (!isPatternSearchModule(module)) {
-    throw new NarsilError(
-      ErrorCodes.CONFIG_INVALID,
-      'registerPatternSearch takes the patternSearch object that "@delali/narsil/pattern" exports'
-    );
-  }
-  registered = module;
-}
-function availablePatternSearch() {
-  return registered ?? platformPatternSearch();
-}
-function requirePatternSearchFor(schema) {
-  for (const { path, pattern } of schemaFieldsOf(schema)) {
-    if (pattern) requirePatternSearch(path);
-  }
-}
-function requirePatternSearch(fieldPath) {
-  const module = availablePatternSearch();
-  if (module !== null) return module;
-  throw new NarsilError(
-    ErrorCodes.CONFIG_INVALID,
-    `Field "${fieldPath}" is a pattern field, but this bundle contains no pattern search code. Add import { patternSearch } from '@delali/narsil/pattern', then call registerPatternSearch(patternSearch) before you create or load the index`,
-    { field: fieldPath }
-  );
-}
-
-// src/core/pattern-index/host.ts
-var PATTERN_HOST = { appendFoldedCodePoint };
-function patternIndexOf(indexes, fieldPath) {
-  let index = indexes.get(fieldPath);
-  if (index === void 0) {
-    index = requirePatternSearch(fieldPath).createIndex(PATTERN_HOST);
-    indexes.set(fieldPath, index);
-  }
-  return index;
-}
-
 // src/geo/haversine.ts
 function haversineDistance(lat1, lon1, lat2, lon2) {
   const phi1 = lat1 * DEG_TO_RAD;
@@ -2035,6 +1986,8 @@ function evaluatePatternTests(fieldPath, filter, context, bitsets) {
   if (typeof filter.startsWith === "string") bitsets.push(matching("startsWith", [filter.startsWith]));
   if (typeof filter.endsWith === "string") bitsets.push(matching("endsWith", [filter.endsWith]));
   if (typeof filter.contains === "string") bitsets.push(matching("contains", [filter.contains]));
+  if (typeof filter.wildcard === "string") bitsets.push(matching("wildcard", [filter.wildcard]));
+  if (typeof filter.regex === "string") bitsets.push(matching("regex", [filter.regex]));
 }
 
 // src/filters/field-tests.ts
@@ -2239,6 +2192,8 @@ var FIELD_FILTER_OPERATORS = [
   "startsWith",
   "endsWith",
   "contains",
+  "wildcard",
+  "regex",
   "caseFold",
   "containsAll",
   "matchesAny",
@@ -3525,7 +3480,7 @@ function platformPartialWordSearch() {
 }
 
 // src/core/partition/partial-words/registry.ts
-var registered2 = null;
+var registered = null;
 function isPartialWordSearchModule(value) {
   return value !== null && typeof value === "object" && value.name === "partialWords" && "createGrowingSlot" in value && typeof value.createGrowingSlot === "function" && "createFrozenSlot" in value && typeof value.createFrozenSlot === "function" && "buildTable" in value && typeof value.buildTable === "function" && "expand" in value && typeof value.expand === "function";
 }
@@ -3536,10 +3491,10 @@ function registerPartialWordSearch(module) {
       'Pass registerPartialWordSearch the partialWordSearch object that "@delali/narsil/partial-words" exports'
     );
   }
-  registered2 = module;
+  registered = module;
 }
 function availablePartialWordSearch() {
-  return registered2 ?? platformPartialWordSearch();
+  return registered ?? platformPartialWordSearch();
 }
 function requirePartialWordSearchFor(schema) {
   const fields = partialFieldsOf(schema);
@@ -11179,4 +11134,4 @@ function createDirectExecutor(options) {
   };
 }
 
-export { CHECKSUM_YIELD_CHUNK_BYTES, CONSECUTIVE_TIMEOUTS_BEFORE_DEATH, DEFAULT_BACKPRESSURE_LIMIT, DEFAULT_REQUEST_TIMEOUT_MS, FIELD_FILTER_OPERATORS, FILTER_EXPRESSION_KEYS, MAX_FILE_BYTES, NATIVE_CRC32_MIN_BYTES, NO_LIVE_VECTOR_AT_POSITION, SHUTDOWN_TIMEOUT_MS, VALID_HNSW_METRICS, VECTOR_FILE_MAX_VECTORS, VECTOR_FILE_PAYLOAD_VERSION, assembleGraphParts, buildSegmentPayload, bytesToVectors, countDead, createDirectExecutor, createFrozenSegment, createPartitionIndex, createPartitionManager, createPartitionRouter, createSharedFrozenSegment, deadBitsWhere, deadBytesFor, decodeEveryStoredDocument, decodeMapWithoutFields, decodeStoredDocument, decodeVectorFilePayload, decodeVectorGraphPartPayload, decodeVectorGraphPayload, decodeVectorIndexParts, encodeSegmentState, freezeSegmentShared, generateId, getStopWords, getTokenizer, graphPartsOfEarlierPayload, hasStopWords, hasTokenizer, isCompositePartition, isDead, mergeFrozenSegments, mergeSegmentContents, missingWorkerEntry, nodeThreadId, partitionStateOf, readHeapStatistics, registerPartialWordSearch, registerPatternSearch, registerStopWords, registerTokenizer, requirePartialWordSearchFor, requirePatternSearchFor, resolveIndexAnalysis, resolvePartitionInsertOptions, resolveRequestThreadCount, resolveWorkerCount, resolveWorkerEntry, segmentArraysOf, selectionKeeps, shareSegmentArrays, spawnNodeWorker, splitWorkerBudget, survivorDocumentsOf, vectorsPerFile, vectorsToBytes, workerResourceLimits };
+export { CHECKSUM_YIELD_CHUNK_BYTES, CONSECUTIVE_TIMEOUTS_BEFORE_DEATH, DEFAULT_BACKPRESSURE_LIMIT, DEFAULT_REQUEST_TIMEOUT_MS, FIELD_FILTER_OPERATORS, FILTER_EXPRESSION_KEYS, MAX_FILE_BYTES, NATIVE_CRC32_MIN_BYTES, NO_LIVE_VECTOR_AT_POSITION, SHUTDOWN_TIMEOUT_MS, VALID_HNSW_METRICS, VECTOR_FILE_MAX_VECTORS, VECTOR_FILE_PAYLOAD_VERSION, assembleGraphParts, buildSegmentPayload, bytesToVectors, countDead, createDirectExecutor, createFrozenSegment, createPartitionIndex, createPartitionManager, createPartitionRouter, createSharedFrozenSegment, deadBitsWhere, deadBytesFor, decodeEveryStoredDocument, decodeMapWithoutFields, decodeStoredDocument, decodeVectorFilePayload, decodeVectorGraphPartPayload, decodeVectorGraphPayload, decodeVectorIndexParts, encodeSegmentState, freezeSegmentShared, generateId, getStopWords, getTokenizer, graphPartsOfEarlierPayload, hasStopWords, hasTokenizer, isCompositePartition, isDead, mergeFrozenSegments, mergeSegmentContents, missingWorkerEntry, nodeThreadId, partitionStateOf, readHeapStatistics, registerPartialWordSearch, registerStopWords, registerTokenizer, requirePartialWordSearchFor, resolveIndexAnalysis, resolvePartitionInsertOptions, resolveRequestThreadCount, resolveWorkerCount, resolveWorkerEntry, segmentArraysOf, selectionKeeps, shareSegmentArrays, spawnNodeWorker, splitWorkerBudget, survivorDocumentsOf, vectorsPerFile, vectorsToBytes, workerResourceLimits };

package/dist/{client-C-I70ArH.d.ts → client-D4sXFS9f.d.ts}

@@ -1,6 +1,6 @@
 import { P as PartitionConfig, A as AnyDocument, a as InsertOptions, W as WriteOptions } from './schema-DgyDijhf.js';
-import { V as VectorMaintenanceResult, B as BatchResult, L as ListParams, h as ListResult, I as IndexInfo, g as IndexStats, P as PartitionStatsResult, Q as QueryParams, e as QueryResult, j as PreflightResult, k as SuggestParams, l as SuggestResult } from './memory-DJEwWoCi.js';
-import { e as TaskRecord, M as MemoryStatsResponse, b as ImportResult, H as HttpIndexConfig, c as TaskListQuery, T as TaskListPage } from './tasks-BVmAKYBh.js';
+import { V as VectorMaintenanceResult, B as BatchResult, L as ListParams, h as ListResult, I as IndexInfo, g as IndexStats, P as PartitionStatsResult, Q as QueryParams, e as QueryResult, j as PreflightResult, k as SuggestParams, l as SuggestResult } from './memory-M9kPmYP8.js';
+import { e as TaskRecord, M as MemoryStatsResponse, b as ImportResult, H as HttpIndexConfig, c as TaskListQuery, T as TaskListPage } from './tasks-A8se0J98.js';
 
 /**
  * This is the `fetch` implementation the client sends every request through.

package/dist/client.d.ts

@@ -1,10 +1,10 @@
-export { C as ClientErrorCode, a as ClientErrorCodes, E as ErrorCode, b as ErrorCodes, N as NarsilError, c as NarsilErrorCode, S as ServerErrorCode, d as ServerErrorCodes, i as isNarsilError } from './memory-DJEwWoCi.js';
-export { A as ASYNC_IMPORT_CAPABILITY, D as DOCUMENT_EXPORT_CAPABILITY, E as ErrorEnvelope, I as INDEX_LIFECYCLE_CAPABILITY, R as REBUILD_ANALYSIS_CAPABILITY, S as SERVER_CAPABILITIES, T as TASK_CANCEL_CAPABILITY, a as TASK_FILTER_CAPABILITY } from './capabilities-CAMzmkAg.js';
-export { C as CapabilitiesResponse, a as CreateIndexEmbedding, H as HttpIndexConfig, I as ImportError, b as ImportResult, M as MemoryStatsResponse, T as TaskListPage, c as TaskListQuery, d as TaskProgress, e as TaskRecord, f as TaskStatus, g as TaskType } from './tasks-BVmAKYBh.js';
-export { A as AdminOperations, B as BulkOperations, D as DocumentOperations, F as FetchFunction, I as ImportSource, a as IndexOperations, N as NarsilClient, b as NarsilClientOptions, P as PutResult, R as RequestOptions, S as SearchOperations, c as ServerOperations, d as ServerVersion, T as TaskOperations, W as WaitForTaskOptions, e as createNarsilClient } from './client-C-I70ArH.js';
+export { C as ClientErrorCode, a as ClientErrorCodes, E as ErrorCode, b as ErrorCodes, N as NarsilError, c as NarsilErrorCode, S as ServerErrorCode, d as ServerErrorCodes, i as isNarsilError } from './memory-M9kPmYP8.js';
+export { A as ASYNC_IMPORT_CAPABILITY, D as DOCUMENT_EXPORT_CAPABILITY, E as ErrorEnvelope, I as INDEX_LIFECYCLE_CAPABILITY, R as REBUILD_ANALYSIS_CAPABILITY, S as SERVER_CAPABILITIES, T as TASK_CANCEL_CAPABILITY, a as TASK_FILTER_CAPABILITY } from './capabilities-BHNksicw.js';
+export { C as CapabilitiesResponse, a as CreateIndexEmbedding, H as HttpIndexConfig, I as ImportError, b as ImportResult, M as MemoryStatsResponse, T as TaskListPage, c as TaskListQuery, d as TaskProgress, e as TaskRecord, f as TaskStatus, g as TaskType } from './tasks-A8se0J98.js';
+export { A as AdminOperations, B as BulkOperations, D as DocumentOperations, F as FetchFunction, I as ImportSource, a as IndexOperations, N as NarsilClient, b as NarsilClientOptions, P as PutResult, R as RequestOptions, S as SearchOperations, c as ServerOperations, d as ServerVersion, T as TaskOperations, W as WaitForTaskOptions, e as createNarsilClient } from './client-D4sXFS9f.js';
 import './schema-DgyDijhf.js';
 import './adapters-79OIa4xm.js';
-import './config-C6Iw5rCX.js';
-import './config-DJPH3IQB.js';
+import './config-B-GRonqw.js';
+import './config-C1kbq1X2.js';
 import './types-ChrVy4lH.js';
 import './types-Cb9kiY5P.js';