@atlaskit/editor-plugin-autocomplete 6.0.0 → 7.0.0

package/dist/cjs/pm-plugins/scoring-pipeline.js

@@ -4,19 +4,27 @@ var _interopRequireDefault = require("@babel/runtime/helpers/interopRequireDefau
 Object.defineProperty(exports, "__esModule", {
   value: true
 });
-exports.STAGE2_WEIGHT = exports.STAGE1_WEIGHT = exports.MIN_STAGE1_SCORE = void 0;
+exports.loadGrammarDataAsync = exports.isGrammarDataLoaded = exports.STAGE2_WEIGHT = exports.STAGE1_WEIGHT = exports.MIN_STAGE1_SCORE = void 0;
 exports.rankCandidates = rankCandidates;
+var _regenerator = _interopRequireDefault(require("@babel/runtime/regenerator"));
+var _defineProperty2 = _interopRequireDefault(require("@babel/runtime/helpers/defineProperty"));
+var _asyncToGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/asyncToGenerator"));
 var _slicedToArray2 = _interopRequireDefault(require("@babel/runtime/helpers/slicedToArray"));
-var _combined_l2_l3_pos_tags = _interopRequireDefault(require("./data/combined_l2_l3_pos_tags.json"));
-var _ghost_pos_tags = _interopRequireDefault(require("./data/ghost_pos_tags.json"));
-var _grammar_transitions_10k = _interopRequireDefault(require("./data/grammar_transitions_10k.json"));
+var _ufo = require("../analytics/ufo");
+var _artifactsManifest = require("./artifacts-manifest");
+var _debugMode = require("./debug-mode");
 function _createForOfIteratorHelper(r, e) { var t = "undefined" != typeof Symbol && r[Symbol.iterator] || r["@@iterator"]; if (!t) { if (Array.isArray(r) || (t = _unsupportedIterableToArray(r)) || e && r && "number" == typeof r.length) { t && (r = t); var _n = 0, F = function F() {}; return { s: F, n: function n() { return _n >= r.length ? { done: !0 } : { done: !1, value: r[_n++] }; }, e: function e(r) { throw r; }, f: F }; } throw new TypeError("Invalid attempt to iterate non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method."); } var o, a = !0, u = !1; return { s: function s() { t = t.call(r); }, n: function n() { var r = t.next(); return a = r.done, r; }, e: function e(r) { u = !0, o = r; }, f: function f() { try { a || null == t.return || t.return(); } finally { if (u) throw o; } } }; }
 function _unsupportedIterableToArray(r, a) { if (r) { if ("string" == typeof r) return _arrayLikeToArray(r, a); var t = {}.toString.call(r).slice(8, -1); return "Object" === t && r.constructor && (t = r.constructor.name), "Map" === t || "Set" === t ? Array.from(r) : "Arguments" === t || /^(?:Ui|I)nt(?:8|16|32)(?:Clamped)?Array$/.test(t) ? _arrayLikeToArray(r, a) : void 0; } }
-function _arrayLikeToArray(r, a) { (null == a || a > r.length) && (a = r.length); for (var e = 0, n = Array(a); e < a; e++) n[e] = r[e]; return n; } /**
+function _arrayLikeToArray(r, a) { (null == a || a > r.length) && (a = r.length); for (var e = 0, n = Array(a); e < a; e++) n[e] = r[e]; return n; }
+function ownKeys(e, r) { var t = Object.keys(e); if (Object.getOwnPropertySymbols) { var o = Object.getOwnPropertySymbols(e); r && (o = o.filter(function (r) { return Object.getOwnPropertyDescriptor(e, r).enumerable; })), t.push.apply(t, o); } return t; }
+function _objectSpread(e) { for (var r = 1; r < arguments.length; r++) { var t = null != arguments[r] ? arguments[r] : {}; r % 2 ? ownKeys(Object(t), !0).forEach(function (r) { (0, _defineProperty2.default)(e, r, t[r]); }) : Object.getOwnPropertyDescriptors ? Object.defineProperties(e, Object.getOwnPropertyDescriptors(t)) : ownKeys(Object(t)).forEach(function (r) { Object.defineProperty(e, r, Object.getOwnPropertyDescriptor(t, r)); }); } return e; } /**
  * Scoring Pipeline: Stage 1 (Semantic + Frequency), Grammar Filter, Stage 2 (LM Re-ranking).
  *
  * Operates synchronously on pre-loaded data. Each stage gracefully degrades
  * when its required data isn't available (cold → warm → full warm).
+ *
+ * The grammar/POS payloads are fetched from the CDN by loadGrammarDataAsync, so
+ * the grammar filter is a pass-through until that resolves.
  */
 // ─── Types ──────────────────────────────────────────────────
 
@@ -37,22 +45,119 @@ var L1_SESSION_CAP = 1.2;
 // at 0.00024) from triggering re-ranking.
 var LM_GATE_THRESHOLD = 0.0005;
 
-// ─── Grammar Data (loaded once on import) ───────────────────
+// ─── Grammar Data (fetched once from the CDN) ────────────────
+
+var posTags = new Map();
+
+/**
+ * GHOST POS DICTIONARY
+ * A mapping of common structural English words that were stripped from the main
+ * domain vocabulary. This allows the grammar filter to understand context
+ * without suggesting these words to the user.
+ */
+var ghostPosTags = {};
 
-var posTags = new Map(Object.entries(_combined_l2_l3_pos_tags.default));
-var grammarTransitions = _grammar_transitions_10k.default;
+/**
+ * Precomputed map from each POS tag to the set of allowed next POS tags, so
+ * applyGrammarFilter never re-iterates the transition rules per call.
+ */
+var precomputedAllowedByPos = new Map();
+var grammarDataPromise;
+var setGrammarData = function setGrammarData(data) {
+  var _data$grammarTransiti;
+  posTags = new Map(Object.entries(data.posTags));
+  ghostPosTags = data.ghostPosTags;
+  precomputedAllowedByPos = new Map(Object.entries((_data$grammarTransiti = data.grammarTransitions.transitions) !== null && _data$grammarTransiti !== void 0 ? _data$grammarTransiti : {}).map(function (_ref) {
+    var _ref2 = (0, _slicedToArray2.default)(_ref, 2),
+      pos = _ref2[0],
+      rule = _ref2[1];
+    return [pos, new Set(rule.allowed)];
+  }));
+};
+var isGrammarDataLoaded = exports.isGrammarDataLoaded = function isGrammarDataLoaded() {
+  return precomputedAllowedByPos.size > 0;
+};
 
 /**
- * Precomputed map from each POS tag to the set of allowed next POS tags.
- * Built once at module load from grammarTransitions so applyGrammarFilter
- * never re-iterates the transition rules per call.
+ * Fetch the POS tag and grammar transition payloads from the CDN.
+ *
+ * Safe to call repeatedly: the in-flight promise is shared, and a failure is not
+ * cached so a later call can retry.
  */
-var precomputedAllowedByPos = new Map(Object.entries(grammarTransitions.transitions).map(function (_ref) {
-  var _ref2 = (0, _slicedToArray2.default)(_ref, 2),
-    pos = _ref2[0],
-    rule = _ref2[1];
-  return [pos, new Set(rule.allowed)];
-}));
+var loadGrammarDataAsync = exports.loadGrammarDataAsync = function loadGrammarDataAsync(options) {
+  var _options$isLocalLLM;
+  if (grammarDataPromise) {
+    return grammarDataPromise;
+  }
+  var isLocalLLM = (_options$isLocalLLM = options === null || options === void 0 ? void 0 : options.isLocalLLM) !== null && _options$isLocalLLM !== void 0 ? _options$isLocalLLM : false;
+  var surface = options === null || options === void 0 ? void 0 : options.surface;
+  grammarDataPromise = (0, _asyncToGenerator2.default)( /*#__PURE__*/_regenerator.default.mark(function _callee() {
+    var _yield$Promise$all, _yield$Promise$all2, posTagsData, ghostPosTagsData, grammarTransitionsData, _t;
+    return _regenerator.default.wrap(function (_context) {
+      while (1) switch (_context.prev = _context.next) {
+        case 0:
+          (0, _ufo.startExp)(_ufo.EXPERIENCE_NAME.LOAD_GRAMMAR, 'singleton', _objectSpread({
+            isLocalLLM: isLocalLLM
+          }, surface ? {
+            surface: surface
+          } : {}));
+          _context.prev = 1;
+          _context.next = 2;
+          return Promise.all([(0, _artifactsManifest.fetchArtifactJson)(_artifactsManifest.ARTIFACT_NAME.POS_TAGS), (0, _artifactsManifest.fetchArtifactJson)(_artifactsManifest.ARTIFACT_NAME.GHOST_POS_TAGS), (0, _artifactsManifest.fetchArtifactJson)(_artifactsManifest.ARTIFACT_NAME.GRAMMAR_TRANSITIONS)]);
+        case 2:
+          _yield$Promise$all = _context.sent;
+          _yield$Promise$all2 = (0, _slicedToArray2.default)(_yield$Promise$all, 3);
+          posTagsData = _yield$Promise$all2[0];
+          ghostPosTagsData = _yield$Promise$all2[1];
+          grammarTransitionsData = _yield$Promise$all2[2];
+          if (!((grammarTransitionsData === null || grammarTransitionsData === void 0 ? void 0 : grammarTransitionsData.transitions) == null)) {
+            _context.next = 3;
+            break;
+          }
+          throw new Error("[scoring-pipeline] ".concat(_artifactsManifest.ARTIFACT_NAME.GRAMMAR_TRANSITIONS, " is missing its `transitions` wrapper"));
+        case 3:
+          setGrammarData({
+            posTags: posTagsData !== null && posTagsData !== void 0 ? posTagsData : {},
+            ghostPosTags: ghostPosTagsData !== null && ghostPosTagsData !== void 0 ? ghostPosTagsData : {},
+            grammarTransitions: grammarTransitionsData
+          });
+          (0, _ufo.succeedExp)(_ufo.EXPERIENCE_NAME.LOAD_GRAMMAR, 'singleton', _objectSpread({
+            isLocalLLM: isLocalLLM,
+            posTagCount: posTags.size,
+            transitionCount: precomputedAllowedByPos.size
+          }, surface ? {
+            surface: surface
+          } : {}));
+          if ((0, _debugMode.isAutocompleteDebugEnabled)()) {
+            // eslint-disable-next-line no-console
+            console.log('[scoring-pipeline] Grammar data loaded:', {
+              posTagCount: posTags.size,
+              ghostPosTagCount: Object.keys(ghostPosTags).length,
+              transitionCount: precomputedAllowedByPos.size
+            });
+          }
+          _context.next = 5;
+          break;
+        case 4:
+          _context.prev = 4;
+          _t = _context["catch"](1);
+          (0, _ufo.failExp)(_ufo.EXPERIENCE_NAME.LOAD_GRAMMAR, 'singleton', _objectSpread({
+            isLocalLLM: isLocalLLM,
+            errorType: 'network'
+          }, surface ? {
+            surface: surface
+          } : {}));
+          // Allow a later call to retry the load rather than caching the failure.
+          grammarDataPromise = undefined;
+          throw _t;
+        case 5:
+        case "end":
+          return _context.stop();
+      }
+    }, _callee, null, [[1, 4]]);
+  }))();
+  return grammarDataPromise;
+};
 
 // ─── Math ───────────────────────────────────────────────────
 
@@ -101,18 +206,15 @@ function scoreStage1(candidate, contextVector, getWordVector, maxTenantFreq) {
 
 // ─── Grammar Filter ─────────────────────────────────────────
 
-/**
- * GHOST POS DICTIONARY
- * A hardcoded mapping of common structural English words that were stripped
- * from the main domain vocabulary. This allows the grammar filter to understand
- * context without suggesting these words to the user.
- */
-var ghostPosTags = _ghost_pos_tags.default;
 function applyGrammarFilter(candidates, previousWord) {
-  if (!previousWord) return {
-    filtered: candidates,
-    grammarMeta: null
-  };
+  // Pass everything through until loadGrammarDataAsync resolves — filtering on a
+  // partially loaded ruleset would drop valid candidates.
+  if (!previousWord || !isGrammarDataLoaded()) {
+    return {
+      filtered: candidates,
+      grammarMeta: null
+    };
+  }
   var lowerPrev = previousWord.toLowerCase();
   var prevTags = ghostPosTags[lowerPrev] || posTags.get(lowerPrev);
   if (!prevTags || prevTags.length === 0) {

package/dist/cjs/pm-plugins/text-predictor.js

@@ -1,23 +1,22 @@
 "use strict";
 
 var _interopRequireDefault = require("@babel/runtime/helpers/interopRequireDefault");
-var _typeof3 = require("@babel/runtime/helpers/typeof");
 Object.defineProperty(exports, "__esModule", {
   value: true
 });
 exports.predict = exports.loadVectorsAsync = exports.loadDefaultVocabulary = exports.initVocabulary = exports.initVectors = exports.initL3Vocabulary = exports.ingestDocumentPage = exports.incrementSessionFreq = exports.getPredictorStatus = exports.getLastPredictionDebug = void 0;
 var _regenerator = _interopRequireDefault(require("@babel/runtime/regenerator"));
-var _asyncToGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/asyncToGenerator"));
 var _typeof2 = _interopRequireDefault(require("@babel/runtime/helpers/typeof"));
+var _asyncToGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/asyncToGenerator"));
 var _slicedToArray2 = _interopRequireDefault(require("@babel/runtime/helpers/slicedToArray"));
 var _createClass2 = _interopRequireDefault(require("@babel/runtime/helpers/createClass"));
 var _classCallCheck2 = _interopRequireDefault(require("@babel/runtime/helpers/classCallCheck"));
 var _defineProperty2 = _interopRequireDefault(require("@babel/runtime/helpers/defineProperty"));
 var _ufo = require("../analytics/ufo");
+var _artifactsManifest = require("./artifacts-manifest");
 var _debugMode = require("./debug-mode");
 var _scoringPipeline = require("./scoring-pipeline");
 var _slowLaneClient = require("./slow-lane-client");
-function _interopRequireWildcard(e, t) { if ("function" == typeof WeakMap) var r = new WeakMap(), n = new WeakMap(); return (_interopRequireWildcard = function _interopRequireWildcard(e, t) { if (!t && e && e.__esModule) return e; var o, i, f = { __proto__: null, default: e }; if (null === e || "object" != _typeof3(e) && "function" != typeof e) return f; if (o = t ? n : r) { if (o.has(e)) return o.get(e); o.set(e, f); } for (var _t4 in e) "default" !== _t4 && {}.hasOwnProperty.call(e, _t4) && ((i = (o = Object.defineProperty) && Object.getOwnPropertyDescriptor(e, _t4)) && (i.get || i.set) ? o(f, _t4, i) : f[_t4] = e[_t4]); return f; })(e, t); }
 function ownKeys(e, r) { var t = Object.keys(e); if (Object.getOwnPropertySymbols) { var o = Object.getOwnPropertySymbols(e); r && (o = o.filter(function (r) { return Object.getOwnPropertyDescriptor(e, r).enumerable; })), t.push.apply(t, o); } return t; }
 function _objectSpread(e) { for (var r = 1; r < arguments.length; r++) { var t = null != arguments[r] ? arguments[r] : {}; r % 2 ? ownKeys(Object(t), !0).forEach(function (r) { (0, _defineProperty2.default)(e, r, t[r]); }) : Object.getOwnPropertyDescriptors ? Object.defineProperties(e, Object.getOwnPropertyDescriptors(t)) : ownKeys(Object(t)).forEach(function (r) { Object.defineProperty(e, r, Object.getOwnPropertyDescriptor(t, r)); }); } return e; }
 function _createForOfIteratorHelper(r, e) { var t = "undefined" != typeof Symbol && r[Symbol.iterator] || r["@@iterator"]; if (!t) { if (Array.isArray(r) || (t = _unsupportedIterableToArray(r)) || e && r && "number" == typeof r.length) { t && (r = t); var _n = 0, F = function F() {}; return { s: F, n: function n() { return _n >= r.length ? { done: !0 } : { done: !1, value: r[_n++] }; }, e: function e(r) { throw r; }, f: F }; } throw new TypeError("Invalid attempt to iterate non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method."); } var o, a = !0, u = !1; return { s: function s() { t = t.call(r); }, n: function n() { var r = t.next(); return a = r.done, r; }, e: function e(r) { u = !0, o = r; }, f: function f() { try { a || null == t.return || t.return(); } finally { if (u) throw o; } } }; }
@@ -37,9 +36,9 @@ function _arrayLikeToArray(r, a) { (null == a || a > r.length) && (a = r.length)
  *
  * Session personalization (L1): words the user types are incrementally boosted
  * via incrementSessionFreq(), called on word boundaries from the plugin.
- */ // The vocabulary, L3 and word-index JSON payloads are dynamically imported in
-// loadDefaultVocabulary / loadVectorsAsync so their (large) contents stay out of
-// the editor's main chunk and only load when autocomplete is initialised.
+ */ // The vocabulary, L3 and word-index payloads are fetched from the CDN in
+// loadDefaultVocabulary / loadVectorsAsync so their (large) contents never enter
+// the editor's bundle and only load when autocomplete is initialised.
 // ─── Constants ───────────────────────────────────────────────────────────────
 
 // eslint-disable-next-line require-unicode-regexp
@@ -689,56 +688,10 @@ var predict = exports.predict = function predict(textBefore) {
 
 // ─── Data Loading ────────────────────────────────────────────────────────────
 
-/**
- * Unwrap a dynamically imported JSON module to its parsed value, handling both
- * interop modes AFM's bundler chain emits: a `.default`-wrapped namespace
- * (classic webpack) and a named-exports namespace (webpack 5 / atlaspack JSON
- * modules, where `default` can be a misleading scalar). Named exports are
- * preferred when present. The caller declares the JSON `shape` because a dense
- * array and a sparse numeric-keyed object are emitted identically as named
- * exports. Kept in lock-step with the matching helper in local-slow-lane-client.ts.
- */
-function unwrapJsonModule(mod, shape) {
-  if (mod == null || (0, _typeof2.default)(mod) !== 'object') {
-    return null;
-  }
-  var namespace = mod;
-  var ownKeys = Object.keys(namespace).filter(function (k) {
-    return k !== 'default' && k !== '__esModule';
-  });
-  if (ownKeys.length > 0) {
-    if (shape === 'array') {
-      var len = ownKeys.length;
-      var arr = new Array(len);
-      for (var i = 0; i < len; i++) {
-        arr[i] = namespace[String(i)];
-      }
-      return arr;
-    }
-    var obj = {};
-    var _iterator1 = _createForOfIteratorHelper(ownKeys),
-      _step1;
-    try {
-      for (_iterator1.s(); !(_step1 = _iterator1.n()).done;) {
-        var k = _step1.value;
-        obj[k] = namespace[k];
-      }
-    } catch (err) {
-      _iterator1.e(err);
-    } finally {
-      _iterator1.f();
-    }
-    return obj;
-  }
-  if ('default' in namespace && namespace.default != null) {
-    return namespace.default;
-  }
-  return null;
-}
 var loadVectorsAsync = exports.loadVectorsAsync = /*#__PURE__*/function () {
   var _ref6 = (0, _asyncToGenerator2.default)( /*#__PURE__*/_regenerator.default.mark(function _callee(options) {
     var _options$isLocalLLM;
-    var isLocalLLM, surface, url, _wordIndexOuter$index, res, buffer, float32, wordIndexModule, wordIndexOuter, wordIndex, nWords, dim, _t, _t2;
+    var isLocalLLM, surface, url, res, buffer, float32, wordIndexPayload, wrappedIndex, wordIndex, nWords, dim, _t, _t2;
     return _regenerator.default.wrap(function (_context) {
       while (1) switch (_context.prev = _context.next) {
         case 0:
@@ -748,14 +701,6 @@ var loadVectorsAsync = exports.loadVectorsAsync = /*#__PURE__*/function () {
           }
           return _context.abrupt("return");
         case 1:
-          if (options !== null && options !== void 0 && options.getBinaryUrl) {
-            _context.next = 2;
-            break;
-          }
-          // eslint-disable-next-line no-console
-          console.warn('[text-predictor] loadVectorsAsync called without a getBinaryUrl — vectors will not load. Pass getVectorsBinaryUrl via plugin options.');
-          return _context.abrupt("return");
-        case 2:
           isLocalLLM = (_options$isLocalLLM = options === null || options === void 0 ? void 0 : options.isLocalLLM) !== null && _options$isLocalLLM !== void 0 ? _options$isLocalLLM : false;
           surface = options === null || options === void 0 ? void 0 : options.surface;
           vectorsLoadStarted = true;
@@ -764,16 +709,16 @@ var loadVectorsAsync = exports.loadVectorsAsync = /*#__PURE__*/function () {
           }, surface ? {
             surface: surface
           } : {}));
-          _context.prev = 3;
-          _context.next = 4;
-          return options.getBinaryUrl();
-        case 4:
+          _context.prev = 2;
+          _context.next = 3;
+          return (0, _artifactsManifest.getArtifactUrl)(_artifactsManifest.ARTIFACT_NAME.WORD_VECTORS);
+        case 3:
           url = _context.sent;
-          _context.next = 6;
+          _context.next = 5;
           break;
-        case 5:
-          _context.prev = 5;
-          _t = _context["catch"](3);
+        case 4:
+          _context.prev = 4;
+          _t = _context["catch"](2);
           vectorsLoadStarted = false;
           (0, _ufo.failExp)(_ufo.EXPERIENCE_NAME.LOAD_VECTORS, 'singleton', _objectSpread({
             isLocalLLM: isLocalLLM,
@@ -784,14 +729,14 @@ var loadVectorsAsync = exports.loadVectorsAsync = /*#__PURE__*/function () {
           // eslint-disable-next-line no-console
           console.warn('[text-predictor] Failed to resolve vectors URL:', _t);
           return _context.abrupt("return");
-        case 6:
-          _context.prev = 6;
-          _context.next = 7;
+        case 5:
+          _context.prev = 5;
+          _context.next = 6;
           return fetch(url);
-        case 7:
+        case 6:
           res = _context.sent;
           if (res.ok) {
-            _context.next = 8;
+            _context.next = 7;
             break;
           }
           vectorsLoadStarted = false;
@@ -805,26 +750,26 @@ var loadVectorsAsync = exports.loadVectorsAsync = /*#__PURE__*/function () {
           // eslint-disable-next-line no-console
           console.warn("[text-predictor] Failed to load vectors: ".concat(res.status));
           return _context.abrupt("return");
+        case 7:
+          _context.next = 8;
+          return res.arrayBuffer();
         case 8:
+          buffer = _context.sent;
+          float32 = new Float32Array(buffer); // The word index is published either bare or wrapped as `{ "index": {…} }`
+          // (the wrapper exists so no real entry, e.g. the word "default", can shadow
+          // a bundler-synthesised ESM export). Detect the wrapper by value type, not
+          // by key presence — "index" is itself a plausible vocabulary word, and in a
+          // bare index every value is a number.
           _context.next = 9;
-          return res.arrayBuffer();
+          return (0, _artifactsManifest.fetchArtifactJson)(_artifactsManifest.ARTIFACT_NAME.WORD_INDEX);
         case 9:
-          buffer = _context.sent;
-          float32 = new Float32Array(buffer); // word_index_10k.json is wrapped as `{ "index": {…} }` so no real entry
-          // (e.g. the word "default") can shadow the synthetic ESM `default` export
-          // the bundler creates for dynamically-imported JSON.
-          _context.next = 10;
-          return Promise.resolve().then(function () {
-            return _interopRequireWildcard(require( /* webpackChunkName: "@atlaskit-internal_editor-plugin-autocomplete-word-index-10k" */'./data/word_index_10k.json'));
-          });
-        case 10:
-          wordIndexModule = _context.sent;
-          wordIndexOuter = unwrapJsonModule(wordIndexModule, 'object');
-          wordIndex = (_wordIndexOuter$index = wordIndexOuter === null || wordIndexOuter === void 0 ? void 0 : wordIndexOuter.index) !== null && _wordIndexOuter$index !== void 0 ? _wordIndexOuter$index : {};
+          wordIndexPayload = _context.sent;
+          wrappedIndex = wordIndexPayload === null || wordIndexPayload === void 0 ? void 0 : wordIndexPayload.index;
+          wordIndex = (0, _typeof2.default)(wrappedIndex) === 'object' && wrappedIndex !== null ? wrappedIndex : wordIndexPayload !== null && wordIndexPayload !== void 0 ? wordIndexPayload : {};
           nWords = Object.keys(wordIndex).length;
           if (nWords === 0) {
             // eslint-disable-next-line no-console
-            console.warn('[text-predictor] word_index_10k.json missing its `index` wrapper — wordIndex is empty, semantic scoring will be a no-op.');
+            console.warn("[text-predictor] ".concat(_artifactsManifest.ARTIFACT_NAME.WORD_INDEX, " was empty \u2014 semantic scoring will be a no-op."));
           }
           dim = float32.length / nWords;
           vectorStore = {
@@ -848,11 +793,11 @@ var loadVectorsAsync = exports.loadVectorsAsync = /*#__PURE__*/function () {
               sizeBytes: float32.byteLength
             });
           }
-          _context.next = 12;
+          _context.next = 11;
           break;
-        case 11:
-          _context.prev = 11;
-          _t2 = _context["catch"](6);
+        case 10:
+          _context.prev = 10;
+          _t2 = _context["catch"](5);
           vectorsLoadStarted = false;
           (0, _ufo.failExp)(_ufo.EXPERIENCE_NAME.LOAD_VECTORS, 'singleton', _objectSpread({
             isLocalLLM: isLocalLLM,
@@ -862,11 +807,11 @@ var loadVectorsAsync = exports.loadVectorsAsync = /*#__PURE__*/function () {
           } : {}));
           // eslint-disable-next-line no-console
           console.warn('[text-predictor] Failed to load vectors:', _t2);
-        case 12:
+        case 11:
         case "end":
           return _context.stop();
       }
-    }, _callee, null, [[3, 5], [6, 11]]);
+    }, _callee, null, [[2, 4], [5, 10]]);
   }));
   return function loadVectorsAsync(_x) {
     return _ref6.apply(this, arguments);
@@ -887,7 +832,7 @@ var loadDefaultVocabulary = exports.loadDefaultVocabulary = function loadDefault
   var isLocalLLM = (_options$isLocalLLM2 = options === null || options === void 0 ? void 0 : options.isLocalLLM) !== null && _options$isLocalLLM2 !== void 0 ? _options$isLocalLLM2 : false;
   var surface = options === null || options === void 0 ? void 0 : options.surface;
   vocabularyLoadPromise = (0, _asyncToGenerator2.default)( /*#__PURE__*/_regenerator.default.mark(function _callee2() {
-    var _yield$Promise$all, _yield$Promise$all2, vocabularyModule, l3VocabularyModule, vocabularyData, l3VocabularyData, terms, _t3;
+    var _yield$Promise$all, _yield$Promise$all2, vocabularyData, l3VocabularyData, terms, _t3;
     return _regenerator.default.wrap(function (_context2) {
       while (1) switch (_context2.prev = _context2.next) {
         case 0:
@@ -898,23 +843,17 @@ var loadDefaultVocabulary = exports.loadDefaultVocabulary = function loadDefault
           } : {}));
           _context2.prev = 1;
           _context2.next = 2;
-          return Promise.all([Promise.resolve().then(function () {
-            return _interopRequireWildcard(require( /* webpackChunkName: "@atlaskit-internal_editor-plugin-autocomplete-vocabulary-10k" */'./data/vocabulary_10k.json'));
-          }), Promise.resolve().then(function () {
-            return _interopRequireWildcard(require( /* webpackChunkName: "@atlaskit-internal_editor-plugin-autocomplete-l3-vocabulary" */'./data/l3_vocabulary.json'));
-          })]);
+          return Promise.all([(0, _artifactsManifest.fetchArtifactJson)(_artifactsManifest.ARTIFACT_NAME.VOCABULARY), (0, _artifactsManifest.fetchArtifactJson)(_artifactsManifest.ARTIFACT_NAME.L3_VOCABULARY)]);
         case 2:
           _yield$Promise$all = _context2.sent;
           _yield$Promise$all2 = (0, _slicedToArray2.default)(_yield$Promise$all, 2);
-          vocabularyModule = _yield$Promise$all2[0];
-          l3VocabularyModule = _yield$Promise$all2[1];
-          vocabularyData = unwrapJsonModule(vocabularyModule, 'object');
-          l3VocabularyData = unwrapJsonModule(l3VocabularyModule, 'array');
+          vocabularyData = _yield$Promise$all2[0];
+          l3VocabularyData = _yield$Promise$all2[1];
           if (!((vocabularyData === null || vocabularyData === void 0 ? void 0 : vocabularyData.words) == null || !Array.isArray(l3VocabularyData))) {
             _context2.next = 3;
             break;
           }
-          throw new Error('[text-predictor] vocabulary JSON modules could not be unwrapped');
+          throw new Error("[text-predictor] unexpected vocabulary payload \u2014 " + "".concat(_artifactsManifest.ARTIFACT_NAME.VOCABULARY, " ").concat((vocabularyData === null || vocabularyData === void 0 ? void 0 : vocabularyData.words) == null ? 'is missing its `words` wrapper' : 'is valid', ", ") + "".concat(_artifactsManifest.ARTIFACT_NAME.L3_VOCABULARY, " ").concat(Array.isArray(l3VocabularyData) ? 'is valid' : 'is not an array'));
         case 3:
           terms = Object.entries(vocabularyData.words).map(function (_ref8) {
             var _ref9 = (0, _slicedToArray2.default)(_ref8, 2),
@@ -947,7 +886,7 @@ var loadDefaultVocabulary = exports.loadDefaultVocabulary = function loadDefault
           _t3 = _context2["catch"](1);
           (0, _ufo.failExp)(_ufo.EXPERIENCE_NAME.LOAD_VOCABULARY, 'singleton', _objectSpread({
             isLocalLLM: isLocalLLM,
-            errorType: 'parse_error'
+            errorType: 'load_error'
           }, surface ? {
             surface: surface
           } : {}));

package/dist/es2019/analytics/ufo.js

@@ -31,7 +31,8 @@ import { ConcurrentExperience, ExperiencePerformanceTypes, ExperienceTypes } fro
 export const EXPERIENCE_NAME = {
   SLOW_LANE_FETCH: 'slow-lane-fetch',
   LOAD_VOCABULARY: 'load-vocabulary',
-  LOAD_VECTORS: 'load-vectors'
+  LOAD_VECTORS: 'load-vectors',
+  LOAD_GRAMMAR: 'load-grammar'
 };
 const isUfoEnabled = () => {
   var _process, _process$env;
@@ -54,7 +55,8 @@ const operationConfig = {
 const experiences = {
   [EXPERIENCE_NAME.SLOW_LANE_FETCH]: new ConcurrentExperience(EXPERIENCE_NAME.SLOW_LANE_FETCH, operationConfig),
   [EXPERIENCE_NAME.LOAD_VOCABULARY]: new ConcurrentExperience(EXPERIENCE_NAME.LOAD_VOCABULARY, operationConfig),
-  [EXPERIENCE_NAME.LOAD_VECTORS]: new ConcurrentExperience(EXPERIENCE_NAME.LOAD_VECTORS, operationConfig)
+  [EXPERIENCE_NAME.LOAD_VECTORS]: new ConcurrentExperience(EXPERIENCE_NAME.LOAD_VECTORS, operationConfig),
+  [EXPERIENCE_NAME.LOAD_GRAMMAR]: new ConcurrentExperience(EXPERIENCE_NAME.LOAD_GRAMMAR, operationConfig)
 };
 export const startExp = (name, id, metadata) => {
   if (!isUfoEnabled()) {

package/dist/es2019/pm-plugins/artifacts-manifest.js

@@ -0,0 +1,126 @@
+/**
+ * Artifacts Manifest Client: resolves the autocomplete model artifacts hosted on
+ * the CDN.
+ *
+ * `GET /gateway/api/v1/autocomplete/artifacts` returns one presigned CDN URL per
+ * artifact for the tenant's active model version:
+ *
+ *   { activeVersion, artifacts: [{ name, url, contentType, sizeBytes, checksum }] }
+ *
+ * Every loader in the plugin resolves its payload through here, so no model data
+ * is bundled with the package.
+ */
+
+import { isAutocompleteDebugEnabled } from './debug-mode';
+
+// ─── Types ───────────────────────────────────────────────────────────────────
+
+/**
+ * Manifest artifact names. These are the manifest's lookup keys and are
+ * hyphenated, unlike the underscored filenames the artifacts are generated from.
+ */
+export const ARTIFACT_NAME = {
+  FIRST_TOKEN_TO_WORDS: 'first-token-to-words.json',
+  GHOST_POS_TAGS: 'ghost-pos-tags.json',
+  GRAMMAR_TRANSITIONS: 'grammar-transitions-10k.json',
+  L3_VOCABULARY: 'l3-vocabulary.json',
+  POS_TAGS: 'combined-l2-l3-pos-tags.json',
+  VOCABULARY: 'vocabulary-10k.json',
+  WORD_INDEX: 'word-index-10k.json',
+  WORD_VECTORS: 'word-vectors-10k.bin'
+};
+
+/** A single entry in the artifacts manifest. */
+
+/** Response body of the artifacts endpoint. */
+
+// ─── Constants ───────────────────────────────────────────────────────────────
+
+const MANIFEST_ENDPOINT = '/gateway/api/v1/autocomplete/artifacts';
+
+/**
+ * Presigned artifact URLs expire ~10 minutes after the manifest is issued, so
+ * the manifest is re-requested well inside that window instead of being cached
+ * for the session and handing out dead URLs to a late loader.
+ */
+const MANIFEST_TTL_MS = 5 * 60 * 1000;
+
+// ─── Manifest ────────────────────────────────────────────────────────────────
+
+let manifestPromise;
+let manifestRequestedAt = 0;
+
+/** Test-only: drops the memoised manifest so the next lookup re-requests it. */
+export const resetArtifactsManifestCache = () => {
+  manifestPromise = undefined;
+  manifestRequestedAt = 0;
+};
+const requestManifest = async () => {
+  const res = await fetch(MANIFEST_ENDPOINT, {
+    method: 'GET',
+    headers: {
+      'x-experience-id': 'confluence-smart-typeahead-artifacts',
+      'x-product': 'confluence'
+    }
+  });
+  if (!res.ok) {
+    throw new Error(`[autocomplete-artifacts] Manifest request failed: ${res.status}`);
+  }
+  const body = await res.json();
+  if (!Array.isArray(body === null || body === void 0 ? void 0 : body.artifacts) || body.artifacts.length === 0) {
+    throw new Error('[autocomplete-artifacts] Manifest response contained no artifacts');
+  }
+  if (isAutocompleteDebugEnabled()) {
+    // eslint-disable-next-line no-console
+    console.log('[autocomplete-artifacts] Manifest loaded:', {
+      activeVersion: body.activeVersion,
+      names: body.artifacts.map(artifact => artifact.name)
+    });
+  }
+  return new Map(body.artifacts.map(artifact => [artifact.name, artifact]));
+};
+const getManifest = () => {
+  if (manifestPromise && Date.now() - manifestRequestedAt < MANIFEST_TTL_MS) {
+    return manifestPromise;
+  }
+  manifestRequestedAt = Date.now();
+  manifestPromise = requestManifest().catch(e => {
+    // Never cache a rejection — a transient failure would otherwise stop every
+    // later loader in the session from retrying.
+    resetArtifactsManifestCache();
+    throw e;
+  });
+  return manifestPromise;
+};
+
+// ─── Public API ──────────────────────────────────────────────────────────────
+
+/**
+ * Resolve the CDN URL for an artifact.
+ *
+ * :raises Error: when the manifest cannot be fetched or omits the artifact.
+ */
+export const getArtifactUrl = async name => {
+  var _manifest$get;
+  const manifest = await getManifest();
+  const url = (_manifest$get = manifest.get(name)) === null || _manifest$get === void 0 ? void 0 : _manifest$get.url;
+  if (!url) {
+    throw new Error(`[autocomplete-artifacts] Manifest has no URL for "${name}"`);
+  }
+  return url;
+};
+
+/**
+ * Download and parse a JSON artifact. The presigned URL carries its own auth
+ * token, so the request is sent without Atlassian headers or credentials.
+ *
+ * :raises Error: when the manifest lookup or the download fails.
+ */
+export const fetchArtifactJson = async name => {
+  const url = await getArtifactUrl(name);
+  const res = await fetch(url);
+  if (!res.ok) {
+    throw new Error(`[autocomplete-artifacts] Failed to download "${name}": ${res.status}`);
+  }
+  return await res.json();
+};

package/dist/es2019/pm-plugins/autocomplete-plugin.js

@@ -7,6 +7,7 @@ import { DecorationSet } from '@atlaskit/editor-prosemirror/view';
 import { isAutocompleteDebugEnabled } from './debug-mode';
 import { createGhostTextDecorationSet } from './ghost-text-decoration';
 import { createLocalSlowLaneClient } from './local-slow-lane-client';
+import { loadGrammarDataAsync } from './scoring-pipeline';
 import { createSlowLaneClient, setDefaultSlowLaneClient, isWordBoundary } from './slow-lane-client';
 import { predict, loadDefaultVocabulary, loadVectorsAsync, incrementSessionFreq, ingestDocumentPage } from './text-predictor';
 export const autocompletePluginKey = new PluginKey('autocomplete');
@@ -665,7 +666,7 @@ export const createAutocompletePlugin = (options, api) => {
           return accepted;
         },
         focus: () => {
-          var _options$useLocalMode, _options$useLocalMode2;
+          var _options$useLocalMode, _options$useLocalMode2, _options$useLocalMode3;
           if (!isAutocompleteEnabled) {
             return false;
           }
@@ -678,7 +679,6 @@ export const createAutocompletePlugin = (options, api) => {
             });
           });
           loadVectorsAsync({
-            getBinaryUrl: options === null || options === void 0 ? void 0 : options.getVectorsBinaryUrl,
             isLocalLLM: (_options$useLocalMode2 = options === null || options === void 0 ? void 0 : options.useLocalModel) !== null && _options$useLocalMode2 !== void 0 ? _options$useLocalMode2 : false,
             surface
           }).catch(error => {
@@ -686,6 +686,14 @@ export const createAutocompletePlugin = (options, api) => {
               location: 'editor-plugin-autocomplete/loadVectorsAsync'
             });
           });
+          loadGrammarDataAsync({
+            isLocalLLM: (_options$useLocalMode3 = options === null || options === void 0 ? void 0 : options.useLocalModel) !== null && _options$useLocalMode3 !== void 0 ? _options$useLocalMode3 : false,
+            surface
+          }).catch(error => {
+            logException(error, {
+              location: 'editor-plugin-autocomplete/loadGrammarDataAsync'
+            });
+          });
           if (!hasIngestedPage) {
             hasIngestedPage = true;
             refreshContext({