@semiont/jobs 0.5.1 → 0.5.3

package/dist/worker-main.js

@@ -1,15 +1,15 @@
-import { InMemorySessionStorage, setStoredSession, kbBackendUrl, SemiontClient, SemiontSession, createJobClaimAdapter } from '@semiont/sdk';
 import { createTomlConfigLoader, baseUrl, RESOURCE_BROADCAST_TYPES, resourceId, validateAndCorrectOffsets, didToAgent, getLocaleEnglishName } from '@semiont/core';
 import { deriveStorageUri } from '@semiont/content';
-import { getLogTraceContext, withSpan, SpanKind, recordJobOutcome } from '@semiont/observability';
+import { withSpan, SpanKind, recordJobOutcome } from '@semiont/observability';
 import { generateAnnotationId } from '@semiont/event-sourcing';
 import { createInferenceClient } from '@semiont/inference';
 import { createServer } from 'http';
 import { existsSync, readFileSync } from 'fs';
 import { homedir, hostname } from 'os';
 import { join } from 'path';
+import { InMemorySessionStorage, setStoredSession, kbBackendUrl, SemiontClient, SemiontSession } from '@semiont/sdk';
 import { HttpTransport, HttpContentTransport } from '@semiont/api-client';
-import winston from 'winston';
+import { createProcessLogger } from '@semiont/observability/process-logger';
 
 var __create = Object.create;
 var __defProp = Object.defineProperty;
@@ -322,16 +322,16 @@ var require_timeoutProvider = __commonJS({
     Object.defineProperty(exports$1, "__esModule", { value: true });
     exports$1.timeoutProvider = void 0;
     exports$1.timeoutProvider = {
-      setTimeout: function(handler, timeout) {
+      setTimeout: function(handler, timeout2) {
         var args = [];
         for (var _i = 2; _i < arguments.length; _i++) {
           args[_i - 2] = arguments[_i];
         }
         var delegate = exports$1.timeoutProvider.delegate;
         if (delegate === null || delegate === void 0 ? void 0 : delegate.setTimeout) {
-          return delegate.setTimeout.apply(delegate, __spreadArray([handler, timeout], __read(args)));
+          return delegate.setTimeout.apply(delegate, __spreadArray([handler, timeout2], __read(args)));
         }
-        return setTimeout.apply(void 0, __spreadArray([handler, timeout], __read(args)));
+        return setTimeout.apply(void 0, __spreadArray([handler, timeout2], __read(args)));
       },
       clearTimeout: function(handle) {
         var delegate = exports$1.timeoutProvider.delegate;
@@ -703,19 +703,19 @@ var require_Observable = __commonJS({
     var config_1 = require_config();
     var isFunction_1 = require_isFunction();
     var errorContext_1 = require_errorContext();
-    var Observable = (function() {
-      function Observable2(subscribe) {
+    var Observable2 = (function() {
+      function Observable3(subscribe) {
         if (subscribe) {
           this._subscribe = subscribe;
         }
       }
-      Observable2.prototype.lift = function(operator) {
-        var observable = new Observable2();
+      Observable3.prototype.lift = function(operator) {
+        var observable = new Observable3();
         observable.source = this;
         observable.operator = operator;
         return observable;
       };
-      Observable2.prototype.subscribe = function(observerOrNext, error, complete) {
+      Observable3.prototype.subscribe = function(observerOrNext, error, complete) {
         var _this = this;
         var subscriber = isSubscriber(observerOrNext) ? observerOrNext : new Subscriber_1.SafeSubscriber(observerOrNext, error, complete);
         errorContext_1.errorContext(function() {
@@ -724,14 +724,14 @@ var require_Observable = __commonJS({
         });
         return subscriber;
       };
-      Observable2.prototype._trySubscribe = function(sink) {
+      Observable3.prototype._trySubscribe = function(sink) {
         try {
           return this._subscribe(sink);
         } catch (err) {
           sink.error(err);
         }
       };
-      Observable2.prototype.forEach = function(next, promiseCtor) {
+      Observable3.prototype.forEach = function(next, promiseCtor) {
         var _this = this;
         promiseCtor = getPromiseCtor(promiseCtor);
         return new promiseCtor(function(resolve, reject) {
@@ -750,21 +750,21 @@ var require_Observable = __commonJS({
           _this.subscribe(subscriber);
         });
       };
-      Observable2.prototype._subscribe = function(subscriber) {
+      Observable3.prototype._subscribe = function(subscriber) {
         var _a;
         return (_a = this.source) === null || _a === void 0 ? void 0 : _a.subscribe(subscriber);
       };
-      Observable2.prototype[observable_1.observable] = function() {
+      Observable3.prototype[observable_1.observable] = function() {
         return this;
       };
-      Observable2.prototype.pipe = function() {
+      Observable3.prototype.pipe = function() {
         var operations = [];
         for (var _i = 0; _i < arguments.length; _i++) {
           operations[_i] = arguments[_i];
         }
         return pipe_1.pipeFromArray(operations)(this);
       };
-      Observable2.prototype.toPromise = function(promiseCtor) {
+      Observable3.prototype.toPromise = function(promiseCtor) {
         var _this = this;
         promiseCtor = getPromiseCtor(promiseCtor);
         return new promiseCtor(function(resolve, reject) {
@@ -778,12 +778,12 @@ var require_Observable = __commonJS({
           });
         });
       };
-      Observable2.create = function(subscribe) {
-        return new Observable2(subscribe);
+      Observable3.create = function(subscribe) {
+        return new Observable3(subscribe);
       };
-      return Observable2;
+      return Observable3;
     })();
-    exports$1.Observable = Observable;
+    exports$1.Observable = Observable2;
     function getPromiseCtor(promiseCtor) {
       var _a;
       return (_a = promiseCtor !== null && promiseCtor !== void 0 ? promiseCtor : config_1.config.Promise) !== null && _a !== void 0 ? _a : Promise;
@@ -1203,9 +1203,9 @@ var require_Subject = __commonJS({
     var ObjectUnsubscribedError_1 = require_ObjectUnsubscribedError();
     var arrRemove_1 = require_arrRemove();
     var errorContext_1 = require_errorContext();
-    var Subject = (function(_super) {
-      __extends(Subject2, _super);
-      function Subject2() {
+    var Subject2 = (function(_super) {
+      __extends(Subject3, _super);
+      function Subject3() {
         var _this = _super.call(this) || this;
         _this.closed = false;
         _this.currentObservers = null;
@@ -1215,17 +1215,17 @@ var require_Subject = __commonJS({
         _this.thrownError = null;
         return _this;
       }
-      Subject2.prototype.lift = function(operator) {
+      Subject3.prototype.lift = function(operator) {
         var subject = new AnonymousSubject(this, this);
         subject.operator = operator;
         return subject;
       };
-      Subject2.prototype._throwIfClosed = function() {
+      Subject3.prototype._throwIfClosed = function() {
         if (this.closed) {
           throw new ObjectUnsubscribedError_1.ObjectUnsubscribedError();
         }
       };
-      Subject2.prototype.next = function(value) {
+      Subject3.prototype.next = function(value) {
         var _this = this;
         errorContext_1.errorContext(function() {
           var e_1, _a;
@@ -1251,7 +1251,7 @@ var require_Subject = __commonJS({
           }
         });
       };
-      Subject2.prototype.error = function(err) {
+      Subject3.prototype.error = function(err) {
         var _this = this;
         errorContext_1.errorContext(function() {
           _this._throwIfClosed();
@@ -1265,7 +1265,7 @@ var require_Subject = __commonJS({
           }
         });
       };
-      Subject2.prototype.complete = function() {
+      Subject3.prototype.complete = function() {
         var _this = this;
         errorContext_1.errorContext(function() {
           _this._throwIfClosed();
@@ -1278,11 +1278,11 @@ var require_Subject = __commonJS({
           }
         });
       };
-      Subject2.prototype.unsubscribe = function() {
+      Subject3.prototype.unsubscribe = function() {
         this.isStopped = this.closed = true;
         this.observers = this.currentObservers = null;
       };
-      Object.defineProperty(Subject2.prototype, "observed", {
+      Object.defineProperty(Subject3.prototype, "observed", {
         get: function() {
           var _a;
           return ((_a = this.observers) === null || _a === void 0 ? void 0 : _a.length) > 0;
@@ -1290,16 +1290,16 @@ var require_Subject = __commonJS({
         enumerable: false,
         configurable: true
       });
-      Subject2.prototype._trySubscribe = function(subscriber) {
+      Subject3.prototype._trySubscribe = function(subscriber) {
         this._throwIfClosed();
         return _super.prototype._trySubscribe.call(this, subscriber);
       };
-      Subject2.prototype._subscribe = function(subscriber) {
+      Subject3.prototype._subscribe = function(subscriber) {
         this._throwIfClosed();
         this._checkFinalizedStatuses(subscriber);
         return this._innerSubscribe(subscriber);
       };
-      Subject2.prototype._innerSubscribe = function(subscriber) {
+      Subject3.prototype._innerSubscribe = function(subscriber) {
         var _this = this;
         var _a = this, hasError = _a.hasError, isStopped = _a.isStopped, observers = _a.observers;
         if (hasError || isStopped) {
@@ -1312,7 +1312,7 @@ var require_Subject = __commonJS({
           arrRemove_1.arrRemove(observers, subscriber);
         });
       };
-      Subject2.prototype._checkFinalizedStatuses = function(subscriber) {
+      Subject3.prototype._checkFinalizedStatuses = function(subscriber) {
         var _a = this, hasError = _a.hasError, thrownError = _a.thrownError, isStopped = _a.isStopped;
         if (hasError) {
           subscriber.error(thrownError);
@@ -1320,17 +1320,17 @@ var require_Subject = __commonJS({
           subscriber.complete();
         }
       };
-      Subject2.prototype.asObservable = function() {
+      Subject3.prototype.asObservable = function() {
         var observable = new Observable_1.Observable();
         observable.source = this;
         return observable;
       };
-      Subject2.create = function(destination, source) {
+      Subject3.create = function(destination, source) {
         return new AnonymousSubject(destination, source);
       };
-      return Subject2;
+      return Subject3;
     })(Observable_1.Observable);
-    exports$1.Subject = Subject;
+    exports$1.Subject = Subject2;
     var AnonymousSubject = (function(_super) {
       __extends(AnonymousSubject2, _super);
       function AnonymousSubject2(destination, source) {
@@ -1356,7 +1356,7 @@ var require_Subject = __commonJS({
         return (_b = (_a = this.source) === null || _a === void 0 ? void 0 : _a.subscribe(subscriber)) !== null && _b !== void 0 ? _b : Subscription_1.EMPTY_SUBSCRIPTION;
       };
       return AnonymousSubject2;
-    })(Subject);
+    })(Subject2);
     exports$1.AnonymousSubject = AnonymousSubject;
   }
 });
@@ -1386,26 +1386,26 @@ var require_BehaviorSubject = __commonJS({
     Object.defineProperty(exports$1, "__esModule", { value: true });
     exports$1.BehaviorSubject = void 0;
     var Subject_1 = require_Subject();
-    var BehaviorSubject2 = (function(_super) {
-      __extends(BehaviorSubject3, _super);
-      function BehaviorSubject3(_value) {
+    var BehaviorSubject3 = (function(_super) {
+      __extends(BehaviorSubject4, _super);
+      function BehaviorSubject4(_value) {
         var _this = _super.call(this) || this;
         _this._value = _value;
         return _this;
       }
-      Object.defineProperty(BehaviorSubject3.prototype, "value", {
+      Object.defineProperty(BehaviorSubject4.prototype, "value", {
         get: function() {
           return this.getValue();
         },
         enumerable: false,
         configurable: true
       });
-      BehaviorSubject3.prototype._subscribe = function(subscriber) {
+      BehaviorSubject4.prototype._subscribe = function(subscriber) {
         var subscription = _super.prototype._subscribe.call(this, subscriber);
         !subscription.closed && subscriber.next(this._value);
         return subscription;
       };
-      BehaviorSubject3.prototype.getValue = function() {
+      BehaviorSubject4.prototype.getValue = function() {
         var _a = this, hasError = _a.hasError, thrownError = _a.thrownError, _value = _a._value;
         if (hasError) {
           throw thrownError;
@@ -1413,12 +1413,12 @@ var require_BehaviorSubject = __commonJS({
         this._throwIfClosed();
         return _value;
       };
-      BehaviorSubject3.prototype.next = function(value) {
+      BehaviorSubject4.prototype.next = function(value) {
         _super.prototype.next.call(this, this._value = value);
       };
-      return BehaviorSubject3;
+      return BehaviorSubject4;
     })(Subject_1.Subject);
-    exports$1.BehaviorSubject = BehaviorSubject2;
+    exports$1.BehaviorSubject = BehaviorSubject3;
   }
 });
 
@@ -1655,16 +1655,16 @@ var require_intervalProvider = __commonJS({
     Object.defineProperty(exports$1, "__esModule", { value: true });
     exports$1.intervalProvider = void 0;
     exports$1.intervalProvider = {
-      setInterval: function(handler, timeout) {
+      setInterval: function(handler, timeout2) {
         var args = [];
         for (var _i = 2; _i < arguments.length; _i++) {
           args[_i - 2] = arguments[_i];
         }
         var delegate = exports$1.intervalProvider.delegate;
         if (delegate === null || delegate === void 0 ? void 0 : delegate.setInterval) {
-          return delegate.setInterval.apply(delegate, __spreadArray([handler, timeout], __read(args)));
+          return delegate.setInterval.apply(delegate, __spreadArray([handler, timeout2], __read(args)));
         }
-        return setInterval.apply(void 0, __spreadArray([handler, timeout], __read(args)));
+        return setInterval.apply(void 0, __spreadArray([handler, timeout2], __read(args)));
       },
       clearInterval: function(handle) {
         var delegate = exports$1.intervalProvider.delegate;
@@ -3562,7 +3562,7 @@ var require_firstValueFrom = __commonJS({
     exports$1.firstValueFrom = void 0;
     var EmptyError_1 = require_EmptyError();
     var Subscriber_1 = require_Subscriber();
-    function firstValueFrom(source, config) {
+    function firstValueFrom2(source, config) {
       var hasConfig = typeof config === "object";
       return new Promise(function(resolve, reject) {
         var subscriber = new Subscriber_1.SafeSubscriber({
@@ -3582,7 +3582,7 @@ var require_firstValueFrom = __commonJS({
         source.subscribe(subscriber);
       });
     }
-    exports$1.firstValueFrom = firstValueFrom;
+    exports$1.firstValueFrom = firstValueFrom2;
   }
 });
 
@@ -3669,7 +3669,7 @@ var require_timeout = __commonJS({
         this.info = info;
       };
     });
-    function timeout(config, schedulerArg) {
+    function timeout2(config, schedulerArg) {
       var _a = isDate_1.isValidDate(config) ? { first: config } : typeof config === "number" ? { each: config } : config, first = _a.first, each = _a.each, _b = _a.with, _with = _b === void 0 ? timeoutErrorFactory : _b, _c = _a.scheduler, scheduler = _c === void 0 ? schedulerArg !== null && schedulerArg !== void 0 ? schedulerArg : async_1.asyncScheduler : _c, _d = _a.meta, meta = _d === void 0 ? null : _d;
       if (first == null && each == null) {
         throw new TypeError("No timeout provided.");
@@ -3707,7 +3707,7 @@ var require_timeout = __commonJS({
         !seen && startTimer(first != null ? typeof first === "number" ? first : +first - scheduler.now() : each);
       });
     }
-    exports$1.timeout = timeout;
+    exports$1.timeout = timeout2;
     function timeoutErrorFactory(info) {
       throw new exports$1.TimeoutError(info);
     }
@@ -3721,7 +3721,7 @@ var require_map = __commonJS({
     exports$1.map = void 0;
     var lift_1 = require_lift();
     var OperatorSubscriber_1 = require_OperatorSubscriber();
-    function map(project, thisArg) {
+    function map2(project, thisArg) {
       return lift_1.operate(function(source, subscriber) {
         var index = 0;
         source.subscribe(OperatorSubscriber_1.createOperatorSubscriber(subscriber, function(value) {
@@ -3729,7 +3729,7 @@ var require_map = __commonJS({
         }));
       });
     }
-    exports$1.map = map;
+    exports$1.map = map2;
   }
 });
 
@@ -4629,7 +4629,7 @@ var require_merge = __commonJS({
     var empty_1 = require_empty();
     var args_1 = require_args();
     var from_1 = require_from();
-    function merge() {
+    function merge2() {
       var args = [];
       for (var _i = 0; _i < arguments.length; _i++) {
         args[_i] = arguments[_i];
@@ -4639,7 +4639,7 @@ var require_merge = __commonJS({
       var sources = args;
       return !sources.length ? empty_1.EMPTY : sources.length === 1 ? innerFrom_1.innerFrom(sources[0]) : mergeAll_1.mergeAll(concurrent)(from_1.from(sources, scheduler));
     }
-    exports$1.merge = merge;
+    exports$1.merge = merge2;
   }
 });
 
@@ -4746,7 +4746,7 @@ var require_filter = __commonJS({
     exports$1.filter = void 0;
     var lift_1 = require_lift();
     var OperatorSubscriber_1 = require_OperatorSubscriber();
-    function filter(predicate, thisArg) {
+    function filter2(predicate, thisArg) {
       return lift_1.operate(function(source, subscriber) {
         var index = 0;
         source.subscribe(OperatorSubscriber_1.createOperatorSubscriber(subscriber, function(value) {
@@ -4754,7 +4754,7 @@ var require_filter = __commonJS({
         }));
       });
     }
-    exports$1.filter = filter;
+    exports$1.filter = filter2;
   }
 });
 
@@ -5881,7 +5881,7 @@ var require_take = __commonJS({
     var empty_1 = require_empty();
     var lift_1 = require_lift();
     var OperatorSubscriber_1 = require_OperatorSubscriber();
-    function take(count) {
+    function take2(count) {
       return count <= 0 ? function() {
         return empty_1.EMPTY;
       } : lift_1.operate(function(source, subscriber) {
@@ -5896,7 +5896,7 @@ var require_take = __commonJS({
         }));
       });
     }
-    exports$1.take = take;
+    exports$1.take = take2;
   }
 });
 
@@ -6688,7 +6688,7 @@ var require_merge2 = __commonJS({
     var mergeAll_1 = require_mergeAll();
     var args_1 = require_args();
     var from_1 = require_from();
-    function merge() {
+    function merge2() {
       var args = [];
       for (var _i = 0; _i < arguments.length; _i++) {
         args[_i] = arguments[_i];
@@ -6699,7 +6699,7 @@ var require_merge2 = __commonJS({
         mergeAll_1.mergeAll(concurrent)(from_1.from(__spreadArray([source], __read(args)), scheduler)).subscribe(subscriber);
       });
     }
-    exports$1.merge = merge;
+    exports$1.merge = merge2;
   }
 });
 
@@ -9201,7 +9201,112 @@ var require_cjs = __commonJS({
   }
 });
 
-// src/workers/detection/motivation-prompts.ts
+// src/job-claim-adapter.ts
+var import_rxjs = __toESM(require_cjs());
+var import_rxjs2 = __toESM(require_cjs());
+function createJobClaimAdapter(options) {
+  const { bus, jobTypes } = options;
+  const activeJob$ = new import_rxjs.BehaviorSubject(null);
+  const isProcessing$ = new import_rxjs.BehaviorSubject(false);
+  const jobsCompleted$ = new import_rxjs.BehaviorSubject(0);
+  const errors$ = new import_rxjs.Subject();
+  let jobSubscription = null;
+  let started = false;
+  const claimJob = async (assignment) => {
+    try {
+      const correlationId = crypto.randomUUID();
+      const result$ = (0, import_rxjs2.merge)(
+        bus.on$("job:claimed").pipe(
+          (0, import_rxjs2.filter)((e) => e.correlationId === correlationId),
+          (0, import_rxjs2.map)((e) => ({ ok: true, response: e.response }))
+        ),
+        bus.on$("job:claim-failed").pipe(
+          (0, import_rxjs2.filter)((e) => e.correlationId === correlationId),
+          (0, import_rxjs2.map)(() => ({ ok: false }))
+        )
+      ).pipe((0, import_rxjs2.take)(1), (0, import_rxjs2.timeout)(1e4));
+      const resultPromise = (0, import_rxjs2.firstValueFrom)(result$);
+      await bus.emit("job:claim", { correlationId, jobId: assignment.jobId });
+      const result = await resultPromise;
+      if (!result.ok) return null;
+      const job = result.response;
+      return {
+        jobId: assignment.jobId,
+        type: assignment.type,
+        resourceId: assignment.resourceId,
+        userId: job.metadata?.userId ?? "",
+        params: job.params ?? {}
+      };
+    } catch {
+      return null;
+    }
+  };
+  return {
+    activeJob$: activeJob$.asObservable(),
+    isProcessing$: isProcessing$.asObservable(),
+    jobsCompleted$: jobsCompleted$.asObservable(),
+    errors$: errors$.asObservable(),
+    start: () => {
+      if (started) return;
+      started = true;
+      bus.addChannels?.(["job:queued"]);
+      jobSubscription = bus.on$("job:queued").subscribe((event) => {
+        const jobType = event.jobType;
+        if (jobTypes.length > 0 && !jobTypes.includes(jobType)) return;
+        if (isProcessing$.getValue()) return;
+        isProcessing$.next(true);
+        claimJob({ jobId: event.jobId, type: jobType, resourceId: event.resourceId }).then((job) => {
+          if (job) {
+            activeJob$.next(job);
+          } else {
+            isProcessing$.next(false);
+          }
+        }).catch(() => {
+          isProcessing$.next(false);
+        });
+      });
+    },
+    stop: () => {
+      jobSubscription?.unsubscribe();
+      jobSubscription = null;
+      started = false;
+    },
+    completeJob: () => {
+      activeJob$.next(null);
+      isProcessing$.next(false);
+      jobsCompleted$.next(jobsCompleted$.getValue() + 1);
+    },
+    failJob: (jid, error) => {
+      activeJob$.next(null);
+      isProcessing$.next(false);
+      errors$.next({ jobId: jid, error });
+    },
+    dispose: () => {
+      jobSubscription?.unsubscribe();
+      jobSubscription = null;
+      started = false;
+      activeJob$.complete();
+      isProcessing$.complete();
+      jobsCompleted$.complete();
+      errors$.complete();
+    }
+  };
+}
+function languageName(tag) {
+  return getLocaleEnglishName(tag) || tag;
+}
+function sourceLanguageGuidance(sourceLanguage) {
+  if (!sourceLanguage) return "";
+  return `
+
+Source text language: ${languageName(sourceLanguage)}.`;
+}
+function bodyLanguageGuidance(language, kind) {
+  if (!language || language === "en") return "";
+  return `
+
+IMPORTANT: Write your ${kind} in ${languageName(language)}.`;
+}
 var MotivationPrompts = class {
   /**
    * Build a prompt for detecting comment-worthy passages
@@ -9212,8 +9317,10 @@ var MotivationPrompts = class {
    * @param density - Optional target number of comments per 2000 words
    * @returns Formatted prompt string
    */
-  static buildCommentPrompt(content, instructions, tone, density) {
+  static buildCommentPrompt(content, instructions, tone, density, language, sourceLanguage) {
     let prompt;
+    const sourceLang = sourceLanguageGuidance(sourceLanguage);
+    const bodyLang = bodyLanguageGuidance(language, "comments");
     if (instructions) {
       const toneGuidance = tone ? ` Use a ${tone} tone.` : "";
       const densityGuidance = density ? `
@@ -9221,7 +9328,7 @@ var MotivationPrompts = class {
 Aim for approximately ${density} comments per 2000 words of text.` : "";
       prompt = `Add comments to passages in this text following these instructions:
 
-${instructions}${toneGuidance}${densityGuidance}
+${instructions}${toneGuidance}${densityGuidance}${sourceLang}${bodyLang}
 
 Text to analyze:
 ---
@@ -9257,7 +9364,7 @@ Guidelines:
 - Provide comments that ADD VALUE beyond restating the text
 - Focus on explanation, background, or connections to other ideas
 - Avoid obvious or trivial comments
-- Keep comments concise (1-3 sentences typically)${densityGuidance}
+- Keep comments concise (1-3 sentences typically)${densityGuidance}${sourceLang}${bodyLang}
 
 Text to analyze:
 ---
@@ -9289,15 +9396,16 @@ Example format:
    * @param density - Optional target number of highlights per 2000 words
    * @returns Formatted prompt string
    */
-  static buildHighlightPrompt(content, instructions, density) {
+  static buildHighlightPrompt(content, instructions, density, sourceLanguage) {
     let prompt;
+    const sourceLang = sourceLanguageGuidance(sourceLanguage);
     if (instructions) {
       const densityGuidance = density ? `
 
 Aim for approximately ${density} highlights per 2000 words of text.` : "";
       prompt = `Identify passages in this text to highlight following these instructions:
 
-${instructions}${densityGuidance}
+${instructions}${densityGuidance}${sourceLang}
 
 Text to analyze:
 ---
@@ -9330,7 +9438,7 @@ Guidelines:
 - Highlight notable quotes or particularly striking statements
 - Highlight critical decisions, action items, or turning points
 - Select passages that are SIGNIFICANT, not just interesting
-- Avoid trivial or obvious content${densityGuidance}
+- Avoid trivial or obvious content${densityGuidance}${sourceLang}
 
 Text to analyze:
 ---
@@ -9362,8 +9470,10 @@ Example format:
    * @param density - Optional target number of assessments per 2000 words
    * @returns Formatted prompt string
    */
-  static buildAssessmentPrompt(content, instructions, tone, density) {
+  static buildAssessmentPrompt(content, instructions, tone, density, language, sourceLanguage) {
     let prompt;
+    const sourceLang = sourceLanguageGuidance(sourceLanguage);
+    const bodyLang = bodyLanguageGuidance(language, "assessments");
     if (instructions) {
       const toneGuidance = tone ? ` Use a ${tone} tone.` : "";
       const densityGuidance = density ? `
@@ -9371,7 +9481,7 @@ Example format:
 Aim for approximately ${density} assessments per 2000 words of text.` : "";
       prompt = `Assess passages in this text following these instructions:
 
-${instructions}${toneGuidance}${densityGuidance}
+${instructions}${toneGuidance}${densityGuidance}${sourceLang}${bodyLang}
 
 Text to analyze:
 ---
@@ -9407,7 +9517,7 @@ Guidelines:
 - Assess evidence quality, logical soundness, or practical implications
 - Provide assessments that ADD INSIGHT beyond restating the text
 - Focus on passages where evaluation would help readers form judgments
-- Keep assessments concise yet substantive (1-3 sentences typically)${densityGuidance}
+- Keep assessments concise yet substantive (1-3 sentences typically)${densityGuidance}${sourceLang}${bodyLang}
 
 Text to analyze:
 ---
@@ -9443,7 +9553,8 @@ Example format:
    * @param categoryExamples - Example questions/guidance for this category
    * @returns Formatted prompt string
    */
-  static buildTagPrompt(content, category, schemaName, schemaDescription, schemaDomain, categoryDescription, categoryExamples) {
+  static buildTagPrompt(content, category, schemaName, schemaDescription, schemaDomain, categoryDescription, categoryExamples, sourceLanguage) {
+    const sourceLang = sourceLanguageGuidance(sourceLanguage);
     const prompt = `You are analyzing a text using the ${schemaName} framework.
 
 Schema: ${schemaDescription}
@@ -9462,7 +9573,7 @@ Guidelines:
 - Look for passages that explicitly fulfill this role
 - Passages can be sentences, paragraphs, or sections
 - Aim for precision - only tag passages that clearly serve this structural role
-- Typical documents have 1-5 instances of each category (some may have 0)
+- Typical documents have 1-5 instances of each category (some may have 0)${sourceLang}
 
 Text to analyze:
 ---
@@ -9693,159 +9804,6 @@ var MotivationParsers = class {
   }
 };
 
-// ../ontology/dist/index.js
-var TAG_SCHEMAS = {
-  "legal-irac": {
-    id: "legal-irac",
-    name: "Legal Analysis (IRAC)",
-    description: "Issue, Rule, Application, Conclusion framework for legal reasoning",
-    domain: "legal",
-    tags: [
-      {
-        name: "Issue",
-        description: "The legal question or problem to be resolved",
-        examples: [
-          "What is the central legal question?",
-          "What must the court decide?",
-          "What is the dispute about?"
-        ]
-      },
-      {
-        name: "Rule",
-        description: "The relevant law, statute, or legal principle",
-        examples: [
-          "What law applies?",
-          "What is the legal standard?",
-          "What statute governs this case?"
-        ]
-      },
-      {
-        name: "Application",
-        description: "How the rule applies to the specific facts",
-        examples: [
-          "How does the law apply to these facts?",
-          "Analysis of the case",
-          "How do the facts satisfy the legal standard?"
-        ]
-      },
-      {
-        name: "Conclusion",
-        description: "The resolution or outcome based on the analysis",
-        examples: [
-          "What is the court's decision?",
-          "What is the final judgment?",
-          "What is the holding?"
-        ]
-      }
-    ]
-  },
-  "scientific-imrad": {
-    id: "scientific-imrad",
-    name: "Scientific Paper (IMRAD)",
-    description: "Introduction, Methods, Results, Discussion structure for research papers",
-    domain: "scientific",
-    tags: [
-      {
-        name: "Introduction",
-        description: "Background, context, and research question",
-        examples: [
-          "What is the research question?",
-          "Why is this important?",
-          "What is the hypothesis?"
-        ]
-      },
-      {
-        name: "Methods",
-        description: "Experimental design and procedures",
-        examples: [
-          "How was the study conducted?",
-          "What methods were used?",
-          "What was the experimental design?"
-        ]
-      },
-      {
-        name: "Results",
-        description: "Findings and observations",
-        examples: [
-          "What did the study find?",
-          "What are the data?",
-          "What were the observations?"
-        ]
-      },
-      {
-        name: "Discussion",
-        description: "Interpretation and implications of results",
-        examples: [
-          "What do the results mean?",
-          "What are the implications?",
-          "How do these findings relate to prior work?"
-        ]
-      }
-    ]
-  },
-  "argument-toulmin": {
-    id: "argument-toulmin",
-    name: "Argument Structure (Toulmin)",
-    description: "Claim, Evidence, Warrant, Counterargument, Rebuttal framework for argumentation",
-    domain: "general",
-    tags: [
-      {
-        name: "Claim",
-        description: "The main assertion or thesis",
-        examples: [
-          "What is being argued?",
-          "What is the main point?",
-          "What position is being taken?"
-        ]
-      },
-      {
-        name: "Evidence",
-        description: "Data or facts supporting the claim",
-        examples: [
-          "What supports this claim?",
-          "What are the facts?",
-          "What data is provided?"
-        ]
-      },
-      {
-        name: "Warrant",
-        description: "Reasoning connecting evidence to claim",
-        examples: [
-          "Why does this evidence support the claim?",
-          "What is the logic?",
-          "How does this reasoning work?"
-        ]
-      },
-      {
-        name: "Counterargument",
-        description: "Opposing viewpoints or objections",
-        examples: [
-          "What are the objections?",
-          "What do critics say?",
-          "What are alternative views?"
-        ]
-      },
-      {
-        name: "Rebuttal",
-        description: "Response to counterarguments",
-        examples: [
-          "How is the objection addressed?",
-          "Why is the counterargument wrong?",
-          "How is the criticism answered?"
-        ]
-      }
-    ]
-  }
-};
-function getTagSchema(schemaId) {
-  return TAG_SCHEMAS[schemaId] || null;
-}
-function getSchemaCategory(schemaId, categoryName) {
-  const schema = getTagSchema(schemaId);
-  if (!schema) return null;
-  return schema.tags.find((tag) => tag.name === categoryName) || null;
-}
-
 // src/workers/annotation-detection.ts
 var AnnotationDetection = class {
   /**
@@ -9864,40 +9822,58 @@ var AnnotationDetection = class {
     return Buffer.concat(chunks).toString("utf-8");
   }
   /**
-   * Detect comments in content
+   * Detect comments in content.
+   *
+   * `language` is the locale the LLM should write comment text in (annotation
+   * body locale). `sourceLanguage` is the locale of the content being analyzed
+   * (source-resource locale). See `types.ts` "Locale conventions" for the
+   * full discussion.
    */
-  static async detectComments(content, client, instructions, tone, density) {
-    const prompt = MotivationPrompts.buildCommentPrompt(content, instructions, tone, density);
+  static async detectComments(content, client, instructions, tone, density, language, sourceLanguage) {
+    const prompt = MotivationPrompts.buildCommentPrompt(content, instructions, tone, density, language, sourceLanguage);
     const response = await client.generateText(prompt, 3e3, 0.4);
     return MotivationParsers.parseComments(response, content);
   }
   /**
-   * Detect highlights in content
+   * Detect highlights in content.
+   *
+   * Highlights have no body — only `sourceLanguage` (source-resource locale)
+   * applies, used in the prompt so the LLM analyzes non-English source
+   * correctly.
    */
-  static async detectHighlights(content, client, instructions, density) {
-    const prompt = MotivationPrompts.buildHighlightPrompt(content, instructions, density);
+  static async detectHighlights(content, client, instructions, density, sourceLanguage) {
+    const prompt = MotivationPrompts.buildHighlightPrompt(content, instructions, density, sourceLanguage);
     const response = await client.generateText(prompt, 2e3, 0.3);
     return MotivationParsers.parseHighlights(response, content);
   }
   /**
-   * Detect assessments in content
+   * Detect assessments in content.
+   *
+   * `language` is the locale the LLM should write assessment text in
+   * (annotation body locale). `sourceLanguage` is the locale of the content
+   * being analyzed (source-resource locale).
    */
-  static async detectAssessments(content, client, instructions, tone, density) {
-    const prompt = MotivationPrompts.buildAssessmentPrompt(content, instructions, tone, density);
+  static async detectAssessments(content, client, instructions, tone, density, language, sourceLanguage) {
+    const prompt = MotivationPrompts.buildAssessmentPrompt(content, instructions, tone, density, language, sourceLanguage);
     const response = await client.generateText(prompt, 3e3, 0.3);
     return MotivationParsers.parseAssessments(response, content);
   }
   /**
-   * Detect tags in content for a specific category
+   * Detect tags in content for a specific category.
+   *
+   * The full `TagSchema` is supplied by the dispatcher (resolved against
+   * the per-KB tag-schema projection at job-creation time) so the worker
+   * is independent of the registry.
+   *
+   * `sourceLanguage` is the locale of the content being analyzed. Body-locale
+   * (`language`) doesn't influence the tag prompt — categories are schema
+   * identifiers, not LLM-generated text — so it's consumed at the body-stamp
+   * site, not here.
    */
-  static async detectTags(content, client, schemaId, category) {
-    const schema = getTagSchema(schemaId);
-    if (!schema) {
-      throw new Error(`Invalid tag schema: ${schemaId}`);
-    }
-    const categoryInfo = getSchemaCategory(schemaId, category);
+  static async detectTags(content, client, schema, category, sourceLanguage) {
+    const categoryInfo = schema.tags.find((t) => t.name === category);
     if (!categoryInfo) {
-      throw new Error(`Invalid category "${category}" for schema ${schemaId}`);
+      throw new Error(`Invalid category "${category}" for schema ${schema.id}`);
     }
     const prompt = MotivationPrompts.buildTagPrompt(
       content,
@@ -9906,16 +9882,15 @@ var AnnotationDetection = class {
       schema.description,
       schema.domain,
       categoryInfo.description,
-      categoryInfo.examples
+      categoryInfo.examples,
+      sourceLanguage
     );
     const response = await client.generateText(prompt, 4e3, 0.2);
     const parsedTags = MotivationParsers.parseTags(response);
     return MotivationParsers.validateTagOffsets(parsedTags, content, category);
   }
 };
-
-// src/workers/detection/entity-extractor.ts
-async function extractEntities(exact, entityTypes, client, includeDescriptiveReferences = false, logger2) {
+async function extractEntities(exact, entityTypes, client, includeDescriptiveReferences = false, logger2, sourceLanguage) {
   const entityTypesDescription = entityTypes.map((et) => {
     if (typeof et === "string") {
       return et;
@@ -9944,8 +9919,11 @@ Examples:
 ` : `
 Find direct mentions only (names, proper nouns). Do not include pronouns or descriptive references.
 `;
+  const sourceLangGuidance = sourceLanguage ? `
+Source text language: ${getLocaleEnglishName(sourceLanguage) || sourceLanguage}.
+` : "";
   const prompt = `Identify entity references in the following text. Look for mentions of: ${entityTypesDescription}.
-${descriptiveReferenceGuidance}
+${descriptiveReferenceGuidance}${sourceLangGuidance}
 Text to analyze:
 """
 ${exact}
@@ -10134,12 +10112,15 @@ Example output:
 function getLanguageName(locale) {
   return getLocaleEnglishName(locale) || locale;
 }
-async function generateResourceFromTopic(topic, entityTypes, client, userPrompt, locale, context, temperature, maxTokens, logger2) {
+async function generateResourceFromTopic(topic, entityTypes, client, userPrompt, locale, context, temperature, maxTokens, logger2, sourceLanguage) {
   const finalTemperature = temperature ?? 0.7;
   const finalMaxTokens = maxTokens ?? 500;
   const languageInstruction = locale && locale !== "en" ? `
 
 IMPORTANT: Write the entire resource in ${getLanguageName(locale)}.` : "";
+  const sourceLanguageInstruction = sourceLanguage ? `
+
+The source resource and embedded context are in ${getLanguageName(sourceLanguage)}.` : "";
   let annotationSection = "";
   if (context) {
     const parts = [];
@@ -10201,7 +10182,7 @@ ${parts.join("\n")}`;
   const structureGuidance = finalMaxTokens >= 1e3 ? "organized into titled sections (## Section) with well-structured paragraphs" : "organized into well-structured paragraphs";
   const prompt = `Generate a concise, informative resource about "${topic}".
 ${entityTypes.length > 0 ? `Focus on these entity types: ${entityTypes.join(", ")}.` : ""}
-${userPrompt ? `Additional context: ${userPrompt}` : ""}${annotationSection}${contextSection}${graphContextSection}${languageInstruction}
+${userPrompt ? `Additional context: ${userPrompt}` : ""}${annotationSection}${contextSection}${graphContextSection}${sourceLanguageInstruction}${languageInstruction}
 
 Requirements:
 - Start with a clear heading (# Title)
@@ -10266,7 +10247,8 @@ async function processHighlightJob(content, inferenceClient, params, userId, gen
     content,
     inferenceClient,
     params.instructions,
-    params.density
+    params.density,
+    params.sourceLanguage
   );
   onProgress(60, `Creating ${highlights.length} annotations...`, "creating");
   const annotations = highlights.map(
@@ -10286,16 +10268,19 @@ async function processCommentJob(content, inferenceClient, params, userId, gener
     inferenceClient,
     params.instructions,
     params.tone,
-    params.density
+    params.density,
+    params.language,
+    params.sourceLanguage
   );
   onProgress(60, `Creating ${comments.length} annotations...`, "creating");
+  const bodyLanguage = params.language ?? "en";
   const annotations = comments.map(
     (c) => (
       // Match the pre-#651 CommentAnnotationWorker: include format and
       // language on the body TextualBody. Optional in the schema, but
       // consumers that do language-aware rendering rely on them.
       buildTextAnnotation(params.resourceId, userId, generator, "commenting", c, [
-        { type: "TextualBody", value: c.comment, purpose: "commenting", format: "text/plain", language: "en" }
+        { type: "TextualBody", value: c.comment, purpose: "commenting", format: "text/plain", language: bodyLanguage }
       ])
     )
   );
@@ -10313,9 +10298,12 @@ async function processAssessmentJob(content, inferenceClient, params, userId, ge
     inferenceClient,
     params.instructions,
     params.tone,
-    params.density
+    params.density,
+    params.language,
+    params.sourceLanguage
   );
   onProgress(60, `Creating ${assessments.length} annotations...`, "creating");
+  const bodyLanguage = params.language ?? "en";
   const annotations = assessments.map(
     (a) => (
       // Single-object body with purpose aligned to motivation, matching the
@@ -10329,7 +10317,7 @@ async function processAssessmentJob(content, inferenceClient, params, userId, ge
         value: a.assessment,
         purpose: "assessing",
         format: "text/plain",
-        language: "en"
+        language: bodyLanguage
       })
     )
   );
@@ -10348,6 +10336,7 @@ async function processReferenceJob(content, inferenceClient, params, userId, gen
   let errors = 0;
   const allAnnotations = [];
   onProgress(10, "Loading resource...", "analyzing", { requestParams });
+  const bodyLanguage = params.language ?? "en";
   for (let i = 0; i < entityTypeNames.length; i++) {
     const entityTypeName = entityTypeNames[i];
     if (!entityTypeName) continue;
@@ -10366,11 +10355,14 @@ async function processReferenceJob(content, inferenceClient, params, userId, gen
       [entityTypeName],
       inferenceClient,
       params.includeDescriptiveReferences ?? false,
-      logger2
+      logger2,
+      params.sourceLanguage
     );
     totalFound += extractedEntities.length;
     completedEntityTypes.push({ entityType: entityTypeName, foundCount: extractedEntities.length });
-    const unresolvedBody = [{ type: "TextualBody", value: entityTypeName, purpose: "tagging" }];
+    const unresolvedBody = [
+      { type: "TextualBody", value: entityTypeName, purpose: "tagging", format: "text/plain", language: bodyLanguage }
+    ];
     for (const entity of extractedEntities) {
       try {
         const validated = validateAndCorrectOffsets(content, entity.startOffset, entity.endOffset, entity.exact);
@@ -10403,20 +10395,22 @@ async function processTagJob(content, inferenceClient, params, userId, generator
     const categoryTags = await AnnotationDetection.detectTags(
       content,
       inferenceClient,
-      params.schemaId,
-      category
+      params.schema,
+      category,
+      params.sourceLanguage
     );
     allTags.push(...categoryTags);
   }
   const tags = allTags;
   onProgress(60, `Creating ${tags.length} tag annotations...`, "creating");
+  const bodyLanguage = params.language ?? "en";
   const byCategory = {};
   const annotations = tags.map((t) => {
     const category = t.category ?? "unknown";
     byCategory[category] = (byCategory[category] ?? 0) + 1;
     return buildTextAnnotation(params.resourceId, userId, generator, "tagging", t, [
-      { type: "TextualBody", value: category, purpose: "tagging", format: "text/plain", language: "en" },
-      { type: "TextualBody", value: params.schemaId, purpose: "classifying", format: "text/plain" }
+      { type: "TextualBody", value: category, purpose: "tagging", format: "text/plain", language: bodyLanguage },
+      { type: "TextualBody", value: params.schema.id, purpose: "classifying", format: "text/plain" }
     ]);
   });
   onProgress(100, `Complete! Created ${annotations.length} tags`, "creating");
@@ -10438,7 +10432,10 @@ async function processGenerationJob(inferenceClient, params, onProgress) {
     params.language,
     params.context,
     params.temperature,
-    params.maxTokens
+    params.maxTokens,
+    void 0,
+    // logger
+    params.sourceLanguage
   );
   onProgress(85, "Creating resource...", "creating");
   return {
@@ -10463,7 +10460,7 @@ function startWorkerProcess(config) {
   const { session, logger: logger2 } = config;
   const httpTransport = session.client.transport;
   const adapter = createJobClaimAdapter({
-    actor: httpTransport.actor,
+    bus: httpTransport.actor,
     jobTypes: config.jobTypes
   });
   adapter.activeJob$.subscribe((job) => {
@@ -10654,6 +10651,7 @@ async function handleJobInner(adapter, config, job) {
       ...genParams.referenceId ? { sourceAnnotationId: genParams.referenceId } : {},
       ...genParams.prompt ? { generationPrompt: genParams.prompt } : {},
       ...genParams.language ? { language: genParams.language } : {},
+      ...genParams.entityTypes && genParams.entityTypes.length > 0 ? { entityTypes: genParams.entityTypes } : {},
       generator
     });
     await emitEvent(session, "job:complete", {
@@ -10667,41 +10665,7 @@ async function handleJobInner(adapter, config, job) {
 }
 
 // src/worker-main.ts
-var import_rxjs = __toESM(require_cjs());
-var traceContextFormat = winston.format((info) => {
-  const trace = getLogTraceContext();
-  if (trace) {
-    info.trace_id = trace.trace_id;
-    info.span_id = trace.span_id;
-  }
-  return info;
-})();
-function createProcessLogger(component) {
-  const level = process.env.LOG_LEVEL ?? "info";
-  const format = process.env.LOG_FORMAT === "simple" ? winston.format.combine(
-    winston.format.timestamp({ format: "YYYY-MM-DD HH:mm:ss" }),
-    winston.format.errors({ stack: true }),
-    traceContextFormat,
-    winston.format.printf(({ level: lvl, message, timestamp, ...meta }) => {
-      const metaStr = Object.keys(meta).length > 0 ? ` ${JSON.stringify(meta)}` : "";
-      return `${timestamp} [${lvl.toUpperCase()}] [${component}] ${message}${metaStr}`;
-    })
-  ) : winston.format.combine(
-    winston.format.timestamp(),
-    winston.format.errors({ stack: true }),
-    traceContextFormat,
-    winston.format.json()
-  );
-  const logger2 = winston.createLogger({
-    level,
-    defaultMeta: { component },
-    format,
-    transports: [new winston.transports.Console()]
-  });
-  return logger2;
-}
-
-// src/worker-main.ts
+var import_rxjs3 = __toESM(require_cjs());
 var ALL_JOB_TYPES = [
   "reference-annotation",
   "generation",
@@ -10804,25 +10768,24 @@ async function main() {
   logger.info("Authenticated");
   const { protocol, host, port } = parseBackendUrl(backendBaseUrl);
   const kbId = `worker-${hostname()}`;
+  const endpoint = { kind: "http", host, port, protocol };
   const kb = {
     id: kbId,
     label: `Worker pool @ ${host}`,
-    host,
-    port,
-    protocol,
-    email: `worker-pool@${host}`
+    email: `worker-pool@${host}`,
+    endpoint
   };
   const storage = new InMemorySessionStorage();
   setStoredSession(storage, kbId, { access: initialToken, refresh: "" });
-  const token$ = new import_rxjs.BehaviorSubject(null);
+  const token$ = new import_rxjs3.BehaviorSubject(null);
   let session;
   const transport = new HttpTransport({
-    baseUrl: baseUrl(kbBackendUrl(kb)),
+    baseUrl: baseUrl(kbBackendUrl(endpoint)),
     token$,
     tokenRefresher: () => session.refresh().then((t) => t ?? null)
   });
   const content = new HttpContentTransport(transport);
-  const client = new SemiontClient(transport, content);
+  const client = new SemiontClient(transport, content, transport);
   session = new SemiontSession({
     kb,
     storage,