@semiont/jobs 0.5.12 → 0.5.14

package/dist/index.d.ts

@@ -691,5 +691,37 @@ declare function generateResourceFromTopic(topic: string, entityTypes: string[],
     content: string;
 }>;
 
-export { AnnotationDetection, FsJobQueue, generateResourceFromTopic, isCancelledJob, isCompleteJob, isFailedJob, isPendingJob, isRunningJob, processAssessmentJob, processCommentJob, processGenerationJob, processHighlightJob, processReferenceJob, processTagJob };
+/**
+ * Worker Runtime — the importable half of the worker host.
+ *
+ * `worker-main.ts` is a process entrypoint (config at module scope,
+ * `main()` at import) and therefore untestable by construction; everything
+ * a unit test needs to reach lives here instead, fully parameterized — no
+ * module-scope env reads, no side effects at import.
+ *
+ * The load-bearing contract this module owns (and the reason it was
+ * extracted): a worker's stamped identity is the DID the
+ * `/api/tokens/agent` exchange MINTED for it, carried verbatim — never
+ * re-derived from the URL the worker happens to dial. One logical agent
+ * previously got two DIDs that way (.plans/bugs/agent-did-host-skew.md).
+ */
+
+/**
+ * Stall watchdog (WORKER-LIVENESS.md P3) — the fail-fast line behind the
+ * inference timeout. There is no poll loop to heartbeat; the honest
+ * stall signal in this push-driven architecture is *processing without
+ * activity*: an agent holding a claimed job whose `lastActivityAt`
+ * (claim / progress / finish) has stopped advancing is wedged — the
+ * adapter ignores every announcement while `isProcessing`, so a wedged
+ * agent never recovers on its own. Silent hang → loud crash → whatever
+ * restart policy the deployment chose.
+ *
+ * Thresholds are fixed by design (no env knobs) and deliberately
+ * layered: inference timeout (10 min, P2) fires first; this watchdog
+ * (15 min) catches the failure modes nobody predicted; the backend's
+ * dead-worker janitor (30 min) re-queues the job regardless.
+ */
+declare const STALL_THRESHOLD_MS: number;
+
+export { AnnotationDetection, FsJobQueue, STALL_THRESHOLD_MS, generateResourceFromTopic, isCancelledJob, isCompleteJob, isFailedJob, isPendingJob, isRunningJob, processAssessmentJob, processCommentJob, processGenerationJob, processHighlightJob, processReferenceJob, processTagJob };
 export type { AnyJob, AssessmentDetectionJob, AssessmentDetectionParams, AssessmentDetectionProgress, AssessmentDetectionResult, CancelledJob, CommentDetectionJob, CommentDetectionParams, CommentDetectionProgress, CommentDetectionResult, CompleteJob, DetectionJob, DetectionParams, DetectionProgress, DetectionResult, FailedJob, GenerationJob, GenerationParams, GenerationResult, HighlightDetectionJob, HighlightDetectionParams, HighlightDetectionProgress, HighlightDetectionResult, JobMetadata, JobQueryFilters, JobQueue, JobStatus, JobType, OnProgress, PendingJob, ProcessorResult, RunningJob, TagDetectionJob, TagDetectionParams, TagDetectionProgress, TagDetectionResult, YieldProgress };

package/dist/index.js

@@ -2,6 +2,10 @@ import { promises } from 'fs';
 import * as path from 'path';
 import { jobId, deriveViews, reconcileSelector, isObject, isString, getLocaleEnglishName, didToAgent, isArray } from '@semiont/core';
 import { generateAnnotationId } from '@semiont/event-sourcing';
+import '@semiont/content';
+import '@semiont/observability';
+import '@semiont/sdk';
+import '@semiont/http-transport';
 
 // src/fs-job-queue.ts
 var REANNOUNCE_INTERVAL_MS = 3e4;
@@ -419,6 +423,41 @@ function isFailedJob(job) {
 function isCancelledJob(job) {
   return job.status === "cancelled";
 }
+
+// src/workers/inference-call.ts
+var INFERENCE_TIMEOUT_MS = 10 * 6e4;
+async function withTimeout(work, label) {
+  let timer;
+  const timedOut = new Promise((_, reject) => {
+    timer = setTimeout(() => {
+      reject(new Error(
+        `Inference call timed out after ${INFERENCE_TIMEOUT_MS / 6e4} minutes (${label}) \u2014 failing the job to keep the claim loop live`
+      ));
+    }, INFERENCE_TIMEOUT_MS);
+    timer.unref?.();
+  });
+  try {
+    return await Promise.race([work, timedOut]);
+  } catch (err) {
+    work.catch(() => {
+    });
+    throw err;
+  } finally {
+    clearTimeout(timer);
+  }
+}
+function boundedGenerate(client, prompt, maxTokens, temperature, options) {
+  return withTimeout(
+    client.generateText(prompt, maxTokens, temperature, options),
+    `${client.type}:${client.modelId}`
+  );
+}
+function boundedGenerateWithMetadata(client, prompt, maxTokens, temperature, options) {
+  return withTimeout(
+    client.generateTextWithMetadata(prompt, maxTokens, temperature, options),
+    `${client.type}:${client.modelId}`
+  );
+}
 function languageName(tag) {
   return getLocaleEnglishName(tag) || tag;
 }
@@ -900,7 +939,7 @@ var AnnotationDetection = class {
    */
   static async detectComments(content, client, instructions, tone, density, language, sourceLanguage) {
     const prompt = MotivationPrompts.buildCommentPrompt(content, instructions, tone, density, language, sourceLanguage);
-    const response = await client.generateTextWithMetadata(prompt, 3e3, 0.4, { format: "json" });
+    const response = await boundedGenerateWithMetadata(client, prompt, 3e3, 0.4, { format: "json" });
     assertNotTruncated(response, "comment");
     return MotivationParsers.parseComments(response.text, content);
   }
@@ -913,7 +952,7 @@ var AnnotationDetection = class {
    */
   static async detectHighlights(content, client, instructions, density, sourceLanguage) {
     const prompt = MotivationPrompts.buildHighlightPrompt(content, instructions, density, sourceLanguage);
-    const response = await client.generateTextWithMetadata(prompt, 2e3, 0.3, { format: "json" });
+    const response = await boundedGenerateWithMetadata(client, prompt, 2e3, 0.3, { format: "json" });
     assertNotTruncated(response, "highlight");
     return MotivationParsers.parseHighlights(response.text, content);
   }
@@ -926,7 +965,7 @@ var AnnotationDetection = class {
    */
   static async detectAssessments(content, client, instructions, tone, density, language, sourceLanguage) {
     const prompt = MotivationPrompts.buildAssessmentPrompt(content, instructions, tone, density, language, sourceLanguage);
-    const response = await client.generateTextWithMetadata(prompt, 3e3, 0.3, { format: "json" });
+    const response = await boundedGenerateWithMetadata(client, prompt, 3e3, 0.3, { format: "json" });
     assertNotTruncated(response, "assessment");
     return MotivationParsers.parseAssessments(response.text, content);
   }
@@ -957,7 +996,7 @@ var AnnotationDetection = class {
       categoryInfo.examples,
       sourceLanguage
     );
-    const response = await client.generateTextWithMetadata(prompt, 4e3, 0.2, { format: "json" });
+    const response = await boundedGenerateWithMetadata(client, prompt, 4e3, 0.2, { format: "json" });
     assertNotTruncated(response, "tag");
     const parsedTags = MotivationParsers.parseTags(response.text);
     return MotivationParsers.validateTagOffsets(parsedTags, content, category);
@@ -1013,7 +1052,8 @@ If no entities are found, respond with an empty array [].
 Example output:
 [{"exact":"Alice","entityType":"Person","prefix":"","suffix":" went to"},{"exact":"Paris","entityType":"Location","prefix":"went to ","suffix":" yesterday"}]`;
   logger.debug("Sending entity extraction request", { entityTypes: entityTypesDescription });
-  const response = await client.generateTextWithMetadata(
+  const response = await boundedGenerateWithMetadata(
+    client,
     prompt,
     4e3,
     // Increased to handle many entities without truncation
@@ -1263,7 +1303,7 @@ ${formatRequirements}`;
     temperature: finalTemperature,
     maxTokens: finalMaxTokens
   });
-  const response = await client.generateText(prompt, finalMaxTokens, finalTemperature);
+  const response = await boundedGenerate(client, prompt, finalMaxTokens, finalTemperature);
   logger.debug("Got response from inference", { responseLength: response.length });
   const result = parseResponse(response);
   logger.debug("Parsed response", {
@@ -1661,7 +1701,8 @@ async function processGenerationJob(inferenceClient, params, onProgress, logger)
     }
   };
 }
+var STALL_THRESHOLD_MS = 15 * 6e4;
 
-export { AnnotationDetection, FsJobQueue, generateResourceFromTopic, isCancelledJob, isCompleteJob, isFailedJob, isPendingJob, isRunningJob, processAssessmentJob, processCommentJob, processGenerationJob, processHighlightJob, processReferenceJob, processTagJob };
+export { AnnotationDetection, FsJobQueue, STALL_THRESHOLD_MS, generateResourceFromTopic, isCancelledJob, isCompleteJob, isFailedJob, isPendingJob, isRunningJob, processAssessmentJob, processCommentJob, processGenerationJob, processHighlightJob, processReferenceJob, processTagJob };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map