@semiont/jobs 0.6.8 → 0.6.9

package/dist/index.d.ts

@@ -1,4 +1,4 @@
-import { JobId, components, UserId, UnitCursor, ResourceId, EntityType, JobReferenceAnnotationResult, GenerationJobParams, JobHighlightAnnotationResult, JobAssessmentAnnotationResult, JobCommentAnnotationResult, TagSchema, JobTagAnnotationResult, Annotation, Logger, SupportedMediaType, GatheredContext, BusRequestErrorCode, BusRequestPrimitive, EventMap } from '@semiont/core';
+import { JobId, components, UserId, UnitCursor, ResourceId, EntityType, JobReferenceAnnotationResult, GenerationJobParams, JobHighlightAnnotationResult, JobAssessmentAnnotationResult, JobCommentAnnotationResult, TagSchema, JobTagAnnotationResult, Annotation, Logger, SupportedMediaType, GatheredContext, JobType as JobType$1, EventMap, BusRequestErrorCode, BusRequestPrimitive } from '@semiont/core';
 import { InferenceClient } from '@semiont/inference';
 import { Observable } from 'rxjs';
 
@@ -142,16 +142,6 @@ type TagDetectionParams = {
     /** Source-resource locale — see locale conventions above. */
     sourceLanguage?: string;
 };
-/**
- * Generation job result
- */
-interface GenerationResult {
-    kind: 'generation';
-    resourceId: ResourceId;
-    resourceName: string;
-    /** True when the model stopped at the maxTokens ceiling — the artifact is cut off, not complete (GENERATE-FROM-RESOURCE D6). */
-    truncated: boolean;
-}
 /**
  * The progress a running job carries: the last its worker reported with
  * `job:report-progress`, or an empty object before the first report. One shape
@@ -216,7 +206,7 @@ type DetectionJob = Job<DetectionParams, JobReferenceAnnotationResult>;
 /** A generation job's params carry the resource the dispatcher derived from the context's focus. */
 type GenerationJob = Job<GenerationJobParams & {
     resourceId: ResourceId;
-}, GenerationResult>;
+}, components['schemas']['JobGenerationResult']>;
 type HighlightDetectionJob = Job<HighlightDetectionParams, JobHighlightAnnotationResult>;
 type AssessmentDetectionJob = Job<AssessmentDetectionParams, JobAssessmentAnnotationResult>;
 type CommentDetectionJob = Job<CommentDetectionParams, JobCommentAnnotationResult>;
@@ -250,7 +240,7 @@ declare function isCancelledJob(job: AnyJob): job is CancelledJob<any>;
 
 interface GenerationCitation {
     /** The cited source resource (validated present in the embedded context). */
-    resourceId: string;
+    resourceId: ResourceId;
     /** The contributing annotation, when the cited excerpt was annotation-derived. */
     annotationId?: string;
     /** Claim span in the FINAL (token-stripped) content; substring(start, end) === exact. */
@@ -259,6 +249,17 @@ interface GenerationCitation {
     exact: string;
 }
 
+/**
+ * Job Processors
+ *
+ * Pure functions that take content + inference client + params,
+ * report progress via callback, and return annotations + results.
+ *
+ * No EventBus, no queue, no side effects except calling inference.
+ * Driven by the remote worker process (worker-process.ts), which claims
+ * jobs over SSE and dispatches by jobType to these functions.
+ */
+
 /** A detected span — offsets into the extracted `.text`, plus optional context. */
 type SpanMatch = {
     exact: string;
@@ -378,7 +379,7 @@ declare function processGenerationJob(inferenceClient: InferenceClient, params:
     title: string;
     format: SupportedMediaType;
     citations: GenerationCitation[];
-    result: GenerationResult;
+    truncated: boolean;
 }>;
 
 /**
@@ -628,12 +629,17 @@ declare function generateResourceFromTopic(topic: string, entityTypes: string[],
  * an in-process worker could pass a shim wrapping `client.bus`.
  */
 
+/** The job a `job:claimed` reply carries, running under this worker: the spec's `JobRunning`. */
+type ClaimedJob = EventMap['job:claimed']['response'];
+/**
+ * The claimed job as the worker holds it: the fields of the claimed record the
+ * worker reads, each with the type the spec gives it there.
+ */
 interface ActiveJob {
-    jobId: string;
-    type: string;
-    resourceId: string;
-    userId: string;
-    params: Record<string, unknown>;
+    jobId: JobId;
+    type: JobType$1;
+    resourceId: ResourceId;
+    params: ClaimedJob['params'];
     /**
      * Entity-type units earlier failed attempts fully emitted
      * (ABANDONED-INFERENCE P2 checkpointed resume) — carried on the claimed
@@ -671,9 +677,9 @@ interface JobClaimAdapterOptions {
  * A claim the dispatcher refused for a reason other than "nothing pending".
  *
  * `code` is the `BusRequestError` code the reply was promoted to, or `null`
- * when the failure was local — an unusable record, a thrown non-bus error —
- * never a manufactured bus code. `bus.none-pending` never appears here: it is
- * the quiet park, and emitting it would make an empty queue look like a fault.
+ * when the failure was local — a thrown non-bus error — never a manufactured
+ * bus code. `bus.none-pending` never appears here: it is the quiet park, and
+ * emitting it would make an empty queue look like a fault.
  */
 interface ClaimRefusal {
     code: BusRequestErrorCode | null;
@@ -819,4 +825,4 @@ declare const WORKER_CONSUMED_BROADCASTS: readonly ["job:queued", "job:cancel-re
 declare const WORKER_CHANNELS: readonly (keyof EventMap)[];
 
 export { AnnotationDetection, STALL_THRESHOLD_MS, WORKER_CHANNELS, WORKER_CONSUMED_BROADCASTS, createJobClaimAdapter, generateResourceFromTopic, isCancelledJob, isCompleteJob, isFailedJob, isPendingJob, isRunningJob, processAssessmentJob, processCommentJob, processGenerationJob, processHighlightJob, processReferenceJob, processTagJob };
-export type { ActiveJob, AnyJob, AssessmentDetectionJob, AssessmentDetectionParams, CancelledJob, ClaimRefusal, CommentDetectionJob, CommentDetectionParams, CompleteJob, DetectionJob, DetectionParams, FailedJob, GenerationJob, GenerationResult, HighlightDetectionJob, HighlightDetectionParams, JobClaimAdapter, JobClaimAdapterOptions, JobMetadata, JobStatus, JobType, OnProgress, PendingJob, ProcessorResult, RunningAnyJob, RunningJob, StoredProgress, TagDetectionJob, TagDetectionParams, WorkerVitals };
+export type { ActiveJob, AnyJob, AssessmentDetectionJob, AssessmentDetectionParams, CancelledJob, ClaimRefusal, CommentDetectionJob, CommentDetectionParams, CompleteJob, DetectionJob, DetectionParams, FailedJob, GenerationJob, HighlightDetectionJob, HighlightDetectionParams, JobClaimAdapter, JobClaimAdapterOptions, JobMetadata, JobStatus, JobType, OnProgress, PendingJob, ProcessorResult, RunningAnyJob, RunningJob, StoredProgress, TagDetectionJob, TagDetectionParams, WorkerVitals };

package/dist/index.js

@@ -1,4 +1,4 @@
-import { JOB_TYPES, replyChannelsFor, deriveViews, GENERATABLE_MEDIA_TYPES, estimateTokens, isObject, isString, reconcileSelector, getLocaleEnglishName, busRequest, BusRequestError, isArray, isNumber, cutChunk, chunkText } from '@semiont/core';
+import { replyChannelsFor, deriveViews, GENERATABLE_MEDIA_TYPES, estimateTokens, isObject, isString, reconcileSelector, getLocaleEnglishName, isResourceId, busRequest, BusRequestError, cutChunk, chunkText } from '@semiont/core';
 import { withSpan, recordAnchorOutcome, recordDetectionCall } from '@semiont/observability';
 import { StructuredReadError } from '@semiont/inference';
 import { execFileSync } from 'child_process';
@@ -3764,9 +3764,9 @@ var require_mapOneOrManyArgs = __commonJS({
     Object.defineProperty(exports, "__esModule", { value: true });
     exports.mapOneOrManyArgs = void 0;
     var map_1 = require_map();
-    var isArray2 = Array.isArray;
+    var isArray = Array.isArray;
     function callOrApply(fn, args) {
-      return isArray2(args) ? fn.apply(void 0, __spreadArray([], __read(args))) : fn(args);
+      return isArray(args) ? fn.apply(void 0, __spreadArray([], __read(args))) : fn(args);
     }
     function mapOneOrManyArgs(fn) {
       return map_1.map(function(args) {
@@ -3911,14 +3911,14 @@ var require_argsArgArrayOrObject = __commonJS({
   "../../node_modules/rxjs/dist/cjs/internal/util/argsArgArrayOrObject.js"(exports) {
     Object.defineProperty(exports, "__esModule", { value: true });
     exports.argsArgArrayOrObject = void 0;
-    var isArray2 = Array.isArray;
+    var isArray = Array.isArray;
     var getPrototypeOf = Object.getPrototypeOf;
     var objectProto = Object.prototype;
     var getKeys = Object.keys;
     function argsArgArrayOrObject(args) {
       if (args.length === 1) {
         var first_1 = args[0];
-        if (isArray2(first_1)) {
+        if (isArray(first_1)) {
           return { args: first_1, keys: null };
         }
         if (isPOJO(first_1)) {
@@ -4666,9 +4666,9 @@ var require_argsOrArgArray = __commonJS({
   "../../node_modules/rxjs/dist/cjs/internal/util/argsOrArgArray.js"(exports) {
     Object.defineProperty(exports, "__esModule", { value: true });
     exports.argsOrArgArray = void 0;
-    var isArray2 = Array.isArray;
+    var isArray = Array.isArray;
     function argsOrArgArray(args) {
-      return args.length === 1 && isArray2(args[0]) ? args[0] : args;
+      return args.length === 1 && isArray(args[0]) ? args[0] : args;
     }
     exports.argsOrArgArray = argsOrArgArray;
   }
@@ -9203,7 +9203,8 @@ var require_cjs = __commonJS({
     } });
   }
 });
-new Set(JOB_TYPES);
+
+// src/types.ts
 function isPendingJob(job) {
   return job.status === "pending";
 }
@@ -10562,8 +10563,6 @@ function compileTypst(source) {
     rmSync(dir, { recursive: true, force: true });
   }
 }
-
-// src/workers/generation/citation-resolver.ts
 var CITATION_TOKEN = /\[\[([^\s[\]/]+)(?:\/([^\s[\]/]+))?\]\]/g;
 function collectContextResourceIds(context) {
   const ids = /* @__PURE__ */ new Set();
@@ -10589,7 +10588,7 @@ function resolveCitationTokens(content, validResourceIds, logger) {
     const annotationId = match[2];
     clean += content.slice(last, match.index).replace(/[ \t]+$/, "");
     last = match.index + token.length;
-    if (!validResourceIds.has(citedResourceId)) {
+    if (!isResourceId(citedResourceId) || !validResourceIds.has(citedResourceId)) {
       logger.warn("Citation token references an id absent from the provided context \u2014 dropped", {
         resourceId: citedResourceId
       });
@@ -11091,12 +11090,7 @@ async function processGenerationJob(inferenceClient, params, onProgress, logger)
       title,
       format: outputMediaType,
       citations: citations2,
-      result: {
-        kind: "generation",
-        resourceId: "",
-        resourceName: title,
-        truncated: generated2.truncated
-      }
+      truncated: generated2.truncated
     };
   }
   onProgress(5, { code: "generating-resource" });
@@ -11132,30 +11126,12 @@ async function processGenerationJob(inferenceClient, params, onProgress, logger)
     title,
     format: outputMediaType,
     citations,
-    result: {
-      kind: "generation",
-      resourceId: "",
-      resourceName: title,
-      truncated: generated.truncated
-    }
+    truncated: generated.truncated
   };
 }
 
 // src/job-claim-adapter.ts
 var import_rxjs = __toESM(require_cjs());
-function readUnitCursors(raw, completedUnits) {
-  if (!isObject(raw)) return {};
-  const done = new Set(completedUnits);
-  const cursors = {};
-  for (const [unit, value] of Object.entries(raw)) {
-    if (done.has(unit) || !isObject(value)) continue;
-    const { next, size, found, emitted } = value;
-    if (!isNumber(next) || !isNumber(size) || next < 0 || size < 1) continue;
-    if (!isNumber(found) || !isNumber(emitted) || found < 0 || emitted < 0) continue;
-    cursors[unit] = { next, size, found, emitted };
-  }
-  return cursors;
-}
 function createJobClaimAdapter(options) {
   const { bus, jobTypes } = options;
   const activeJob$ = new import_rxjs.BehaviorSubject(null);
@@ -11174,9 +11150,9 @@ function createJobClaimAdapter(options) {
   let activeSince = null;
   const iso = (t) => t === null ? null : new Date(t).toISOString();
   const claimNext = async () => {
-    let record;
+    let claimed;
     try {
-      record = await busRequest(bus, "job:claim", { types: jobTypes }, 1e4);
+      claimed = await busRequest(bus, "job:claim", { types: jobTypes }, 1e4);
     } catch (error) {
       if (error instanceof BusRequestError) {
         if (error.code === "bus.none-pending") return { declined: true };
@@ -11184,25 +11160,19 @@ function createJobClaimAdapter(options) {
       }
       return { refused: { code: null, message: error instanceof Error ? error.message : String(error) } };
     }
-    if (!isString(record.metadata?.id) || !isString(record.metadata?.type)) {
-      return { refused: { code: null, message: "claimed record carries no job id or type" } };
-    }
-    const completedUnits = isArray(record.metadata?.completedUnits) ? record.metadata.completedUnits.filter(isString) : [];
-    const unitCursors = readUnitCursors(record.metadata?.unitCursors, completedUnits);
-    const params = record.params ?? {};
+    const { metadata, params } = claimed;
     return {
       job: {
-        jobId: record.metadata.id,
-        type: record.metadata.type,
-        resourceId: isString(params.resourceId) ? params.resourceId : "",
-        userId: record.metadata?.userId ?? "",
+        jobId: metadata.id,
+        type: metadata.type,
+        resourceId: params.resourceId,
         params,
-        completedUnits,
-        unitCursors,
-        // Absent or malformed metadata reads as "no budget left" — a worker
-        // that cannot see the budget must not claim a retry is coming.
-        retryCount: isNumber(record.metadata?.retryCount) ? record.metadata.retryCount : 0,
-        maxRetries: isNumber(record.metadata?.maxRetries) ? record.metadata.maxRetries : 0
+        // Both appear on the record once an attempt has checkpointed, and a
+        // finished unit has no cursor: the dispatcher drops it at the merge.
+        completedUnits: metadata.completedUnits ?? [],
+        unitCursors: metadata.unitCursors ?? {},
+        retryCount: metadata.retryCount,
+        maxRetries: metadata.maxRetries
       }
     };
   };