@semiont/jobs 0.6.8 → 0.6.10

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";
 }
@@ -9759,7 +9760,8 @@ Example format:
   /**
    * Build a prompt for detecting structural tags
    *
-   * @param content - The full text content to analyze (NOT truncated for structural analysis)
+   * @param content - The text content to analyze — a chunk sized by the
+   *   caller from derived provider limits; NEVER re-truncated here
    * @param category - The specific category to detect
    * @param schemaName - Human-readable schema name
    * @param schemaDescription - Schema description
@@ -10187,7 +10189,7 @@ ${piece}
   }
   if (items.length * YIELD_COLLAPSE_BAND < counted) {
     throw new YieldCollapseError(
-      `Extraction found ${items.length} entities where a count call reports ~${counted} mentions (band \xD7${YIELD_COLLAPSE_BAND}) on a ${piece.length}-char chunk \u2014 silent yield collapse (F7): deterministic \u2014 a same-size retry returns the identical under-report.`,
+      `Extraction found ${items.length} entities where a count call reports ~${counted} mentions (band \xD7${YIELD_COLLAPSE_BAND}) on a ${piece.length}-char chunk \u2014 silent yield collapse: deterministic \u2014 a same-size retry returns the identical under-report.`,
       [...items],
       { found: items.length, counted, pieceChars: piece.length }
     );
@@ -10562,8 +10564,6 @@ 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 +10589,7 @@ 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
       });
@@ -10738,9 +10738,8 @@ async function processCommentJob(content, inferenceClient, params, buildAnnotati
       found += comments.length;
       const fresh = dedupe(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.
+          // Format and language go on the body TextualBody: optional in the
+          // schema, but consumers that do language-aware rendering rely on them.
           buildAnnotation("commenting", c, [
             { type: "TextualBody", value: c.comment, purpose: "commenting", format: "text/plain", language: bodyLanguage }
           ])
@@ -10781,8 +10780,7 @@ async function processAssessmentJob(content, inferenceClient, params, buildAnnot
       const fresh = dedupe(assessments.map(
         (a) => (
           // Single-object body with purpose aligned to motivation, matching the
-          // pre-#651 AssessmentAnnotationWorker's shape and the majority of
-          // persisted assessments. Do not switch to an array or to
+          // majority of persisted assessments. Do not switch to an array or to
           // purpose='describing' — that loses the "this is an assessment, not
           // a description" signal and breaks existing readers that access
           // `body.value` directly on the object.
@@ -10850,11 +10848,11 @@ async function processReferenceJob(content, inferenceClient, params, buildAnnota
       params.includeDescriptiveReferences ?? false,
       logger,
       params.sourceLanguage,
-      // Liveness heartbeat (DETECTION-HEARTBEAT): fires at chunk boundaries and
-      // every ~15 s while a call is in flight, so a long single-chunk call is
-      // not silent. It repeats the current position rather than inventing an
-      // advance — the stall watchdog, janitor and client timeout need a signal,
-      // not a monotone.
+      // Liveness heartbeat: fires at chunk boundaries and every ~15 s while a
+      // call is in flight, so a long single-chunk call is not silent. It
+      // repeats the current position rather than inventing an advance — the
+      // stall watchdog, janitor and client timeout need a signal, not a
+      // monotone.
       () => emitTypeProgress(entityTypeName),
       (verdict) => {
         underReported = {
@@ -11011,7 +11009,7 @@ async function processGenerationJob(inferenceClient, params, onProgress, logger)
       `Unsupported outputMediaType for generation: ${outputMediaType}. Generation produces ${GENERATABLE_MEDIA_TYPES.join(" or ")}.`
     );
   }
-  const title = params.title ?? "Untitled";
+  const title = params.title;
   const entityTypes = (params.entityTypes ?? []).map(String);
   if (outputMediaType === "application/pdf") {
     onProgress(5, { code: "generating-resource" });
@@ -11091,12 +11089,7 @@ 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 +11125,12 @@ 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 +11149,9 @@ 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 +11159,19 @@ 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
       }
     };
   };
@@ -11319,26 +11288,25 @@ var WORKER_AWAITED_OPERATIONS = [
   "job:claim",
   "browse:resource-requested",
   // Canonical geometry for a geometry-bearing detection: the consult behind
-  // `ConsultAnchoredText` (SMELTER-OWNS-OCR P2). Its omission broke every
-  // PDF detection job at the transport probe
-  // (.plans/WORKER-ANCHORED-TEXT-CHANNEL.md); the census below now fails the
+  // `ConsultAnchoredText`, answered by the Smelter. Without it every PDF
+  // detection job fails at the transport probe; the census below fails the
   // BUILD when this list and the declared awaits drift.
   "browse:anchored-text-requested",
-  // Durability acknowledgement for a unit's annotations (JOB-RESTART-SAFETY
-  // P6). The worker AWAITS this one — a unit may not advance until its
-  // annotations are in the event log — so its replies must be in the narrow
-  // channel set or every commit fails fast with `bus.unsubscribed`.
+  // Durability acknowledgement for a unit's annotations. The worker AWAITS
+  // this one — a unit may not advance until its annotations are in the event
+  // log — so its replies must be in the narrow channel set or every commit
+  // fails fast with `bus.unsubscribed`.
   "mark:commit",
-  // The durability probe for a commit whose acknowledgement never routed
-  // (COMMIT-ACK-FALSE-FAILURE F1). SINGULAR by design: the annotation LIST
-  // channel is the multi-MB fan-out this narrowing exists to keep out, and a
-  // rare error path is no reason to let it back in.
+  // The durability probe for a commit whose acknowledgement never routed.
+  // SINGULAR by design: the annotation LIST channel is the multi-MB fan-out
+  // this narrowing exists to keep out, and a rare error path is no reason to
+  // let it in.
   "browse:annotation-requested"
 ];
 var WORKER_CONSUMED_BROADCASTS = [
   // The queue announcement the claim adapter races for.
   "job:queued",
-  // Cooperative cancellation of the ACTIVE job (JOB-RESTART-SAFETY P4).
+  // Cooperative cancellation of the ACTIVE job: it stops at a unit boundary.
   "job:cancel-requested"
 ];
 var WORKER_CHANNELS = [