@semiont/jobs 0.5.2 → 0.5.3

package/dist/index.d.ts

@@ -1,6 +1,6 @@
 import { Readable } from 'stream';
 import * as _semiont_core from '@semiont/core';
-import { ResourceId, JobId, UserId, EntityType, AnnotationId, Annotation, GatheredContext, Logger, EventBus, components } from '@semiont/core';
+import { ResourceId, JobId, UserId, EntityType, AnnotationId, Annotation, GatheredContext, TagSchema, Logger, EventBus, components } from '@semiont/core';
 import { SemiontProject } from '@semiont/core/node';
 import { InferenceClient } from '@semiont/inference';
 
@@ -132,11 +132,16 @@ interface CommentDetectionParams {
     sourceLanguage?: string;
 }
 /**
- * Tag detection job parameters
+ * Tag detection job parameters.
+ *
+ * Carries the *full* `TagSchema` (not just an id). The dispatcher resolves
+ * the caller-supplied `schemaId` against the per-KB tag-schema projection
+ * at job-creation time and embeds the resolved schema here, keeping the
+ * worker independent of the registry.
  */
 interface TagDetectionParams {
     resourceId: ResourceId;
-    schemaId: string;
+    schema: TagSchema;
     categories: string[];
     /** Annotation body locale — see locale conventions above. */
     language?: string;
@@ -591,12 +596,16 @@ declare class AnnotationDetection {
     /**
      * 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 detectTags(content: string, client: InferenceClient, schemaId: string, category: string, sourceLanguage?: string): Promise<TagMatch[]>;
+    static detectTags(content: string, client: InferenceClient, schema: TagSchema, category: string, sourceLanguage?: string): Promise<TagMatch[]>;
 }
 
 /**

package/dist/index.js

@@ -947,159 +947,6 @@ 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 {
   /**
@@ -1157,19 +1004,19 @@ var AnnotationDetection = class {
   /**
    * 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, sourceLanguage) {
-    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,
@@ -1718,7 +1565,7 @@ async function processTagJob(content, inferenceClient, params, userId, generator
     const categoryTags = await AnnotationDetection.detectTags(
       content,
       inferenceClient,
-      params.schemaId,
+      params.schema,
       category,
       params.sourceLanguage
     );
@@ -1733,7 +1580,7 @@ async function processTagJob(content, inferenceClient, params, userId, generator
     byCategory[category] = (byCategory[category] ?? 0) + 1;
     return buildTextAnnotation(params.resourceId, userId, generator, "tagging", t, [
       { type: "TextualBody", value: category, purpose: "tagging", format: "text/plain", language: bodyLanguage },
-      { type: "TextualBody", value: params.schemaId, purpose: "classifying", format: "text/plain" }
+      { type: "TextualBody", value: params.schema.id, purpose: "classifying", format: "text/plain" }
     ]);
   });
   onProgress(100, `Complete! Created ${annotations.length} tags`, "creating");