feat: make extract input structured

src/ingestion/llm-extractor.ts

@@ -15,25 +15,10 @@ export class LlmFactExtractor implements FactExtractor {
   constructor(private readonly options: LlmFactExtractorOptions) {}
 
   async extract(input: string): Promise<ExtractedFact> {
-    const prompt = this.buildPrompt(input);
+    return this.options.model.generateText({
-    return this.options.model.generateText(prompt);
+      instruction: DEFAULT_INSTRUCTIONS,
-  }
+      input,
-
+      additionalInstruction: this.options.additionalInstructions,
-  private buildPrompt(input: string): string {
+    });
-    if (this.options.promptBuilder) {
-      return this.options.promptBuilder(input, this.options.instructions);
-    }
-
-    const instructions = this.options.instructions?.trim();
-
-    return [
-      DEFAULT_INSTRUCTIONS,
-      instructions && instructions.length > 0
-        ? `Additional instructions:\n${instructions}`
-        : null,
-      `Input:\n${input.trim()}`,
-    ]
-      .filter((value): value is string => value !== null)
-      .join("\n\n");
   }
 }

src/ingestion/types.ts

@@ -2,14 +2,14 @@ import type {
   AddFactInput,
   EmbeddingProvider,
   TopicLinkInput,
-} from '../types/api';
+} from "../types/api";
 
 export interface ExtractedFact {
   statement?: string;
   summary?: string | null;
   source?: string | null;
   confidence?: number | null;
-  metadata?: AddFactInput['metadata'];
+  metadata?: AddFactInput["metadata"];
   topics: TopicLinkInput[];
 }
 
@@ -17,14 +17,19 @@ export interface FactExtractor {
   extract(input: string): Promise<ExtractedFact>;
 }
 
+export interface LlmTextGenerationModelInput {
+  instruction: string;
+  input: string;
+  additionalInstruction?: string | undefined;
+}
+
 export interface LlmTextGenerationModel {
-  generateText(prompt: string): Promise<ExtractedFact>;
+  generateText(prompt: LlmTextGenerationModelInput): Promise<ExtractedFact>;
 }
 
 export interface LlmFactExtractorOptions {
   model: LlmTextGenerationModel;
-  instructions?: string;
+  additionalInstructions?: string | undefined;
-  promptBuilder?: (input: string, instructions?: string) => string;
 }
 
 export interface IngestStatementOptions {