add types and constants ported from QMD pipeline

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/constants.ts

@@ -0,0 +1,43 @@
+// Chunking
+export const CHUNK_SIZE_TOKENS = 900;
+export const CHUNK_OVERLAP_TOKENS = 135; // 15%
+export const CHUNK_SIZE_CHARS = 3600; // ~4 chars/token
+export const CHUNK_OVERLAP_CHARS = 540;
+
+// RRF
+export const RRF_K = 60;
+export const RRF_PRIMARY_WEIGHT = 2.0;
+export const RRF_SECONDARY_WEIGHT = 1.0;
+export const RRF_RANK1_BONUS = 0.05;
+export const RRF_RANK2_BONUS = 0.02;
+
+// Score Blending
+export const BLEND_TOP3_RRF = 0.75;
+export const BLEND_MID_RRF = 0.6;
+export const BLEND_TAIL_RRF = 0.4;
+
+// BM25
+export const BM25_K1 = 1.2;
+export const BM25_B = 0.75;
+
+// Search
+export const RERANK_CANDIDATE_LIMIT = 40;
+export const RERANK_CONTEXT_TOKENS = 2048;
+
+// Embedding templates (embeddinggemma format)
+export const EMBED_QUERY_TEMPLATE = (query: string) =>
+  `task: search result | query: ${query}`;
+export const EMBED_DOC_TEMPLATE = (title: string, body: string) =>
+  `title: ${title} | text: ${body}`;
+
+// Model IDs for Transformers.js
+export const MODEL_EMBEDDING = "onnx-community/embeddinggemma-300m-ONNX";
+export const MODEL_RERANKER = "onnx-community/Qwen3-Reranker-0.6B-ONNX";
+export const MODEL_EXPANSION = "shreyask/qmd-query-expansion-1.7B-ONNX"; // placeholder
+
+// Example queries
+export const EXAMPLE_QUERIES = [
+  "API versioning best practices",
+  "distributed consensus algorithms",
+  "gradient descent optimization",
+];

src/types.ts

@@ -0,0 +1,100 @@
+// A document loaded into the demo
+export interface Document {
+  id: string; // filename-based
+  title: string; // extracted from H1 or filename
+  body: string; // full text content
+  filepath: string; // e.g. "api-design-principles.md"
+}
+
+// A chunk of a document
+export interface Chunk {
+  docId: string;
+  chunkIndex: number;
+  text: string;
+  startChar: number; // offset in original doc
+  title: string; // parent doc title
+}
+
+// A chunk with its embedding vector
+export interface EmbeddedChunk extends Chunk {
+  embedding: Float32Array;
+}
+
+// Search result from BM25 or vector search
+export interface ScoredChunk {
+  chunk: Chunk;
+  score: number;
+  source: "bm25" | "vector";
+}
+
+// Result after RRF fusion
+export interface RRFResult {
+  docId: string;
+  filepath: string;
+  title: string;
+  bestChunk: string;
+  score: number;
+  contributions: RRFContribution[];
+}
+
+export interface RRFContribution {
+  source: "bm25" | "vector";
+  queryType: "original" | "lex" | "vec" | "hyde";
+  query: string;
+  rank: number;
+  weight: number;
+  rrfContribution: number;
+}
+
+// Result after reranking
+export interface RerankedResult extends RRFResult {
+  rerankScore: number;
+  blendedScore: number;
+}
+
+// Final result
+export interface FinalResult {
+  filepath: string;
+  title: string;
+  bestChunk: string;
+  score: number; // blended score
+  docId: string;
+}
+
+// Query expansion output
+export interface ExpandedQuery {
+  hyde: string; // hypothetical document snippet
+  vec: string[]; // dense retrieval sentences
+  lex: string; // BM25 keywords
+}
+
+// Pipeline events for React UI
+export type PipelineStage = "expansion" | "search" | "rrf" | "rerank" | "blend";
+export type PipelineStatus = "idle" | "running" | "done" | "error";
+
+export type PipelineEvent =
+  | { stage: "expansion"; status: "running" }
+  | { stage: "expansion"; status: "done"; data: ExpandedQuery }
+  | { stage: "expansion"; status: "error"; error: string }
+  | { stage: "search"; status: "running" }
+  | {
+      stage: "search";
+      status: "done";
+      data: { bm25Hits: ScoredChunk[]; vectorHits: ScoredChunk[] };
+    }
+  | { stage: "rrf"; status: "done"; data: { merged: RRFResult[] } }
+  | { stage: "rerank"; status: "running" }
+  | {
+      stage: "rerank";
+      status: "done";
+      data: { before: RRFResult[]; after: RerankedResult[] };
+    }
+  | { stage: "blend"; status: "done"; data: { finalResults: FinalResult[] } };
+
+// Model loading state
+export interface ModelState {
+  name: string;
+  status: "pending" | "downloading" | "loading" | "ready" | "error";
+  progress: number; // 0-1
+  error?: string;
+}