testing with IP

Author: wjohnsto

services/video-search/src/config.ts

@@ -10,25 +10,29 @@ let {
     LOG_STREAM,
     NODE_ENV,
     YOUTUBE_VIDEOS,
+    YOUTUBE_TRANSCRIPT_PREFIX,
+    YOUTUBE_VIDEO_INFO_PREFIX,
     REDIS_URL,
     SEARCHAPI_API_KEY,
     GOOGLE_VIDEO_INDEX_NAME,
     GOOGLE_VIDEO_PREFIX,
+    GOOGLE_ANSWER_INDEX_NAME,
+    GOOGLE_ANSWER_PREFIX,
     GOOGLE_API_KEY,
     GOOGLE_EMBEDDING_MODEL,
     GOOGLE_SUMMARY_MODEL,
     GOOGLE_VECTOR_SET,
     GOOGLE_SUMMARY_PREFIX,
-    GOOGLE_ANSWER_PREFIX,
     OPENAI_VIDEO_INDEX_NAME,
     OPENAI_VIDEO_PREFIX,
+    OPENAI_ANSWER_INDEX_NAME,
+    OPENAI_ANSWER_PREFIX,
     OPENAI_API_KEY,
     OPENAI_ORGANIZATION,
     OPENAI_EMBEDDING_MODEL,
     OPENAI_SUMMARY_MODEL,
     OPENAI_VECTOR_SET,
     OPENAI_SUMMARY_PREFIX,
-    OPENAI_ANSWER_PREFIX,
     USE,
 } = process.env;
 
@@ -41,6 +45,9 @@ const DEFAULT_VIDEO_PREFIX = 'video:';
 const DEFAULT_VECTOR_SET = 'video-vectors';
 const DEFAULT_SUMMARY_PREFIX = 'video-summary:';
 const DEFAULT_ANSWER_PREFIX = 'video-answer:';
+const DEFAULT_ANSWER_INDEX_NAME = 'idx-answers';
+const DEFAULT_YOUTUBE_TRANSCRIPT_PREFIX = 'transcripts:';
+const DEFAULT_YOUTUBE_VIDEO_INFO_PREFIX = 'yt-videos:';
 
 export default {
     app: {
@@ -68,8 +75,10 @@ export default {
             YOUTUBE_VIDEOS ??
             'AJhTduDOVCs,c9Rr--1r6pk,FQzlq91g7mg,I-ohlZXXaxs,KUfufrwpBkM,LaiQFZ5bXaM,SzcpwtLRgyk,Z8qcpXyMAiA'
         ).split(','),
-        TRANSCRIPT_PREFIX: 'transcripts:',
-        VIDEO_INFO_PREFIX: 'yt-videos:',
+        TRANSCRIPT_PREFIX:
+            YOUTUBE_TRANSCRIPT_PREFIX ?? DEFAULT_YOUTUBE_TRANSCRIPT_PREFIX,
+        VIDEO_INFO_PREFIX:
+            YOUTUBE_VIDEO_INFO_PREFIX ?? DEFAULT_YOUTUBE_VIDEO_INFO_PREFIX,
     },
     redis: {
         REDIS_URL: REDIS_URL ?? 'redis://localhost:6379',
@@ -87,6 +96,8 @@ export default {
         VECTOR_SET: GOOGLE_VECTOR_SET ?? `google-${DEFAULT_VECTOR_SET}`,
         SUMMARY_PREFIX:
             GOOGLE_SUMMARY_PREFIX ?? `google-${DEFAULT_SUMMARY_PREFIX}`,
+        ANSWER_INDEX_NAME:
+            GOOGLE_ANSWER_INDEX_NAME ?? `google-${DEFAULT_ANSWER_INDEX_NAME}`,
         ANSWER_PREFIX:
             GOOGLE_ANSWER_PREFIX ?? `google-${DEFAULT_ANSWER_PREFIX}`,
     },
@@ -101,6 +112,8 @@ export default {
         VECTOR_SET: OPENAI_VECTOR_SET ?? `openai-${DEFAULT_VECTOR_SET}`,
         SUMMARY_PREFIX:
             OPENAI_SUMMARY_PREFIX ?? `openai-${DEFAULT_SUMMARY_PREFIX}`,
+        ANSWER_INDEX_NAME:
+            OPENAI_ANSWER_INDEX_NAME ?? `openai-${DEFAULT_ANSWER_INDEX_NAME}`,
         ANSWER_PREFIX:
             OPENAI_ANSWER_PREFIX ?? `openai-${DEFAULT_ANSWER_PREFIX}`,
     },

services/video-search/src/google/config.ts

@@ -6,6 +6,7 @@ import {
     GoogleGenerativeAIEmbeddings,
 } from '@langchain/google-genai';
 import { TaskType } from '@google/generative-ai';
+import { VectorAlgorithms } from 'redis';
 
 export const llm = new ChatGoogleGenerativeAI({
     apiKey: config.google.API_KEY,
@@ -25,4 +26,18 @@ export const vectorStore = new RedisVectorStore(getEmbeddings(), {
     redisClient: client,
     indexName: config.google.VIDEO_INDEX_NAME,
     keyPrefix: config.google.VIDEO_PREFIX,
+    indexOptions: {
+        ALGORITHM: VectorAlgorithms.HNSW,
+        DISTANCE_METRIC: 'COSINE',
+    },
+});
+
+export const answerVectorStore = new RedisVectorStore(getEmbeddings(), {
+    redisClient: client,
+    indexName: config.google.ANSWER_INDEX_NAME,
+    keyPrefix: config.google.ANSWER_PREFIX,
+    indexOptions: {
+        ALGORITHM: VectorAlgorithms.HNSW,
+        DISTANCE_METRIC: 'L2',
+    },
 });

services/video-search/src/google/search.ts

@@ -1,11 +1,11 @@
+import { Document } from 'langchain/document';
 import * as summarize from './summarize.js';
 import { type VideoDocument } from '../transcripts/index.js';
-import { llm, vectorStore } from './config.js';
+import { llm, vectorStore, answerVectorStore } from './config.js';
 import log from '../log.js';
 import config from '../config.js';
 import { ANSWER_PROMPT } from '../templates/answers.js';
 import { StringOutputParser } from 'langchain/schema/output_parser';
-import { cacheAside } from '../db.js';
 
 // eslint-disable-next-line @typescript-eslint/no-unsafe-argument
 const questionAnswerChain = ANSWER_PROMPT.pipe(llm as any).pipe(
@@ -29,29 +29,45 @@ async function getVideos(question: string) {
     >);
 }
 
-const answerCache = cacheAside(config.google.ANSWER_PREFIX);
-
 async function getAnswer(question: string, videos: VideoDocument[]) {
     log.debug(`Getting answer for question: ${question}`, {
         location: 'google.search.getAnswer',
     });
 
-    const cachedAnswer = await answerCache.get(question);
+    const haveAnswers = await answerVectorStore.checkIndexExists();
 
-    if (typeof cachedAnswer === 'string') {
-        log.debug(`Found cached answer for question: ${question}`, {
+    if (haveAnswers) {
+        log.debug(`Searching for closest answer to question: ${question}`, {
             location: 'google.search.getAnswer',
+            question,
         });
 
-        return cachedAnswer;
+        const [result] = await answerVectorStore.similaritySearchWithScore(
+            question,
+            1,
+        );
+
+        log.debug(`Found closest answer with score: ${String(result[1])}`, {
+            location: 'google.search.getAnswer',
+            answer: result[0],
+            score: result[1],
+        });
     }
 
     const answer = (await questionAnswerChain.invoke({
         question,
         data: JSON.stringify(videos),
     })) as string;
 
-    await answerCache.set(question, answer);
+    await answerVectorStore.addDocuments([
+        new Document({
+            metadata: {
+                videos,
+                answer,
+            },
+            pageContent: question,
+        }),
+    ]);
 
     return answer;
 }

services/video-search/src/openai/config.ts

@@ -27,3 +27,9 @@ export const vectorStore = new RedisVectorStore(getEmbeddings(), {
     indexName: config.openai.VIDEO_INDEX_NAME,
     keyPrefix: config.openai.VIDEO_PREFIX,
 });
+
+export const answerVectorStore = new RedisVectorStore(getEmbeddings(), {
+    redisClient: client,
+    indexName: config.openai.ANSWER_INDEX_NAME,
+    keyPrefix: config.openai.ANSWER_PREFIX,
+});

services/video-search/src/templates/questions.ts

@@ -2,7 +2,7 @@ import { PromptTemplate } from 'langchain/prompts';
 
 const questionTemplate = `
 You are an expert in summarizing questions.
-Your goal is to reduce a question down to its simplest form while still retaining the semantic meaning.
+Your goal is to reduce a question down to its simplest form while still retaining the semantic meaning. Try to be as deterministic as possible.
 Below you find the question:
 --------
 {question}