sk/Perplexica
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge remote-tracking branch 'upstream/master' into refac/improve-searxng-queries

Browse source
This commit is contained in:
Dave 2025-08-17 15:30:48 +02:00
commit 5885cf6d98
44 changed files with 999 additions and 628 deletions
Download patch file Download diff file Expand all files Collapse all files

51
src/lib/chains/imageSearchAgent.ts
View file

@ -3,32 +3,18 @@ import {
RunnableMap,
RunnableLambda,
} from '@langchain/core/runnables';
import { PromptTemplate } from '@langchain/core/prompts';
import { ChatPromptTemplate } from '@langchain/core/prompts';
import formatChatHistoryAsString from '../utils/formatHistory';
import { BaseMessage } from '@langchain/core/messages';
import { StringOutputParser } from '@langchain/core/output_parsers';
import { searchSearxng } from '../searxng';
import type { BaseChatModel } from '@langchain/core/language_models/chat_models';
import LineOutputParser from '../outputParsers/lineOutputParser';
const imageSearchChainPrompt = `
You will be given a conversation below and a follow up question. You need to rephrase the follow-up question so it is a standalone question that can be used by the LLM to search the web for images.
You need to make sure the rephrased question agrees with the conversation and is relevant to the conversation.
Example:
1. Follow up question: What is a cat?
Rephrased: A cat
2. Follow up question: What is a car? How does it works?
Rephrased: Car working
3. Follow up question: How does an AC work?
Rephrased: AC working
Conversation:
{chat_history}
Follow up question: {query}
Rephrased question:
Output only the rephrased query wrapped in an XML <query> element. Do not include any explanation or additional text.
`;
type ImageSearchChainInput = {
@ -54,12 +40,39 @@ const createImageSearchChain = (llm: BaseChatModel) => {
return input.query;
},
}),
PromptTemplate.fromTemplate(imageSearchChainPrompt),
ChatPromptTemplate.fromMessages([
['system', imageSearchChainPrompt],
[
'user',
'<conversation>\n</conversation>\n<follow_up>\nWhat is a cat?\n</follow_up>',
],
['assistant', '<query>A cat</query>'],
[
'user',
'<conversation>\n</conversation>\n<follow_up>\nWhat is a car? How does it work?\n</follow_up>',
],
['assistant', '<query>Car working</query>'],
[
'user',
'<conversation>\n</conversation>\n<follow_up>\nHow does an AC work?\n</follow_up>',
],
['assistant', '<query>AC working</query>'],
[
'user',
'<conversation>{chat_history}</conversation>\n<follow_up>\n{query}\n</follow_up>',
],
]),
llm,
strParser,
RunnableLambda.from(async (input: string) => {
input = input.replace(/<think>.*?<\/think>/g, '');
const queryParser = new LineOutputParser({
key: 'query',
});
return await queryParser.parse(input);
}),
RunnableLambda.from(async (input: string) => {
const res = await searchSearxng(input, {
engines: ['bing images', 'google images'],
});

63
src/lib/chains/videoSearchAgent.ts
View file

@ -3,33 +3,19 @@ import {
RunnableMap,
RunnableLambda,
} from '@langchain/core/runnables';
import { PromptTemplate } from '@langchain/core/prompts';
import { ChatPromptTemplate } from '@langchain/core/prompts';
import formatChatHistoryAsString from '../utils/formatHistory';
import { BaseMessage } from '@langchain/core/messages';
import { StringOutputParser } from '@langchain/core/output_parsers';
import { searchSearxng } from '../searxng';
import type { BaseChatModel } from '@langchain/core/language_models/chat_models';
import LineOutputParser from '../outputParsers/lineOutputParser';
const VideoSearchChainPrompt = `
You will be given a conversation below and a follow up question. You need to rephrase the follow-up question so it is a standalone question that can be used by the LLM to search Youtube for videos.
You need to make sure the rephrased question agrees with the conversation and is relevant to the conversation.
Example:
1. Follow up question: How does a car work?
Rephrased: How does a car work?
2. Follow up question: What is the theory of relativity?
Rephrased: What is theory of relativity
3. Follow up question: How does an AC work?
Rephrased: How does an AC work
Conversation:
{chat_history}
Follow up question: {query}
Rephrased question:
`;
const videoSearchChainPrompt = `
You will be given a conversation below and a follow up question. You need to rephrase the follow-up question so it is a standalone question that can be used by the LLM to search Youtube for videos.
You need to make sure the rephrased question agrees with the conversation and is relevant to the conversation.
Output only the rephrased query wrapped in an XML <query> element. Do not include any explanation or additional text.
`;
type VideoSearchChainInput = {
chat_history: BaseMessage[];
@ -55,12 +41,37 @@ const createVideoSearchChain = (llm: BaseChatModel) => {
return input.query;
},
}),
PromptTemplate.fromTemplate(VideoSearchChainPrompt),
ChatPromptTemplate.fromMessages([
['system', videoSearchChainPrompt],
[
'user',
'<conversation>\n</conversation>\n<follow_up>\nHow does a car work?\n</follow_up>',
],
['assistant', '<query>How does a car work?</query>'],
[
'user',
'<conversation>\n</conversation>\n<follow_up>\nWhat is the theory of relativity?\n</follow_up>',
],
['assistant', '<query>Theory of relativity</query>'],
[
'user',
'<conversation>\n</conversation>\n<follow_up>\nHow does an AC work?\n</follow_up>',
],
['assistant', '<query>AC working</query>'],
[
'user',
'<conversation>{chat_history}</conversation>\n<follow_up>\n{query}\n</follow_up>',
],
]),
llm,
strParser,
RunnableLambda.from(async (input: string) => {
input = input.replace(/<think>.*?<\/think>/g, '');
const queryParser = new LineOutputParser({
key: 'query',
});
return await queryParser.parse(input);
}),
RunnableLambda.from(async (input: string) => {
const res = await searchSearxng(input, {
engines: ['youtube'],
});
@ -92,8 +103,8 @@ const handleVideoSearch = (
input: VideoSearchChainInput,
llm: BaseChatModel,
) => {
const VideoSearchChain = createVideoSearchChain(llm);
return VideoSearchChain.invoke(input);
const videoSearchChain = createVideoSearchChain(llm);
return videoSearchChain.invoke(input);
};
export default handleVideoSearch;

5
src/lib/config.ts
View file

@ -35,6 +35,9 @@ interface Config {
DEEPSEEK: {
API_KEY: string;
};
AIMLAPI: {
API_KEY: string;
};
LM_STUDIO: {
API_URL: string;
};
@ -85,6 +88,8 @@ export const getOllamaApiEndpoint = () => loadConfig().MODELS.OLLAMA.API_URL;
export const getDeepseekApiKey = () => loadConfig().MODELS.DEEPSEEK.API_KEY;
export const getAimlApiKey = () => loadConfig().MODELS.AIMLAPI.API_KEY;
export const getCustomOpenaiApiKey = () =>
loadConfig().MODELS.CUSTOM_OPENAI.API_KEY;

94
src/lib/providers/aimlapi.ts Normal file
View file

@ -0,0 +1,94 @@
import { ChatOpenAI, OpenAIEmbeddings } from '@langchain/openai';
import { getAimlApiKey } from '../config';
import { ChatModel, EmbeddingModel } from '.';
import { BaseChatModel } from '@langchain/core/language_models/chat_models';
import { Embeddings } from '@langchain/core/embeddings';
import axios from 'axios';
export const PROVIDER_INFO = {
key: 'aimlapi',
displayName: 'AI/ML API',
};
interface AimlApiModel {
id: string;
name?: string;
type?: string;
}
const API_URL = 'https://api.aimlapi.com';
export const loadAimlApiChatModels = async () => {
const apiKey = getAimlApiKey();
if (!apiKey) return {};
try {
const response = await axios.get(`${API_URL}/models`, {
headers: {
'Content-Type': 'application/json',
Authorization: `Bearer ${apiKey}`,
},
});
const chatModels: Record<string, ChatModel> = {};
response.data.data.forEach((model: AimlApiModel) => {
if (model.type === 'chat-completion') {
chatModels[model.id] = {
displayName: model.name || model.id,
model: new ChatOpenAI({
apiKey: apiKey,
modelName: model.id,
temperature: 0.7,
configuration: {
baseURL: API_URL,
},
}) as unknown as BaseChatModel,
};
}
});
return chatModels;
} catch (err) {
console.error(`Error loading AI/ML API models: ${err}`);
return {};
}
};
export const loadAimlApiEmbeddingModels = async () => {
const apiKey = getAimlApiKey();
if (!apiKey) return {};
try {
const response = await axios.get(`${API_URL}/models`, {
headers: {
'Content-Type': 'application/json',
Authorization: `Bearer ${apiKey}`,
},
});
const embeddingModels: Record<string, EmbeddingModel> = {};
response.data.data.forEach((model: AimlApiModel) => {
if (model.type === 'embedding') {
embeddingModels[model.id] = {
displayName: model.name || model.id,
model: new OpenAIEmbeddings({
apiKey: apiKey,
modelName: model.id,
configuration: {
baseURL: API_URL,
},
}) as unknown as Embeddings,
};
}
});
return embeddingModels;
} catch (err) {
console.error(`Error loading AI/ML API embeddings models: ${err}`);
return {};
}
};

12
src/lib/providers/anthropic.ts
View file

@ -9,6 +9,18 @@ export const PROVIDER_INFO = {
import { BaseChatModel } from '@langchain/core/language_models/chat_models';
const anthropicChatModels: Record<string, string>[] = [
{
displayName: 'Claude 4.1 Opus',
key: 'claude-opus-4-1-20250805',
},
{
displayName: 'Claude 4 Opus',
key: 'claude-opus-4-20250514',
},
{
displayName: 'Claude 4 Sonnet',
key: 'claude-sonnet-4-20250514',
},
{
displayName: 'Claude 3.7 Sonnet',
key: 'claude-3-7-sonnet-20250219',

2
src/lib/providers/deepseek.ts
View file

@ -31,7 +31,7 @@ export const loadDeepseekChatModels = async () => {
chatModels[model.key] = {
displayName: model.displayName,
model: new ChatOpenAI({
openAIApiKey: deepseekApiKey,
apiKey: deepseekApiKey,
modelName: model.key,
temperature: 0.7,
configuration: {

16
src/lib/providers/gemini.ts
View file

@ -14,8 +14,16 @@ import { Embeddings } from '@langchain/core/embeddings';
const geminiChatModels: Record<string, string>[] = [
{
displayName: 'Gemini 2.5 Pro Experimental',
key: 'gemini-2.5-pro-exp-03-25',
displayName: 'Gemini 2.5 Flash',
key: 'gemini-2.5-flash',
},
{
displayName: 'Gemini 2.5 Flash-Lite',
key: 'gemini-2.5-flash-lite',
},
{
displayName: 'Gemini 2.5 Pro',
key: 'gemini-2.5-pro',
},
{
displayName: 'Gemini 2.0 Flash',
@ -67,7 +75,7 @@ export const loadGeminiChatModels = async () => {
displayName: model.displayName,
model: new ChatGoogleGenerativeAI({
apiKey: geminiApiKey,
modelName: model.key,
model: model.key,
temperature: 0.7,
}) as unknown as BaseChatModel,
};
@ -100,7 +108,7 @@ export const loadGeminiEmbeddingModels = async () => {
return embeddingModels;
} catch (err) {
console.error(`Error loading OpenAI embeddings models: ${err}`);
console.error(`Error loading Gemini embeddings models: ${err}`);
return {};
}
};

11
src/lib/providers/groq.ts
View file

@ -1,4 +1,4 @@
import { ChatOpenAI } from '@langchain/openai';
import { ChatGroq } from '@langchain/groq';
import { getGroqApiKey } from '../config';
import { ChatModel } from '.';
@ -28,13 +28,10 @@ export const loadGroqChatModels = async () => {
groqChatModels.forEach((model: any) => {
chatModels[model.id] = {
displayName: model.id,
model: new ChatOpenAI({
openAIApiKey: groqApiKey,
modelName: model.id,
model: new ChatGroq({
apiKey: groqApiKey,
model: model.id,
temperature: 0.7,
configuration: {
baseURL: 'https://api.groq.com/openai/v1',
},
}) as unknown as BaseChatModel,
};
});

10
src/lib/providers/index.ts
View file

@ -35,6 +35,11 @@ import {
loadDeepseekChatModels,
PROVIDER_INFO as DeepseekInfo,
} from './deepseek';
import {
loadAimlApiChatModels,
loadAimlApiEmbeddingModels,
PROVIDER_INFO as AimlApiInfo,
} from './aimlapi';
import {
loadLMStudioChatModels,
loadLMStudioEmbeddingsModels,
@ -49,6 +54,7 @@ export const PROVIDER_METADATA = {
gemini: GeminiInfo,
transformers: TransformersInfo,
deepseek: DeepseekInfo,
aimlapi: AimlApiInfo,
lmstudio: LMStudioInfo,
custom_openai: {
key: 'custom_openai',
@ -76,6 +82,7 @@ export const chatModelProviders: Record<
anthropic: loadAnthropicChatModels,
gemini: loadGeminiChatModels,
deepseek: loadDeepseekChatModels,
aimlapi: loadAimlApiChatModels,
lmstudio: loadLMStudioChatModels,
};
@ -87,6 +94,7 @@ export const embeddingModelProviders: Record<
ollama: loadOllamaEmbeddingModels,
gemini: loadGeminiEmbeddingModels,
transformers: loadTransformersEmbeddingsModels,
aimlapi: loadAimlApiEmbeddingModels,
lmstudio: loadLMStudioEmbeddingsModels,
};
@ -110,7 +118,7 @@ export const getAvailableChatModelProviders = async () => {
[customOpenAiModelName]: {
displayName: customOpenAiModelName,
model: new ChatOpenAI({
openAIApiKey: customOpenAiApiKey,
apiKey: customOpenAiApiKey,
modelName: customOpenAiModelName,
temperature: 0.7,
configuration: {

4
src/lib/providers/lmstudio.ts
View file

@ -47,7 +47,7 @@ export const loadLMStudioChatModels = async () => {
chatModels[model.id] = {
displayName: model.name || model.id,
model: new ChatOpenAI({
openAIApiKey: 'lm-studio',
apiKey: 'lm-studio',
configuration: {
baseURL: ensureV1Endpoint(endpoint),
},
@ -83,7 +83,7 @@ export const loadLMStudioEmbeddingsModels = async () => {
embeddingsModels[model.id] = {
displayName: model.name || model.id,
model: new OpenAIEmbeddings({
openAIApiKey: 'lm-studio',
apiKey: 'lm-studio',
configuration: {
baseURL: ensureV1Endpoint(endpoint),
},

4
src/lib/providers/ollama.ts
View file

@ -6,8 +6,8 @@ export const PROVIDER_INFO = {
key: 'ollama',
displayName: 'Ollama',
};
import { ChatOllama } from '@langchain/community/chat_models/ollama';
import { OllamaEmbeddings } from '@langchain/community/embeddings/ollama';
import { ChatOllama } from '@langchain/ollama';
import { OllamaEmbeddings } from '@langchain/ollama';
export const loadOllamaChatModels = async () => {
const ollamaApiEndpoint = getOllamaApiEndpoint();

18
src/lib/providers/openai.ts
View file

@ -42,6 +42,18 @@ const openaiChatModels: Record<string, string>[] = [
displayName: 'GPT 4.1',
key: 'gpt-4.1',
},
{
displayName: 'GPT 5 nano',
key: 'gpt-5-nano',
},
{
displayName: 'GPT 5 mini',
key: 'gpt-5-mini',
},
{
displayName: 'GPT 5',
key: 'gpt-5',
},
];
const openaiEmbeddingModels: Record<string, string>[] = [
@ -67,9 +79,9 @@ export const loadOpenAIChatModels = async () => {
chatModels[model.key] = {
displayName: model.displayName,
model: new ChatOpenAI({
openAIApiKey: openaiApiKey,
apiKey: openaiApiKey,
modelName: model.key,
temperature: 0.7,
temperature: model.key.includes('gpt-5') ? 1 : 0.7,
}) as unknown as BaseChatModel,
};
});
@ -93,7 +105,7 @@ export const loadOpenAIEmbeddingModels = async () => {
embeddingModels[model.key] = {
displayName: model.displayName,
model: new OpenAIEmbeddings({
openAIApiKey: openaiApiKey,
apiKey: openaiApiKey,
modelName: model.key,
}) as unknown as Embeddings,
};

7
src/lib/utils/formatHistory.ts
View file

@ -1,8 +1,11 @@
import { BaseMessage } from '@langchain/core/messages';
import { BaseMessage, isAIMessage } from '@langchain/core/messages';
const formatChatHistoryAsString = (history: BaseMessage[]) => {
return history
.map((message) => `${message._getType()}: ${message.content}`)
.map(
(message) =>
`${isAIMessage(message) ? 'AI' : 'User'}: ${message.content}`,
)
.join('\n');
};