From cf403302a93f8a058c96b3bf6cee42ca8337cb9d Mon Sep 17 00:00:00 2001 From: Han Xiao Date: Tue, 25 Feb 2025 15:56:16 +0800 Subject: [PATCH] refactor: schemas --- src/agent.ts | 9 ++-- src/app.ts | 2 +- src/tools/evaluator.ts | 10 +++-- src/utils/action-tracker.ts | 18 +++++--- src/utils/i18n.json | 86 +++++++++++++++++++++++++++++++++++++ src/utils/schemas.ts | 2 +- src/utils/text-tools.ts | 36 ++++++++++++++++ 7 files changed, 145 insertions(+), 18 deletions(-) create mode 100644 src/utils/i18n.json diff --git a/src/agent.ts b/src/agent.ts index ca9792ae..51422339 100644 --- a/src/agent.ts +++ b/src/agent.ts @@ -181,7 +181,6 @@ ${allKeywords.join('\n')} actionSections.push(` - For greetings, casual conversation, or general knowledge questions, answer directly without references. -- If the question is clearly within your knowledge cutoff (i.e. Aug. 2024) and requires no up-to-date knowledge to get better answer, then provide a confident answer directly. - For all other questions, provide a verified answer with references. Each reference must include exactQuote and url. - If uncertain, use @@ -356,11 +355,11 @@ export async function getResponse(question?: string, return { exactQuote: ref.exactQuote, title: allURLs[ref.url]?.title, - url: normalizeUrl(ref.url) + url: ref.url ? normalizeUrl(ref.url) : '' } }); - context.actionTracker.trackThink(`But wait, let me evaluate the answer first.`) + context.actionTracker.trackThink('eval_first', SchemaGen.languageCode) const evaluation = await evaluateAnswer(currentQuestion, thisStep, evaluationMetrics[currentQuestion], @@ -506,7 +505,7 @@ But then you realized you have asked them before. You decided to to think out of if (keywordsQueries.length > 0) { - context.actionTracker.trackThink(`Let me search for "${keywordsQueries.join(', ')}" to gather more information.`) + context.actionTracker.trackThink('search_for', SchemaGen.languageCode, {keywords: keywordsQueries.join(', ')}); for (const query of keywordsQueries) { console.log(`Search query: ${query}`); @@ -594,7 +593,7 @@ You decided to think out of the box or cut from a completely different angle. const uniqueURLs = thisStep.URLTargets; if (uniqueURLs.length > 0) { - context.actionTracker.trackThink(`Let me read ${uniqueURLs.join(', ')} to gather more information.`); + context.actionTracker.trackThink('read_for', SchemaGen.languageCode, {urls: uniqueURLs.join(', ')}); const urlResults = await Promise.all( uniqueURLs.map(async url => { diff --git a/src/app.ts b/src/app.ts index d16a24e0..87d90acd 100644 --- a/src/app.ts +++ b/src/app.ts @@ -132,7 +132,7 @@ function calculateDelay(chunk: string, burstMode: boolean): number { // Special handling for CJK characters if (/^[\u4e00-\u9fff\u3040-\u30ff\uac00-\ud7af]$/.test(chunk)) { - return Math.random() * 50 + 10; // Longer delay for individual CJK characters + return Math.random() * 25 + 10; // Longer delay for individual CJK characters } // Base delay calculation diff --git a/src/tools/evaluator.ts b/src/tools/evaluator.ts index 0526ed62..9fad9869 100644 --- a/src/tools/evaluator.ts +++ b/src/tools/evaluator.ts @@ -474,7 +474,9 @@ Question: "请解释赤壁之战的历史背景、主要参与者以及战略意 Now evaluate this question: -Question: ${question}`; +Question: ${question} + +NOTE: "think" field should be in the same language as the question`; } const TOOL_NAME = 'evaluator'; @@ -560,7 +562,7 @@ export async function evaluateAnswer( case 'attribution': { // Safely handle references and ensure we have content - const allKnowledge = await fetchSourceContent(uniqueNewURLs, trackers); + const allKnowledge = await fetchSourceContent(uniqueNewURLs, trackers, schemaGen); visitedURLs.push(...uniqueNewURLs); if (allKnowledge.trim().length === 0) { @@ -610,9 +612,9 @@ export async function evaluateAnswer( } // Helper function to fetch and combine source content -async function fetchSourceContent(urls: string[], trackers: TrackerContext): Promise { +async function fetchSourceContent(urls: string[], trackers: TrackerContext, schemaGen: Schemas): Promise { if (!urls.length) return ''; - trackers.actionTracker.trackThink('Let me fetch the source content to verify the answer.'); + trackers.actionTracker.trackThink('read_for_verify', schemaGen.languageCode); try { const results = await Promise.all( urls.map(async (url) => { diff --git a/src/utils/action-tracker.ts b/src/utils/action-tracker.ts index c00fa405..2e0b77a6 100644 --- a/src/utils/action-tracker.ts +++ b/src/utils/action-tracker.ts @@ -1,5 +1,6 @@ -import { EventEmitter } from 'events'; -import { StepAction } from '../types'; +import {EventEmitter} from 'events'; +import {StepAction} from '../types'; +import {getI18nText} from "./text-tools"; interface ActionState { thisStep: StepAction; @@ -8,6 +9,7 @@ interface ActionState { totalStep: number; } + export class ActionTracker extends EventEmitter { private state: ActionState = { thisStep: {action: 'answer', answer: '', references: [], think: ''}, @@ -17,18 +19,20 @@ export class ActionTracker extends EventEmitter { }; trackAction(newState: Partial) { - this.state = { ...this.state, ...newState }; + this.state = {...this.state, ...newState}; this.emit('action', this.state.thisStep); } - trackThink(think: string) { - // only update the think field of the current state - this.state = { ...this.state, thisStep: { ...this.state.thisStep, think } }; + trackThink(think: string, lang?: string, params = {}) { + if (lang) { + think = getI18nText(think, lang, params); + } + this.state = {...this.state, thisStep: {...this.state.thisStep, think}}; this.emit('action', this.state.thisStep); } getState(): ActionState { - return { ...this.state }; + return {...this.state}; } reset() { diff --git a/src/utils/i18n.json b/src/utils/i18n.json new file mode 100644 index 00000000..ccf99315 --- /dev/null +++ b/src/utils/i18n.json @@ -0,0 +1,86 @@ +{ + "en": { + "eval_first": "But wait, let me evaluate the answer first.", + "search_for": "Let me search for ${keywords} to gather more information.", + "read_for": "Let me read ${urls} to gather more information.", + "read_for_verify": "Let me fetch the source content to verify the answer." + }, + "zh-CN": { + "eval_first": "等等，让我先自己评估一下答案。", + "search_for": "让我搜索${keywords}来获取更多信息。", + "read_for": "让我读取网页${urls}来获取更多信息。", + "read_for_verify": "让我读取源网页内容来验证答案。" + }, + "zh-TW": { + "eval_first": "等等，讓我先評估一下答案。", + "search_for": "讓我搜索${keywords}來獲取更多信息。", + "read_for": "讓我閱讀${urls}來獲取更多信息。", + "read_for_verify": "讓我獲取源內容來驗證答案。" + }, + "ja": { + "eval_first": "ちょっと待って、まず答えを評価します。", + "search_for": "キーワード${keywords}で検索して、情報を集めます。", + "read_for": "URL${urls}を読んで、情報を集めます。", + "read_for_verify": "答えを確認するために、ソースコンテンツを取得します。" + }, + "ko": { + "eval_first": "잠시만요, 먼저 답변을 평가해 보겠습니다.", + "search_for": "키워드 ${keywords}로 검색하여 더 많은 정보를 수집하겠습니다.", + "read_for": "URL ${urls}을 읽어 더 많은 정보를 수집하겠습니다.", + "read_for_verify": "답변을 확인하기 위해 소스 콘텐츠를 가져오겠습니다." + }, + "fr": { + "eval_first": "Un instant, je vais d'abord évaluer la réponse.", + "search_for": "Je vais rechercher ${keywords} pour obtenir plus d'informations.", + "read_for": "Je vais lire ${urls} pour obtenir plus d'informations.", + "read_for_verify": "Je vais récupérer le contenu source pour vérifier la réponse." + }, + "de": { + "eval_first": "Einen Moment, ich werde die Antwort zuerst evaluieren.", + "search_for": "Ich werde nach ${keywords} suchen, um weitere Informationen zu sammeln.", + "read_for": "Ich werde ${urls} lesen, um weitere Informationen zu sammeln.", + "read_for_verify": "Ich werde den Quellinhalt abrufen, um die Antwort zu überprüfen." + }, + "es": { + "eval_first": "Un momento, voy a evaluar la respuesta primero.", + "search_for": "Voy a buscar ${keywords} para recopilar más información.", + "read_for": "Voy a leer ${urls} para recopilar más información.", + "read_for_verify": "Voy a obtener el contenido fuente para verificar la respuesta." + }, + "it": { + "eval_first": "Un attimo, valuterò prima la risposta.", + "search_for": "Cercherò ${keywords} per raccogliere ulteriori informazioni.", + "read_for": "Leggerò ${urls} per raccogliere ulteriori informazioni.", + "read_for_verify": "Recupererò il contenuto sorgente per verificare la risposta." + }, + "pt": { + "eval_first": "Um momento, vou avaliar a resposta primeiro.", + "search_for": "Vou pesquisar ${keywords} para reunir mais informações.", + "read_for": "Vou ler ${urls} para reunir mais informações.", + "read_for_verify": "Vou buscar o conteúdo da fonte para verificar a resposta." + }, + "ru": { + "eval_first": "Подождите, я сначала оценю ответ.", + "search_for": "Дайте мне поискать ${keywords} для сбора дополнительной информации.", + "read_for": "Дайте мне прочитать ${urls} для сбора дополнительной информации.", + "read_for_verify": "Дайте мне получить исходный контент для проверки ответа." + }, + "ar": { + "eval_first": "لكن انتظر، دعني أقوم بتقييم الإجابة أولاً.", + "search_for": "دعني أبحث عن ${keywords} لجمع المزيد من المعلومات.", + "read_for": "دعني أقرأ ${urls} لجمع المزيد من المعلومات.", + "read_for_verify": "دعني أحضر محتوى المصدر للتحقق من الإجابة." + }, + "nl": { + "eval_first": "Een moment, ik zal het antwoord eerst evalueren.", + "search_for": "Ik zal zoeken naar ${keywords} om meer informatie te verzamelen.", + "read_for": "Ik zal ${urls} lezen om meer informatie te verzamelen.", + "read_for_verify": "Ik zal de broninhoud ophalen om het antwoord te verifiëren." + }, + "zh": { + "eval_first": "等等，让我先评估一下答案。", + "search_for": "让我搜索${keywords}来获取更多信息。", + "read_for": "让我阅读${urls}来获取更多信息。", + "read_for_verify": "让我获取源内容来验证答案。" + } +} \ No newline at end of file diff --git a/src/utils/schemas.ts b/src/utils/schemas.ts index 796e433a..b14b4fdf 100644 --- a/src/utils/schemas.ts +++ b/src/utils/schemas.ts @@ -61,7 +61,7 @@ ${question}`; export class Schemas { private languageStyle: string = 'formal English'; - private languageCode: string = 'en'; + public languageCode: string = 'en'; constructor(query: string) { diff --git a/src/utils/text-tools.ts b/src/utils/text-tools.ts index c045b560..3e3a2547 100644 --- a/src/utils/text-tools.ts +++ b/src/utils/text-tools.ts @@ -125,3 +125,39 @@ export function chooseK(a: string[], k: number) { export function removeHTMLtags(text: string) { return text.replace(/<[^>]*>?/gm, ''); } + + +export function getI18nText(key: string, lang = 'en', params: Record = {}) { + // 获取i18n数据 + const i18nData = require('./i18n.json'); + + // 确保语言代码存在，如果不存在则使用英语作为后备 + if (!i18nData[lang]) { + console.error(`Language '${lang}' not found, falling back to English.`); + lang = 'en'; + } + + // 获取对应语言的文本 + let text = i18nData[lang][key]; + + // 如果文本不存在，则使用英语作为后备 + if (!text) { + console.error(`Key '${key}' not found for language '${lang}', falling back to English.`); + text = i18nData['en'][key]; + + // 如果英语版本也不存在，则返回键名 + if (!text) { + console.error(`Key '${key}' not found for English either.`); + return key; + } + } + + // 替换模板中的变量 + if (params) { + Object.keys(params).forEach(paramKey => { + text = text.replace(`\${${paramKey}}`, params[paramKey]); + }); + } + + return text; +} \ No newline at end of file