feat: 可以进行多轮文献查询
This commit is contained in:
liuweiqing
parent
edfb1c5475
commit
9d799f1736
|
|
@ -6,3 +6,4 @@ NEXT_PUBLIC_PAPER_URL=/api/paper
|
|||
NEXT_PUBLIC_SEMANTIC_API_KEY=hEQvK6ARe84dzDPcMnpzX4n9jfoqztkMfaftPWnb
|
||||
NEXT_PUBLIC_PUBMED_API_KEY=057616e7ce6c722f2ae8679e38a8be9b1a09
|
||||
VERCEL_URL=https://www.paperai.life
|
||||
NODE_ENV=production
|
||||
|
|
|
|||
|
|
@ -19,7 +19,7 @@ const initialState: APIState = {
|
|||
2.文献引用：只引用与主题紧密相关的论文。在引用文献时，文末应使用方括号内的数字来标注引用来源，如 [1]。。请确保每个引用在文章中都有其对应的编号，*无需在文章末尾提供参考文献列表*。*每个文献对应的序号只应该出现一次，比如说引用了第一篇文献文中就只能出现一次[1]*。
|
||||
3.忽略无关文献：对于与主题无关的论文，请不要包含在您的写作中。只关注对理解和阐述主题有实质性帮助的资料。
|
||||
4.来源明确：在文章中，清楚地指出每个引用的具体来源。引用的信息应准确无误，确保读者能够追溯到原始文献。
|
||||
5.使用中文完成回答,不超过三百字
|
||||
5.使用用户所说的语言完成回答，不超过五百字
|
||||
6.只能对给出的文献进行引用，坚决不能虚构文献。
|
||||
返回格式举例：
|
||||
在某个方面，某论文实现了以下突破...[1],在另一篇论文中，研究了...[2]`,
|
||||
|
|
|
|||
|
|
@ -38,12 +38,13 @@ interface Author {
|
|||
async function getArxivPapers(
|
||||
query: string,
|
||||
maxResults = 5,
|
||||
offset = -1,
|
||||
sortBy = "submittedDate",
|
||||
sortOrder = "descending"
|
||||
) {
|
||||
const maxOffset = 30 - maxResults; // 假设总记录数为 100
|
||||
const start = getRandomOffset(maxOffset);
|
||||
const url = `https://export.arxiv.org/api/query?search_query=${query}&start=${start}&max_results=${maxResults}&sortBy=${sortBy}&sortOrder=${sortOrder}`;
|
||||
const maxOffset = 20 - maxResults; // 假设总记录数为 20
|
||||
if (offset === -1) offset = getRandomOffset(maxOffset);
|
||||
const url = `https://export.arxiv.org/api/query?search_query=${query}&start=${offset}&max_results=${maxResults}&sortBy=${sortBy}&sortOrder=${sortOrder}`;
|
||||
|
||||
try {
|
||||
const response = await axios.get(url);
|
||||
|
|
|
|||
|
|
@ -10,14 +10,20 @@ type PubMedID = string;
|
|||
// 定义idList为PubMedID数组
|
||||
type IDList = PubMedID[];
|
||||
|
||||
async function getPubMedPapers(query: string, year: number, limit = 2) {
|
||||
async function getPubMedPapers(
|
||||
query: string,
|
||||
year: number,
|
||||
offset = -1,
|
||||
limit = 2
|
||||
) {
|
||||
try {
|
||||
const baseURL =
|
||||
"https://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi";
|
||||
const db = "pubmed"; // 设定搜索的数据库为PubMed
|
||||
const retMax = limit; // 检索的最大记录数
|
||||
const retStart = getRandomOffset(20 - limit); // 假设每页最多30条，根据需要随机偏移
|
||||
const url = `${baseURL}?db=${db}&term=${query}[Title/Abstract]+AND+2018:3000[Date - Publication]&retMax=${retMax}&retStart=${retStart}&api_key=${process.env.NEXT_PUBLIC_PUBMED_API_KEY}`;
|
||||
const maxOffset = 20 - limit; // 假设总记录数为 20
|
||||
if (offset === -1) offset = getRandomOffset(maxOffset);
|
||||
const url = `${baseURL}?db=${db}&term=${query}[Title/Abstract]+AND+2018:3000[Date - Publication]&retMax=${retMax}&retStart=${offset}&api_key=${process.env.NEXT_PUBLIC_PUBMED_API_KEY}`;
|
||||
const response = await axios.get(url, { responseType: "text" });
|
||||
console.log(response.data);
|
||||
// 解析XML数据
|
||||
|
|
@ -155,9 +161,14 @@ async function getPubMedPaperDetails(idList: IDList) {
|
|||
}
|
||||
|
||||
// 示例：使用这些函数
|
||||
async function fetchPubMedData(query: string, year: number, limit: number) {
|
||||
async function fetchPubMedData(
|
||||
query: string,
|
||||
year: number,
|
||||
limit: number,
|
||||
offset: number
|
||||
) {
|
||||
try {
|
||||
const idList = await getPubMedPapers(query, year, limit);
|
||||
const idList = await getPubMedPapers(query, year, offset, limit);
|
||||
if (idList && idList.length > 0) {
|
||||
const paperDetails = await getPubMedPaperDetails(idList);
|
||||
console.log("fetchPubMedData", paperDetails); // 处理或显示文章详情
|
||||
|
|
|
|||
|
|
@ -15,10 +15,15 @@ interface Paper {
|
|||
url: string;
|
||||
}
|
||||
|
||||
async function getSemanticPapers(query: string, year: string, limit = 2) {
|
||||
async function getSemanticPapers(
|
||||
query: string,
|
||||
year: string,
|
||||
offset = -1,
|
||||
limit = 2
|
||||
) {
|
||||
try {
|
||||
const maxOffset = 20 - limit; // 假设总记录数为 20
|
||||
const offset = getRandomOffset(maxOffset);
|
||||
if (offset === -1) offset = getRandomOffset(maxOffset);
|
||||
const url = `https://api.semanticscholar.org/graph/v1/paper/search`;
|
||||
const response = await axios.get(url, {
|
||||
headers: {
|
||||
|
|
|
|||
|
|
@ -101,6 +101,11 @@ const QEditor = ({ lng }) => {
|
|||
"gpt语言模型",
|
||||
"gpt-4"
|
||||
); // 默认选项
|
||||
const [generatedPaperNumber, setGeneratedPaperNumber] = useLocalStorage(
|
||||
"生成次数",
|
||||
1
|
||||
); // 初始值设为1
|
||||
|
||||
//redux
|
||||
const dispatch = useAppDispatch();
|
||||
const references = useAppSelector((state) => state.auth.referencesRedux);
|
||||
|
|
@ -231,16 +236,20 @@ const QEditor = ({ lng }) => {
|
|||
const handleInputChange = (event: any) => {
|
||||
setUserInput(event.target.value);
|
||||
};
|
||||
|
||||
// 处理输入generatedPaperNumber变化的函数
|
||||
const handleGeneratedPaperNumberChange = (event: any) => {
|
||||
const newValue = parseInt(event.target.value, 10);
|
||||
setGeneratedPaperNumber(newValue);
|
||||
};
|
||||
// 处理AI写作
|
||||
const handleAIWrite = async () => {
|
||||
quill.setSelection(cursorPosition, 0); // 将光标移动到原来的位置
|
||||
quill!.setSelection(cursorPosition!, 0); // 将光标移动到原来的位置
|
||||
|
||||
const prompt = "请帮助用户完成论文写作，使用用户所说的语言完成";
|
||||
await sendMessageToOpenAI(
|
||||
userInput,
|
||||
quill,
|
||||
selectedModel,
|
||||
quill!,
|
||||
selectedModel!,
|
||||
apiKey,
|
||||
upsreamUrl,
|
||||
prompt
|
||||
|
|
@ -248,20 +257,22 @@ const QEditor = ({ lng }) => {
|
|||
// 清空input内容
|
||||
setUserInput("");
|
||||
// 重新获取更新后的内容并更新 Redux store
|
||||
const updatedContent = quill.root.innerHTML;
|
||||
const updatedContent = quill!.root.innerHTML;
|
||||
dispatch(setEditorContent(updatedContent));
|
||||
};
|
||||
|
||||
// 处理paper2AI
|
||||
async function paper2AI(topic: string) {
|
||||
quill.setSelection(cursorPosition, 0); // 将光标移动到原来的位置
|
||||
|
||||
quill!.setSelection(cursorPosition!, 0); // 将光标移动到原来的位置
|
||||
let offset = -1;
|
||||
if (generatedPaperNumber) offset = 0;
|
||||
for (let i = 0; i < generatedPaperNumber!; i++) {
|
||||
try {
|
||||
if (!topic) {
|
||||
//使用ai提取当前要请求的论文主题
|
||||
const prompt =
|
||||
"As a topic extraction assistant, you can help me extract the current discussion of the paper topic, I will enter the content of the paper, you extract the paper topic , no more than two, Hyphenated query terms yield no matches (replace it with space to find matches) return format is: topic1 topic2";
|
||||
const userMessage = getTextBeforeCursor(quill, 2000);
|
||||
const userMessage = getTextBeforeCursor(quill!, 2000);
|
||||
topic = await getTopicFromAI(userMessage, prompt, apiKey);
|
||||
console.log("topic in AI before removeSpecialCharacters", topic);
|
||||
topic = removeSpecialCharacters(topic);
|
||||
|
|
@ -277,21 +288,21 @@ const QEditor = ({ lng }) => {
|
|||
rawData = await getArxivPapers(topic);
|
||||
console.log("arxiv rawdata:", rawData);
|
||||
// 将 rawData 转换为引用数组
|
||||
newReferences = rawData.map((entry) => ({
|
||||
newReferences = rawData.map((entry: any) => ({
|
||||
url: entry.id,
|
||||
title: entry.title,
|
||||
year: entry.published,
|
||||
author: entry.authors?.slice(0, 3).join(", "),
|
||||
}));
|
||||
dataString = rawData
|
||||
.map((entry) => {
|
||||
.map((entry: any) => {
|
||||
return `ID: ${entry.id}\nTime: ${entry.published}\nTitle: ${entry.title}\nSummary: ${entry.summary}\n\n`;
|
||||
})
|
||||
.join("");
|
||||
} else if (selectedSource === "semanticScholar") {
|
||||
rawData = await getSemanticPapers(topic, "2015-2023");
|
||||
rawData = await getSemanticPapers(topic, "2015-2023", offset);
|
||||
// 将 rawData 转换为引用数组
|
||||
newReferences = rawData.map((entry) => ({
|
||||
newReferences = rawData.map((entry: any) => ({
|
||||
url: entry.url,
|
||||
title: entry.title,
|
||||
year: entry.year,
|
||||
|
|
@ -300,16 +311,16 @@ const QEditor = ({ lng }) => {
|
|||
journal: formatJournalReference(entry),
|
||||
}));
|
||||
dataString = rawData
|
||||
.map((entry) => {
|
||||
.map((entry: any) => {
|
||||
return `Time: ${entry.year}\nTitle: ${entry.title}\nSummary: ${entry.abstract}\n\n`;
|
||||
})
|
||||
.join("");
|
||||
} else if (selectedSource === "pubmed") {
|
||||
rawData = await fetchPubMedData(topic, 2020, 2);
|
||||
rawData = await fetchPubMedData(topic, 2020, offset, 2);
|
||||
if (!rawData) {
|
||||
throw new Error("未搜索到文献 from PubMed.");
|
||||
}
|
||||
newReferences = rawData.map((entry) => ({
|
||||
newReferences = rawData.map((entry: any) => ({
|
||||
id: entry.id, // 文章的 PubMed ID
|
||||
title: entry.title, // 文章的标题
|
||||
abstract: entry.abstract, // 文章的摘要
|
||||
|
|
@ -324,13 +335,13 @@ const QEditor = ({ lng }) => {
|
|||
console.log(newReferences);
|
||||
|
||||
dataString = rawData
|
||||
.map((entry) => {
|
||||
.map((entry: any) => {
|
||||
return `Time: ${entry.year}\nTitle: ${entry.title}\nSummary: ${entry.abstract}\n\n`;
|
||||
})
|
||||
.join("");
|
||||
}
|
||||
//在对应的位置添加文献
|
||||
const nearestNumber = getNumberBeforeCursor(quill);
|
||||
const nearestNumber = getNumberBeforeCursor(quill!);
|
||||
dispatch(
|
||||
addReferencesRedux({
|
||||
references: newReferences,
|
||||
|
|
@ -346,20 +357,20 @@ const QEditor = ({ lng }) => {
|
|||
// editorValue
|
||||
// )}`;
|
||||
const content = `之前用户已经完成的内容上下文：${getTextBeforeCursor(
|
||||
quill,
|
||||
500
|
||||
quill!,
|
||||
900
|
||||
)},搜索到的论文内容:${trimmedMessage},需要完成的论文主题：${topic},请根据搜索到的论文内容完成用户的论文`;
|
||||
await sendMessageToOpenAI(
|
||||
content,
|
||||
quill,
|
||||
selectedModel,
|
||||
quill!,
|
||||
selectedModel!,
|
||||
apiKey,
|
||||
upsreamUrl,
|
||||
systemPrompt
|
||||
);
|
||||
setUserInput("");
|
||||
// 重新获取更新后的内容并更新 Redux store
|
||||
const updatedContent = quill.root.innerHTML;
|
||||
const updatedContent = quill!.root.innerHTML;
|
||||
dispatch(setEditorContent(updatedContent));
|
||||
if (isVip) {
|
||||
//在云端同步supabase
|
||||
|
|
@ -370,12 +381,15 @@ const QEditor = ({ lng }) => {
|
|||
paperNumberRedux
|
||||
);
|
||||
}
|
||||
//修改offset使得按照接下来的顺序进行获取文献
|
||||
offset += 2;
|
||||
} catch (error) {
|
||||
// console.error("Error fetching data:", error);
|
||||
// 在处理错误后，再次抛出这个错误
|
||||
throw new Error(`Paper2AI出现错误: ${error}`);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return (
|
||||
<div className="flex flex-col ">
|
||||
|
|
@ -420,6 +434,12 @@ const QEditor = ({ lng }) => {
|
|||
<option value="gpt-4">gpt-4</option>
|
||||
<option value="deepseek-chat">deepseek-chat</option>
|
||||
</select>
|
||||
<input
|
||||
type="number"
|
||||
value={generatedPaperNumber}
|
||||
onChange={handleGeneratedPaperNumberChange}
|
||||
className="border border-gray-300 text-gray-700 text-sm p-1 rounded w-16"
|
||||
/>
|
||||
<button
|
||||
onClick={() => formatTextInEditor(quill)} // 假设 updateIndex 是处理更新操作的函数
|
||||
className="bg-gray-300 hover:bg-gray-400 text-black font-bold py-2 px-4 rounded"
|
||||
|
|
|
|||
|
|
@ -1,5 +1,7 @@
|
|||
import { Transforms } from "slate";
|
||||
import { Editor } from "slate";
|
||||
import Quill from "quill";
|
||||
|
||||
import { extractText } from "@/utils/others/slateutils";
|
||||
import {
|
||||
updateBracketNumbersInDeltaKeepSelection,
|
||||
|
|
@ -20,8 +22,8 @@ function isValidApiKey(apiKey: string) {
|
|||
|
||||
const sendMessageToOpenAI = async (
|
||||
content: string,
|
||||
editor: Editor,
|
||||
selectedModel: "gpt3.5",
|
||||
editor: Quill,
|
||||
selectedModel: string,
|
||||
apiKey: string,
|
||||
upsreamUrl: string,
|
||||
prompt?: string
|
||||
|
|
@ -54,7 +56,7 @@ const sendMessageToOpenAI = async (
|
|||
2.文献引用：只引用与主题紧密相关的论文。在引用文献时，文末应使用方括号内的数字来标注引用来源，如 [1]。。请确保每个引用在文章中都有其对应的编号，*无需在文章末尾提供参考文献列表*。*每个文献对应的序号只应该出现一次，比如说引用了第一篇文献文中就只能出现一次[1]*。
|
||||
3.忽略无关文献：对于与主题无关的论文，请不要包含在您的写作中。只关注对理解和阐述主题有实质性帮助的资料。
|
||||
4.来源明确：在文章中，清楚地指出每个引用的具体来源。引用的信息应准确无误，确保读者能够追溯到原始文献。
|
||||
5.使用用户所说的语言完成回答,不超过三百字
|
||||
5.使用用户所说的语言完成回答，不超过五百字
|
||||
6.只能对给出的文献进行引用，坚决不能虚构文献。
|
||||
返回格式举例：
|
||||
在某个方面，某论文实现了以下突破...[1],在另一篇论文中，研究了...[2]`,
|
||||
|
|
|
|||
|
|
@ -4,6 +4,7 @@
|
|||
|
||||
import * as Sentry from "@sentry/nextjs";
|
||||
|
||||
if (process.env.NODE_ENV === "production") {
|
||||
Sentry.init({
|
||||
dsn: "https://523c4056ba48d012c62a377dfc49f647@o4506728662564864.ingest.sentry.io/4506728672264192",
|
||||
|
||||
|
|
@ -32,3 +33,4 @@ Sentry.init({
|
|||
}),
|
||||
],
|
||||
});
|
||||
}
|
||||
|
|
|
|||
|
|
@ -4,7 +4,7 @@
|
|||
// https://docs.sentry.io/platforms/javascript/guides/nextjs/
|
||||
|
||||
import * as Sentry from "@sentry/nextjs";
|
||||
|
||||
if (process.env.NODE_ENV === "production") {
|
||||
Sentry.init({
|
||||
dsn: "https://523c4056ba48d012c62a377dfc49f647@o4506728662564864.ingest.sentry.io/4506728672264192",
|
||||
|
||||
|
|
@ -14,3 +14,4 @@ Sentry.init({
|
|||
// Setting this option to true will print useful information to the console while you're setting up Sentry.
|
||||
debug: false,
|
||||
});
|
||||
}
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
// https://docs.sentry.io/platforms/javascript/guides/nextjs/
|
||||
|
||||
import * as Sentry from "@sentry/nextjs";
|
||||
|
||||
if (process.env.NODE_ENV === "production") {
|
||||
Sentry.init({
|
||||
dsn: "https://523c4056ba48d012c62a377dfc49f647@o4506728662564864.ingest.sentry.io/4506728672264192",
|
||||
|
||||
|
|
@ -13,3 +13,4 @@ Sentry.init({
|
|||
// Setting this option to true will print useful information to the console while you're setting up Sentry.
|
||||
debug: false,
|
||||
});
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in New Issue
Block a user