feat(前端): 聊天钩子支持文档上传和解析
- 集成文档类型检测工具 - 支持PDF/Word/Excel文件上传 - PDF文档转Base64传给后端API - Office文档(Word/Excel)传给后端解析 - 文本文件直接读取内容 - 添加文档大小验证
This commit is contained in:
gaoziman
parent
2344024e84
commit
c776fb95b7
@ -2,6 +2,16 @@
|
||||
|
||||
import { useState, useCallback, useRef } from 'react';
|
||||
import { executePythonInPyodide, type LoadingCallback } from '@/services/tools/pyodideRunner';
|
||||
import {
|
||||
detectDocumentType,
|
||||
isPdfFile,
|
||||
isOfficeDocument,
|
||||
validateDocumentSize,
|
||||
fileToBase64 as documentFileToBase64,
|
||||
type PdfDocumentData,
|
||||
type OfficeDocumentData,
|
||||
getFileMimeType,
|
||||
} from '@/utils/document-utils';
|
||||
|
||||
export interface StreamMessage {
|
||||
type: 'thinking' | 'text' | 'tool_use_start' | 'tool_execution_result' | 'tool_search_images' | 'tool_search_videos' | 'pyodide_execution_required' | 'tool_used' | 'done' | 'error';
|
||||
@ -263,16 +273,23 @@ export function useStreamChat() {
|
||||
const uploadedDocuments: UploadedDocument[] = [];
|
||||
const imageContents: { type: 'image'; media_type: string; data: string }[] = [];
|
||||
const documentContents: { name: string; content: string }[] = [];
|
||||
// PDF 文档（直接传给 Claude API）
|
||||
const pdfDocuments: PdfDocumentData[] = [];
|
||||
// Office 文档（Word/Excel，需要后端解析）
|
||||
const officeDocuments: OfficeDocumentData[] = [];
|
||||
|
||||
if (files && files.length > 0) {
|
||||
console.log('[useStreamChat] Processing files:', files.length);
|
||||
for (const fileInfo of files) {
|
||||
const docType = detectDocumentType(fileInfo.file);
|
||||
console.log('[useStreamChat] File info:', {
|
||||
name: fileInfo.file.name,
|
||||
type: fileInfo.file.type,
|
||||
size: fileInfo.file.size,
|
||||
isImage: fileInfo.file.type.startsWith('image/'),
|
||||
docType: docType,
|
||||
});
|
||||
|
||||
// 处理图片文件
|
||||
if (fileInfo.file.type.startsWith('image/')) {
|
||||
try {
|
||||
@ -291,6 +308,67 @@ export function useStreamChat() {
|
||||
console.error('Failed to convert image to base64:', err);
|
||||
}
|
||||
}
|
||||
// 处理 PDF 文件（使用 Claude 原生支持）
|
||||
else if (isPdfFile(fileInfo.file)) {
|
||||
// 验证文件大小
|
||||
const validation = validateDocumentSize(fileInfo.file);
|
||||
if (!validation.valid) {
|
||||
console.error('[useStreamChat] PDF validation failed:', validation.error);
|
||||
// 可以选择抛出错误或显示提示
|
||||
continue;
|
||||
}
|
||||
|
||||
try {
|
||||
const base64 = await documentFileToBase64(fileInfo.file);
|
||||
console.log('[useStreamChat] PDF converted to base64, length:', base64.length);
|
||||
pdfDocuments.push({
|
||||
name: fileInfo.file.name,
|
||||
size: fileInfo.file.size,
|
||||
data: base64,
|
||||
media_type: 'application/pdf',
|
||||
});
|
||||
// 保存到 uploadedDocuments 用于前端显示
|
||||
uploadedDocuments.push({
|
||||
name: fileInfo.file.name,
|
||||
size: fileInfo.file.size,
|
||||
type: 'pdf',
|
||||
content: `[PDF 文档: ${fileInfo.file.name}]`, // PDF 内容由 Claude 直接处理
|
||||
});
|
||||
} catch (err) {
|
||||
console.error('Failed to convert PDF to base64:', err);
|
||||
}
|
||||
}
|
||||
// 处理 Office 文档（Word/Excel，需要后端解析）
|
||||
else if (isOfficeDocument(fileInfo.file)) {
|
||||
// 验证文件大小
|
||||
const validation = validateDocumentSize(fileInfo.file);
|
||||
if (!validation.valid) {
|
||||
console.error('[useStreamChat] Office document validation failed:', validation.error);
|
||||
continue;
|
||||
}
|
||||
|
||||
try {
|
||||
const base64 = await documentFileToBase64(fileInfo.file);
|
||||
const mimeType = getFileMimeType(fileInfo.file);
|
||||
console.log('[useStreamChat] Office document converted to base64, length:', base64.length, 'type:', docType);
|
||||
officeDocuments.push({
|
||||
name: fileInfo.file.name,
|
||||
size: fileInfo.file.size,
|
||||
data: base64,
|
||||
type: docType as 'word' | 'excel',
|
||||
mimeType: mimeType,
|
||||
});
|
||||
// 保存到 uploadedDocuments 用于前端显示
|
||||
uploadedDocuments.push({
|
||||
name: fileInfo.file.name,
|
||||
size: fileInfo.file.size,
|
||||
type: docType,
|
||||
content: `[${docType === 'word' ? 'Word' : 'Excel'} 文档: ${fileInfo.file.name}]`, // 内容由后端解析后补充
|
||||
});
|
||||
} catch (err) {
|
||||
console.error('Failed to convert Office document to base64:', err);
|
||||
}
|
||||
}
|
||||
// 处理文本类文件
|
||||
else if (isTextFile(fileInfo.file)) {
|
||||
try {
|
||||
@ -348,7 +426,7 @@ export function useStreamChat() {
|
||||
abortControllerRef.current = new AbortController();
|
||||
|
||||
try {
|
||||
// 调试日志：确认图片数据
|
||||
// 调试日志：确认图片和文档数据
|
||||
console.log('[useStreamChat] Sending request with:', {
|
||||
conversationId,
|
||||
messageLength: finalMessage.length,
|
||||
@ -356,11 +434,24 @@ export function useStreamChat() {
|
||||
tools,
|
||||
enableThinking,
|
||||
imagesCount: imageContents.length,
|
||||
pdfDocumentsCount: pdfDocuments.length,
|
||||
officeDocumentsCount: officeDocuments.length,
|
||||
images: imageContents.length > 0 ? imageContents.map(img => ({
|
||||
type: img.type,
|
||||
media_type: img.media_type,
|
||||
dataLength: img.data.length,
|
||||
})) : undefined,
|
||||
pdfDocuments: pdfDocuments.length > 0 ? pdfDocuments.map(doc => ({
|
||||
name: doc.name,
|
||||
size: doc.size,
|
||||
dataLength: doc.data.length,
|
||||
})) : undefined,
|
||||
officeDocuments: officeDocuments.length > 0 ? officeDocuments.map(doc => ({
|
||||
name: doc.name,
|
||||
size: doc.size,
|
||||
type: doc.type,
|
||||
dataLength: doc.data.length,
|
||||
})) : undefined,
|
||||
});
|
||||
|
||||
const response = await fetch('/api/chat', {
|
||||
@ -381,6 +472,10 @@ export function useStreamChat() {
|
||||
uploadedImages: uploadedImages.length > 0 ? uploadedImages : undefined,
|
||||
// 传递上传的文档用于保存到数据库
|
||||
uploadedDocuments: uploadedDocuments.length > 0 ? uploadedDocuments : undefined,
|
||||
// 传递 PDF 文档给后端（用于 Claude 原生 document 类型）
|
||||
pdfDocuments: pdfDocuments.length > 0 ? pdfDocuments : undefined,
|
||||
// 传递 Office 文档给后端（需要后端解析）
|
||||
officeDocuments: officeDocuments.length > 0 ? officeDocuments : undefined,
|
||||
}),
|
||||
signal: abortControllerRef.current.signal,
|
||||
});
|
||||
|
||||
Loading…
Reference in New Issue
Block a user