TanWenyan/leaudit-platform-frontend
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

备份

Browse Source
This commit is contained in:
yorn
2025-11-25 11:02:40 +08:00
parent 93bae2de17
commit 0ed6f0aaf4
5 changed files with 194 additions and 156 deletions
Download Patch File Download Diff File Expand all files Collapse all files
app/routes/monaco-demo.tsx
+162 -65
View File
@@ -13,6 +13,7 @@ import { useState, useRef, useEffect } from "react";
import { DiffEditor } from "@monaco-editor/react"; import { DiffEditor } from "@monaco-editor/react";
import type { editor } from "monaco-editor"; import type { editor } from "monaco-editor";
import { pdfjs } from 'react-pdf'; import { pdfjs } from 'react-pdf';
import mammoth from 'mammoth';
import { toastService } from '~/components/ui/Toast'; import { toastService } from '~/components/ui/Toast';
// 设置 PDF.js worker（与 pdf-demo.tsx 相同） // 设置 PDF.js worker（与 pdf-demo.tsx 相同）
@@ -25,14 +26,26 @@ export const meta: MetaFunction = () => {
]; ];
}; };
// 文档类型枚举
type DocumentType = 'pdf' | 'docx' | 'unknown';
// PDF 类型枚举 // PDF 类型枚举
type PdfType = 'text' | 'scanned' | 'unknown'; type PdfType = 'text' | 'scanned' | 'unknown';
// PDF 信息接口 // PDF 信息接口（内部使用）
interface PdfInfo { interface PdfInfo {
type: PdfType; type: PdfType;
numPages: number; numPages: number;
textLength: number; textLength: number;
confidence: number;
}
// 文档信息接口
interface DocumentInfo {
fileType: DocumentType;
pdfType?: PdfType; // 只有 PDF 才有
numPages?: number; // PDF 页数
textLength: number;
confidence: number; // 文本提取置信度 (0-1) confidence: number; // 文本提取置信度 (0-1)
} }
@@ -120,15 +133,23 @@ export default function MonacoDemoPage() {
const [diffCount, setDiffCount] = useState<number>(0); const [diffCount, setDiffCount] = useState<number>(0);
const [currentDiff, setCurrentDiff] = useState<number>(0); const [currentDiff, setCurrentDiff] = useState<number>(0);
// PDF相关状态 // 文档相关状态
const [pdf1Url, setPdf1Url] = useState<string>(''); const [doc1Url, setDoc1Url] = useState<string>('');
const [pdf2Url, setPdf2Url] = useState<string>(''); const [doc2Url, setDoc2Url] = useState<string>('');
const [pdf1Info, setPdf1Info] = useState<PdfInfo | null>(null); const [doc1Info, setDoc1Info] = useState<DocumentInfo | null>(null);
const [pdf2Info, setPdf2Info] = useState<PdfInfo | null>(null); const [doc2Info, setDoc2Info] = useState<DocumentInfo | null>(null);
const [isLoadingPdf1, setIsLoadingPdf1] = useState(false); const [isLoadingDoc1, setIsLoadingDoc1] = useState(false);
const [isLoadingPdf2, setIsLoadingPdf2] = useState(false); const [isLoadingDoc2, setIsLoadingDoc2] = useState(false);
const [useExample, setUseExample] = useState(true); const [useExample, setUseExample] = useState(true);
// 检测文件类型（根据文件路径）
const detectFileType = (filePath: string): DocumentType => {
const lowerPath = filePath.toLowerCase();
if (lowerPath.endsWith('.pdf')) return 'pdf';
if (lowerPath.endsWith('.docx') || lowerPath.endsWith('.doc')) return 'docx';
return 'unknown';
};
// PDF类型检测函数 // PDF类型检测函数
const detectPdfType = async (pdfUrl: string): Promise<PdfInfo> => { const detectPdfType = async (pdfUrl: string): Promise<PdfInfo> => {
const loadingTask = pdfjs.getDocument(pdfUrl); const loadingTask = pdfjs.getDocument(pdfUrl);
@@ -189,32 +210,81 @@ export default function MonacoDemoPage() {
return fullText; return fullText;
}; };
// 加载PDF并提取文本 // Word文档文本提取函数
const loadPdfAndExtractText = async (pdfUrl: string, setPdfInfo: (info: PdfInfo | null) => void, setLoading: (loading: boolean) => void, setTextContent: (text: string) => void) => { const extractTextFromWord = async (docUrl: string): Promise<string> => {
// 通过 fetch 获取文件
const response = await fetch(docUrl);
if (!response.ok) {
throw new Error(`无法加载文档: ${response.statusText}`);
}
// 获取 ArrayBuffer
const arrayBuffer = await response.arrayBuffer();
// 使用 mammoth 提取纯文本
const result = await mammoth.extractRawText({ arrayBuffer });
return result.value;
};
// 加载文档并提取文本（支持 PDF 和 Word）
const loadDocumentAndExtractText = async (
docUrl: string,
filePath: string,
setDocInfo: (info: DocumentInfo | null) => void,
setLoading: (loading: boolean) => void,
setTextContent: (text: string) => void
) => {
try { try {
setLoading(true); setLoading(true);
// 1. 检测PDF类型 // 1. 检测文件类型
const pdfInfo = await detectPdfType(pdfUrl); const fileType = detectFileType(filePath);
setPdfInfo(pdfInfo);
// 2. 提取文本 if (fileType === 'pdf') {
if (pdfInfo.type === 'text') { // PDF 处理
const text = await extractTextFromPdf(pdfUrl); const pdfInfo = await detectPdfType(docUrl);
const text = await extractTextFromPdf(docUrl);
const docInfo: DocumentInfo = {
fileType: 'pdf',
pdfType: pdfInfo.type,
numPages: pdfInfo.numPages,
textLength: pdfInfo.textLength,
confidence: pdfInfo.confidence
};
setDocInfo(docInfo);
setTextContent(text); setTextContent(text);
if (pdfInfo.type === 'text') {
toastService.success(`PDF加载成功！共 ${pdfInfo.numPages} 页，提取了 ${pdfInfo.textLength} 个字符`); toastService.success(`PDF加载成功！共 ${pdfInfo.numPages} 页，提取了 ${pdfInfo.textLength} 个字符`);
} else if (pdfInfo.type === 'scanned') { } else if (pdfInfo.type === 'scanned') {
toastService.warning('检测到扫描版PDF，文本提取质量可能较低'); toastService.warning('检测到扫描版PDF，文本提取质量可能较低');
const text = await extractTextFromPdf(pdfUrl);
setTextContent(text);
} else { } else {
toastService.error('无法识别PDF类型，可能是图片PDF'); toastService.error('无法识别PDF类型，可能是图片PDF');
}
} else if (fileType === 'docx') {
// Word 处理
const text = await extractTextFromWord(docUrl);
const docInfo: DocumentInfo = {
fileType: 'docx',
textLength: text.length,
confidence: 1.0 // Word 文档文本提取置信度为 100%
};
setDocInfo(docInfo);
setTextContent(text);
toastService.success(`Word文档加载成功！提取了 ${text.length} 个字符`);
} else {
toastService.error('不支持的文件类型');
setTextContent(''); setTextContent('');
} }
} catch (error) { } catch (error) {
console.error('PDF加载失败:', error); console.error('文档加载失败:', error);
toastService.error('PDF加载失败，请检查文件路径'); toastService.error(`文档加载失败: ${error instanceof Error ? error.message : '未知错误'}`);
setPdfInfo(null); setDocInfo(null);
setTextContent(''); setTextContent('');
} finally { } finally {
setLoading(false); setLoading(false);
@@ -281,8 +351,8 @@ export default function MonacoDemoPage() {
setModifiedText(CONTRACT_B); setModifiedText(CONTRACT_B);
setCurrentDiff(0); setCurrentDiff(0);
setUseExample(true); setUseExample(true);
setPdf1Info(null); setDoc1Info(null);
setPdf2Info(null); setDoc2Info(null);
// 重新计算差异数量 // 重新计算差异数量
setTimeout(() => { setTimeout(() => {
@@ -295,34 +365,50 @@ export default function MonacoDemoPage() {
}, 100); }, 100);
}; };
// 从URL参数加载PDF // 构建文件访问 URL
const loadPdfsFromUrl = () => { const buildFileUrl = (filePath: string): string => {
// 如果路径以 public/ 开头或者以已知的 public 子目录开头（如 testWork/）
// 则直接使用静态资源路径
if (filePath.startsWith('public/')) {
// 去掉 public/ 前缀，直接访问静态资源
return '/' + filePath.substring(7);
} else if (filePath.startsWith('testWork/') || filePath.startsWith('testPDF/')) {
// testWork 和 testPDF 目录在 public 下，直接作为静态资源访问
return '/' + filePath;
} else {
// 其他路径通过 api/pdf-proxy 代理访问（从 MinIO 获取）
return `/api/pdf-proxy?path=${encodeURIComponent(filePath)}`;
}
};
// 从URL参数加载文档（支持 PDF 和 Word）
const loadDocumentsFromUrl = () => {
if (typeof window === 'undefined') return; if (typeof window === 'undefined') return;
const searchParams = new URLSearchParams(window.location.search); const searchParams = new URLSearchParams(window.location.search);
const pdf1Path = searchParams.get('pdf1'); const doc1Path = searchParams.get('doc1') || searchParams.get('pdf1'); // 兼容旧参数名
const pdf2Path = searchParams.get('pdf2'); const doc2Path = searchParams.get('doc2') || searchParams.get('pdf2'); // 兼容旧参数名
if (pdf1Path || pdf2Path) { if (doc1Path || doc2Path) {
setUseExample(false); setUseExample(false);
if (pdf1Path) { if (doc1Path) {
const fullUrl = `/api/pdf-proxy?path=${encodeURIComponent(pdf1Path)}`; const fullUrl = buildFileUrl(doc1Path);
setPdf1Url(fullUrl); setDoc1Url(fullUrl);
loadPdfAndExtractText(fullUrl, setPdf1Info, setIsLoadingPdf1, setOriginalText); loadDocumentAndExtractText(fullUrl, doc1Path, setDoc1Info, setIsLoadingDoc1, setOriginalText);
} }
if (pdf2Path) { if (doc2Path) {
const fullUrl = `/api/pdf-proxy?path=${encodeURIComponent(pdf2Path)}`; const fullUrl = buildFileUrl(doc2Path);
setPdf2Url(fullUrl); setDoc2Url(fullUrl);
loadPdfAndExtractText(fullUrl, setPdf2Info, setIsLoadingPdf2, setModifiedText); loadDocumentAndExtractText(fullUrl, doc2Path, setDoc2Info, setIsLoadingDoc2, setModifiedText);
} }
} }
}; };
// 组件挂载时读取URL参数 // 组件挂载时读取URL参数
useEffect(() => { useEffect(() => {
loadPdfsFromUrl(); loadDocumentsFromUrl();
// eslint-disable-next-line react-hooks/exhaustive-deps // eslint-disable-next-line react-hooks/exhaustive-deps
}, []); }, []);
@@ -452,8 +538,8 @@ export default function MonacoDemoPage() {
</div> </div>
</div> </div>
{/* PDF加载信息 */} {/* 文档加载信息 */}
{!useExample && (pdf1Info || pdf2Info || isLoadingPdf1 || isLoadingPdf2) && ( {!useExample && (doc1Info || doc2Info || isLoadingDoc1 || isLoadingDoc2) && (
<div style={{ <div style={{
padding: '12px 24px', padding: '12px 24px',
backgroundColor: '#fff3cd', backgroundColor: '#fff3cd',
@@ -462,48 +548,58 @@ export default function MonacoDemoPage() {
color: '#856404' color: '#856404'
}}> }}>
<div style={{ display: 'flex', alignItems: 'flex-start', gap: '24px' }}> <div style={{ display: 'flex', alignItems: 'flex-start', gap: '24px' }}>
<i className="ri-file-pdf-line" style={{ fontSize: '18px', marginTop: '2px' }}></i> <i className="ri-file-text-line" style={{ fontSize: '18px', marginTop: '2px' }}></i>
<div style={{ flex: 1 }}> <div style={{ flex: 1 }}>
<strong>PDF文档信息：</strong> <strong>文档信息：</strong>
<div style={{ display: 'flex', gap: '24px', marginTop: '8px' }}> <div style={{ display: 'flex', gap: '24px', marginTop: '8px' }}>
{/* PDF 1 信息 */} {/* 文档 1 信息 */}
<div style={{ flex: 1 }}> <div style={{ flex: 1 }}>
<div style={{ fontWeight: 'bold', marginBottom: '4px' }}>📄 文档1（左侧/原始）</div> <div style={{ fontWeight: 'bold', marginBottom: '4px' }}>📄 文档1（左侧/原始）</div>
{isLoadingPdf1 ? ( {isLoadingDoc1 ? (
<div style={{ color: '#666' }}>⏳ 加载中...</div> <div style={{ color: '#666' }}>⏳ 加载中...</div>
) : pdf1Info ? ( ) : doc1Info ? (
<div> <div>
<div>类型: <span style={{ <div>类型: <span style={{
color: pdf1Info.type === 'text' ? '#28a745' : pdf1Info.type === 'scanned' ? '#ffc107' : '#dc3545', color: doc1Info.fileType === 'pdf' ? '#007bff' : doc1Info.fileType === 'docx' ? '#28a745' : '#dc3545',
fontWeight: 'bold' fontWeight: 'bold'
}}> }}>
{pdf1Info.type === 'text' ? '✅ 文本PDF' : pdf1Info.type === 'scanned' ? '⚠️ 扫描PDF' : '❌ 未知类型'} {doc1Info.fileType === 'pdf' ? '📕 PDF文档' : doc1Info.fileType === 'docx' ? '📘 Word文档' : '❌ 未知类型'}
</span></div> </span></div>
<div>页数: {pdf1Info.numPages} 页</div> {doc1Info.fileType === 'pdf' && doc1Info.numPages && (
<div>字符数: {pdf1Info.textLength} 个</div> <div>页数: {doc1Info.numPages} 页</div>
<div>置信度: {(pdf1Info.confidence * 100).toFixed(0)}%</div> )}
{doc1Info.fileType === 'pdf' && doc1Info.pdfType && (
<div>PDF类型: {doc1Info.pdfType === 'text' ? '✅ 文本' : doc1Info.pdfType === 'scanned' ? '⚠️ 扫描' : '❌ 未知'}</div>
)}
<div>字符数: {doc1Info.textLength} 个</div>
<div>置信度: {(doc1Info.confidence * 100).toFixed(0)}%</div>
</div> </div>
) : ( ) : (
<div style={{ color: '#999' }}>未加载</div> <div style={{ color: '#999' }}>未加载</div>
)} )}
</div> </div>
{/* PDF 2 信息 */} {/* 文档 2 信息 */}
<div style={{ flex: 1 }}> <div style={{ flex: 1 }}>
<div style={{ fontWeight: 'bold', marginBottom: '4px' }}>📄 文档2（右侧/修改）</div> <div style={{ fontWeight: 'bold', marginBottom: '4px' }}>📄 文档2（右侧/修改）</div>
{isLoadingPdf2 ? ( {isLoadingDoc2 ? (
<div style={{ color: '#666' }}>⏳ 加载中...</div> <div style={{ color: '#666' }}>⏳ 加载中...</div>
) : pdf2Info ? ( ) : doc2Info ? (
<div> <div>
<div>类型: <span style={{ <div>类型: <span style={{
color: pdf2Info.type === 'text' ? '#28a745' : pdf2Info.type === 'scanned' ? '#ffc107' : '#dc3545', color: doc2Info.fileType === 'pdf' ? '#007bff' : doc2Info.fileType === 'docx' ? '#28a745' : '#dc3545',
fontWeight: 'bold' fontWeight: 'bold'
}}> }}>
{pdf2Info.type === 'text' ? '✅ 文本PDF' : pdf2Info.type === 'scanned' ? '⚠️ 扫描PDF' : '❌ 未知类型'} {doc2Info.fileType === 'pdf' ? '📕 PDF文档' : doc2Info.fileType === 'docx' ? '📘 Word文档' : '❌ 未知类型'}
</span></div> </span></div>
<div>页数: {pdf2Info.numPages} 页</div> {doc2Info.fileType === 'pdf' && doc2Info.numPages && (
<div>字符数: {pdf2Info.textLength} 个</div> <div>页数: {doc2Info.numPages} 页</div>
<div>置信度: {(pdf2Info.confidence * 100).toFixed(0)}%</div> )}
{doc2Info.fileType === 'pdf' && doc2Info.pdfType && (
<div>PDF类型: {doc2Info.pdfType === 'text' ? '✅ 文本' : doc2Info.pdfType === 'scanned' ? '⚠️ 扫描' : '❌ 未知'}</div>
)}
<div>字符数: {doc2Info.textLength} 个</div>
<div>置信度: {(doc2Info.confidence * 100).toFixed(0)}%</div>
</div> </div>
) : ( ) : (
<div style={{ color: '#999' }}>未加载</div> <div style={{ color: '#999' }}>未加载</div>
@@ -537,7 +633,7 @@ export default function MonacoDemoPage() {
<div style={{ marginTop: '12px', paddingTop: '12px', borderTop: '1px solid #b3d9ff' }}> <div style={{ marginTop: '12px', paddingTop: '12px', borderTop: '1px solid #b3d9ff' }}>
<strong>💡 使用提示：</strong> <strong>💡 使用提示：</strong>
<div style={{ marginTop: '4px' }}> <div style={{ marginTop: '4px' }}>
您可以通过URL参数加载PDF文档进行对比： 您可以通过URL参数加载文档进行对比（支持 PDF 和 Word）：
<code style={{ <code style={{
display: 'block', display: 'block',
marginTop: '4px', marginTop: '4px',
@@ -547,10 +643,11 @@ export default function MonacoDemoPage() {
fontSize: '12px', fontSize: '12px',
wordBreak: 'break-all' wordBreak: 'break-all'
}}> }}>
/monaco-demo?pdf1=路径1&pdf2=路径2 /monaco-demo?doc1=路径1&doc2=路径2
</code> </code>
<div style={{ marginTop: '4px', fontSize: '12px' }}> <div style={{ marginTop: '4px', fontSize: '12px' }}>
示例: <code>/monaco-demo?pdf1=documents/contract_v1.pdf&pdf2=documents/contract_v2.pdf</code> <div>PDF示例: <code>/monaco-demo?doc1=documents/contract_v1.pdf&doc2=documents/contract_v2.pdf</code></div>
<div style={{ marginTop: '2px' }}>Word示例: <code>/monaco-demo?doc1=testWork/(最终版)智慧法务平台建设采购项目合同(1).docx&doc2=testWork/(最终版)智慧法务平台建设采购项目合同(2).docx</code></div>
</div> </div>
</div> </div>
</div> </div>
@@ -591,8 +688,8 @@ export default function MonacoDemoPage() {
}} }}
/> />
{/* PDF加载中的遮罩层 */} {/* 文档加载中的遮罩层 */}
{(isLoadingPdf1 || isLoadingPdf2) && ( {(isLoadingDoc1 || isLoadingDoc2) && (
<div style={{ <div style={{
position: 'absolute', position: 'absolute',
top: 0, top: 0,
@@ -616,10 +713,10 @@ export default function MonacoDemoPage() {
margin: '0 auto 16px' margin: '0 auto 16px'
}}></div> }}></div>
<div style={{ fontSize: '16px', color: '#333' }}> <div style={{ fontSize: '16px', color: '#333' }}>
正在加载PDF文档并提取文本... 正在加载文档并提取文本...
</div> </div>
{isLoadingPdf1 && <div style={{ fontSize: '14px', color: '#666', marginTop: '8px' }}>📄 加载文档1</div>} {isLoadingDoc1 && <div style={{ fontSize: '14px', color: '#666', marginTop: '8px' }}>📄 加载文档1</div>}
{isLoadingPdf2 && <div style={{ fontSize: '14px', color: '#666', marginTop: '8px' }}>📄 加载文档2</div>} {isLoadingDoc2 && <div style={{ fontSize: '14px', color: '#666', marginTop: '8px' }}>📄 加载文档2</div>}
</div> </div>
</div> </div>
)} )}
app/routes/pdf-demo.tsx
+13 -72
View File
@@ -39,7 +39,7 @@ interface HighlightArea {
// 基于坐标的字符数据 // 基于坐标的字符数据
interface CharacterBox { interface CharacterBox {
box: [number, number][]; // 4个点：左上、右上、右下、左下 box: [number, number][];
char: string; char: string;
page: number; page: number;
} }
@@ -70,7 +70,7 @@ export default function PdfDemo() {
// PDF文件URL（使用示例PDF） // PDF文件URL（使用示例PDF）
// const [pdfUrl] = useState('/testPDF/sample.pdf'); // 使用包含真实文本层的PDF // const [pdfUrl] = useState('/testPDF/sample.pdf'); // 使用包含真实文本层的PDF
// const [pdfUrl] = useState('/api/pdf-proxy?path=documents/mz/行政处罚决定书/2025/11月13日/第71号--未在当地烟草专卖批发企业进货_02时58分36秒/第71号--未在当地烟草专卖批发企业进货.pdf'); // 使用项目中的示例PDF // const [pdfUrl] = useState('/api/pdf-proxy?path=documents/mz/行政处罚决定书/2025/11月13日/第71号--未在当地烟草专卖批发企业进货_02时58分36秒/第71号--未在当地烟草专卖批发企业进货.pdf'); // 使用项目中的示例PDF
const [pdfUrl] = useState('/api/pdf-proxy?path=documents/mz/行政处罚决定书/2025/11月22日/第35号--无烟草专卖品准运证运输烟草专卖品_15时15分24秒/第35号--无烟草专卖品准运证运输烟草专卖品.pdf') const [pdfUrl] = useState('/api/pdf-proxy?path=documents/mz/测试示范类型/2025/11月24日/第37号--涉嫌生产、销售伪劣产品罪_12时19分10秒/第37号--涉嫌生产、销售伪劣产品罪.pdf')
// PDF状态 // PDF状态
const [numPages, setNumPages] = useState<number | null>(null); const [numPages, setNumPages] = useState<number | null>(null);
@@ -227,87 +227,28 @@ export default function PdfDemo() {
// 获取Page容器（SVG实际渲染的坐标空间） // 获取Page容器（SVG实际渲染的坐标空间）
const pageContainer = canvas?.closest('.react-pdf__Page') as HTMLElement; const pageContainer = canvas?.closest('.react-pdf__Page') as HTMLElement;
if (canvas && pageContainer && pdfOriginalWidthPt) { if (canvas && pdfOriginalWidthPt) {
// Canvas 内部绘制尺寸（考虑了 devicePixelRatio）
const canvasInternalWidth = canvas.width;
const canvasInternalHeight = canvas.height;
// Canvas 显示尺寸（浏览器中实际占用的像素） // Canvas 显示尺寸（浏览器中实际占用的像素）
const canvasDisplayWidth = canvas.offsetWidth; const canvasDisplayWidth = canvas.offsetWidth;
const canvasDisplayHeight = canvas.offsetHeight; const canvasDisplayHeight = canvas.offsetHeight;
// Page容器尺寸（SVG高亮渲染的实际坐标空间） // 计算坐标缩放比例：Canvas显示尺寸 / PDF原始尺寸
const pageContainerWidth = pageContainer.offsetWidth; const autoScale = canvasDisplayWidth / pdfOriginalWidthPt;
const pageContainerHeight = pageContainer.offsetHeight;
// 尝试多种计算方式 console.log('📏 PDF尺寸信息:');
const scale1_canvasDisplay = canvasDisplayWidth / pdfOriginalWidthPt; console.log(' - PDF原始尺寸 (page.view):', pdfOriginalWidthPt, 'x', pdfOriginalHeightPt, 'pt');
const scale2_canvasInternal = canvasInternalWidth / pdfOriginalWidthPt; console.log(' - Canvas显示尺寸 (offsetWidth):', canvasDisplayWidth, 'x', canvasDisplayHeight, 'px');
const scale3_pageContainer = pageContainerWidth / pdfOriginalWidthPt; console.log(' - 用户缩放 (scale):', scale);
console.log(' - devicePixelRatio:', window.devicePixelRatio || 1);
// 尝试反向计算：如果OCR尺寸比渲染尺寸大（需要缩小） console.log('🎯 自动计算坐标缩放:', autoScale.toFixed(3), 'x');
const scale4_inverseCanvasInternal = canvasDisplayWidth / canvasInternalWidth; console.log(' 公式: Canvas显示宽度 / PDF原始宽度 =', canvasDisplayWidth, '/', pdfOriginalWidthPt);
const scale5_inversePage = canvasDisplayWidth / pageContainerWidth;
// 计算如果要达到 0.83 的缩放比例，OCR原始尺寸应该是多少
const expectedOcrWidth = canvasDisplayWidth / 0.83;
console.log('📏 尺寸信息汇总:');
console.log(' 1️⃣ PDF原始尺寸 (page.view):', pdfOriginalWidthPt, 'x', pdfOriginalHeightPt, 'pt');
console.log(' 2️⃣ Page容器尺寸:', pageContainerWidth, 'x', pageContainerHeight, 'px');
console.log(' 3️⃣ Canvas显示尺寸:', canvasDisplayWidth, 'x', canvasDisplayHeight, 'px');
console.log(' 4️⃣ Canvas内部尺寸:', canvasInternalWidth, 'x', canvasInternalHeight, 'px');
console.log(' 5️⃣ 用户缩放 (scale):', scale);
console.log(' 6️⃣ devicePixelRatio:', window.devicePixelRatio || 1);
console.log('');
console.log('🎯 各种计算方式:');
console.log(' 方案1️⃣: Canvas显示 / PDF原始 =', scale1_canvasDisplay.toFixed(3), 'x');
console.log(' 方案2️⃣: Canvas内部 / PDF原始 =', scale2_canvasInternal.toFixed(3), 'x');
console.log(' 方案3️⃣: Page容器 / PDF原始 =', scale3_pageContainer.toFixed(3), 'x');
console.log(' 方案4️⃣: Canvas显示 / Canvas内部 =', scale4_inverseCanvasInternal.toFixed(3), 'x ⬅ 可能是这个！');
console.log(' 方案5️⃣: Canvas显示 / Page容器 =', scale5_inversePage.toFixed(3), 'x');
console.log('');
console.log('🔍 目标值分析:');
console.log(' - 手动校准的正确值: 0.83');
console.log(' - 反推OCR图像尺寸:', expectedOcrWidth.toFixed(0), 'x', (canvasDisplayHeight / 0.83).toFixed(0), 'px');
console.log(' - 比较: ', expectedOcrWidth.toFixed(0), 'vs Canvas内部', canvasInternalWidth);
// 使用最接近0.83的方案
let autoScale = scale1_canvasDisplay;
let scaleMethod = '方案1 (Canvas显示/PDF原始)';
// 检查哪个方案最接近0.83
const diff1 = Math.abs(scale1_canvasDisplay - 0.83);
const diff2 = Math.abs(scale2_canvasInternal - 0.83);
const diff3 = Math.abs(scale3_pageContainer - 0.83);
const diff4 = Math.abs(scale4_inverseCanvasInternal - 0.83);
const diff5 = Math.abs(scale5_inversePage - 0.83);
const minDiff = Math.min(diff1, diff2, diff3, diff4, diff5);
if (minDiff === diff4) {
autoScale = scale4_inverseCanvasInternal;
scaleMethod = '方案4 (Canvas显示/Canvas内部)';
} else if (minDiff === diff5) {
autoScale = scale5_inversePage;
scaleMethod = '方案5 (Canvas显示/Page容器)';
} else if (minDiff === diff2) {
autoScale = scale2_canvasInternal;
scaleMethod = '方案2 (Canvas内部/PDF原始)';
} else if (minDiff === diff3) {
autoScale = scale3_pageContainer;
scaleMethod = '方案3 (Page容器/PDF原始)';
}
console.log('');
console.log('✅ 自动选择:', scaleMethod, '=', autoScale.toFixed(3), 'x (最接近0.83)');
// 保存原始宽度和自动计算的缩放比例 // 保存原始宽度和自动计算的缩放比例
setPdfOriginalWidth(pdfOriginalWidthPt); setPdfOriginalWidth(pdfOriginalWidthPt);
setCoordinateScale(autoScale); setCoordinateScale(autoScale);
setIsScaleAutoCalculated(true); setIsScaleAutoCalculated(true);
toastService.success(`自动校准完成: ${autoScale.toFixed(3)}x (${scaleMethod})`); toastService.success(`自动校准完成: ${autoScale.toFixed(3)}x`);
} else { } else {
console.warn('⚠️ 无法获取Canvas元素、Page容器或原始尺寸'); console.warn('⚠️ 无法获取Canvas元素、Page容器或原始尺寸');
console.log('调试信息:', { console.log('调试信息:', {
public/testWork/（最终版）智慧法务平台建设采购项目合同(1).docx
BIN
View File
Binary file not shown.
public/testWork/（最终版）智慧法务平台建设采购项目合同(2).docx
BIN
View File
Binary file not shown.