Upload 2 files
Browse files- app.js +1403 -0
- index.html +0 -19
app.js
ADDED
@@ -0,0 +1,1403 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
// 全局变量
|
2 |
+
let pdfFile = null;
|
3 |
+
let markdownContent = '';
|
4 |
+
let translationContent = '';
|
5 |
+
let imagesData = [];
|
6 |
+
|
7 |
+
// DOM 元素
|
8 |
+
const mistralApiKeyInput = document.getElementById('mistralApiKey');
|
9 |
+
const toggleMistralKeyBtn = document.getElementById('toggleMistralKey');
|
10 |
+
const rememberMistralKeyCheckbox = document.getElementById('rememberMistralKey');
|
11 |
+
const translationApiKeyInput = document.getElementById('translationApiKey');
|
12 |
+
const toggleTranslationKeyBtn = document.getElementById('toggleTranslationKey');
|
13 |
+
const rememberTranslationKeyCheckbox = document.getElementById('rememberTranslationKey');
|
14 |
+
|
15 |
+
const translationModelSelect = document.getElementById('translationModel');
|
16 |
+
const customModelSettings = document.getElementById('customModelSettings');
|
17 |
+
|
18 |
+
// 高级设置相关
|
19 |
+
const advancedSettingsToggle = document.getElementById('advancedSettingsToggle');
|
20 |
+
const advancedSettings = document.getElementById('advancedSettings');
|
21 |
+
const advancedSettingsIcon = document.getElementById('advancedSettingsIcon');
|
22 |
+
const maxTokensPerChunk = document.getElementById('maxTokensPerChunk');
|
23 |
+
const maxTokensPerChunkValue = document.getElementById('maxTokensPerChunkValue');
|
24 |
+
|
25 |
+
// 文件上传相关
|
26 |
+
const dropZone = document.getElementById('dropZone');
|
27 |
+
const pdfFileInput = document.getElementById('pdfFileInput');
|
28 |
+
const browseFilesBtn = document.getElementById('browseFilesBtn');
|
29 |
+
const fileInfo = document.getElementById('fileInfo');
|
30 |
+
const fileName = document.getElementById('fileName');
|
31 |
+
const fileSize = document.getElementById('fileSize');
|
32 |
+
const removeFileBtn = document.getElementById('removeFileBtn');
|
33 |
+
|
34 |
+
// 翻译相关
|
35 |
+
const targetLanguage = document.getElementById('targetLanguage');
|
36 |
+
|
37 |
+
// 按钮
|
38 |
+
const processBtn = document.getElementById('processBtn');
|
39 |
+
const downloadMarkdownBtn = document.getElementById('downloadMarkdownBtn');
|
40 |
+
const downloadTranslationBtn = document.getElementById('downloadTranslationBtn');
|
41 |
+
|
42 |
+
// 结果展示
|
43 |
+
const resultsSection = document.getElementById('resultsSection');
|
44 |
+
const markdownPreview = document.getElementById('markdownPreview');
|
45 |
+
const translationPreview = document.getElementById('translationPreview');
|
46 |
+
const translationResultCard = document.getElementById('translationResultCard');
|
47 |
+
|
48 |
+
// 进度相关
|
49 |
+
const progressSection = document.getElementById('progressSection');
|
50 |
+
const progressStep = document.getElementById('progressStep');
|
51 |
+
const progressPercentage = document.getElementById('progressPercentage');
|
52 |
+
const progressBar = document.getElementById('progressBar');
|
53 |
+
const progressLog = document.getElementById('progressLog');
|
54 |
+
|
55 |
+
document.addEventListener('DOMContentLoaded', () => {
|
56 |
+
// 初始化 - 从本地存储加载 API Key
|
57 |
+
if (localStorage.getItem('mistralApiKey')) {
|
58 |
+
mistralApiKeyInput.value = localStorage.getItem('mistralApiKey');
|
59 |
+
rememberMistralKeyCheckbox.checked = true;
|
60 |
+
}
|
61 |
+
|
62 |
+
if (localStorage.getItem('translationApiKey')) {
|
63 |
+
translationApiKeyInput.value = localStorage.getItem('translationApiKey');
|
64 |
+
rememberTranslationKeyCheckbox.checked = true;
|
65 |
+
}
|
66 |
+
|
67 |
+
// 加载设置
|
68 |
+
loadSettings();
|
69 |
+
|
70 |
+
// API Key 显示/隐藏切换
|
71 |
+
toggleMistralKeyBtn.addEventListener('click', () => {
|
72 |
+
if (mistralApiKeyInput.type === 'password') {
|
73 |
+
mistralApiKeyInput.type = 'text';
|
74 |
+
toggleMistralKeyBtn.innerHTML = '<iconify-icon icon="carbon:view-off" width="20"></iconify-icon>';
|
75 |
+
} else {
|
76 |
+
mistralApiKeyInput.type = 'password';
|
77 |
+
toggleMistralKeyBtn.innerHTML = '<iconify-icon icon="carbon:view" width="20"></iconify-icon>';
|
78 |
+
}
|
79 |
+
});
|
80 |
+
|
81 |
+
toggleTranslationKeyBtn.addEventListener('click', () => {
|
82 |
+
if (translationApiKeyInput.type === 'password') {
|
83 |
+
translationApiKeyInput.type = 'text';
|
84 |
+
toggleTranslationKeyBtn.innerHTML = '<iconify-icon icon="carbon:view-off" width="20"></iconify-icon>';
|
85 |
+
} else {
|
86 |
+
translationApiKeyInput.type = 'password';
|
87 |
+
toggleTranslationKeyBtn.innerHTML = '<iconify-icon icon="carbon:view" width="20"></iconify-icon>';
|
88 |
+
}
|
89 |
+
});
|
90 |
+
|
91 |
+
// API Key 记住选项
|
92 |
+
rememberMistralKeyCheckbox.addEventListener('change', () => {
|
93 |
+
if (rememberMistralKeyCheckbox.checked) {
|
94 |
+
localStorage.setItem('mistralApiKey', mistralApiKeyInput.value);
|
95 |
+
} else {
|
96 |
+
localStorage.removeItem('mistralApiKey');
|
97 |
+
}
|
98 |
+
});
|
99 |
+
|
100 |
+
rememberTranslationKeyCheckbox.addEventListener('change', () => {
|
101 |
+
if (rememberTranslationKeyCheckbox.checked) {
|
102 |
+
localStorage.setItem('translationApiKey', translationApiKeyInput.value);
|
103 |
+
} else {
|
104 |
+
localStorage.removeItem('translationApiKey');
|
105 |
+
}
|
106 |
+
});
|
107 |
+
|
108 |
+
mistralApiKeyInput.addEventListener('input', () => {
|
109 |
+
if (rememberMistralKeyCheckbox.checked) {
|
110 |
+
localStorage.setItem('mistralApiKey', mistralApiKeyInput.value);
|
111 |
+
}
|
112 |
+
});
|
113 |
+
|
114 |
+
translationApiKeyInput.addEventListener('input', () => {
|
115 |
+
if (rememberTranslationKeyCheckbox.checked) {
|
116 |
+
localStorage.setItem('translationApiKey', translationApiKeyInput.value);
|
117 |
+
}
|
118 |
+
});
|
119 |
+
|
120 |
+
// PDF 文件拖放上传
|
121 |
+
dropZone.addEventListener('dragover', (e) => {
|
122 |
+
e.preventDefault();
|
123 |
+
dropZone.classList.add('border-blue-500', 'bg-blue-50');
|
124 |
+
});
|
125 |
+
|
126 |
+
dropZone.addEventListener('dragleave', () => {
|
127 |
+
dropZone.classList.remove('border-blue-500', 'bg-blue-50');
|
128 |
+
});
|
129 |
+
|
130 |
+
dropZone.addEventListener('drop', (e) => {
|
131 |
+
e.preventDefault();
|
132 |
+
dropZone.classList.remove('border-blue-500', 'bg-blue-50');
|
133 |
+
|
134 |
+
if (e.dataTransfer.files.length > 0 && e.dataTransfer.files[0].type === 'application/pdf') {
|
135 |
+
handleFileSelection(e.dataTransfer.files[0]);
|
136 |
+
} else {
|
137 |
+
showNotification('请上传PDF文件', 'error');
|
138 |
+
}
|
139 |
+
});
|
140 |
+
|
141 |
+
// 浏览文件按钮
|
142 |
+
browseFilesBtn.addEventListener('click', () => {
|
143 |
+
pdfFileInput.click();
|
144 |
+
});
|
145 |
+
|
146 |
+
// 文件选择处理
|
147 |
+
pdfFileInput.addEventListener('change', (e) => {
|
148 |
+
if (e.target.files.length > 0) {
|
149 |
+
handleFileSelection(e.target.files[0]);
|
150 |
+
}
|
151 |
+
});
|
152 |
+
|
153 |
+
// 移除文件
|
154 |
+
removeFileBtn.addEventListener('click', () => {
|
155 |
+
pdfFile = null;
|
156 |
+
fileInfo.classList.add('hidden');
|
157 |
+
pdfFileInput.value = '';
|
158 |
+
updateProcessButtonState();
|
159 |
+
});
|
160 |
+
|
161 |
+
// 处理按钮
|
162 |
+
processBtn.addEventListener('click', async () => {
|
163 |
+
try {
|
164 |
+
const mistralKey = mistralApiKeyInput.value.trim();
|
165 |
+
|
166 |
+
if (!mistralKey) {
|
167 |
+
showNotification('请输入Mistral API Key', 'error');
|
168 |
+
return;
|
169 |
+
}
|
170 |
+
|
171 |
+
if (!pdfFile) {
|
172 |
+
showNotification('请上传PDF文件', 'error');
|
173 |
+
return;
|
174 |
+
}
|
175 |
+
|
176 |
+
// 开始处理
|
177 |
+
processBtn.disabled = true;
|
178 |
+
showProgressSection();
|
179 |
+
updateProgress('开始处理...', 5);
|
180 |
+
addProgressLog('开始OCR处理...');
|
181 |
+
|
182 |
+
try {
|
183 |
+
// 执行OCR处理
|
184 |
+
await processPdfWithMistral(mistralKey);
|
185 |
+
|
186 |
+
// 如果选择了翻译,则执行翻译
|
187 |
+
if (translationModelSelect.value !== 'none') {
|
188 |
+
const translationKey = translationApiKeyInput.value.trim();
|
189 |
+
if (translationModelSelect.value !== 'none' && !translationKey) {
|
190 |
+
showNotification('请输入翻译API Key', 'error');
|
191 |
+
updateProgress('翻译需要API Key', 100);
|
192 |
+
addProgressLog('错误: 缺少翻译API Key');
|
193 |
+
processBtn.disabled = false;
|
194 |
+
return;
|
195 |
+
}
|
196 |
+
updateProgress('开始翻译...', 60);
|
197 |
+
addProgressLog(`使用${translationModelSelect.value}模型进行翻译...`);
|
198 |
+
|
199 |
+
// 获取文档大小估计
|
200 |
+
const estimatedTokens = estimateTokenCount(markdownContent);
|
201 |
+
const tokenLimit = 12000; // 设置一个安全的token限制
|
202 |
+
|
203 |
+
if (estimatedTokens > tokenLimit) {
|
204 |
+
// 使用分段翻译
|
205 |
+
addProgressLog(`文档较大(~${Math.round(estimatedTokens/1000)}K tokens),将进行分段翻译`);
|
206 |
+
translationContent = await translateLongDocument(markdownContent, targetLanguage.value, translationModelSelect.value, translationKey);
|
207 |
+
} else {
|
208 |
+
// 直接翻译
|
209 |
+
addProgressLog(`文档较小(~${Math.round(estimatedTokens/1000)}K tokens),不分段直接翻译`);
|
210 |
+
translationContent = await translateMarkdown(markdownContent, targetLanguage.value, translationModelSelect.value, translationKey);
|
211 |
+
}
|
212 |
+
}
|
213 |
+
|
214 |
+
// 显示结果
|
215 |
+
updateProgress('处理完成!', 100);
|
216 |
+
addProgressLog('全部处理完成!');
|
217 |
+
showResultsSection();
|
218 |
+
} catch (error) {
|
219 |
+
console.error('处理错误:', error);
|
220 |
+
showNotification('处理过程中出错: ' + error.message, 'error');
|
221 |
+
addProgressLog('错误: ' + error.message);
|
222 |
+
updateProgress('处理失败', 100);
|
223 |
+
} finally {
|
224 |
+
processBtn.disabled = false;
|
225 |
+
}
|
226 |
+
} catch (error) {
|
227 |
+
console.error('处理错误:', error);
|
228 |
+
showNotification('处理过程中出错: ' + error.message, 'error');
|
229 |
+
addProgressLog('错误: ' + error.message);
|
230 |
+
updateProgress('处理失败', 100);
|
231 |
+
processBtn.disabled = false;
|
232 |
+
}
|
233 |
+
});
|
234 |
+
|
235 |
+
// 下载按钮
|
236 |
+
downloadMarkdownBtn.addEventListener('click', () => {
|
237 |
+
if (markdownContent) {
|
238 |
+
downloadMarkdownWithImages();
|
239 |
+
}
|
240 |
+
});
|
241 |
+
|
242 |
+
downloadTranslationBtn.addEventListener('click', () => {
|
243 |
+
if (translationContent) {
|
244 |
+
//downloadText(translationContent, 'translation.md');
|
245 |
+
downloadTranslationWithImages();
|
246 |
+
}
|
247 |
+
});
|
248 |
+
|
249 |
+
// 翻译模型变更
|
250 |
+
translationModelSelect.addEventListener('change', function() {
|
251 |
+
if (this.value === 'custom') {
|
252 |
+
customModelSettings.classList.remove('hidden');
|
253 |
+
} else {
|
254 |
+
customModelSettings.classList.add('hidden');
|
255 |
+
}
|
256 |
+
|
257 |
+
// 更新翻译界面可见性
|
258 |
+
updateTranslationUIVisibility();
|
259 |
+
|
260 |
+
// 保存设置
|
261 |
+
saveSettings();
|
262 |
+
});
|
263 |
+
|
264 |
+
// 高级设置开关
|
265 |
+
advancedSettingsToggle.addEventListener('click', function() {
|
266 |
+
advancedSettings.classList.toggle('hidden');
|
267 |
+
|
268 |
+
// 更新图标方向
|
269 |
+
if (advancedSettings.classList.contains('hidden')) {
|
270 |
+
advancedSettingsIcon.setAttribute('icon', 'carbon:chevron-down');
|
271 |
+
} else {
|
272 |
+
advancedSettingsIcon.setAttribute('icon', 'carbon:chevron-up');
|
273 |
+
}
|
274 |
+
|
275 |
+
// 保存设置
|
276 |
+
saveSettings();
|
277 |
+
});
|
278 |
+
|
279 |
+
// 最大Token数设置滑动条
|
280 |
+
maxTokensPerChunk.addEventListener('input', function() {
|
281 |
+
maxTokensPerChunkValue.textContent = this.value;
|
282 |
+
saveSettings();
|
283 |
+
});
|
284 |
+
|
285 |
+
// 为自定义模型设置添加变更事件监听器
|
286 |
+
const customModelInputs = [
|
287 |
+
document.getElementById('customModelName'),
|
288 |
+
document.getElementById('customApiEndpoint'),
|
289 |
+
document.getElementById('customModelId'),
|
290 |
+
document.getElementById('customRequestFormat')
|
291 |
+
];
|
292 |
+
|
293 |
+
customModelInputs.forEach(input => {
|
294 |
+
input.addEventListener('change', function() {
|
295 |
+
saveSettings();
|
296 |
+
});
|
297 |
+
// 同时监听输入事件,实时保存
|
298 |
+
input.addEventListener('input', function() {
|
299 |
+
saveSettings();
|
300 |
+
});
|
301 |
+
});
|
302 |
+
|
303 |
+
// 初始化 UI 状态
|
304 |
+
updateProcessButtonState();
|
305 |
+
updateTranslationUIVisibility();
|
306 |
+
});
|
307 |
+
|
308 |
+
// 辅助函数
|
309 |
+
function handleFileSelection(file) {
|
310 |
+
pdfFile = file;
|
311 |
+
document.getElementById('fileName').textContent = file.name;
|
312 |
+
document.getElementById('fileSize').textContent = formatFileSize(file.size);
|
313 |
+
document.getElementById('fileInfo').classList.remove('hidden');
|
314 |
+
updateProcessButtonState();
|
315 |
+
}
|
316 |
+
|
317 |
+
function updateProcessButtonState() {
|
318 |
+
const mistralKey = document.getElementById('mistralApiKey').value.trim();
|
319 |
+
const processBtn = document.getElementById('processBtn');
|
320 |
+
processBtn.disabled = !pdfFile || !mistralKey;
|
321 |
+
}
|
322 |
+
|
323 |
+
function updateTranslationUIVisibility() {
|
324 |
+
const translationModelValue = translationModelSelect.value;
|
325 |
+
|
326 |
+
// 如果选择了翻译模型,显示API Key输入框和提示
|
327 |
+
const translationApiKeySection = document.querySelector('#translationApiKey').closest('div').parentNode;
|
328 |
+
if (translationModelValue !== 'none') {
|
329 |
+
translationApiKeySection.style.display = 'block';
|
330 |
+
} else {
|
331 |
+
translationApiKeySection.style.display = 'none';
|
332 |
+
}
|
333 |
+
}
|
334 |
+
|
335 |
+
function showResultsSection() {
|
336 |
+
document.getElementById('progressSection').classList.add('hidden');
|
337 |
+
document.getElementById('resultsSection').classList.remove('hidden');
|
338 |
+
|
339 |
+
// 显示Markdown内容
|
340 |
+
document.getElementById('markdownPreview').textContent = markdownContent.substring(0, 500) + '...';
|
341 |
+
|
342 |
+
// 显示翻译内容(如果有)
|
343 |
+
if (translationContent) {
|
344 |
+
document.getElementById('translationPreview').textContent = translationContent.substring(0, 500) + '...';
|
345 |
+
document.getElementById('translationResultCard').classList.remove('hidden');
|
346 |
+
} else {
|
347 |
+
document.getElementById('translationResultCard').classList.add('hidden');
|
348 |
+
}
|
349 |
+
|
350 |
+
window.scrollTo({
|
351 |
+
top: document.getElementById('resultsSection').offsetTop - 20,
|
352 |
+
behavior: 'smooth'
|
353 |
+
});
|
354 |
+
}
|
355 |
+
|
356 |
+
function showProgressSection() {
|
357 |
+
document.getElementById('resultsSection').classList.add('hidden');
|
358 |
+
document.getElementById('progressSection').classList.remove('hidden');
|
359 |
+
document.getElementById('progressLog').innerHTML = '';
|
360 |
+
updateProgress('初始化...', 0);
|
361 |
+
|
362 |
+
window.scrollTo({
|
363 |
+
top: document.getElementById('progressSection').offsetTop - 20,
|
364 |
+
behavior: 'smooth'
|
365 |
+
});
|
366 |
+
}
|
367 |
+
|
368 |
+
function updateProgress(stepText, percentage) {
|
369 |
+
document.getElementById('progressStep').textContent = stepText;
|
370 |
+
document.getElementById('progressPercentage').textContent = `${percentage}%`;
|
371 |
+
document.getElementById('progressBar').style.width = `${percentage}%`;
|
372 |
+
}
|
373 |
+
|
374 |
+
function addProgressLog(text) {
|
375 |
+
const logElement = document.getElementById('progressLog');
|
376 |
+
const timestamp = new Date().toLocaleTimeString();
|
377 |
+
logElement.innerHTML += `<div>[${timestamp}] ${text}</div>`;
|
378 |
+
logElement.scrollTop = logElement.scrollHeight;
|
379 |
+
}
|
380 |
+
|
381 |
+
function showNotification(message, type = 'info') {
|
382 |
+
// 创建通知元素
|
383 |
+
const notification = document.createElement('div');
|
384 |
+
|
385 |
+
// ��据类型设置样式和图标
|
386 |
+
let bgColor, iconName, textColor;
|
387 |
+
switch (type) {
|
388 |
+
case 'success':
|
389 |
+
bgColor = 'bg-green-50 border-green-500';
|
390 |
+
textColor = 'text-green-800';
|
391 |
+
iconName = 'carbon:checkmark-filled';
|
392 |
+
break;
|
393 |
+
case 'error':
|
394 |
+
bgColor = 'bg-red-50 border-red-500';
|
395 |
+
textColor = 'text-red-800';
|
396 |
+
iconName = 'carbon:error-filled';
|
397 |
+
break;
|
398 |
+
case 'warning':
|
399 |
+
bgColor = 'bg-yellow-50 border-yellow-500';
|
400 |
+
textColor = 'text-yellow-800';
|
401 |
+
iconName = 'carbon:warning-filled';
|
402 |
+
break;
|
403 |
+
default: // info
|
404 |
+
bgColor = 'bg-blue-50 border-blue-500';
|
405 |
+
textColor = 'text-blue-800';
|
406 |
+
iconName = 'carbon:information-filled';
|
407 |
+
}
|
408 |
+
|
409 |
+
// 设置通知样式
|
410 |
+
notification.className = `flex items-center p-4 mb-4 max-w-md border-l-4 ${bgColor} ${textColor} shadow-md rounded-r-lg transform transition-all duration-300 ease-in-out`;
|
411 |
+
notification.style.opacity = '0';
|
412 |
+
notification.style.transform = 'translateX(100%)';
|
413 |
+
|
414 |
+
// 设置通知内容
|
415 |
+
notification.innerHTML = `
|
416 |
+
<iconify-icon icon="${iconName}" class="flex-shrink-0 w-5 h-5 mr-2"></iconify-icon>
|
417 |
+
<div class="ml-3 text-sm font-medium flex-grow">${message}</div>
|
418 |
+
<button type="button" class="ml-auto -mx-1.5 -my-1.5 rounded-lg p-1.5 inline-flex h-8 w-8 hover:bg-gray-200 focus:ring-2 focus:ring-gray-400">
|
419 |
+
<iconify-icon icon="carbon:close" class="w-5 h-5"></iconify-icon>
|
420 |
+
</button>
|
421 |
+
`;
|
422 |
+
|
423 |
+
// 获取通知容器
|
424 |
+
const container = document.getElementById('notification-container');
|
425 |
+
container.appendChild(notification);
|
426 |
+
|
427 |
+
// 显示动画
|
428 |
+
setTimeout(() => {
|
429 |
+
notification.style.opacity = '1';
|
430 |
+
notification.style.transform = 'translateX(0)';
|
431 |
+
}, 10);
|
432 |
+
|
433 |
+
// 添加关闭按钮点击事件
|
434 |
+
const closeButton = notification.querySelector('button');
|
435 |
+
closeButton.addEventListener('click', () => {
|
436 |
+
closeNotification(notification);
|
437 |
+
});
|
438 |
+
|
439 |
+
// 自动关闭(5秒后)
|
440 |
+
const timeout = setTimeout(() => {
|
441 |
+
closeNotification(notification);
|
442 |
+
}, 5000);
|
443 |
+
|
444 |
+
// 保存timeout引用,以便可以在手动关闭时清除
|
445 |
+
notification.dataset.timeout = timeout;
|
446 |
+
|
447 |
+
// 返回通知元素,以便可以手动关闭
|
448 |
+
return notification;
|
449 |
+
}
|
450 |
+
|
451 |
+
// 关闭通知的辅助函数
|
452 |
+
function closeNotification(notification) {
|
453 |
+
// 清除自动关闭的timeout
|
454 |
+
clearTimeout(notification.dataset.timeout);
|
455 |
+
|
456 |
+
// 淡出动画
|
457 |
+
notification.style.opacity = '0';
|
458 |
+
notification.style.transform = 'translateX(100%)';
|
459 |
+
|
460 |
+
// 动画完成后移除元素
|
461 |
+
setTimeout(() => {
|
462 |
+
if (notification.parentNode) {
|
463 |
+
notification.parentNode.removeChild(notification);
|
464 |
+
}
|
465 |
+
}, 300);
|
466 |
+
}
|
467 |
+
|
468 |
+
function formatFileSize(bytes) {
|
469 |
+
if (bytes < 1024) return bytes + ' B';
|
470 |
+
else if (bytes < 1048576) return (bytes / 1024).toFixed(2) + ' KB';
|
471 |
+
else return (bytes / 1048576).toFixed(2) + ' MB';
|
472 |
+
}
|
473 |
+
|
474 |
+
// Mistral OCR 处理
|
475 |
+
async function processPdfWithMistral(apiKey) {
|
476 |
+
try {
|
477 |
+
addProgressLog('准备PDF文件...');
|
478 |
+
updateProgress('PDF处理准备中', 10);
|
479 |
+
|
480 |
+
// 检查API密钥长度
|
481 |
+
if (apiKey.length < 30) {
|
482 |
+
throw new Error('Mistral API密钥格式可能不正确,请检查');
|
483 |
+
}
|
484 |
+
|
485 |
+
// 从上传文件到获取OCR结果的完整流程
|
486 |
+
const formData = new FormData();
|
487 |
+
// 关键点:文件上传字段名必须是file
|
488 |
+
formData.append('file', pdfFile);
|
489 |
+
formData.append('purpose', 'ocr');
|
490 |
+
|
491 |
+
addProgressLog('准备上传PDF文件...');
|
492 |
+
updateProgress('上传文件中...', 20);
|
493 |
+
addProgressLog('开始上传到Mistral...');
|
494 |
+
|
495 |
+
console.log('开始上传文件,文件名:', pdfFile.name, '文件大小:', pdfFile.size);
|
496 |
+
|
497 |
+
// 尝试上传文件
|
498 |
+
let response;
|
499 |
+
try {
|
500 |
+
response = await fetch('https://api.mistral.ai/v1/files', {
|
501 |
+
method: 'POST',
|
502 |
+
headers: {
|
503 |
+
'Authorization': `Bearer ${apiKey}`
|
504 |
+
// 不要设置Content-Type,让浏览器自动设置multipart/form-data和boundary
|
505 |
+
},
|
506 |
+
body: formData
|
507 |
+
});
|
508 |
+
} catch (uploadError) {
|
509 |
+
console.error('上传错误详情:', uploadError);
|
510 |
+
addProgressLog(`网络错误: ${uploadError.message || '未知网络错误'}`);
|
511 |
+
throw new Error(`文件上传失败,网络错误: ${uploadError.message || '未知网络错误'}`);
|
512 |
+
}
|
513 |
+
|
514 |
+
if (!response.ok) {
|
515 |
+
let errorInfo = '未知错误';
|
516 |
+
try {
|
517 |
+
const responseText = await response.text();
|
518 |
+
console.error('上传失败原始响应:', responseText);
|
519 |
+
try {
|
520 |
+
const jsonError = JSON.parse(responseText);
|
521 |
+
errorInfo = jsonError.error?.message || jsonError.message || jsonError.detail || responseText;
|
522 |
+
} catch (e) {
|
523 |
+
errorInfo = responseText || `HTTP错误: ${response.status} ${response.statusText}`;
|
524 |
+
}
|
525 |
+
} catch (e) {
|
526 |
+
errorInfo = `HTTP错误: ${response.status} ${response.statusText}`;
|
527 |
+
}
|
528 |
+
|
529 |
+
addProgressLog(`上传失败: ${response.status} - ${errorInfo}`);
|
530 |
+
|
531 |
+
if (response.status === 401) {
|
532 |
+
throw new Error('API密钥无效或未授权,请检查您的Mistral API密钥');
|
533 |
+
} else {
|
534 |
+
throw new Error(`文件上传失败 (${response.status}): ${errorInfo}`);
|
535 |
+
}
|
536 |
+
}
|
537 |
+
|
538 |
+
let fileData;
|
539 |
+
try {
|
540 |
+
fileData = await response.json();
|
541 |
+
console.log('文件上传响应:', JSON.stringify(fileData));
|
542 |
+
} catch (e) {
|
543 |
+
console.error('解析文件数据错误:', e);
|
544 |
+
throw new Error('无法解析文件上传响应数据');
|
545 |
+
}
|
546 |
+
|
547 |
+
if (!fileData || !fileData.id) {
|
548 |
+
console.error('文件数据无效:', fileData);
|
549 |
+
throw new Error('上传成功但未返回有效的文件ID');
|
550 |
+
}
|
551 |
+
|
552 |
+
const fileId = fileData.id;
|
553 |
+
addProgressLog(`文件上传成功,ID: ${fileId}`);
|
554 |
+
updateProgress('获取文件访问权限...', 30);
|
555 |
+
|
556 |
+
// 确保fileId是有效的字符串
|
557 |
+
if (typeof fileId !== 'string' || fileId.trim() === '') {
|
558 |
+
throw new Error('文件ID无效,无法继续处理');
|
559 |
+
}
|
560 |
+
|
561 |
+
// 等待一下确保文件已经处理完成
|
562 |
+
await new Promise(resolve => setTimeout(resolve, 1000));
|
563 |
+
|
564 |
+
// 获取签名URL - 使用文档中的确切格式
|
565 |
+
try {
|
566 |
+
// 使用/url端点并传递expiry参数
|
567 |
+
const urlEndpoint = `https://api.mistral.ai/v1/files/${fileId}/url?expiry=24`;
|
568 |
+
console.log('请求签名URL:', urlEndpoint);
|
569 |
+
|
570 |
+
response = await fetch(urlEndpoint, {
|
571 |
+
method: 'GET',
|
572 |
+
headers: {
|
573 |
+
'Authorization': `Bearer ${apiKey}`,
|
574 |
+
'Accept': 'application/json'
|
575 |
+
}
|
576 |
+
});
|
577 |
+
} catch (urlError) {
|
578 |
+
console.error('获取URL错误详情:', urlError);
|
579 |
+
addProgressLog(`获取URL错误: ${urlError.message || '未知网络错误'}`);
|
580 |
+
throw new Error(`获取签名URL失败,网络错误: ${urlError.message || '未知网络错误'}`);
|
581 |
+
}
|
582 |
+
|
583 |
+
if (!response.ok) {
|
584 |
+
let errorInfo = '未知错误';
|
585 |
+
try {
|
586 |
+
const responseText = await response.text();
|
587 |
+
console.error('获取URL失败原始响应:', responseText);
|
588 |
+
try {
|
589 |
+
const jsonError = JSON.parse(responseText);
|
590 |
+
errorInfo = jsonError.error?.message || jsonError.message || jsonError.detail || responseText;
|
591 |
+
} catch (e) {
|
592 |
+
errorInfo = responseText || `HTTP错误: ${response.status} ${response.statusText}`;
|
593 |
+
}
|
594 |
+
} catch (e) {
|
595 |
+
errorInfo = `HTTP错误: ${response.status} ${response.statusText}`;
|
596 |
+
}
|
597 |
+
|
598 |
+
addProgressLog(`获取签名URL失败: ${response.status} - ${errorInfo}`);
|
599 |
+
throw new Error(`获取签名URL失败 (${response.status}): ${errorInfo}`);
|
600 |
+
}
|
601 |
+
|
602 |
+
let urlData;
|
603 |
+
try {
|
604 |
+
urlData = await response.json();
|
605 |
+
console.log('签名URL响应:', JSON.stringify(urlData));
|
606 |
+
} catch (e) {
|
607 |
+
console.error('解析URL数据错误:', e);
|
608 |
+
throw new Error('无法解析签名URL响应数据');
|
609 |
+
}
|
610 |
+
|
611 |
+
if (!urlData || !urlData.url) {
|
612 |
+
console.error('URL数据无效:', urlData);
|
613 |
+
addProgressLog('返回的URL格式不正确');
|
614 |
+
throw new Error('获取的签名URL格式不正确');
|
615 |
+
}
|
616 |
+
|
617 |
+
const signedUrl = urlData.url;
|
618 |
+
addProgressLog('成功获取文件访问URL');
|
619 |
+
updateProgress('开始OCR处理...', 40);
|
620 |
+
|
621 |
+
// 进行OCR处理 - 请求体需要匹配最新文档
|
622 |
+
try {
|
623 |
+
response = await fetch('https://api.mistral.ai/v1/ocr', {
|
624 |
+
method: 'POST',
|
625 |
+
headers: {
|
626 |
+
'Authorization': `Bearer ${apiKey}`,
|
627 |
+
'Content-Type': 'application/json',
|
628 |
+
'Accept': 'application/json'
|
629 |
+
},
|
630 |
+
body: JSON.stringify({
|
631 |
+
// 完全匹配文档示例
|
632 |
+
model: 'mistral-ocr-latest',
|
633 |
+
document: {
|
634 |
+
type: "document_url",
|
635 |
+
document_url: signedUrl
|
636 |
+
},
|
637 |
+
include_image_base64: true
|
638 |
+
})
|
639 |
+
});
|
640 |
+
} catch (ocrError) {
|
641 |
+
console.error('OCR错误详情:', ocrError);
|
642 |
+
addProgressLog(`OCR处理网络错误: ${ocrError.message || '未知网络错误'}`);
|
643 |
+
throw new Error(`OCR处理失败,网络错误: ${ocrError.message || '未知网络错误'}`);
|
644 |
+
}
|
645 |
+
|
646 |
+
if (!response.ok) {
|
647 |
+
let errorInfo = '未知错误';
|
648 |
+
try {
|
649 |
+
const responseText = await response.text();
|
650 |
+
console.error('OCR处理失败原始响应:', responseText);
|
651 |
+
try {
|
652 |
+
const jsonError = JSON.parse(responseText);
|
653 |
+
errorInfo = jsonError.error?.message || jsonError.message || jsonError.detail || responseText;
|
654 |
+
} catch (e) {
|
655 |
+
errorInfo = responseText || `HTTP错误: ${response.status} ${response.statusText}`;
|
656 |
+
}
|
657 |
+
} catch (e) {
|
658 |
+
errorInfo = `HTTP错误: ${response.status} ${response.statusText}`;
|
659 |
+
}
|
660 |
+
|
661 |
+
addProgressLog(`OCR处理失败: ${response.status} - ${errorInfo}`);
|
662 |
+
throw new Error(`OCR处理失败 (${response.status}): ${errorInfo}`);
|
663 |
+
}
|
664 |
+
|
665 |
+
let ocrData;
|
666 |
+
try {
|
667 |
+
ocrData = await response.json();
|
668 |
+
console.log('OCR处理成功,返回数据类型:', typeof ocrData);
|
669 |
+
} catch (e) {
|
670 |
+
console.error('解析OCR数据错误:', e);
|
671 |
+
throw new Error('无法解析OCR处理响应数据');
|
672 |
+
}
|
673 |
+
|
674 |
+
if (!ocrData || !ocrData.pages) {
|
675 |
+
console.error('OCR数据无效:', ocrData);
|
676 |
+
throw new Error('OCR处理成功但返回的数据格式不正确');
|
677 |
+
}
|
678 |
+
|
679 |
+
addProgressLog('OCR处理完成,开始生成Markdown');
|
680 |
+
updateProgress('生成Markdown...', 50);
|
681 |
+
|
682 |
+
// 处理OCR结果
|
683 |
+
await processOcrResults(ocrData);
|
684 |
+
addProgressLog('Markdown生成完成');
|
685 |
+
|
686 |
+
return true;
|
687 |
+
} catch (error) {
|
688 |
+
console.error('Mistral OCR处理错误:', error);
|
689 |
+
addProgressLog(`处理失败: ${error.message || '未知错误'}`);
|
690 |
+
throw error;
|
691 |
+
}
|
692 |
+
}
|
693 |
+
|
694 |
+
// 处理OCR结果
|
695 |
+
async function processOcrResults(ocrResponse) {
|
696 |
+
try {
|
697 |
+
markdownContent = '';
|
698 |
+
imagesData = [];
|
699 |
+
|
700 |
+
// 处理每一页
|
701 |
+
for (const page of ocrResponse.pages) {
|
702 |
+
// 处理图片
|
703 |
+
const pageImages = {};
|
704 |
+
|
705 |
+
for (const img of page.images) {
|
706 |
+
const imgId = img.id;
|
707 |
+
const imgData = img.image_base64;
|
708 |
+
imagesData.push({
|
709 |
+
id: imgId,
|
710 |
+
data: imgData
|
711 |
+
});
|
712 |
+
pageImages[imgId] = `images/${imgId}.png`;
|
713 |
+
}
|
714 |
+
|
715 |
+
// 替换Markdown中的图片引用
|
716 |
+
let pageMarkdown = page.markdown;
|
717 |
+
for (const [imgName, imgPath] of Object.entries(pageImages)) {
|
718 |
+
pageMarkdown = pageMarkdown.replace(
|
719 |
+
new RegExp(`!\\[${imgName}\\]\\(${imgName}\\)`, 'g'),
|
720 |
+
``
|
721 |
+
);
|
722 |
+
}
|
723 |
+
|
724 |
+
markdownContent += pageMarkdown + '\n\n';
|
725 |
+
}
|
726 |
+
|
727 |
+
return true;
|
728 |
+
} catch (error) {
|
729 |
+
console.error('处理OCR结果错误:', error);
|
730 |
+
throw new Error('处理OCR结果失败: ' + error.message);
|
731 |
+
}
|
732 |
+
}
|
733 |
+
|
734 |
+
// 翻译Markdown
|
735 |
+
async function translateMarkdown(markdownText, targetLang, model, apiKey) {
|
736 |
+
try {
|
737 |
+
// 允许使用全局变量或传入参数
|
738 |
+
const content = markdownText || markdownContent;
|
739 |
+
const lang = targetLang || document.getElementById('targetLanguage').value;
|
740 |
+
const selectedModel = model || document.getElementById('translationModel').value;
|
741 |
+
const key = apiKey || document.getElementById('translationApiKey').value.trim();
|
742 |
+
|
743 |
+
if (!content) {
|
744 |
+
throw new Error('没有要翻译的内容');
|
745 |
+
}
|
746 |
+
|
747 |
+
if (!key) {
|
748 |
+
throw new Error('未提供API密钥');
|
749 |
+
}
|
750 |
+
|
751 |
+
if (selectedModel === 'none') {
|
752 |
+
return content; // 不需要翻译
|
753 |
+
}
|
754 |
+
|
755 |
+
// 修正targetLanguage值
|
756 |
+
const actualLang = lang === 'chinese' ? 'zh' : lang;
|
757 |
+
|
758 |
+
// 构建统一的翻译提示词
|
759 |
+
const translationPromptTemplate = `请将以下${actualLang === 'zh' ? '英文' : '中文'}内容翻译为${actualLang === 'zh' ? '中文' : '英文'},
|
760 |
+
要求:
|
761 |
+
|
762 |
+
1. 保持所有Markdown语法元素不变(如#标题, *斜体*, **粗体**, [链接](), ![图片]()等)
|
763 |
+
|
764 |
+
2. 学术/专业术语应准确翻译,必要时可保留英文原文在括号中
|
765 |
+
|
766 |
+
3. 保持原文的段落结构和格式
|
767 |
+
|
768 |
+
4. 仅翻译内容,不要添加额外解释
|
769 |
+
|
770 |
+
5. 对于行间公式,使用:
|
771 |
+
$$
|
772 |
+
...
|
773 |
+
$$
|
774 |
+
标记
|
775 |
+
|
776 |
+
文档内容��
|
777 |
+
|
778 |
+
${content}`;
|
779 |
+
|
780 |
+
//对温度等参数做统一默认设置, 若未单独设置, 则使用默认值
|
781 |
+
const temperature = 0.5;
|
782 |
+
const maxTokens = 100000;
|
783 |
+
const sys_prompt = "你是一个专业的文档翻译助手,擅长保持原文档格式进行精确翻译。";
|
784 |
+
|
785 |
+
// 配置各种翻译API
|
786 |
+
const apiConfigs = {
|
787 |
+
'deepseek': {
|
788 |
+
endpoint: 'https://api.deepseek.com/v1/chat/completions',
|
789 |
+
modelName: 'DeepSeek v3 (deepseek-v3)',
|
790 |
+
headers: {
|
791 |
+
'Content-Type': 'application/json',
|
792 |
+
'Authorization': `Bearer ${key}`
|
793 |
+
},
|
794 |
+
bodyBuilder: () => ({
|
795 |
+
model: "deepseek-v3",
|
796 |
+
messages: [
|
797 |
+
{ role: "system", content: sys_prompt },
|
798 |
+
{ role: "user", content: translationPromptTemplate }
|
799 |
+
],
|
800 |
+
temperature: temperature,
|
801 |
+
max_tokens: maxTokens
|
802 |
+
}),
|
803 |
+
responseExtractor: (data) => data.choices[0].message.content
|
804 |
+
},
|
805 |
+
'gemini': {
|
806 |
+
endpoint: `https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash:generateContent?key=${key}`,
|
807 |
+
modelName: 'Google Gemini 2.0 Flash',
|
808 |
+
headers: { 'Content-Type': 'application/json' },
|
809 |
+
bodyBuilder: () => ({
|
810 |
+
contents: [
|
811 |
+
{
|
812 |
+
role: "user",
|
813 |
+
parts: [{ text: translationPromptTemplate }]
|
814 |
+
}
|
815 |
+
],
|
816 |
+
generationConfig: {
|
817 |
+
temperature: temperature,
|
818 |
+
maxOutputTokens: maxTokens
|
819 |
+
}
|
820 |
+
}),
|
821 |
+
responseExtractor: (data) => data.candidates[0].content.parts[0].text
|
822 |
+
},
|
823 |
+
'claude': {
|
824 |
+
endpoint: 'https://api.anthropic.com/v1/messages',
|
825 |
+
modelName: 'Claude 3.5 Sonnet',
|
826 |
+
headers: {
|
827 |
+
'Content-Type': 'application/json',
|
828 |
+
'x-api-key': key,
|
829 |
+
'anthropic-version': '2023-06-01'
|
830 |
+
},
|
831 |
+
bodyBuilder: () => ({
|
832 |
+
model: "claude-3-5-sonnet",
|
833 |
+
max_tokens: maxTokens,
|
834 |
+
messages: [
|
835 |
+
{ role: "user", content: translationPromptTemplate }
|
836 |
+
]
|
837 |
+
}),
|
838 |
+
responseExtractor: (data) => data.content[0].text
|
839 |
+
},
|
840 |
+
'mistral': {
|
841 |
+
endpoint: 'https://api.mistral.ai/v1/chat/completions',
|
842 |
+
modelName: 'Mistral Large (mistral-large-latest)',
|
843 |
+
headers: {
|
844 |
+
'Content-Type': 'application/json',
|
845 |
+
'Authorization': `Bearer ${key}`
|
846 |
+
},
|
847 |
+
bodyBuilder: () => ({
|
848 |
+
model: "mistral-large-latest",
|
849 |
+
messages: [
|
850 |
+
{ role: "system", content: sys_prompt },
|
851 |
+
{ role: "user", content: translationPromptTemplate }
|
852 |
+
],
|
853 |
+
temperature: temperature,
|
854 |
+
max_tokens: maxTokens
|
855 |
+
}),
|
856 |
+
responseExtractor: (data) => data.choices[0].message.content
|
857 |
+
},
|
858 |
+
'tongyi-deepseek-v3': {
|
859 |
+
endpoint: 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions',
|
860 |
+
modelName: '阿里云通义百炼 DeepSeek v3',
|
861 |
+
headers: {
|
862 |
+
'Content-Type': 'application/json',
|
863 |
+
'Authorization': `Bearer ${key}`
|
864 |
+
},
|
865 |
+
bodyBuilder: () => ({
|
866 |
+
model: "deepseek-v3",
|
867 |
+
messages: [
|
868 |
+
{ role: "system", content: sys_prompt },
|
869 |
+
{ role: "user", content: translationPromptTemplate }
|
870 |
+
],
|
871 |
+
temperature: temperature,
|
872 |
+
max_tokens: maxTokens
|
873 |
+
}),
|
874 |
+
responseExtractor: (data) => data.choices[0].message.content
|
875 |
+
},
|
876 |
+
'volcano-deepseek-v3': {
|
877 |
+
endpoint: 'https://api.volcengine.com/ml/api/v1/open/llm/inference',
|
878 |
+
modelName: '火山引擎 DeepSeek v3',
|
879 |
+
headers: {
|
880 |
+
'Content-Type': 'application/json',
|
881 |
+
'Authorization': `Bearer ${key}`
|
882 |
+
},
|
883 |
+
bodyBuilder: () => ({
|
884 |
+
model: "deepseek-v3",
|
885 |
+
messages: [
|
886 |
+
{ role: "system", content: sys_prompt },
|
887 |
+
{ role: "user", content: translationPromptTemplate }
|
888 |
+
],
|
889 |
+
parameters: {
|
890 |
+
temperature: temperature,
|
891 |
+
max_tokens: maxTokens
|
892 |
+
}
|
893 |
+
}),
|
894 |
+
responseExtractor: (data) => data.choices[0].message.content
|
895 |
+
},
|
896 |
+
'custom': {
|
897 |
+
// 自定义模型配置将动态生成
|
898 |
+
createConfig: () => {
|
899 |
+
// 获取用户设置的参数
|
900 |
+
const customModelName = document.getElementById('customModelName').value.trim();
|
901 |
+
const customApiEndpoint = document.getElementById('customApiEndpoint').value.trim();
|
902 |
+
const customModelId = document.getElementById('customModelId').value.trim();
|
903 |
+
const customRequestFormat = document.getElementById('customRequestFormat').value;
|
904 |
+
|
905 |
+
if (!customModelName || !customApiEndpoint || !customModelId) {
|
906 |
+
throw new Error('请填写完整的自定义模型信息');
|
907 |
+
}
|
908 |
+
|
909 |
+
// 根据选择的请求格式创建不同的配置
|
910 |
+
const config = {
|
911 |
+
endpoint: customApiEndpoint,
|
912 |
+
modelName: customModelName,
|
913 |
+
headers: {
|
914 |
+
'Content-Type': 'application/json'
|
915 |
+
},
|
916 |
+
bodyBuilder: null,
|
917 |
+
responseExtractor: null
|
918 |
+
};
|
919 |
+
|
920 |
+
// 添加授权头
|
921 |
+
if (customApiEndpoint.includes('anthropic')) {
|
922 |
+
config.headers['x-api-key'] = key;
|
923 |
+
config.headers['anthropic-version'] = '2023-06-01';
|
924 |
+
} else {
|
925 |
+
config.headers['Authorization'] = `Bearer ${key}`;
|
926 |
+
}
|
927 |
+
|
928 |
+
// 根据格式设置请求体构建器和响应提取器
|
929 |
+
switch (customRequestFormat) {
|
930 |
+
case 'openai':
|
931 |
+
config.bodyBuilder = () => ({
|
932 |
+
model: customModelId,
|
933 |
+
messages: [
|
934 |
+
{ role: "system", content: sys_prompt },
|
935 |
+
{ role: "user", content: translationPromptTemplate }
|
936 |
+
],
|
937 |
+
temperature: temperature,
|
938 |
+
max_tokens: maxTokens
|
939 |
+
});
|
940 |
+
config.responseExtractor = (data) => data.choices[0].message.content;
|
941 |
+
break;
|
942 |
+
|
943 |
+
case 'anthropic':
|
944 |
+
config.bodyBuilder = () => ({
|
945 |
+
model: customModelId,
|
946 |
+
max_tokens: maxTokens,
|
947 |
+
messages: [
|
948 |
+
{ role: "user", content: translationPromptTemplate }
|
949 |
+
]
|
950 |
+
});
|
951 |
+
config.responseExtractor = (data) => data.content[0].text;
|
952 |
+
break;
|
953 |
+
|
954 |
+
case 'gemini':
|
955 |
+
config.bodyBuilder = () => ({
|
956 |
+
contents: [
|
957 |
+
{
|
958 |
+
role: "user",
|
959 |
+
parts: [{ text: translationPromptTemplate }]
|
960 |
+
}
|
961 |
+
],
|
962 |
+
generationConfig: {
|
963 |
+
temperature: temperature,
|
964 |
+
maxOutputTokens: maxTokens
|
965 |
+
}
|
966 |
+
});
|
967 |
+
config.responseExtractor = (data) => data.candidates[0].content.parts[0].text;
|
968 |
+
break;
|
969 |
+
}
|
970 |
+
|
971 |
+
return config;
|
972 |
+
}
|
973 |
+
}
|
974 |
+
};
|
975 |
+
|
976 |
+
// 选择API配置
|
977 |
+
const apiConfig = apiConfigs[selectedModel];
|
978 |
+
|
979 |
+
if (!apiConfig) {
|
980 |
+
throw new Error(`不支持的翻译模型: ${selectedModel}`);
|
981 |
+
}
|
982 |
+
|
983 |
+
addProgressLog(`正在调用${apiConfig.modelName || selectedModel}翻译API...`);
|
984 |
+
let response;
|
985 |
+
// 使用常规模型配置
|
986 |
+
if (selectedModel !== 'custom') {
|
987 |
+
response = await fetch(apiConfig.endpoint, {
|
988 |
+
method: 'POST',
|
989 |
+
headers: apiConfig.headers,
|
990 |
+
body: JSON.stringify(apiConfig.bodyBuilder())
|
991 |
+
});
|
992 |
+
} else {
|
993 |
+
// 使用自定义模型配置
|
994 |
+
const customConfig = apiConfig.createConfig();
|
995 |
+
response = await fetch(customConfig.endpoint, {
|
996 |
+
method: 'POST',
|
997 |
+
headers: customConfig.headers,
|
998 |
+
body: JSON.stringify(customConfig.bodyBuilder())
|
999 |
+
});
|
1000 |
+
}
|
1001 |
+
|
1002 |
+
if (!response.ok) {
|
1003 |
+
let errorText;
|
1004 |
+
try {
|
1005 |
+
const errorJson = await response.json();
|
1006 |
+
errorText = JSON.stringify(errorJson);
|
1007 |
+
} catch (e) {
|
1008 |
+
errorText = await response.text();
|
1009 |
+
}
|
1010 |
+
|
1011 |
+
console.error(`API错误 (${response.status}): ${errorText}`);
|
1012 |
+
throw new Error(`翻译API返回错误 (${response.status}): ${errorText.substring(0, 200)}`);
|
1013 |
+
}
|
1014 |
+
|
1015 |
+
const data = await response.json();
|
1016 |
+
|
1017 |
+
// 提取翻译后的内容
|
1018 |
+
let translatedContent;
|
1019 |
+
|
1020 |
+
if (selectedModel !== 'custom') {
|
1021 |
+
// 使用预定义模型的响应提取器
|
1022 |
+
translatedContent = apiConfig.responseExtractor(data);
|
1023 |
+
} else {
|
1024 |
+
// 使用自定义模型的响应提取器
|
1025 |
+
const customConfig = apiConfig.createConfig();
|
1026 |
+
translatedContent = customConfig.responseExtractor(data);
|
1027 |
+
}
|
1028 |
+
|
1029 |
+
try {
|
1030 |
+
// 提取翻译结果
|
1031 |
+
if (!translatedContent) {
|
1032 |
+
throw new Error('译文为空');
|
1033 |
+
}
|
1034 |
+
|
1035 |
+
return translatedContent;
|
1036 |
+
} catch (error) {
|
1037 |
+
console.error('提取翻译结果错误:', error, '原始响应:', data);
|
1038 |
+
throw new Error(`提取翻译结果失败: ${error.message}`);
|
1039 |
+
}
|
1040 |
+
} catch (error) {
|
1041 |
+
console.error('翻译错误:', error);
|
1042 |
+
throw new Error(`调用${model}翻译API失败: ${error.message}`);
|
1043 |
+
}
|
1044 |
+
}
|
1045 |
+
|
1046 |
+
// 长文档翻译函数
|
1047 |
+
async function translateLongDocument(markdownText, targetLang, model, apiKey) {
|
1048 |
+
const parts = splitMarkdownIntoChunks(markdownText);
|
1049 |
+
console.log(`将文档分割为${parts.length}个部分进行翻译`);
|
1050 |
+
addProgressLog(`文档被分割为${parts.length}个部分进行翻译`);
|
1051 |
+
|
1052 |
+
let translatedContent = '';
|
1053 |
+
|
1054 |
+
for (let i = 0; i < parts.length; i++) {
|
1055 |
+
updateProgress(`翻译第 ${i+1}/${parts.length} 部分...`, 60 + Math.floor((i / parts.length) * 30));
|
1056 |
+
addProgressLog(`正在翻译第 ${i+1}/${parts.length} 部分...`);
|
1057 |
+
|
1058 |
+
try {
|
1059 |
+
// 翻译当前部分
|
1060 |
+
const partResult = await translateMarkdown(parts[i], targetLang, model, apiKey);
|
1061 |
+
translatedContent += partResult;
|
1062 |
+
|
1063 |
+
// 添加分隔符(如果不是最后一部分)
|
1064 |
+
if (i < parts.length - 1) {
|
1065 |
+
translatedContent += '\n\n';
|
1066 |
+
}
|
1067 |
+
|
1068 |
+
// 简单的节流,避免API速率限制
|
1069 |
+
if (i < parts.length - 1) {
|
1070 |
+
await new Promise(resolve => setTimeout(resolve, 1000));
|
1071 |
+
}
|
1072 |
+
} catch (error) {
|
1073 |
+
console.error(`第 ${i+1} 部分翻译失败:`, error);
|
1074 |
+
addProgressLog(`第 ${i+1} 部分翻译失败: ${error.message}`);
|
1075 |
+
|
1076 |
+
// 继续尝试其余部分
|
1077 |
+
translatedContent += `\n\n> **翻译错误 (第 ${i+1} 部分), 使用原语言**: ${error.message}\n\n${parts[i]}\n\n`;
|
1078 |
+
}
|
1079 |
+
}
|
1080 |
+
|
1081 |
+
|
1082 |
+
return translatedContent;
|
1083 |
+
}
|
1084 |
+
|
1085 |
+
// 智能分割Markdown为多个片段
|
1086 |
+
function splitMarkdownIntoChunks(markdown) {
|
1087 |
+
// 估计每个标记的平均长度
|
1088 |
+
const estimatedTokens = estimateTokenCount(markdown);
|
1089 |
+
// 从用户设置获取最大token数限制
|
1090 |
+
const tokenLimit = parseInt(maxTokensPerChunk.value) || 2000;
|
1091 |
+
|
1092 |
+
// 如果文档足够小,不需要分割
|
1093 |
+
if (estimatedTokens <= tokenLimit) {
|
1094 |
+
return [markdown];
|
1095 |
+
}
|
1096 |
+
|
1097 |
+
// 按章节分割
|
1098 |
+
const chunks = [];
|
1099 |
+
const lines = markdown.split('\n');
|
1100 |
+
let currentChunk = [];
|
1101 |
+
let currentTokenCount = 0;
|
1102 |
+
let inCodeBlock = false;
|
1103 |
+
|
1104 |
+
// 定义标题行的正则表达式
|
1105 |
+
const headingRegex = /^#{1,6}\s+.+$/;
|
1106 |
+
|
1107 |
+
for (let i = 0; i < lines.length; i++) {
|
1108 |
+
const line = lines[i];
|
1109 |
+
|
1110 |
+
// 检测代码块
|
1111 |
+
if (line.trim().startsWith('```')) {
|
1112 |
+
inCodeBlock = !inCodeBlock;
|
1113 |
+
}
|
1114 |
+
|
1115 |
+
// 估计当前行的token数
|
1116 |
+
const lineTokens = estimateTokenCount(line);
|
1117 |
+
|
1118 |
+
// 判断是否应该在这里分割
|
1119 |
+
const isHeading = headingRegex.test(line) && !inCodeBlock;
|
1120 |
+
const wouldExceedLimit = currentTokenCount + lineTokens > tokenLimit;
|
1121 |
+
|
1122 |
+
if (isHeading && currentChunk.length > 0 && (wouldExceedLimit || currentTokenCount > tokenLimit * 0.7)) {
|
1123 |
+
// 在遇到标题且当前段已积累足够内容时分割
|
1124 |
+
chunks.push(currentChunk.join('\n'));
|
1125 |
+
currentChunk = [];
|
1126 |
+
currentTokenCount = 0;
|
1127 |
+
}
|
1128 |
+
|
1129 |
+
// 如果当前段落即使加上这一行也超过限制,而且已经有内容了
|
1130 |
+
if (!isHeading && wouldExceedLimit && currentChunk.length > 0) {
|
1131 |
+
chunks.push(currentChunk.join('\n'));
|
1132 |
+
currentChunk = [];
|
1133 |
+
currentTokenCount = 0;
|
1134 |
+
}
|
1135 |
+
|
1136 |
+
// 添加当前行到当前段落
|
1137 |
+
currentChunk.push(line);
|
1138 |
+
currentTokenCount += lineTokens;
|
1139 |
+
}
|
1140 |
+
|
1141 |
+
// 添加最后一段
|
1142 |
+
if (currentChunk.length > 0) {
|
1143 |
+
chunks.push(currentChunk.join('\n'));
|
1144 |
+
}
|
1145 |
+
|
1146 |
+
// 处理过大的段落(可能是因为没有标题或标记导致的)
|
1147 |
+
const finalChunks = [];
|
1148 |
+
for (const chunk of chunks) {
|
1149 |
+
const chunkTokens = estimateTokenCount(chunk);
|
1150 |
+
if (chunkTokens > tokenLimit) {
|
1151 |
+
// 如果段落仍然过大,按段落分割
|
1152 |
+
const subChunks = splitByParagraphs(chunk, tokenLimit);
|
1153 |
+
finalChunks.push(...subChunks);
|
1154 |
+
} else {
|
1155 |
+
finalChunks.push(chunk);
|
1156 |
+
}
|
1157 |
+
}
|
1158 |
+
|
1159 |
+
return finalChunks;
|
1160 |
+
}
|
1161 |
+
|
1162 |
+
// 按段落分割过大的文本块
|
1163 |
+
function splitByParagraphs(text, tokenLimit) {
|
1164 |
+
const paragraphs = text.split('\n\n');
|
1165 |
+
const chunks = [];
|
1166 |
+
let currentChunk = [];
|
1167 |
+
let currentTokenCount = 0;
|
1168 |
+
|
1169 |
+
for (const paragraph of paragraphs) {
|
1170 |
+
const paragraphTokens = estimateTokenCount(paragraph);
|
1171 |
+
|
1172 |
+
// 如果单个段落就超过了限制,则需要进一步分割
|
1173 |
+
if (paragraphTokens > tokenLimit) {
|
1174 |
+
// 如果当前段已有内容,先保存
|
1175 |
+
if (currentChunk.length > 0) {
|
1176 |
+
chunks.push(currentChunk.join('\n\n'));
|
1177 |
+
currentChunk = [];
|
1178 |
+
currentTokenCount = 0;
|
1179 |
+
}
|
1180 |
+
|
1181 |
+
// 按句子分割大段落
|
1182 |
+
const sentenceChunks = splitBySentences(paragraph, tokenLimit);
|
1183 |
+
chunks.push(...sentenceChunks);
|
1184 |
+
continue;
|
1185 |
+
}
|
1186 |
+
|
1187 |
+
// 检查是否加上这个段落会超出限制
|
1188 |
+
if (currentTokenCount + paragraphTokens > tokenLimit && currentChunk.length > 0) {
|
1189 |
+
chunks.push(currentChunk.join('\n\n'));
|
1190 |
+
currentChunk = [];
|
1191 |
+
currentTokenCount = 0;
|
1192 |
+
}
|
1193 |
+
|
1194 |
+
currentChunk.push(paragraph);
|
1195 |
+
currentTokenCount += paragraphTokens;
|
1196 |
+
}
|
1197 |
+
|
1198 |
+
// 添加最后一段
|
1199 |
+
if (currentChunk.length > 0) {
|
1200 |
+
chunks.push(currentChunk.join('\n\n'));
|
1201 |
+
}
|
1202 |
+
|
1203 |
+
return chunks;
|
1204 |
+
}
|
1205 |
+
|
1206 |
+
// 按句子分割过大的段落
|
1207 |
+
function splitBySentences(paragraph, tokenLimit) {
|
1208 |
+
// 简单的句子分割规则(这里可以根据需要改进)
|
1209 |
+
const sentences = paragraph.replace(/([.!?。!?])\s*/g, "$1\n").split('\n');
|
1210 |
+
const chunks = [];
|
1211 |
+
let currentChunk = [];
|
1212 |
+
let currentTokenCount = 0;
|
1213 |
+
|
1214 |
+
for (const sentence of sentences) {
|
1215 |
+
if (!sentence.trim()) continue;
|
1216 |
+
|
1217 |
+
const sentenceTokens = estimateTokenCount(sentence);
|
1218 |
+
|
1219 |
+
// 检查是否加上这个句子会超出限制
|
1220 |
+
if (currentTokenCount + sentenceTokens > tokenLimit && currentChunk.length > 0) {
|
1221 |
+
chunks.push(currentChunk.join(' '));
|
1222 |
+
currentChunk = [];
|
1223 |
+
currentTokenCount = 0;
|
1224 |
+
}
|
1225 |
+
|
1226 |
+
currentChunk.push(sentence);
|
1227 |
+
currentTokenCount += sentenceTokens;
|
1228 |
+
}
|
1229 |
+
|
1230 |
+
// 添加最后一段
|
1231 |
+
if (currentChunk.length > 0) {
|
1232 |
+
chunks.push(currentChunk.join(' '));
|
1233 |
+
}
|
1234 |
+
|
1235 |
+
return chunks;
|
1236 |
+
}
|
1237 |
+
|
1238 |
+
// 合并翻译后的片段
|
1239 |
+
function mergeTranslatedChunks(chunks) {
|
1240 |
+
// 简单合并,可以根据需要添加更复杂的处理逻辑
|
1241 |
+
return chunks.join('\n\n');
|
1242 |
+
}
|
1243 |
+
|
1244 |
+
// 估计文本的token数量(粗略估计)
|
1245 |
+
function estimateTokenCount(text) {
|
1246 |
+
if (!text) return 0;
|
1247 |
+
|
1248 |
+
// 中文和其他语言的处理方式不同
|
1249 |
+
// 英文大约每75个字符对应20个token
|
1250 |
+
// 中文大约每字符对应1.5个token
|
1251 |
+
|
1252 |
+
// 检测是否包含大量中文
|
1253 |
+
const chineseRatio = (text.match(/[\u4e00-\u9fa5]/g) || []).length / text.length;
|
1254 |
+
|
1255 |
+
if (chineseRatio > 0.5) {
|
1256 |
+
// 主要是中文文本
|
1257 |
+
return Math.ceil(text.length * 1.5);
|
1258 |
+
} else {
|
1259 |
+
// 主要是英文或其他文本
|
1260 |
+
return Math.ceil(text.length / 3.75);
|
1261 |
+
}
|
1262 |
+
}
|
1263 |
+
|
1264 |
+
// 将文件转换为Base64
|
1265 |
+
function fileToBase64(file) {
|
1266 |
+
return new Promise((resolve, reject) => {
|
1267 |
+
const reader = new FileReader();
|
1268 |
+
reader.readAsDataURL(file);
|
1269 |
+
reader.onload = () => resolve(reader.result);
|
1270 |
+
reader.onerror = error => reject(error);
|
1271 |
+
});
|
1272 |
+
}
|
1273 |
+
|
1274 |
+
// 下载单个文本文件
|
1275 |
+
function downloadText(content, filename) {
|
1276 |
+
const blob = new Blob([content], { type: 'text/markdown' });
|
1277 |
+
saveAs(blob, filename);
|
1278 |
+
}
|
1279 |
+
|
1280 |
+
// 下载包含图像的Markdown
|
1281 |
+
async function downloadMarkdownWithImages() {
|
1282 |
+
try {
|
1283 |
+
const zip = new JSZip();
|
1284 |
+
|
1285 |
+
// 添加Markdown文件
|
1286 |
+
zip.file('document.md', markdownContent);
|
1287 |
+
|
1288 |
+
// 创建images文件夹
|
1289 |
+
const imagesFolder = zip.folder('images');
|
1290 |
+
|
1291 |
+
// 添加图片
|
1292 |
+
for (const img of imagesData) {
|
1293 |
+
const imgData = img.data.split(',')[1];
|
1294 |
+
imagesFolder.file(`${img.id}.png`, imgData, { base64: true });
|
1295 |
+
}
|
1296 |
+
|
1297 |
+
// 生成并下载zip文件
|
1298 |
+
const zipBlob = await zip.generateAsync({ type: 'blob' });
|
1299 |
+
const pdfName = pdfFile ? pdfFile.name.replace('.pdf', '') : 'document';
|
1300 |
+
saveAs(zipBlob, `${pdfName}_markdown.zip`);
|
1301 |
+
} catch (error) {
|
1302 |
+
console.error('创建ZIP文件失败:', error);
|
1303 |
+
showNotification('下载失败: ' + error.message, 'error');
|
1304 |
+
}
|
1305 |
+
}
|
1306 |
+
|
1307 |
+
downloadTranslationWithImages = async () => {
|
1308 |
+
try {
|
1309 |
+
const zip = new JSZip();
|
1310 |
+
|
1311 |
+
// 直接添加声明到翻译内容
|
1312 |
+
const currentDate = new Date().toISOString().split('T')[0];
|
1313 |
+
const headerDeclaration = `> *本文档由 Paper Burner 工具制作 (${currentDate})。内容由 AI 大模型翻译生成,不保证翻译内容的准确性和完整性。*\n\n`;
|
1314 |
+
const footerDeclaration = `\n\n---\n> *免责声明:本文档内容由大模型API自动翻译生成,Paper Burner 工具不对翻译内容的准确性、完整性和合法性负责。*`;
|
1315 |
+
|
1316 |
+
// 添加Markdown文件,包含声明
|
1317 |
+
const contentToDownload = headerDeclaration + translationContent + footerDeclaration;
|
1318 |
+
zip.file('document.md', contentToDownload);
|
1319 |
+
|
1320 |
+
// 创建images文件夹
|
1321 |
+
const imagesFolder = zip.folder('images');
|
1322 |
+
|
1323 |
+
// 添加图片
|
1324 |
+
for (const img of imagesData) {
|
1325 |
+
const imgData = img.data.split(',')[1];
|
1326 |
+
imagesFolder.file(`${img.id}.png`, imgData, { base64: true });
|
1327 |
+
}
|
1328 |
+
|
1329 |
+
// 生成并下载zip文件
|
1330 |
+
const zipBlob = await zip.generateAsync({ type: 'blob' });
|
1331 |
+
const pdfName = pdfFile ? pdfFile.name.replace('.pdf', '') : 'document';
|
1332 |
+
saveAs(zipBlob, `${pdfName}_translation.zip`);
|
1333 |
+
} catch (error) {
|
1334 |
+
console.error('创建ZIP文件失败:', error);
|
1335 |
+
showNotification('下载失败: ' + error.message, 'error');
|
1336 |
+
}
|
1337 |
+
}
|
1338 |
+
|
1339 |
+
// 保存设置
|
1340 |
+
function saveSettings() {
|
1341 |
+
// 保存高级设置到本地存储
|
1342 |
+
localStorage.setItem('advancedSettings', JSON.stringify({
|
1343 |
+
maxTokensPerChunk: maxTokensPerChunk.value
|
1344 |
+
}));
|
1345 |
+
|
1346 |
+
// 如果是自定义模型,保存自定义模型设置
|
1347 |
+
if (translationModelSelect.value === 'custom') {
|
1348 |
+
localStorage.setItem('customModelSettings', JSON.stringify({
|
1349 |
+
modelName: document.getElementById('customModelName').value,
|
1350 |
+
apiEndpoint: document.getElementById('customApiEndpoint').value,
|
1351 |
+
modelId: document.getElementById('customModelId').value,
|
1352 |
+
requestFormat: document.getElementById('customRequestFormat').value
|
1353 |
+
}));
|
1354 |
+
}
|
1355 |
+
|
1356 |
+
// 保存选中的翻译模型
|
1357 |
+
localStorage.setItem('selectedTranslationModel', translationModelSelect.value);
|
1358 |
+
}
|
1359 |
+
|
1360 |
+
// 加载设置
|
1361 |
+
function loadSettings() {
|
1362 |
+
// 加载高级设置
|
1363 |
+
try {
|
1364 |
+
const advancedSettingsData = localStorage.getItem('advancedSettings');
|
1365 |
+
if (advancedSettingsData) {
|
1366 |
+
const settings = JSON.parse(advancedSettingsData);
|
1367 |
+
if (settings.maxTokensPerChunk) {
|
1368 |
+
maxTokensPerChunk.value = settings.maxTokensPerChunk;
|
1369 |
+
maxTokensPerChunkValue.textContent = settings.maxTokensPerChunk;
|
1370 |
+
}
|
1371 |
+
}
|
1372 |
+
} catch (e) {
|
1373 |
+
console.error('加载高级设置失败:', e);
|
1374 |
+
}
|
1375 |
+
|
1376 |
+
// 加载自定义模型设置
|
1377 |
+
try {
|
1378 |
+
const customModelData = localStorage.getItem('customModelSettings');
|
1379 |
+
if (customModelData) {
|
1380 |
+
const settings = JSON.parse(customModelData);
|
1381 |
+
document.getElementById('customModelName').value = settings.modelName || '';
|
1382 |
+
document.getElementById('customApiEndpoint').value = settings.apiEndpoint || '';
|
1383 |
+
document.getElementById('customModelId').value = settings.modelId || '';
|
1384 |
+
document.getElementById('customRequestFormat').value = settings.requestFormat || 'openai';
|
1385 |
+
}
|
1386 |
+
} catch (e) {
|
1387 |
+
console.error('加载自定义模型设置失败:', e);
|
1388 |
+
}
|
1389 |
+
|
1390 |
+
// 加载选中的翻译模型
|
1391 |
+
try {
|
1392 |
+
const selectedModel = localStorage.getItem('selectedTranslationModel');
|
1393 |
+
if (selectedModel) {
|
1394 |
+
translationModelSelect.value = selectedModel;
|
1395 |
+
// 如果是自定义模型,显示自定义设置
|
1396 |
+
if (selectedModel === 'custom') {
|
1397 |
+
customModelSettings.classList.remove('hidden');
|
1398 |
+
}
|
1399 |
+
}
|
1400 |
+
} catch (e) {
|
1401 |
+
console.error('加载选中的翻译模型失败:', e);
|
1402 |
+
}
|
1403 |
+
}
|
index.html
CHANGED
@@ -1,19 +0,0 @@
|
|
1 |
-
<!doctype html>
|
2 |
-
<html>
|
3 |
-
<head>
|
4 |
-
<meta charset="utf-8" />
|
5 |
-
<meta name="viewport" content="width=device-width" />
|
6 |
-
<title>My static Space</title>
|
7 |
-
<link rel="stylesheet" href="style.css" />
|
8 |
-
</head>
|
9 |
-
<body>
|
10 |
-
<div class="card">
|
11 |
-
<h1>Welcome to your static Space!</h1>
|
12 |
-
<p>You can modify this app directly by editing <i>index.html</i> in the Files and versions tab.</p>
|
13 |
-
<p>
|
14 |
-
Also don't forget to check the
|
15 |
-
<a href="https://huggingface.co/docs/hub/spaces" target="_blank">Spaces documentation</a>.
|
16 |
-
</p>
|
17 |
-
</div>
|
18 |
-
</body>
|
19 |
-
</html>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|