Spaces:
Running
Running
Upload 3 files
Browse files- app.py +191 -104
- translator.py +118 -99
app.py
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
import os
|
|
|
|
| 2 |
import gradio as gr
|
| 3 |
import huggingface_hub
|
| 4 |
import numpy as np
|
|
@@ -7,7 +8,6 @@ import pandas as pd
|
|
| 7 |
from PIL import Image
|
| 8 |
from huggingface_hub import login
|
| 9 |
|
| 10 |
-
# 导入修改后的翻译函数
|
| 11 |
from translator import translate_texts
|
| 12 |
|
| 13 |
# ------------------------------------------------------------------
|
|
@@ -19,10 +19,7 @@ LABEL_FILENAME = "selected_tags.csv"
|
|
| 19 |
|
| 20 |
HF_TOKEN = os.environ.get("HF_TOKEN", "")
|
| 21 |
if HF_TOKEN:
|
| 22 |
-
|
| 23 |
-
login(token=HF_TOKEN)
|
| 24 |
-
except Exception as e:
|
| 25 |
-
print(f"Hugging Face登录失败: {e}")
|
| 26 |
else:
|
| 27 |
print("⚠️ 未检测到 HF_TOKEN,私有模型可能下载失败")
|
| 28 |
|
|
@@ -125,38 +122,114 @@ except RuntimeError as e:
|
|
| 125 |
# Gradio UI
|
| 126 |
# ------------------------------------------------------------------
|
| 127 |
custom_css = """
|
| 128 |
-
.label-container {
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 135 |
"""
|
| 136 |
|
| 137 |
_js_functions = """
|
| 138 |
function copyToClipboard(text) {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 139 |
if (typeof text === 'undefined' || text === null) {
|
| 140 |
-
console.warn('copyToClipboard was called with undefined or null text.');
|
|
|
|
| 141 |
return;
|
| 142 |
}
|
|
|
|
| 143 |
navigator.clipboard.writeText(text).then(() => {
|
|
|
|
| 144 |
const feedback = document.createElement('div');
|
| 145 |
-
|
|
|
|
|
|
|
| 146 |
displayText = displayText.substring(0, 30) + (displayText.length > 30 ? '...' : '');
|
|
|
|
| 147 |
feedback.textContent = '已复制: ' + displayText;
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 153 |
document.body.appendChild(feedback);
|
| 154 |
setTimeout(() => {
|
| 155 |
feedback.style.opacity = '0';
|
| 156 |
-
setTimeout(() => {
|
|
|
|
|
|
|
|
|
|
|
|
|
| 157 |
}, 1500);
|
| 158 |
}).catch(err => {
|
| 159 |
-
console.error('Failed to copy tag. Error:', err, '
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 160 |
});
|
| 161 |
}
|
| 162 |
"""
|
|
@@ -180,20 +253,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title="AI 图像标签分析器", css=cus
|
|
| 180 |
gen_slider = gr.Slider(0, 1, value=0.35, step=0.01, label="通用标签阈值", info="越高 → 标签更少更准")
|
| 181 |
char_slider = gr.Slider(0, 1, value=0.85, step=0.01, label="角色标签阈值", info="推荐保持较高阈值")
|
| 182 |
show_tag_scores = gr.Checkbox(True, label="在列表中显示标签置信度")
|
| 183 |
-
|
| 184 |
-
with gr.Accordion("🔑 翻译服务配置", open=False):
|
| 185 |
-
enable_translation_cb = gr.Checkbox(label="启用翻译", value=True, info="取消勾选则不进行翻译")
|
| 186 |
-
gr.Markdown("提供 **系统访问密钥** 或 **自定义API密钥** 来启用翻译功能。如果两者均未提供或不正确,将不进行翻译。")
|
| 187 |
-
|
| 188 |
-
with gr.Tabs():
|
| 189 |
-
with gr.TabItem("使用系统密钥"):
|
| 190 |
-
system_key_input = gr.Textbox(label="系统访问密钥", type="password", placeholder="输入管理员提供的密钥")
|
| 191 |
-
with gr.TabItem("使用自定义API"):
|
| 192 |
-
gr.Markdown("在此处填入你自己的翻译API密钥。")
|
| 193 |
-
tencent_id_input = gr.Textbox(label="腾讯云 SecretId", type="password")
|
| 194 |
-
tencent_key_input = gr.Textbox(label="腾讯云 SecretKey", type="password")
|
| 195 |
-
baidu_json_input = gr.Textbox(label="百度翻译凭证 (JSON格式)", type="password", placeholder='[{"app_id":"...", "secret_key":"..."}]')
|
| 196 |
-
|
| 197 |
with gr.Accordion("📊 标签汇总设置", open=True):
|
| 198 |
gr.Markdown("选择要包含在下方汇总文本框中的标签类别:")
|
| 199 |
with gr.Row():
|
|
@@ -223,24 +283,27 @@ with gr.Blocks(theme=gr.themes.Soft(), title="AI 图像标签分析器", css=cus
|
|
| 223 |
show_copy_button=True
|
| 224 |
)
|
| 225 |
|
| 226 |
-
|
|
|
|
| 227 |
if not tags_dict:
|
| 228 |
return "<p>暂无标签</p>"
|
| 229 |
|
| 230 |
html = '<div class="label-container">'
|
|
|
|
|
|
|
|
|
|
|
|
|
| 231 |
tag_keys = list(tags_dict.keys())
|
| 232 |
|
| 233 |
for i, tag in enumerate(tag_keys):
|
| 234 |
score = tags_dict[tag]
|
| 235 |
-
escaped_tag = tag.replace("'", "\\'")
|
| 236 |
|
| 237 |
html += '<div class="tag-item">'
|
| 238 |
tag_display_html = f'<span class="tag-en" onclick="copyToClipboard(\'{escaped_tag}\')">{tag}</span>'
|
| 239 |
|
| 240 |
-
|
| 241 |
-
|
| 242 |
-
if show_translation_in_list and translation_text and translation_text != tag:
|
| 243 |
-
tag_display_html += f'<span class="tag-zh">({translation_text})</span>'
|
| 244 |
|
| 245 |
html += f'<div>{tag_display_html}</div>'
|
| 246 |
if show_scores:
|
|
@@ -253,17 +316,20 @@ with gr.Blocks(theme=gr.themes.Soft(), title="AI 图像标签分析器", css=cus
|
|
| 253 |
current_res, current_translations_dict,
|
| 254 |
s_gen, s_char, s_rat, s_sep_type, s_show_zh
|
| 255 |
):
|
| 256 |
-
if not current_res:
|
|
|
|
| 257 |
|
| 258 |
summary_parts = []
|
| 259 |
-
|
|
|
|
| 260 |
|
| 261 |
categories_to_summarize = []
|
| 262 |
if s_gen: categories_to_summarize.append("general")
|
| 263 |
if s_char: categories_to_summarize.append("characters")
|
| 264 |
if s_rat: categories_to_summarize.append("ratings")
|
| 265 |
|
| 266 |
-
if not categories_to_summarize:
|
|
|
|
| 267 |
|
| 268 |
for cat_key in categories_to_summarize:
|
| 269 |
if current_res.get(cat_key):
|
|
@@ -272,85 +338,97 @@ with gr.Blocks(theme=gr.themes.Soft(), title="AI 图像标签分析器", css=cus
|
|
| 272 |
cat_translations = current_translations_dict.get(cat_key, [])
|
| 273 |
|
| 274 |
for i, en_tag in enumerate(cat_tags_en):
|
| 275 |
-
|
| 276 |
-
|
| 277 |
-
if s_show_zh and translation_text and translation_text != en_tag:
|
| 278 |
-
tags_to_join.append(f"{en_tag}({translation_text})")
|
| 279 |
else:
|
| 280 |
tags_to_join.append(en_tag)
|
| 281 |
-
if tags_to_join:
|
| 282 |
summary_parts.append(separator.join(tags_to_join))
|
| 283 |
|
| 284 |
-
|
|
|
|
|
|
|
| 285 |
final_summary = joiner.join(summary_parts)
|
| 286 |
return final_summary if final_summary else "选定的类别中没有找到标签。"
|
| 287 |
|
|
|
|
|
|
|
| 288 |
def process_image_and_generate_outputs(
|
| 289 |
img, g_th, c_th, s_scores, # Main inputs
|
| 290 |
-
s_gen, s_char, s_rat, s_sep, s_zh_in_sum
|
| 291 |
-
# New translation controls
|
| 292 |
-
enable_translation, sys_key, tc_id, tc_key, baidu_json
|
| 293 |
):
|
| 294 |
-
initial_yield_state = (
|
| 295 |
-
gr.update(interactive=True, value="🚀 开始分析"), # btn
|
| 296 |
-
"", "", "", "", # html outputs
|
| 297 |
-
gr.update(placeholder="分析失败..."), # summary
|
| 298 |
-
{}, {}, {} # states
|
| 299 |
-
)
|
| 300 |
if img is None:
|
| 301 |
-
yield (
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 302 |
return
|
| 303 |
|
| 304 |
if tagger_instance is None:
|
| 305 |
-
yield (
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 306 |
return
|
| 307 |
|
| 308 |
yield (
|
| 309 |
gr.update(interactive=False, value="🔄 处理中..."),
|
| 310 |
gr.update(visible=True, value="🔄 正在分析图像,请稍候..."),
|
| 311 |
-
gr.HTML(value="<p>分析中...</p>"),
|
| 312 |
-
gr.
|
|
|
|
|
|
|
|
|
|
| 313 |
)
|
| 314 |
|
| 315 |
try:
|
|
|
|
| 316 |
res, tag_categories_original_order = tagger_instance.predict(img, g_th, c_th)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 317 |
|
| 318 |
current_translations_dict = {}
|
| 319 |
-
|
| 320 |
-
|
| 321 |
-
|
| 322 |
-
|
| 323 |
-
|
| 324 |
-
|
| 325 |
-
if all_tags_to_translate:
|
| 326 |
-
# 使用新的参数调用翻译函数
|
| 327 |
-
all_translations_flat = translate_texts(
|
| 328 |
-
texts=all_tags_to_translate,
|
| 329 |
-
system_key_input=sys_key,
|
| 330 |
-
tencent_id=tc_id,
|
| 331 |
-
tencent_key=tc_key,
|
| 332 |
-
baidu_creds_json_str=baidu_json
|
| 333 |
-
)
|
| 334 |
-
|
| 335 |
-
offset = 0
|
| 336 |
-
for cat_key in ["general", "characters", "ratings"]:
|
| 337 |
-
num_tags_in_cat = len(tag_categories_original_order.get(cat_key, []))
|
| 338 |
-
current_translations_dict[cat_key] = all_translations_flat[offset : offset + num_tags_in_cat] if num_tags_in_cat > 0 else []
|
| 339 |
offset += num_tags_in_cat
|
| 340 |
-
|
| 341 |
-
|
| 342 |
-
current_translations_dict[cat_key] = []
|
| 343 |
|
| 344 |
-
general_html = format_tags_html(res.get("general", {}), current_translations_dict.get("general", []), s_scores,
|
| 345 |
-
char_html = format_tags_html(res.get("characters", {}), current_translations_dict.get("characters", []), s_scores,
|
| 346 |
-
rating_html = format_tags_html(res.get("ratings", {}), current_translations_dict.get("ratings", []), s_scores,
|
| 347 |
|
| 348 |
-
summary_text = generate_summary_text_content(
|
|
|
|
|
|
|
|
|
|
| 349 |
|
| 350 |
yield (
|
| 351 |
-
gr.update(interactive=True, value="🚀 开始分析"),
|
| 352 |
-
|
| 353 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 354 |
)
|
| 355 |
|
| 356 |
except Exception as e:
|
|
@@ -358,13 +436,14 @@ with gr.Blocks(theme=gr.themes.Soft(), title="AI 图像标签分析器", css=cus
|
|
| 358 |
tb_str = traceback.format_exc()
|
| 359 |
print(f"处理时发生错误: {e}\n{tb_str}")
|
| 360 |
yield (
|
| 361 |
-
gr.update(visible=True, value=f"❌ 处理失败: {str(e)}"),
|
| 362 |
gr.update(interactive=True, value="🚀 开始分析"),
|
|
|
|
| 363 |
"<p>处理出错</p>", "<p>处理出错</p>", "<p>处理出错</p>",
|
| 364 |
gr.update(value=f"错误: {str(e)}", placeholder="分析失败..."),
|
| 365 |
{}, {}, {}
|
| 366 |
)
|
| 367 |
|
|
|
|
| 368 |
def update_summary_display(
|
| 369 |
s_gen, s_char, s_rat, s_sep, s_zh_in_sum,
|
| 370 |
current_res_from_state, current_translations_from_state
|
|
@@ -378,16 +457,20 @@ with gr.Blocks(theme=gr.themes.Soft(), title="AI 图像标签分析器", css=cus
|
|
| 378 |
)
|
| 379 |
return gr.update(value=new_summary_text)
|
| 380 |
|
| 381 |
-
|
| 382 |
-
translation_inputs = [enable_translation_cb, system_key_input, tencent_id_input, tencent_key_input, baidu_json_input]
|
| 383 |
-
|
| 384 |
btn.click(
|
| 385 |
process_image_and_generate_outputs,
|
| 386 |
-
inputs=[
|
| 387 |
-
|
| 388 |
-
|
| 389 |
-
|
| 390 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 391 |
)
|
| 392 |
|
| 393 |
summary_controls = [sum_general, sum_char, sum_rating, sum_sep, sum_show_zh]
|
|
@@ -395,9 +478,13 @@ with gr.Blocks(theme=gr.themes.Soft(), title="AI 图像标签分析器", css=cus
|
|
| 395 |
ctrl.change(
|
| 396 |
fn=update_summary_display,
|
| 397 |
inputs=summary_controls + [state_res, state_translations_dict],
|
| 398 |
-
outputs=[out_summary]
|
|
|
|
| 399 |
)
|
| 400 |
-
|
|
|
|
|
|
|
|
|
|
| 401 |
if __name__ == "__main__":
|
| 402 |
if tagger_instance is None:
|
| 403 |
print("CRITICAL: Tagger 未能初始化,应用功能将受限。请检查之前的错误信息。")
|
|
|
|
| 1 |
import os
|
| 2 |
+
import json
|
| 3 |
import gradio as gr
|
| 4 |
import huggingface_hub
|
| 5 |
import numpy as np
|
|
|
|
| 8 |
from PIL import Image
|
| 9 |
from huggingface_hub import login
|
| 10 |
|
|
|
|
| 11 |
from translator import translate_texts
|
| 12 |
|
| 13 |
# ------------------------------------------------------------------
|
|
|
|
| 19 |
|
| 20 |
HF_TOKEN = os.environ.get("HF_TOKEN", "")
|
| 21 |
if HF_TOKEN:
|
| 22 |
+
login(token=HF_TOKEN)
|
|
|
|
|
|
|
|
|
|
| 23 |
else:
|
| 24 |
print("⚠️ 未检测到 HF_TOKEN,私有模型可能下载失败")
|
| 25 |
|
|
|
|
| 122 |
# Gradio UI
|
| 123 |
# ------------------------------------------------------------------
|
| 124 |
custom_css = """
|
| 125 |
+
.label-container {
|
| 126 |
+
max-height: 300px;
|
| 127 |
+
overflow-y: auto;
|
| 128 |
+
border: 1px solid #ddd;
|
| 129 |
+
padding: 10px;
|
| 130 |
+
border-radius: 5px;
|
| 131 |
+
background-color: #f9f9f9;
|
| 132 |
+
}
|
| 133 |
+
.tag-item {
|
| 134 |
+
display: flex;
|
| 135 |
+
justify-content: space-between;
|
| 136 |
+
align-items: center;
|
| 137 |
+
margin: 2px 0;
|
| 138 |
+
padding: 2px 5px;
|
| 139 |
+
border-radius: 3px;
|
| 140 |
+
background-color: #fff;
|
| 141 |
+
transition: background-color 0.2s;
|
| 142 |
+
}
|
| 143 |
+
.tag-item:hover {
|
| 144 |
+
background-color: #f0f0f0;
|
| 145 |
+
}
|
| 146 |
+
.tag-en {
|
| 147 |
+
font-weight: bold;
|
| 148 |
+
color: #333;
|
| 149 |
+
cursor: pointer; /* Indicates clickable */
|
| 150 |
+
}
|
| 151 |
+
.tag-zh {
|
| 152 |
+
color: #666;
|
| 153 |
+
margin-left: 10px;
|
| 154 |
+
}
|
| 155 |
+
.tag-score {
|
| 156 |
+
color: #999;
|
| 157 |
+
font-size: 0.9em;
|
| 158 |
+
}
|
| 159 |
+
.btn-analyze-container { /* Custom class for analyze button container */
|
| 160 |
+
margin-top: 15px;
|
| 161 |
+
margin-bottom: 15px;
|
| 162 |
+
}
|
| 163 |
"""
|
| 164 |
|
| 165 |
_js_functions = """
|
| 166 |
function copyToClipboard(text) {
|
| 167 |
+
// --- 调试信息 ---
|
| 168 |
+
console.log('copyToClipboard function was called.');
|
| 169 |
+
console.log('Received text:', text);
|
| 170 |
+
// console.trace(); // 如果需要更详细的调用栈信息,可以取消这行注释
|
| 171 |
+
|
| 172 |
+
// --- 保护性检查 ---
|
| 173 |
+
// 如果 text 未定义或为 null,则不执行后续操作,并打印警告
|
| 174 |
if (typeof text === 'undefined' || text === null) {
|
| 175 |
+
console.warn('copyToClipboard was called with undefined or null text. Aborting this specific copy operation.');
|
| 176 |
+
// 在这种情况下,我们不应该尝试复制,也不应该显示“已复制”的提示
|
| 177 |
return;
|
| 178 |
}
|
| 179 |
+
|
| 180 |
navigator.clipboard.writeText(text).then(() => {
|
| 181 |
+
// console.log('Tag copied to clipboard: ' + text); // 成功复制的日志(可选)
|
| 182 |
const feedback = document.createElement('div');
|
| 183 |
+
|
| 184 |
+
// 确保 text 是字符串类型,再进行 substring 操作
|
| 185 |
+
let displayText = String(text); // 将 text 转换为字符串以防万一
|
| 186 |
displayText = displayText.substring(0, 30) + (displayText.length > 30 ? '...' : '');
|
| 187 |
+
|
| 188 |
feedback.textContent = '已复制: ' + displayText;
|
| 189 |
+
feedback.style.position = 'fixed';
|
| 190 |
+
feedback.style.bottom = '20px';
|
| 191 |
+
feedback.style.left = '50%';
|
| 192 |
+
feedback.style.transform = 'translateX(-50%)';
|
| 193 |
+
feedback.style.backgroundColor = '#4CAF50';
|
| 194 |
+
feedback.style.color = 'white';
|
| 195 |
+
feedback.style.padding = '10px 20px';
|
| 196 |
+
feedback.style.borderRadius = '5px';
|
| 197 |
+
feedback.style.zIndex = '10000';
|
| 198 |
+
feedback.style.transition = 'opacity 0.5s ease-out';
|
| 199 |
document.body.appendChild(feedback);
|
| 200 |
setTimeout(() => {
|
| 201 |
feedback.style.opacity = '0';
|
| 202 |
+
setTimeout(() => {
|
| 203 |
+
if (document.body.contains(feedback)) { // 确保元素还在DOM中
|
| 204 |
+
document.body.removeChild(feedback);
|
| 205 |
+
}
|
| 206 |
+
}, 500);
|
| 207 |
}, 1500);
|
| 208 |
}).catch(err => {
|
| 209 |
+
console.error('Failed to copy tag. Error:', err, 'Attempted to copy text:', text);
|
| 210 |
+
// 可以考虑也给用户一��错误提示,但原版 alert 可能体验不佳
|
| 211 |
+
// alert('复制失败: ' + err);
|
| 212 |
+
const errorFeedback = document.createElement('div');
|
| 213 |
+
errorFeedback.textContent = '复制操作失败!'; // 更友好的错误提示
|
| 214 |
+
errorFeedback.style.position = 'fixed';
|
| 215 |
+
errorFeedback.style.bottom = '20px';
|
| 216 |
+
errorFeedback.style.left = '50%';
|
| 217 |
+
errorFeedback.style.transform = 'translateX(-50%)';
|
| 218 |
+
errorFeedback.style.backgroundColor = '#D32F2F'; // 红色背景表示错误
|
| 219 |
+
errorFeedback.style.color = 'white';
|
| 220 |
+
errorFeedback.style.padding = '10px 20px';
|
| 221 |
+
errorFeedback.style.borderRadius = '5px';
|
| 222 |
+
errorFeedback.style.zIndex = '10000';
|
| 223 |
+
errorFeedback.style.transition = 'opacity 0.5s ease-out';
|
| 224 |
+
document.body.appendChild(errorFeedback);
|
| 225 |
+
setTimeout(() => {
|
| 226 |
+
errorFeedback.style.opacity = '0';
|
| 227 |
+
setTimeout(() => {
|
| 228 |
+
if (document.body.contains(errorFeedback)) {
|
| 229 |
+
document.body.removeChild(errorFeedback);
|
| 230 |
+
}
|
| 231 |
+
}, 500);
|
| 232 |
+
}, 2500);
|
| 233 |
});
|
| 234 |
}
|
| 235 |
"""
|
|
|
|
| 253 |
gen_slider = gr.Slider(0, 1, value=0.35, step=0.01, label="通用标签阈值", info="越高 → 标签更少更准")
|
| 254 |
char_slider = gr.Slider(0, 1, value=0.85, step=0.01, label="角色标签阈值", info="推荐保持较高阈值")
|
| 255 |
show_tag_scores = gr.Checkbox(True, label="在列表中显示标签置信度")
|
| 256 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 257 |
with gr.Accordion("📊 标签汇总设置", open=True):
|
| 258 |
gr.Markdown("选择要包含在下方汇总文本框中的标签类别:")
|
| 259 |
with gr.Row():
|
|
|
|
| 283 |
show_copy_button=True
|
| 284 |
)
|
| 285 |
|
| 286 |
+
# ----------------- 辅助函数 -----------------
|
| 287 |
+
def format_tags_html(tags_dict, translations_list, category_name, show_scores=True, show_translation_in_list=True):
|
| 288 |
if not tags_dict:
|
| 289 |
return "<p>暂无标签</p>"
|
| 290 |
|
| 291 |
html = '<div class="label-container">'
|
| 292 |
+
|
| 293 |
+
if not isinstance(translations_list, list):
|
| 294 |
+
translations_list = []
|
| 295 |
+
|
| 296 |
tag_keys = list(tags_dict.keys())
|
| 297 |
|
| 298 |
for i, tag in enumerate(tag_keys):
|
| 299 |
score = tags_dict[tag]
|
| 300 |
+
escaped_tag = tag.replace("'", "\\'") # Escape for JS
|
| 301 |
|
| 302 |
html += '<div class="tag-item">'
|
| 303 |
tag_display_html = f'<span class="tag-en" onclick="copyToClipboard(\'{escaped_tag}\')">{tag}</span>'
|
| 304 |
|
| 305 |
+
if show_translation_in_list and i < len(translations_list) and translations_list[i]:
|
| 306 |
+
tag_display_html += f'<span class="tag-zh">({translations_list[i]})</span>'
|
|
|
|
|
|
|
| 307 |
|
| 308 |
html += f'<div>{tag_display_html}</div>'
|
| 309 |
if show_scores:
|
|
|
|
| 316 |
current_res, current_translations_dict,
|
| 317 |
s_gen, s_char, s_rat, s_sep_type, s_show_zh
|
| 318 |
):
|
| 319 |
+
if not current_res:
|
| 320 |
+
return "请先分析图像或选择要汇总的标签类别。"
|
| 321 |
|
| 322 |
summary_parts = []
|
| 323 |
+
separators = {"逗号": ", ", "换行": "\n", "空格": " "}
|
| 324 |
+
separator = separators.get(s_sep_type, ", ")
|
| 325 |
|
| 326 |
categories_to_summarize = []
|
| 327 |
if s_gen: categories_to_summarize.append("general")
|
| 328 |
if s_char: categories_to_summarize.append("characters")
|
| 329 |
if s_rat: categories_to_summarize.append("ratings")
|
| 330 |
|
| 331 |
+
if not categories_to_summarize:
|
| 332 |
+
return "请至少选择一个标签类别进行汇总。"
|
| 333 |
|
| 334 |
for cat_key in categories_to_summarize:
|
| 335 |
if current_res.get(cat_key):
|
|
|
|
| 338 |
cat_translations = current_translations_dict.get(cat_key, [])
|
| 339 |
|
| 340 |
for i, en_tag in enumerate(cat_tags_en):
|
| 341 |
+
if s_show_zh and i < len(cat_translations) and cat_translations[i]:
|
| 342 |
+
tags_to_join.append(f"{en_tag}({cat_translations[i]})")
|
|
|
|
|
|
|
| 343 |
else:
|
| 344 |
tags_to_join.append(en_tag)
|
| 345 |
+
if tags_to_join: # only add if there are tags for this category
|
| 346 |
summary_parts.append(separator.join(tags_to_join))
|
| 347 |
|
| 348 |
+
# Join parts with double newline for readability if multiple categories present and separator is not newline
|
| 349 |
+
joiner = "\n\n" if separator != "\n" and len(summary_parts) > 1 else separator if separator == "\n" else " "
|
| 350 |
+
|
| 351 |
final_summary = joiner.join(summary_parts)
|
| 352 |
return final_summary if final_summary else "选定的类别中没有找到标签。"
|
| 353 |
|
| 354 |
+
|
| 355 |
+
# ----------------- 主要处理回调 -----------------
|
| 356 |
def process_image_and_generate_outputs(
|
| 357 |
img, g_th, c_th, s_scores, # Main inputs
|
| 358 |
+
s_gen, s_char, s_rat, s_sep, s_zh_in_sum
|
|
|
|
|
|
|
| 359 |
):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 360 |
if img is None:
|
| 361 |
+
yield (
|
| 362 |
+
gr.update(interactive=True, value="🚀 开始分析"),
|
| 363 |
+
gr.update(visible=True, value="❌ 请先上传图片。"),
|
| 364 |
+
"", "", "", "",
|
| 365 |
+
gr.update(placeholder="请先上传图片并开始分析..."),
|
| 366 |
+
{}, {}, {}
|
| 367 |
+
)
|
| 368 |
return
|
| 369 |
|
| 370 |
if tagger_instance is None:
|
| 371 |
+
yield (
|
| 372 |
+
gr.update(interactive=True, value="🚀 开始分析"),
|
| 373 |
+
gr.update(visible=True, value="❌ 分析器未成功初始化,请检查控制台错误。"),
|
| 374 |
+
"", "", "", "",
|
| 375 |
+
gr.update(placeholder="分析器初始化失败..."),
|
| 376 |
+
{}, {}, {}
|
| 377 |
+
)
|
| 378 |
return
|
| 379 |
|
| 380 |
yield (
|
| 381 |
gr.update(interactive=False, value="🔄 处理中..."),
|
| 382 |
gr.update(visible=True, value="🔄 正在分析图像,请稍候..."),
|
| 383 |
+
gr.HTML(value="<p>分析中...</p>"), # General
|
| 384 |
+
gr.HTML(value="<p>分析中...</p>"), # Character
|
| 385 |
+
gr.HTML(value="<p>分析中...</p>"), # Rating
|
| 386 |
+
gr.update(value="分析中,请稍候..."), # Summary
|
| 387 |
+
{}, {}, {} # Clear states initially
|
| 388 |
)
|
| 389 |
|
| 390 |
try:
|
| 391 |
+
# 1. Predict tags
|
| 392 |
res, tag_categories_original_order = tagger_instance.predict(img, g_th, c_th)
|
| 393 |
+
|
| 394 |
+
all_tags_to_translate = []
|
| 395 |
+
for cat_key in ["general", "characters", "ratings"]:
|
| 396 |
+
all_tags_to_translate.extend(tag_categories_original_order.get(cat_key, []))
|
| 397 |
+
|
| 398 |
+
all_translations_flat = []
|
| 399 |
+
if all_tags_to_translate:
|
| 400 |
+
all_translations_flat = translate_texts(all_tags_to_translate, src_lang="auto", tgt_lang="zh")
|
| 401 |
|
| 402 |
current_translations_dict = {}
|
| 403 |
+
offset = 0
|
| 404 |
+
for cat_key in ["general", "characters", "ratings"]:
|
| 405 |
+
cat_original_tags = tag_categories_original_order.get(cat_key, [])
|
| 406 |
+
num_tags_in_cat = len(cat_original_tags)
|
| 407 |
+
if num_tags_in_cat > 0:
|
| 408 |
+
current_translations_dict[cat_key] = all_translations_flat[offset : offset + num_tags_in_cat]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 409 |
offset += num_tags_in_cat
|
| 410 |
+
else:
|
| 411 |
+
current_translations_dict[cat_key] = []
|
|
|
|
| 412 |
|
| 413 |
+
general_html = format_tags_html(res.get("general", {}), current_translations_dict.get("general", []), "general", s_scores, True)
|
| 414 |
+
char_html = format_tags_html(res.get("characters", {}), current_translations_dict.get("characters", []), "characters", s_scores, True)
|
| 415 |
+
rating_html = format_tags_html(res.get("ratings", {}), current_translations_dict.get("ratings", []), "ratings", s_scores, True)
|
| 416 |
|
| 417 |
+
summary_text = generate_summary_text_content(
|
| 418 |
+
res, current_translations_dict,
|
| 419 |
+
s_gen, s_char, s_rat, s_sep, s_zh_in_sum
|
| 420 |
+
)
|
| 421 |
|
| 422 |
yield (
|
| 423 |
+
gr.update(interactive=True, value="🚀 开始分析"),
|
| 424 |
+
gr.update(visible=True, value="✅ 分析完成!"),
|
| 425 |
+
general_html,
|
| 426 |
+
char_html,
|
| 427 |
+
rating_html,
|
| 428 |
+
gr.update(value=summary_text),
|
| 429 |
+
res,
|
| 430 |
+
current_translations_dict,
|
| 431 |
+
tag_categories_original_order
|
| 432 |
)
|
| 433 |
|
| 434 |
except Exception as e:
|
|
|
|
| 436 |
tb_str = traceback.format_exc()
|
| 437 |
print(f"处理时发生错误: {e}\n{tb_str}")
|
| 438 |
yield (
|
|
|
|
| 439 |
gr.update(interactive=True, value="🚀 开始分析"),
|
| 440 |
+
gr.update(visible=True, value=f"❌ 处理失败: {str(e)}"),
|
| 441 |
"<p>处理出错</p>", "<p>处理出错</p>", "<p>处理出错</p>",
|
| 442 |
gr.update(value=f"错误: {str(e)}", placeholder="分析失败..."),
|
| 443 |
{}, {}, {}
|
| 444 |
)
|
| 445 |
|
| 446 |
+
# ----------------- 更新汇总文本的回调 -----------------
|
| 447 |
def update_summary_display(
|
| 448 |
s_gen, s_char, s_rat, s_sep, s_zh_in_sum,
|
| 449 |
current_res_from_state, current_translations_from_state
|
|
|
|
| 457 |
)
|
| 458 |
return gr.update(value=new_summary_text)
|
| 459 |
|
| 460 |
+
# ----------------- 绑定事件 -----------------
|
|
|
|
|
|
|
| 461 |
btn.click(
|
| 462 |
process_image_and_generate_outputs,
|
| 463 |
+
inputs=[
|
| 464 |
+
img_in, gen_slider, char_slider, show_tag_scores,
|
| 465 |
+
sum_general, sum_char, sum_rating, sum_sep, sum_show_zh
|
| 466 |
+
],
|
| 467 |
+
outputs=[
|
| 468 |
+
btn, processing_info,
|
| 469 |
+
out_general, out_char, out_rating,
|
| 470 |
+
out_summary,
|
| 471 |
+
state_res, state_translations_dict, state_tag_categories_for_translation
|
| 472 |
+
],
|
| 473 |
+
# show_progress="full" # Gradio's built-in progress
|
| 474 |
)
|
| 475 |
|
| 476 |
summary_controls = [sum_general, sum_char, sum_rating, sum_sep, sum_show_zh]
|
|
|
|
| 478 |
ctrl.change(
|
| 479 |
fn=update_summary_display,
|
| 480 |
inputs=summary_controls + [state_res, state_translations_dict],
|
| 481 |
+
outputs=[out_summary],
|
| 482 |
+
# show_progress=False # Typically fast, no need for progress indicator
|
| 483 |
)
|
| 484 |
+
|
| 485 |
+
# ------------------------------------------------------------------
|
| 486 |
+
# 启动
|
| 487 |
+
# ------------------------------------------------------------------
|
| 488 |
if __name__ == "__main__":
|
| 489 |
if tagger_instance is None:
|
| 490 |
print("CRITICAL: Tagger 未能初始化,应用功能将受限。请检查之前的错误信息。")
|
translator.py
CHANGED
|
@@ -1,141 +1,160 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import hashlib, hmac, json, os, random, time
|
| 2 |
from datetime import datetime
|
| 3 |
-
from typing import List, Sequence, Optional
|
| 4 |
|
| 5 |
import requests
|
| 6 |
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
|
|
|
| 10 |
TENCENT_SECRET_KEY = os.environ.get("TENCENT_SECRET_KEY")
|
| 11 |
TENCENT_TRANSLATE_URL = os.environ.get("TENCENT_TRANSLATE_URL", "https://tmt.tencentcloudapi.com")
|
| 12 |
|
| 13 |
BAIDU_TRANSLATE_URL = os.environ.get("BAIDU_TRANSLATE_URL", "https://fanyi-api.baidu.com/api/trans/vip/translate")
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
|
|
|
|
|
|
|
|
|
|
| 19 |
def _sign(key: bytes, msg: str) -> bytes:
|
| 20 |
return hmac.new(key, msg.encode("utf-8"), hashlib.sha256).digest()
|
| 21 |
|
| 22 |
def _tc3_signature(secret_key: str, date: str, service: str, string_to_sign: str) -> str:
|
| 23 |
-
secret_date
|
| 24 |
-
secret_service
|
| 25 |
-
secret_signing
|
| 26 |
return hmac.new(secret_signing, string_to_sign.encode("utf-8"), hashlib.sha256).hexdigest()
|
| 27 |
|
| 28 |
-
def _translate_with_tencent(texts: Sequence[str], src
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
payload_str = json.dumps(payload, ensure_ascii=False)
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
headers = {
|
| 53 |
-
"Authorization":
|
| 54 |
-
"
|
| 55 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
}
|
| 57 |
|
|
|
|
| 58 |
try:
|
| 59 |
resp = requests.post(TENCENT_TRANSLATE_URL, headers=headers, data=payload_str, timeout=8)
|
| 60 |
resp.raise_for_status()
|
| 61 |
data = resp.json()
|
| 62 |
-
|
| 63 |
-
return data["Response"]["TargetText"].split("\n")
|
| 64 |
-
else:
|
| 65 |
-
print(f"[translator] Tencent API abnormal response: {data}")
|
| 66 |
-
return None
|
| 67 |
except Exception as e:
|
| 68 |
print(f"[translator] Tencent API error → {e}")
|
| 69 |
return None
|
| 70 |
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
salt
|
| 80 |
query = "\n".join(texts)
|
| 81 |
-
sign
|
| 82 |
-
params = {
|
| 83 |
-
|
|
|
|
|
|
|
| 84 |
try:
|
| 85 |
resp = requests.get(BAIDU_TRANSLATE_URL, params=params, timeout=8)
|
| 86 |
resp.raise_for_status()
|
| 87 |
data = resp.json()
|
| 88 |
-
|
| 89 |
-
return [item["dst"] for item in data["trans_result"]]
|
| 90 |
-
else:
|
| 91 |
-
print(f"[translator] Baidu API abnormal response: {data}")
|
| 92 |
-
return None
|
| 93 |
except Exception as e:
|
| 94 |
print(f"[translator] Baidu API error → {e}")
|
| 95 |
return None
|
| 96 |
|
| 97 |
-
|
|
|
|
|
|
|
| 98 |
def translate_texts(texts: Sequence[str],
|
| 99 |
src_lang: str = "auto",
|
| 100 |
-
tgt_lang: str = "zh"
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
baidu_creds_json_str: Optional[str] = None) -> List[str]:
|
| 105 |
if not texts:
|
| 106 |
return []
|
| 107 |
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
-
|
| 112 |
-
use_tencent_id, use_tencent_key = tencent_id, tencent_key
|
| 113 |
-
print("[translator] Using custom Tencent API key.")
|
| 114 |
-
if baidu_creds_json_str:
|
| 115 |
-
try:
|
| 116 |
-
creds = json.loads(baidu_creds_json_str)
|
| 117 |
-
if isinstance(creds, list) and all(isinstance(d, dict) for d in creds):
|
| 118 |
-
use_baidu_creds = creds
|
| 119 |
-
print("[translator] Using custom Baidu API key(s).")
|
| 120 |
-
else:
|
| 121 |
-
print("[translator] Warning: Custom Baidu credentials format is incorrect.")
|
| 122 |
-
except json.JSONDecodeError:
|
| 123 |
-
print("[translator] Warning: Failed to parse custom Baidu credentials JSON.")
|
| 124 |
-
|
| 125 |
-
elif TRANSLATOR_ACCESS_KEY and system_key_input == TRANSLATOR_ACCESS_KEY:
|
| 126 |
-
print("[translator] System access key validated. Using system-configured API keys.")
|
| 127 |
-
use_tencent_id, use_tencent_key = TENCENT_SECRET_ID, TENCENT_SECRET_KEY
|
| 128 |
-
use_baidu_creds = BAIDU_CREDENTIALS_DEFAULT
|
| 129 |
-
|
| 130 |
-
else:
|
| 131 |
-
print("[translator] Translation disabled: No valid API keys or system key provided.")
|
| 132 |
-
return list(texts)
|
| 133 |
-
|
| 134 |
-
translated_texts = None
|
| 135 |
-
if use_tencent_id and use_tencent_key:
|
| 136 |
-
translated_texts = _translate_with_tencent(texts, src_lang, tgt_lang, use_tencent_id, use_tencent_key)
|
| 137 |
-
|
| 138 |
-
if translated_texts is None and use_baidu_creds:
|
| 139 |
-
translated_texts = _translate_with_baidu(texts, src_lang, tgt_lang, use_baidu_creds)
|
| 140 |
-
|
| 141 |
-
return translated_texts or list(texts)
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
translator.py
|
| 3 |
+
腾讯云 + 百度翻译 API 轮询封装
|
| 4 |
+
⚠️ 需在 HF 空间的 “Variables” 页设置以下环境变量
|
| 5 |
+
------------------------------------------------------------------
|
| 6 |
+
TENCENT_SECRET_ID 腾讯云 SecretId
|
| 7 |
+
TENCENT_SECRET_KEY 腾讯云 SecretKey
|
| 8 |
+
TENCENT_TRANSLATE_URL (可选) 默认 https://tmt.tencentcloudapi.com
|
| 9 |
+
BAIDU_TRANSLATE_URL (可选) 默认 https://fanyi-api.baidu.com/api/trans/vip/translate
|
| 10 |
+
BAIDU_CREDENTIALS_JSON 形如:
|
| 11 |
+
[
|
| 12 |
+
{"app_id": "xxxx", "secret_key": "yyyy"},
|
| 13 |
+
{"app_id": "aaaa", "secret_key": "bbbb"}
|
| 14 |
+
]
|
| 15 |
+
------------------------------------------------------------------
|
| 16 |
+
"""
|
| 17 |
import hashlib, hmac, json, os, random, time
|
| 18 |
from datetime import datetime
|
| 19 |
+
from typing import List, Sequence, Optional
|
| 20 |
|
| 21 |
import requests
|
| 22 |
|
| 23 |
+
# ------------------------------------------------------------------
|
| 24 |
+
# 读取环境变量
|
| 25 |
+
# ------------------------------------------------------------------
|
| 26 |
+
TENCENT_SECRET_ID = os.environ.get("TENCENT_SECRET_ID")
|
| 27 |
TENCENT_SECRET_KEY = os.environ.get("TENCENT_SECRET_KEY")
|
| 28 |
TENCENT_TRANSLATE_URL = os.environ.get("TENCENT_TRANSLATE_URL", "https://tmt.tencentcloudapi.com")
|
| 29 |
|
| 30 |
BAIDU_TRANSLATE_URL = os.environ.get("BAIDU_TRANSLATE_URL", "https://fanyi-api.baidu.com/api/trans/vip/translate")
|
| 31 |
+
BAIDU_CREDENTIALS = json.loads(os.environ.get("BAIDU_CREDENTIALS_JSON", "[]"))
|
| 32 |
+
|
| 33 |
+
# 内部轮询索引
|
| 34 |
+
_baidu_idx: int = 0
|
| 35 |
+
def _next_baidu_cred():
|
| 36 |
+
global _baidu_idx
|
| 37 |
+
if not BAIDU_CREDENTIALS:
|
| 38 |
+
return None
|
| 39 |
+
cred = BAIDU_CREDENTIALS[_baidu_idx]
|
| 40 |
+
_baidu_idx = (_baidu_idx + 1) % len(BAIDU_CREDENTIALS)
|
| 41 |
+
return cred
|
| 42 |
|
| 43 |
+
# ------------------------------------------------------------------
|
| 44 |
+
# 腾讯翻译
|
| 45 |
+
# ------------------------------------------------------------------
|
| 46 |
def _sign(key: bytes, msg: str) -> bytes:
|
| 47 |
return hmac.new(key, msg.encode("utf-8"), hashlib.sha256).digest()
|
| 48 |
|
| 49 |
def _tc3_signature(secret_key: str, date: str, service: str, string_to_sign: str) -> str:
|
| 50 |
+
secret_date = _sign(("TC3" + secret_key).encode(), date)
|
| 51 |
+
secret_service = _sign(secret_date, service)
|
| 52 |
+
secret_signing = _sign(secret_service, "tc3_request")
|
| 53 |
return hmac.new(secret_signing, string_to_sign.encode("utf-8"), hashlib.sha256).hexdigest()
|
| 54 |
|
| 55 |
+
def _translate_with_tencent(texts: Sequence[str], src="auto", tgt="zh") -> Optional[List[str]]:
|
| 56 |
+
"""优先使用腾讯云翻译。失败返回 None"""
|
| 57 |
+
if not (TENCENT_SECRET_ID and TENCENT_SECRET_KEY):
|
| 58 |
+
return None # 未配置凭证
|
| 59 |
+
service = "tmt"
|
| 60 |
+
host = "tmt.tencentcloudapi.com"
|
| 61 |
+
action = "TextTranslate"
|
| 62 |
+
version = "2018-03-21"
|
| 63 |
+
region = "ap-beijing"
|
| 64 |
+
ts = int(time.time())
|
| 65 |
+
date = datetime.utcfromtimestamp(ts).strftime("%Y-%m-%d")
|
| 66 |
+
algorithm = "TC3-HMAC-SHA256"
|
| 67 |
+
|
| 68 |
+
payload = {
|
| 69 |
+
"SourceText": "\n".join(texts),
|
| 70 |
+
"Source": src,
|
| 71 |
+
"Target": tgt,
|
| 72 |
+
"ProjectId": 0,
|
| 73 |
+
}
|
| 74 |
payload_str = json.dumps(payload, ensure_ascii=False)
|
| 75 |
+
|
| 76 |
+
# ---------- step‑1 canonical request ----------
|
| 77 |
+
canonical_request = "\n".join([
|
| 78 |
+
"POST",
|
| 79 |
+
"/",
|
| 80 |
+
"",
|
| 81 |
+
f"content-type:application/json; charset=utf-8\nhost:{host}\nx-tc-action:{action.lower()}\n",
|
| 82 |
+
"content-type;host;x-tc-action",
|
| 83 |
+
hashlib.sha256(payload_str.encode()).hexdigest(),
|
| 84 |
+
])
|
| 85 |
+
|
| 86 |
+
# ---------- step‑2 string to sign ----------
|
| 87 |
+
credential_scope = f"{date}/{service}/tc3_request"
|
| 88 |
+
string_to_sign = "\n".join([
|
| 89 |
+
algorithm, str(ts), credential_scope,
|
| 90 |
+
hashlib.sha256(canonical_request.encode()).hexdigest(),
|
| 91 |
+
])
|
| 92 |
+
|
| 93 |
+
# ---------- step‑3 signature ----------
|
| 94 |
+
signature = _tc3_signature(TENCENT_SECRET_KEY, date, service, string_to_sign)
|
| 95 |
+
|
| 96 |
+
# ---------- step‑4 headers ----------
|
| 97 |
+
authorization = (
|
| 98 |
+
f"{algorithm} Credential={TENCENT_SECRET_ID}/{credential_scope}, "
|
| 99 |
+
f"SignedHeaders=content-type;host;x-tc-action, Signature={signature}"
|
| 100 |
+
)
|
| 101 |
headers = {
|
| 102 |
+
"Authorization": authorization,
|
| 103 |
+
"Content-Type": "application/json; charset=utf-8",
|
| 104 |
+
"Host": host,
|
| 105 |
+
"X-TC-Action": action,
|
| 106 |
+
"X-TC-Timestamp": str(ts),
|
| 107 |
+
"X-TC-Version": version,
|
| 108 |
+
"X-TC-Region": region,
|
| 109 |
}
|
| 110 |
|
| 111 |
+
# ---------- request ----------
|
| 112 |
try:
|
| 113 |
resp = requests.post(TENCENT_TRANSLATE_URL, headers=headers, data=payload_str, timeout=8)
|
| 114 |
resp.raise_for_status()
|
| 115 |
data = resp.json()
|
| 116 |
+
return data["Response"]["TargetText"].split("\n")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 117 |
except Exception as e:
|
| 118 |
print(f"[translator] Tencent API error → {e}")
|
| 119 |
return None
|
| 120 |
|
| 121 |
+
# ------------------------------------------------------------------
|
| 122 |
+
# 百度翻译
|
| 123 |
+
# ------------------------------------------------------------------
|
| 124 |
+
def _translate_with_baidu(texts: Sequence[str], src="auto", tgt="zh") -> Optional[List[str]]:
|
| 125 |
+
creds = _next_baidu_cred()
|
| 126 |
+
if creds is None:
|
| 127 |
+
return None # 未配置凭证
|
| 128 |
+
app_id, secret_key = creds["app_id"], creds["secret_key"]
|
| 129 |
+
salt = random.randint(32768, 65536)
|
| 130 |
query = "\n".join(texts)
|
| 131 |
+
sign = hashlib.md5((app_id + query + str(salt) + secret_key).encode()).hexdigest()
|
| 132 |
+
params = {
|
| 133 |
+
"q": query, "from": src, "to": tgt,
|
| 134 |
+
"appid": app_id, "salt": salt, "sign": sign,
|
| 135 |
+
}
|
| 136 |
try:
|
| 137 |
resp = requests.get(BAIDU_TRANSLATE_URL, params=params, timeout=8)
|
| 138 |
resp.raise_for_status()
|
| 139 |
data = resp.json()
|
| 140 |
+
return [item["dst"] for item in data["trans_result"]]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 141 |
except Exception as e:
|
| 142 |
print(f"[translator] Baidu API error → {e}")
|
| 143 |
return None
|
| 144 |
|
| 145 |
+
# ------------------------------------------------------------------
|
| 146 |
+
# 对外统一函数
|
| 147 |
+
# ------------------------------------------------------------------
|
| 148 |
def translate_texts(texts: Sequence[str],
|
| 149 |
src_lang: str = "auto",
|
| 150 |
+
tgt_lang: str = "zh") -> List[str]:
|
| 151 |
+
"""
|
| 152 |
+
优先 Tencent → 失败再 Baidu → 如果都失败,返回原文。
|
| 153 |
+
"""
|
|
|
|
| 154 |
if not texts:
|
| 155 |
return []
|
| 156 |
|
| 157 |
+
out = _translate_with_tencent(texts, src_lang, tgt_lang)
|
| 158 |
+
if out is None:
|
| 159 |
+
out = _translate_with_baidu(texts, src_lang, tgt_lang)
|
| 160 |
+
return out or list(texts)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|