Update app.py
Browse files
app.py
CHANGED
|
@@ -4,39 +4,61 @@ from bs4 import BeautifulSoup
|
|
| 4 |
import re
|
| 5 |
import os
|
| 6 |
import tempfile
|
| 7 |
-
import
|
| 8 |
-
from PIL import Image, ImageEnhance, ImageFilter
|
| 9 |
import numpy as np
|
| 10 |
from urllib.parse import urlparse
|
| 11 |
import time
|
| 12 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
# 确保中文显示正常
|
| 14 |
import matplotlib.pyplot as plt
|
| 15 |
plt.rcParams["font.family"] = ["SimHei", "WenQuanYi Micro Hei", "Heiti TC"]
|
| 16 |
|
| 17 |
-
# 设置Tesseract OCR路径(Hugging Face Spaces上已预安装)
|
| 18 |
-
try:
|
| 19 |
-
pytesseract.pytesseract.tesseract_cmd = r'/usr/bin/tesseract'
|
| 20 |
-
except:
|
| 21 |
-
pass # 在Windows上可能需要手动设置
|
| 22 |
-
|
| 23 |
def extract_gif_urls(html_content):
|
| 24 |
"""从HTML内容中提取符合条件的GIF图片URL"""
|
| 25 |
soup = BeautifulSoup(html_content, 'html.parser')
|
| 26 |
img_tags = soup.find_all('img')
|
| 27 |
|
| 28 |
gif_urls = []
|
| 29 |
-
# 匹配霹雳布袋戏相关的GIF格式,特别是0101.gif这类序列
|
| 30 |
# 放宽正则匹配条件,确保能识别到相关GIF
|
| 31 |
-
pattern = r'
|
| 32 |
|
| 33 |
for img in img_tags:
|
| 34 |
src = img.get('src', '')
|
| 35 |
-
|
| 36 |
-
if src and re.search(r'\d+\.gif$', src, re.IGNORECASE):
|
| 37 |
# 处理相对路径
|
| 38 |
if not src.startswith(('http://', 'https://')):
|
| 39 |
-
# 尝试补全相对路径(针对常见情况)
|
| 40 |
if src.startswith('/'):
|
| 41 |
parsed_url = urlparse(html_content.url) if hasattr(html_content, 'url') else None
|
| 42 |
if parsed_url:
|
|
@@ -47,17 +69,16 @@ def extract_gif_urls(html_content):
|
|
| 47 |
continue
|
| 48 |
gif_urls.append(src)
|
| 49 |
|
| 50 |
-
#
|
| 51 |
try:
|
| 52 |
gif_urls.sort(key=lambda x: int(re.search(r'(\d+)\.gif', x).group(1)))
|
| 53 |
except:
|
| 54 |
-
pass
|
| 55 |
return gif_urls
|
| 56 |
|
| 57 |
def download_gif(url, save_path):
|
| 58 |
"""下载GIF图片"""
|
| 59 |
try:
|
| 60 |
-
# 添加请求头,模拟浏览器行为
|
| 61 |
headers = {
|
| 62 |
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
|
| 63 |
}
|
|
@@ -73,20 +94,21 @@ def download_gif(url, save_path):
|
|
| 73 |
|
| 74 |
def process_gif_for_ocr(gif_path):
|
| 75 |
"""处理GIF图片以提高OCR识别率"""
|
| 76 |
-
|
|
|
|
|
|
|
| 77 |
try:
|
| 78 |
gif = Image.open(gif_path)
|
| 79 |
|
| 80 |
-
#
|
| 81 |
frames = []
|
| 82 |
try:
|
| 83 |
-
for i in range(10):
|
| 84 |
gif.seek(i)
|
| 85 |
-
frames.append(gif.convert('L'))
|
| 86 |
except EOFError:
|
| 87 |
pass
|
| 88 |
|
| 89 |
-
# 如果没有获取到帧,返回None
|
| 90 |
if not frames:
|
| 91 |
return None
|
| 92 |
|
|
@@ -100,7 +122,7 @@ def process_gif_for_ocr(gif_path):
|
|
| 100 |
# 轻微锐化
|
| 101 |
frame = frame.filter(ImageFilter.SHARPEN)
|
| 102 |
|
| 103 |
-
#
|
| 104 |
threshold = 140
|
| 105 |
frame = frame.point(lambda p: p > threshold and 255)
|
| 106 |
|
|
@@ -111,11 +133,10 @@ def process_gif_for_ocr(gif_path):
|
|
| 111 |
|
| 112 |
def ocr_image(image):
|
| 113 |
"""对处理后的图像进行OCR识别"""
|
| 114 |
-
if image is None:
|
| 115 |
-
return ""
|
| 116 |
|
| 117 |
try:
|
| 118 |
-
# 使用Tesseract进行OCR,指定中文识别,增加更多配置参数
|
| 119 |
custom_config = r'--oem 3 --psm 3 -l chi_sim+eng'
|
| 120 |
text = pytesseract.image_to_string(image, config=custom_config)
|
| 121 |
|
|
@@ -124,22 +145,23 @@ def ocr_image(image):
|
|
| 124 |
return text
|
| 125 |
except Exception as e:
|
| 126 |
print(f"OCR识别失败: {str(e)}")
|
| 127 |
-
return "OCR
|
| 128 |
|
| 129 |
def extract_text_from_url(url, progress=gr.Progress()):
|
| 130 |
"""从指定URL提取GIF并识别文本"""
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
try:
|
| 132 |
-
# 创建临时目录
|
| 133 |
with tempfile.TemporaryDirectory() as temp_dir:
|
| 134 |
progress(0, desc="正在获取网页内容...")
|
| 135 |
|
| 136 |
-
# 获取网页内容,添加请求头模拟浏览器
|
| 137 |
headers = {
|
| 138 |
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
|
| 139 |
}
|
| 140 |
response = requests.get(url, timeout=15, headers=headers)
|
| 141 |
if response.status_code != 200:
|
| 142 |
-
# 确保返回两个值:错误信息和空列表
|
| 143 |
return f"无法访问网页,状态码:{response.status_code}", []
|
| 144 |
|
| 145 |
# 提取GIF URL
|
|
@@ -147,7 +169,6 @@ def extract_text_from_url(url, progress=gr.Progress()):
|
|
| 147 |
gif_urls = extract_gif_urls(response.text)
|
| 148 |
|
| 149 |
if not gif_urls:
|
| 150 |
-
# 确保返回两个值:提示信息和空列表
|
| 151 |
return "未找到符合条件的GIF图片", []
|
| 152 |
|
| 153 |
progress(0.3, desc=f"找到{len(gif_urls)}个GIF图片,开始处理...")
|
|
@@ -159,11 +180,9 @@ def extract_text_from_url(url, progress=gr.Progress()):
|
|
| 159 |
current_progress = 0.3
|
| 160 |
|
| 161 |
for i, gif_url in enumerate(gif_urls):
|
| 162 |
-
# 更新进度
|
| 163 |
current_progress += step
|
| 164 |
progress(current_progress, desc=f"处理第{i+1}/{len(gif_urls)}个GIF...")
|
| 165 |
|
| 166 |
-
# 提取文件名
|
| 167 |
parsed_url = urlparse(gif_url)
|
| 168 |
filename = os.path.basename(parsed_url.path)
|
| 169 |
|
|
@@ -173,13 +192,13 @@ def extract_text_from_url(url, progress=gr.Progress()):
|
|
| 173 |
all_text.append(f"【{filename}】下载失败")
|
| 174 |
continue
|
| 175 |
|
| 176 |
-
# 处理GIF
|
| 177 |
processed_image = process_gif_for_ocr(gif_path)
|
| 178 |
if processed_image is None:
|
| 179 |
all_text.append(f"【{filename}】处理失败")
|
| 180 |
continue
|
| 181 |
|
| 182 |
-
#
|
| 183 |
processed_path = os.path.join(temp_dir, f"processed_{filename}.png")
|
| 184 |
processed_image.save(processed_path)
|
| 185 |
gif_images.append(Image.open(processed_path))
|
|
@@ -188,17 +207,13 @@ def extract_text_from_url(url, progress=gr.Progress()):
|
|
| 188 |
text = ocr_image(processed_image)
|
| 189 |
all_text.append(f"【{filename}】\n{text}")
|
| 190 |
|
| 191 |
-
# 避免请求过于频繁
|
| 192 |
time.sleep(0.5)
|
| 193 |
|
| 194 |
-
# 拼接所有文本
|
| 195 |
result_text = "\n\n".join(all_text)
|
| 196 |
-
|
| 197 |
progress(1.0, desc="处理完成")
|
| 198 |
return result_text, gif_images
|
| 199 |
|
| 200 |
except Exception as e:
|
| 201 |
-
# 确保返回两个值:错误信息和空列表
|
| 202 |
return f"处理过程出错:{str(e)}", []
|
| 203 |
|
| 204 |
def create_interface():
|
|
@@ -208,13 +223,16 @@ def create_interface():
|
|
| 208 |
# 霹雳布袋戏GIF文本提取工具
|
| 209 |
|
| 210 |
这个工具可以从指定的霹雳布袋戏相关网页中提取GIF图片,并识别其中的文本内容。
|
| 211 |
-
|
| 212 |
-
## 使用方法:
|
| 213 |
-
1. 输入包含GIF的网页URL(例如:https://pilicreateworld.tw-blog.com/PILI/PILI69/01.HTM)
|
| 214 |
-
2. 点击"提取文本"按钮
|
| 215 |
-
3. 等待处理完成,查看识别结果
|
| 216 |
""")
|
| 217 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 218 |
with gr.Row():
|
| 219 |
url_input = gr.Textbox(
|
| 220 |
label="网页URL",
|
|
@@ -234,16 +252,16 @@ def create_interface():
|
|
| 234 |
label="处理后的GIF帧",
|
| 235 |
show_label=True,
|
| 236 |
elem_id="gallery",
|
| 237 |
-
columns=2,
|
| 238 |
height="auto"
|
| 239 |
)
|
| 240 |
|
| 241 |
with gr.Row():
|
| 242 |
gr.Markdown("""
|
| 243 |
## 注意事项:
|
|
|
|
| 244 |
- 识别 accuracy 取决于GIF图片的清晰度
|
| 245 |
- 处理可能需要几分钟时间,请耐心等待
|
| 246 |
-
- 如遇网络问题,请检查URL是否正确或稍后重试
|
| 247 |
""")
|
| 248 |
|
| 249 |
# 设置事件
|
|
|
|
| 4 |
import re
|
| 5 |
import os
|
| 6 |
import tempfile
|
| 7 |
+
import subprocess
|
|
|
|
| 8 |
import numpy as np
|
| 9 |
from urllib.parse import urlparse
|
| 10 |
import time
|
| 11 |
|
| 12 |
+
# 尝试安装Tesseract(仅在Hugging Face Spaces环境中有效)
|
| 13 |
+
def install_tesseract():
|
| 14 |
+
try:
|
| 15 |
+
# 检查Tesseract是否已安装
|
| 16 |
+
subprocess.run(['tesseract', '--version'], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
| 17 |
+
return True
|
| 18 |
+
except (FileNotFoundError, subprocess.CalledProcessError):
|
| 19 |
+
print("Tesseract未安装,尝试自动安装...")
|
| 20 |
+
try:
|
| 21 |
+
# 在Ubuntu/Debian系统上安装Tesseract
|
| 22 |
+
subprocess.run(['apt-get', 'update'], check=True)
|
| 23 |
+
subprocess.run(['apt-get', 'install', '-y', 'tesseract-ocr', 'tesseract-ocr-chi-sim'], check=True)
|
| 24 |
+
# 安装Python绑定
|
| 25 |
+
subprocess.run(['pip', 'install', 'pytesseract'], check=True)
|
| 26 |
+
return True
|
| 27 |
+
except Exception as e:
|
| 28 |
+
print(f"自动安装Tesseract失败: {str(e)}")
|
| 29 |
+
return False
|
| 30 |
+
|
| 31 |
+
# 检查并安装Tesseract
|
| 32 |
+
tesseract_available = install_tesseract()
|
| 33 |
+
|
| 34 |
+
# 只有在Tesseract可用时才导入相关库
|
| 35 |
+
if tesseract_available:
|
| 36 |
+
import pytesseract
|
| 37 |
+
from PIL import Image, ImageEnhance, ImageFilter
|
| 38 |
+
# 设置Tesseract OCR路径
|
| 39 |
+
try:
|
| 40 |
+
pytesseract.pytesseract.tesseract_cmd = subprocess.check_output(['which', 'tesseract']).decode().strip()
|
| 41 |
+
except:
|
| 42 |
+
pytesseract.pytesseract.tesseract_cmd = r'/usr/bin/tesseract'
|
| 43 |
+
|
| 44 |
# 确保中文显示正常
|
| 45 |
import matplotlib.pyplot as plt
|
| 46 |
plt.rcParams["font.family"] = ["SimHei", "WenQuanYi Micro Hei", "Heiti TC"]
|
| 47 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
def extract_gif_urls(html_content):
|
| 49 |
"""从HTML内容中提取符合条件的GIF图片URL"""
|
| 50 |
soup = BeautifulSoup(html_content, 'html.parser')
|
| 51 |
img_tags = soup.find_all('img')
|
| 52 |
|
| 53 |
gif_urls = []
|
|
|
|
| 54 |
# 放宽正则匹配条件,确保能识别到相关GIF
|
| 55 |
+
pattern = r'\d+\.gif$'
|
| 56 |
|
| 57 |
for img in img_tags:
|
| 58 |
src = img.get('src', '')
|
| 59 |
+
if src and re.search(pattern, src, re.IGNORECASE):
|
|
|
|
| 60 |
# 处理相对路径
|
| 61 |
if not src.startswith(('http://', 'https://')):
|
|
|
|
| 62 |
if src.startswith('/'):
|
| 63 |
parsed_url = urlparse(html_content.url) if hasattr(html_content, 'url') else None
|
| 64 |
if parsed_url:
|
|
|
|
| 69 |
continue
|
| 70 |
gif_urls.append(src)
|
| 71 |
|
| 72 |
+
# 按文件名排序
|
| 73 |
try:
|
| 74 |
gif_urls.sort(key=lambda x: int(re.search(r'(\d+)\.gif', x).group(1)))
|
| 75 |
except:
|
| 76 |
+
pass
|
| 77 |
return gif_urls
|
| 78 |
|
| 79 |
def download_gif(url, save_path):
|
| 80 |
"""下载GIF图片"""
|
| 81 |
try:
|
|
|
|
| 82 |
headers = {
|
| 83 |
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
|
| 84 |
}
|
|
|
|
| 94 |
|
| 95 |
def process_gif_for_ocr(gif_path):
|
| 96 |
"""处理GIF图片以提高OCR识别率"""
|
| 97 |
+
if not tesseract_available:
|
| 98 |
+
return None
|
| 99 |
+
|
| 100 |
try:
|
| 101 |
gif = Image.open(gif_path)
|
| 102 |
|
| 103 |
+
# 尝试提取多个帧
|
| 104 |
frames = []
|
| 105 |
try:
|
| 106 |
+
for i in range(10):
|
| 107 |
gif.seek(i)
|
| 108 |
+
frames.append(gif.convert('L'))
|
| 109 |
except EOFError:
|
| 110 |
pass
|
| 111 |
|
|
|
|
| 112 |
if not frames:
|
| 113 |
return None
|
| 114 |
|
|
|
|
| 122 |
# 轻微锐化
|
| 123 |
frame = frame.filter(ImageFilter.SHARPEN)
|
| 124 |
|
| 125 |
+
# 二值化处理
|
| 126 |
threshold = 140
|
| 127 |
frame = frame.point(lambda p: p > threshold and 255)
|
| 128 |
|
|
|
|
| 133 |
|
| 134 |
def ocr_image(image):
|
| 135 |
"""对处理后的图像进行OCR识别"""
|
| 136 |
+
if not tesseract_available or image is None:
|
| 137 |
+
return "Tesseract OCR未安装,无法识别文本"
|
| 138 |
|
| 139 |
try:
|
|
|
|
| 140 |
custom_config = r'--oem 3 --psm 3 -l chi_sim+eng'
|
| 141 |
text = pytesseract.image_to_string(image, config=custom_config)
|
| 142 |
|
|
|
|
| 145 |
return text
|
| 146 |
except Exception as e:
|
| 147 |
print(f"OCR识别失败: {str(e)}")
|
| 148 |
+
return f"OCR识别失败: {str(e)}"
|
| 149 |
|
| 150 |
def extract_text_from_url(url, progress=gr.Progress()):
|
| 151 |
"""从指定URL提取GIF并识别文本"""
|
| 152 |
+
# 检查Tesseract是否可用
|
| 153 |
+
if not tesseract_available:
|
| 154 |
+
return "Tesseract OCR安装失败,无法进行文本识别。请联系管理员解决此问题。", []
|
| 155 |
+
|
| 156 |
try:
|
|
|
|
| 157 |
with tempfile.TemporaryDirectory() as temp_dir:
|
| 158 |
progress(0, desc="正在获取网页内容...")
|
| 159 |
|
|
|
|
| 160 |
headers = {
|
| 161 |
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
|
| 162 |
}
|
| 163 |
response = requests.get(url, timeout=15, headers=headers)
|
| 164 |
if response.status_code != 200:
|
|
|
|
| 165 |
return f"无法访问网页,状态码:{response.status_code}", []
|
| 166 |
|
| 167 |
# 提取GIF URL
|
|
|
|
| 169 |
gif_urls = extract_gif_urls(response.text)
|
| 170 |
|
| 171 |
if not gif_urls:
|
|
|
|
| 172 |
return "未找到符合条件的GIF图片", []
|
| 173 |
|
| 174 |
progress(0.3, desc=f"找到{len(gif_urls)}个GIF图片,开始处理...")
|
|
|
|
| 180 |
current_progress = 0.3
|
| 181 |
|
| 182 |
for i, gif_url in enumerate(gif_urls):
|
|
|
|
| 183 |
current_progress += step
|
| 184 |
progress(current_progress, desc=f"处理第{i+1}/{len(gif_urls)}个GIF...")
|
| 185 |
|
|
|
|
| 186 |
parsed_url = urlparse(gif_url)
|
| 187 |
filename = os.path.basename(parsed_url.path)
|
| 188 |
|
|
|
|
| 192 |
all_text.append(f"【{filename}】下载失败")
|
| 193 |
continue
|
| 194 |
|
| 195 |
+
# 处理GIF
|
| 196 |
processed_image = process_gif_for_ocr(gif_path)
|
| 197 |
if processed_image is None:
|
| 198 |
all_text.append(f"【{filename}】处理失败")
|
| 199 |
continue
|
| 200 |
|
| 201 |
+
# 保存处理后的图像
|
| 202 |
processed_path = os.path.join(temp_dir, f"processed_{filename}.png")
|
| 203 |
processed_image.save(processed_path)
|
| 204 |
gif_images.append(Image.open(processed_path))
|
|
|
|
| 207 |
text = ocr_image(processed_image)
|
| 208 |
all_text.append(f"【{filename}】\n{text}")
|
| 209 |
|
|
|
|
| 210 |
time.sleep(0.5)
|
| 211 |
|
|
|
|
| 212 |
result_text = "\n\n".join(all_text)
|
|
|
|
| 213 |
progress(1.0, desc="处理完成")
|
| 214 |
return result_text, gif_images
|
| 215 |
|
| 216 |
except Exception as e:
|
|
|
|
| 217 |
return f"处理过程出错:{str(e)}", []
|
| 218 |
|
| 219 |
def create_interface():
|
|
|
|
| 223 |
# 霹雳布袋戏GIF文本提取工具
|
| 224 |
|
| 225 |
这个工具可以从指定的霹雳布袋戏相关网页中提取GIF图片,并识别其中的文本内容。
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 226 |
""")
|
| 227 |
|
| 228 |
+
# 显示Tesseract状态
|
| 229 |
+
if not tesseract_available:
|
| 230 |
+
gr.Markdown("""
|
| 231 |
+
<div style="background-color: #ffebee; padding: 10px; border-radius: 5px; color: #b71c1c;">
|
| 232 |
+
⚠️ 注意:Tesseract OCR引擎安装失败,可能无法正常识别文本。
|
| 233 |
+
</div>
|
| 234 |
+
""")
|
| 235 |
+
|
| 236 |
with gr.Row():
|
| 237 |
url_input = gr.Textbox(
|
| 238 |
label="网页URL",
|
|
|
|
| 252 |
label="处理后的GIF帧",
|
| 253 |
show_label=True,
|
| 254 |
elem_id="gallery",
|
| 255 |
+
columns=2,
|
| 256 |
height="auto"
|
| 257 |
)
|
| 258 |
|
| 259 |
with gr.Row():
|
| 260 |
gr.Markdown("""
|
| 261 |
## 注意事项:
|
| 262 |
+
- 首次使用可能需要时间安装OCR组件
|
| 263 |
- 识别 accuracy 取决于GIF图片的清晰度
|
| 264 |
- 处理可能需要几分钟时间,请耐心等待
|
|
|
|
| 265 |
""")
|
| 266 |
|
| 267 |
# 设置事件
|