kitjesen commited on
Commit
a973131
·
verified ·
1 Parent(s): dd2116b

Upload download_models.py

Browse files
Files changed (1) hide show
  1. download_models.py +21 -13
download_models.py CHANGED
@@ -6,8 +6,9 @@ from modelscope import snapshot_download
6
 
7
 
8
  def download_json(url):
 
9
  response = requests.get(url)
10
- response.raise_for_status()
11
  return response.json()
12
 
13
 
@@ -30,11 +31,6 @@ def download_and_modify_json(url, local_filename, modifications):
30
 
31
 
32
  if __name__ == '__main__':
33
- # 在 Docker 环境中使用 /app 目录
34
- base_dir = "/app"
35
- models_dir = os.path.join(base_dir, "models")
36
- os.makedirs(models_dir, exist_ok=True)
37
-
38
  mineru_patterns = [
39
  "models/Layout/LayoutLMv3/*",
40
  "models/Layout/YOLO/*",
@@ -44,18 +40,30 @@ if __name__ == '__main__':
44
  "models/TabRec/StructEqTable/*",
45
  ]
46
 
47
- # 下载模型到 Docker 容器中的固定路径
48
- model_dir = snapshot_download('opendatalab/PDF-Extract-Kit-1.0',
49
- cache_dir=models_dir,
50
- allow_patterns=mineru_patterns)
 
 
 
 
 
 
 
 
 
 
 
 
 
51
  layoutreader_model_dir = snapshot_download('ppaanngggg/layoutreader',
52
- cache_dir=models_dir)
53
 
54
- model_dir = os.path.join(model_dir, 'models')
55
  print(f'model_dir is: {model_dir}')
56
  print(f'layoutreader_model_dir is: {layoutreader_model_dir}')
57
 
58
- # 配置文件保存到 Docker 容器的固定路径
59
  json_url = 'https://gitee.com/myhloli/MinerU/raw/master/magic-pdf.template.json'
60
  config_file_name = 'magic-pdf.json'
61
  config_file = os.path.join(base_dir, config_file_name)
 
6
 
7
 
8
  def download_json(url):
9
+ # 下载JSON文件
10
  response = requests.get(url)
11
+ response.raise_for_status() # 检查请求是否成功
12
  return response.json()
13
 
14
 
 
31
 
32
 
33
  if __name__ == '__main__':
 
 
 
 
 
34
  mineru_patterns = [
35
  "models/Layout/LayoutLMv3/*",
36
  "models/Layout/YOLO/*",
 
40
  "models/TabRec/StructEqTable/*",
41
  ]
42
 
43
+ # Docker 环境中使用 /app 目录
44
+ base_dir = "/app"
45
+ cache_dir = os.path.join(base_dir, "model_cache")
46
+ os.makedirs(cache_dir, exist_ok=True)
47
+
48
+ # 下载模型,使用与原代码相同的参数
49
+ try:
50
+ model_dir = snapshot_download('opendatalab/PDF-Extract-Kit-1.0',
51
+ cache_dir=cache_dir,
52
+ revision='master')
53
+ model_dir = os.path.join(model_dir, 'models')
54
+ except TypeError:
55
+ # 如果 allow_patterns 不支持,则下载完整模型
56
+ model_dir = snapshot_download('opendatalab/PDF-Extract-Kit-1.0',
57
+ cache_dir=cache_dir)
58
+ model_dir = os.path.join(model_dir, 'models')
59
+
60
  layoutreader_model_dir = snapshot_download('ppaanngggg/layoutreader',
61
+ cache_dir=cache_dir)
62
 
 
63
  print(f'model_dir is: {model_dir}')
64
  print(f'layoutreader_model_dir is: {layoutreader_model_dir}')
65
 
66
+ # 配置文件保存到 Docker 容器的应用目录
67
  json_url = 'https://gitee.com/myhloli/MinerU/raw/master/magic-pdf.template.json'
68
  config_file_name = 'magic-pdf.json'
69
  config_file = os.path.join(base_dir, config_file_name)