fix: update mineru setup
Browse files- requirements.txt +4 -2
- utils.py +61 -7
requirements.txt
CHANGED
@@ -5,10 +5,10 @@ rapid-table>=1.0.3,<2.0.0
|
|
5 |
rapidocr-paddle
|
6 |
rapidocr-onnxruntime
|
7 |
gradio-pdf>=0.0.21
|
8 |
-
git+https://github.com/opendatalab/MinerU.git@
|
9 |
git+https://github.com/VikParuchuri/marker
|
10 |
docling
|
11 |
-
PyMuPDF
|
12 |
pymupdf4llm
|
13 |
unstructured[pdf]
|
14 |
ultralytics>=8.3.48
|
@@ -22,3 +22,5 @@ unimernet==0.2.3
|
|
22 |
transformers<5.0.0,>=4.45.2
|
23 |
pypdf
|
24 |
opencv-contrib-python
|
|
|
|
|
|
5 |
rapidocr-paddle
|
6 |
rapidocr-onnxruntime
|
7 |
gradio-pdf>=0.0.21
|
8 |
+
git+https://github.com/opendatalab/MinerU.git@release-1.3.4
|
9 |
git+https://github.com/VikParuchuri/marker
|
10 |
docling
|
11 |
+
PyMuPDF
|
12 |
pymupdf4llm
|
13 |
unstructured[pdf]
|
14 |
ultralytics>=8.3.48
|
|
|
22 |
transformers<5.0.0,>=4.45.2
|
23 |
pypdf
|
24 |
opencv-contrib-python
|
25 |
+
numpy==1.26.4
|
26 |
+
pdfminer.six
|
utils.py
CHANGED
@@ -62,12 +62,71 @@ def fix_problematic_imports():
|
|
62 |
sys.modules[
|
63 |
"magic_pdf.model.sub_modules.mfr.unimernet.Unimernet"
|
64 |
] = fake_unimernet_module
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
|
66 |
|
67 |
def prepare_env_mineru():
|
68 |
import json
|
69 |
import os
|
70 |
-
|
71 |
import nltk
|
72 |
|
73 |
# download nltk data
|
@@ -82,12 +141,7 @@ def prepare_env_mineru():
|
|
82 |
return
|
83 |
|
84 |
# download models
|
85 |
-
|
86 |
-
"wget https://raw.githubusercontent.com/opendatalab/MinerU/"
|
87 |
-
"refs/heads/release-1.3.12/scripts/download_models_hf.py"
|
88 |
-
" -O download_models_hf.py"
|
89 |
-
)
|
90 |
-
os.system("python3 download_models_hf.py")
|
91 |
|
92 |
with open(config_path, "r") as file:
|
93 |
data = json.load(file)
|
|
|
62 |
sys.modules[
|
63 |
"magic_pdf.model.sub_modules.mfr.unimernet.Unimernet"
|
64 |
] = fake_unimernet_module
|
65 |
+
|
66 |
+
|
67 |
+
def setup_mineru_config():
|
68 |
+
import json
|
69 |
+
import os
|
70 |
+
import requests
|
71 |
+
from huggingface_hub import snapshot_download
|
72 |
+
|
73 |
+
|
74 |
+
def download_json(url):
|
75 |
+
response = requests.get(url)
|
76 |
+
response.raise_for_status()
|
77 |
+
return response.json()
|
78 |
+
|
79 |
+
|
80 |
+
def download_and_modify_json(url, local_filename, modifications):
|
81 |
+
if os.path.exists(local_filename):
|
82 |
+
data = json.load(open(local_filename))
|
83 |
+
config_version = data.get('config_version', '0.0.0')
|
84 |
+
if config_version < '1.2.0':
|
85 |
+
data = download_json(url)
|
86 |
+
else:
|
87 |
+
data = download_json(url)
|
88 |
+
|
89 |
+
for key, value in modifications.items():
|
90 |
+
data[key] = value
|
91 |
+
|
92 |
+
with open(local_filename, 'w', encoding='utf-8') as f:
|
93 |
+
json.dump(data, f, ensure_ascii=False, indent=4)
|
94 |
+
|
95 |
+
mineru_patterns = [
|
96 |
+
"models/Layout/YOLO/*",
|
97 |
+
"models/MFD/YOLO/*",
|
98 |
+
"models/MFR/unimernet_hf_small_2503/*",
|
99 |
+
"models/OCR/paddleocr_torch/*",
|
100 |
+
]
|
101 |
+
model_dir = snapshot_download('opendatalab/PDF-Extract-Kit-1.0', allow_patterns=mineru_patterns)
|
102 |
+
|
103 |
+
layoutreader_pattern = [
|
104 |
+
"*.json",
|
105 |
+
"*.safetensors",
|
106 |
+
]
|
107 |
+
layoutreader_model_dir = snapshot_download('hantian/layoutreader', allow_patterns=layoutreader_pattern)
|
108 |
+
|
109 |
+
model_dir = model_dir + '/models'
|
110 |
+
print(f'model_dir is: {model_dir}')
|
111 |
+
print(f'layoutreader_model_dir is: {layoutreader_model_dir}')
|
112 |
+
|
113 |
+
json_url = 'https://raw.githubusercontent.com/opendatalab/MinerU/refs/heads/release-1.3.12/magic-pdf.template.json'
|
114 |
+
config_file_name = 'magic-pdf.json'
|
115 |
+
home_dir = os.path.expanduser('~')
|
116 |
+
config_file = os.path.join(home_dir, config_file_name)
|
117 |
+
|
118 |
+
json_mods = {
|
119 |
+
'models-dir': model_dir,
|
120 |
+
'layoutreader-model-dir': layoutreader_model_dir,
|
121 |
+
}
|
122 |
+
|
123 |
+
download_and_modify_json(json_url, config_file, json_mods)
|
124 |
+
print(f'The configuration file has been configured successfully, the path is: {config_file}')
|
125 |
|
126 |
|
127 |
def prepare_env_mineru():
|
128 |
import json
|
129 |
import os
|
|
|
130 |
import nltk
|
131 |
|
132 |
# download nltk data
|
|
|
141 |
return
|
142 |
|
143 |
# download models
|
144 |
+
setup_mineru_config()
|
|
|
|
|
|
|
|
|
|
|
145 |
|
146 |
with open(config_path, "r") as file:
|
147 |
data = json.load(file)
|