ftshijt commited on
Commit
73d9817
·
1 Parent(s): a4807ff

remove fix for unusual issue

Browse files
Files changed (2) hide show
  1. Dockerfile +0 -4
  2. fix_nltk_permissions.py +0 -81
Dockerfile CHANGED
@@ -38,10 +38,6 @@ RUN git clone https://github.com/wavlab-speech/versa.git && \
38
  RUN mkdir -p /app/data/configs /app/data/uploads /app/data/results && \
39
  chmod -R 777 /app/data
40
 
41
- # Copy and run the NLTK fix script
42
- COPY fix_nltk_permissions.py .
43
- RUN python fix_nltk_permissions.py
44
-
45
  # Copy universal metrics YAML file
46
  COPY universal_metrics.yaml /app/data/configs/
47
 
 
38
  RUN mkdir -p /app/data/configs /app/data/uploads /app/data/results && \
39
  chmod -R 777 /app/data
40
 
 
 
 
 
41
  # Copy universal metrics YAML file
42
  COPY universal_metrics.yaml /app/data/configs/
43
 
fix_nltk_permissions.py DELETED
@@ -1,81 +0,0 @@
1
- #!/usr/bin/env python3
2
- """
3
- Script to pre-download NLTK data to a writable location and patch
4
- discrete_speech_metrics to use this location.
5
- """
6
-
7
- import os
8
- import sys
9
- import nltk
10
- import importlib.util
11
- import subprocess
12
-
13
- # Set NLTK data directory to a writable location
14
- nltk_data_dir = "/usr/local/share/nltk_data"
15
- os.makedirs(nltk_data_dir, exist_ok=True)
16
- os.environ["NLTK_DATA"] = nltk_data_dir
17
-
18
- print(f"Setting NLTK data directory to: {nltk_data_dir}")
19
- print(f"Directory exists: {os.path.exists(nltk_data_dir)}")
20
- print(f"Directory permissions: {oct(os.stat(nltk_data_dir).st_mode)}")
21
- print(f"User ID: {os.getuid()}, Group ID: {os.getgid()}")
22
-
23
- # Make the directory world-writable
24
- try:
25
- os.chmod(nltk_data_dir, 0o777)
26
- print(f"Changed permissions on {nltk_data_dir} to 0o777")
27
- except Exception as e:
28
- print(f"Failed to change permissions: {e}")
29
-
30
- # Pre-download necessary NLTK data
31
- for package in ['punkt', 'stopwords', 'wordnet']:
32
- print(f"Downloading NLTK package: {package}")
33
- try:
34
- nltk.download(package, download_dir=nltk_data_dir, quiet=False)
35
- print(f"Successfully downloaded {package}")
36
- except Exception as e:
37
- print(f"Error downloading {package}: {e}")
38
-
39
- # Try to modify the discrete_speech_metrics module to use our NLTK data dir
40
- try:
41
- # Find discrete_speech_metrics module
42
- spec = importlib.util.find_spec("discrete_speech_metrics")
43
- if spec is not None:
44
- module_path = os.path.dirname(spec.origin)
45
- print(f"Found discrete_speech_metrics at: {module_path}")
46
-
47
- # Files to patch
48
- files_to_patch = [
49
- os.path.join(module_path, "speechbleu.py"),
50
- os.path.join(module_path, "speechbert.py"),
51
- os.path.join(module_path, "__init__.py")
52
- ]
53
-
54
- for file_path in files_to_patch:
55
- if os.path.exists(file_path):
56
- print(f"Patching file: {file_path}")
57
-
58
- # Read file content
59
- with open(file_path, 'r') as file:
60
- content = file.read()
61
-
62
- # Replace nltk.download with our custom version
63
- if "nltk.download" in content:
64
- patched_content = content.replace(
65
- 'nltk.download(',
66
- f'nltk.download(download_dir="{nltk_data_dir}", '
67
- )
68
-
69
- # Write patched content back
70
- with open(file_path, 'w') as file:
71
- file.write(patched_content)
72
-
73
- print(f"Successfully patched {file_path}")
74
- else:
75
- print(f"No nltk.download calls found in {file_path}")
76
- else:
77
- print("discrete_speech_metrics module not found")
78
- except Exception as e:
79
- print(f"Error patching discrete_speech_metrics: {e}")
80
-
81
- print("NLTK setup complete")