Spaces:
Sleeping
Sleeping
Fix NLTK download syntax error in Dockerfile by using separate Python script
Browse files- Dockerfile +1 -15
- download_nltk_data.py +20 -0
Dockerfile
CHANGED
|
@@ -23,21 +23,7 @@ RUN pip install --no-cache-dir --upgrade pip setuptools wheel && \
|
|
| 23 |
COPY . .
|
| 24 |
|
| 25 |
# Download NLTK data during build with error handling
|
| 26 |
-
RUN python
|
| 27 |
-
nltk_data_dir = '/root/nltk_data'; \
|
| 28 |
-
os.makedirs(nltk_data_dir, exist_ok=True); \
|
| 29 |
-
nltk.data.path.append(nltk_data_dir); \
|
| 30 |
-
try: \
|
| 31 |
-
nltk.download('punkt', quiet=True); \
|
| 32 |
-
print('Downloaded punkt'); \
|
| 33 |
-
except Exception as e: \
|
| 34 |
-
print(f'Failed to download punkt: {e}'); \
|
| 35 |
-
try: \
|
| 36 |
-
nltk.download('vader_lexicon', quiet=True); \
|
| 37 |
-
print('Downloaded vader_lexicon'); \
|
| 38 |
-
except Exception as e: \
|
| 39 |
-
print(f'Failed to download vader_lexicon: {e}'); \
|
| 40 |
-
print('NLTK download step completed')"
|
| 41 |
|
| 42 |
# Expose the port the app runs on
|
| 43 |
EXPOSE 7860
|
|
|
|
| 23 |
COPY . .
|
| 24 |
|
| 25 |
# Download NLTK data during build with error handling
|
| 26 |
+
RUN python download_nltk_data.py
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
|
| 28 |
# Expose the port the app runs on
|
| 29 |
EXPOSE 7860
|
download_nltk_data.py
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""Script to download NLTK data during Docker build"""
|
| 2 |
+
import nltk
|
| 3 |
+
import os
|
| 4 |
+
|
| 5 |
+
# Set NLTK data directory
|
| 6 |
+
nltk_data_dir = '/root/nltk_data'
|
| 7 |
+
os.makedirs(nltk_data_dir, exist_ok=True)
|
| 8 |
+
nltk.data.path.append(nltk_data_dir)
|
| 9 |
+
|
| 10 |
+
# Download required NLTK data
|
| 11 |
+
required_data = ['punkt', 'vader_lexicon']
|
| 12 |
+
for data_name in required_data:
|
| 13 |
+
try:
|
| 14 |
+
nltk.download(data_name, quiet=True)
|
| 15 |
+
print(f'Downloaded {data_name}')
|
| 16 |
+
except Exception as e:
|
| 17 |
+
print(f'Failed to download {data_name}: {e}')
|
| 18 |
+
|
| 19 |
+
print('NLTK download step completed')
|
| 20 |
+
|