s

Dockerfile

@@ -6,6 +6,13 @@ COPY ./requirements.txt /code/requirements.txt
 
 RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
 
+# Set an environment variable for the NLTK data directory
+ENV NLTK_DATA /code/nltk_data
+
+# Create the directory for NLTK data and download the 'punkt' tokenizer models
+RUN mkdir -p $NLTK_DATA && \
+    python -m nltk.downloader -d $NLTK_DATA punkt
+
 COPY . .
 
 CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

main.py

@@ -6,7 +6,10 @@ import uvicorn
 from typing import Generator
 import json  # Asegúrate de que esta línea esté al principio del archivo
 import nltk
-nltk.download('punkt')
+import os
+
+
+nltk.data.path.append(os.getenv('NLTK_DATA'))
 
 app = FastAPI()