nltk_download

Dockerfile

@@ -6,11 +6,19 @@ WORKDIR /app
 COPY . .
 
 # This is only for building docker in huggingface spaces
-ENV HF_HOME=/data/.huggingface
+ENV HF_HOME=/data/.huggingface           \
+    NLTK_DATA=/opt/nltk_data
 
 RUN conda env create -f environment_sate_1.5.yml
 
-RUN conda run -n SATEv1.5 python -c "import nltk; nltk.download('punkt'); nltk.download('punkt_tab')"
+RUN mkdir -p $NLTK_DATA && \
+    conda run -n SATEv1.5 python - <<'PY'
+import nltk, os, json, textwrap
+target = os.environ["NLTK_DATA"]
+for pkg in ("punkt", "punkt_tab"):
+    nltk.download(pkg, download_dir=target, quiet=True)
+print(f"Downloaded {pkg} to {target}")
+PY
 
 RUN mkdir -p /app/session_data && chown -R user:user /app/session_data