Update Dockerfile

Dockerfile

@@ -1,29 +1,34 @@
-# Usa uma imagem base mais recente e suportada
+# Use a more recent and supported base image
 FROM python:3.9-slim-bookworm
 
-# Define o diretório de trabalho
+# Set the working directory
 WORKDIR /app
 
-# --- ETAPA 1: INSTALAR O GIT E CLONAR O SEU DATASET COMPLETO ---
+# --- STEP 1: INSTALL GIT AND CLONE YOUR DATASET ---
 RUN apt-get update && apt-get install -y git git-lfs && git-lfs install
 RUN git clone https://huggingface.co/spaces/Finish-him/prometheus-embedding-generator ./dados && cd dados && git lfs pull
 
-# --- ETAPA 2: CONFIGURAR O AMBIENTE PYTHON E PRÉ-AQUECER O CACHE ---
+# --- STEP 2: SET UP PYTHON ENVIRONMENT & PRE-WARM CACHE ---
 COPY requirements.txt .
 RUN pip install --no-cache-dir --upgrade -r requirements.txt
 
-# Define o caminho da cache
+# Define cache path
 ENV HF_HOME=/app/cache/huggingface
 ENV SENTENCE_TRANSFORMERS_HOME=/app/cache/torch
-RUN mkdir -p $HF_HOME && mkdir -p $SENTENCE_TRANSFORMERS_HOME && chown -R 1000:1000 /app/cache
+RUN mkdir -p $HF_HOME && mkdir -p $SENTENCE_TRANSFORMERS_HOME
 
-# --- CORREÇÃO PRINCIPAL: PRÉ-AQUECIMENTO DO CACHE ---
-# Adiciona "import os;" ao início do comando python
+# Pre-warm the cache with the e5-large model
 RUN python -c "import os; from sentence_transformers import SentenceTransformer; SentenceTransformer('intfloat/multilingual-e5-large', cache_folder=os.environ.get('SENTENCE_TRANSFORMERS_HOME', '/app/cache/torch'))"
 
-# --- ETAPA 3: EXECUTAR O SCRIPT ---
-# Copia o script de treinamento
+# --- STEP 3: PREPARE RUN DIRECTORIES & COPY SCRIPT ---
+
+# *** THE FIX IS HERE ***
+# Create the extraction directory and give the default user (1000) ownership
+RUN mkdir -p /app/dados_extraidos && chown -R 1000:1000 /app/dados_extraidos && chown -R 1000:1000 /app/cache
+
+# Copy the training script
 COPY train.py .
 
-# Executa o script e, após o término, mantém o contêiner a correr.
+# --- STEP 4: RUN THE SCRIPT ---
+# Execute the script and keep the container alive afterward
 CMD ["sh", "-c", "python train.py && sleep infinity"]