Upload 4 files

.envor.txt

@@ -0,0 +1 @@
+ROUTER_API_KEY=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

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+LegisMiner.pdf filter=lfs diff=lfs merge=lfs -text

LegisMiner.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6702876fa520e29805546201280dfba74e4b6cf7e86c8d6dc03c3f485e7293ec
+size 51675358

app.py.py

@@ -0,0 +1,196 @@
+# -*- coding: utf-8 -*-
+"""app
+
+Automatically generated by Colab.
+
+Original file is located at
+    https://colab.research.google.com/drive/1PhcQoTZvxdPQe6E1HMx_Nl4Zs_tY7J_y
+"""
+
+
+
+import gradio as gr
+import os
+from dotenv import load_dotenv
+from langchain_community.chat_models import ChatOpenAI
+from langchain_community.document_loaders import PyPDFLoader
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_community.vectorstores import FAISS
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.chains import RetrievalQA
+import tempfile
+import numpy as np
+from sklearn.metrics.pairwise import cosine_similarity
+
+# ✅ Carrega variáveis de ambiente
+load_dotenv()
+OPENROUTER_API_KEY = os.getenv("ROUTER_API_KEY")
+
+if not OPENROUTER_API_KEY:
+    raise ValueError("❌ A variável de ambiente ROUTER_API_KEY não está definida. Verifique o arquivo .env.")
+
+# Inicialização
+embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+qa_chain = None
+processed_file = None
+
+# 🔁 Carrega automaticamente o legisMiner.pdf ao iniciar
+def load_default_pdf():
+    global qa_chain, processed_file
+    try:
+        loader = PyPDFLoader("legisMiner.pdf")
+        documents = loader.load()
+
+        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
+        docs = text_splitter.split_documents(documents)
+
+        vectorstore = FAISS.from_documents(docs, embeddings)
+
+        llm = ChatOpenAI(
+            openai_api_key=OPENROUTER_API_KEY,
+            openai_api_base="https://openrouter.ai/api/v1",
+            model="deepseek/deepseek-r1-0528:free",
+            temperature=0.7
+        )
+
+        qa_chain = RetrievalQA.from_chain_type(
+            llm=llm,
+            retriever=vectorstore.as_retriever(),
+            return_source_documents=True
+        )
+
+        processed_file = "LegisMiner.pdf"
+        print("✅ LegisMiner.pdf carregado automaticamente.")
+    except Exception as e:
+        print(f"❌ Erro ao carregar LegisMiner.pdf automaticamente: {e}")
+
+def calculate_rag_metrics(query, response, source_docs):
+    metrics = {}
+    try:
+        query_embedding = embeddings.embed_query(query)
+        response_embedding = embeddings.embed_query(response)
+        metrics["query_response_similarity"] = cosine_similarity(
+            [query_embedding], [response_embedding]
+        )[0][0]
+
+        doc_similarities = []
+        for doc in source_docs:
+            doc_embedding = embeddings.embed_query(doc.page_content[:1000])
+            similarity = cosine_similarity([response_embedding], [doc_embedding])[0][0]
+            doc_similarities.append(similarity)
+
+        metrics["avg_response_source_similarity"] = np.mean(doc_similarities) if doc_similarities else 0
+        metrics["max_response_source_similarity"] = max(doc_similarities) if doc_similarities else 0
+        metrics["num_source_documents"] = len(source_docs)
+
+    except Exception as e:
+        metrics["error"] = str(e)
+
+    return metrics
+
+def process_pdf(file):
+    global qa_chain, processed_file
+
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
+        tmp.write(file)
+        pdf_path = tmp.name
+
+    try:
+        loader = PyPDFLoader(pdf_path)
+        documents = loader.load()
+
+        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
+        docs = text_splitter.split_documents(documents)
+
+        vectorstore = FAISS.from_documents(docs, embeddings)
+
+        llm = ChatOpenAI(
+            openai_api_key=OPENROUTER_API_KEY,
+            openai_api_base="https://openrouter.ai/api/v1",
+            model="deepseek/deepseek-r1-0528-qwen3-8b:free",
+            temperature=0.7
+        )
+
+        qa_chain = RetrievalQA.from_chain_type(
+            llm=llm,
+            retriever=vectorstore.as_retriever(),
+            return_source_documents=True
+        )
+
+        processed_file = os.path.basename(pdf_path)
+        return f"✅ PDF processado com sucesso: {processed_file}"
+
+    except Exception as e:
+        return f"❌ Erro ao processar PDF: {str(e)}"
+
+def ask_question(question):
+    global qa_chain
+
+    if qa_chain is None:
+        return "⚠️ Por favor, carregue um PDF primeiro", "", {}
+
+    try:
+        resposta = qa_chain.invoke({"query": question})
+
+        sources = "\n\n".join(
+            [f"📄 Fonte {i+1}:\n{doc.page_content[:500]}..."
+             for i, doc in enumerate(resposta['source_documents'])]
+        )
+
+        metrics = calculate_rag_metrics(
+            question,
+            resposta['result'],
+            resposta['source_documents']
+        )
+
+        metrics_text = "\n".join(
+            [f"{k.replace('_', ' ').title()}: {v:.2f}" if isinstance(v, float) else f"{k.replace('_', ' ').title()}: {v}"
+             for k, v in metrics.items() if k != "error"]
+        )
+
+        return resposta['result'], sources, metrics_text
+
+    except Exception as e:
+        return f"❌ Erro ao processar pergunta: {str(e)}", "", {}
+
+# Interface Gradio
+with gr.Blocks(title="Chat com PDF usando OpenRouter", theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 🧠 Artificial Intelligence Applied to Regulatory Standard Processing in Mining\n### 💡 Development of a Decision Support Tool")
+
+    with gr.Row():
+        with gr.Column(scale=1):
+            file_input = gr.File(label="📤 Envie um PDF", type="binary")
+            process_btn = gr.Button("Processar PDF", variant="primary")
+            status_output = gr.Textbox(label="Status")
+
+        with gr.Column(scale=2):
+            question_input = gr.Textbox(label="Faça uma pergunta sobre Normas da Mineração", lines=3)
+            ask_btn = gr.Button("Enviar Pergunta", variant="primary")
+            answer_output = gr.Textbox(label="✅ Resposta", interactive=False)
+
+        with gr.Accordion("📄 Fontes usadas", open=False):
+                sources_output = gr.Textbox(label="Trechos relevantes", lines=10)
+
+        with gr.Accordion("📊 Métricas RAG", open=False):
+                metrics_output = gr.Textbox(label="Métricas", lines=4)
+
+
+
+    process_btn.click(
+        fn=process_pdf,
+        inputs=file_input,
+        outputs=status_output
+    )
+
+    ask_btn.click(
+        fn=ask_question,
+        inputs=question_input,
+        outputs=[answer_output, sources_output, metrics_output]
+    )
+
+# 🔁 Carrega o PDF fixo ao iniciar
+load_default_pdf()
+
+# Compartilhamento opcional no Colab ou Hugging Face
+share = True if 'COLAB_JUPYTER_TRANSPORT' in os.environ else False
+demo.launch(share=share, debug=False)

requirements.txt

@@ -0,0 +1,7 @@
+gradio
+python-dotenv
+langchain
+faiss-cpu
+sentence-transformers
+scikit-learn
+openai