Initial Commit

.github/prompts/pdfExtractor.prompt.md

@@ -0,0 +1,6 @@
+# Extrator de pdf
+
+O extrator é um agente de IA construído com o langchain em python, seguindo as seguintes instruções:
+  Dentro de um PDF deve extrair as seguintes informações:
+- Os índices dos capítulos retornando as linhas de início e do fim
+- Separar cada capítulo, criando um PDF diferente para cada um

.github/prompts/projeto.prompt.md

@@ -0,0 +1,8 @@
+# Escopo do projeto:
+
+Vai consumir um modelo llm para realizar consultas em um banco de dados, possuindo as seguintes características:
+- Agnóstico de banco de dados, podendo ser usado para qualquer tipo
+- A resposta retorna a lógica do que foi feito junto com o resultado do seu sql exceto:
+    - Caso altere qualquer tipo de dado no banco, retorne apenas o sql para o próprio usuário executar
+    - Caso sela apenas consulta realize a consulta, retorne qual foi o sql e o resultado da consulta
+- Usará llm local, com o llama cpp e suporte a vulkan

.vscode/settings.json

@@ -0,0 +1,11 @@
+{
+    "python.testing.unittestArgs": [
+        "-v",
+        "-s",
+        "./test",
+        "-p",
+        "test_*.py"
+    ],
+    "python.testing.pytestEnabled": false,
+    "python.testing.unittestEnabled": true
+}

agentsConfig.ini

@@ -0,0 +1,3 @@
+[PDFAgent]
+model = qwen2.5:1.5b
+modelProvider = ollama

src/__init__.py

@@ -0,0 +1 @@
+# Este arquivo transforma o diretório em um pacote Python

src/interfaces/agents/agent.py

@@ -0,0 +1,23 @@
+from abc import ABC, abstractmethod
+from typing import List, Dict, Any
+
+class Agent(ABC):
+    @abstractmethod
+    def get_agent_tools(self) -> List[Dict[str, Any]]:
+        """
+        Get the tools available for the agent.
+        
+        Returns:
+            List[Dict[str, Any]]: A list of dictionaries representing the tools.
+        """
+        pass
+    
+    @abstractmethod
+    def get_agent_details(self) -> Dict[str, Any]:
+        """
+        Get the details of the agent.
+        
+        Returns:
+            Dict[str, Any]: A dictionary containing agent details.
+        """
+        pass

src/interfaces/models/modelInference.py

@@ -0,0 +1,16 @@
+from abc import ABC, abstractmethod
+from ast import Dict
+from typing import List, Dict, Any, Optional
+
+
+class ModelManager(ABC):
+    
+    @abstractmethod
+    def get_available_models(self) -> List[Dict[str, Any]]:
+        pass
+    
+    @abstractmethod
+    def get_model_details(self, model_name: str) -> Optional[Dict[str, Any]]:
+        pass
+    
+    

src/main.py

@@ -0,0 +1,9 @@
+from service.ollama.ollamaModelManager import OllamaModelManager
+
+# Utilizando nossa classe para obter os modelos
+ollama_manager = OllamaModelManager()
+models = ollama_manager.get_available_models()
+# print(models)
+
+details = ollama_manager.get_model_details("qwen2.5:1.5b")
+print(details)

src/modules/pdf/pdfAgent.py

@@ -0,0 +1,33 @@
+from langchain.agents import AgentExecutor
+from langchain_community.tools import BaseTool
+
+import fitz
+import os
+
+from src.interfaces.agents.agent import Agent
+
+
+class PDFAgent(Agent): 
+    
+    def __init__(self):
+        """
+        Initialize a PDF agent.
+        This initializes the PDF agent which handles operations related to PDF documents.
+        The agent attribute needs to be set with an appropriate agent implementation
+        during instantiation.
+        Attributes:
+            agent: The agent implementation for PDF operations.
+        """
+        
+        
+        self.agent 
+        
+    def get_agent_details(self):
+        return super().get_agent_details()
+    
+    def get_agent_tools(self):
+        return super().get_agent_tools()
+
+
+
+        

src/modules/pdf/tools/metadata.pdf.tool.py

@@ -0,0 +1,15 @@
+import fitz
+from langchain_community.tools import BaseTool
+
+class PDFMetadataTool(BaseTool):
+    name="pdf_metadata"
+    description="Extract metadata from PDF file"
+    
+    def _run(self, path: str):
+        try: 
+            with fitz.open(path) as doc:
+                metadata = {
+                    "páginas": len(doc)
+                }
+        except Exception as e:
+            return f"Error: {e}"

src/modules/pdf/tools/textExtract.pdf.tool.py

@@ -0,0 +1,10 @@
+from langchain_community.tools import BaseTool
+
+
+class PDFextractTextTool(BaseTool):
+    name="pdf_text_extract"
+    description="Extract text from PDF file"
+    
+    def _run(self, path: str):
+        try: 
+            

src/modules/pdf/utils.py

@@ -0,0 +1,18 @@
+import fitz
+import os
+
+
+class PDFUtils:
+    @staticmethod
+    def get_total_lines(pdf_path: str):
+        pdf = fitz.open(pdf_path)
+        return len(pdf)
+        
+        
+# Get the user's home directory and construct the path to the PDF
+home_dir = os.path.expanduser("~")
+pdf_path = os.path.join(home_dir, "Downloads", "Manuais.pdf")
+
+# Call the method with the full path
+total_pages = PDFUtils.get_total_lines(pdf_path)
+print(f"Total pages in PDF: {total_pages}")

src/service/ollama/ollamaModelManager.py

@@ -0,0 +1,46 @@
+import requests
+
+from src.interfaces.models.modelInference import ModelManager
+
+
+class OllamaModelManager(ModelManager):
+    def __init__(self, base_url = "http://localhost:11434"):
+        self.base_url = base_url
+        
+    def get_available_models(self):
+        try:
+            print(requests.__file__)
+            print(self.base_url)
+            response = requests.get(f"{self.base_url}/api/tags")
+            data = response.json()
+            return data.get('models', [])
+        except Exception as e:
+            print(f"Erro ao obter modelos: {e}")
+            return []
+
+    def get_model_details(self, model_name):
+        try:
+            
+           # Obter detalhes completos do modelo via API
+            response = requests.post(
+                f"{self.base_url}/api/show",
+                json={"name": model_name}
+            )
+            model_details = response.json()
+            
+            # O tamanho do contexto geralmente está disponível em model_details['parameters']['context_length']
+            # ou em outro campo similar dependendo do modelo
+            context_size = model_details.get('parameters', {}).get('context_length', 'Não disponível')
+            
+            # Podemos adicionar outros detalhes relevantes
+            details = {
+                "name": model_name,
+                "context_size": context_size,
+                "model_type": model_details.get('modelfile', {}).get('parameter', 'Não disponível'),
+                "license": model_details.get('license', 'Não disponível')
+            }
+            
+            return details
+        except Exception as e:
+            print(f"Error getting model details: {e}")
+            return {"name": model_name, "error": str(e)}

src/utils/__init__.py

@@ -0,0 +1 @@
+# Este arquivo transforma o diretório em um pacote Python

src/utils/getModels.py

@@ -0,0 +1,9 @@
+
+
+from service.ollama.ollamaModelManager import OllamaModelManager
+
+
+ollama_service = OllamaModelManager()
+models = ollama_service.get_available_models()
+
+print(models)

test/test_ollamaModelManager.py

@@ -0,0 +1,81 @@
+import unittest
+from unittest.mock import patch, MagicMock
+
+from src.service.ollama.ollamaModelManager import OllamaModelManager
+
+class TestOllamaModelManager(unittest.TestCase):
+    
+    def setUp(self):
+        self.model_manager = OllamaModelManager()
+        self.model_manager.base_url = "http://test-url:11434"
+    
+    @patch('service.ollama.ollamaModelManager.requests.get')
+    def test_get_available_models_success(self, mock_get):
+        # Setup mock response
+        mock_response = MagicMock()
+        mock_response.json.return_value = {
+            'models': [
+                {'name': 'model1'},
+                {'name': 'model2'}
+            ]
+        }
+        mock_get.return_value = mock_response
+        
+        # Call method
+        result = self.model_manager.get_available_models()
+        
+        # Assertions
+        mock_get.assert_called_once_with("http://test-url:11434/api/tags")
+        self.assertEqual(len(result), 2)
+        self.assertEqual(result, [{'name': 'model1'}, {'name': 'model2'}])
+    
+    @patch('service.ollama.ollamaModelManager.requests.get')
+    def test_get_available_models_exception(self, mock_get):
+        # Setup mock to raise exception
+        mock_get.side_effect = Exception("Connection error")
+        
+        # Call method
+        result = self.model_manager.get_available_models()
+        
+        # Assertions
+        self.assertEqual(result, [])
+    
+    @patch('service.ollama.ollamaModelManager.requests.post')
+    def test_get_model_details_success(self, mock_post):
+        # Setup mock response
+        mock_response = MagicMock()
+        mock_response.json.return_value = {
+            'parameters': {'context_length': 4096},
+            'modelfile': {'parameter': 'llama2'},
+            'license': 'Apache 2.0'
+        }
+        mock_post.return_value = mock_response
+        
+        # Call method
+        result = self.model_manager.get_model_details('llama2')
+        
+        # Assertions
+        mock_post.assert_called_once_with(
+            "http://test-url:11434/api/show",
+            json={"name": "llama2"}
+        )
+        self.assertEqual(result['name'], 'llama2')
+        self.assertEqual(result['context_size'], 4096)
+        self.assertEqual(result['license'], 'Apache 2.0')
+    
+    @patch('service.ollama.ollamaModelManager.requests.post')
+    def test_get_model_details_exception(self, mock_post):
+        # Setup mock to raise exception
+        mock_post.side_effect = Exception("API error")
+        
+        # Call method
+        result = self.model_manager.get_model_details('unknown_model')
+        
+        # Assertions
+        self.assertEqual(result['name'], 'unknown_model')
+        self.assertTrue('error' in result)
+        self.assertEqual(result['error'], 'API error')
+
+
+if __name__ == '__main__':
+    unittest.main()