rpa-suite 1.6.0__py3-none-any.whl → 1.6.1__py3-none-any.whl

rpa_suite/__init__.py

@@ -66,7 +66,7 @@ Módulos disponíveis:
     ``Iris``: Objeto Iris Automação de funções para converter documentos com OCR + IA baseado em ``docling``
 """
 
-__version__ = "1.5.9"
+__version__ = "1.6.1"
 
 # allows importing the rpa_suite module without the package name
 from .suite import rpa

rpa_suite/core/__init__.py

@@ -36,13 +36,16 @@ from .parallel import ParallelRunner
 from .asyncrun import AsyncRunner
 
 
-# On this case, we are importing the Browser class only if the selenium and webdriver_manager modules are installed.
-# This is useful to avoid unnecessary imports and dependencies if the user does not need the Browser functionality.
+# On this case, we are importing the (Browser|Iris) class only if the (selenium and webdriver_manager| docling) modules are installed.
+# This is useful to avoid unnecessary imports and dependencies if the user does not need the (Browser|Iris) functionality.
 import importlib.util
 
 # from .browser import Browser
 if importlib.util.find_spec("selenium") and importlib.util.find_spec("webdriver_manager"):
     from .browser import Browser
 
+# from .iris import Iris
+if importlib.util.find_spec("docling"):
+    from .iris import Iris
 
-__version__ = "1.5.5"
+__version__ = "1.6.1"

rpa_suite/core/iris.py

@@ -1,327 +1,140 @@
 # rpa_suite/core/iris.py
+
 """
-Iris (OCR-IA) module for document conversion using DocLing.
+Iris (OCR-IA) módulo para conversão de documentos usando DocLing.
 
-This module provides a simplified interface for converting documents
-into various formats, optimized for RPA automation use.
+Este módulo fornece uma interface simplificada para converter documentos
+em vários formatos, otimizado para uso em automação RPA.
 """
 
+# imports externos
+try:
+    from docling.document_converter import DocumentConverter
+except ImportError as e:
+    raise ImportError("Iris - Error: Não foi possível importar 'docling.document_converter'. Certifique-se de que a biblioteca 'docling' está instalada.") from e
+
+# imports de terceiros
 from enum import Enum
 from pathlib import Path
 from typing import Any, Dict, List, Optional, Union
 
-# imports internal
+# imports internos
 from rpa_suite.functions._printer import alert_print, error_print, success_print
 
+class IrisError(Exception):
+    """Exceção personalizada para erros do Iris."""
+    def __init__(self, message):
+        super().__init__(f'Iris - Error: {message}')
 
-class ExportFormat(Enum):
-    """Supported export formats for document conversion."""
-    
+class ExportFormats(Enum):
+    """Formatos de exportação suportados para conversão de documentos."""
     MARKDOWN = "markdown"
     DICT = "dict"
     DOCTAGS = "doctags"
     HTML = "html"
     TEXT = "text"
 
-
-class IrisError(Exception):
-    """Custom exception for Iris class errors."""
-    
-    pass
-
-
 class Iris:
     """
     Iris (OCR-IA)
-    Document converter using the DocLing library.
-    
-    This class provides a simplified interface for converting documents
-    in various formats (PDF, images, text) to structured formats such as
-    Markdown, HTML, plain text, among others.
-    
-    Attributes:
-        engine: Instance of DocLing's DocumentConverter.
-        last_result: Last processed conversion result.
-        
-    Example:
+    Conversor de documentos usando a biblioteca DocLing.
+
+    Esta classe fornece uma interface simplificada para converter documentos
+    em vários formatos (PDF, imagens, texto) para formatos estruturados como
+    Markdown, HTML, texto simples, entre outros.
+
+    Atributos:
+        engine: Instância do DocumentConverter do DocLing.
+        last_result: Último resultado de conversão processado.
+
+    Exemplo:
         >>> iris = Iris()
-        >>> content = iris.read_document("document.pdf", ExportFormat.MARKDOWN)
+        >>> content = iris.read_document("document.pdf", ExportFormats.MARKDOWN)
         >>> print(content)
     """
-    
-    # Supported file extensions
-    SUPPORTED_EXTENSIONS = {
-        '.pdf', '.txt', '.docx', '.doc', '.png', '.jpg', '.jpeg', 
-        '.tiff', '.bmp', '.webp', '.pptx', '.xlsx'
-    }
-    
-    def __init__(self, display_message: bool = False) -> None:
-        """
-        Initializes the Iris class with the document converter.
-        
-        Raises:
-            IrisError: If the DocLing library is not installed.
-        """
-        self._engine: Optional[Any] = None
-        self._last_result: Optional[Any] = None
-        self.display_message: bool = display_message
-        self._initialize_engine()
-    
-    def _initialize_engine(self) -> None:
-        """
-        Initializes the DocumentConverter engine.
-        
-        Raises:
-            IrisError: If the DocLing library is not available.
-        """
-        try:
-            from docling.document_converter import DocumentConverter
-            self._engine = DocumentConverter()
-            if self.display_message: success_print("Iris engine initialized successfully")
-        except ImportError as e:
-            error_msg = (
-                "The 'docling' library is not installed. "
-                "Run: python -m pip install docling"
-            )
-            error_print(f"Iris - {error_msg}")
-            error_print(f"Error importing DocLing: {e}")
-            raise IrisError(error_msg) from e
-    
-    @property
-    def engine(self) -> Any:
-        """Returns the DocumentConverter engine instance."""
-        return self._engine
-    
-    @property
-    def last_result(self) -> Optional[Any]:
-        """Returns the last processed conversion result."""
-        return self._last_result
-    
-    def _validate_file_path(self, file_path: Union[str, Path]) -> Path:
-        """
-        Validates the file path and returns a Path object.
-        
-        Args:
-            file_path: Path to the file.
-            
-        Returns:
-            Path: Validated Path object.
-            
-        Raises:
-            IrisError: If the file does not exist or is not supported.
-        """
-        path_obj = Path(file_path)
-        
-        if not path_obj.exists():
-            raise IrisError(f"File not found: {file_path}")
-        
-        if not path_obj.is_file():
-            raise IrisError(f"Path does not point to a file: {file_path}")
-        
-        if path_obj.suffix.lower() not in self.SUPPORTED_EXTENSIONS:
-            supported = ", ".join(sorted(self.SUPPORTED_EXTENSIONS))
-            raise IrisError(
-                f"Extension '{path_obj.suffix}' is not supported. "
-                f"Supported extensions: {supported}"
-            )
-        
-        return path_obj
-    
-    def _convert_document(self, file_path: Path) -> Any:
+
+    engine: Optional[DocumentConverter]
+
+    def __init__(self) -> None:
         """
-        Converts the document using DocumentConverter.
-        
-        Args:
-            file_path: Path to the file.
-            
-        Returns:
-            Result of the DocLing conversion.
-            
-        Raises:
-            IrisError: If the conversion fails.
+        Inicializa a classe Iris com o conversor de documentos.
+
+        Levanta:
+            IrisError: Se a biblioteca DocLing não estiver instalada.
         """
         try:
-            if self.display_message: success_print(f"Starting conversion of file: {file_path}")
-            result = self._engine.convert(str(file_path))
-            self._last_result = result
-            if self.display_message: success_print("Conversion completed successfully")
-            return result
+            self.iris_converter = DocumentConverter()
+            self.engine = self.iris_converter
+            self.path_file = None
+            self.result_converted = None
+            self.result_formatted = None
         except Exception as e:
-            error_msg = f"Error converting document '{file_path}': {e}"
-            error_print(f"Iris - {error_msg}")
-            error_print(error_msg)
-            raise IrisError(error_msg) from e
-    
-    def _export_to_format(self, document: Any, export_format: ExportFormat) -> Any:
+            error_print("Iris - Error: Falha ao inicializar o DocumentConverter.")
+            raise IrisError(f"Falha ao inicializar o DocumentConverter: {e}")
+
+    def __convert_document(self, path_file: str = None):
         """
-        Exports the document to the specified format.
-        
-        Args:
-            document: Document converted by DocLing.
-            export_format: Desired export format.
-            
-        Returns:
-            Document in the specified format.
-            
-        Raises:
-            IrisError: If the export fails.
+        Converte o documento informado pelo caminho.
+
+        Levanta:
+            IrisError: Se ocorrer erro na conversão do documento.
         """
-        export_methods = {
-            ExportFormat.MARKDOWN: document.export_to_markdown,
-            ExportFormat.DICT: document.export_to_dict,
-            ExportFormat.DOCTAGS: document.export_to_doctags,
-            ExportFormat.HTML: document.export_to_html,
-            ExportFormat.TEXT: document.export_to_text,
-        }
-        
         try:
-            export_method = export_methods[export_format]
-            return export_method()
-        except KeyError:
-            available_formats = ", ".join([fmt.value for fmt in ExportFormat])
-            raise IrisError(
-                f"Format '{export_format.value}' is not supported. "
-                f"Available formats: {available_formats}"
-            )
+            if not path_file:
+                raise IrisError("Caminho do arquivo não informado para conversão.")
+            self.result_converted = self.iris_converter.convert(path_file)
         except Exception as e:
-            error_msg = f"Error exporting to format '{export_format.value}': {e}"
-            error_print(error_msg)
-            raise IrisError(error_msg) from e
-    
-    def read_document(
-        self,
-        file_path: Union[str, Path],
-        export_format: ExportFormat = ExportFormat.MARKDOWN,
-        verbose: bool = False,
-    ) -> Optional[Any]:
+            error_print(f"Iris - Error: Falha ao converter o documento: {e}")
+            raise IrisError(f"Falha ao converter o documento: {e}")
+
+    def read_document(self, file_path: str = None, result_format=ExportFormats.MARKDOWN, verbose: bool = False) -> Optional[Union[str, dict]]:
         """
-        Reads and converts a document to the specified format.
-        
+        Lê e converte um documento para o formato especificado.
+
         Args:
-            file_path: Path to the document file.
-            export_format: Desired export format.
-            verbose: If True, displays success messages.
-            
-        Returns:
-            Document converted to the specified format, or None if it fails.
-            
-        Raises:
-            IrisError: If an error occurs during validation, conversion, or export.
-            
-        Example:
+            file_path: Caminho para o arquivo do documento.
+            result_format: Formato de exportação desejado.
+            verbose: Se True, exibe mensagens de sucesso.
+
+        Retorna:
+            Documento convertido para o formato especificado, ou None se falhar.
+
+        Levanta:
+            IrisError: Se ocorrer erro durante validação, conversão ou exportação.
+
+        Exemplo:
             >>> iris = Iris()
-            >>> content = iris.read_document("doc.pdf", ExportFormat.TEXT)
+            >>> content = iris.read_document("doc.pdf", ExportFormats.TEXT)
             >>> print(content)
         """
         try:
-            # File validation
-            validated_path = self._validate_file_path(file_path)
-            
-            # Document conversion
-            conversion_result = self._convert_document(validated_path)
-            
-            # Conversion result check
-            if not conversion_result or not hasattr(conversion_result, 'document'):
-                raise IrisError("Invalid conversion result or document not found")
-            
-            # Export to desired format
-            formatted_result = self._export_to_format(
-                conversion_result.document, 
-                export_format
-            )
-            
+            self.__convert_document(file_path)
+
+            if not self.result_converted or not hasattr(self.result_converted, 'document'):
+                raise IrisError("Conversão falhou ou objeto retornado inválido.")
+
+            if result_format == ExportFormats.MARKDOWN:
+                self.result_formatted = self.result_converted.document.export_to_markdown()
+            elif result_format == ExportFormats.DICT:
+                self.result_formatted = self.result_converted.document.export_to_dict()
+            elif result_format == ExportFormats.DOCTAGS:
+                self.result_formatted = self.result_converted.document.export_to_doctags()
+            elif result_format == ExportFormats.HTML:
+                self.result_formatted = self.result_converted.document.export_to_html()
+            elif result_format == ExportFormats.TEXT:
+                self.result_formatted = self.result_converted.document.export_to_text()
+            else:
+                alert_print(f'Iris - Error: Formato não suportado: {result_format}.')
+                raise IrisError(f"Formato não suportado: {result_format}.")
+
             if verbose:
-                success_print("Iris - Conversion completed successfully")
-            
-            success_print(
-                f"Document '{validated_path.name}' converted to '{export_format.value}'"
-            )
-            
-            return formatted_result
-            
-        except IrisError:
-            # Re-raise exceptions from the class itself
-            raise
+                success_print('Irir - Convertido com sucesso!')
+
+            return self.result_formatted
+
+        except IrisError as ie:
+            error_print(str(ie))
+            return None
         except Exception as e:
-            error_msg = f"Unexpected error while processing document: {e}"
-            error_print(f"Iris - {error_msg}")
-            error_print(error_msg)
-            raise IrisError(error_msg) from e
-    
-    def read_multiple_documents(
-        self,
-        file_paths: List[Union[str, Path]],
-        export_format: ExportFormat = ExportFormat.MARKDOWN,
-        verbose: bool = False,
-    ) -> Dict[str, Optional[Any]]:
-        """
-        Reads and converts multiple documents.
-        
-        Args:
-            file_paths: List of file paths.
-            export_format: Desired export format.
-            verbose: If True, displays detailed messages.
-            
-        Returns:
-            Dictionary with the file name as key and converted content as value.
-            
-        Example:
-            >>> iris = Iris()
-            >>> files = ["doc1.pdf", "doc2.txt"]
-            >>> results = iris.read_multiple_documents(files, ExportFormat.TEXT)
-            >>> for filename, content in results.items():
-            ...     print(f"{filename}: {len(content) if content else 0} characters")
-        """
-        results = {}
-        successful_conversions = 0
-        
-        for file_path in file_paths:
-            try:
-                content = self.read_document(file_path, export_format, verbose=False)
-                filename = Path(file_path).name
-                results[filename] = content
-                successful_conversions += 1
-                
-                if verbose:
-                    if self.display_message: success_print(f"Iris - '{filename}' converted successfully")
-                    
-            except IrisError as e:
-                filename = Path(file_path).name
-                results[filename] = None
-                if verbose:
-                    error_print(f"Iris - Error converting '{filename}': {e}")
-                alert_print(f"Failed to convert '{filename}': {e}")
-        
-        if verbose:
-            total_files = len(file_paths)
-            if self.display_message: success_print(
-                f"Iris - Processing completed: {successful_conversions}/{total_files} "
-                f"files converted successfully"
-            )
-        
-        return results
-    
-    def get_supported_extensions(self) -> List[str]:
-        """
-        Returns the list of supported file extensions.
-        
-        Returns:
-            Sorted list of supported extensions.
-        """
-        return sorted(list(self.SUPPORTED_EXTENSIONS))
-    
-    def is_file_supported(self, file_path: Union[str, Path]) -> bool:
-        """
-        Checks if a file is supported by the class.
-        
-        Args:
-            file_path: Path to the file.
-            
-        Returns:
-            True if the file is supported, False otherwise.
-        """
-        try:
-            path_obj = Path(file_path)
-            return path_obj.suffix.lower() in self.SUPPORTED_EXTENSIONS
-        except Exception:
-            return False
+            error_print(f"Iris - Error: Erro inesperado ao ler o documento: {e}")
+            raise IrisError(f"Erro inesperado ao ler o documento: {e}")

rpa_suite/functions/__init__.py

@@ -1,3 +1,3 @@
 # rpa_suite/functions/__init__.py
 
-__version__ = "1.5.5"
+__version__ = "1.6.1"

rpa_suite/suite.py

@@ -155,7 +155,8 @@ class Suite:
     Parallel: ParallelRunner = ParallelRunner
     Asyn: AsyncRunner = AsyncRunner
 
-    # On this case, we are importing the (Browser | Iris) class only if the selenium and webdriver_manager modules are installed.
+
+    # On this case, we are importing the (Browser | Iris) class only if the (selenium and webdriver_manager| docling) modules are installed.
     # This is useful to avoid unnecessary imports and dependencies if the user does not need the (Browser | Iris) functionality.
     import importlib.util
 
@@ -334,7 +335,7 @@ class Suite:
             "setuptools",
             "wheel",
             "pyperclip",
-            "pywin32"
+            "pywin32",
             "colorama",
             "colorlog",
             "email_validator",

rpa_suite/utils/__init__.py

@@ -12,4 +12,4 @@ O módulo de utilitários da rpa-suite fornece uma coleção de funções e clas
 from .system import Tools
 
 
-__version__ = "1.5.5"
+__version__ = "1.6.1"

{rpa_suite-1.6.0.dist-info → rpa_suite-1.6.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rpa_suite
-Version: 1.6.0
+Version: 1.6.1
 Summary: Conjunto de ferramentas essenciais para Automação RPA com Python, que facilitam o dia a dia de desenvolvimento.
 Author: Camilo Costa de Carvalho
 Author-email: camilo.carvalho@vettracode.com
@@ -186,12 +186,12 @@ No setup do nosso projeto já estão inclusas as dependências, só será necess
 - pillow
 - pyautogui
 - typing
-- setuptools
 
-  opcionalmente para automação de navegador:
+  opcionalmente para usar todas funcionalidades:
 
   - selenium
   - webdriver_manager
+  - docling
 
 <br>
 <hr>
@@ -285,14 +285,12 @@ O módulo principal do rpa-suite é dividido em categorias. Cada categoria cont
   **Iris (OCR-IA)**
 
   - **read_document** - Reads and converts a document to the specified format.
-  - **read_multiple_documents** - Reads and converts multiple documents.
-  - **get_supported_extensions** - Returns the list of supported file extensions
-  - **is_file_supported** - Checks if a file is supported by the class.
 
+<br>
 
 ## Release Notes
 
-### Versão: **Beta 1.6.0**
+### Versão: **Beta 1.6.1**
 
 - **Data de Lançamento:** *20/02/2024*
 - **Última Atualização:** 08/06/2025

{rpa_suite-1.6.0.dist-info → rpa_suite-1.6.1.dist-info}/RECORD

@@ -1,6 +1,6 @@
-rpa_suite/__init__.py,sha256=WqIndloIi1jP_YsEf0lGNOxgr7Htdgt6XuL5ti5znb8,2915
-rpa_suite/suite.py,sha256=j07ZkaJttJ9WhfIKcc7j6h11ZssRzWdpcvXQDBYZT_s,11804
-rpa_suite/core/__init__.py,sha256=2KWotqRNuCNwVhACACB4zhrXnTWR9H77_6U6j0WTJK0,1738
+rpa_suite/__init__.py,sha256=olciiOHwT6n0OtAxGp2QH-GTLuS0IvfvnnJOObiKIMo,2915
+rpa_suite/suite.py,sha256=ylGDXRhyIWM3wibmG_R0UbeHKz5-hYUaTT92LvA9IaU,11818
+rpa_suite/core/__init__.py,sha256=dmW1RHmaX17QDJ-7lZgZys6JcY1o0kaoplAWSOnZpXY,1858
 rpa_suite/core/asyncrun.py,sha256=gRKsqvT4QAwg906BkLQXHi-oMbjM30D3yRWV1qAqj1Y,4192
 rpa_suite/core/browser.py,sha256=NeJk8lWDKZcGR9ULfWkDZ4WmFujU-DVr5-QH0qUSSgU,14725
 rpa_suite/core/clock.py,sha256=ELhgehLoqrf5bjkDpJ8wkcha9YmsnIfLb0qQW7OKrzw,13161
@@ -8,19 +8,19 @@ rpa_suite/core/date.py,sha256=nnAktYMZNjcN4e6HEiYJgdMLD5VZluaOjfyfSPaz71c,6307
 rpa_suite/core/dir.py,sha256=ZfgFeCkl8iB8Tc5dST35olImpj4PoWThovNYvtpwnu8,10329
 rpa_suite/core/email.py,sha256=D69vPmoBJYwSTgDu5tvXhakvsYprXr0BAFRYeaVicx0,8473
 rpa_suite/core/file.py,sha256=hCXoWiEGtxRfp5Uq33p0f2eDwKUv3dEiUSajOhpNwbc,11317
-rpa_suite/core/iris.py,sha256=Z8aP96y78-DnzUxwtxJZiIXXOLXWTCsaWpdMoxROfQY,11467
+rpa_suite/core/iris.py,sha256=x-CvbIVtThFpA1AtmVlm-Ps97jcCuk-De-22WF3_qRo,5461
 rpa_suite/core/log.py,sha256=9dPDnV8e4p9lwZoyd1ICb6CjJiiSXTXVJseQkdtdRuQ,6542
 rpa_suite/core/parallel.py,sha256=a_aEqvoJ9jxsFg1H42wsPT2pCS3WApqbGc2PETgBBEs,11460
 rpa_suite/core/print.py,sha256=i1icdpNreQf2DCO6uLQKuuUD0vsrsOnYSpiQGaGNJi4,5780
 rpa_suite/core/regex.py,sha256=IHQF-xHVacDuloQqcBJdTCjd7oXVqDdbGa50Mb803Bk,3321
 rpa_suite/core/validate.py,sha256=Msk_bL9pBuenuUzFv7Wg9L_z3zXq0lOHsDavkwfaAn0,10620
 rpa_suite/functions/__create_ss_dir.py,sha256=kaRotLlYDCQGKtv9nd8zZUorQmHYGbHmOEWJ1DZBBYc,3426
-rpa_suite/functions/__init__.py,sha256=Y9Kp8tTmyCcQ4sErjb0c2cbDNTAAoTArEF2pYl7mt5s,57
+rpa_suite/functions/__init__.py,sha256=7u63cRyow2OYQMt6Ph5uYImM_aUeMqdMaOvvO5v698Y,57
 rpa_suite/functions/_printer.py,sha256=gj7dwOt4roSj2iwOGWeGgUD3JVr7h4UESyCg9CmrieA,3946
-rpa_suite/utils/__init__.py,sha256=bqxq5kckulcQzNCn1tHwHj0WMIQBTUYNDzMzBhLtbIY,729
+rpa_suite/utils/__init__.py,sha256=VAPzxR_aW-8kWsAUYzvrFdkH_aRLXywTUvj_qah9GwM,729
 rpa_suite/utils/system.py,sha256=kkTsjwBQ-8_G_6l-0tuwkpmeI3KVssRZ7QAiYlR3vt0,5185
-rpa_suite-1.6.0.dist-info/licenses/LICENSE,sha256=5D8PIbs31iGd9i1_MDNg4SzaQnp9sEIULALh2y3WyMI,1102
-rpa_suite-1.6.0.dist-info/METADATA,sha256=sIX1G91PeiHFKxPIrckrZFQ32BgGfv5Mf4YwKFsM4sQ,13546
-rpa_suite-1.6.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-rpa_suite-1.6.0.dist-info/top_level.txt,sha256=HYkDtg-kJNAr3F2XAIPyJ-QBbNhk7q6jrqsFt10lz4Y,10
-rpa_suite-1.6.0.dist-info/RECORD,,
+rpa_suite-1.6.1.dist-info/licenses/LICENSE,sha256=5D8PIbs31iGd9i1_MDNg4SzaQnp9sEIULALh2y3WyMI,1102
+rpa_suite-1.6.1.dist-info/METADATA,sha256=wLNug0aSQf_uGvoVplzWELy6a9LL7Vc_BOh98MIwRt0,13324
+rpa_suite-1.6.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+rpa_suite-1.6.1.dist-info/top_level.txt,sha256=HYkDtg-kJNAr3F2XAIPyJ-QBbNhk7q6jrqsFt10lz4Y,10
+rpa_suite-1.6.1.dist-info/RECORD,,