comparison COBRAxy/utils/general_utils.py @ 380:03a7ba63813f draft

Uploaded

379:6f0aed06098c

    PDF    = ("pdf",) # this is also a common output format for images, as it's required in publications.
    
    # Updated to include compressed variants
    XML    = ("xml", "xml.gz", "xml.zip", "xml.bz2") # SBML files are XML files, sometimes compressed
    JSON   = ("json", "json.gz", "json.zip", "json.bz2") # COBRA models can be stored as JSON files, sometimes compressed
    
    TXT = ("txt",) # this is how most output data is written
    PICKLE = ("pickle", "pk", "p") # this is how all runtime data structures are saved

    def __init__(self, *extensions):
        self.extensions = extensions
        # Store original extension when set via fromExt

        """
        (Private) converts to str representation. Good practice for usage with argparse.
        Returns:
            str : the string representation of the file extension.
        """
        # If we have an original extension stored (for compressed files), use it
        if hasattr(self, '_original_extension') and self._original_extension:
            return self._original_extension
        
        # For XML and JSON without original extension, use the base extension
        if self == FileFormat.XML:
            return "xml"
        elif self == FileFormat.JSON:
            return "json"
        
        return self.value[-1]

class FilePath():
    """

            if str(ext) in FileFormat.JSON.value:
                # Compressed files are not automatically handled by cobra
                if(ext == "json"):
                    return cobra.io.load_json_model(file_path.show())
                else: 
                    return self.extract_json_model(file_path, ext)

        except Exception as e: raise DataErr(file_path, e.__str__())
        raise DataErr(file_path,
            f"Fomat \"{file_path.ext}\" is not recognized, only JSON and XML files are supported.")
    

    def extract_json_model(file_path:FilePath, ext :FileFormat) -> cobra.Model:
        """
        Extract json COBRA model from a compressed file (zip, gz, bz2).
        
        Args:
            file_path: File path of the model
            ext: File extensions of class FileFormat (should be .zip, .gz or .bz2)
            

        try:
            if '.zip' in ext_str:
                with zipfile.ZipFile(file_path.show(), 'r') as zip_ref:
                    with zip_ref.open(zip_ref.namelist()[0]) as json_file:
                        content = json_file.read().decode('utf-8')
                        return cobra.io.load_json_model(StringIO(content))
            elif '.gz' in ext_str:
                with gzip.open(file_path.show(), 'rt', encoding='utf-8') as gz_ref:
                    return cobra.io.load_json_model(gz_ref)
            elif '.bz2' in ext_str:
                with bz2.open(file_path.show(), 'rt', encoding='utf-8') as bz2_ref:
                    return cobra.io.load_json_model(bz2_ref)
            else:
                raise ValueError(f"Compression format not supported: {ext_str}. Supported: .zip, .gz and .bz2")
            
        except Exception as e:
            raise Exception(f"Error during model extraction: {str(e)}")

380:03a7ba63813f

    PDF    = ("pdf",) # this is also a common output format for images, as it's required in publications.
    
    # Updated to include compressed variants
    XML    = ("xml", "xml.gz", "xml.zip", "xml.bz2") # SBML files are XML files, sometimes compressed
    JSON   = ("json", "json.gz", "json.zip", "json.bz2") # COBRA models can be stored as JSON files, sometimes compressed
    MAT    = ("mat", "mat.gz", "mat.zip", "mat.bz2") # COBRA models can be stored as MAT files, sometimes compressed
    YML    = ("yml", "yml.gz", "yml.zip", "yml.bz2") # COBRA models can be stored as YML files, sometimes compressed

    TXT    = ("txt",) # this is how most output data is written
    PICKLE = ("pickle", "pk", "p") # this is how all runtime data structures are saved

    def __init__(self, *extensions):
        self.extensions = extensions
        # Store original extension when set via fromExt

        """
        (Private) converts to str representation. Good practice for usage with argparse.
        Returns:
            str : the string representation of the file extension.
        """
        # If we have an original extension stored (for compressed files only), use it
        if hasattr(self, '_original_extension') and self._original_extension:
            return self._original_extension
        
        # For XML, JSON, MAT and YML without original extension, use the base extension
        if self == FileFormat.XML:
            return "xml"
        elif self == FileFormat.JSON:
            return "json"
        elif self == FileFormat.MAT:
            return "mat"
        elif self == FileFormat.YML:
            return "yml"
        
        return self.value[-1]

class FilePath():
    """

            if str(ext) in FileFormat.JSON.value:
                # Compressed files are not automatically handled by cobra
                if(ext == "json"):
                    return cobra.io.load_json_model(file_path.show())
                else: 
                    return self.extract_model(file_path, ext, "json")

            if str(ext) in FileFormat.MAT.value:
                # Compressed files are not automatically handled by cobra
                if(ext == "mat"):
                    return cobra.io.load_matlab_model(file_path.show())
                else: 
                    return self.extract_model(file_path, ext, "mat")

            if str(ext) in FileFormat.YML.value:
                # Compressed files are not automatically handled by cobra
                if(ext == "yml"):
                    return cobra.io.load_yaml_model(file_path.show())
                else: 
                    return self.extract_model(file_path, ext, "yml")

        except Exception as e: raise DataErr(file_path, e.__str__())
        raise DataErr(file_path,
            f"Fomat \"{file_path.ext}\" is not recognized, only JSON, XML, MAT and YAML (.yml) files are supported.")
    

    def extract_model(file_path:FilePath, ext :FileFormat, model_encoding:Literal["json", "mat", "yml"]) -> cobra.Model:
        """
        Extract JSON, MAT and YAML COBRA model from a compressed file (zip, gz, bz2).
        
        Args:
            file_path: File path of the model
            ext: File extensions of class FileFormat (should be .zip, .gz or .bz2)
            

        try:
            if '.zip' in ext_str:
                with zipfile.ZipFile(file_path.show(), 'r') as zip_ref:
                    with zip_ref.open(zip_ref.namelist()[0]) as json_file:
                        content = json_file.read().decode('utf-8')
                        if model_encoding == "json":
                            return cobra.io.load_json_model(StringIO(content))
                        elif model_encoding == "mat":
                            return cobra.io.load_matlab_model(StringIO(content))
                        elif model_encoding == "yml":
                            return cobra.io.load_yaml_model(StringIO(content))
                        else:
                            raise ValueError(f"Unsupported model encoding: {model_encoding}. Supported: json, mat, yml")
            elif '.gz' in ext_str:
                with gzip.open(file_path.show(), 'rt', encoding='utf-8') as gz_ref:
                    if model_encoding == "json":
                        return cobra.io.load_json_model(gz_ref)
                    elif model_encoding == "mat":
                        return cobra.io.load_matlab_model(gz_ref)
                    elif model_encoding == "yml":
                        return cobra.io.load_yaml_model(gz_ref)
                    else:
                        raise ValueError(f"Unsupported model encoding: {model_encoding}. Supported: json, mat, yml")
            elif '.bz2' in ext_str:
                with bz2.open(file_path.show(), 'rt', encoding='utf-8') as bz2_ref:
                    if model_encoding == "json":
                        return cobra.io.load_json_model(bz2_ref)
                    elif model_encoding == "mat":
                        return cobra.io.load_matlab_model(bz2_ref)
                    elif model_encoding == "yml":
                        return cobra.io.load_yaml_model(bz2_ref)
                    else:
                        raise ValueError(f"Unsupported model encoding: {model_encoding}. Supported: json, mat, yml")
            else:
                raise ValueError(f"Compression format not supported: {ext_str}. Supported: .zip, .gz and .bz2")
            
        except Exception as e:
            raise Exception(f"Error during model extraction: {str(e)}")