Input Specification

As it was mentioned in Shell Command Tasks, the user can customize the input and output for the ShellCommandTask. In this section, more examples of the input specification will be provided.

Let’s start from the previous example:

bet_input_spec = SpecInfo(
    ( "in_file", File,
      { "help_string": "input file ...",
        "position": 1,
        "mandatory": True } ),
    ( "out_file", str,
      { "help_string": "name of output ...",
        "position": 2,
                          "{in_file}_br" } ),
    ( "mask", bool,
      { "help_string": "create binary mask",
        "argstr": "-m", } ) ],
    bases=(ShellSpec,) )


In order to create an input specification, a new SpecInfo object has to be created. The field name specifies the type of the spec and it should be always “Input” for the input specification. The field bases specifies the “base specification” you want to use (can think about it as a parent class) and it will usually contains ShellSpec only, unless you want to build on top of your other specification (this will not be cover in this section). The part that should be always customised is the fields part. Each element of the fields is a separate input field that is added to the specification. In this example, three-elements tuples - with name, type and dictionary with additional information - are used. But this is only one of the supported syntax, more options will be described below.

Adding a New Field to the Spec

Pydra uses attr classes to represent the input specification, and the full syntax for each field is:

field1 = ("field1_name", attr.ib(type=<'field1_type'>, metadata=<'dictionary with metadata'>)

However, we allow for shorter syntax, that does not include attr.ib:

  • providing only name and the type

field1 = ("field1_name", <'field1_type'>)
  • providing name, type and metadata (as in the example above)

field1 = ("field1_name", <'field1_type'>, <'dictionary with metadata'>))
  • providing name, type and default value

field1 = ("field1_name", <'field1_type'>, <'default value'>)
  • providing name, type, default value and metadata

field1 = ("field1_name", <'field1_type'>, <'default value', <'dictionary with metadata'>))

Each of the shorter versions will be converted to the (name, attr.ib(…).


Type can be provided as a simple python type (e.g. str, int, float, etc.) or can be more complex by using typing.List, typing.Dict and typing.Union.

There are also special types provided by Pydra:

  • File and Directory - should be used in input_spec if the field is an existing file or directory. Pydra checks if the file or directory exists, and returns an error if it doesn’t exist.

  • MultiInputObj - a special type that takes a any value and if the value is not a list it converts value to a 1-element list (it could be used together with MultiOutputObj in the output_spec to reverse the conversion of the output values).


In the example we used multiple keys in the metadata dictionary including help_string, position, etc. In this section all allowed key will be described:

help_string (str, mandatory):

A short description of the input field.

mandatory (bool, default: False):

If True user has to provide a value for the field.

sep (str):

A separator if a list is provided as a value.

argstr (str):

A flag or string that is used in the command before the value, e.g. -v or -v {inp_field}, but it could be and empty string, “”. If are used, e.g. -v…, the flag is used before every element if a list is provided as a value. If no argstr is used the field is not part of the command.

position (int):

Position of the field in the command, could be nonnegative or negative integer. If nothing is provided the field will be inserted between all fields with nonnegative positions and fields with negative positions.

allowed_values (list):

List of allowed values for the field.

requires (list):

List of field names that are required together with the field.

xor (list):

List of field names that are mutually exclusive with the field.

copyfile (bool, default: False):

If True, a hard link is created for the input file in the output directory. If hard link not possible, the file is copied to the output directory.

container_path (bool, default: False, only for ContainerTask):

If True a path will be consider as a path inside the container (and not as a local path).

output_file_template (str):

If provided, the field is treated also as an output field and it is added to the output spec. The template can use other fields, e.g. {file1}. Used in order to create an output specification.

output_field_name (str, used together with output_file_template)

If provided the field is added to the output spec with changed name. Used in order to create an output specification.

keep_extension (bool, default: True):

A flag that specifies if the file extension should be removed from the field value. Used in order to create an output specification.

readonly (bool, default: False):

If True the input field can’t be provided by the user but it aggregates other input fields (for example the fields with argstr: -o {fldA} {fldB}).

formatter (function):

If provided the argstr of the field is created using the function. This function can for example be used to combine several inputs into one command argument. The function can take field (this input field will be passed to the function), inputs (entire inputs will be passed) or any input field name (a specific input field will be sent).


Pydra allows for using simple validator for types and allowev_values. The validators are disabled by default, but can be enabled by calling pydra.set_input_validator(flag=True).