Merge pull request #22 from illusional/parsing

Primitive parsing of CWL

Author: illusional

cwlgen/__init__.py

@@ -14,7 +14,7 @@
 
 from .utils import literal, literal_presenter
 
-from .import_cwl import parse_cwl
+from .import_cwl import parse_cwl, parse_cwl_dict
 
 logging.basicConfig(level=logging.INFO)
 _LOGGER = logging.getLogger(__name__)
@@ -24,6 +24,7 @@
 from .common import *
 from .commandlinetool import *
 from .workflow import *
+from .workflowdeps import *
 from .commandlinebinding import CommandLineBinding
 from .requirements import *
 

cwlgen/commandlinetool.py

@@ -3,6 +3,7 @@
 # External libraries
 import ruamel.yaml
 
+from cwlgen.commandlinebinding import CommandLineBinding
 from .common import CWL_VERSIONS, DEF_VERSION, CWL_SHEBANG, Namespaces, Parameter
 from .requirements import *
 from .utils import literal, literal_presenter, Serializable
@@ -15,32 +16,128 @@
 #  Class(es)  ------------------------------
 
 
+class CommandOutputBinding(Serializable):
+    '''
+    Describes how to generate an output parameter based on the files produced.
+    '''
+
+    def __init__(self, glob=None, load_contents=None, output_eval=None):
+        '''
+        :param glob: Find corresponding file(s)
+        :type glob: STRING
+        :param load_contents: For each file matched, read up to the 1st 64 KiB of text and
+                              place it in the contents field
+        :type load_contents: BOOLEAN
+        :param output_eval: Evaluate an expression to generate the output value
+        :type output_eval: STRING
+        '''
+        self.glob = glob
+        self.loadContents = load_contents
+        self.outputEval = output_eval
+
+
+class CommandInputParameter(Parameter):
+    """
+    An input parameter for a :class:`cwlgen.CommandLineTool`.
+    """
+
+    parse_types = {"inputBinding": [CommandLineBinding]}
+
+    def __init__(self, param_id, label=None, secondary_files=None, param_format=None,
+                 streamable=None, doc=None, input_binding=None, default=None, param_type=None):
+        '''
+        :param param_id: unique identifier for this parameter
+        :type param_id: STRING
+        :param label: short, human-readable label
+        :type label: STRING
+        :param secondary_files: If type is a file, describes files that must be
+                                included alongside the primary file(s)
+        :type secondary_files: STRING
+        :param param_format: If type is a file, uri to ontology of the format or exact format.
+        :type param_format: STRING
+        :param streamable: If type is a file, true indicates that the file is read or written
+                           sequentially without seeking
+        :type streamable: BOOLEAN
+        :param doc: documentation
+        :type doc: STRING
+        :param input_binding: describes how to handle the input
+        :type input_binding: :class:`cwlgen.CommandLineBinding` object
+        :param default: default value
+        :type default: STRING
+        :param param_type: type of data assigned to the parameter corresponding to CWLType
+        :type param_type: STRING
+        '''
+        Parameter.__init__(self, param_id=param_id, label=label,
+                           secondary_files=secondary_files, param_format=param_format,
+                           streamable=streamable, doc=doc, param_type=param_type)
+        self.inputBinding = input_binding
+        self.default = default
+
+
+class CommandOutputParameter(Parameter):
+    '''
+    An output parameter for a :class:`cwlgen.CommandLineTool`.
+    '''
+
+    parse_types = {"outputBinding": [CommandOutputBinding]}
+
+    def __init__(self, param_id, label=None, secondary_files=None, param_format=None,
+                 streamable=None, doc=None, output_binding=None, param_type=None):
+        '''
+        :param param_id: unique identifier for this parameter
+        :type param_id: STRING
+        :param label: short, human-readable label
+        :type label: STRING
+        :param secondary_files: If type is a file, describes files that must be
+                                included alongside the primary file(s)
+        :type secondary_files: STRING
+        :param param_format: If type is a file, uri to ontology of the format or exact format
+        :type param_format: STRING
+        :param streamable: If type is a file, true indicates that the file is read or written
+                           sequentially without seeking
+        :type streamable: BOOLEAN
+        :param doc: documentation
+        :type doc: STRING
+        :param output_binding: describes how to handle the output
+        :type output_binding: :class:`cwlgen.CommandOutputBinding` object
+        :param param_type: type of data assigned to the parameter corresponding to CWLType
+        :type param_type: STRING
+        '''
+        Parameter.__init__(self, param_id, label, secondary_files, param_format, streamable,
+                           doc, param_type)
+        self.outputBinding = output_binding
+
+
 class CommandLineTool(Serializable):
     '''
     Contain all informations to describe a CWL command line tool.
     '''
 
     __CLASS__ = 'CommandLineTool'
 
+    parse_types = {'inputs': [[CommandInputParameter]], "outputs": [[CommandOutputParameter]]}
+    ignore_fields_on_parse = ["namespaces", "class"]
+    ignore_fields_on_convert = ["namespaces", "class", "metadata", "requirements"]
+
     def __init__(self, tool_id=None, base_command=None, label=None, doc=None,
                  cwl_version=None, stdin=None, stderr=None, stdout=None, path=None):
         '''
         :param tool_id: unique identifier for this tool
-        :type tool_id: STRING
+        :type tool_id: str
         :param base_command: command line for the tool
-        :type base_command: STRING | list[STRING]
+        :type base_command: str | list[STRING]
         :param label: label of this tool
-        :type label: STRING
+        :type label: str
         :param doc: documentation for the tool, usually longer than the label
-        :type doc: STRING
+        :type doc: str
         :param cwl_version: version of the CWL tool
-        :type cwl_version: STRING
+        :type cwl_version: str
         :param stdin: path to a file whose contents must be piped into stdin
-        :type stdin: STRING
+        :type stdin: str
         :param stderr: capture stderr into the given file
-        :type stderr: STRING
+        :type stderr: str
         :param stdout: capture stdout into the given file
-        :type stdout: STRING
+        :type stdout: str
 
         inputs (:class:`cwlgen.CommandInputParameter` objects),
         outputs (:class:`cwlgen.CommandOutputParameter` objects),
@@ -72,7 +169,6 @@ def __init__(self, tool_id=None, base_command=None, label=None, doc=None,
         self._path = path
         self.namespaces = Namespaces()
         self.metadata = {}
-        self.ignore_attributes = ["namespaces"]
 
     def get_dict(self):
 
@@ -113,91 +209,3 @@ def export(self, outfile=None):
             out_write.write(CWL_SHEBANG + '\n\n')
             out_write.write(rep)
             out_write.close()
-
-
-class CommandInputParameter(Parameter):
-    '''
-    An input parameter for a :class:`cwlgen.CommandLineTool`.
-    '''
-
-    def __init__(self, param_id, label=None, secondary_files=None, param_format=None,
-                 streamable=None, doc=None, input_binding=None, default=None, param_type=None):
-        '''
-        :param param_id: unique identifier for this parameter
-        :type param_id: STRING
-        :param label: short, human-readable label
-        :type label: STRING
-        :param secondary_files: If type is a file, describes files that must be
-                                included alongside the primary file(s)
-        :type secondary_files: STRING
-        :param param_format: If type is a file, uri to ontology of the format or exact format.
-        :type param_format: STRING
-        :param streamable: If type is a file, true indicates that the file is read or written
-                           sequentially without seeking
-        :type streamable: BOOLEAN
-        :param doc: documentation
-        :type doc: STRING
-        :param input_binding: describes how to handle the input
-        :type input_binding: :class:`cwlgen.CommandLineBinding` object
-        :param default: default value
-        :type default: STRING
-        :param param_type: type of data assigned to the parameter corresponding to CWLType
-        :type param_type: STRING
-        '''
-        Parameter.__init__(self, param_id=param_id, label=label,
-                           secondary_files=secondary_files, param_format=param_format,
-                           streamable=streamable, doc=doc, param_type=param_type)
-        self.inputBinding = input_binding
-        self.default = default
-
-
-class CommandOutputParameter(Parameter):
-    '''
-    An output parameter for a :class:`cwlgen.CommandLineTool`.
-    '''
-
-    def __init__(self, param_id, label=None, secondary_files=None, param_format=None,
-                 streamable=None, doc=None, output_binding=None, param_type=None):
-        '''
-        :param param_id: unique identifier for this parameter
-        :type param_id: STRING
-        :param label: short, human-readable label
-        :type label: STRING
-        :param secondary_files: If type is a file, describes files that must be
-                                included alongside the primary file(s)
-        :type secondary_files: STRING
-        :param param_format: If type is a file, uri to ontology of the format or exact format
-        :type param_format: STRING
-        :param streamable: If type is a file, true indicates that the file is read or written
-                           sequentially without seeking
-        :type streamable: BOOLEAN
-        :param doc: documentation
-        :type doc: STRING
-        :param output_binding: describes how to handle the output
-        :type output_binding: :class:`cwlgen.CommandOutputBinding` object
-        :param param_type: type of data assigned to the parameter corresponding to CWLType
-        :type param_type: STRING
-        '''
-        Parameter.__init__(self, param_id, label, secondary_files, param_format, streamable,
-                           doc, param_type)
-        self.outputBinding = output_binding
-
-
-class CommandOutputBinding(Serializable):
-    '''
-    Describes how to generate an output parameter based on the files produced.
-    '''
-
-    def __init__(self, glob=None, load_contents=None, output_eval=None):
-        '''
-        :param glob: Find corresponding file(s)
-        :type glob: STRING
-        :param load_contents: For each file matched, read up to the 1st 64 KiB of text and
-                              place it in the contents field
-        :type load_contents: BOOLEAN
-        :param output_eval: Evaluate an expression to generate the output value
-        :type output_eval: STRING
-        '''
-        self.glob = glob
-        self.loadContents = load_contents
-        self.outputEval = output_eval

cwlgen/common.py

@@ -116,7 +116,7 @@ class CwlTypes:
 
 # functions
 
-class Namespaces(object):
+class Namespaces(Serializable):
     """
     Define different namespace for the description.
     """
@@ -172,6 +172,20 @@ def __init__(self, param_id, label=None, secondary_files=None, param_format=None
         self.doc = doc
         self.type = parse_type(param_type, requires_type)
 
+    @classmethod
+    def parse_with_id(cls, d, identifier):
+        d["id"] = identifier
+        return super(Parameter, cls).parse_dict(d)
+
+    # @classmethod
+    # def parse_dict(cls, d):
+    #     self = super(Parameter, cls).parse_dict(d)
+    #     secs = d.get("secondaryFiles")
+    #     self.secondaryFiles = []
+    #     if secs:
+    #         self.secondaryFiles = secs if isinstance(secs, list) else [secs]
+    #     return self
+
 
 class CommandInputArraySchema(Serializable):
     '''