Adds files to files entry

Author: tainagdcoleman

wfcommons/common/workflow.py

@@ -177,9 +177,18 @@ def write_json(self, json_file_path: Optional[pathlib.Path] = None) -> None:
                     machines_list.append(machine.name)
                     workflow_machines.append(machine.as_dict())
 
+            # add files to the workflow json object (input and output)
+            for file in task.input_files:
+                files.append(file.as_dict())
+            for file in task.output_files:
+                files.append(file.as_dict())
+
         if workflow_machines:
             workflow_json["workflow"]["execution"]["machines"] = workflow_machines
 
+        if files and len(files) > 0:
+            workflow_json["workflow"]["specification"]["files"] = files
+
         # write to file
         if not json_file_path:
             json_file_path = pathlib.Path(f"{self.name.lower()}.json")

wfcommons/wfgen/abstract_recipe.py

@@ -205,8 +205,6 @@ def _generate_task_files(self, task: Task) -> List[File]:
         output_files_list = self._generate_files(task.task_id, task_recipe['output'], FileLink.OUTPUT)
         task.output_files = self.tasks_files[task.task_id]
 
-        
-
         # obtain input files from parents
         input_files = []
 
@@ -221,13 +219,14 @@ def _generate_task_files(self, task: Task) -> List[File]:
             if input_file not in self.tasks_files_names[task.task_id]:
                 self.tasks_files[task.task_id].append(File(name=input_file,
                                                            link=FileLink.INPUT,
-                                                            size=input_file.size))
+                                                           size=input_file.size))
                 self.tasks_files_names[task.task_id].append(input_file)
 
         # generate additional input files
         self._generate_files(task.task_id, task_recipe['input'], FileLink.INPUT)
         task.input_files = [ifile for ifile in self.tasks_files[task.task_id] if ifile.link == FileLink.INPUT]
 
+
         return output_files_list
 
     def _generate_files(self, task_id: str, recipe: Dict[str, Any], link: FileLink) -> List[File]:
@@ -248,9 +247,9 @@ def _generate_files(self, task_id: str, recipe: Dict[str, Any], link: FileLink)
         extension_list: List[str] = []
         for f in self.tasks_files[task_id]:
             if f.link == link:
+                extension_list.append(path.splitext(f.file_id)[1] if '.' in f.file_id else f.file_id)
                 files_list.append(f)
-                extension_list.append(path.splitext(f.name)[1] if '.' in f.name else f.name)
-
+            
         for extension in recipe:
             if extension not in extension_list:
                 file = self._generate_file(extension, recipe, link)
@@ -278,10 +277,12 @@ def _generate_file(self, extension: str, recipe: Dict[str, Any], link: FileLink)
                     else self.output_file_size_factor) * generate_rvs(recipe[extension]['distribution'],
                                                                       recipe[extension]['min'],
                                                                       recipe[extension]['max']))
+        
         return File(file_id=str(uuid.uuid4()) + extension,
                     link=link,
                     size=size)
 
+
     def _get_files_by_task_and_link(self, task_id: str, link: FileLink) -> List[File]:
         """
         Get the list of files for a task ID and link type.

wfcommons/wfinstances/instance_analyzer.py

@@ -109,7 +109,7 @@ def build_summary(self,
                 # For each input_file and output_file, append the file size to the dictionary
                 for infile in task.input_files:
                     extension: str = path.splitext(infile.file_id)[1] if '.' in infile.file_id else infile.file_id
-                    # print(f"file {infile.file_id} extension: {extension}")
+                    
                     if extension[1:].isnumeric():
                         extension = path.splitext(infile.file_id.replace(extension, ''))[1]