small fixes

Author: rafaelfsilva

workflowhub/common/machine.py

@@ -74,9 +74,9 @@ def __init__(self,
         self.hashcode = hashcode
 
         self.cpu_cores: int = cpu['count']
-        self.cpu_speed: int = cpu['speed']
-        self.cpu_flops: int = cpu['count'] * cpu['speed'] * 10 ^ 6
-        self.cpu_vendor: str = cpu['vendor']
+        self.cpu_speed: int = cpu['speed'] if 'speed' in cpu else 0
+        self.cpu_flops: int = cpu['count'] * cpu['speed'] * 10 ^ 6 if 'speed' in cpu else 0
+        self.cpu_vendor: str = cpu['vendor'] if 'vendor' in cpu else None
 
         self.logger.debug("created machine: {0} with {1} cores and {2} FLOPS.".format(
             self.name, self.cpu_cores, self.cpu_flops)

workflowhub/trace/logs/makeflow.py

@@ -118,7 +118,7 @@ def _parse_workflow_file(self):
 
                 elif len(line.strip()) > 0:
                     # task execution command
-                    prefix = line.split()[1 if 'LOCAL' in line else 0].replace('./', '')
+                    prefix = line.replace('./', '').replace('perl', '').strip().split()[1 if 'LOCAL' in line else 0]
                     task_name = "{}_ID{:06d}".format(prefix, task_id_counter)
                     task_id_counter += 1
 
@@ -128,7 +128,7 @@ def _parse_workflow_file(self):
                     list_files.extend(self._create_files(inputs, FileLink.INPUT, task_name))
 
                     # create task
-                    args = line.replace('LOCAL', '').strip()
+                    args = ' '.join(line.replace('LOCAL', '').replace('perl', '').strip().split())
                     task = Task(name=task_name,
                                 task_type=TaskType.COMPUTE,
                                 runtime=0,
@@ -210,7 +210,7 @@ def _parse_resource_monitor_logs(self):
                 data = json.load(f)
 
                 # task
-                task = self.args_map[data['command']]
+                task = self.args_map[data['command'].replace('perl', '').strip()]
                 task.runtime = float(data['wall_time'][0])
                 task.cores = int(data['cores'][0])
                 task.memory = int(data['memory'][0]) * 1000  # MB to KB

workflowhub/trace/trace.py

@@ -88,7 +88,7 @@ def __init__(self, input_trace: str, schema_file: Optional[str] = None, logger:
             machine['nodeName']: Machine(
                 name=machine['nodeName'],
                 cpu={k: v for k, v in machine['cpu'].items()},
-                system=MachineSystem(machine.get('system', None)),
+                system=MachineSystem(machine.get('system', None)) if machine.get('system', None) else None,
                 architecture=machine.get('architecture', None),
                 memory=machine.get('memory', None),
                 release=machine.get('release', None),