Merge pull request #580 from rcali21/PROV_io

djarecka · web-flow · commit 5bbb4a96b852 · 2022-10-28T13:45:29.000-04:00
WIP-Adding software version retrieval
diff --git a/pydra/engine/audit.py b/pydra/engine/audit.py
@@ -170,21 +170,48 @@ def audit_check(self, flag):
         return self.audit_flags & flag
 
     def audit_task(self, task):
+        import subprocess as sp
+
         label = task.name
         if hasattr(task.inputs, "executable"):
             command = task.cmdline
         # assume function task
         else:
-            # work on changing this to function name
             command = None
+        # if hasattr(task.inputs, "in_file"):
+        #     input_file = task.inputs.in_file
+        # else:
+        #     input_file = None
+
+        if command is not None:
+            cmd_name = command.split()[0]
+            software = f"{cmd_name} --version"
+            # take the first word of command as the
+            # name of the executable
+            # (this may not always be the case)
+            version_cmd = sp.run(software, shell=True, stdout=sp.PIPE).stdout.decode(
+                "utf-8"
+            )
+            try:
+                version_cmd = version_cmd.splitlines()[0]
+
+            except IndexError:
+                version_cmd = f"{cmd_name} -- Version unknown"
+
+        else:
+            version_cmd = None
 
         start_message = {
             "@id": self.aid,
             "@type": "task",
-            "label": label,
-            "command": command,
-            "startedAtTime": now(),
+            "Label": label,
+            "Command": command,
+            "StartedAtTime": now(),
+            "AssociatedWith": version_cmd,
         }
+
+        # new code to be added here for i/o tracking - WIP
+
         self.audit_message(start_message, AuditFlag.PROV)
 
         # add more fields according to BEP208 doc
diff --git a/pydra/engine/tests/test_task.py b/pydra/engine/tests/test_task.py
@@ -1003,9 +1003,12 @@ def testfunc(a: int, b: float = 0.1) -> ty.NamedTuple("Output", [("out", float)]
     for file in glob(str(message_path) + "/*.jsonld"):
         with open(file, "r") as f:
             data = json.load(f)
-            if "label" in data:
+            print(data)
+            if "Label" in data:
                 json_content.append(True)
-                assert "testfunc" == data["label"]
+                assert "testfunc" == data["Label"]
+                if "AssociatedWith" in data:
+                    assert None == data["AssociatedWith"]
     assert any(json_content)
 
 
@@ -1031,16 +1034,50 @@ def test_audit_shellcommandtask(tmpdir):
     for file in glob(str(message_path) + "/*.jsonld"):
         with open(file, "r") as f:
             data = json.load(f)
-            if "label" in data:
+            print(data)
+            if "Label" in data:
                 label_content.append(True)
-            if "command" in data:
+            if "Command" in data:
                 command_content.append(True)
-                assert "ls -l" == data["command"]
+                assert "ls -l" == data["Command"]
 
     print(command_content)
     assert any(label_content)
 
 
+def test_audit_shellcommandtask_version(tmpdir):
+    import subprocess as sp
+
+    version_cmd = sp.run("less --version", shell=True, stdout=sp.PIPE).stdout.decode(
+        "utf-8"
+    )
+    version_cmd = version_cmd.splitlines()[0]
+    cmd = "less"
+    shelly = ShellCommandTask(
+        name="shelly",
+        executable=cmd,
+        args="test_task.py",
+        audit_flags=AuditFlag.PROV,
+        messengers=FileMessenger(),
+    )
+
+    import glob
+
+    shelly.cache_dir = tmpdir
+    shelly()
+    message_path = tmpdir / shelly.checksum / "messages"
+    # go through each jsonld file in message_path and check if the label field exists
+    version_content = []
+    for file in glob.glob(str(message_path) + "/*.jsonld"):
+        with open(file, "r") as f:
+            data = json.load(f)
+            if "AssociatedWith" in data:
+                if version_cmd in data["AssociatedWith"]:
+                    version_content.append(True)
+
+    assert any(version_content)
+
+
 def test_audit_prov_messdir_1(tmpdir, use_validator):
     """customized messenger dir"""
 
