Merge pull request #662 from tclose/type-coercion

Merging now ahead of alpha release 🥳 

(NB: I left out the relaxation to allow Parent (out) -> Child (in) to pass type-checking as it is a bit involved, and can wait until before we make a full release)

Author: tclose

.github/workflows/testslurm.yml

@@ -21,6 +21,8 @@ jobs:
         docker pull $DOCKER_IMAGE
         # Have image running in background
         docker run `bash <(curl -s https://codecov.io/env)` -itd -h ernie --name slurm -v `pwd`:/pydra -e NO_ET=$NO_ET $DOCKER_IMAGE
+    - name: Update python
+      run: docker exec slurm bash -c "conda install python==3.8.15"
     - name: Display previous jobs with sacct
       run: |
         echo "Allowing ports/daemons time to start" && sleep 10

.gitignore

@@ -15,6 +15,7 @@ cov.xml
 .*.swp
 *~
 .idea
+*.venv
 
 .DS_Store
 

docs/components.rst

@@ -169,15 +169,17 @@ the Task execution, the user can set splitter and combiner attributes of the Sta
 .. code-block:: python
 
   task_with_state =
-        add2(x=[1, 5]).split("x").combine("x")
+        add2().split(x=[1, 5]).combine("x")
 
 In this example, the ``State`` class is responsible for creating a list of two
 separate inputs, *[{x: 1}, {x:5}]*, each run of the *Task* should get one
-element from the list.
-The results are grouped back when returning the result from the *Task*.
-While this example
-illustrates mapping and grouping of results over a single parameter, *Pydra*
-extends this to arbitrary combinations of input fields and downstream grouping
+element from the list. Note that in this case the value for `x` is set in the `split()`
+method, not at the task's initialisation.
+The `combine()` method, specifies that the results are grouped back when returning the
+result from the *Task*.
+
+While this example illustrates mapping and grouping of results over a single parameter,
+*Pydra* extends this to arbitrary combinations of input fields and downstream grouping
 over nested dataflows. Details of how splitters and combiners power *Pydra*'s
 scalable dataflows are described in the next section.
 

pydra/__init__.py

@@ -47,11 +47,3 @@ def check_latest_version():
 
     if TaskBase._etelemetry_version_data is None:
         TaskBase._etelemetry_version_data = check_latest_version()
-
-
-# attr run_validators is set to False, but could be changed using use_validator
-attr.set_run_validators(False)
-
-
-def set_input_validator(flag=False):
-    attr.set_run_validators(flag)

pydra/engine/audit.py

@@ -3,8 +3,10 @@
 import json
 import attr
 from ..utils.messenger import send_message, make_message, gen_uuid, now, AuditFlag
-from .helpers import ensure_list, gather_runtime_info, hash_file
-from .specs import attr_fields, File, Directory
+from ..utils.hash import hash_function
+from .helpers import ensure_list, gather_runtime_info
+from .specs import attr_fields
+from fileformats.core import FileSet
 
 try:
     import importlib_resources
@@ -181,10 +183,11 @@ def audit_task(self, task):
         command = task.cmdline if hasattr(task.inputs, "executable") else None
         attr_list = attr_fields(task.inputs)
         for attrs in attr_list:
-            if attrs.type in [File, Directory]:
-                input_name = attrs.name
-                input_path = os.path.abspath(getattr(task.inputs, input_name))
-                file_hash = hash_file(input_path)
+            input_name = attrs.name
+            value = getattr(task.inputs, input_name)
+            if isinstance(value, FileSet):
+                input_path = os.path.abspath(value)
+                file_hash = hash_function(value)
                 entity_id = f"uid:{gen_uuid()}"
                 entity_message = {
                     "@id": entity_id,