Merge pull request #796 from crim-ca/patch-default-literal

CHANGES.rst

@@ -12,7 +12,10 @@ Changes
 
 Changes:
 --------
-- No change.
+- Add resilient handling of `I/O` literal ``default`` values when parsing remote `OGC API - Processes` descriptions.
+  Due to varying definitions from the standard revisions, some implementations could indicate a single literal default
+  value as an array representation (e.g.: ``default: [1.23]``), leading to parsing "errors" in `Weaver` that expects a
+  strict match between the ``default`` value and its ``type``.
 
 Fixes:
 ------

setup.cfg

@@ -118,6 +118,7 @@ exclude =
 	env,
 	parts,
 	examples,
+	reports,
 	node_modules,
 
 [doc8]

tests/processes/test_convert.py

@@ -1,6 +1,9 @@
 """
 Unit tests of functions within :mod:`weaver.processes.convert`.
 """
+import inspect
+import logging
+
 # pylint: disable=R1729  # ignore non-generator representation employed for displaying test log results
 
 import json
@@ -1006,6 +1009,89 @@ def test_cwl2wps_io_raise_mixed_types(test_type):
         cwl2wps_io(io_info, IO_INPUT)
 
 
+@pytest.mark.parametrize(
+    ["io_default", "io_expect", "is_erroneous"],
+    [
+        (
+            {"name": "data", "type": "string", "default": ["test"]},
+            "test",
+            True,
+        ),
+        (
+            {"name": "data", "type": "integer", "default": [1]},
+            1,
+            True,
+        ),
+        (
+            {"name": "data", "type": "float", "default": [3.1416]},
+            3.1416,
+            True,
+        ),
+        (
+            {"name": "data", "type": "boolean", "default": [True]},
+            True,
+            True,
+        ),
+        (
+            # empty considered as "null default"
+            {"name": "data", "type": "integer", "default": []},
+            None,
+            True,
+        ),
+        (
+            # explicitly "null default"
+            {"name": "data", "type": "integer", "default": [None]},
+            None,
+            True,
+        ),
+        (
+            # only 1 array element allowed for silent handling
+            {"name": "data", "type": "integer", "default": [1, 2, 3]},
+            PackageTypeError,
+            True,
+        ),
+        (
+            # don't allow nested lists, even if literal inside
+            # assume it is too much "lists" involved, and the intention is most probably wrong at this point
+            {"name": "data", "type": "integer", "default": [[1]]},
+            PackageTypeError,
+            True,
+        ),
+        (
+            # don't allow non-literals within list, even if single element
+            {"name": "data", "type": "integer", "default": [{}]},
+            PackageTypeError,
+            True,
+        ),
+        (
+            # sanity check valid case
+            {"name": "data", "type": "string", "default": "test"},
+            "test",
+            False,
+        ),
+        (
+            # sanity check valid case
+            {"name": "data", "type": "float", "default": 3.1416},
+            3.1416,
+            False,
+        )
+    ]
+)
+def test_cwl2wps_io_literal_handle_erroneous_defaults(io_default, io_expect, is_erroneous, caplog):
+    """
+    Given a remote :term:`OAP` process description with an erroneous ``default``, ensure parsing can mitigate silently.
+    """
+
+    if inspect.isclass(io_expect) and issubclass(io_expect, Exception):
+        with pytest.raises(io_expect):
+            cwl2wps_io(io_default, IO_INPUT)
+    else:
+        with caplog.at_level(logging.WARNING):
+            result = cwl2wps_io(io_default, IO_INPUT)
+            assert result.data == io_expect
+        assert not is_erroneous or (is_erroneous and "detected badly formed 'default'" in caplog.text.lower())
+
+
 def test_cwl2wps_io_record_format():
     """
     Validate handling of alternative representation by CWL I/O record after parsing contents into a tool instance.

weaver/processes/convert.py

@@ -1774,7 +1774,25 @@ def cwl2wps_io(io_info, io_select):
             # anything later on when CWL/WPS merging is attempted
             if io_def.symbols is not AnyValue:
                 kw["allowed_values"] = io_def.symbols
-            kw["default"] = io_info.get("default", None)
+            io_default = io_info.get("default", None)
+            # resilience against badly formed process descriptions
+            # (in case the reference is a remote process we have no control over)
+            if isinstance(io_default, list):
+                LOGGER.warning(
+                    "Detected badly formed 'default' as array for literal input [%s]. "
+                    "Attempting patch of the definition to make it valid.",
+                    io_def.name,
+                )
+                if len(io_default) == 0:
+                    io_default = None
+                elif len(io_default) == 1:
+                    io_default = io_default[0]
+            if not isinstance(io_default, (str, float, int, bool, type(None))):
+                raise PackageTypeError(
+                    f"Unsupported I/O info definition: '{io_info!r}' with '{io_select}' "
+                    f"specifies a 'default' that does not correspond to a valid literal value."
+                )
+            kw["default"] = io_default
             kw["min_occurs"] = io_def.min_occurs
             kw["max_occurs"] = io_def.max_occurs
         return io_literal(**kw)