Add more TypedDict definitions

This commit adds more `TypedDict` definitions for `File`, `Dirent`, and
`Directory` objects. In addition, it adds the `exitCode` entry to the
`CWLRuntimeParameterContext`, which was missing.

Author: GlassOfWhiskey

cwl_utils/expression.py

@@ -125,7 +125,7 @@ def evaluator(
         try:
             if first_symbol in ("inputs", "self", "runtime"):
                 symbol = cast(
-                    Literal["inputs"] | Literal["self"] | Literal["runtime"],
+                    Literal["inputs", "self", "runtime"],
                     first_symbol,
                 )
                 if inspect.iscoroutinefunction(js_engine.regex_eval):

cwl_utils/sandboxjs.py

@@ -16,13 +16,13 @@
 from importlib.resources import files
 from io import BytesIO
 from pathlib import Path
-from typing import Any, Deque, cast
+from typing import Any, Deque, Literal, cast
 
 from schema_salad.utils import json_dumps
 
 from cwl_utils.errors import JavascriptException, WorkflowException
 from cwl_utils.loghandler import _logger
-from cwl_utils.types import CWLOutputType
+from cwl_utils.types import CWLOutputType, is_directory, is_file
 from cwl_utils.utils import singularity_supports_userns
 
 default_timeout = 20
@@ -549,11 +549,81 @@ def regex_eval(
 
             if isinstance(current_value, Mapping):
                 try:
-                    return self.regex_eval(
-                        parsed_string + remaining_string,
-                        remaining_string[m.end(1) :],
-                        cast(CWLOutputType, current_value[cast(str, key)]),
-                    )
+                    if is_directory(current_value):
+                        if key in ("class", "location", "path", "basename", "listing"):
+                            return self.regex_eval(
+                                parsed_string + remaining_string,
+                                remaining_string[m.end(1) :],
+                                cast(
+                                    CWLOutputType,
+                                    current_value[
+                                        cast(
+                                            Literal[
+                                                "class",
+                                                "location",
+                                                "path",
+                                                "basename",
+                                                "listing",
+                                            ],
+                                            key,
+                                        )
+                                    ],
+                                ),
+                            )
+                        else:
+                            raise WorkflowException(f"Key {key} is unexpected.")
+                    elif is_file(current_value):
+                        if key in (
+                            "class",
+                            "location",
+                            "path",
+                            "basename",
+                            "dirname",
+                            "nameroot",
+                            "nameext",
+                            "checksum",
+                            "size",
+                            "secondaryFiles",
+                            "format",
+                            "contents",
+                        ):
+                            return self.regex_eval(
+                                parsed_string + remaining_string,
+                                remaining_string[m.end(1) :],
+                                cast(
+                                    CWLOutputType,
+                                    current_value[
+                                        cast(
+                                            Literal[
+                                                "class",
+                                                "location",
+                                                "path",
+                                                "basename",
+                                                "dirname",
+                                                "nameroot",
+                                                "nameext",
+                                                "checksum",
+                                                "size",
+                                                "secondaryFiles",
+                                                "format",
+                                                "contents",
+                                            ],
+                                            key,
+                                        )
+                                    ],
+                                ),
+                            )
+                        else:
+                            raise WorkflowException(f"Key {key} is unexpected.")
+                    else:
+                        return self.regex_eval(
+                            parsed_string + remaining_string,
+                            remaining_string[m.end(1) :],
+                            cast(
+                                CWLOutputType,
+                                cast(MutableMapping[str, Any], current_value)[cast(str, key)],
+                            ),
+                        )
                 except KeyError as exc:
                     raise WorkflowException(
                         f"{parsed_string!r} doesn't have property {key!r}."

cwl_utils/types.py

@@ -1,8 +1,15 @@
 # SPDX-License-Identifier: Apache-2.0
 # From https://github.com/rabix/sbpack/blob/b8404a0859ffcbe1edae6d8f934e51847b003320/sbpack/lib.py
 """Shared Python type definitions for commons JSON like CWL objects."""
-from collections.abc import MutableMapping, MutableSequence
-from typing import TypeAlias, TypedDict
+import sys
+from collections.abc import Mapping, MutableMapping, MutableSequence
+from typing import Any, Literal, TypeAlias, TypeGuard, TypedDict
+
+if sys.version_info >= (3, 11):
+    from typing import Required
+else:
+    from typing_extensions import Required
+
 
 built_in_types: list[str] = [
     "null",
@@ -21,12 +28,46 @@
 ]
 
 
+DirectoryDict = TypedDict(
+    "DirectoryDict",
+    {
+        "class": Required[Literal["Directory"]],
+        "location": str,
+        "path": str,
+        "basename": str,
+        "listing": MutableSequence["FileDict | DirectoryDict"],
+    },
+    total=False,
+)
+
+
+FileDict = TypedDict(
+    "FileDict",
+    {
+        "class": Required[Literal["File"]],
+        "location": str,
+        "path": str,
+        "basename": str,
+        "dirname": str,
+        "nameroot": str,
+        "nameext": str,
+        "checksum": str,
+        "size": int,
+        "secondaryFiles": MutableSequence["FileDict | DirectoryDict"],
+        "format": str,
+        "contents": str,
+    },
+    total=False,
+)
+
+
 CWLOutputAtomType: TypeAlias = (
     None
     | bool
     | str
     | int
-    | float
+    | FileDict
+    | DirectoryDict
     | MutableSequence["CWLOutputAtomType"]
     | MutableMapping[str, "CWLOutputAtomType"]
 )
@@ -35,6 +76,8 @@
     | str
     | int
     | float
+    | FileDict
+    | DirectoryDict
     | MutableSequence[CWLOutputAtomType]
     | MutableMapping[str, CWLOutputAtomType]
 )
@@ -49,9 +92,30 @@ class CWLRuntimeParameterContext(TypedDict, total=False):
     ram: float | str
     outdirSize: float | str
     tmpdirSize: float | str
+    exitCode: int
 
 
 class CWLParameterContext(TypedDict, total=False):
     inputs: CWLObjectType
     self: CWLOutputType | None
     runtime: CWLRuntimeParameterContext
+
+
+class DirentDict(TypedDict, total=False):
+    entry: Required[str]
+    entryname: str
+    writable: bool
+
+
+def is_directory(value: Any) -> TypeGuard[DirectoryDict]:
+    return isinstance(value, Mapping) and value.get("class") == "Directory"
+
+
+def is_file(value: Any) -> TypeGuard[FileDict]:
+    return isinstance(value, Mapping) and value.get("class") == "File"
+
+
+def is_file_or_directory(
+    value: Any,
+) -> TypeGuard[FileDict | DirectoryDict]:
+    return isinstance(value, Mapping) and value.get("class") in ("File", "Directory")