Migrate format checking routines from cwltool (#141)

* GA: fix pydocstyle test

Co-authored-by: Michael R. Crusoe <[email protected]>

Author: GlassOfWhiskey

MANIFEST.in

@@ -9,6 +9,8 @@ include cwl_utils/cwlNodeEngineWithContext.js
 include testdata/*.cwl
 include testdata/*.yaml
 include testdata/*.input
+include testdata/*.ttl
+include testdata/*.owl
 include cwl_utils/py.typed
 global-exclude *~
 global-exclude *.pyc

cwl_utils/file_formats.py

@@ -0,0 +1,71 @@
+"""
+CWL file formats utilities.
+
+For more information, please visit https://www.commonwl.org/user_guide/16-file-formats/
+"""
+
+from typing import List, Optional, Set, Union
+
+from rdflib import OWL, RDFS, Graph, URIRef
+from schema_salad.exceptions import ValidationException
+from schema_salad.utils import aslist, json_dumps
+
+from cwl_utils.types import CWLObjectType
+
+
+def formatSubclassOf(
+    fmt: str, cls: str, ontology: Optional[Graph], visited: Set[str]
+) -> bool:
+    """Determine if `fmt` is a subclass of `cls`."""
+    if URIRef(fmt) == URIRef(cls):
+        return True
+
+    if ontology is None:
+        return False
+
+    if fmt in visited:
+        return False
+
+    visited.add(fmt)
+
+    uriRefFmt = URIRef(fmt)
+
+    for _s, _p, o in ontology.triples((uriRefFmt, RDFS.subClassOf, None)):
+        # Find parent classes of `fmt` and search upward
+        if formatSubclassOf(o, cls, ontology, visited):
+            return True
+
+    for _s, _p, o in ontology.triples((uriRefFmt, OWL.equivalentClass, None)):
+        # Find equivalent classes of `fmt` and search horizontally
+        if formatSubclassOf(o, cls, ontology, visited):
+            return True
+
+    for s, _p, _o in ontology.triples((None, OWL.equivalentClass, uriRefFmt)):
+        # Find equivalent classes of `fmt` and search horizontally
+        if formatSubclassOf(s, cls, ontology, visited):
+            return True
+
+    return False
+
+
+def check_format(
+    actual_file: Union[CWLObjectType, List[CWLObjectType]],
+    input_formats: Union[List[str], str],
+    ontology: Optional[Graph],
+) -> None:
+    """Confirm that the format present is valid for the allowed formats."""
+    for afile in aslist(actual_file):
+        if not afile:
+            continue
+        if "format" not in afile:
+            raise ValidationException(
+                f"File has no 'format' defined: {json_dumps(afile, indent=4)}"
+            )
+        for inpf in aslist(input_formats):
+            if afile["format"] == inpf or formatSubclassOf(
+                afile["format"], inpf, ontology, set()
+            ):
+                return
+        raise ValidationException(
+            f"File has an incompatible format: {json_dumps(afile, indent=4)}"
+        )

cwl_utils/graph_split.py

@@ -14,9 +14,9 @@
 from typing import IO, Any, MutableMapping, Set, Union, cast
 
 from cwlformat.formatter import stringify_dict
+from ruamel.yaml.dumper import RoundTripDumper
 from ruamel.yaml.main import YAML, dump
 from ruamel.yaml.representer import RoundTripRepresenter
-from ruamel.yaml.dumper import RoundTripDumper
 from schema_salad.sourceline import SourceLine, add_lc_filename
 
 

cwl_utils/utils.py

@@ -6,10 +6,6 @@
 import urllib.parse
 import urllib.request
 from copy import deepcopy
-
-from ruamel.yaml.main import YAML
-from ruamel.yaml.parser import ParserError
-from ruamel.yaml.scanner import ScannerError
 from typing import (
     Any,
     Dict,
@@ -21,6 +17,10 @@
     Union,
 )
 
+from ruamel.yaml.main import YAML
+from ruamel.yaml.parser import ParserError
+from ruamel.yaml.scanner import ScannerError
+
 from cwl_utils.errors import MissingKeyField
 from cwl_utils.loghandler import _logger
 

requirements.txt

@@ -1,4 +1,6 @@
 cwl-upgrader >= 1.2.3
 packaging
+rdflib
 requests
+CacheControl
 schema-salad >= 8.2, < 9