adjust process builtin id retrieval + tmp fix pywps replacing file/da…

…ta value cause by auto-fetch (geopython/pywps#526, #91)

weaver/processes/builtin/__init__.py

@@ -27,7 +27,7 @@
 if TYPE_CHECKING:
     from weaver.typedefs import AnyDatabaseContainer, CWL   # noqa: F401
     from cwltool.context import RuntimeContext              # noqa: F401
-    from typing import AnyStr, Dict, Type, Union            # noqa: F401
+    from typing import Any, AnyStr, Dict, Type, Union       # noqa: F401
 
 LOGGER = logging.getLogger(__name__)
 
@@ -136,7 +136,10 @@ def register_builtin_processes(container):
 
 class BuiltinProcessJobBase(CommandLineJob):
     def __init__(self, builder, joborder, make_path_mapper, requirements, hints, name):
-        self.process = [h.get("process") for h in hints][0]
+        process_hints = [h for h in hints if "process" in h]
+        if not process_hints or len(process_hints) != 1:
+            raise PackageNotFound("Could not extract referenced process in job.")
+        self.process = process_hints[0]["process"]
         super(BuiltinProcessJobBase, self).__init__(builder, joborder, make_path_mapper, requirements, hints, name)
 
     def _validate_process(self):
@@ -149,11 +152,11 @@ def _validate_process(self):
             raise PackageExecutionError("Invalid package is not of type '{}'".format(PROCESS_BUILTIN))
 
     # pylint: disable=W0221,arguments-differ    # naming using python like arguments
-    def run(self, runtime_context):
-        # type: (RuntimeContext) -> None
+    def run(self, runtime_context, **kwargs):
+        # type: (RuntimeContext, Any) -> None
         try:
             self._validate_process()
-            super(BuiltinProcessJobBase, self).run(runtime_context)
+            super(BuiltinProcessJobBase, self).run(runtime_context, **kwargs)
         except Exception as err:
             LOGGER.warning(u"Failed to run process:\n%s", err, exc_info=runtime_context.debug)
             self.output_callback({}, "permanentFail")

weaver/processes/wps_package.py

@@ -647,12 +647,12 @@ def _cwl2wps_io(io_info, io_select):
         if "format" in io_info:
             io_formats = [io_info["format"]] if isinstance(io_info["format"], six.string_types) else io_info["format"]
             kw["supported_formats"] = [Format(clean_mime_type_format(str(fmt))) for fmt in io_formats]
-            kw["mode"] = MODE.SIMPLE
+            kw["mode"] = MODE.SIMPLE  # only validate the extension (not file contents)
         else:
             # we need to minimally add 1 format, otherwise empty list is evaluated as None by pywps
             # when "supported_formats" is None, the process's json property raises because of it cannot iterate formats
             kw["supported_formats"] = [DefaultFormat]
-            kw["mode"] = MODE.NONE
+            kw["mode"] = MODE.NONE  # don't validate anything as default is only raw text
         if is_output:
             if io_type == "Directory":
                 kw["as_reference"] = True
@@ -1821,6 +1821,13 @@ def _handler(self, request, response):
                             cwl_inputs[input_id] = [self.make_location_input(data, input_type, input_def)
                                                     for data, input_def in zip(input_data, input_occurs)]
                         else:
+                            # FIXME: we don't want auto fetch because we pass down value, but this is PyWPS bug
+                            #   (https://github.com/geopython/pywps/issues/526)
+                            #   (https://github.com/crim-ca/weaver/issues/91)
+                            #   href are sometime already handled (pulled and staged locally), use it directly
+                            #   validate using the internal '_file' instead of 'file' otherwise we trigger the fetch
+                            if input_i.file and os.path.isfile(input_i._file):  # noqa:W0212
+                                input_data = input_i.file
                             cwl_inputs[input_id] = self.make_location_input(input_data, input_type, input_i)
                     elif isinstance(input_i, (LiteralInput, BoundingBoxInput)):
                         cwl_inputs[input_id] = input_data