Merge pull request PolusAI#207 from jfennick/source_means_var_not_val

jfennick · web-flow · commit e6f59d6d5b62 · 2024-04-09T12:00:53.000-10:00
Source means var not val
diff --git a/src/wic/api/pythonapi.py b/src/wic/api/pythonapi.py
@@ -201,9 +201,9 @@ def set_input_Step_Workflow(process_self: Any, __name: str, __value: Any) -> Any
                     # (Very useful for regression testing!)
                     # NOTE: process_name is either clt name or workflow name
                     tmp = __value.value if __value.value else f"{__name}{process_self.process_name}"
-                    alias_dict = {'wic_alias': {'key': tmp}}
+                    alias_dict = {'wic_alias': tmp}
                     local_input._set_value(alias_dict, linked=True)
-                    anchor_dict = {'wic_anchor': {'key': tmp}}
+                    anchor_dict = {'wic_anchor': tmp}
                     __value._set_value(anchor_dict, linked=True)
             except BaseException as exc:
                 raise exc
@@ -223,7 +223,7 @@ def set_input_Step_Workflow(process_self: Any, __name: str, __value: Any) -> Any
                     __value._set_value(f"{tmp}", linked=True)
                 else:
                     anchor_dict = __value.value
-                    alias_dict = {'wic_alias': {'key': anchor_dict['wic_anchor']['key']}}
+                    alias_dict = {'wic_alias': anchor_dict['wic_anchor']}
                     local_input._set_value(alias_dict, linked=True)
             except BaseException as exc:
                 raise exc
@@ -237,7 +237,7 @@ def set_input_Step_Workflow(process_self: Any, __name: str, __value: Any) -> Any
                 f"got {__value.__class__.__name__}, "
                 f"expected {obj.inp_type.__name__}"
             )
-        ii_dict = {'wic_inline_input': {'key': __value}}
+        ii_dict = {'wic_inline_input': __value}
         process_self.inputs[index]._set_value(ii_dict)
 
 
@@ -382,12 +382,12 @@ def _yml(self) -> dict:
                 if isinstance(inp.value, Path):
                     # Special case for Path since it does not inherit from YAMLObject
                     in_dict[inp.name] = str(inp.value)
-                elif isinstance(inp.value, dict) and isinstance(inp.value.get('wic_alias', {}).get('key', {}), Path):
+                elif isinstance(inp.value, dict) and isinstance(inp.value.get('wic_alias', {}), Path):
                     # Special case for Path since it does not inherit from YAMLObject
-                    in_dict[inp.name] = {'wic_alias': {'key': str(inp.value['wic_alias']['key'])}}
-                elif isinstance(inp.value, dict) and isinstance(inp.value.get('wic_inline_input', {}).get('key', {}), Path):
+                    in_dict[inp.name] = {'wic_alias': str(inp.value['wic_alias'])}
+                elif isinstance(inp.value, dict) and isinstance(inp.value.get('wic_inline_input', {}), Path):
                     # Special case for Path since it does not inherit from YAMLObject
-                    in_dict[inp.name] = {'wic_inline_input': {'key': str(inp.value['wic_inline_input']['key'])}}
+                    in_dict[inp.name] = {'wic_inline_input': str(inp.value['wic_inline_input'])}
                 elif isinstance(inp.value, str):
                     in_dict[inp.name] = inp.value  # Obviously strings are serializable
                 elif isinstance(inp.value, yaml.YAMLObject):
diff --git a/src/wic/ast.py b/src/wic/ast.py
@@ -287,12 +287,12 @@ def python_script_generate_cwl(yaml_tree_tuple: YamlTree,
                 yml_args = copy.deepcopy(steps[i][step_key]['in'])
                 python_script_path = yml_args.get('script', '')
                 if isinstance(python_script_path, dict) and 'wic_inline_input' in python_script_path:
-                    python_script_path = python_script_path['wic_inline_input']['key']
+                    python_script_path = python_script_path['wic_inline_input']
                 # NOTE: The existence of the script: tag should now be guaranteed in the schema
                 del yml_args['script']
                 python_script_docker_pull = yml_args.get('dockerPull', '')  # Optional
                 if isinstance(python_script_docker_pull, dict) and 'wic_inline_input' in python_script_docker_pull:
-                    python_script_docker_pull = python_script_docker_pull['wic_inline_input']['key']
+                    python_script_docker_pull = python_script_docker_pull['wic_inline_input']
                 if 'dockerPull' in yml_args:
                     del yml_args['dockerPull']
                     del steps[i][step_key]['in']['dockerPull']
diff --git a/src/wic/compiler.py b/src/wic/compiler.py
@@ -451,8 +451,7 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
                     out_key = keys[0]
                     out_val = out_val[out_key]
                     if isinstance(out_val, Dict) and 'wic_anchor' in out_val:
-                        out_val = out_val['wic_anchor']
-                        edgedef = out_val['key']
+                        edgedef = out_val['wic_anchor']
 
                         # NOTE: There can only be one definition, but multiple call sites.
                         if not explicit_edge_defs_copy.get(edgedef):
@@ -469,13 +468,12 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
             # Extract input value into separate yml file
             # Replace it here with a new variable name
             arg_val = steps[i][step_key]['in'][arg_key]
+
             # Convert native YAML to a JSON-encoded string for specific tags.
             tags = ['config']
             if arg_key in tags and isinstance(arg_val, Dict) and ('wic_inline_input' in arg_val):
-                # Do NOT wrap config: in {'source': ...}
-                arg_val = {'wic_inline_input': {'key': json.dumps(arg_val['wic_inline_input']['key'])}}
-            elif isinstance(arg_val, str):
-                arg_val = {'source': arg_val}
+                arg_val = {'wic_inline_input': json.dumps(arg_val['wic_inline_input'])}
+
             # Use triple underscore for namespacing so we can split later
             in_name = f'{step_name_i}___{arg_key}'  # {step_name_i}_input___{arg_key}
 
@@ -495,21 +493,21 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
                 # NOTE: Exclude cwl_watcher from explicit edge dereferences.
                 # Since cwl_watcher requires explicit filenames for globbing,
                 # we do not want to replace them with internal CWL dependencies!
-                if not explicit_edge_defs_copy.get(arg_val['key']):
+                if not explicit_edge_defs_copy.get(arg_val):
                     if is_root and not testing:
                         # Even if is_root, we don't want to raise an Exception
                         # here because in test_cwl_embedding_independence, we
                         # recompile all subworkflows as if they were root. That
                         # will cause this code path to be taken but it is not
                         # actually an error. Add a CWL input for testing only.
-                        raise Exception(f"Error! No definition found for &{arg_val['key']}!")
+                        raise Exception(f"Error! No definition found for &{arg_val}!")
                     inputs_workflow.update({in_name: in_dict})
                     steps[i][step_key]['in'][arg_key] = {'source': in_name}
                     # Add a 'dummy' value to explicit_edge_calls anyway, because
                     # that determines sub_args_provided when the recursion returns.
                     explicit_edge_calls_copy.update({in_name: (namespaces + [step_name_i], arg_key)})
                 else:
-                    (nss_def_init, var) = explicit_edge_defs_copy[arg_val['key']]
+                    (nss_def_init, var) = explicit_edge_defs_copy[arg_val]
 
                     nss_def_embedded = var.split('___')[:-1]
                     nss_call_embedded = arg_key.split('___')[:-1]
@@ -546,7 +544,7 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
                     elif len(nss_call_tails) > 1:
                         inputs_workflow.update({in_name: in_dict})
                         # Store explicit edge call site info up through the recursion.
-                        d = {in_name: explicit_edge_defs_copy[arg_val['key']]}
+                        d = {in_name: explicit_edge_defs_copy[arg_val]}
                         # d = {in_name, (namespaces + [step_name_i], var)} # ???
                         explicit_edge_calls_copy.update(d)
                         steps[i][step_key]['in'][arg_key] = {'source': in_name}
@@ -601,7 +599,7 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
 
                         utils_graphs.add_graph_edge(args, graph_init, nss_def, nss_call, label, color='blue')
             elif isinstance(arg_val, Dict) and 'wic_inline_input' in arg_val:
-                arg_val = arg_val['wic_inline_input']['key']
+                arg_val = arg_val['wic_inline_input']
 
                 if arg_key in steps[i][step_key].get('scatter', []):
                     # Promote scattered input types to arrays
@@ -624,6 +622,7 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
                     graphdata.nodes.append((input_node_name, attrs))
                     graphdata.edges.append((input_node_name, step_node_name, {}))
             else:
+                arg_var: str = arg_val
                 # Leave un-evaluated, i.e. allow the user to inject raw CWL.
                 # The un-evaluated string should refer to either an inputs: variable
                 # or an internal CWL dependency, i.e. an output from a previous step.
@@ -638,20 +637,20 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
                 # (yet) be inlined. Somehow, if they are not marked with
                 # inlineable: False, test_inline_subworkflows can still pass.
                 # This Exception will (correctly) cause such inlineing tests to fail.
-                if arg_val['source'] not in yaml_tree.get('inputs', {}):
+                if arg_var not in yaml_tree.get('inputs', {}):
                     if not args.allow_raw_cwl:
-                        print(f"Warning! Did you forget to use !ii before {arg_val['source']} in {yaml_stem}.yml?")
+                        print(f"Warning! Did you forget to use !ii before {arg_var} in {yaml_stem}.yml?")
                         print('If you want to compile the workflow anyway, use --allow_raw_cwl')
                         sys.exit(1)
 
                     inputs = yaml_tree.get('inputs', {})
                     unbound_lit_var = 'Error! Unbound literal variable'
                     if inputs == {}:
-                        raise Exception(f"{unbound_lit_var}{arg_val['source']} not in inputs: tag in {yaml_stem}.yml")
+                        raise Exception(f"{unbound_lit_var}{arg_var} not in inputs: tag in {yaml_stem}.yml")
                     inputs_dump = yaml.dump({'inputs': inputs})
-                    raise Exception(f"{unbound_lit_var}{arg_val['source']} not in\n{inputs_dump}\nin {yaml_stem}.yml")
+                    raise Exception(f"{unbound_lit_var}{arg_var} not in\n{inputs_dump}\nin {yaml_stem}.yml")
 
-                inputs_key_dict = yaml_tree['inputs'][arg_val['source']]
+                inputs_key_dict = yaml_tree['inputs'][arg_var]
                 if 'doc' in inputs_key_dict:
                     inputs_key_dict['doc'] += '\\n' + in_dict.get('doc', '')
                 else:
@@ -661,12 +660,12 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
                 else:
                     inputs_key_dict['label'] = in_dict.get('label', '')
 
-                if arg_val['source'] in input_mapping_copy:
-                    input_mapping_copy[arg_val['source']].append(in_name)
+                if arg_var in input_mapping_copy:
+                    input_mapping_copy[arg_var].append(in_name)
                 else:
-                    input_mapping_copy[arg_val['source']] = [in_name]
+                    input_mapping_copy[arg_var] = [in_name]
                 # TODO: We can use un-evaluated variable names for input mapping; no notation for output mapping!
-                steps[i][step_key]['in'][arg_key] = arg_val  # Leave un-evaluated
+                steps[i][step_key]['in'][arg_key] = {'source': arg_var}  # Leave un-evaluated
 
         for arg_key in args_required:
             # print('arg_key', arg_key)
@@ -854,8 +853,7 @@ def compile_workflow_once(yaml_tree_ast: YamlTree,
         else:
             # We cannot store string values as a dict, so use type: ignore
             arg_val = in_dict['value']
-            new_val = arg_val['source'] if isinstance(arg_val, Dict) and 'source' in arg_val else arg_val
-            new_keyval = {key: new_val}
+            new_keyval = {key: arg_val}
         # else:
         #    raise Exception(f"Error! Unknown type: {in_dict['type']}")
         yaml_inputs.update(new_keyval)
diff --git a/src/wic/inference.py b/src/wic/inference.py
@@ -219,7 +219,6 @@ def perform_edge_inference(args: argparse.Namespace,
             else:
                 vars_workflow_output_internal.append(f'{step_name_j}/{out_key}')
 
-            # arg_val = {'source': f'{step_name_j}/{out_key}'}
             arg_val = f'{step_name_j}/{out_key}'
             arg_keyval = {arg_key: arg_val}
             steps_i = utils_cwl.add_yamldict_keyval_in(steps[i], step_key, arg_keyval)
@@ -332,7 +331,6 @@ def perform_edge_inference(args: argparse.Namespace,
         # which should match in the parent workflow.
         inputs_workflow.update({in_name: in_dict})
 
-        # arg_keyval = {arg_key: {'source': in_name}}
         arg_keyval = {arg_key: in_name}
         steps_i = utils_cwl.add_yamldict_keyval_in(steps[i], step_key, arg_keyval)
         return steps_i
diff --git a/src/wic/input_output.py b/src/wic/input_output.py
@@ -68,20 +68,6 @@ def write_to_disk(rose_tree: RoseTree, path: Path, relative_run_path: bool) -> N
     cwl_tree = node_data.compiled_cwl
     yaml_inputs = node_data.workflow_inputs_file
 
-    # NOTE: As part of the scatter feature we introduced the use of 'source',
-    # but in some cases (biobb 'config' tag) it is not being removed correctly
-    # in the compiler, so as a last resort remove it here.
-    yaml_inputs_no_source = {}
-    for key, val in yaml_inputs.items():
-        try:
-            if isinstance(val, str):
-                val_dict = json.loads(val)
-                if 'source' in val_dict:
-                    val = val_dict['source']
-        except Exception as e:
-            pass
-        yaml_inputs_no_source[key] = val
-
     path.mkdir(parents=True, exist_ok=True)
     if relative_run_path:
         filename_cwl = f'{yaml_stem}.cwl'
@@ -98,7 +84,7 @@ def write_to_disk(rose_tree: RoseTree, path: Path, relative_run_path: bool) -> N
         w.write(auto_gen_header)
         w.write(''.join(yaml_content))
 
-    yaml_content = yaml.dump(yaml_inputs_no_source, sort_keys=False, line_break='\n', indent=2, Dumper=NoAliasDumper)
+    yaml_content = yaml.dump(yaml_inputs, sort_keys=False, line_break='\n', indent=2, Dumper=NoAliasDumper)
     with open(path / filename_yml, mode='w', encoding='utf-8') as inp:
         inp.write(auto_gen_header)
         inp.write(yaml_content)
@@ -220,10 +206,10 @@ def write_absolute_yaml_tags(args: argparse.Namespace, in_dict_in: Yaml, namespa
     # we don't want users' home directories in the yml files.
     cachedir_path = Path(args.cachedir).absolute()
     # print('setting cachedir_path to', cachedir_path)
-    in_dict_in['root_workflow_yml_path'] = {'wic_inline_input': {'key': str(Path(args.yaml).parent.absolute())}}
+    in_dict_in['root_workflow_yml_path'] = {'wic_inline_input': str(Path(args.yaml).parent.absolute())}
 
-    in_dict_in['cachedir_path'] = {'wic_inline_input': {'key': str(cachedir_path)}}
-    in_dict_in['homedir'] = {'wic_inline_input': {'key': args.homedir}}
+    in_dict_in['cachedir_path'] = {'wic_inline_input': str(cachedir_path)}
+    in_dict_in['homedir'] = {'wic_inline_input': args.homedir}
 
     # Add a 'dummy' values to explicit_edge_calls, because
     # that determines sub_args_provided when the recursion returns.
diff --git a/src/wic/schemas/wic_schema.py b/src/wic/schemas/wic_schema.py
@@ -173,17 +173,11 @@ def cwl_schema(name: str, cwl: Json, id_prefix: str) -> Json:
     anytype: Dict[Any, Any] = {}
 
     # See utils_yaml.py
-    aliasprops = default_schema()
-    aliasprops['properties'] = {'key': str_nonempty}
-    aliasprops['required'] = ['key']
     alias = default_schema()
-    alias['properties'] = {'wic_alias': aliasprops}  # !*
+    alias['properties'] = {'wic_alias': str_nonempty}  # !*
 
-    iiprops = default_schema()
-    iiprops['properties'] = {'key': anytype}
-    iiprops['required'] = ['key']
     ii = default_schema()
-    ii['properties'] = {'wic_inline_input': iiprops}  # !ii
+    ii['properties'] = {'wic_inline_input': anytype}  # !ii
 
     # required = []
     for key, val in cwl['inputs'].items():
@@ -211,11 +205,8 @@ def cwl_schema(name: str, cwl: Json, id_prefix: str) -> Json:
         if key == 'config' and name == 'config_tag_mdp':
             grompp = config_schemas.get('grompp', {})
 
-            iiprops_mdp = default_schema()
-            iiprops_mdp['properties'] = {'key': grompp}
-            iiprops_mdp['required'] = ['key']
             ii_mdp = default_schema()
-            ii_mdp['properties'] = {'wic_inline_input': iiprops_mdp}  # !ii
+            ii_mdp['properties'] = {'wic_inline_input': grompp}  # !ii
 
             inputs_props[key] = ii_mdp
             continue
@@ -272,10 +263,8 @@ def cwl_schema(name: str, cwl: Json, id_prefix: str) -> Json:
     outputs['properties'] = outputs_props
 
     # See utils_yaml.py
-    anchorprops = default_schema()
-    anchorprops['properties'] = {'key': str_nonempty}
     anchor = default_schema()
-    anchor['properties'] = {'wic_anchor': anchorprops}  # !&
+    anchor['properties'] = {'wic_anchor': str_nonempty}  # !&
 
     keys_anchors: Json = {}
     for key in cwl['outputs'].keys():
@@ -461,10 +450,8 @@ def wic_main_schema(tools_cwl: Tools, yml_stems: List[str], schema_store: Dict[s
         in_schema['properties'] = {'script': str_nonempty}
 
         # See utils_yaml.py
-        anchorprops = default_schema()
-        anchorprops['properties'] = {'key': str_nonempty}
         anchor = default_schema()
-        anchor['properties'] = {'wic_anchor': anchorprops}  # !&
+        anchor['properties'] = {'wic_anchor': str_nonempty}  # !&
 
         # NOTE: We do not know the specific keys statically, so we have to use str_nonempty
         out_schema: Json = {'type': 'array', 'items': {'anyOf': [str_nonempty, anchor]}}
diff --git a/src/wic/utils_yaml.py b/src/wic/utils_yaml.py
@@ -8,16 +8,16 @@
 # because then these constructors will fire again.
 
 
-def anchor_constructor(loader: yaml.SafeLoader, node: yaml.nodes.ScalarNode) -> Dict[str, Dict[str, Any]]:
-    key = loader.construct_scalar(node)
+def anchor_constructor(loader: yaml.SafeLoader, node: yaml.nodes.ScalarNode) -> Dict[str, Any]:
+    val = loader.construct_scalar(node)
     name = 'wic_anchor'  # NOT '!&'
-    return {name: {'key': key}}
+    return {name: val}
 
 
-def alias_constructor(loader: yaml.SafeLoader, node: yaml.nodes.ScalarNode) -> Dict[str, Dict[str, Any]]:
-    key = loader.construct_scalar(node)
+def alias_constructor(loader: yaml.SafeLoader, node: yaml.nodes.ScalarNode) -> Dict[str, Any]:
+    val = loader.construct_scalar(node)
     name = 'wic_alias'  # NOT '!*'
-    return {name: {'key': key}}
+    return {name: val}
 
 
 def inlineinput_constructor(loader: yaml.SafeLoader, node: yaml.nodes.Node) -> Dict[str, Dict[str, Any]]:
@@ -37,7 +37,7 @@ def inlineinput_constructor(loader: yaml.SafeLoader, node: yaml.nodes.Node) -> D
     else:
         raise Exception(f'Unknown yaml node type! {node}')
     name = 'wic_inline_input'  # NOT '!ii'
-    return {name: {'key': val}}
+    return {name: val}
 
 
 def wic_loader() -> Type[yaml.SafeLoader]:
