Remove YAML Converter

src/converter/__init__.py

@@ -1,2 +1 @@
-from .linkml_converter import *
-from .yaml_converter import *
+from .converter import *

src/converter/converter.py

@@ -1,68 +1,139 @@
 import os
 
+from src.utils.helpers import merge_dict_list, load_yaml
 
-class ConverterTrait:
+
+class LinkMLConverter:
 
     def __init__(self, benchmark_file):
         self.stage_order_map = None
         self.benchmark_file = os.path.abspath(benchmark_file)
+        self.benchmark = load_yaml(benchmark_file)
 
-    def get_stage_id(self, stage):
-        raise NotImplementedError("Method not implemented yet")
-
-    def get_module_id(self, module):
-        raise NotImplementedError("Method not implemented yet")
+    def get_benchmark_name(self):
+        return self.benchmark.name if self.benchmark.name else self.benchmark.id
 
     def get_benchmark_definition(self):
-        raise NotImplementedError("Method not implemented yet")
+        return self.benchmark
 
-    def get_benchmark_definition_file(self):
-        raise NotImplementedError("Method not implemented yet")
+    def get_stage_id(self, stage):
+        return stage.id
+
+    def get_module_id(self, module):
+        return module.id
 
     def get_benchmark_stages(self):
-        raise NotImplementedError("Method not implemented yet")
+        return dict([(x.id, x) for x in self.benchmark.stages])
 
     def get_benchmark_stage(self, stage_id):
-        raise NotImplementedError("Method not implemented yet")
+        stages = self.get_benchmark_stages().values()
+        return next(stage for stage in stages if stage.id == stage_id)
 
     def get_modules_by_stage(self, stage):
-        raise NotImplementedError("Method not implemented yet")
+        return dict([(x.id, x) for x in stage.modules])
 
     def get_stage_implicit_inputs(self, stage):
-        raise NotImplementedError("Method not implemented yet")
+        if isinstance(stage, str):
+            stage = self.get_benchmark_stages()[stage]
 
-    def get_inputs_stage(self, implicit_inputs):
-        raise NotImplementedError("Method not implemented yet")
+        return [input.entries for input in stage.inputs]
 
-    def get_stage_explicit_inputs(self, stage):
-        raise NotImplementedError("Method not implemented yet")
+    def get_inputs_stage(self, implicit_inputs):
+        stages_map = {key: None for key in implicit_inputs}
+        if implicit_inputs is not None:
+            all_stages = self.get_benchmark_stages()
+            all_stages_outputs = []
+            for stage_id in all_stages:
+                outputs = self.get_stage_outputs(stage=stage_id)
+                outputs = {key: stage_id for key, value in outputs.items()}
+                all_stages_outputs.append(outputs)
+
+            all_stages_outputs = merge_dict_list(all_stages_outputs)
+            for in_deliverable in implicit_inputs:
+                # beware stage needs to be substituted
+                curr_output = all_stages_outputs[in_deliverable]
+
+                stages_map[in_deliverable] = curr_output
+
+        return stages_map
+
+    def get_stage_explicit_inputs(self, implicit_inputs):
+        explicit = {key: None for key in implicit_inputs}
+        if implicit_inputs is not None:
+            all_stages = self.get_benchmark_stages()
+            all_stages_outputs = []
+            for stage_id in all_stages:
+                outputs = self.get_stage_outputs(stage=stage_id)
+                outputs = {
+                    key: value.format(
+                        input="{input}",
+                        stage=stage_id,
+                        module="{module}",
+                        params="{params}",
+                        dataset="{dataset}",
+                    )
+                    for key, value in outputs.items()
+                }
+                all_stages_outputs.append(outputs)
+
+            all_stages_outputs = merge_dict_list(all_stages_outputs)
+            for in_deliverable in implicit_inputs:
+                # beware stage needs to be substituted
+                curr_output = all_stages_outputs[in_deliverable]
+
+                explicit[in_deliverable] = curr_output
+
+        return explicit
 
     def get_stage_outputs(self, stage):
-        raise NotImplementedError("Method not implemented yet")
+        if isinstance(stage, str):
+            stage = self.get_benchmark_stages()[stage]
+
+        return dict([(output.id, output.path) for output in stage.outputs])
 
     def get_module_excludes(self, module):
-        raise NotImplementedError("Method not implemented yet")
+        if isinstance(module, str):
+            module = self.get_benchmark_modules()[module]
+
+        return module.exclude
 
     def get_module_parameters(self, module):
-        raise NotImplementedError("Method not implemented yet")
+        params = None
+        if module.parameters is not None:
+            params = [x.values for x in module.parameters]
+
+        return params
 
     def get_module_repository(self, module):
-        raise NotImplementedError("Method not implemented yet")
+        return module.repository
 
     def is_initial(self, stage):
-        raise NotImplementedError("Method not implemented yet")
+        if stage.inputs is None or len(stage.inputs) == 0:
+            return True
+        else:
+            return False
 
     def get_after(self, stage):
-        raise NotImplementedError("Method not implemented yet")
+        return stage.after
 
     def get_stage_ids(self):
-        raise NotImplementedError("Method not implemented yet")
+        return [x.id for x in self.benchmark.stages]
 
     def get_module_ids(self):
-        raise NotImplementedError("Method not implemented yet")
+        module_ids = []
+        for stage in self.benchmark.stages:
+            for module in stage.modules:
+                module_ids.append(module.id)
+
+        return module_ids
 
     def get_output_ids(self):
-        raise NotImplementedError("Method not implemented yet")
+        output_ids = []
+        for stage in self.benchmark.stages:
+            for output in stage.outputs:
+                output_ids.append(output.id)
+
+        return output_ids
 
     def get_initial_datasets(self):
         stages = self.get_benchmark_stages()