From c9a15fcea6df7e511d199d9f539cf4e45cc61f67 Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Wed, 8 Nov 2023 16:30:24 +0100
Subject: [PATCH] convert: add encodingFormat to files when available

---
 src/runcrate/convert.py             | 13 +++++++++----
 tests/test_cwlprov_crate_builder.py |  1 +
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/src/runcrate/convert.py b/src/runcrate/convert.py
index 6957b56..ff48935 100644
--- a/src/runcrate/convert.py
+++ b/src/runcrate/convert.py
@@ -742,9 +742,9 @@ def patch_workflow_input_collection(self, crate, wf=None):
             if "ComputationalWorkflow" in as_list(tool.type):
                 self.patch_workflow_input_collection(crate, wf=tool)
 
-    def _map_input_data(self, data):
+    def _map_input_data(self, crate, data):
         if isinstance(data, list):
-            return [self._map_input_data(_) for _ in data]
+            return [self._map_input_data(crate, _) for _ in data]
         if isinstance(data, dict):
             rval = {}
             for k, v in data.items():
@@ -756,8 +756,13 @@ def _map_input_data(self, data):
                         source_k = str(source)
                     dest = self.file_map.get(source_k)
                     rval[k] = str(dest) if dest else v
+                    fmt = data.get("format")
+                    if fmt:
+                        entity = crate.get(str(dest))
+                        if entity:
+                            entity["encodingFormat"] = fmt
                 else:
-                    rval[k] = self._map_input_data(v)
+                    rval[k] = self._map_input_data(crate, v)
             return rval
         return data
 
@@ -766,7 +771,7 @@ def add_inputs_file(self, crate):
         if path.is_file():
             with open(path) as f:
                 data = json.load(f)
-            data = self._map_input_data(data)
+            data = self._map_input_data(crate, data)
             source = StringIO(json.dumps(data, indent=4))
             crate.add_file(source, path.name, properties={
                 "name": "input object document",
diff --git a/tests/test_cwlprov_crate_builder.py b/tests/test_cwlprov_crate_builder.py
index 2ec6f9c..0f6404d 100644
--- a/tests/test_cwlprov_crate_builder.py
+++ b/tests/test_cwlprov_crate_builder.py
@@ -96,6 +96,7 @@ def test_revsort(data_dir, tmpdir):
             assert entity["alternateName"] == "whale.txt"
             assert entity["sha1"] == entity.id.rsplit("/")[-1]
             assert entity["contentSize"] == "1111"
+            assert "encodingFormat" in entity
             wf_input_file = entity
     wf_output_file = wf_results[0]
     assert wf_output_file["alternateName"] == "output.txt"