From ea2d487abb7ddbce24bd09640ac370d8cd2ec9cb Mon Sep 17 00:00:00 2001
From: Nazim Azeli <nazim.azeli@crim.ca>
Date: Tue, 10 Sep 2024 14:11:09 -0400
Subject: [PATCH] Fixing typedef schema, adding test for softwareVersion  and
 adding content type parsing for codeRepository

---
 tests/processes/test_wps_package.py | 32 +++++++++++++++++++++++------
 weaver/processes/wps_package.py     | 21 +++++++++++++++----
 weaver/typedefs.py                  |  3 ++-
 3 files changed, 45 insertions(+), 11 deletions(-)

diff --git a/tests/processes/test_wps_package.py b/tests/processes/test_wps_package.py
index d09d5b665..879bc4c9e 100644
--- a/tests/processes/test_wps_package.py
+++ b/tests/processes/test_wps_package.py
@@ -678,11 +678,13 @@ def test_format_extension_validator_basic(data_input, mode, expect):
                 "s:author": [
                     {"class": "s:Person", "s:name": "John Doe", "s:affiliation": "Example Inc."}
                 ],
-
             },
             "wps_package_metadata": {}
         },
         {
+            "abstract": "",
+            "title": "",
+            "version": "",
             "metadata": [
                 {
                     "role": "author",
@@ -704,8 +706,12 @@ def test_format_extension_validator_basic(data_input, mode, expect):
             "wps_package_metadata": {}
         },
         {
+            "abstract": "",
+            "title": "",
+            "version": "",
             "metadata": [
                 {
+                    "type": "text/html",
                     "rel": "codeRepository",
                     "href": "https://gitlab.com/"
                 }
@@ -717,11 +723,11 @@ def test_format_extension_validator_basic(data_input, mode, expect):
         {
             "cwl_package_package": {
                 "s:version": "1.0"
-
             },
             "wps_package_metadata": {
                 "metadata": [
                     {
+                        "type": "text/html",
                         "rel": "codeRepository",
                         "href": "https://gitlab.com/"
                     }
@@ -729,16 +735,32 @@ def test_format_extension_validator_basic(data_input, mode, expect):
             }
         },
         {
+            "abstract": "",
+            "title": "",
             "metadata": [
                 {
+                    "type": "text/html",
                     "rel": "codeRepository",
                     "href": "https://gitlab.com/"
                 },
-
             ],
             "version": "1.0"
         }
     ),
+    (
+        # Test softwareVersion
+        {
+            "cwl_package_package": {
+                "s:softwareVersion": "1.0.0"
+            },
+            "wps_package_metadata": {}
+        },
+        {
+            "abstract": "",
+            "title": "",
+            "version": "1.0.0"
+        }
+    ),
 ])
 def test_process_metadata(original, expected):
     # type: (CWL, CWL) -> None
@@ -746,6 +768,4 @@ def test_process_metadata(original, expected):
     wps_package_metadata = original["wps_package_metadata"]
     _update_package_metadata(wps_package_metadata, cwl_package_package)
     # Assertions
-    for key in expected:
-        assert key in wps_package_metadata
-        assert sorted(wps_package_metadata[key]) == sorted(expected[key])
+    assert wps_package_metadata == expected
diff --git a/weaver/processes/wps_package.py b/weaver/processes/wps_package.py
index 112840d04..8e4598def 100644
--- a/weaver/processes/wps_package.py
+++ b/weaver/processes/wps_package.py
@@ -719,17 +719,30 @@ def _update_package_metadata(wps_package_metadata, cwl_package_package):
         )
 
     # specific use case with a different mapping
-    if "s:version" in cwl_package_package:
+    # https://docs.ogc.org/bp/20-089r1.html#toc31
+    if "s:version" in cwl_package_package or "s:softwareVersion" in cwl_package_package:
         wps_package_metadata["version"] = str(
-             str(cwl_package_package.get("s:version", "")) or str(wps_package_metadata.get("version"))
+            str(wps_package_metadata.get("version", ""))
+            or cwl_package_package.get("s:version")
+            or cwl_package_package.get("s:softwareVersion")
         )
+    else:
+        wps_package_metadata["version"] = str(wps_package_metadata.get("version", ""))
 
     for metadata_mapping in SUPPORTED_METADATA_MAPPING:
         if metadata_mapping in cwl_package_package:
             metadata = wps_package_metadata.get("metadata", [])
-            if (isinstance((cwl_package_package[metadata_mapping]), str)
-                    and urlparse(cwl_package_package[metadata_mapping]) != ""):
+            if (
+                isinstance((cwl_package_package[metadata_mapping]), str)
+                and urlparse(cwl_package_package[metadata_mapping]).scheme != ""
+            ):
+                url = cwl_package_package[metadata_mapping]
+                if metadata_mapping == "s:codeRepository":
+                    type = "text/html"
+                else:
+                    type = get_content_type(os.path.splitext(url)[-1], default=ContentType.TEXT_PLAIN)
                 metadata.append({
+                    "type": type,
                     "rel": metadata_mapping.strip("s:"),
                     "href": cwl_package_package[metadata_mapping]
                 })
diff --git a/weaver/typedefs.py b/weaver/typedefs.py
index b90aadd48..00203d778 100644
--- a/weaver/typedefs.py
+++ b/weaver/typedefs.py
@@ -221,13 +221,14 @@ class CWL_SchemaName(Protocol):
         "$schemas": NotRequired[Dict[str, str]],
         "$graph": NotRequired[CWL_Graph],
         "s:author": NotRequired[List[Dict[str, JSON]]],
-        "s:citation": NotRequired[List[Dict[str, str]]],
+        "s:citation": NotRequired[List[Union[str, JSON]]],
         "s:codeRepository": NotRequired[Link],
         "s:contributor": NotRequired[List[Dict[str, JSON]]],
         "s:dateCreated": NotRequired[datetime],
         "s:license": NotRequired[Optional[Union[Dict[str, JSON], Link]]],
         "s:releaseNotes": NotRequired[Optional[Union[str, Link]]],
         "s:version": NotRequired[Optional[Union[str, Number]]],
+        "s:softwareVersion": NotRequired[Optional[Union[str, Number]]],
     }, total=False)
     CWL_WorkflowStepPackage = TypedDict("CWL_WorkflowStepPackage", {
         "id": str,          # reference ID of the package