Merge branch 'main' of https://github.com/bigbio/sdrf-pipelines into dev

.github/workflows/ci.yml

@@ -15,8 +15,21 @@ jobs:
     steps:
       - uses: actions/checkout@v2
 
+      # Set up the specific Python version for Black
+      - name: Set up Python 3.12
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.12  # Set your desired Python version here
+
+      # Install Black and dependencies if necessary
+      - name: Install Black
+        run: |
+          python -m pip install --upgrade pip
+          pip install black
+
+      # Run Black linting
       - name: Check code lints with Black
-        uses: psf/black@stable
+        run: black . --check
 
       # If the above check failed, post a comment on the PR explaining the failure
       - name: Post PR comment

recipe/meta.yaml

@@ -1,7 +1,7 @@
 # recipe/meta.yaml
 package:
   name: sdrf-pipelines
-  version: "0.0.29"
+  version: "0.0.30"
 
 source:
   path: ../
@@ -17,6 +17,7 @@ requirements:
   host:
     - pip
     - python >=3.5
+    - setuptools
   run:
     - click
     - requests

requirements-dev.txt

@@ -2,4 +2,5 @@ black
 isort
 pre-commit
 pytest
-pytest-datadir
+pytest-datadir
+setuptools

requirements.txt

@@ -11,4 +11,5 @@ numpy
 defusedxml
 pyarrow
 duckdb
-rdflib
+rdflib
+setuptools

sdrf_pipelines/sdrf/sdrf_schema.py

@@ -56,13 +56,15 @@ def ontology_term_parser(cell_value: str = None):
         term[TERM_NAME] = values[0].lower()
     else:
         for name in values:
-            value_terms = name.split("=")
-            if len(value_terms) == 2:
-                term[value_terms[0].strip().upper()] = value_terms[1].strip().lower()
-            else:
+            value_terms = name.split("=", 1)
+            if len(value_terms) == 1:
+                raise ValueError("Not a key-value pair: " + name)
+            if "=" in value_terms[1] and value_terms[0].lower() != "cs":
                 raise ValueError(
                     f"Invalid term: {name} after splitting by '=', please check the prefix (e.g. AC, NT, " f"TA..)"
                 )
+            term[value_terms[0].strip().upper()] = value_terms[1].strip().lower()
+
     return term
 
 

setup.py

@@ -49,6 +49,7 @@ def get_version(rel_path):
         "pyarrow",
         "duckdb",
         "rdflib",
+        "setuptools",
     ],
     entry_points={"console_scripts": ["parse_sdrf = sdrf_pipelines.parse_sdrf:main"]},
     platforms=["any"],