pycodegen: safe identifiers don't have periods nor start with digits

Author: mr-c

schema_salad/python_codegen.py

@@ -97,10 +97,12 @@ def safe_name(name: str) -> str:
         avn = schema.avro_field_name(name)
         if avn.startswith("anon."):
             avn = avn[5:]
-        if avn in ("class", "in"):
+        elif avn[0].isdigit():
+            avn = f"_{avn}"
+        elif avn in ("class", "in"):
             # reserved words
-            avn = avn + "_"
-        return avn
+            avn = f"{avn}_"
+        return avn.replace(".", "_")
 
     def prologue(self) -> None:
         """Trigger to generate the prolouge code."""

schema_salad/tests/test_python_codegen.py

@@ -12,12 +12,18 @@
 from schema_salad import codegen
 from schema_salad.avro.schema import Names
 from schema_salad.fetcher import DefaultFetcher
+from schema_salad.python_codegen import PythonCodeGen
 from schema_salad.python_codegen_support import LoadingOptions
 from schema_salad.schema import load_schema
 
 from .util import basket_file_uri, cwl_file_uri, get_data, metaschema_file_uri
 
 
+def test_safe_identifiers() -> None:
+    """Affirm correct construction of identifiers safe for Python."""
+    assert PythonCodeGen.safe_name("5.8s_pattern") == "_5_8s_pattern"
+
+
 def test_cwl_gen(tmp_path: Path) -> None:
     src_target = tmp_path / "src.py"
     python_codegen(cwl_file_uri, src_target)