Support XML Schema Definition Language (XSD) import

pyproject.toml

@@ -56,6 +56,7 @@ rdflib = "^7.1.1"
 jsonasobj2 = "^1.0.4"
 deprecation = "^2.1.0"
 numpy = "<2.0"
+lxml = "^5.3.0"
 pydbml = "^1.1.2"
 pyyaml = "^6.0.2"
 llm = {version = "^0.21", optional = true}
@@ -68,6 +69,10 @@ sphinx-click = ">=3.1.0"
 sphinxcontrib-mermaid = ">=0.9.2"
 myst-parser = "*"
 jupyter = ">=1.0.0"
+lxml-stubs = "^0.5.1"
+
+[tool.poetry.group.llm.dependencies]
+llm = ">=0.12"
 lxml = ">=4.9.1"
 deptry = "^0.23.0"
 

schema_automator/cli.py

@@ -32,6 +32,7 @@
 from schema_automator.importers.rdfs_import_engine import RdfsImportEngine
 from schema_automator.importers.sql_import_engine import SqlImportEngine
 from schema_automator.importers.tabular_import_engine import TableImportEngine
+from schema_automator.importers.xsd_import_engine import XsdImportEngine
 from schema_automator.utils.schemautils import write_schema
 from schema_automator import __version__
 
@@ -497,6 +498,23 @@ def import_rdfs(rdfsfile: str, output: str, metamodel_mappings: str, schema_name
     schema = sie.convert(rdfsfile, name=schema_name, **args)
     write_schema(schema, output)
 
+@main.command()
+@click.argument('xsd')
+@output_option
+@schema_name_option
+@click.option('--output', '-o', help="Path to saved yaml schema")
+def import_xsd(xsd: str, output: str, **kwargs):
+    """
+    Import an XML Schema Definition Language (XSD) schema to LinkML
+
+    Example:
+
+        schemauto import-xsd schema.xml -o prov.yaml
+    """
+    engine = XsdImportEngine()
+    schema = engine.convert(xsd, **kwargs)
+    write_schema(schema, output)
+
 @main.command()
 @click.argument('rdffile')
 @output_option

schema_automator/importers/__init__.py

@@ -3,3 +3,13 @@
 from schema_automator.importers.dosdp_import_engine import DOSDPImportEngine
 from schema_automator.importers.frictionless_import_engine import FrictionlessImportEngine
 from schema_automator.importers.cadsr_import_engine import CADSRImportEngine
+from schema_automator.importers.xsd_import_engine import XsdImportEngine
+
+__all__ = [
+    'JsonSchemaImportEngine',
+    'OwlImportEngine',
+    'DOSDPImportEngine',
+    'FrictionlessImportEngine',
+    'CADSRImportEngine',
+    'XsdImportEngine'
+]