soap: handle recursive complexType (#81643)

Reference to already converted complexType are converted to JSON schema
references.

passerelle/apps/soap/models.py

@@ -15,6 +15,7 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 import collections
+import hashlib
 
 import zeep
 import zeep.helpers
@@ -184,42 +185,61 @@ class SOAPConnector(BaseResource, HTTPResource):
             operations_and_schemas.append((name, input_schema, output_schema))
         return operations_and_schemas
 
-    def type2schema(self, xsd_type, keep_root=False, compress=False):
-        # simplify schema: when a type contains a unique element, it will try
-        # to match any dict or list with it on input and will flatten the
-        # schema on output.
-        if (
-            isinstance(xsd_type, zeep.xsd.ComplexType)
-            and len(xsd_type.elements) == 1
-            and not keep_root
-            and compress
-        ):
-            if xsd_type.elements[0][1].max_occurs != 1:
+    @classmethod
+    def type2schema(cls, xsd_type, keep_root=False, compress=False):
+        seen = set()
+
+        def to_id(s):
+            return f'ref-{hashlib.md5(str(s).encode()).hexdigest()}'
+
+        def t2s(xsd_type):
+            type_name = xsd_type.qname or xsd_type.name
+            if isinstance(xsd_type, zeep.xsd.ComplexType):
+                if type_name in seen:
+                    return {'$ref': '#' + to_id(type_name)}
+                seen.add(type_name)
+            # simplify schema: when a type contains a unique element, it will try
+            # to match any dict or list with it on input and will flatten the
+            # schema on output.
+            if (
+                isinstance(xsd_type, zeep.xsd.ComplexType)
+                and len(xsd_type.elements) == 1
+                and not keep_root
+                and compress
+                # and is not recursive
+                and xsd_type.elements[0][1].type != xsd_type
+            ):
+                if xsd_type.elements[0][1].max_occurs != 1:
+                    schema = {
+                        'type': 'array',
+                        'items': t2s(xsd_type.elements[0][1].type),
+                    }
+                else:
+                    schema = t2s(xsd_type.elements[0][1].type)
+            elif isinstance(xsd_type, zeep.xsd.ComplexType):
+                properties = collections.OrderedDict()
                 schema = {
-                    'type': 'array',
-                    'items': self.type2schema(xsd_type.elements[0][1].type, compress=compress),
+                    'type': 'object',
+                    'properties': properties,
+                    '$anchor': to_id(type_name),
                 }
+                for key, element in xsd_type.elements:
+                    if element.min_occurs > 0:
+                        schema.setdefault('required', []).append(key)
+                    element_schema = t2s(element.type)
+                    if element.max_occurs == 'unbounded' or element.max_occurs > 1:
+                        element_schema = {'type': 'array', 'items': element_schema}
+                    properties[key] = element_schema
+                if not properties:
+                    schema = {'type': 'null'}
+            elif isinstance(xsd_type, zeep.xsd.BuiltinType):
+                schema = {'type': 'string'}
             else:
-                schema = self.type2schema(xsd_type.elements[0][1].type, compress=compress)
-        elif isinstance(xsd_type, zeep.xsd.ComplexType):
-            properties = collections.OrderedDict()
-            schema = {
-                'type': 'object',
-                'properties': properties,
-            }
-            for key, element in xsd_type.elements:
-                if element.min_occurs > 0:
-                    schema.setdefault('required', []).append(key)
-                element_schema = self.type2schema(element.type, compress=compress)
-                if element.max_occurs == 'unbounded' or element.max_occurs > 1:
-                    element_schema = {'type': 'array', 'items': element_schema}
-                properties[key] = element_schema
-            if not properties:
-                schema = {'type': 'null'}
-        elif isinstance(xsd_type, zeep.xsd.BuiltinType):
-            schema = {'type': 'string'}
-        else:
-            schema = {}
-        if xsd_type.qname:
-            schema['description'] = str(xsd_type.qname).replace('{http://www.w3.org/2001/XMLSchema}', 'xsd:')
-        return schema
+                schema = {}
+            if xsd_type.qname:
+                schema['description'] = str(xsd_type.qname).replace(
+                    '{http://www.w3.org/2001/XMLSchema}', 'xsd:'
+                )
+            return schema
+
+        return t2s(xsd_type)

tests/test_soap.py

@@ -109,8 +109,10 @@ xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
    </soap:Body>
 </soap:Envelope>'''
     INPUT_SCHEMA = {
+        '$anchor': 'ref-6adf97f83acf6453d4a6a4b1070f3754',
         'properties': {
             'firstName': {
+                '$anchor': 'ref-dbd3a37522045c54032a5b96864a500d',
                 'description': '{http://www.examples.com/wsdl/HelloService.wsdl}firstName',
                 'properties': {
                     'string': {'items': {'type': 'string', 'description': 'xsd:string'}, 'type': 'array'},
@@ -124,6 +126,7 @@ xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
         'type': 'object',
     }
     OUTPUT_SCHEMA = {
+        '$anchor': 'ref-6adf97f83acf6453d4a6a4b1070f3754',
         'properties': {
             'greeting': {'type': 'string', 'description': 'xsd:string'},
             'who': {'type': 'string', 'description': 'xsd:string'},
@@ -157,6 +160,11 @@ class SOAP12(SOAP11):
     <xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema"
       xmlns:tns="urn:examples:helloservice"
       targetNamespace="urn:examples:helloservice">
+      <xsd:complexType name="recurse">
+        <xsd:sequence>
+          <xsd:element name="anotherme" type="tns:recurse" minOccurs="0"/>
+        </xsd:sequence>
+      </xsd:complexType>
       <xsd:element name="sayHello">
         <xsd:complexType>
           <xsd:sequence>
@@ -170,6 +178,7 @@ class SOAP12(SOAP11):
           <xsd:sequence>
             <xsd:element name="greeting" type="xsd:string"/>
             <xsd:element name="who" type="xsd:string" maxOccurs="unbounded"/>
+            <xsd:element name="recursion" type="tns:recurse" minOccurs="0"/>
           </xsd:sequence>
         </xsd:complexType>
       </xsd:element>
@@ -223,10 +232,12 @@ class SOAP12(SOAP11):
      <sayHelloResponse xmlns="urn:examples:helloservice">
        <greeting>Hello</greeting>
        <who>John!</who>
+       <recursion><anotherme/></recursion>
      </sayHelloResponse>
   </soap:Body>
 </soap:Envelope>'''
     INPUT_SCHEMA = {
+        '$anchor': 'ref-5b712371a9c9bf61f983831c2ed3f364',
         'type': 'object',
         'properties': {
             'firstName': {'type': 'array', 'items': {'type': 'string', 'description': 'xsd:string'}},
@@ -236,6 +247,7 @@ class SOAP12(SOAP11):
         'description': '{urn:examples:helloservice}sayHello',
     }
     OUTPUT_SCHEMA = {
+        '$anchor': 'ref-5e505e086d14d5417f2799da5c085712',
         'description': '{urn:examples:helloservice}sayHelloResponse',
         'properties': {
             'greeting': {'type': 'string', 'description': 'xsd:string'},
@@ -243,6 +255,16 @@ class SOAP12(SOAP11):
                 'type': 'array',
                 'items': {'type': 'string', 'description': 'xsd:string'},
             },
+            'recursion': {
+                '$anchor': 'ref-63d3d62358d2daf62cd2ebd07640165e',
+                'description': '{urn:examples:helloservice}recurse',
+                'type': 'object',
+                'properties': {
+                    'anotherme': {
+                        '$ref': '#ref-63d3d62358d2daf62cd2ebd07640165e',
+                    }
+                },
+            },
         },
         'required': ['greeting', 'who'],
         'type': 'object',
@@ -255,6 +277,7 @@ class SOAP12(SOAP11):
     OUTPUT_DATA = {
         'greeting': 'Hello',
         'who': ['John!'],
+        'recursion': {'anotherme': None},
     }
     VALIDATION_ERROR = 'Expected at least 1 items (minOccurs check) 0 items found. (sayHello.firstName)'