Document strict flag and also use it to enable recover xml parsing

Author: mvantellingen

CHANGES

@@ -1,5 +1,9 @@
 1.2.0 (unreleased)
 ------------------
+ - Add flag to disable strict mode in the Client. This allows zeep to better
+   work with non standard compliant SOAP Servers. See the documentation for
+   usage and potential downsides.
+ - Minor refactor of resolving of elements for improved performance
  - Support the SOAP 1.2 'http://www.w3.org/2003/05/soap/bindings/HTTP/' 
    transport uri (#355)
  - Fallback to matching wsdl lookups to matching when the target namespace is
@@ -11,6 +15,7 @@
  - Filter out duplicate types and elements when dump the wsdl schema (#360)
  - Add ``zeep.CachingClient()`` which enables the SqliteCache by default
 
+
 1.1.0 (2017-02-18)
 ------------------
  - Fix an attribute error when an complexType used xsd:anyType as base 

docs/client.rst

@@ -12,6 +12,20 @@ Otherwise the first service and first port within that service are used as the
 default.
 
 
+Strict mode
+-----------
+By default zeep will operate in 'strict' mode. This can be disabled if you are
+working with a SOAP server which is not standards compliant by passing the
+kwarg ``strict=False`` to the ``Client``.  Disabling strict mode will change
+the following behaviour:
+
+ - The XML is parsed with the recover mode enabled
+ - Non optional elements are allowed to be missing in xsd:sequences
+
+Note that disabling strict mode should be considered a last resort since it
+might result in data-loss between the XML and the returned response.
+
+
 The ServiceProxy object
 -----------------------
 

src/zeep/__main__.py

@@ -27,6 +27,9 @@ def parse_arguments(args=None):
         '--verbose', action='store_true', help='Enable verbose output')
     parser.add_argument(
         '--profile', help="Enable profiling and save output to given file")
+    parser.add_argument(
+        '--no-strict', action='store_true', default=False,
+        help="Disable strict mode")
     return parser.parse_args(args)
 
 
@@ -72,7 +75,9 @@ def main(args):
 
     transport = Transport(cache=cache, session=session)
     st = time.time()
-    client = Client(args.wsdl_file, transport=transport)
+
+    strict = not args.no_strict
+    client = Client(args.wsdl_file, transport=transport, strict=strict)
     logger.debug("Loading WSDL took %sms", (time.time() - st) * 1000)
 
     if args.profile:

src/zeep/loader.py

@@ -18,7 +18,7 @@ def resolve(self, url, pubid, context):
             return self.resolve_string(content, context)
 
 
-def parse_xml(content, transport, base_url=None, recover=False):
+def parse_xml(content, transport, base_url=None, strict=False):
     """Parse an XML string and return the root Element.
 
     :param content: The XML string
@@ -28,27 +28,34 @@ def parse_xml(content, transport, base_url=None, recover=False):
     :param base_url: The base url of the document, used to make relative
       lookups absolute.
     :type base_url: str
-    :param recover: boolean to indicate if the lxml recover mode should be
-      enabled. Recover mode tries to parse invalid XML as best as it can.
-    :type recover: boolean
+    :param strict: boolean to indicate if the lxml should be parsed a 'strict'.
+      If false then the recover mode is enabled which tries to parse invalid
+      XML as best as it can.
+    :type strict: boolean
     :returns: The document root
     :rtype: lxml.etree._Element
 
     """
-    parser = etree.XMLParser(remove_comments=True, resolve_entities=False)
+    recover = not strict
+    parser = etree.XMLParser(
+        remove_comments=True, resolve_entities=False, recover=recover)
     parser.resolvers.add(ImportResolver(transport))
     try:
         return fromstring(content, parser=parser, base_url=base_url)
     except etree.XMLSyntaxError as exc:
         raise XMLSyntaxError("Invalid XML content received (%s)" % exc.message)
 
 
-def load_external(url, transport, base_url=None):
+def load_external(url, transport, base_url=None, strict=True):
     """Load an external XML document.
 
     :param url:
     :param transport:
     :param base_url:
+    :param strict: boolean to indicate if the lxml should be parsed a 'strict'.
+      If false then the recover mode is enabled which tries to parse invalid
+      XML as best as it can.
+    :type strict: boolean
 
     """
     if hasattr(url, 'read'):
@@ -57,7 +64,7 @@ def load_external(url, transport, base_url=None):
         if base_url:
             url = absolute_location(url, base_url)
         content = transport.load(url)
-    return parse_xml(content, transport, base_url)
+    return parse_xml(content, transport, base_url, strict=strict)
 
 
 def absolute_location(location, base):

src/zeep/wsdl/wsdl.py

@@ -8,8 +8,7 @@
 import six
 from lxml import etree
 
-from zeep.loader import (
-    absolute_location, is_relative_path, load_external, parse_xml)
+from zeep.loader import absolute_location, is_relative_path, load_external
 from zeep.utils import findall_multiple_ns
 from zeep.wsdl import parse
 from zeep.xsd import Schema
@@ -125,7 +124,8 @@ def _get_xml_document(self, location):
         :type location: string
 
         """
-        return load_external(location, self.transport, self.location)
+        return load_external(
+            location, self.transport, self.location, strict=self.strict)
 
     def _add_definition(self, definition):
         key = (definition.target_namespace, definition.location)

src/zeep/xsd/visitor.py

@@ -35,7 +35,13 @@ class SchemaVisitor(object):
     """Visitor which processes XSD files and registers global elements and
     types in the given schema.
 
+    :param schema:
+    :type schema: zeep.xsd.schema.Schema
+    :param document:
+    :type document: zeep.xsd.schema.SchemaDocument
+
     """
+
     def __init__(self, schema, document):
         self.document = document
         self.schema = schema
@@ -183,7 +189,10 @@ def visit_import(self, node, parent):
             return
 
         # Load the XML
-        schema_node = load_external(location, self.schema._transport)
+        schema_node = load_external(
+            location,
+            self.schema._transport,
+            strict=self.schema.strict)
 
         # Check if the xsd:import namespace matches the targetNamespace. If
         # the xsd:import statement didn't specify a namespace then make sure
@@ -227,7 +236,9 @@ def visit_include(self, node, parent):
             return
 
         schema_node = load_external(
-            location, self.schema._transport, base_url=self.document._base_url)
+            location, self.schema._transport,
+            base_url=self.document._base_url,
+            strict=self.schema.strict)
         self._includes.add(location)
 
         # When the included document has no default namespace defined but the