summaryrefslogtreecommitdiff
path: root/libxsd-frontend/xsd-frontend/parser.cxx
diff options
context:
space:
mode:
Diffstat (limited to 'libxsd-frontend/xsd-frontend/parser.cxx')
-rw-r--r--libxsd-frontend/xsd-frontend/parser.cxx5164
1 files changed, 5164 insertions, 0 deletions
diff --git a/libxsd-frontend/xsd-frontend/parser.cxx b/libxsd-frontend/xsd-frontend/parser.cxx
new file mode 100644
index 0000000..6b582ab
--- /dev/null
+++ b/libxsd-frontend/xsd-frontend/parser.cxx
@@ -0,0 +1,5164 @@
+// file : xsd-frontend/parser.cxx
+// copyright : Copyright (c) 2005-2014 Code Synthesis Tools CC
+// license : GNU GPL v2 + exceptions; see accompanying LICENSE file
+
+#include <map>
+#include <stack>
+#include <vector>
+#include <iostream>
+#include <sstream>
+
+#include <cutl/compiler/type-id.hxx>
+
+#include <xsd-frontend/version.hxx> // Check Xerces-C++ version.
+#include <xsd-frontend/xml.hxx>
+#include <xsd-frontend/parser.hxx>
+#include <xsd-frontend/schema-dom-parser.hxx>
+
+#include <xsd-frontend/semantic-graph.hxx>
+#include <xsd-frontend/traversal.hxx>
+
+//@@ Do i need this?
+//
+#include <xercesc/dom/DOM.hpp>
+
+#include <xercesc/sax/ErrorHandler.hpp>
+#include <xercesc/sax/SAXParseException.hpp>
+
+#include <xercesc/sax2/SAX2XMLReader.hpp>
+#include <xercesc/sax2/XMLReaderFactory.hpp>
+
+#include <xercesc/util/XMLUniDefs.hpp>
+#include <xercesc/util/XMLString.hpp>
+#include <xercesc/util/PlatformUtils.hpp>
+#include <xercesc/util/BinInputStream.hpp>
+#include <xercesc/util/BinFileInputStream.hpp>
+
+#include <xercesc/validators/common/Grammar.hpp>
+
+#include <xercesc/sax/InputSource.hpp>
+#include <xercesc/framework/LocalFileInputSource.hpp>
+#include <xercesc/framework/Wrapper4InputSource.hpp>
+
+using namespace std;
+
+using cutl::compiler::type_id;
+
+namespace XSDFrontend
+{
+ namespace Xerces = XML::Xerces;
+ using namespace SemanticGraph;
+
+ //@@ Port to tracing facility.
+ //
+ bool trace_ = false;
+
+ String const xsd = L"http://www.w3.org/2001/XMLSchema";
+ String const xse = L"http://www.codesynthesis.com/xmlns/xml-schema-extension";
+
+ namespace
+ {
+ //
+ // Exceptions.
+ //
+
+ struct NotNamespace
+ {
+ NotNamespace (String const& ns)
+ : ns_ (ns)
+ {
+ }
+
+ String const&
+ ns () const
+ {
+ return ns_;
+ }
+
+ private:
+ String ns_;
+ };
+
+ struct NotName
+ {
+ NotName (String const& ns, String const& name)
+ : ns_ (ns), name_ (name)
+ {
+ }
+
+ String const&
+ ns () const
+ {
+ return ns_;
+ }
+
+ String const&
+ name () const
+ {
+ return name_;
+ }
+
+ private:
+ String ns_;
+ String name_;
+ };
+
+ // Trim leading and trailing whitespaces.
+ //
+ template <typename C>
+ StringTemplate<C>
+ trim (StringTemplate<C> const& s)
+ {
+ typedef StringTemplate<C> String;
+
+ size_t size (s.size ());
+
+ if (size == 0)
+ return s;
+
+ C const* f (s.c_str ());
+ C const* l (f + size);
+
+ C const* of (f);
+
+ while (f < l &&
+ (*f == C (0x20) || *f == C (0x0A) ||
+ *f == C (0x0D) || *f == C (0x09)))
+ ++f;
+
+ --l;
+
+ C const* ol (l);
+
+ while (l > f &&
+ (*l == C (0x20) || *l == C (0x0A) ||
+ *l == C (0x0D) || *l == C (0x09)))
+ --l;
+
+ if (f != of || l != ol)
+ return f <= l ? String (f, l - f + 1) : String ();
+ else
+ return s;
+ }
+
+ // Name cache. We only support maximum two nodes with the same
+ // name in the cache (e.g., element and type). For (rare) cases
+ // where there is three or more names, there will be a cache miss.
+ //
+ struct CacheNodes
+ {
+ CacheNodes () : first (0), second (0) {}
+
+ Nameable* first;
+ Nameable* second;
+ };
+
+ typedef std::map<String, CacheNodes> NodeMap;
+ typedef std::map<String, NodeMap> NamespaceMap;
+ typedef std::vector<SemanticGraph::Member*> DefaultValues;
+
+ template <typename X>
+ X&
+ resolve (String const& ns_name,
+ String const& uq_name,
+ Schema& s_,
+ NamespaceMap& cache)
+ {
+ // First check the cache.
+ //
+ NamespaceMap::iterator i (cache.find (ns_name));
+
+ if (i != cache.end ())
+ {
+ NodeMap::iterator j (i->second.find (uq_name));
+
+ if (j != i->second.end ())
+ {
+ X* x;
+
+ if ((x = dynamic_cast<X*> (j->second.first)) ||
+ (x = dynamic_cast<X*> (j->second.second)))
+ return *x;
+ }
+ }
+
+ Scope::NamesIteratorPair nss (s_.find (ns_name));
+
+ if (nss.first == nss.second)
+ throw NotNamespace (ns_name);
+
+ for (; nss.first != nss.second; ++nss.first)
+ {
+ Namespace& ns (dynamic_cast<Namespace&> (nss.first->named ()));
+
+ Scope::NamesIteratorPair types (ns.find (uq_name));
+
+ for (; types.first != types.second; ++types.first)
+ {
+ if (X* x = dynamic_cast<X*> (&types.first->named ()))
+ {
+ if (trace_)
+ wcout << "successfully resolved '" << ns_name << '#' << uq_name
+ << "'" << endl;
+
+ // Add to the cache if there are free slots.
+ //
+ NodeMap& m (i != cache.end () ? i->second : cache[ns_name]);
+ CacheNodes& n (m[uq_name]);
+
+ if (n.first == 0)
+ n.first = x;
+ else if (n.second == 0)
+ n.second = x;
+
+ return *x;
+ }
+ }
+ }
+
+ throw NotName (ns_name, uq_name);
+ }
+
+ //
+ //
+ typedef std::map<String, String> Facets;
+
+ void
+ copy_facets (Restricts& r, Facets const& f)
+ {
+ for (Facets::const_iterator i (f.begin ()), e (f.end ()); i != e; ++i)
+ r.facet_insert (i->first, i->second);
+ }
+
+ //
+ //
+ struct UnionMemberType
+ {
+ UnionMemberType (String const& ns, String const& uq)
+ : ns_name (ns), uq_name (uq)
+ {
+ }
+
+ String ns_name;
+ String uq_name;
+ };
+
+ typedef std::vector<UnionMemberType> UnionMemberTypes;
+
+ //
+ //
+ struct ElementGroupRef
+ {
+ ElementGroupRef (String const& uq_name_, String const& ns_name_,
+ unsigned long min_, unsigned long max_,
+ Compositor& compositor, Scope& scope)
+ : uq_name (uq_name_), ns_name (ns_name_),
+ min (min_), max (max_)
+ {
+ contains_pos = compositor.contains_end ();
+ if (compositor.contains_begin () != contains_pos)
+ --contains_pos;
+
+ names_pos = scope.names_end ();
+ if (scope.names_begin () != names_pos)
+ --names_pos;
+ }
+
+ ElementGroupRef (String const& uq_name_, String const& ns_name_,
+ unsigned long min_, unsigned long max_,
+ Scope& scope)
+ : uq_name (uq_name_), ns_name (ns_name_),
+ min (min_), max (max_)
+ {
+ names_pos = scope.names_end ();
+ if (scope.names_begin () != names_pos)
+ --names_pos;
+ }
+
+ String uq_name;
+ String ns_name;
+ unsigned long min, max;
+ Compositor::ContainsIterator contains_pos;
+ Scope::NamesIterator names_pos;
+ };
+
+ typedef std::vector<ElementGroupRef> ElementGroupRefs;
+
+ //
+ //
+ struct AttributeGroupRef
+ {
+ AttributeGroupRef (String const& uq_name_,
+ String const& ns_name_,
+ Scope& scope)
+ : uq_name (uq_name_), ns_name (ns_name_)
+ {
+ names_pos = scope.names_end ();
+ if (scope.names_begin () != names_pos)
+ --names_pos;
+ }
+
+ String uq_name;
+ String ns_name;
+ Scope::NamesIterator names_pos;
+ };
+
+ typedef std::vector<AttributeGroupRef> AttributeGroupRefs;
+
+
+ //
+ //
+ template <typename N, typename A>
+ struct NodeArgs
+ {
+ NodeArgs (N& node, A arg)
+ : node_ (node), arg_ (arg)
+ {
+ }
+
+ operator N& () const
+ {
+ return node_;
+ }
+
+ template <typename E>
+ void
+ add_edge_left (E& e)
+ {
+ node_.add_edge_left (e, arg_);
+ }
+
+ template <typename E>
+ void
+ add_edge_right (E& e)
+ {
+ node_.add_edge_right (e, arg_);
+ }
+
+ private:
+ N& node_;
+ A arg_;
+ };
+
+
+ //
+ //
+ struct Resolver : Traversal::Element,
+ Traversal::Attribute,
+ Traversal::Fundamental::IdRef,
+ Traversal::Fundamental::IdRefs,
+ Traversal::List,
+ Traversal::Union,
+ Traversal::Complex,
+ Traversal::Enumeration,
+ Traversal::ElementGroup,
+ Traversal::AttributeGroup,
+ Traversal::Compositor
+ {
+ Resolver (Schema& s,
+ bool& valid,
+ NamespaceMap& cache,
+ DefaultValues& default_values)
+ : s_ (s),
+ valid_ (valid),
+ cache_ (cache),
+ default_values_ (default_values)
+ {
+ *this >> contains_compositor >> *this;
+ }
+
+ void
+ traverse (SemanticGraph::Attribute& a)
+ {
+ // Avoid traversing attribute more than once.
+ //
+ if (!a.context ().count ("attribute-traversed"))
+ {
+ a.context ().set ("attribute-traversed", true);
+ SemanticGraph::Member& m (a);
+ resolve_member (m);
+ }
+ }
+
+ void
+ traverse (SemanticGraph::Element& e)
+ {
+ resolve_element (e);
+ }
+
+ void
+ resolve_element (SemanticGraph::Element& e)
+ {
+ // Avoid resolving element more than once.
+ //
+ if (e.context ().count ("element-resolved"))
+ return;
+
+ e.context ().set ("element-resolved", true);
+
+ {
+ SemanticGraph::Member& m (e);
+ resolve_member (m);
+ }
+
+ if (e.context ().count ("substitution-ns-name"))
+ {
+ String ns_name (e.context ().get<String> ("substitution-ns-name"));
+ String uq_name (e.context ().get<String> ("substitution-uq-name"));
+
+ e.context ().remove ("substitution-ns-name");
+ e.context ().remove ("substitution-uq-name");
+
+ try
+ {
+ SemanticGraph::Element& root (
+ resolve<SemanticGraph::Element> (ns_name, uq_name, s_, cache_));
+
+ s_.new_edge<Substitutes> (e, root);
+
+ // See if we need to derive the type of this element from the
+ // one it substitutes.
+ //
+ if (!e.typed_p ())
+ {
+ resolve_member (root); // Make sure the type is resolved.
+ s_.new_edge<Belongs> (e, root.type ());
+ }
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+ abort ();
+ }
+ }
+ catch (NotName const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve name '" << ex.name ()
+ << "' inside namespace '" << ex.ns () << "'" <<endl;
+ abort ();
+ }
+ }
+ }
+ }
+
+ void
+ resolve_member (SemanticGraph::Member& m)
+ {
+ using SemanticGraph::Member;
+ using SemanticGraph::Element;
+ using SemanticGraph::Attribute;
+
+ try
+ {
+ String ns_name;
+ String uq_name;
+
+ if (m.context ().count ("type-ns-name"))
+ {
+ ns_name = m.context ().get<String> ("type-ns-name");
+ uq_name = m.context ().get<String> ("type-uq-name");
+
+ m.context ().remove ("type-ns-name");
+ m.context ().remove ("type-uq-name");
+ m.context ().remove ("edge-type-id");
+
+ s_.new_edge<Belongs> (
+ m, resolve<SemanticGraph::Type> (ns_name, uq_name, s_, cache_));
+ }
+ else if (m.context ().count ("instance-ns-name"))
+ {
+ ns_name = m.context ().get<String> ("instance-ns-name");
+ uq_name = m.context ().get<String> ("instance-uq-name");
+
+ m.context ().remove ("instance-ns-name");
+ m.context ().remove ("instance-uq-name");
+
+ // Resolve the name to the same type. It is legal to have
+ // an element and an attribute with the same name.
+ //
+ Member& ref (
+ m.is_a<Element> ()
+ ? static_cast<Member&> (
+ resolve<Element> (ns_name, uq_name, s_, cache_))
+ : static_cast<Member&> (
+ resolve<Attribute> (ns_name, uq_name, s_, cache_)));
+
+ // Make sure the referenced member is fully resolved.
+ // @@ Substitutes edge won't be resolved.
+ //
+ resolve_member (ref);
+
+
+ // Substitution group info. We have to test for both resolved
+ // and unresolved cases since we don't know whether it was
+ // resolved or not.
+ //
+ if (ref.is_a<Element> ())
+ {
+ Element& m_e (dynamic_cast<Element&> (m));
+ Element& ref_e (dynamic_cast<Element&> (ref));
+
+ if (ref_e.substitutes_p ())
+ {
+ s_.new_edge<Substitutes> (m_e, ref_e.substitutes ().root ());
+ }
+ else if (ref_e.context ().count ("substitution-ns-name"))
+ {
+ m_e.context ().set (
+ "substitution-ns-name",
+ ref_e.context ().get<String> ("substitution-ns-name"));
+
+ m_e.context ().set (
+ "substitution-uq-name",
+ ref_e.context ().get<String> ("substitution-uq-name"));
+ }
+ }
+
+ //
+ //
+ s_.new_edge<BelongsToNamespace> (m, ref.namespace_ ());
+
+ // Transfer default and fixed values if we haven't already
+ // gotten them.
+ //
+ if (!m.default_p ())
+ {
+ if (ref.fixed_p ())
+ m.fixed (ref.value ());
+ else if (ref.default_p ())
+ {
+ // Default value applies only if the attribute is optional.
+ //
+ if (Attribute* a = dynamic_cast<Attribute*> (&m))
+ {
+ if (a->optional_p ())
+ m.default_ (ref.value ());
+ }
+ else
+ m.default_ (ref.value ());
+ }
+
+ if (m.default_p ())
+ {
+ m.context ().set (
+ "dom-node",
+ ref.context ().get<Xerces::DOMElement*> ("dom-node"));
+ default_values_.push_back (&m);
+ }
+ }
+
+ // Transfer annotation if we haven't already gotten it.
+ //
+ if (!m.annotated_p () && ref.annotated_p ())
+ s_.new_edge<Annotates> (ref.annotation (), m);
+
+ // Type info. Can be missing for a substitution group member.
+ //
+ if (ref.typed_p ())
+ s_.new_edge<Belongs> (m, ref.type ());
+ }
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+ abort ();
+ }
+ }
+ catch (NotName const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve name '" << ex.name ()
+ << "' inside namespace '" << ex.ns () << "'" <<endl;
+ abort ();
+ }
+ }
+ }
+
+ void
+ traverse (SemanticGraph::Fundamental::IdRef& i)
+ {
+ ref_type (i);
+ }
+
+ void
+ traverse (SemanticGraph::Fundamental::IdRefs& i)
+ {
+ ref_type (i);
+ }
+
+ void
+ ref_type (SemanticGraph::Specialization& s)
+ {
+ if (s.context ().count ("type-ns-name"))
+ {
+ String ns_name (s.context ().get<String> ("type-ns-name"));
+ String uq_name (s.context ().get<String> ("type-uq-name"));
+
+ s.context ().remove ("type-ns-name");
+ s.context ().remove ("type-uq-name");
+ s.context ().remove ("edge-type-id");
+
+ try
+ {
+ s_.new_edge<Arguments> (
+ resolve<SemanticGraph::Type> (ns_name, uq_name, s_, cache_), s);
+ }
+ catch (NotName const& ex)
+ {
+ wcerr << s.file () << ":" << s.line () << ":" << s.column () << ": "
+ << "error: unable to resolve type '" << uq_name << "' "
+ << "in namespace '" << ns_name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+
+ void
+ traverse (SemanticGraph::List& l)
+ {
+ if (l.context ().count ("type-ns-name"))
+ {
+ String ns_name (l.context ().get<String> ("type-ns-name"));
+ String uq_name (l.context ().get<String> ("type-uq-name"));
+
+ l.context ().remove ("type-ns-name");
+ l.context ().remove ("type-uq-name");
+ l.context ().remove ("edge-type-id");
+
+ try
+ {
+ s_.new_edge<Arguments> (
+ resolve<SemanticGraph::Type> (ns_name, uq_name, s_, cache_), l);
+ }
+ catch (NotName const& ex)
+ {
+ wcerr << l.file () << ":" << l.line () << ":" << l.column () << ": "
+ << "error: unable to resolve item type '" << uq_name << "' "
+ << "in namespace '" << ns_name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ Traversal::List::traverse (l);
+ }
+
+ void
+ traverse (SemanticGraph::Union& u)
+ {
+ using SemanticGraph::Union;
+
+ if (u.context ().count ("union-member-types"))
+ {
+ UnionMemberTypes const& m (
+ u.context ().get<UnionMemberTypes> ("union-member-types"));
+
+ // Process it backwards so that we can just insert each
+ // edge in the front.
+ //
+ for (UnionMemberTypes::const_reverse_iterator i (m.rbegin ());
+ i != m.rend (); i++)
+ {
+ try
+ {
+ NodeArgs<Union, Union::ArgumentedIterator> na (
+ u, u.argumented_begin ());
+
+ s_.new_edge<Arguments> (
+ resolve<SemanticGraph::Type> (
+ i->ns_name, i->uq_name, s_, cache_), na);
+ }
+ catch (NotName const& ex)
+ {
+ wcerr << u.file () << ":" << u.line () << ":" << u.column () << ": "
+ << "error: unable to resolve item type '" << i->uq_name << "' "
+ << "in namespace '" << i->ns_name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ u.context ().remove ("union-member-types");
+ }
+
+ Traversal::Union::traverse (u);
+ }
+
+ void
+ traverse (SemanticGraph::Complex& c)
+ {
+ // Avoid traversing complex type more than once.
+ //
+ if (c.context ().count ("complex-type-resolved"))
+ return;
+
+ c.context ().set ("complex-type-resolved", true);
+
+ // Resolve base type if any.
+ //
+ if (c.context ().count ("type-ns-name"))
+ {
+ String ns_name (c.context ().get<String> ("type-ns-name"));
+ String uq_name (c.context ().get<String> ("type-uq-name"));
+ type_id edge_id (c.context ().get<type_id> ("edge-type-id"));
+
+ c.context ().remove ("type-ns-name");
+ c.context ().remove ("type-uq-name");
+ c.context ().remove ("edge-type-id");
+
+ try
+ {
+ if (edge_id == typeid (Extends))
+ {
+ s_.new_edge<Extends> (
+ c, resolve<SemanticGraph::Type> (
+ ns_name, uq_name, s_, cache_));
+ }
+ else if (edge_id == typeid (Restricts))
+ {
+ Restricts& r (
+ s_.new_edge<Restricts> (
+ c, resolve<SemanticGraph::Type> (
+ ns_name, uq_name, s_, cache_)));
+
+ if (c.context ().count ("facets"))
+ {
+ Facets const& f (c.context ().get<Facets> ("facets"));
+ copy_facets (r, f);
+ c.context ().remove ("facets");
+ }
+ }
+ else
+ assert (false);
+ }
+ catch (NotName const& ex)
+ {
+ wcerr << c.file () << ":" << c.line () << ":" << c.column () << ": "
+ << "error: unable to resolve base type '" << uq_name << "' "
+ << "in namespace '" << ns_name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ // Resolve attribute-group-refs. Do it before element-group-refs
+ // so that if the scope was empty they end up at the end.
+ //
+ if (c.context ().count ("attribute-group-refs"))
+ {
+ AttributeGroupRefs& refs (
+ c.context ().get<AttributeGroupRefs> ("attribute-group-refs"));
+
+ // Handle refs from last to first so that multiple insertions
+ // to an empty list (always front) end up in proper order.
+ //
+ for (AttributeGroupRefs::reverse_iterator i (refs.rbegin ());
+ i != refs.rend (); ++i)
+ {
+ clone_attribute_group_content (*i, c);
+ }
+
+ c.context ().remove ("attribute-group-refs");
+ }
+
+ // Resolve element-group-ref if any.
+ //
+ if (c.context ().count ("element-group-ref"))
+ {
+ using SemanticGraph::Compositor;
+
+ ElementGroupRef& ref (
+ c.context ().get<ElementGroupRef> ("element-group-ref"));
+
+ Compositor* comp (clone_element_group_content (c, ref));
+
+ // Create ContainsCompositor edge.
+ //
+ if (comp)
+ s_.new_edge<ContainsCompositor> (c, *comp, ref.min, ref.max);
+
+ c.context ().remove ("element-group-ref");
+ }
+
+ Traversal::Complex::traverse (c);
+ }
+
+ void
+ traverse (SemanticGraph::Enumeration& e)
+ {
+ // Resolve base type if any.
+ //
+ if (e.context ().count ("type-ns-name"))
+ {
+ String ns_name (e.context ().get<String> ("type-ns-name"));
+ String uq_name (e.context ().get<String> ("type-uq-name"));
+
+ e.context ().remove ("type-ns-name");
+ e.context ().remove ("type-uq-name");
+ e.context ().remove ("edge-type-id");
+
+ try
+ {
+ Restricts& r (
+ s_.new_edge<Restricts> (
+ e, resolve<SemanticGraph::Type> (
+ ns_name, uq_name, s_, cache_)));
+
+ if (e.context ().count ("facets"))
+ {
+ Facets const& f (e.context ().get<Facets> ("facets"));
+ copy_facets (r, f);
+ e.context ().remove ("facets");
+ }
+ }
+ catch (NotName const& ex)
+ {
+ wcerr << e.file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unable to resolve base type '" << uq_name << "' "
+ << "in namespace '" << ns_name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ Traversal::Enumeration::traverse (e);
+ }
+
+ void
+ traverse (SemanticGraph::ElementGroup& g)
+ {
+ // Avoid traversing groups more than once.
+ //
+ if (!g.context ().count ("element-group-traversed"))
+ {
+ g.context ().set ("element-group-traversed", true);
+ Traversal::ElementGroup::traverse (g);
+
+ // Note that setting element-group-resolved after traversing
+ // the group allows for a recursive shallow resolution using
+ // resolve_element_group.
+ //
+ g.context ().set ("element-group-resolved", true);
+ }
+ }
+
+ // We need a "shallow" resolve to break possible recursing:
+ // group->element->complexType->group.
+ //
+ void
+ resolve_element_group (SemanticGraph::ElementGroup& g)
+ {
+ using SemanticGraph::Scope;
+ using SemanticGraph::Element;
+
+ // Avoid resolving groups more than once.
+ //
+ if (!g.context ().count ("element-group-resolved"))
+ {
+ g.context ().set ("element-group-resolved", true);
+
+ for (Scope::NamesIterator i (g.names_begin ());
+ i != g.names_end (); ++i)
+ {
+ if (Element* e = dynamic_cast<Element*> (&i->named ()))
+ resolve_element (*e);
+ }
+
+ traverse (g.contains_compositor ().compositor ());
+ }
+ }
+
+ void
+ traverse (SemanticGraph::AttributeGroup& g)
+ {
+ // Avoid traversing groups more than once.
+ //
+ if (g.context ().count ("attribute-group-resolved"))
+ return;
+
+ g.context ().set ("attribute-group-resolved", true);
+
+ // Resolve attribute-group-refs.
+ //
+ if (g.context ().count ("attribute-group-refs"))
+ {
+ AttributeGroupRefs& refs (
+ g.context ().get<AttributeGroupRefs> ("attribute-group-refs"));
+
+ // Handle refs from last to first so that multiple insertions
+ // to an empty list (always front) end up in proper order.
+ //
+ for (AttributeGroupRefs::reverse_iterator i (refs.rbegin ());
+ i != refs.rend (); ++i)
+ {
+ clone_attribute_group_content (*i, g);
+ }
+
+ g.context ().remove ("attribute-group-refs");
+ }
+
+ Traversal::AttributeGroup::traverse (g);
+ }
+
+ void
+ traverse (SemanticGraph::Compositor& c)
+ {
+ using SemanticGraph::Compositor;
+
+ // Resolve element-group-refs if any.
+ //
+ if (c.context ().count ("element-group-refs"))
+ {
+ using SemanticGraph::Scope;
+
+ ElementGroupRefs& refs (
+ c.context ().get<ElementGroupRefs> ("element-group-refs"));
+
+ // Handle refs from last to first so that multiple insertions
+ // to an empty list (always front) end up in proper order.
+ //
+ for (ElementGroupRefs::reverse_iterator i (refs.rbegin ());
+ i != refs.rend (); ++i)
+ {
+ // Find our scope.
+ //
+ Compositor* j (&c);
+
+ while(!j->contained_compositor_p ())
+ j = &j->contained_particle ().compositor ();
+
+ Compositor* comp (
+ clone_element_group_content (
+ dynamic_cast<Scope&> (j->contained_compositor ().container ()),
+ *i));
+
+ // Create ContainsParticle edge.
+ //
+ if (comp)
+ {
+ NodeArgs<Compositor, Compositor::ContainsIterator> na (
+ c, i->contains_pos);
+ s_.new_edge<ContainsParticle> (na, *comp, i->min, i->max);
+ }
+ }
+
+ c.context ().remove ("element-group-refs");
+ }
+
+ // Traverse recursively but only particles that are compositors.
+ // This way we won't trigger anonymous type traversal (via member)
+ // and therefore can call this functions from resolve_element_group
+ // to completely resolve a group.
+ //
+ for (Compositor::ContainsIterator i (c.contains_begin ()),
+ e (c.contains_end ()); i != e; ++i)
+ {
+ SemanticGraph::Particle& p (i->particle ());
+
+ if (p.is_a<Compositor> ())
+ dispatch (p);
+ }
+
+ // Traversal::Compositor::traverse (c);
+ }
+
+ SemanticGraph::Compositor*
+ clone_element_group_content (SemanticGraph::Scope& s,
+ ElementGroupRef const& ref)
+ {
+ using SemanticGraph::Scope;
+ using SemanticGraph::Compositor;
+ using SemanticGraph::ElementGroup;
+
+ try
+ {
+ ElementGroup& g (
+ resolve<ElementGroup> (ref.ns_name, ref.uq_name, s_, cache_));
+
+ // Make sure the group and all its content are fully resolved.
+ //
+ resolve_element_group (g);
+
+ Scope::NamesIterator pos (ref.names_pos);
+ Compositor& root (g.contains_compositor ().compositor ());
+ Compositor& copy (clone_compositor (root, s, pos));
+
+ return &copy;
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+ abort ();
+ }
+ }
+ catch (NotName const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve name '" << ex.name ()
+ << "' inside namespace '" << ex.ns () << "'" << endl;
+ abort ();
+ }
+ }
+
+ return 0;
+ }
+
+ SemanticGraph::Compositor&
+ clone_compositor (SemanticGraph::Compositor& c,
+ SemanticGraph::Scope& scope,
+ SemanticGraph::Scope::NamesIterator& pos)
+ {
+ using SemanticGraph::Any;
+ using SemanticGraph::Element;
+ using SemanticGraph::Particle;
+ using SemanticGraph::Compositor;
+
+ Compositor* tmp (0);
+
+ if (c.is_a<All> ())
+ tmp = &s_.new_node<All> (c.file (), c.line (), c.column ());
+ else if (c.is_a<Choice> ())
+ tmp = &s_.new_node<Choice> (c.file (), c.line (), c.column ());
+ else if (c.is_a<Sequence> ())
+ tmp = &s_.new_node<Sequence> (c.file (), c.line (), c.column ());
+ else
+ assert (false);
+
+ Compositor& copy (*tmp);
+
+ // Copy annotation.
+ //
+ if (c.annotated_p ())
+ s_.new_edge<Annotates> (c.annotation (), copy);
+
+ for (Compositor::ContainsIterator i (c.contains_begin ());
+ i != c.contains_end (); ++i)
+ {
+ Particle& p (i->particle ());
+
+ if (p.is_a<Compositor> ())
+ {
+ Compositor& c (dynamic_cast<Compositor&> (p));
+ Compositor& cc (clone_compositor (c, scope, pos));
+
+ s_.new_edge<ContainsParticle> (copy, cc, i->min (), i->max ());
+ }
+ else if (p.is_a<Element> ())
+ {
+ Element& e (dynamic_cast<Element&> (p));
+ Element& ec (clone_element (e));
+
+ s_.new_edge<ContainsParticle> (copy, ec, i->min (), i->max ());
+
+ NodeArgs<Scope, Scope::NamesIterator> na (scope, pos);
+ s_.new_edge<Names> (na, ec, e.name ());
+ ++pos;
+ }
+ else if (p.is_a<Any> ())
+ {
+ Any& a (dynamic_cast<Any&> (p));
+ Any& ac (
+ s_.new_node<Any> (a.file (), a.line (), a.column (),
+ a.namespace_begin (), a.namespace_end ()));
+
+ ac.prototype (a);
+
+ s_.new_edge<ContainsParticle> (copy, ac, i->min (), i->max ());
+
+ // Transfer annotation.
+ //
+ if (a.annotated_p ())
+ s_.new_edge<Annotates> (a.annotation (), ac);
+
+ // Any has no name so we have to come up with a fake one in
+ // order to put it into the scope. Note that we cannot reuse
+ // the name from the prototype.
+
+ unsigned long count;
+ SemanticGraph::Context& ctx (scope.context ());
+
+ if (!ctx.count ("any-name-count"))
+ {
+ count = 0;
+ ctx.set ("any-name-count", count);
+ }
+ else
+ count = ++(ctx.get<unsigned long> ("any-name-count"));
+
+ std::basic_ostringstream<wchar_t> os;
+ os << "any #" << count;
+
+ NodeArgs<Scope, Scope::NamesIterator> na (scope, pos);
+ s_.new_edge<Names> (na, ac, os.str ());
+ ++pos;
+ }
+ else
+ assert (false);
+ }
+
+ return copy;
+ }
+
+ // Clone a fully-resolved element. Note that it cannot be used as
+ // is to clone ref'ed element (default/fixed value, etc).
+ //
+ SemanticGraph::Element&
+ clone_element (SemanticGraph::Element& e)
+ {
+ using SemanticGraph::Element;
+
+ Element& copy (
+ s_.new_node<Element> (
+ e.file (), e.line (), e.column (), e.global_p (), e.qualified_p ()));
+
+ if (e.qualified_p ())
+ s_.new_edge<BelongsToNamespace> (copy, e.namespace_ ());
+
+ // Transfer default and fixed values.
+ //
+ if (e.fixed_p ())
+ copy.fixed (e.value ());
+ else if (e.default_p ())
+ copy.default_ (e.value ());
+
+ if (copy.default_p ())
+ {
+ copy.context ().set (
+ "dom-node",
+ e.context ().get<Xerces::DOMElement*> ("dom-node"));
+ default_values_.push_back (&copy);
+ }
+
+ // Transfer annotation.
+ //
+ if (e.annotated_p ())
+ s_.new_edge<Annotates> (e.annotation (), copy);
+
+ // Belongs edge.
+ //
+ if (e.typed_p ())
+ s_.new_edge<Belongs> (copy, e.type ());
+ else
+ assert (!valid_);
+
+ // Substitutes edge.
+ //
+ if (e.substitutes_p ())
+ s_.new_edge<Substitutes> (copy, e.substitutes ().root ());
+
+ return copy;
+ }
+
+ void
+ clone_attribute_group_content (AttributeGroupRef& ref,
+ SemanticGraph::Scope& s)
+ {
+ using SemanticGraph::Scope;
+ using SemanticGraph::Attribute;
+ using SemanticGraph::AttributeGroup;
+
+ try
+ {
+ AttributeGroup& g (
+ resolve<AttributeGroup> (ref.ns_name, ref.uq_name, s_, cache_));
+
+ // Make sure the group and all its content are fully resolved.
+ //
+ traverse (g);
+
+ Scope::NamesIterator pos (ref.names_pos);
+
+ for (Scope::NamesIterator i (g.names_begin ());
+ i != g.names_end (); ++i)
+ {
+ if (Attribute* p = dynamic_cast<Attribute*> (&i->named ()))
+ {
+ Attribute& a (
+ s_.new_node<Attribute> (p->file (),
+ p->line (),
+ p->column (),
+ p->optional_p (),
+ p->global_p (),
+ p->qualified_p ()));
+
+ NodeArgs<Scope, Scope::NamesIterator> na (s, pos);
+ s_.new_edge<Names> (na, a, p->name ());
+ ++pos;
+
+ if (p->qualified_p ())
+ s_.new_edge<BelongsToNamespace> (a, p->namespace_ ());
+
+ // Transfer default and fixed values if any.
+ //
+ if (p->fixed_p ())
+ a.fixed (p->value ());
+ else if (p->default_p ())
+ a.default_ (p->value ());
+
+ if (a.default_p ())
+ {
+ a.context ().set (
+ "dom-node",
+ p->context ().get<Xerces::DOMElement*> ("dom-node"));
+ default_values_.push_back (&a);
+ }
+
+ // Transfer annotation.
+ //
+ if (p->annotated_p ())
+ s_.new_edge<Annotates> (p->annotation (), a);
+
+ // Belongs edge.
+ //
+ if (p->typed_p ())
+ s_.new_edge<Belongs> (a, p->type ());
+ else
+ assert (!valid_);
+ }
+ else if (
+ AnyAttribute* p = dynamic_cast<AnyAttribute*> (&i->named ()))
+ {
+ AnyAttribute& any (
+ s_.new_node<AnyAttribute> (p->file (),
+ p->line (),
+ p->column (),
+ p->namespace_begin (),
+ p->namespace_end ()));
+
+ any.prototype (*p);
+
+ // Transfer annotation.
+ //
+ if (p->annotated_p ())
+ s_.new_edge<Annotates> (p->annotation (), any);
+
+ // AnyAttribute has no name so we have to come up with a fake
+ // one in order to put it into the scope. Note that we cannot
+ // reuse the name from the attribute group.
+
+ unsigned long count;
+ SemanticGraph::Context& ctx (s.context ());
+
+ if (!ctx.count ("any-attribute-name-count"))
+ {
+ count = 0;
+ ctx.set ("any-attribute-name-count", count);
+ }
+ else
+ count = ++(ctx.get<unsigned long> ("any-attribute-name-count"));
+
+ std::basic_ostringstream<wchar_t> os;
+ os << "any-attribute #" << count;
+
+ NodeArgs<Scope, Scope::NamesIterator> na (s, pos);
+ s_.new_edge<Names> (na, any, os.str ());
+ ++pos;
+ }
+ }
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+ abort ();
+ }
+ }
+ catch (NotName const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << "ice: unable to resolve attribute group name '"
+ << ex.name () << "' inside namespace '" << ex.ns () << "'"
+ << endl;
+ abort ();
+ }
+ }
+ }
+
+ private:
+ Schema& s_;
+ bool& valid_;
+ NamespaceMap& cache_;
+ DefaultValues& default_values_;
+
+ private:
+ //Traversal::ContainsParticle contains_particle;
+ Traversal::ContainsCompositor contains_compositor;
+ };
+ }
+
+ // Parser::Impl
+ //
+
+ class Parser::Impl
+ {
+ Impl (Impl const&);
+ Impl& operator= (Impl const&);
+
+ public:
+ ~Impl ();
+
+ Impl (bool proper_restriction,
+ bool multiple_imports,
+ bool full_schema_check,
+ LocationTranslator*,
+ const WarningSet*);
+
+ auto_ptr<Schema>
+ parse (Path const&);
+
+ auto_ptr<Schema>
+ parse (Paths const&);
+
+ auto_ptr<Schema>
+ xml_schema (Path const&);
+
+ private:
+ void
+ fill_xml_schema (Schema&, Path const&);
+
+ private:
+ XML::AutoPtr<Xerces::DOMDocument>
+ dom (SemanticGraph::Path const&, bool validate);
+
+ void
+ schema (XML::Element const&);
+
+ SemanticGraph::Annotation*
+ annotation (bool process);
+
+ void
+ import (XML::Element const&);
+
+ void
+ include (XML::Element const&);
+
+ void
+ element_group (XML::Element const&, bool in_compositor);
+
+ SemanticGraph::Type*
+ simple_type (XML::Element const&);
+
+ SemanticGraph::Type*
+ list (XML::Element const& l, XML::Element const& type);
+
+ SemanticGraph::Type*
+ union_ (XML::Element const& u, XML::Element const& type);
+
+ SemanticGraph::Type*
+ restriction (XML::Element const& r, XML::Element const& type);
+
+ void
+ enumeration (XML::Element const&);
+
+ SemanticGraph::Type*
+ complex_type (XML::Element const&);
+
+ All*
+ all (XML::Element const&);
+
+ Choice*
+ choice (XML::Element const&, bool in_compositor);
+
+ Sequence*
+ sequence (XML::Element const&, bool in_compositor);
+
+ void
+ simple_content (XML::Element const&);
+
+ void
+ complex_content (XML::Element const&, Complex&);
+
+ void
+ simple_content_extension (XML::Element const&);
+
+ void
+ simple_content_restriction (XML::Element const&);
+
+ void
+ complex_content_extension (XML::Element const&, Complex&);
+
+ void
+ complex_content_restriction (XML::Element const&, Complex&);
+
+ void
+ element (XML::Element const&, bool global);
+
+ void
+ attribute (XML::Element const&, bool global);
+
+ void
+ attribute_group (XML::Element const&);
+
+ void
+ any (XML::Element const&);
+
+ void
+ any_attribute (XML::Element const&);
+
+ private:
+ bool
+ is_disabled (char const* warning)
+ {
+ return disabled_warnings_all_ ||
+ (disabled_warnings_ &&
+ disabled_warnings_->find (warning) != disabled_warnings_->end ());
+ }
+
+ private:
+ bool
+ more () const
+ {
+ iterator const& it (iteration_state_.top ());
+
+ return it.l_->getLength () > it.i_;
+ }
+
+ XML::Element
+ next ()
+ {
+ iterator& it (iteration_state_.top ());
+
+ return XML::Element (
+ dynamic_cast<Xerces::DOMElement*> (it.l_->item (it.i_++)));
+ }
+
+ void
+ prev ()
+ {
+ iterator& it (iteration_state_.top ());
+
+ if (it.i_)
+ --it.i_;
+ }
+
+ void
+ push (XML::Element const& e)
+ {
+ iteration_state_.push (e.dom_element ());
+ }
+
+ void
+ pop ()
+ {
+ iteration_state_.pop ();
+ }
+
+ private:
+ void
+ push_scope (SemanticGraph::Scope& s)
+ {
+ scope_stack_.push (&s);
+ }
+
+ void
+ pop_scope ()
+ {
+ scope_stack_.pop ();
+ }
+
+ SemanticGraph::Scope&
+ scope () const
+ {
+ return *(scope_stack_.top ());
+ }
+
+ private:
+ void
+ push_compositor (SemanticGraph::Compositor& c)
+ {
+ compositor_stack_.push (&c);
+ }
+
+ void
+ pop_compositor ()
+ {
+ assert (!compositor_stack_.empty ());
+ compositor_stack_.pop ();
+ }
+
+ SemanticGraph::Compositor&
+ compositor () const
+ {
+ assert (!compositor_stack_.empty ());
+ return *(compositor_stack_.top ());
+ }
+
+ private:
+ static unsigned long const unbounded = ~static_cast<unsigned long> (0);
+
+ unsigned long
+ parse_min (String const& m)
+ {
+ if (m.empty ())
+ return 1;
+
+ unsigned long v;
+ std::basic_istringstream<wchar_t> is (m);
+
+ is >> v;
+ return v;
+ }
+
+ unsigned long
+ parse_max (String const& m)
+ {
+ if (m.empty ())
+ return 1;
+
+ if (m == L"unbounded")
+ return unbounded;
+
+ unsigned long v;
+ std::basic_istringstream<wchar_t> is (m);
+
+ is >> v;
+ return v;
+ }
+
+ private:
+ SemanticGraph::Namespace&
+ cur_ns () const
+ {
+ // Here I am using the fact that each Schema Names only one
+ // Namespace.
+ //
+ return dynamic_cast<Namespace&> (cur_->names_begin ()->named ());
+ }
+
+ private:
+ String
+ unqualified_name (String const& n)
+ {
+ return XML::uq_name (n);
+ }
+
+ String
+ namespace_name (XML::Element const& e, String const& n)
+ {
+ try
+ {
+ String p (XML::prefix (n));
+
+ // If we are currently handling a chameleon-included schema then
+ // the empty prefix is logically translated into acquired target
+ // namespace.
+ //
+ if (cur_chameleon_ && p.empty ())
+ return cur_ns ().name ();
+
+ // We have to try to resolve even the empty prefix since it can
+ // be assigned to a namespace (which takes precedence over names
+ // without a namespace).
+ //
+ return XML::ns_name (e.dom_element (), p);
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ if (ex.prefix ().empty ())
+ return String ();
+ else
+ throw;
+ }
+ }
+
+ SemanticGraph::Type&
+ ultimate_base (SemanticGraph::Type& t)
+ {
+ using namespace SemanticGraph;
+
+ Complex* c = dynamic_cast<Complex*> (&t);
+
+ if (c != 0 && c->inherits_p ())
+ {
+ Type* b (&c->inherits ().base ());
+
+ while (true)
+ {
+ Complex* cb (dynamic_cast<Complex*> (b));
+
+ if (cb != 0 && cb->inherits_p ())
+ {
+ b = &cb->inherits ().base ();
+ continue;
+ }
+
+ break;
+ }
+
+ return *b;
+ }
+ else
+ return t;
+ }
+
+ private:
+ template <typename Edge, typename Node>
+ Edge*
+ set_type (String const& type, XML::Element const& e, Node& node);
+
+ private:
+ XML::PtrVector<Xerces::DOMDocument>* dom_docs_;
+
+ struct iterator
+ {
+ iterator (Xerces::DOMElement* e)
+ : l_ (e->getChildNodes ()), i_ (0)
+ {
+ }
+
+ Xerces::DOMNodeList* l_;
+ size_t i_;
+ };
+
+ std::stack<iterator> iteration_state_;
+ SemanticGraph::Schema* s_; // root schema file
+ SemanticGraph::Schema* cur_; // current schema file
+ bool cur_chameleon_; // whethere cur_ is chameleon
+
+ SemanticGraph::Schema* xml_schema_; // XML Schema file
+ SemanticGraph::Path xml_schema_path_;
+
+ //
+ //
+ std::stack<SemanticGraph::Scope*> scope_stack_;
+
+ //
+ //
+ std::stack<SemanticGraph::Compositor*> compositor_stack_;
+
+
+ // Map of absolute file path and namespace pair to a Schema node.
+ //
+ struct SchemaId
+ {
+ SchemaId (SemanticGraph::Path const& path, String const& ns)
+ : path_ (path), ns_ (ns)
+ {
+ }
+
+
+ friend bool
+ operator< (SchemaId const& x, SchemaId const& y)
+ {
+ return x.path_ < y.path_ || (x.path_ == y.path_ && x.ns_ < y.ns_);
+ }
+
+ private:
+ SemanticGraph::Path path_;
+ String ns_;
+ };
+
+
+ typedef std::map<SchemaId, SemanticGraph::Schema*> SchemaMap;
+ SchemaMap schema_map_;
+
+ // Path stack for diagnostic.
+ //
+ struct PathPair
+ {
+ PathPair (SemanticGraph::Path const& r, SemanticGraph::Path const& a)
+ : rel (r), abs (a)
+ {
+ }
+
+ SemanticGraph::Path rel, abs;
+ };
+
+ std::stack<PathPair> file_stack_;
+
+ SemanticGraph::Path const&
+ file ()
+ {
+ return file_stack_.top ().rel;
+ }
+
+ SemanticGraph::Path const&
+ abs_file ()
+ {
+ return file_stack_.top ().abs;
+ }
+
+ // Members with default/fixed values (needed for QName handling).
+ //
+ DefaultValues default_values_;
+
+ private:
+ bool qualify_attribute_;
+ bool qualify_element_;
+
+ bool valid_;
+
+ bool proper_restriction_;
+ bool multiple_imports_;
+ bool full_schema_check_;
+ LocationTranslator* loc_translator_;
+ const WarningSet* disabled_warnings_;
+ bool disabled_warnings_all_;
+
+ NamespaceMap* cache_;
+ };
+
+
+ Parser::Impl::
+ Impl (bool proper_restriction,
+ bool multiple_imports,
+ bool full_schema_check,
+ LocationTranslator* t,
+ const WarningSet* dw)
+ : s_ (0),
+ cur_ (0),
+ cur_chameleon_ (false),
+ xml_schema_path_ ("XMLSchema.xsd"),
+ qualify_attribute_ (false),
+ qualify_element_ (false),
+ proper_restriction_ (proper_restriction),
+ multiple_imports_ (multiple_imports),
+ full_schema_check_ (full_schema_check),
+ loc_translator_ (t),
+ disabled_warnings_ (dw),
+ disabled_warnings_all_ (false)
+ {
+ if (dw && dw->find ("all") != dw->end ())
+ disabled_warnings_all_ = true;
+
+ // Initialize the Xerces-C++ runtime.
+ //
+ Xerces::XMLPlatformUtils::Initialize ();
+ }
+
+ Parser::Impl::
+ ~Impl ()
+ {
+ // Terminate the Xerces-C++ runtime.
+ //
+ Xerces::XMLPlatformUtils::Terminate ();
+ }
+
+ template<typename T> T&
+ add_type (Schema& s, Namespace& ns, String name)
+ {
+ Path path ("XMLSchema.xsd");
+ T& node (s.new_node<T> (path, 0, 0));
+ s.new_edge<Names> (ns, node, name);
+
+ return node;
+ }
+
+ void Parser::Impl::
+ fill_xml_schema (Schema& s, Path const& path)
+ {
+ Namespace& ns (s.new_node<Namespace> (path, 1, 1));
+ s.new_edge<Names> (s, ns, xsd);
+
+ // anyType and & anySimpleType
+ //
+ AnyType& any_type (
+ add_type<AnyType> (s, ns, L"anyType"));
+ add_type<AnySimpleType> (s, ns, L"anySimpleType");
+
+ // Integers.
+ //
+ add_type<Fundamental::Byte> (s, ns, L"byte");
+ add_type<Fundamental::UnsignedByte> (s, ns, L"unsignedByte");
+ add_type<Fundamental::Short> (s, ns, L"short");
+ add_type<Fundamental::UnsignedShort> (s, ns, L"unsignedShort");
+ add_type<Fundamental::Int> (s, ns, L"int");
+ add_type<Fundamental::UnsignedInt> (s, ns, L"unsignedInt");
+ add_type<Fundamental::Long> (s, ns, L"long");
+ add_type<Fundamental::UnsignedLong> (s, ns, L"unsignedLong");
+ add_type<Fundamental::Integer> (s, ns, L"integer");
+ add_type<Fundamental::NonPositiveInteger> (s, ns, L"nonPositiveInteger");
+ add_type<Fundamental::NonNegativeInteger> (s, ns, L"nonNegativeInteger");
+ add_type<Fundamental::PositiveInteger> (s, ns, L"positiveInteger");
+ add_type<Fundamental::NegativeInteger> (s, ns, L"negativeInteger");
+
+ // Boolean.
+ //
+ add_type<Fundamental::Boolean> (s, ns, L"boolean");
+
+ // Floats.
+ //
+ add_type<Fundamental::Float> (s, ns, L"float");
+ add_type<Fundamental::Double> (s, ns, L"double");
+ add_type<Fundamental::Decimal> (s, ns, L"decimal");
+
+ // Strings
+ //
+ add_type<Fundamental::String> (s, ns, L"string");
+ add_type<Fundamental::NormalizedString> (s, ns, L"normalizedString");
+ add_type<Fundamental::Token> (s, ns, L"token");
+ add_type<Fundamental::Name> (s, ns, L"Name");
+ add_type<Fundamental::NameToken> (s, ns, L"NMTOKEN");
+ add_type<Fundamental::NameTokens> (s, ns, L"NMTOKENS");
+ add_type<Fundamental::NCName> (s, ns, L"NCName");
+ add_type<Fundamental::Language> (s, ns, L"language");
+
+ // ID/IDREF.
+ //
+ add_type<Fundamental::Id> (s, ns, L"ID");
+
+ Fundamental::IdRef& id_ref (
+ s.new_node<Fundamental::IdRef> (path, 0, 0));
+ s.new_edge<Names> (ns, id_ref, L"IDREF");
+ s.new_edge<Arguments> (any_type, id_ref);
+
+ Fundamental::IdRefs& id_refs (
+ s.new_node<Fundamental::IdRefs> (path, 0, 0));
+ s.new_edge<Names> (ns, id_refs, L"IDREFS");
+ s.new_edge<Arguments> (any_type, id_refs);
+
+ // URI.
+ //
+ add_type<Fundamental::AnyURI> (s, ns, L"anyURI");
+
+ // Qualified name.
+ //
+ add_type<Fundamental::QName> (s, ns, L"QName");
+
+ // Binary.
+ //
+ add_type<Fundamental::Base64Binary> (s, ns, L"base64Binary");
+ add_type<Fundamental::HexBinary> (s, ns, L"hexBinary");
+
+ // Date/time.
+ //
+ add_type<Fundamental::Date> (s, ns, L"date");
+ add_type<Fundamental::DateTime> (s, ns, L"dateTime");
+ add_type<Fundamental::Duration> (s, ns, L"duration");
+ add_type<Fundamental::Day> (s, ns, L"gDay");
+ add_type<Fundamental::Month> (s, ns, L"gMonth");
+ add_type<Fundamental::MonthDay> (s, ns, L"gMonthDay");
+ add_type<Fundamental::Year> (s, ns, L"gYear");
+ add_type<Fundamental::YearMonth> (s, ns, L"gYearMonth");
+ add_type<Fundamental::Time> (s, ns, L"time");
+
+ // Entity.
+ //
+ add_type<Fundamental::Entity> (s, ns, L"ENTITY");
+ add_type<Fundamental::Entities> (s, ns, L"ENTITIES");
+
+ // Notation.
+ //
+ add_type<Fundamental::Notation> (s, ns, L"NOTATION");
+ }
+
+
+ auto_ptr<Schema> Parser::Impl::
+ xml_schema (Path const& tu)
+ {
+ valid_ = true;
+
+ auto_ptr<Schema> rs (new Schema (tu, 1, 1));
+ fill_xml_schema (*rs, tu);
+
+ if (!valid_)
+ throw InvalidSchema ();
+
+ return rs;
+ }
+ auto_ptr<Schema> Parser::Impl::
+ parse (Path const& tu)
+ {
+ valid_ = true;
+ schema_map_.clear ();
+ default_values_.clear ();
+
+ XML::PtrVector<Xerces::DOMDocument> dom_docs;
+ dom_docs_ = &dom_docs;
+
+ NamespaceMap cache;
+ cache_ = &cache;
+
+ XML::AutoPtr<Xerces::DOMDocument> d (dom (tu, true));
+
+ if (!d)
+ throw InvalidSchema ();
+
+ XML::Element root (d->getDocumentElement ());
+ String ns (trim (root["targetNamespace"]));
+
+ if (trace_)
+ wcout << "target namespace: " << ns << endl;
+
+ auto_ptr<Schema> rs (new Schema (tu, root.line (), root.column ()));
+
+ // Implied schema with fundamental types.
+ //
+ xml_schema_ = &rs->new_node<Schema> (xml_schema_path_, 1, 1);
+ rs->new_edge<Implies> (*rs, *xml_schema_, xml_schema_path_);
+
+ fill_xml_schema (*xml_schema_, xml_schema_path_);
+
+ // Parse.
+ //
+ {
+ // Enter the file into schema_map_. Do normalize() before
+ // complete() to avoid hitting system path limits with '..'
+ // directories.
+ //
+ Path abs_path (tu);
+ abs_path.normalize ().complete ();
+ schema_map_[SchemaId (abs_path, ns)] = rs.get ();
+ rs->context ().set ("absolute-path", abs_path);
+
+ s_ = cur_ = rs.get ();
+ {
+ file_stack_.push (PathPair (tu, abs_path));
+
+ {
+ push_scope (
+ s_->new_node<Namespace> (
+ file (), root.line (), root.column ()));
+ s_->new_edge<Names> (*cur_, scope (), ns);
+
+ {
+ schema (root);
+ }
+
+ pop_scope ();
+ }
+
+ file_stack_.pop ();
+ }
+
+ s_ = cur_ = 0;
+ }
+
+ dom_docs_->push_back (d);
+
+ // Second pass to resolve forward references to types, elements,
+ // attributes and groups.
+ //
+ if (valid_)
+ {
+ Traversal::Schema schema;
+
+ struct Uses: Traversal::Uses
+ {
+ virtual void
+ traverse (Type& u)
+ {
+ Schema& s (u.schema ());
+
+ if (!s.context ().count ("schema-resolved"))
+ {
+ s.context ().set ("schema-resolved", true);
+ Traversal::Uses::traverse (u);
+ }
+ }
+ } uses;
+
+ Traversal::Names schema_names;
+ Traversal::Namespace ns;
+ Traversal::Names ns_names;
+
+ schema >> uses >> schema;
+ schema >> schema_names >> ns >> ns_names;
+
+ Resolver resolver (*rs, valid_, *cache_, default_values_);
+
+ struct AnonymousMember: Traversal::Attribute,
+ Traversal::Element,
+ Traversal::Member
+ {
+ AnonymousMember (Traversal::NodeDispatcher& d)
+ {
+ belongs_.node_traverser (d);
+ }
+
+ virtual void
+ traverse (SemanticGraph::Attribute& a)
+ {
+ traverse_member (a);
+ }
+
+ virtual void
+ traverse (SemanticGraph::Element& e)
+ {
+ traverse_member (e);
+ }
+
+ void
+ traverse_member (SemanticGraph::Member& m)
+ {
+ if (m.typed_p () &&
+ !m.type ().named_p () &&
+ !m.type ().context ().count ("seen"))
+ {
+ m.type().context ().set ("seen", true);
+
+ Traversal::Member::belongs (m, belongs_);
+
+ m.type ().context ().remove ("seen");
+ }
+ }
+
+ private:
+ Traversal::Belongs belongs_;
+ } anonymous_member (resolver);
+
+ struct AnonymousBase: Traversal::Type
+ {
+ AnonymousBase (Traversal::NodeDispatcher& d)
+ : base_ (d)
+ {
+ }
+
+ virtual void
+ traverse (SemanticGraph::Type& t)
+ {
+ if (!t.named_p ())
+ base_.dispatch (t);
+ }
+
+ private:
+ Traversal::NodeDispatcher& base_;
+ } anonymous_base (resolver);
+
+ ns_names >> resolver;
+ ns_names >> anonymous_member;
+
+ Traversal::Names names;
+ Traversal::Inherits inherits;
+ Traversal::Argumented argumented;
+ resolver >> names >> resolver;
+ names >> anonymous_member;
+ resolver >> inherits >> anonymous_base;
+ resolver >> argumented >> anonymous_base;
+
+ if (trace_)
+ wcout << "starting resolution pass" << endl;
+
+ schema.dispatch (*rs);
+ }
+
+ // Resolve default/fixed values of QName type.
+ //
+ if (valid_)
+ {
+ for (DefaultValues::const_iterator i (default_values_.begin ()),
+ e (default_values_.end ()); i != e; ++i)
+ {
+ SemanticGraph::Member& m (**i);
+ SemanticGraph::Type& t (m.type ());
+ SemanticGraph::Context& c (m.context ());
+
+ if (ultimate_base (t).is_a<SemanticGraph::Fundamental::QName> ())
+ {
+ String v (m.value ());
+ Xerces::DOMElement* e (c.get<Xerces::DOMElement*> ("dom-node"));
+
+ try
+ {
+ // We have to try to resolve even the empty prefix since it can
+ // be assigned to a namespace (which takes precedence over names
+ // without a namespace).
+ //
+ String ns (XML::ns_name (e, XML::prefix (v)));
+
+ if (m.fixed_p ())
+ m.fixed (ns + L'#' + v);
+ else
+ m.default_ (ns + L'#' + v);
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ if (!ex.prefix ().empty ())
+ {
+ wcerr << m.file () << ":" << m.line () << ":" << m.column ()
+ << ": error: unable to resolve namespace for prefix '"
+ << ex.prefix () << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+
+ c.remove ("dom-node");
+ }
+ }
+
+ if (!valid_)
+ throw InvalidSchema ();
+
+ return rs;
+ }
+
+ auto_ptr<Schema> Parser::Impl::
+ parse (Paths const& paths)
+ {
+ valid_ = true;
+ schema_map_.clear ();
+ default_values_.clear ();
+
+ XML::PtrVector<Xerces::DOMDocument> dom_docs;
+ dom_docs_ = &dom_docs;
+
+ NamespaceMap cache;
+ cache_ = &cache;
+
+ auto_ptr<Schema> rs (new Schema (Path (), 0, 0));
+
+ // Implied schema with fundamental types.
+ //
+ xml_schema_ = &rs->new_node<Schema> (xml_schema_path_, 1, 1);
+ rs->new_edge<Implies> (*rs, *xml_schema_, xml_schema_path_);
+
+ fill_xml_schema (*xml_schema_, xml_schema_path_);
+
+ // Parse individual schemas.
+ //
+ s_ = rs.get ();
+
+ for (Paths::const_iterator i (paths.begin ()); i != paths.end (); ++i)
+ {
+ Path const& tu (*i);
+ XML::AutoPtr<Xerces::DOMDocument> d (dom (tu, true));
+
+ if (!d)
+ throw InvalidSchema ();
+
+ XML::Element root (d->getDocumentElement ());
+ String ns (trim (root["targetNamespace"]));
+
+ if (trace_)
+ wcout << "target namespace: " << ns << endl;
+
+ // Check if we already have this schema. Do normalize() before
+ // complete() to avoid hitting system path limits with '..'
+ // directories.
+ //
+ Path abs_path (tu);
+ abs_path.normalize ().complete ();
+ SchemaId schema_id (abs_path, ns);
+
+ if (schema_map_.find (schema_id) != schema_map_.end ())
+ continue;
+
+ Schema& s (s_->new_node<Schema> (tu, root.line (), root.column ()));
+ s_->new_edge<Implies> (s, *xml_schema_, xml_schema_path_);
+ s_->new_edge<Imports> (*s_, s, tu);
+
+ // Enter the file into schema_map_.
+ //
+ schema_map_[schema_id] = &s;
+ s.context ().set ("absolute-path", abs_path);
+
+ cur_ = &s;
+
+ {
+ file_stack_.push (PathPair (tu, abs_path));
+
+ {
+ push_scope (
+ s_->new_node<Namespace> (
+ file (), root.line (), root.column ()));
+ s_->new_edge<Names> (*cur_, scope (), ns);
+
+ {
+ schema (root);
+ }
+
+ pop_scope ();
+ }
+
+ file_stack_.pop ();
+ }
+
+ cur_ = 0;
+
+ dom_docs_->push_back (d);
+
+ if (!valid_)
+ break;
+ }
+
+ s_ = 0;
+
+ // Second pass to resolve forward references to types, elements,
+ // attributes and groups.
+ //
+ if (valid_)
+ {
+ Traversal::Schema schema;
+
+ struct Uses: Traversal::Uses
+ {
+ virtual void
+ traverse (Type& u)
+ {
+ Schema& s (u.schema ());
+
+ if (!s.context ().count ("schema-resolved"))
+ {
+ s.context ().set ("schema-resolved", true);
+ Traversal::Uses::traverse (u);
+ }
+ }
+ } uses;
+
+ Traversal::Names schema_names;
+ Traversal::Namespace ns;
+ Traversal::Names ns_names;
+
+ schema >> uses >> schema;
+ schema >> schema_names >> ns >> ns_names;
+
+ Resolver resolver (*rs, valid_, *cache_, default_values_);
+
+ struct AnonymousMember: Traversal::Attribute,
+ Traversal::Element,
+ Traversal::Member
+ {
+ AnonymousMember (Traversal::NodeDispatcher& d)
+ {
+ belongs_.node_traverser (d);
+ }
+
+ virtual void
+ traverse (SemanticGraph::Attribute& a)
+ {
+ traverse_member (a);
+ }
+
+ virtual void
+ traverse (SemanticGraph::Element& e)
+ {
+ traverse_member (e);
+ }
+
+ virtual void
+ traverse_member (SemanticGraph::Member& m)
+ {
+ if (m.typed_p () &&
+ !m.type ().named_p () &&
+ !m.type ().context ().count ("seen"))
+ {
+ m.type().context ().set ("seen", true);
+
+ Traversal::Member::belongs (m, belongs_);
+
+ m.type ().context ().remove ("seen");
+ }
+ }
+
+ private:
+ Traversal::Belongs belongs_;
+ } anonymous_member (resolver);
+
+ struct AnonymousBase: Traversal::Type
+ {
+ AnonymousBase (Traversal::NodeDispatcher& d)
+ : base_ (d)
+ {
+ }
+
+ virtual void
+ traverse (SemanticGraph::Type& t)
+ {
+ if (!t.named_p ())
+ base_.dispatch (t);
+ }
+
+ private:
+ Traversal::NodeDispatcher& base_;
+ } anonymous_base (resolver);
+
+ ns_names >> resolver;
+ ns_names >> anonymous_member;
+
+ Traversal::Names names;
+ Traversal::Inherits inherits;
+ Traversal::Argumented argumented;
+ resolver >> names >> resolver;
+ names >> anonymous_member;
+ resolver >> inherits >> anonymous_base;
+ resolver >> argumented >> anonymous_base;
+
+ if (trace_)
+ wcout << "starting resolution pass" << endl;
+
+ schema.dispatch (*rs);
+ }
+
+ // Resolve default/fixed values of QName type.
+ //
+ if (valid_)
+ {
+ for (DefaultValues::const_iterator i (default_values_.begin ()),
+ e (default_values_.end ()); i != e; ++i)
+ {
+ SemanticGraph::Member& m (**i);
+ SemanticGraph::Type& t (m.type ());
+ SemanticGraph::Context& c (m.context ());
+
+ if (ultimate_base (t).is_a<SemanticGraph::Fundamental::QName> ())
+ {
+ String v (m.value ());
+ Xerces::DOMElement* e (c.get<Xerces::DOMElement*> ("dom-node"));
+
+ try
+ {
+ // We have to try to resolve even the empty prefix since it can
+ // be assigned to a namespace (which takes precedence over names
+ // without a namespace).
+ //
+ String ns (XML::ns_name (e, XML::prefix (v)));
+
+ if (m.fixed_p ())
+ m.fixed (ns + L'#' + v);
+ else
+ m.default_ (ns + L'#' + v);
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ if (!ex.prefix ().empty ())
+ {
+ wcerr << m.file () << ":" << m.line () << ":" << m.column ()
+ << ": error: unable to resolve namespace for prefix '"
+ << ex.prefix () << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+
+ c.remove ("dom-node");
+ }
+ }
+
+ if (!valid_)
+ throw InvalidSchema ();
+
+ return rs;
+ }
+
+ void Parser::Impl::
+ schema (XML::Element const& s)
+ {
+ bool old_qa (qualify_attribute_);
+ bool old_qe (qualify_element_);
+
+ if (String af = trim (s["attributeFormDefault"]))
+ qualify_attribute_ = af == L"qualified";
+ else
+ qualify_attribute_ = false;
+
+ if (String ef = trim (s["elementFormDefault"]))
+ qualify_element_ = ef == L"qualified";
+ else
+ qualify_element_ = false;
+
+ push (s);
+
+ // Parse leading annotation if any and add it as an annotation for
+ // this schema.
+ //
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, *cur_);
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ if (name == L"import") import (e); else
+ if (name == L"include") include (e); else
+ if (name == L"element") element (e, true); else
+ if (name == L"attribute") attribute (e, true); else
+ if (name == L"simpleType") simple_type (e); else
+ if (name == L"annotation"); else
+ if (name == L"complexType") complex_type (e); else
+ if (name == L"group") element_group (e, false); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unexpected top-level element: '" << name << "'"
+ << endl;
+
+ valid_ = false;
+ }
+ }
+
+ pop ();
+
+ qualify_attribute_ = old_qa;
+ qualify_element_ = old_qe;
+ }
+
+ void Parser::Impl::
+ import (XML::Element const& i)
+ {
+ NarrowString loc (
+ trim (
+ XML::transcode_to_narrow (
+ i.dom_element ()->getAttribute (
+ XML::XMLChString ("schemaLocation").c_str ()))));
+
+ if (loc_translator_)
+ loc = loc_translator_->translate (loc);
+
+ String ins (trim (i["namespace"]));
+
+ // Ignore empty <import>.
+ //
+ if (!loc && !ins)
+ return;
+
+ Path path, rel_path, abs_path;
+ try
+ {
+ path = Path (loc);
+
+ if (path.absolute ())
+ {
+ abs_path = rel_path = path;
+ abs_path.normalize ();
+ }
+ else
+ {
+ // Do normalize() before complete() to avoid hitting system path
+ // limits with '..' directories.
+ //
+ abs_path = rel_path = file ().directory () / path;
+ abs_path.normalize ().complete ();
+ }
+ }
+ catch (InvalidPath const&)
+ {
+ wcerr << file () << ":" << i.line () << ":" << i.column () << ": "
+ << "error: '" << loc.c_str () << "' is not a valid "
+ << "filesystem path" << endl;
+
+ valid_ = false;
+ return;
+ }
+
+ SchemaId schema_id (abs_path, ins);
+
+ if (schema_map_.find (schema_id) != schema_map_.end ())
+ {
+ s_->new_edge<Imports> (*cur_, *schema_map_[schema_id], path);
+ return;
+ }
+
+ if (trace_)
+ wcout << "importing " << rel_path << endl;
+
+ if (XML::AutoPtr<Xerces::DOMDocument> d = dom (abs_path, false))
+ {
+ XML::Element r (d->getDocumentElement ());
+ String ns (trim (r["targetNamespace"]));
+
+ if (trace_)
+ wcout << "target namespace: " << ns << endl;
+
+ Schema& s (s_->new_node<Schema> (rel_path, r.line (), r.column ()));
+ s_->new_edge<Implies> (s, *xml_schema_, xml_schema_path_);
+ s_->new_edge<Imports> (*cur_, s, path);
+
+ schema_map_[schema_id] = &s;
+ s.context ().set ("absolute-path", abs_path);
+
+ Schema* old_cur (cur_);
+ bool old_cur_chameleon (cur_chameleon_);
+ cur_ = &s;
+ cur_chameleon_ = false;
+
+ {
+ file_stack_.push (PathPair (rel_path, abs_path));
+
+ {
+ push_scope (
+ s_->new_node<Namespace> (file (), r.line (), r.column ()));
+ s_->new_edge<Names> (*cur_, scope (), ns);
+
+ {
+ schema (r);
+ }
+
+ pop_scope ();
+ }
+
+ file_stack_.pop ();
+ }
+
+ cur_chameleon_ = old_cur_chameleon;
+ cur_ = old_cur;
+
+ dom_docs_->push_back (d);
+ }
+ }
+
+ void Parser::Impl::
+ include (XML::Element const& i)
+ {
+ NarrowString loc (
+ trim (
+ XML::transcode_to_narrow (
+ i.dom_element ()->getAttribute (
+ XML::XMLChString ("schemaLocation").c_str ()))));
+
+ if (loc_translator_)
+ loc = loc_translator_->translate (loc);
+
+ Path path, rel_path, abs_path;
+ try
+ {
+ path = Path (loc);
+
+ if (path.absolute ())
+ {
+ abs_path = rel_path = path;
+ abs_path.normalize ();
+ }
+ else
+ {
+ // Do normalize() before complete() to avoid hitting system path
+ // limits with '..' directories.
+ //
+ abs_path = rel_path = file ().directory () / path;
+ abs_path.normalize ().complete ();
+ }
+ }
+ catch (InvalidPath const&)
+ {
+ wcerr << file () << ":" << i.line () << ":" << i.column () << ": "
+ << "error: '" << loc.c_str () << "' is not a valid "
+ << "filesystem path" << endl;
+
+ valid_ = false;
+ return;
+ }
+
+ // Included schema should have the same namespace as ours.
+ //
+ SchemaId schema_id (abs_path, cur_ns ().name ());
+
+ if (schema_map_.find (schema_id) != schema_map_.end ())
+ {
+ Schema& s (*schema_map_[schema_id]);
+
+ // Chemeleon inclusion results in a new Schema node for every
+ // namespace. As a result, such a Schema node can only be
+ // Source'ed. I use this property to decide which edge to use.
+ //
+
+ if (s.used_p () && s.used_begin ()->is_a<Sources> ())
+ s_->new_edge<Sources> (*cur_, s, path);
+ else
+ s_->new_edge<Includes> (*cur_, s, path);
+
+ return;
+ }
+
+ if (trace_)
+ wcout << "including " << rel_path << endl;
+
+ if (XML::AutoPtr<Xerces::DOMDocument> d = dom (abs_path, false))
+ {
+ XML::Element r (d->getDocumentElement ());
+ String ns (trim (r["targetNamespace"])), cur_ns;
+
+ Schema& s (s_->new_node<Schema> (rel_path, r.line (), r.column ()));
+ s_->new_edge<Implies> (s, *xml_schema_, xml_schema_path_);
+
+ schema_map_[schema_id] = &s;
+ s.context ().set ("absolute-path", abs_path);
+
+ bool chameleon (false);
+
+ if (ns.empty () && !(cur_ns = (cur_->names_begin ())->name ()).empty ())
+ {
+ // Chameleon.
+ //
+ ns = cur_ns;
+ s_->new_edge<Sources> (*cur_, s, path);
+ chameleon = true;
+
+ if (trace_)
+ wcout << "handling chameleon schema" << endl;
+ }
+ else
+ s_->new_edge<Includes> (*cur_, s, path);
+
+ if (trace_)
+ wcout << "target namespace: " << ns << endl;
+
+ Schema* old_cur (cur_);
+ bool old_cur_chameleon (cur_chameleon_);
+ cur_ = &s;
+ cur_chameleon_ = chameleon;
+
+ {
+ file_stack_.push (PathPair (rel_path, abs_path));
+
+ {
+ push_scope (
+ s_->new_node<Namespace> (file (), r.line (), r.column ()));
+ s_->new_edge<Names> (*cur_, scope (), ns);
+
+ {
+ schema (r);
+ }
+
+ pop_scope ();
+ }
+
+ file_stack_.pop ();
+ }
+
+ cur_chameleon_ = old_cur_chameleon;
+ cur_ = old_cur;
+
+ dom_docs_->push_back (d);
+ }
+ }
+
+ void Parser::Impl::
+ element_group (XML::Element const& g, bool in_compositor)
+ {
+ if (String name = trim (g["name"]))
+ {
+ ElementGroup& group (
+ s_->new_node<ElementGroup> (file (), g.line (), g.column ()));
+
+ s_->new_edge<Names> (scope (), group, name);
+
+ push_scope (group);
+ push (g);
+
+ annotation (false);
+
+ XML::Element e (next ());
+
+ name = e.name ();
+
+ if (trace_)
+ wcout << name << endl;
+
+ Compositor* c (0);
+
+ if (name == L"all") c = all (e); else
+ if (name == L"choice") c = choice (e, false); else
+ if (name == L"sequence") c = sequence (e, false); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'all', 'choice' or 'sequence' "
+ << "instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+
+ // Group's immediate compositor always has cardinality 1,1.
+ //
+ if (c)
+ s_->new_edge<ContainsCompositor> (group, *c, 1, 1);
+
+ pop ();
+ pop_scope ();
+ }
+ else if (String ref = trim (g["ref"]))
+ {
+ if (trace_)
+ wcout << "element-group-ref " << ref << endl;
+
+ try
+ {
+ String uq_name (unqualified_name (ref));
+ String ns_name (namespace_name (g, ref));
+
+ // In order to avoid code duplication we are going to let the
+ // resolver handle this case.
+ //
+ if (trace_)
+ wcout << "deferring resolution of group name '" << uq_name
+ << "' inside namespace '" << ns_name << "'"
+ << " until later" << endl;
+
+ if (in_compositor)
+ {
+ Compositor& c (compositor ());
+
+ unsigned long min (parse_min (trim (g["minOccurs"])));
+ unsigned long max (parse_max (trim (g["maxOccurs"])));
+
+ ElementGroupRef ref (
+ uq_name, ns_name,
+ min,
+ max == unbounded ? 0 : max,
+ c,
+ scope ());
+
+ if (!c.context ().count ("element-group-refs"))
+ c.context ().set ("element-group-refs", ElementGroupRefs ());
+
+ c.context ().get<ElementGroupRefs> (
+ "element-group-refs").push_back (ref);
+ }
+ else
+ {
+ // This is a group-ref directly in complexType.
+ //
+
+ Scope& s (scope ());
+
+ unsigned long min (parse_min (trim (g["minOccurs"])));
+ unsigned long max (parse_max (trim (g["maxOccurs"])));
+
+ ElementGroupRef ref (
+ uq_name, ns_name, min, max == unbounded ? 0 : max, s);
+
+ s.context ().set ("element-group-ref", ref);
+ }
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << file () << ":" << g.line () << ":" << g.column () << ": "
+ << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+
+ abort ();
+ }
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ wcerr << file () << ":" << g.line () << ":" << g.column () << ": "
+ << "error: unable to resolve namespace prefix '" << ex.prefix ()
+ << "' in '" << ref << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ else
+ {
+ wcerr << file () << ":" << g.line () << ":" << g.column () << ": "
+ << "error: 'name' or 'ref' attribute is missing in group "
+ << "declaration" << endl;
+
+ valid_ = false;
+
+ return;
+ }
+ }
+
+ //@@ Need RAII for push/pop.
+ //
+
+ Type* Parser::Impl::
+ simple_type (XML::Element const& t)
+ {
+ Type* r (0);
+
+ push (t);
+
+ Annotation* a (annotation (true));
+
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (name == L"list") r = list (e, t); else
+ if (name == L"union") r = union_ (e, t); else
+ if (name == L"restriction") r = restriction (e, t); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'list', 'union', or 'restriction' "
+ << "instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (r != 0 && a != 0)
+ s_->new_edge<Annotates> (*a, *r);
+
+ pop ();
+
+ return r;
+ }
+
+ SemanticGraph::Type* Parser::Impl::
+ list (XML::Element const& l, XML::Element const& t)
+ {
+ if (trace_)
+ wcout << "list" << endl;
+
+ List& node (s_->new_node<List> (file (), t.line (), t.column ()));
+
+ if (String item_type = trim (l["itemType"]))
+ {
+ if (trace_)
+ wcout << "item type: " << fq_name (l, item_type) << endl;
+
+ set_type<Arguments> (item_type, l, node);
+ }
+ else
+ {
+ // Anonymous list item type.
+ //
+ push (l);
+
+ annotation (false);
+
+ if (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ Type* t (0);
+
+ if (name == L"simpleType") t = simple_type (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'simpleType' instead of "
+ << "'" << e.name () << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (t)
+ s_->new_edge<Arguments> (*t, node);
+ }
+ else
+ {
+ wcerr << file () << ":" << l.line () << ":" << l.column () << ": "
+ << "error: expected 'itemType' attribute or 'simpleType' "
+ << "nested element" << endl;
+
+ valid_ = false;
+ }
+
+ pop ();
+ }
+
+ if (String name = trim (t["name"]))
+ s_->new_edge<Names> (scope (), node, name);
+
+ return &node;
+ }
+
+ namespace
+ {
+ //
+ // List parsing utility functions.
+ //
+
+ // Find first non-space character.
+ //
+ size_t
+ find_ns (const wchar_t* s, size_t size, size_t pos)
+ {
+ while (pos < size &&
+ (s[pos] == 0x20 || // space
+ s[pos] == 0x0D || // carriage return
+ s[pos] == 0x09 || // tab
+ s[pos] == 0x0A))
+ ++pos;
+
+ return pos < size ? pos : String::npos;
+ }
+
+ // Find first space character.
+ //
+ size_t
+ find_s (const wchar_t* s, size_t size, size_t pos)
+ {
+ while (pos < size &&
+ s[pos] != 0x20 && // space
+ s[pos] != 0x0D && // carriage return
+ s[pos] != 0x09 && // tab
+ s[pos] != 0x0A)
+ ++pos;
+
+ return pos < size ? pos : String::npos;
+ }
+ }
+
+ SemanticGraph::Type* Parser::Impl::
+ union_ (XML::Element const& u, XML::Element const& t)
+ {
+ if (trace_)
+ wcout << "union" << endl;
+
+ Union& node (s_->new_node<Union> (file (), t.line (), t.column ()));
+
+ bool has_members (false);
+
+ if (String members = trim (u["memberTypes"]))
+ {
+ // Don't bother trying to resolve member types at this point
+ // since the order is important so we would have to insert
+ // the late resolutions into specific places. It is simpler
+ // to just do the whole resolution later.
+ //
+ const wchar_t* data (members.c_str ());
+ size_t size (members.size ());
+
+ UnionMemberTypes* m (0);
+
+ // Traverse the type list while logically collapsing spaces.
+ //
+ for (size_t i (find_ns (data, size, 0)); i != String::npos;)
+ {
+ String s;
+ size_t j (find_s (data, size, i));
+
+ if (j != String::npos)
+ {
+ s = String (data + i, j - i);
+ i = find_ns (data, size, j);
+ }
+ else
+ {
+ // Last item.
+ //
+ s = String (data + i, size - i);
+ i = String::npos;
+ }
+
+ if (trace_)
+ wcout << "member type: " << fq_name (u, s) << endl;
+
+ if (m == 0)
+ {
+ node.context ().set ("union-member-types", UnionMemberTypes ());
+ m = &node.context ().get<UnionMemberTypes> ("union-member-types");
+ }
+
+ try
+ {
+ m->push_back (
+ UnionMemberType (
+ namespace_name (u, s), unqualified_name (s)));
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ wcerr << file () << ":" << u.line () << ":" << u.column () << ": "
+ << "error: unable to resolve namespace prefix "
+ << "'" << ex.prefix () << "' in '" << s << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ has_members = (m != 0);
+ }
+
+ // Handle anonymous members.
+ //
+ push (u);
+
+ annotation (false);
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ Type* t (0);
+
+ if (name == L"simpleType") t = simple_type (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'simpleType' instead of "
+ << "'" << e.name () << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (t)
+ s_->new_edge<Arguments> (*t, node);
+ }
+
+ pop ();
+
+ if (node.argumented_begin () == node.argumented_end () && !has_members)
+ {
+ wcerr << file () << ":" << u.line () << ":" << u.column () << ": "
+ << "error: expected 'memberTypes' attribute or 'simpleType' "
+ << "nested element" << endl;
+
+ valid_ = false;
+ }
+
+ if (String name = trim (t["name"]))
+ s_->new_edge<Names> (scope (), node, name);
+
+ return &node;
+ }
+
+ Type* Parser::Impl::
+ restriction (XML::Element const& r, XML::Element const& t)
+ {
+ String base (trim (r["base"]));
+ Type* base_type (0);
+
+ if (base)
+ {
+ if (trace_)
+ wcout << "restriction base: " << fq_name (r, base) << endl;
+ }
+
+ Type* rv (0);
+
+ push (r);
+
+ annotation (false);
+
+ bool enum_ (false);
+
+ if (!base)
+ {
+ // Anonymous base type.
+ //
+ if (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ if (name == L"simpleType") base_type = simple_type (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'simpleType' instead of "
+ << "'" << e.name () << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ else
+ {
+ wcerr << file () << ":" << r.line () << ":" << r.column () << ": "
+ << "error: expected 'base' attribute or 'simpleType' "
+ << "nested element" << endl;
+
+ valid_ = false;
+ }
+
+ if (!valid_)
+ {
+ pop ();
+ return 0;
+ }
+ }
+
+ Facets facets;
+ Restricts* restricts (0);
+ String pattern;
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"enumeration")
+ {
+ // Enumeration
+ //
+ if (enum_)
+ enumeration (e);
+ else
+ {
+ // First
+ //
+ enum_ = true;
+
+ Enumeration& node (
+ s_->new_node<Enumeration> (file (), t.line (), t.column ()));
+
+ if (base_type)
+ restricts = &s_->new_edge<Restricts> (node, *base_type);
+ else
+ restricts = set_type<Restricts> (base, r, node);
+
+ if (String name = trim (t["name"]))
+ s_->new_edge<Names> (scope (), static_cast<Nameable&> (node), name);
+
+ rv = &node;
+ push_scope (node);
+ enumeration (e);
+ }
+ }
+ else if (name == L"minExclusive" ||
+ name == L"minInclusive" ||
+ name == L"maxExclusive" ||
+ name == L"maxInclusive" ||
+ name == L"totalDigits" ||
+ name == L"fractionDigits" ||
+ name == L"length" ||
+ name == L"minLength" ||
+ name == L"maxLength" ||
+ name == L"whiteSpace")
+ {
+ facets[name] = trim (e["value"]);
+ }
+ else if (name == L"pattern")
+ {
+ if (pattern)
+ pattern += L'|';
+
+ pattern += e["value"];
+ }
+ else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unexpected element '" << name << "' in "
+ << "simple type restriction" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ if (pattern)
+ facets[L"pattern"] = pattern;
+
+ if (enum_)
+ pop_scope ();
+ else
+ {
+ Complex& node (s_->new_node<Complex> (file (), t.line (), t.column ()));
+
+ if (base_type)
+ restricts = &s_->new_edge<Restricts> (node, *base_type);
+ else
+ restricts = set_type<Restricts> (base, r, node);
+
+ if (String name = trim (t["name"]))
+ s_->new_edge<Names> (scope (), node, name);
+
+ rv = &node;
+ }
+
+ if (!facets.empty ())
+ {
+ if (restricts)
+ copy_facets (*restricts, facets);
+ else
+ rv->context ().set ("facets", facets);
+ }
+
+ pop ();
+
+ return rv;
+ }
+
+ void Parser::Impl::
+ enumeration (XML::Element const& e)
+ {
+ String value (e["value"]);
+
+ if (trace_)
+ wcout << "enumeration value: " << value << endl;
+
+ push (e);
+ Annotation* a (annotation (true));
+ pop ();
+
+ Enumerator& node (
+ s_->new_node<Enumerator> (file (), e.line (), e.column ()));
+
+ s_->new_edge<Names> (scope (), node, value);
+ s_->new_edge<Belongs> (node, dynamic_cast<Type&>(scope ()));
+
+ if (a != 0)
+ s_->new_edge<Annotates> (*a, node);
+
+ }
+
+ Type* Parser::Impl::
+ complex_type (XML::Element const& t)
+ {
+ Type* r (0);
+
+ Complex& node (s_->new_node<Complex> (file (), t.line (), t.column ()));
+
+ if (String m = trim (t["mixed"]))
+ node.mixed_p (m == L"true" || m == L"1");
+
+ if (String name = trim (t["name"]))
+ s_->new_edge<Names> (scope (), node, name);
+
+ r = &node;
+
+ push_scope (node);
+ push (t);
+
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, node);
+
+ if (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ if (name == L"simpleContent") simple_content (e); else
+ if (name == L"complexContent") complex_content (e, node); else
+ {
+ Compositor* c (0);
+
+ if (name == L"all") c = all (e); else
+ if (name == L"choice") c = choice (e, false); else
+ if (name == L"sequence") c = sequence (e, false); else
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"group") element_group (e, false); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unexpected element '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (c)
+ {
+ unsigned long min (parse_min (trim (e["minOccurs"])));
+ unsigned long max (parse_max (trim (e["maxOccurs"])));
+
+ if (!(min == 0 && max == 0))
+ s_->new_edge<ContainsCompositor> (
+ node, *c, min, max == unbounded ? 0 : max);
+ }
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'attribute', 'anyAttribute', or "
+ << "'attributeGroup' instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+ }
+
+ pop ();
+ pop_scope ();
+
+ return r;
+ }
+
+ All* Parser::Impl::
+ all (XML::Element const& a)
+ {
+ // 'all' cannot be nested inside 'choice' or 'sequence', nor
+ // can it contain any of those. The only valid cardinality
+ // values for 'all' are min=0,1 and max=1.
+ //
+ All& node (s_->new_node<All> (file (), a.line (), a.column ()));
+
+ push_compositor (node);
+ push (a);
+
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, node);
+
+ while (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (name == L"element") element (e, false); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'element' "
+ << "instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ pop ();
+ pop_compositor ();
+
+ return &node;
+ }
+
+ Choice* Parser::Impl::
+ choice (XML::Element const& c, bool in_compositor)
+ {
+ Choice& node (s_->new_node<Choice> (file (), c.line (), c.column ()));
+
+ if (in_compositor)
+ {
+ unsigned long min (parse_min (trim (c["minOccurs"])));
+ unsigned long max (parse_max (trim (c["maxOccurs"])));
+
+ if (!(min == 0 && max == 0))
+ s_->new_edge<ContainsParticle> (
+ compositor (), node, min, max == unbounded ? 0 : max);
+ }
+
+ push_compositor (node);
+ push (c);
+
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, node);
+
+ while (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (name == L"any") any (e); else
+ if (name == L"choice") choice (e, true); else
+ if (name == L"element") element (e, false); else
+ if (name == L"sequence") sequence (e, true); else
+ if (name == L"group") element_group (e, true); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'any', 'group', 'choice', 'sequence', "
+ << "or 'element' instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ pop ();
+ pop_compositor ();
+
+ return &node;
+ }
+
+ Sequence* Parser::Impl::
+ sequence (XML::Element const& s, bool in_compositor)
+ {
+ Sequence& node (s_->new_node<Sequence> (file (), s.line (), s.column ()));
+
+ if (in_compositor)
+ {
+ unsigned long min (parse_min (trim (s["minOccurs"])));
+ unsigned long max (parse_max (trim (s["maxOccurs"])));
+
+ if (!(min == 0 && max == 0))
+ s_->new_edge<ContainsParticle> (
+ compositor (), node, min, max == unbounded ? 0 : max);
+ }
+
+ push_compositor (node);
+ push (s);
+
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, node);
+
+ while (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (name == L"any") any (e); else
+ if (name == L"choice") choice (e, true); else
+ if (name == L"element") element (e, false); else
+ if (name == L"sequence") sequence (e, true); else
+ if (name == L"group") element_group (e, true); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'any', 'group', 'choice', 'sequence', "
+ << "or 'element' instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ pop ();
+ pop_compositor ();
+
+ return &node;
+ }
+
+ void Parser::Impl::
+ simple_content (XML::Element const& c)
+ {
+ push (c);
+
+ annotation (false);
+
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"extension") simple_content_extension (e); else
+ if (name == L"restriction") simple_content_restriction (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'extension' or 'restriction' instead of "
+ << "'" << name << "'" << endl;
+
+ valid_ = false;
+ }
+
+ pop ();
+ }
+
+ void Parser::Impl::
+ complex_content (XML::Element const& c, Complex& type)
+ {
+ if (String m = trim (c["mixed"]))
+ type.mixed_p (m == L"true" || m == L"1");
+
+ push (c);
+
+ annotation (false);
+
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"extension") complex_content_extension (e, type); else
+ if (name == L"restriction") complex_content_restriction (e, type); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'extension' or 'restriction' instead of "
+ << "'" << name << "'" << endl;
+
+ valid_ = false;
+ }
+
+ pop ();
+ }
+
+ void Parser::Impl::
+ simple_content_extension (XML::Element const& e)
+ {
+ if (trace_)
+ wcout << "extension base: " << fq_name (e, e["base"]) << endl;
+
+ set_type<Extends> (trim (e["base"]), e, dynamic_cast<Complex&> (scope ()));
+
+ push (e);
+
+ annotation (false);
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'attribute', 'anyAttribute', or "
+ << "'attributeGroup' instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ pop ();
+ }
+
+ void Parser::Impl::
+ simple_content_restriction (XML::Element const& r)
+ {
+ String base (trim (r["base"]));
+
+ if (trace_ && base)
+ wcout << "restriction base: " << fq_name (r, base) << endl;
+
+ push (r);
+ annotation (false);
+
+ if (!base)
+ {
+ // Anonymous base type.
+ //
+ if (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ if (name == L"simpleType") simple_type (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'simpleType' instead of "
+ << "'" << e.name () << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ else
+ {
+ wcerr << file () << ":" << r.line () << ":" << r.column () << ": "
+ << "error: expected 'base' attribute or 'simpleType' "
+ << "nested element" << endl;
+
+ valid_ = false;
+ }
+
+ if (!valid_)
+ {
+ pop ();
+ return;
+ }
+ }
+
+ Facets facets;
+ String pattern;
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"simpleType")
+ {
+ // This is a "superimposed" restriction where the base
+ // content is restricted by specifying another simple
+ // type. The attributes are restricted in the ussual
+ // way. So in effect we have kind of two base classes.
+ // I guess the way to handle this one day would be to
+ // copy all the facets from the base-to-this-type
+ // part of the hierarchy (will need to "know" facets
+ // for the built-in type restrictions as well). For
+ // now just ignore it.
+ //
+ }
+ else if (name == L"enumeration")
+ {
+ // Right now our sementic graph cannot represent enumerations
+ // with attributes so we are going to ignore enumerators for
+ // now.
+ //
+ }
+ else if (name == L"minExclusive" ||
+ name == L"minInclusive" ||
+ name == L"maxExclusive" ||
+ name == L"maxInclusive" ||
+ name == L"totalDigits" ||
+ name == L"fractionDigits" ||
+ name == L"length" ||
+ name == L"minLength" ||
+ name == L"maxLength" ||
+ name == L"whiteSpace")
+ {
+ facets[name] = trim (e["value"]);
+ }
+ else if (name == L"pattern")
+ {
+ if (pattern)
+ pattern += L'|';
+
+ pattern += e["value"];
+ }
+ else if (name == L"attribute")
+ {
+ if (proper_restriction_)
+ attribute (e, false);
+ }
+ else if (name == L"anyAttribute")
+ {
+ if (proper_restriction_)
+ any_attribute (e);
+ }
+ else if (name == L"attributeGroup")
+ {
+ if (proper_restriction_)
+ attribute_group (e);
+ }
+ else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unexpected element '" << name << "' in "
+ << "simple content restriction" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ if (pattern)
+ facets[L"pattern"] = pattern;
+
+ Complex& type (dynamic_cast<Complex&> (scope ()));
+ Restricts* restricts = set_type<Restricts> (base, r, type);
+
+ if (!facets.empty ())
+ {
+ if (restricts)
+ copy_facets (*restricts, facets);
+ else
+ type.context ().set ("facets", facets);
+ }
+
+ pop ();
+ }
+
+ void Parser::Impl::
+ complex_content_extension (XML::Element const& e, Complex& type)
+ {
+ if (trace_)
+ wcout << "extension base: " << fq_name (e, e["base"]) << endl;
+
+ set_type<Extends> (trim (e["base"]), e, dynamic_cast<Complex&> (scope ()));
+
+ push (e);
+
+ annotation (false);
+
+ if (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+ Compositor* c (0);
+
+ if (name == L"all") c = all (e); else
+ if (name == L"choice") c = choice (e, false); else
+ if (name == L"sequence") c = sequence (e, false); else
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"group") element_group (e, false); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unexpected element '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (c)
+ {
+ unsigned long min (parse_min (trim (e["minOccurs"])));
+ unsigned long max (parse_max (trim (e["maxOccurs"])));
+
+ if (!(min == 0 && max == 0))
+ s_->new_edge<ContainsCompositor> (
+ type, *c, min, max == unbounded ? 0 : max);
+ }
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'attribute', 'anyAttribute', or "
+ << "'attributeGroup' instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+
+ pop ();
+ }
+
+ void Parser::Impl::
+ complex_content_restriction (XML::Element const& e, Complex& type)
+ {
+ if (trace_)
+ wcout << "restriction base: " << fq_name (e, e["base"]) << endl;
+
+ set_type<Restricts> (
+ trim (e["base"]),
+ e,
+ dynamic_cast<Complex&> (scope ()));
+
+ // @@
+ // For now we simply skip the contents unless the base is anyType
+ // (or a trivial alias thereof). Checking for the trivial alias
+ // is further complicated by the fact that it might not be defined
+ // at this stage (forward inheritnace) so we will ignore that case
+ // as well for now.
+ //
+ if (!proper_restriction_)
+ {
+ String base (trim (e["base"]));
+ String uq_name (unqualified_name (base));
+ String ns_name (namespace_name (e, base));
+
+ if (ns_name != xsd || uq_name != L"anyType")
+ return;
+ }
+
+ push (e);
+
+ annotation (false);
+
+ if (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+ Compositor* c (0);
+
+ if (name == L"all") c = all (e); else
+ if (name == L"choice") c = choice (e, false); else
+ if (name == L"sequence") c = sequence (e, false); else
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"group") element_group (e, false); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unexpected element '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (c)
+ {
+ unsigned long min (parse_min (trim (e["minOccurs"])));
+ unsigned long max (parse_max (trim (e["maxOccurs"])));
+
+ if (!(min == 0 && max == 0))
+ s_->new_edge<ContainsCompositor> (
+ type, *c, min, max == unbounded ? 0 : max);
+ }
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'attribute', 'anyAttribute', or "
+ << "'attributeGroup' instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+
+ pop ();
+ }
+
+ void Parser::Impl::
+ element (XML::Element const& e, bool global)
+ {
+ bool qualified (global ? true : qualify_element_);
+
+ if (String form = trim (e["form"]))
+ qualified = form == L"qualified";
+
+ if (trace_)
+ wcout << "element qualified: " << qualified << endl;
+
+ if (String name = trim (e["name"]))
+ {
+ if (trace_)
+ wcout << "element name '" << name << "'" << endl;
+
+ Element& node (
+ s_->new_node<Element> (
+ file (), e.line (), e.column (), global, qualified));
+
+ if (!global)
+ {
+ unsigned long min (parse_min (trim (e["minOccurs"])));
+ unsigned long max (parse_max (trim (e["maxOccurs"])));
+
+ if (!(min == 0 && max == 0))
+ {
+ s_->new_edge<Names> (scope (), node, name);
+
+ s_->new_edge<ContainsParticle> (
+ compositor (), node, min, max == unbounded ? 0 : max);
+ }
+ }
+ else
+ s_->new_edge<Names> (scope (), node, name);
+
+ if (qualified)
+ s_->new_edge<BelongsToNamespace> (node, cur_ns ());
+
+ // Default and fixed values are mutually exclusive.
+ //
+ if (e.attribute_p ("fixed"))
+ node.fixed (e.attribute ("fixed"));
+ else if (e.attribute_p ("default"))
+ node.default_ (e.attribute ("default"));
+
+ if (node.default_p ())
+ {
+ node.context ().set ("dom-node", e.dom_element ());
+ default_values_.push_back (&node);
+ }
+
+ bool subst (false);
+ if (global)
+ {
+ if (String sg = trim (e["substitutionGroup"]))
+ {
+ if (trace_)
+ wcout << "substitutes " << sg << endl;
+
+ subst = true;
+
+ try
+ {
+ String uq_name (unqualified_name (sg));
+ String ns_name (namespace_name (e, sg));
+
+ node.context ().set ("substitution-ns-name", ns_name);
+ node.context ().set ("substitution-uq-name", uq_name);
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unable to resolve namespace prefix '"
+ << ex.prefix () << "' in '" << sg << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+
+ if (String type = trim (e["type"]))
+ {
+ if (trace_)
+ wcout << "element type " << fq_name (e, type) << endl;
+
+ set_type<Belongs> (type, e, node);
+
+ // Parse annotation.
+ //
+ push (e);
+
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, node);
+
+ pop ();
+ }
+ else
+ {
+ // Looks like an anonymous type.
+ //
+ push (e);
+
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, node);
+
+ if (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ Type* t (0);
+
+ if (name == L"simpleType") t = simple_type (e); else
+ if (name == L"complexType") t = complex_type (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'simpleType' or 'complexType' "
+ << "instead of '" << e.name () << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (t)
+ s_->new_edge<Belongs> (node, *t);
+ }
+ // By default the type is anyType unless this element is a
+ // member of a substitution group, in which case it has the
+ // same type as the element it substiutes.
+ //
+ else if (!subst)
+ {
+ if (!is_disabled ("F001"))
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "warning F001: element '" << name << "' is implicitly "
+ << "of anyType" << endl;
+
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "info: did you forget to specify 'type' attribute?"
+ << endl;
+ }
+
+ String prefix (ns_prefix (e, xsd));
+ type = prefix + (prefix.empty () ? L"" : L":") + L"anyType";
+
+ set_type<Belongs> (type, e, node);
+ }
+
+ pop ();
+ }
+ }
+ else if (String ref = trim (e["ref"]))
+ {
+ Element& node (
+ s_->new_node<Element> (
+ file (), e.line (), e.column (), true, true));
+
+ unsigned long min (parse_min (trim (e["minOccurs"])));
+ unsigned long max (parse_max (trim (e["maxOccurs"])));
+
+ // Default and fixed values are mutually exclusive.
+ //
+ if (e.attribute_p ("fixed"))
+ node.fixed (e.attribute ("fixed"));
+ else if (e.attribute_p ("default"))
+ node.default_ (e.attribute ("default"));
+
+ if (node.default_p ())
+ {
+ node.context ().set ("dom-node", e.dom_element ());
+ default_values_.push_back (&node);
+ }
+
+ // Parse annotation.
+ //
+ push (e);
+
+ if (Annotation* a = annotation (true))
+ s_->new_edge<Annotates> (*a, node);
+
+ pop ();
+
+ if (!(min == 0 && max == 0))
+ {
+ // Ref can only be in compositor.
+ //
+ s_->new_edge<ContainsParticle> (
+ compositor (), node, min, max == unbounded ? 0 : max);
+
+ // Try to resolve the prototype.
+ //
+ try
+ {
+ String uq_name (unqualified_name (ref));
+ String ns_name (namespace_name (e, ref));
+
+ s_->new_edge<Names> (scope (), node, uq_name);
+
+ Element& prot (resolve<Element> (ns_name, uq_name, *s_, *cache_));
+ s_->new_edge<BelongsToNamespace> (node, prot.namespace_ ());
+
+ // Copy substitution group information if any.
+ //
+ if (prot.context ().count ("substitution-ns-name"))
+ {
+ node.context ().set (
+ "substitution-ns-name",
+ prot.context ().get<String> ("substitution-ns-name"));
+
+ node.context ().set (
+ "substitution-uq-name",
+ prot.context ().get<String> ("substitution-uq-name"));
+ }
+
+ // Transfer default and fixed values if the ref declaration hasn't
+ // defined its own.
+ //
+ if (!node.default_p ())
+ {
+ if (prot.fixed_p ())
+ node.fixed (prot.value ());
+ else if (prot.default_p ())
+ node.default_ (prot.value ());
+
+ if (node.default_p ())
+ {
+ node.context ().set (
+ "dom-node",
+ prot.context ().get<Xerces::DOMElement*> ("dom-node"));
+ default_values_.push_back (&node);
+ }
+ }
+
+ // Transfer annotation if the ref declaration hasn't defined its own.
+ //
+ if (!node.annotated_p () && prot.annotated_p ())
+ s_->new_edge<Annotates> (prot.annotation (), node);
+
+ // Set type information.
+ //
+ if (prot.typed_p ())
+ {
+ s_->new_edge<Belongs> (node, prot.type ());
+ }
+ else if (prot.context ().count ("type-ns-name"))
+ {
+ String ns_name (prot.context ().get<String> ("type-ns-name"));
+ String uq_name (prot.context ().get<String> ("type-uq-name"));
+
+ node.context ().set ("type-ns-name", ns_name);
+ node.context ().set ("type-uq-name", uq_name);
+ node.context ().set ("edge-type-id", type_id (typeid (Belongs)));
+
+ if (trace_)
+ wcout << "element '" << ref << "' is not typed" << endl
+ << "deferring resolution until later" << endl;
+ }
+ else
+ {
+ // This could be a recursive reference to an element who's
+ // (anonymous) type is being defined. We are going to let
+ // resolver sort out this case.
+ //
+ node.context ().set ("instance-ns-name", ns_name);
+ node.context ().set ("instance-uq-name", uq_name);
+
+ if (trace_)
+ wcout << "looks like a recursive reference to an element '"
+ << ns_name << "#" << uq_name << "' which is being "
+ << "defined" << endl
+ << "deferring resolution until later" << endl;
+ }
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+
+ abort ();
+ }
+ }
+ catch (NotName const& ex)
+ {
+ node.context ().set ("instance-ns-name", ex.ns ());
+ node.context ().set ("instance-uq-name", ex.name ());
+
+ if (trace_)
+ wcout << "unable to resolve name '" << ex.name ()
+ << "' inside namespace '" << ex.ns () << "'" << endl
+ << "deferring resolution until later" << endl;
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unable to resolve namespace prefix '"
+ << ex.prefix () << "' in '" << ref << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ }
+ else
+ {
+ if (valid_)
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: 'name' or 'ref' attribute is missing in element "
+ << "declaration" << endl;
+ }
+ }
+ }
+
+ SemanticGraph::Annotation* Parser::Impl::
+ annotation (bool process)
+ {
+ Annotation* r (0);
+
+ if (more ())
+ {
+ XML::Element e (next ());
+
+ if (e.name () == L"annotation")
+ {
+ if (process)
+ {
+ push (e);
+
+ while (more ())
+ {
+ XML::Element doc (next ());
+
+ if (doc.name () == L"documentation")
+ {
+ using Xerces::DOMNode;
+ using Xerces::DOMText;
+ using Xerces::DOMElement;
+
+ // Use first non-structured (text only) documentation element.
+ //
+ String text;
+ bool struc (false);
+ DOMElement* de (doc.dom_element());
+
+ for (DOMNode* n (de->getFirstChild ());
+ n != 0 && !struc;
+ n = n->getNextSibling ())
+ {
+ switch (n->getNodeType ())
+ {
+ case DOMNode::TEXT_NODE:
+ case DOMNode::CDATA_SECTION_NODE:
+ {
+ DOMText* t (static_cast<DOMText*> (n));
+ text += XML::transcode (t->getData ());
+ break;
+ }
+ case DOMNode::ELEMENT_NODE:
+ {
+ struc = true;
+ break;
+ }
+ default:
+ break; // ignore
+ }
+ }
+
+ if (struc)
+ continue;
+
+ r = &s_->new_node<Annotation> (
+ file (), e.line (), e.column (), text);
+ break;
+ }
+ }
+
+ pop ();
+ }
+ }
+ else
+ prev ();
+ }
+
+ return r;
+ }
+
+
+ void Parser::Impl::
+ attribute (XML::Element const& a, bool global)
+ {
+ bool optional (true);
+
+ String use (trim (a["use"]));
+
+ if (use == L"prohibited")
+ return;
+ else if (use == L"required")
+ optional = false;
+
+ bool qualified (global ? true : qualify_attribute_);
+
+ if (String form = trim (a["form"]))
+ qualified = form == L"qualified";
+
+ if (String name = trim (a["name"]))
+ {
+ if (trace_)
+ wcout << "attribute '" << name << "'" << endl;
+
+ Attribute& node (
+ s_->new_node<Attribute> (
+ file (), a.line (), a.column (), optional, global, qualified));
+
+ s_->new_edge<Names> (scope (), node, name);
+
+ if (qualified)
+ s_->new_edge<BelongsToNamespace> (node, cur_ns ());
+
+
+ // Default and fixed values are mutually exclusive.
+ //
+ if (a.attribute_p ("fixed"))
+ node.fixed (a.attribute ("fixed"));
+ else if (a.attribute_p ("default"))
+ node.default_ (a.attribute ("default"));
+
+ if (node.default_p ())
+ {
+ node.context ().set ("dom-node", a.dom_element ());
+ default_values_.push_back (&node);
+ }
+
+ if (String type = trim (a["type"]))
+ {
+ if (trace_)
+ wcout << "attribute type: '" << fq_name (a, type) << "'" << endl;
+
+ set_type<Belongs> (type, a, node);
+
+ // Parse annotation.
+ //
+ push (a);
+
+ if (Annotation* ann = annotation (true))
+ s_->new_edge<Annotates> (*ann, node);
+
+ pop ();
+ }
+ else
+ {
+ // Looks like an anonymous type.
+ //
+ push (a);
+
+ if (Annotation* ann = annotation (true))
+ s_->new_edge<Annotates> (*ann, node);
+
+ if (more ())
+ {
+ XML::Element e (next ());
+
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ Type* t (0);
+
+ if (name == L"simpleType") t = simple_type (e); else
+ {
+ wcerr << file () << ":" << a.line () << ":" << a.column () << ": "
+ << "error: expected 'simpleType' instead of '" << e.name ()
+ << "'" << endl;
+
+ valid_ = false;
+ }
+
+ if (t)
+ s_->new_edge<Belongs> (node, *t);
+ }
+ else
+ {
+ if (!is_disabled ("F002"))
+ {
+ wcerr << file () << ":" << a.line () << ":" << a.column () << ": "
+ << "warning F002: attribute '" << name << "' is implicitly "
+ << "of anySimpleType" << endl;
+
+ wcerr << file () << ":" << a.line () << ":" << a.column () << ": "
+ << "info: did you forget to specify 'type' attribute?"
+ << endl;
+ }
+
+ // anySimpleType
+ //
+ String prefix (ns_prefix (a, xsd));
+ type = prefix + (prefix.empty () ? L"" : L":") + L"anySimpleType";
+
+ set_type<Belongs> (type, a, node);
+ }
+
+ pop ();
+ }
+ }
+ else if (String ref = trim (a["ref"]))
+ {
+ Attribute& node (
+ s_->new_node<Attribute> (
+ file (), a.line (), a.column (), optional, true, true));
+
+
+ // Default and fixed values are mutually exclusive.
+ //
+ if (a.attribute_p ("fixed"))
+ node.fixed (a.attribute ("fixed"));
+ else if (a.attribute_p ("default"))
+ node.default_ (a.attribute ("default"));
+
+ if (node.default_p ())
+ {
+ node.context ().set ("dom-node", a.dom_element ());
+ default_values_.push_back (&node);
+ }
+
+ // Parse annotation.
+ //
+ push (a);
+
+ if (Annotation* ann = annotation (true))
+ s_->new_edge<Annotates> (*ann, node);
+
+ pop ();
+
+ try
+ {
+ String uq_name (unqualified_name (ref));
+ String ns_name (namespace_name (a, ref));
+
+ s_->new_edge<Names> (scope (), node, uq_name);
+
+ Attribute& prot (resolve<Attribute> (ns_name, uq_name, *s_, *cache_));
+ s_->new_edge<BelongsToNamespace> (node, prot.namespace_ ());
+
+ // Transfer default and fixed values if the ref declaration hasn't
+ // defined its own.
+ //
+ if (!node.default_p ())
+ {
+ // Default value applies only if this attribute is optional.
+ //
+ if (prot.fixed_p ())
+ node.fixed (prot.value ());
+ else if (optional && prot.default_p ())
+ node.default_ (prot.value ());
+
+ if (node.default_p ())
+ {
+ node.context ().set (
+ "dom-node",
+ prot.context ().get<Xerces::DOMElement*> ("dom-node"));
+ default_values_.push_back (&node);
+ }
+ }
+
+ // Transfer annotation if the ref declaration hasn't defined its own.
+ //
+ if (!node.annotated_p () && prot.annotated_p ())
+ s_->new_edge<Annotates> (prot.annotation (), node);
+
+ // Set type.
+ //
+ if (prot.typed_p ())
+ {
+ s_->new_edge<Belongs> (node, prot.type ());
+ }
+ else if (prot.context ().count ("type-ns-name"))
+ {
+ String ns_name (prot.context ().get<String> ("type-ns-name"));
+ String uq_name (prot.context ().get<String> ("type-uq-name"));
+
+ node.context ().set ("type-ns-name", ns_name);
+ node.context ().set ("type-uq-name", uq_name);
+ node.context ().set ("edge-type-id", type_id (typeid (Belongs)));
+
+ if (trace_)
+ wcout << "attribute '" << ref << "' is not typed" << endl
+ << "deferring resolution until later" << endl;
+ }
+ else
+ {
+ // This could be a recursive reference to an attribute who's
+ // (anonymous) type is being defined. We are going to let
+ // resolver sort out this case.
+ //
+ node.context ().set ("instance-ns-name", ns_name);
+ node.context ().set ("instance-uq-name", uq_name);
+
+ if (trace_)
+ wcout << "looks like a recursive reference to an attribute '"
+ << ns_name << "#" << uq_name << "' which is being "
+ << "defined" << endl
+ << "deferring resolution until later" << endl;
+ }
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << file () << ":" << a.line () << ":" << a.column () << ": "
+ << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+ abort ();
+ }
+ }
+ catch (NotName const& ex)
+ {
+ node.context ().set ("instance-ns-name", ex.ns ());
+ node.context ().set ("instance-uq-name", ex.name ());
+
+ if (trace_)
+ wcout << "unable to resolve name '" << ex.name ()
+ << "' inside namespace '" << ex.ns () << "'" << endl
+ << "deferring resolution until later" << endl;
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ wcerr << file () << ":" << a.line () << ":" << a.column () << ": "
+ << "error: unable to resolve namespace prefix '"
+ << ex.prefix () << "' in '" << ref << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ else
+ {
+ if (valid_)
+ {
+ wcerr << file () << ":" << a.line () << ":" << a.column () << ": "
+ << "error: 'name' or 'ref' attribute is missing in attribute "
+ << "declaration" << endl;
+ }
+ }
+ }
+
+ void Parser::Impl::
+ attribute_group (XML::Element const& g)
+ {
+ if (String name = trim (g["name"]))
+ {
+ // Global definition.
+ //
+ if (trace_)
+ wcout << "attributeGroup '" << name << "'" << endl;
+
+ AttributeGroup& group (
+ s_->new_node<AttributeGroup> (file (), g.line (), g.column ()));
+ s_->new_edge<Names> (scope (), group, name);
+
+ push_scope (group);
+ push (g);
+
+ annotation (false);
+
+ while (more ())
+ {
+ XML::Element e (next ());
+ String name (e.name ());
+
+ if (trace_)
+ wcout << name << endl;
+
+ if (name == L"attribute") attribute (e, false); else
+ if (name == L"anyAttribute") any_attribute (e); else
+ if (name == L"attributeGroup") attribute_group (e); else
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: expected 'attribute', 'anyAttribute', or "
+ << "'attributeGroup' instead of '" << name << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+
+ pop ();
+ pop_scope ();
+ }
+ else if (String ref = trim (g["ref"]))
+ {
+ if (trace_)
+ wcout << "attribute-group-ref " << ref << endl;
+
+ try
+ {
+ String uq_name (unqualified_name (ref));
+ String ns_name (namespace_name (g, ref));
+
+ // In order to avoid code duplication we are going to let the
+ // resolver handle this case.
+ //
+ if (trace_)
+ wcout << "deferring resolution of group name '" << uq_name
+ << "' inside namespace '" << ns_name << "'"
+ << " until later" << endl;
+
+ Scope& s (scope ());
+ AttributeGroupRef ref (uq_name, ns_name, s);
+
+ if (!s.context ().count ("attribute-group-refs"))
+ s.context ().set ("attribute-group-refs", AttributeGroupRefs ());
+
+ s.context ().get<AttributeGroupRefs> (
+ "attribute-group-refs").push_back (ref);
+ }
+ catch (NotNamespace const& ex)
+ {
+ if (valid_)
+ {
+ wcerr << file () << ":" << g.line () << ":" << g.column () << ": "
+ << "ice: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+ abort ();
+ }
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ wcerr << file () << ":" << g.line () << ":" << g.column () << ": "
+ << "error: unable to resolve namespace prefix '"
+ << ex.prefix () << "' in '" << ref << "'" << endl;
+
+ valid_ = false;
+ }
+ }
+ else
+ {
+ wcerr << file () << ":" << g.line () << ":" << g.column () << ": "
+ << "error: 'name' or 'ref' attribute is missing in "
+ << "attributeGroup declaration" << endl;
+
+ valid_ = false;
+ return;
+ }
+ }
+
+ void Parser::Impl::
+ any (XML::Element const& a)
+ {
+ if (trace_)
+ wcout << "any" << endl;
+
+ String namespaces (trim (a["namespace"]));
+
+ if (!namespaces)
+ namespaces = L"##any";
+
+ Any& any (
+ s_->new_node<Any> (file (), a.line (), a.column (), namespaces));
+
+ unsigned long min (parse_min (trim (a["minOccurs"])));
+ unsigned long max (parse_max (trim (a["maxOccurs"])));
+
+ // Parse annotation.
+ //
+ push (a);
+
+ if (Annotation* ann = annotation (true))
+ s_->new_edge<Annotates> (*ann, any);
+
+ pop ();
+
+ if (!(min == 0 && max == 0))
+ {
+ s_->new_edge<ContainsParticle> (
+ compositor (), any, min, max == unbounded ? 0 : max);
+
+ // Any has no name so we have to come up with a fake one in order to
+ // put it into the scope.
+ //
+ unsigned long count;
+ SemanticGraph::Context& ctx (scope ().context ());
+
+ if (!ctx.count ("any-name-count"))
+ {
+ count = 0;
+ ctx.set ("any-name-count", count);
+ }
+ else
+ count = ++(ctx.get<unsigned long> ("any-name-count"));
+
+ std::basic_ostringstream<wchar_t> os;
+ os << "any #" << count;
+
+ s_->new_edge<Names> (scope (), any, os.str ());
+ }
+ }
+
+ void Parser::Impl::
+ any_attribute (XML::Element const& a)
+ {
+ if (trace_)
+ wcout << "anyAttribute" << endl;
+
+ String namespaces (trim (a["namespace"]));
+
+ if (!namespaces)
+ namespaces = L"##any";
+
+ AnyAttribute& any (
+ s_->new_node<AnyAttribute> (
+ file (), a.line (), a.column (), namespaces));
+
+ // Parse annotation.
+ //
+ push (a);
+
+ if (Annotation* ann = annotation (true))
+ s_->new_edge<Annotates> (*ann, any);
+
+ pop ();
+
+ // AnyAttribute has no name so we have to come up with a fake one
+ // in order to put it into the scope.
+ //
+
+ unsigned long count;
+ SemanticGraph::Context& ctx (scope ().context ());
+
+ if (!ctx.count ("any-attribute-name-count"))
+ {
+ count = 0;
+ ctx.set ("any-attribute-name-count", count);
+ }
+ else
+ count = ++(ctx.get<unsigned long> ("any-attribute-name-count"));
+
+ std::basic_ostringstream<wchar_t> os;
+ os << "any-attribute #" << count;
+
+ s_->new_edge<Names> (scope (), any, os.str ());
+ }
+
+ // Some specializations to get edge orientations right.
+ //
+
+ template <typename Edge, typename Node>
+ struct Orientation
+ {
+ static Edge&
+ set_edge (Schema& s, Node& node, Type& type)
+ {
+ // By default it is node->edge
+ //
+ return s.template new_edge<Edge> (node, type);
+ }
+ };
+
+ template <typename Node>
+ struct Orientation<Arguments, Node>
+ {
+ static Arguments&
+ set_edge (Schema& s, Node& node, Type& type)
+ {
+ // For Arguments it is type->node.
+ //
+ return s.template new_edge<Arguments> (type, node);
+ }
+ };
+
+ template <typename Edge, typename Node>
+ Edge* Parser::Impl::
+ set_type (String const& type, XML::Element const& e, Node& node)
+ {
+ Edge* r (0);
+
+ try
+ {
+ String uq_name (unqualified_name (type));
+ String ns_name (namespace_name (e, type));
+
+ Type& t (resolve<Type> (ns_name, uq_name, *s_, *cache_));
+
+ // See if it is an IDREF specialization.
+ //
+ if (ns_name == xsd && (uq_name == L"IDREF" || uq_name == L"IDREFS"))
+ {
+ // See if we've got 'xse:refType' attribute.
+ //
+ if (String ref_type = trim (e.attribute (xse, "refType")))
+ {
+ if (trace_)
+ wcout << "found refType attribute '" << ref_type << "'" << endl;
+
+ //@@ It is a bit wasteful to create a new spcialization for
+ // each refType. Instead we could lookup the target type
+ // and then navigate through Arguments edges to see if this
+ // type already arguments specialization that we are intersted
+ // in. But for now I will simplify the logic by creating a new
+ // specialization every time.
+ //
+
+ Specialization* spec (0);
+
+ if (uq_name == L"IDREF")
+ spec = &s_->new_node<Fundamental::IdRef> (
+ file (), e.line (), e.column ());
+ else
+ spec = &s_->new_node<Fundamental::IdRefs> (
+ file (), e.line (), e.column ());
+
+ r = &Orientation<Edge, Node>::set_edge (*s_, node, *spec);
+
+ set_type<Arguments> (ref_type, e, *spec);
+ }
+ else
+ r = &Orientation<Edge, Node>::set_edge (*s_, node, t);
+ }
+ else
+ r = &Orientation<Edge, Node>::set_edge (*s_, node, t);
+ }
+ catch (NotNamespace const& ex)
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unable to resolve namespace '" << ex.ns () << "'"
+ << endl;
+
+ valid_ = false;
+
+ }
+ catch (NotName const& ex)
+ {
+ node.context ().set ("type-ns-name", ex.ns ());
+ node.context ().set ("type-uq-name", ex.name ());
+ node.context ().set ("edge-type-id", type_id (typeid (Edge)));
+
+ if (trace_)
+ wcout << "unable to resolve name '" << ex.name ()
+ << "' inside namespace '" << ex.ns () << "'" << endl
+ << "deferring resolution until later" << endl;
+ }
+ catch (XML::NoMapping const& ex)
+ {
+ wcerr << file () << ":" << e.line () << ":" << e.column () << ": "
+ << "error: unable to resolve namespace prefix "
+ << "'" << ex.prefix () << "' in '" << type << "'" << endl;
+
+ valid_ = false;
+ }
+
+ return r;
+ }
+
+ // Xerces has a provision to associate a public id with input streams
+ // that can later be used in diagnostics. Unfortunately, it doesn't
+ // work. So we will have to keep our own track.
+ //
+ struct Context
+ {
+ Context () {}
+
+ // File map for diagnostic.
+ //
+ Path const&
+ file (Path const& abs) const
+ {
+ FileMap::const_iterator i (file_map_.find (abs));
+
+ if (i != file_map_.end ())
+ {
+ return i->second;
+ }
+ else
+ {
+ return abs;
+ }
+ }
+
+ void
+ map_file (Path const& abs, Path const& rel)
+ {
+ file_map_[abs] = rel;
+ }
+
+ private:
+ Context (Context const&);
+ Context& operator= (Context const&);
+
+ private:
+ typedef std::map<Path, Path> FileMap;
+ FileMap file_map_;
+ };
+
+ //
+ //
+ class ErrorHandler : public Xerces::DOMErrorHandler
+ {
+ public:
+ ErrorHandler (bool& valid, XSDFrontend::Context const& ctx)
+ : valid_ (valid),
+ ctx_ (ctx)
+ {
+ }
+
+ virtual bool
+ handleError (Xerces::DOMError const& e)
+ {
+ // Xerces likes to say "Fatal error encountered during schema scan".
+ // We don't need this junk.
+ //
+ if (!valid_
+ && e.getLocation ()->getLineNumber () == 0
+ && e.getLocation ()->getColumnNumber () == 0)
+ return true;
+
+
+ XSDFrontend::SemanticGraph::Path abs_path (
+ XML::transcode_to_narrow (e.getLocation ()->getURI ()));
+
+ XSDFrontend::SemanticGraph::Path rel_path (ctx_.file (abs_path));
+
+ wcerr << rel_path << ':'
+ << e.getLocation ()->getLineNumber () << ':'
+ << e.getLocation ()->getColumnNumber () << ": ";
+
+ switch (e.getSeverity ())
+ {
+ case Xerces::DOMError::DOM_SEVERITY_WARNING:
+ {
+ wcerr << "warning: ";
+ break;
+ }
+ default:
+ {
+ wcerr << "error: ";
+ valid_ = false;
+ break;
+ }
+ }
+
+ wcerr << e.getMessage () << endl;
+
+ return true;
+ }
+
+ private:
+ bool& valid_;
+ XSDFrontend::Context const& ctx_;
+ };
+
+
+ // Failed to open resource.
+ //
+ struct Open {};
+
+ class InputSource: public Xerces::InputSource
+ {
+ public:
+ InputSource (
+ Path const& abs,
+ Path const& rel,
+ Path const& base,
+ XSDFrontend::Context const& ctx,
+ Xerces::MemoryManager* mm = Xerces::XMLPlatformUtils::fgMemoryManager)
+ : Xerces::InputSource (mm),
+ abs_ (abs),
+ rel_ (rel),
+ base_ (base),
+ ctx_ (ctx)
+ {
+ setSystemId (XML::XMLChString (String (abs_.string ())).c_str ());
+ }
+
+ virtual Xerces::BinInputStream*
+ makeStream () const
+ {
+ using namespace Xerces;
+
+ BinFileInputStream* is (
+ new (getMemoryManager ())
+ BinFileInputStream (getSystemId (), getMemoryManager ()));
+
+ if (!is->getIsOpen ())
+ {
+ delete is;
+
+ wcerr << ctx_.file (base_) << ": error: "
+ << "'" << rel_ << "': unable to open in read mode"
+ << endl;
+
+ throw Open ();
+ }
+
+ return is;
+ }
+
+ private:
+ Path abs_;
+ Path rel_;
+ Path base_;
+ XSDFrontend::Context const& ctx_;
+ };
+
+
+ class EntityResolver: public Xerces::XMemory,
+ public Xerces::DOMLSResourceResolver
+ {
+ public:
+ EntityResolver (XSDFrontend::Context& ctx, LocationTranslator* t)
+ : ctx_ (ctx), loc_translator_ (t)
+ {
+ }
+
+ virtual Xerces::DOMLSInput*
+ resolveResource(XMLCh const* const,
+ XMLCh const* const,
+ XMLCh const* const /*pub_id*/,
+ XMLCh const* const prv_id,
+ XMLCh const* const base_uri)
+ {
+ /*
+ XMLCh empty[1];
+ empty[0] = 0;
+
+ wcerr << "resolve entity:" << endl
+ << " pub_id " << (pub_id ? pub_id : empty) << endl
+ << " prv_id " << (prv_id ? prv_id : empty) << endl
+ << " uri " << (base_uri ? base_uri : empty) << endl;
+ */
+
+ // base_uri should be a valid path by now.
+ //
+ Path base (XML::transcode_to_narrow (base_uri));
+
+ if (prv_id == 0)
+ {
+ //@@ How can I get the line/column numbers for this?
+ //
+ wcerr << ctx_.file (base) << ": error: "
+ << "unable to guess which schema to open"
+ << endl;
+
+ wcerr << ctx_.file (base) << ": info: "
+ << "did you forget to specify schemaLocation for import/include?"
+ << endl;
+
+ throw Open ();
+ }
+
+ NarrowString path_str (XML::transcode_to_narrow (prv_id));
+
+ if (loc_translator_)
+ path_str = loc_translator_->translate (path_str);
+
+ try
+ {
+ Path path (path_str);
+ Path base_dir (base.directory ());
+
+ Path abs_path, rel_path;
+
+ if (path.absolute ())
+ {
+ abs_path = rel_path = path;
+ }
+ else
+ {
+ abs_path = base_dir / path;
+ rel_path = ctx_.file (base).directory () / path;
+ }
+
+ abs_path.normalize ();
+
+ ctx_.map_file (abs_path, rel_path);
+
+ using namespace Xerces;
+
+ InputSource* is (
+ new (XMLPlatformUtils::fgMemoryManager)
+ InputSource (abs_path, rel_path, base, ctx_));
+
+ // Note that I can't use XMLPlatformUtils::fgMemoryManager here
+ // since Wrapper4InputSource is-not-an XMemory.
+ //
+ return new Wrapper4InputSource (is);
+ }
+ catch (InvalidPath const&)
+ {
+ wcerr << ctx_.file (base) << ": error: "
+ << "'" << path_str.c_str () << "' is not a valid filesystem path"
+ << endl;
+ throw;
+ }
+
+ // Will never reach.
+ //
+ return 0;
+ }
+
+ private:
+ XSDFrontend::Context& ctx_;
+ LocationTranslator* loc_translator_;
+ };
+
+
+ XML::AutoPtr<Xerces::DOMDocument> Parser::Impl::
+ dom (Path const& tu, bool validate)
+ {
+ using namespace Xerces;
+
+ try
+ {
+ XSDFrontend::Context ctx;
+
+ // Do normalize() before complete() to avoid hitting system path
+ // limits with '..' directories.
+ //
+ Path abs_path (tu);
+ abs_path.normalize ().complete ();
+ ctx.map_file (abs_path, tu);
+
+ InputSource input_source (abs_path, tu, abs_path, ctx);
+
+ // First validate the schema with Xerces.
+ //
+ if (validate)
+ {
+ // Instantiate the DOM parser.
+ //
+ XMLCh const gLS[] = {chLatin_L, chLatin_S, chNull };
+
+ // Get an implementation of the Load-Store (LS) interface.
+ //
+ DOMImplementationLS* impl (
+ static_cast<DOMImplementationLS*> (
+ DOMImplementationRegistry::getDOMImplementation (gLS)));
+
+ // Create a DOMBuilder.
+ //
+ XML::AutoPtr<DOMLSParser> parser (
+ impl->createLSParser (DOMImplementationLS::MODE_SYNCHRONOUS, 0));
+
+ DOMConfiguration* conf (parser->getDomConfig ());
+
+ conf->setParameter (XMLUni::fgDOMComments, false);
+ conf->setParameter (XMLUni::fgDOMDatatypeNormalization, true);
+ conf->setParameter (XMLUni::fgDOMEntities, false);
+ conf->setParameter (XMLUni::fgDOMNamespaces, true);
+ conf->setParameter (XMLUni::fgDOMValidate, true);
+ conf->setParameter (XMLUni::fgDOMElementContentWhitespace, false);
+ conf->setParameter (XMLUni::fgXercesSchema, true);
+
+ // Xerces-C++ 3.1.0 is the first version with working multi import
+ // support.
+ //
+#if _XERCES_VERSION >= 30100
+ conf->setParameter (XMLUni::fgXercesHandleMultipleImports, multiple_imports_);
+#endif
+
+ conf->setParameter (XMLUni::fgXercesSchemaFullChecking, full_schema_check_);
+ conf->setParameter (XMLUni::fgXercesValidationErrorAsFatal, true);
+
+ ErrorHandler eh (valid_, ctx);
+ conf->setParameter (XMLUni::fgDOMErrorHandler, &eh);
+
+ EntityResolver er (ctx, loc_translator_);
+ conf->setParameter (XMLUni::fgDOMResourceResolver, &er);
+
+ Wrapper4InputSource wrap (&input_source, false);
+ parser->loadGrammar (&wrap, Grammar::SchemaGrammarType);
+ }
+
+ if (!valid_)
+ return XML::AutoPtr<DOMDocument> (0);
+
+ // Now do our own parsing.
+ //
+ std::auto_ptr<XML::SchemaDOMParser> xsd_parser (
+ new (XMLPlatformUtils::fgMemoryManager) XML::SchemaDOMParser ());
+
+ xsd_parser->parse (input_source);
+
+ XML::AutoPtr<DOMDocument> doc (xsd_parser->adoptDocument());
+
+ return doc;
+ }
+ catch (Xerces::XMLException const& e)
+ {
+ wcerr << tu << ": ice: Xerces::XMLException: " << e.getMessage ()
+ << endl;
+
+ abort ();
+ }
+ catch (Xerces::DOMException const& e)
+ {
+ size_t const size = 2047;
+ XMLCh text[size + 1];
+
+ wcerr << tu << ": ice: Xerces::DOMException: ";
+
+ if (DOMImplementation::loadDOMExceptionMsg (e.code, text, size))
+ wcerr << text << endl;
+ else
+ wcerr << "no message available, error code: " << e.code << endl;
+
+ abort ();
+ }
+ catch (InvalidPath const&)
+ {
+ // Diagnostics has already been issued.
+ //
+ valid_ = false;
+ }
+ catch (Open const&)
+ {
+ // Diagnostics has already been issued.
+ //
+ valid_ = false;
+ }
+
+ return XML::AutoPtr<DOMDocument> (0);
+ }
+
+ // LocationTranslator
+ //
+ LocationTranslator::
+ ~LocationTranslator ()
+ {
+ }
+
+ // Parser
+ //
+ Parser::
+ ~Parser ()
+ {
+ }
+
+ Parser::
+ Parser (bool proper_restriction,
+ bool multiple_imports,
+ bool full_schema_check)
+ : impl_ (new Impl (proper_restriction,
+ multiple_imports,
+ full_schema_check,
+ 0,
+ 0))
+ {
+ }
+
+ Parser::
+ Parser (bool proper_restriction,
+ bool multiple_imports,
+ bool full_schema_check,
+ LocationTranslator& t,
+ const WarningSet& d)
+ : impl_ (new Impl (proper_restriction,
+ multiple_imports,
+ full_schema_check,
+ &t,
+ &d))
+ {
+ }
+
+ auto_ptr<SemanticGraph::Schema> Parser::
+ parse (SemanticGraph::Path const& path)
+ {
+ return impl_->parse (path);
+ }
+
+ auto_ptr<SemanticGraph::Schema> Parser::
+ parse (SemanticGraph::Paths const& paths)
+ {
+ return impl_->parse (paths);
+ }
+
+ auto_ptr<SemanticGraph::Schema> Parser::
+ xml_schema (SemanticGraph::Path const& path)
+ {
+ return impl_->xml_schema (path);
+ }
+}
generated by cgit v1.2.3 (git 2.25.1) at 2025-10-31 19:53:47 +0000