Schema and metadata attribute handling fixes for langAware decoding
[shibboleth/cpp-sp.git] / shibsp / attribute / resolver / impl / XMLAttributeExtractor.cpp
1 /**
2  * Licensed to the University Corporation for Advanced Internet
3  * Development, Inc. (UCAID) under one or more contributor license
4  * agreements. See the NOTICE file distributed with this work for
5  * additional information regarding copyright ownership.
6  *
7  * UCAID licenses this file to you under the Apache License,
8  * Version 2.0 (the "License"); you may not use this file except
9  * in compliance with the License. You may obtain a copy of the
10  * License at
11  *
12  * http://www.apache.org/licenses/LICENSE-2.0
13  *
14  * Unless required by applicable law or agreed to in writing,
15  * software distributed under the License is distributed on an
16  * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
17  * either express or implied. See the License for the specific
18  * language governing permissions and limitations under the License.
19  */
20
21 /**
22  * XMLAttributeExtractor.cpp
23  *
24  * AttributeExtractor based on an XML mapping file.
25  */
26
27 #include "internal.h"
28 #include "exceptions.h"
29 #include "Application.h"
30 #include "ServiceProvider.h"
31 #include "attribute/Attribute.h"
32 #include "attribute/AttributeDecoder.h"
33 #include "attribute/filtering/AttributeFilter.h"
34 #include "attribute/filtering/BasicFilteringContext.h"
35 #include "attribute/resolver/AttributeExtractor.h"
36 #include "remoting/ddf.h"
37 #include "security/SecurityPolicy.h"
38 #include "util/SPConstants.h"
39
40 #include <boost/bind.hpp>
41 #include <boost/shared_ptr.hpp>
42 #include <boost/algorithm/string.hpp>
43 #include <boost/iterator/indirect_iterator.hpp>
44 #include <boost/ptr_container/ptr_vector.hpp>
45 #include <boost/tuple/tuple.hpp>
46 #include <saml/SAMLConfig.h>
47 #include <saml/saml1/core/Assertions.h>
48 #include <saml/saml2/core/Assertions.h>
49 #include <saml/saml2/metadata/Metadata.h>
50 #include <saml/saml2/metadata/MetadataCredentialCriteria.h>
51 #include <saml/saml2/metadata/ObservableMetadataProvider.h>
52 #include <xmltooling/XMLToolingConfig.h>
53 #include <xmltooling/security/TrustEngine.h>
54 #include <xmltooling/util/NDC.h>
55 #include <xmltooling/util/ReloadableXMLFile.h>
56 #include <xmltooling/util/Threads.h>
57 #include <xmltooling/util/XMLHelper.h>
58 #include <xercesc/util/XMLUniDefs.hpp>
59
60 using namespace shibsp;
61 using namespace opensaml::saml2md;
62 using namespace opensaml;
63 using namespace xmltooling;
64 using namespace boost;
65 using namespace std;
66 using saml1::NameIdentifier;
67 using saml2::NameID;
68 using saml2::EncryptedAttribute;
69
70 namespace shibsp {
71
72 #if defined (_MSC_VER)
73     #pragma warning( push )
74     #pragma warning( disable : 4250 )
75 #endif
76
77     class XMLExtractorImpl : public ObservableMetadataProvider::Observer
78     {
79     public:
80         XMLExtractorImpl(const DOMElement* e, Category& log);
81         ~XMLExtractorImpl() {
82             for (map<const ObservableMetadataProvider*,decoded_t>::iterator i=m_decodedMap.begin(); i!=m_decodedMap.end(); ++i) {
83                 i->first->removeObserver(this);
84                 for (decoded_t::iterator attrs = i->second.begin(); attrs!=i->second.end(); ++attrs)
85                     for_each(attrs->second.begin(), attrs->second.end(), mem_fun_ref<DDF&,DDF>(&DDF::destroy));
86             }
87             if (m_document)
88                 m_document->release();
89         }
90
91         void setDocument(DOMDocument* doc) {
92             m_document = doc;
93         }
94
95         void onEvent(const ObservableMetadataProvider& metadata) const {
96             // Destroy attributes we cached from this provider.
97             m_attrLock->wrlock();
98             SharedLock wrapper(m_attrLock, false);
99             decoded_t& d = m_decodedMap[&metadata];
100             for (decoded_t::iterator a = d.begin(); a!=d.end(); ++a)
101                 for_each(a->second.begin(), a->second.end(), mem_fun_ref<DDF&,DDF>(&DDF::destroy));
102             d.clear();
103         }
104
105         void extractAttributes(const Application&, const char*, const char*, const NameIdentifier&, ptr_vector<Attribute>&) const;
106         void extractAttributes(const Application&, const char*, const char*, const NameID&, ptr_vector<Attribute>&) const;
107         void extractAttributes(const Application&, const GenericRequest*, const char*, const char*, const saml1::Attribute&, ptr_vector<Attribute>&) const;
108         void extractAttributes(const Application&, const GenericRequest*, const char*, const char*, const saml2::Attribute&, ptr_vector<Attribute>&) const;
109         void extractAttributes(const Application&, const GenericRequest*, const char*, const char*, const saml1::AttributeStatement&, ptr_vector<Attribute>&) const;
110         void extractAttributes(const Application&, const GenericRequest*, const char*, const char*, const saml2::AttributeStatement&, ptr_vector<Attribute>&) const;
111         void extractAttributes(
112             const Application&, const GenericRequest*, const ObservableMetadataProvider*, const XMLCh*, const char*, const Extensions&, ptr_vector<Attribute>&
113             ) const;
114
115         void getAttributeIds(vector<string>& attributes) const {
116             attributes.insert(attributes.end(), m_attributeIds.begin(), m_attributeIds.end());
117         }
118
119         void generateMetadata(SPSSODescriptor& role) const;
120
121     private:
122         Category& m_log;
123         DOMDocument* m_document;
124         typedef map< pair<xstring,xstring>,pair< boost::shared_ptr<AttributeDecoder>,vector<string> > > attrmap_t;
125         attrmap_t m_attrMap;
126         vector<string> m_attributeIds;
127         vector< tuple<xstring,xstring,bool> > m_requestedAttrs;
128
129         // settings for embedded assertions in metadata
130         string m_policyId;
131         scoped_ptr<AttributeFilter> m_filter;
132         scoped_ptr<MetadataProvider> m_metadata;
133         scoped_ptr<TrustEngine> m_trust;
134         bool m_entityAssertions,m_metaAttrCaching;
135
136         // manages caching of decoded Attributes
137         scoped_ptr<RWLock> m_attrLock;
138         typedef map< const EntityAttributes*,vector<DDF> > decoded_t;
139         mutable map<const ObservableMetadataProvider*,decoded_t> m_decodedMap;
140     };
141
142     class XMLExtractor : public AttributeExtractor, public ReloadableXMLFile
143     {
144     public:
145         XMLExtractor(const DOMElement* e) : ReloadableXMLFile(e, Category::getInstance(SHIBSP_LOGCAT".AttributeExtractor.XML")) {
146             background_load();
147         }
148         ~XMLExtractor() {
149             shutdown();
150         }
151
152         // deprecated method
153         void extractAttributes(
154             const Application& application, const RoleDescriptor* issuer, const XMLObject& xmlObject, vector<Attribute*>& attributes
155             ) const {
156             extractAttributes(application, nullptr, issuer, xmlObject, attributes);
157         }
158
159         void extractAttributes(const Application&, const GenericRequest*, const RoleDescriptor*, const XMLObject&, vector<Attribute*>&) const;
160
161         void getAttributeIds(std::vector<std::string>& attributes) const {
162             if (m_impl)
163                 m_impl->getAttributeIds(attributes);
164         }
165
166         void generateMetadata(SPSSODescriptor& role) const {
167             if (m_impl)
168                 m_impl->generateMetadata(role);
169         }
170
171     protected:
172         pair<bool,DOMElement*> background_load();
173
174     private:
175         scoped_ptr<XMLExtractorImpl> m_impl;
176
177         void extractAttributes(const Application&, const GenericRequest*, const RoleDescriptor*, const XMLObject&, ptr_vector<Attribute>&) const;
178     };
179
180 #if defined (_MSC_VER)
181     #pragma warning( pop )
182 #endif
183
184     AttributeExtractor* SHIBSP_DLLLOCAL XMLAttributeExtractorFactory(const DOMElement* const & e)
185     {
186         return new XMLExtractor(e);
187     }
188
189     static const XMLCh _aliases[] =                 UNICODE_LITERAL_7(a,l,i,a,s,e,s);
190     static const XMLCh _AttributeDecoder[] =        UNICODE_LITERAL_16(A,t,t,r,i,b,u,t,e,D,e,c,o,d,e,r);
191     static const XMLCh _AttributeFilter[] =         UNICODE_LITERAL_15(A,t,t,r,i,b,u,t,e,F,i,l,t,e,r);
192     static const XMLCh Attributes[] =               UNICODE_LITERAL_10(A,t,t,r,i,b,u,t,e,s);
193     static const XMLCh _id[] =                      UNICODE_LITERAL_2(i,d);
194     static const XMLCh isRequested[] =              UNICODE_LITERAL_11(i,s,R,e,q,u,e,s,t,e,d);
195     static const XMLCh _MetadataProvider[] =        UNICODE_LITERAL_16(M,e,t,a,d,a,t,a,P,r,o,v,i,d,e,r);
196     static const XMLCh metadataAttributeCaching[] = UNICODE_LITERAL_24(m,e,t,a,d,a,t,a,A,t,t,r,i,b,u,t,e,C,a,c,h,i,n,g);
197     static const XMLCh metadataPolicyId[] =         UNICODE_LITERAL_16(m,e,t,a,d,a,t,a,P,o,l,i,c,y,I,d);
198     static const XMLCh _name[] =                    UNICODE_LITERAL_4(n,a,m,e);
199     static const XMLCh nameFormat[] =               UNICODE_LITERAL_10(n,a,m,e,F,o,r,m,a,t);
200     static const XMLCh _TrustEngine[] =             UNICODE_LITERAL_11(T,r,u,s,t,E,n,g,i,n,e);
201     static const XMLCh _type[] =                    UNICODE_LITERAL_4(t,y,p,e);
202 };
203
204 XMLExtractorImpl::XMLExtractorImpl(const DOMElement* e, Category& log)
205     : m_log(log),
206         m_document(nullptr),
207         m_policyId(XMLHelper::getAttrString(e, nullptr, metadataPolicyId)),
208         m_entityAssertions(true),
209         m_metaAttrCaching(XMLHelper::getAttrBool(e, true, metadataAttributeCaching))
210 {
211 #ifdef _DEBUG
212     xmltooling::NDC ndc("XMLExtractorImpl");
213 #endif
214
215     if (!XMLHelper::isNodeNamed(e, shibspconstants::SHIB2ATTRIBUTEMAP_NS, Attributes))
216         throw ConfigurationException("XML AttributeExtractor requires am:Attributes at root of configuration.");
217
218     DOMElement* child = XMLHelper::getFirstChildElement(e, shibspconstants::SHIB2ATTRIBUTEMAP_NS, _MetadataProvider);
219     if (child) {
220         try {
221             string t(XMLHelper::getAttrString(child, nullptr, _type));
222             if (t.empty())
223                 throw ConfigurationException("MetadataProvider element missing type attribute.");
224             m_log.info("building MetadataProvider of type %s...", t.c_str());
225             m_metadata.reset(SAMLConfig::getConfig().MetadataProviderManager.newPlugin(t.c_str(), child));
226             m_metadata->init();
227         }
228         catch (std::exception& ex) {
229             m_metadata.reset();
230             m_entityAssertions = false;
231             m_log.crit("error building/initializing dedicated MetadataProvider: %s", ex.what());
232             m_log.crit("disabling support for Assertions in EntityAttributes extension");
233         }
234     }
235
236     if (m_entityAssertions) {
237         child = XMLHelper::getFirstChildElement(e, shibspconstants::SHIB2ATTRIBUTEMAP_NS, _TrustEngine);
238         if (child) {
239             try {
240                 string t(XMLHelper::getAttrString(child, nullptr, _type));
241                 if (t.empty())
242                     throw ConfigurationException("TrustEngine element missing type attribute.");
243                 m_log.info("building TrustEngine of type %s...", t.c_str());
244                 m_trust.reset(XMLToolingConfig::getConfig().TrustEngineManager.newPlugin(t.c_str(), child));
245             }
246             catch (std::exception& ex) {
247                 m_entityAssertions = false;
248                 m_log.crit("error building/initializing dedicated TrustEngine: %s", ex.what());
249                 m_log.crit("disabling support for Assertions in EntityAttributes extension");
250             }
251         }
252     }
253
254     if (m_entityAssertions) {
255         child = XMLHelper::getFirstChildElement(e, shibspconstants::SHIB2ATTRIBUTEMAP_NS, _AttributeFilter);
256         if (child) {
257             try {
258                 string t(XMLHelper::getAttrString(child, nullptr, _type));
259                 if (t.empty())
260                     throw ConfigurationException("AttributeFilter element missing type attribute.");
261                 m_log.info("building AttributeFilter of type %s...", t.c_str());
262                 m_filter.reset(SPConfig::getConfig().AttributeFilterManager.newPlugin(t.c_str(), child));
263             }
264             catch (std::exception& ex) {
265                 m_entityAssertions = false;
266                 m_log.crit("error building/initializing dedicated AttributeFilter: %s", ex.what());
267                 m_log.crit("disabling support for Assertions in EntityAttributes extension");
268             }
269         }
270     }
271
272     child = XMLHelper::getFirstChildElement(e, shibspconstants::SHIB2ATTRIBUTEMAP_NS, saml1::Attribute::LOCAL_NAME);
273     while (child) {
274         // Check for missing name or id.
275         const XMLCh* name = child->getAttributeNS(nullptr, _name);
276         if (!name || !*name) {
277             m_log.warn("skipping Attribute with no name");
278             child = XMLHelper::getNextSiblingElement(child, shibspconstants::SHIB2ATTRIBUTEMAP_NS, saml1::Attribute::LOCAL_NAME);
279             continue;
280         }
281
282         auto_ptr_char id(child->getAttributeNS(nullptr, _id));
283         if (!id.get() || !*id.get()) {
284             m_log.warn("skipping Attribute with no id");
285             child = XMLHelper::getNextSiblingElement(child, shibspconstants::SHIB2ATTRIBUTEMAP_NS, saml1::Attribute::LOCAL_NAME);
286             continue;
287         }
288         else if (!strcmp(id.get(), "REMOTE_USER")) {
289             m_log.warn("skipping Attribute, id of REMOTE_USER is a reserved name");
290             child = XMLHelper::getNextSiblingElement(child, shibspconstants::SHIB2ATTRIBUTEMAP_NS, saml1::Attribute::LOCAL_NAME);
291             continue;
292         }
293
294         boost::shared_ptr<AttributeDecoder> decoder;
295         try {
296             DOMElement* dchild = XMLHelper::getFirstChildElement(child, shibspconstants::SHIB2ATTRIBUTEMAP_NS, _AttributeDecoder);
297             if (dchild) {
298                 auto_ptr<xmltooling::QName> q(XMLHelper::getXSIType(dchild));
299                 if (q.get())
300                     decoder.reset(SPConfig::getConfig().AttributeDecoderManager.newPlugin(*q.get(), dchild));
301             }
302             if (!decoder)
303                 decoder.reset(SPConfig::getConfig().AttributeDecoderManager.newPlugin(StringAttributeDecoderType, nullptr));
304         }
305         catch (std::exception& ex) {
306             m_log.error("skipping Attribute (%s), error building AttributeDecoder: %s", id.get(), ex.what());
307         }
308
309         if (!decoder) {
310             child = XMLHelper::getNextSiblingElement(child, shibspconstants::SHIB2ATTRIBUTEMAP_NS, saml1::Attribute::LOCAL_NAME);
311             continue;
312         }
313
314         // Empty NameFormat implies the usual Shib URI naming defaults.
315         const XMLCh* format = child->getAttributeNS(nullptr, nameFormat);
316         if (!format || XMLString::equals(format, shibspconstants::SHIB1_ATTRIBUTE_NAMESPACE_URI) ||
317                 XMLString::equals(format, saml2::Attribute::URI_REFERENCE))
318             format = &chNull;  // ignore default Format/Namespace values
319
320         // Fetch/create the map entry and see if it's a duplicate rule.
321         pair< boost::shared_ptr<AttributeDecoder>,vector<string> >& decl = m_attrMap[pair<xstring,xstring>(name,format)];
322         if (decl.first) {
323             m_log.warn("skipping duplicate Attribute mapping (same name and nameFormat)");
324             child = XMLHelper::getNextSiblingElement(child, shibspconstants::SHIB2ATTRIBUTEMAP_NS, saml1::Attribute::LOCAL_NAME);
325             continue;
326         }
327
328         if (m_log.isInfoEnabled()) {
329             auto_ptr_char n(name);
330             auto_ptr_char f(format);
331             m_log.info("creating mapping for Attribute %s%s%s", n.get(), *f.get() ? ", Format/Namespace:" : "", f.get());
332         }
333
334         decl.first = decoder;
335         decl.second.push_back(id.get());
336         m_attributeIds.push_back(id.get());
337
338         // Check for isRequired/isRequested.
339         bool requested = XMLHelper::getAttrBool(child, false, isRequested);
340         bool required = XMLHelper::getAttrBool(child, false, RequestedAttribute::ISREQUIRED_ATTRIB_NAME);
341         if (required || requested)
342             m_requestedAttrs.push_back(tuple<xstring,xstring,bool>(name,format,required));
343
344         name = child->getAttributeNS(nullptr, _aliases);
345         if (name && *name) {
346             auto_ptr_char aliases(name);
347             string dup(aliases.get());
348             set<string> new_aliases;
349             split(new_aliases, dup, is_space(), algorithm::token_compress_on);
350             set<string>::iterator ru = new_aliases.find("REMOTE_USER");
351             if (ru != new_aliases.end()) {
352                 m_log.warn("skipping alias, REMOTE_USER is a reserved name");
353                 new_aliases.erase(ru);
354             }
355             m_attributeIds.insert(m_attributeIds.end(), new_aliases.begin(), new_aliases.end());
356         }
357
358         child = XMLHelper::getNextSiblingElement(child, shibspconstants::SHIB2ATTRIBUTEMAP_NS, saml1::Attribute::LOCAL_NAME);
359     }
360
361     if (m_metaAttrCaching)
362         m_attrLock.reset(RWLock::create());
363 }
364
365 void XMLExtractorImpl::generateMetadata(SPSSODescriptor& role) const
366 {
367     if (m_requestedAttrs.empty())
368         return;
369     int index = 1;
370     const vector<AttributeConsumingService*>& svcs = const_cast<const SPSSODescriptor*>(&role)->getAttributeConsumingServices();
371     for (vector<AttributeConsumingService*>::const_iterator s =svcs.begin(); s != svcs.end(); ++s) {
372         pair<bool,int> i = (*s)->getIndex();
373         if (i.first && index == i.second)
374             index = i.second + 1;
375     }
376     AttributeConsumingService* svc = AttributeConsumingServiceBuilder::buildAttributeConsumingService();
377     role.getAttributeConsumingServices().push_back(svc);
378     svc->setIndex(index);
379     ServiceName* sn = ServiceNameBuilder::buildServiceName();
380     svc->getServiceNames().push_back(sn);
381     sn->setName(dynamic_cast<EntityDescriptor*>(role.getParent())->getEntityID());
382     static const XMLCh english[] = UNICODE_LITERAL_2(e,n);
383     sn->setLang(english);
384
385     for (vector< tuple<xstring,xstring,bool> >::const_iterator i = m_requestedAttrs.begin(); i != m_requestedAttrs.end(); ++i) {
386         RequestedAttribute* req = RequestedAttributeBuilder::buildRequestedAttribute();
387         svc->getRequestedAttributes().push_back(req);
388         req->setName(i->get<0>().c_str());
389         if (i->get<1>().empty())
390             req->setNameFormat(saml2::Attribute::URI_REFERENCE);
391         else
392             req->setNameFormat(i->get<1>().c_str());
393         if (i->get<2>())
394             req->isRequired(true);
395     }
396 }
397
398 void XMLExtractorImpl::extractAttributes(
399     const Application& application,
400     const char* assertingParty,
401     const char* relyingParty,
402     const NameIdentifier& nameid,
403     ptr_vector<Attribute>& attributes
404     ) const
405 {
406     const XMLCh* format = nameid.getFormat();
407     if (!format || !*format)
408         format = NameIdentifier::UNSPECIFIED;
409     attrmap_t::const_iterator rule;
410     if ((rule = m_attrMap.find(pair<xstring,xstring>(format,xstring()))) != m_attrMap.end()) {
411         auto_ptr<Attribute> a(rule->second.first->decode(nullptr, rule->second.second, &nameid, assertingParty, relyingParty));
412         if (a.get()) {
413             attributes.push_back(a.get());
414             a.release();
415         }
416     }
417     else if (m_log.isDebugEnabled()) {
418         auto_ptr_char temp(format);
419         m_log.debug("skipping unmapped NameIdentifier with format (%s)", temp.get());
420     }
421 }
422
423 void XMLExtractorImpl::extractAttributes(
424     const Application& application,
425     const char* assertingParty,
426     const char* relyingParty,
427     const NameID& nameid,
428     ptr_vector<Attribute>& attributes
429     ) const
430 {
431     const XMLCh* format = nameid.getFormat();
432     if (!format || !*format)
433         format = NameID::UNSPECIFIED;
434     attrmap_t::const_iterator rule;
435     if ((rule = m_attrMap.find(pair<xstring,xstring>(format,xstring()))) != m_attrMap.end()) {
436         auto_ptr<Attribute> a(rule->second.first->decode(nullptr, rule->second.second, &nameid, assertingParty, relyingParty));
437         if (a.get()) {
438             attributes.push_back(a.get());
439             a.release();
440         }
441     }
442     else if (m_log.isDebugEnabled()) {
443         auto_ptr_char temp(format);
444         m_log.debug("skipping unmapped NameID with format (%s)", temp.get());
445     }
446 }
447
448 void XMLExtractorImpl::extractAttributes(
449     const Application& application,
450     const GenericRequest* request,
451     const char* assertingParty,
452     const char* relyingParty,
453     const saml1::Attribute& attr,
454     ptr_vector<Attribute>& attributes
455     ) const
456 {
457     const XMLCh* name = attr.getAttributeName();
458     const XMLCh* format = attr.getAttributeNamespace();
459     if (!name || !*name)
460         return;
461     if (!format || XMLString::equals(format, shibspconstants::SHIB1_ATTRIBUTE_NAMESPACE_URI))
462         format = &chNull;
463     attrmap_t::const_iterator rule;
464     if ((rule = m_attrMap.find(pair<xstring,xstring>(name,format))) != m_attrMap.end()) {
465         auto_ptr<Attribute> a(rule->second.first->decode(request, rule->second.second, &attr, assertingParty, relyingParty));
466         if (a.get()) {
467             attributes.push_back(a.get());
468             a.release();
469         }
470     }
471     else if (m_log.isInfoEnabled()) {
472         auto_ptr_char temp1(name);
473         auto_ptr_char temp2(format);
474         m_log.info("skipping unmapped SAML 1.x Attribute with Name: %s%s%s", temp1.get(), *temp2.get() ? ", Namespace:" : "", temp2.get());
475     }
476 }
477
478 void XMLExtractorImpl::extractAttributes(
479     const Application& application,
480     const GenericRequest* request,
481     const char* assertingParty,
482     const char* relyingParty,
483     const saml2::Attribute& attr,
484     ptr_vector<Attribute>& attributes
485     ) const
486 {
487     const XMLCh* name = attr.getName();
488     const XMLCh* format = attr.getNameFormat();
489     if (!name || !*name)
490         return;
491     if (!format || !*format)
492         format = saml2::Attribute::UNSPECIFIED;
493     else if (XMLString::equals(format, saml2::Attribute::URI_REFERENCE))
494         format = &chNull;
495     attrmap_t::const_iterator rule;
496     if ((rule = m_attrMap.find(pair<xstring,xstring>(name,format))) != m_attrMap.end()) {
497         auto_ptr<Attribute> a(rule->second.first->decode(request, rule->second.second, &attr, assertingParty, relyingParty));
498         if (a.get()) {
499             attributes.push_back(a.get());
500             a.release();
501             return;
502         }
503     }
504     else if (XMLString::equals(format, saml2::Attribute::UNSPECIFIED)) {
505         // As a fallback, if the format is "unspecified", null out the value and re-map.
506         if ((rule = m_attrMap.find(pair<xstring,xstring>(name,xstring()))) != m_attrMap.end()) {
507             auto_ptr<Attribute> a(rule->second.first->decode(request, rule->second.second, &attr, assertingParty, relyingParty));
508             if (a.get()) {
509                 attributes.push_back(a.get());
510                 a.release();
511                 return;
512             }
513         }
514     }
515
516     if (m_log.isInfoEnabled()) {
517         auto_ptr_char temp1(name);
518         auto_ptr_char temp2(format);
519         m_log.info("skipping unmapped SAML 2.0 Attribute with Name: %s%s%s", temp1.get(), *temp2.get() ? ", Format:" : "", temp2.get());
520     }
521 }
522
523 void XMLExtractorImpl::extractAttributes(
524     const Application& application,
525     const GenericRequest* request,
526     const char* assertingParty,
527     const char* relyingParty,
528     const saml1::AttributeStatement& statement,
529     ptr_vector<Attribute>& attributes
530     ) const
531 {
532     static void (XMLExtractorImpl::* extract)(
533         const Application&, const GenericRequest*, const char*, const char*, const saml1::Attribute&, ptr_vector<Attribute>&
534         ) const = &XMLExtractorImpl::extractAttributes;
535     for_each(
536         make_indirect_iterator(statement.getAttributes().begin()), make_indirect_iterator(statement.getAttributes().end()),
537         boost::bind(extract, this, boost::cref(application), request, assertingParty, relyingParty, _1, boost::ref(attributes))
538         );
539 }
540
541 void XMLExtractorImpl::extractAttributes(
542     const Application& application,
543     const GenericRequest* request,
544     const char* assertingParty,
545     const char* relyingParty,
546     const saml2::AttributeStatement& statement,
547     ptr_vector<Attribute>& attributes
548     ) const
549 {
550     static void (XMLExtractorImpl::* extract)(
551         const Application&, const GenericRequest*, const char*, const char*, const saml2::Attribute&, ptr_vector<Attribute>&
552         ) const = &XMLExtractorImpl::extractAttributes;
553     for_each(
554         make_indirect_iterator(statement.getAttributes().begin()), make_indirect_iterator(statement.getAttributes().end()),
555         boost::bind(extract, this, boost::cref(application), request, assertingParty, relyingParty, _1, boost::ref(attributes))
556         );
557 }
558
559 void XMLExtractorImpl::extractAttributes(
560     const Application& application,
561     const GenericRequest* request,
562     const ObservableMetadataProvider* observable,
563     const XMLCh* entityID,
564     const char* relyingParty,
565     const Extensions& ext,
566     ptr_vector<Attribute>& attributes
567     ) const
568 {
569     const vector<XMLObject*>& exts = ext.getUnknownXMLObjects();
570     for (vector<XMLObject*>::const_iterator i = exts.begin(); i != exts.end(); ++i) {
571         const EntityAttributes* container = dynamic_cast<const EntityAttributes*>(*i);
572         if (!container)
573             continue;
574
575         bool useCache = false;
576         map<const ObservableMetadataProvider*,decoded_t>::iterator cacheEntry;
577
578         // Check for cached result.
579         if (observable && m_metaAttrCaching) {
580             m_attrLock->rdlock();
581             cacheEntry = m_decodedMap.find(observable);
582             if (cacheEntry == m_decodedMap.end()) {
583                 // We need to elevate the lock and retry.
584                 m_attrLock->unlock();
585                 m_attrLock->wrlock();
586                 cacheEntry = m_decodedMap.find(observable);
587                 if (cacheEntry == m_decodedMap.end()) {
588                     SharedLock locker(m_attrLock, false);   // guard in case these throw
589
590                     // It's still brand new, so hook it for cache activation.
591                     observable->addObserver(this);
592
593                     // Prime the map reference with an empty decoded map.
594                     cacheEntry = m_decodedMap.insert(make_pair(observable,decoded_t())).first;
595
596                     // Downgrade the lock.
597                     // We don't have to recheck because we never erase the master map entry entirely, even on changes.
598                     locker.release();   // unguard for lock downgrade
599                     m_attrLock->unlock();
600                     m_attrLock->rdlock();
601                 }
602             }
603             useCache = true;
604         }
605
606         if (useCache) {
607             // We're holding the lock, so check the cache.
608             decoded_t::iterator d = cacheEntry->second.find(container);
609             if (d != cacheEntry->second.end()) {
610                 SharedLock locker(m_attrLock, false);   // pop the lock when we're done
611                 for (vector<DDF>::iterator obj = d->second.begin(); obj != d->second.end(); ++obj) {
612                     auto_ptr<Attribute> wrapper(Attribute::unmarshall(*obj));
613                     m_log.debug("recovered cached metadata attribute (%s)", wrapper->getId());
614                     attributes.push_back(wrapper.get());
615                     wrapper.release();
616                 }
617                 break;
618             }
619         }
620
621         // Add a guard for the lock if we're caching.
622         SharedLock locker(useCache ? m_attrLock.get() : nullptr, false);
623
624         // Use a holding area to support caching.
625         ptr_vector<Attribute> holding;
626
627         // Extract attributes into holding area with no asserting party set.
628         static void (XMLExtractorImpl::* extractV2Attr)(
629             const Application&, const GenericRequest*, const char*, const char*, const saml2::Attribute&, ptr_vector<Attribute>&
630             ) const = &XMLExtractorImpl::extractAttributes;
631         for_each(
632             make_indirect_iterator(container->getAttributes().begin()), make_indirect_iterator(container->getAttributes().end()),
633             boost::bind(extractV2Attr, this, boost::ref(application), request, (const char*)nullptr, relyingParty, _1, boost::ref(holding))
634             );
635
636         if (entityID && m_entityAssertions) {
637             const vector<saml2::Assertion*>& asserts = container->getAssertions();
638             for (indirect_iterator<vector<saml2::Assertion*>::const_iterator> assert = make_indirect_iterator(asserts.begin());
639                     assert != make_indirect_iterator(asserts.end()); ++assert) {
640                 if (!(assert->getSignature())) {
641                     if (m_log.isDebugEnabled()) {
642                         auto_ptr_char eid(entityID);
643                         m_log.debug("skipping unsigned assertion in metadata extension for entity (%s)", eid.get());
644                     }
645                     continue;
646                 }
647                 else if (assert->getAttributeStatements().empty()) {
648                     if (m_log.isDebugEnabled()) {
649                         auto_ptr_char eid(entityID);
650                         m_log.debug("skipping assertion with no AttributeStatement in metadata extension for entity (%s)", eid.get());
651                     }
652                     continue;
653                 }
654                 else {
655                     // Check subject.
656                     const NameID* subject = assert->getSubject() ? assert->getSubject()->getNameID() : nullptr;
657                     if (!subject ||
658                             !XMLString::equals(subject->getFormat(), NameID::ENTITY) ||
659                             !XMLString::equals(subject->getName(), entityID)) {
660                         if (m_log.isDebugEnabled()) {
661                             auto_ptr_char eid(entityID);
662                             m_log.debug("skipping assertion with improper Subject in metadata extension for entity (%s)", eid.get());
663                         }
664                         continue;
665                     }
666                 }
667
668                 try {
669                     // Set up and evaluate a policy for an AA asserting attributes to us.
670                     shibsp::SecurityPolicy policy(application, &AttributeAuthorityDescriptor::ELEMENT_QNAME, false, m_policyId.c_str());
671                     Locker locker(m_metadata.get());
672                     if (m_metadata)
673                         policy.setMetadataProvider(m_metadata.get());
674                     if (m_trust)
675                         policy.setTrustEngine(m_trust.get());
676                     // Populate recipient as audience.
677                     const XMLCh* issuer = assert->getIssuer() ? assert->getIssuer()->getName() : nullptr;
678                     policy.getAudiences().push_back(application.getRelyingParty(issuer)->getXMLString("entityID").second);
679
680                     // Extract assertion information for policy.
681                     policy.setMessageID(assert->getID());
682                     policy.setIssueInstant(assert->getIssueInstantEpoch());
683                     policy.setIssuer(assert->getIssuer());
684
685                     // Look up metadata for issuer.
686                     if (policy.getIssuer() && policy.getMetadataProvider()) {
687                         if (policy.getIssuer()->getFormat() && !XMLString::equals(policy.getIssuer()->getFormat(), saml2::NameIDType::ENTITY)) {
688                             m_log.debug("non-system entity issuer, skipping metadata lookup");
689                         }
690                         else {
691                             m_log.debug("searching metadata for entity assertion issuer...");
692                             pair<const EntityDescriptor*,const RoleDescriptor*> lookup;
693                             MetadataProvider::Criteria& mc = policy.getMetadataProviderCriteria();
694                             mc.entityID_unicode = policy.getIssuer()->getName();
695                             mc.role = &AttributeAuthorityDescriptor::ELEMENT_QNAME;
696                             mc.protocol = samlconstants::SAML20P_NS;
697                             lookup = policy.getMetadataProvider()->getEntityDescriptor(mc);
698                             if (!lookup.first) {
699                                 auto_ptr_char iname(policy.getIssuer()->getName());
700                                 m_log.debug("no metadata found, can't establish identity of issuer (%s)", iname.get());
701                             }
702                             else if (!lookup.second) {
703                                 m_log.debug("unable to find compatible AA role in metadata");
704                             }
705                             else {
706                                 policy.setIssuerMetadata(lookup.second);
707                             }
708                         }
709                     }
710
711                     // Authenticate the assertion. We have to clone and marshall it to establish the signature for verification.
712                     scoped_ptr<saml2::Assertion> tokencopy(assert->cloneAssertion());
713                     tokencopy->marshall();
714                     policy.evaluate(*tokencopy);
715                     if (!policy.isAuthenticated()) {
716                         if (m_log.isDebugEnabled()) {
717                             auto_ptr_char tempid(tokencopy->getID());
718                             auto_ptr_char eid(entityID);
719                             m_log.debug(
720                                 "failed to authenticate assertion (%s) in metadata extension for entity (%s)", tempid.get(), eid.get()
721                                 );
722                         }
723                         continue;
724                     }
725
726                     // Override the asserting/relying party names based on this new issuer.
727                     const EntityDescriptor* inlineEntity =
728                         policy.getIssuerMetadata() ? dynamic_cast<const EntityDescriptor*>(policy.getIssuerMetadata()->getParent()) : nullptr;
729                     auto_ptr_char inlineAssertingParty(inlineEntity ? inlineEntity->getEntityID() : nullptr);
730                     relyingParty = application.getRelyingParty(inlineEntity)->getString("entityID").second;
731
732                     // Use a private holding area for filtering purposes.
733                     ptr_vector<Attribute> holding2;
734                     const vector<saml2::Attribute*>& attrs2 =
735                         const_cast<const saml2::AttributeStatement*>(tokencopy->getAttributeStatements().front())->getAttributes();
736                     for_each(
737                         make_indirect_iterator(attrs2.begin()), make_indirect_iterator(attrs2.end()),
738                         boost::bind(extractV2Attr, this, boost::ref(application), request, inlineAssertingParty.get(), relyingParty, _1, boost::ref(holding2))
739                         );
740
741                     // Now we locally filter the attributes so that the actual issuer can be properly set.
742                     // If we relied on outside filtering, the attributes couldn't be distinguished from the
743                     // ones that come from the user's IdP.
744                     if (m_filter && !holding2.empty()) {
745
746                         // The filter API uses an unsafe container, so we have to transfer everything into one and back.
747                         vector<Attribute*> unsafe_holding2;
748
749                         // Use a local exception context since the container is unsafe.
750                         try {
751                             while (!holding2.empty()) {
752                                 ptr_vector<Attribute>::auto_type ptr = holding2.pop_back();
753                                 unsafe_holding2.push_back(ptr.get());
754                                 ptr.release();
755                             }
756                             BasicFilteringContext fc(application, unsafe_holding2, policy.getIssuerMetadata());
757                             Locker filtlocker(m_filter.get());
758                             m_filter->filterAttributes(fc, unsafe_holding2);
759
760                             // Transfer back to safe container
761                             while (!unsafe_holding2.empty()) {
762                                 auto_ptr<Attribute> ptr(unsafe_holding2.back());
763                                 unsafe_holding2.pop_back();
764                                 holding2.push_back(ptr.get());
765                                 ptr.release();
766                             }
767                         }
768                         catch (std::exception& ex) {
769                             m_log.error("caught exception filtering attributes: %s", ex.what());
770                             m_log.error("dumping extracted attributes due to filtering exception");
771                             for_each(unsafe_holding2.begin(), unsafe_holding2.end(), xmltooling::cleanup<Attribute>());
772                             holding2.clear();   // in case the exception was during transfer between containers
773                         }
774                     }
775
776                     if (!holding2.empty()) {
777                         // Copy them over to the main holding tank, which transfers ownership.
778                         holding.transfer(holding.end(), holding2);
779                     }
780                 }
781                 catch (std::exception& ex) {
782                     // Known exceptions are handled gracefully by skipping the assertion.
783                     if (m_log.isDebugEnabled()) {
784                         auto_ptr_char tempid(assert->getID());
785                         auto_ptr_char eid(entityID);
786                         m_log.debug(
787                             "exception authenticating assertion (%s) in metadata extension for entity (%s): %s",
788                             tempid.get(),
789                             eid.get(),
790                             ex.what()
791                             );
792                     }
793                     continue;
794                 }
795             }
796         }
797
798         if (!holding.empty()) {
799             if (useCache) {
800                 locker.release();   // unguard to upgrade lock
801                 m_attrLock->unlock();
802                 m_attrLock->wrlock();
803                 SharedLock locker2(m_attrLock, false);   // pop the lock when we're done
804                 if (cacheEntry->second.count(container) == 0) {
805                     static void (vector<DDF>::* push_back)(DDF const &) = &vector<DDF>::push_back;
806                     vector<DDF>& marshalled = cacheEntry->second[container];
807                     for_each(
808                         holding.begin(), holding.end(),
809                         boost::bind(push_back, boost::ref(marshalled), boost::bind(&Attribute::marshall, _1))
810                         );
811                 }
812             }
813
814             // Copy them to the output parameter, which transfers ownership.
815             attributes.transfer(attributes.end(), holding);
816         }
817
818         // If the lock is held, it's guarded.
819
820         break;  // only process a single extension element
821     }
822 }
823
824 void XMLExtractor::extractAttributes(
825     const Application& application, const GenericRequest* request, const RoleDescriptor* issuer, const XMLObject& xmlObject, vector<Attribute*>& attributes
826     ) const
827 {
828     if (!m_impl)
829         return;
830
831     ptr_vector<Attribute> holding;
832     extractAttributes(application, request, issuer, xmlObject, holding);
833
834     // Transfer ownership from the ptr_vector to the unsafe vector for API compatibility.
835     // Any throws should leave each container in a consistent state. The holding container
836     // is freed by us, and the result container by the caller.
837     while (!holding.empty()) {
838         ptr_vector<Attribute>::auto_type ptr = holding.pop_back();
839         attributes.push_back(ptr.get());
840         ptr.release();
841     }
842 }
843
844 void XMLExtractor::extractAttributes(
845     const Application& application, const GenericRequest* request, const RoleDescriptor* issuer, const XMLObject& xmlObject, ptr_vector<Attribute>& attributes
846     ) const
847 {
848     static void (XMLExtractor::* extractEncrypted)(
849         const Application&, const GenericRequest*, const RoleDescriptor*, const XMLObject&, ptr_vector<Attribute>&
850         ) const = &XMLExtractor::extractAttributes;
851     static void (XMLExtractorImpl::* extractV1Statement)(
852         const Application&, const GenericRequest*, const char*, const char*, const saml1::AttributeStatement&, ptr_vector<Attribute>&
853         ) const = &XMLExtractorImpl::extractAttributes;
854
855     const EntityDescriptor* entity = issuer ? dynamic_cast<const EntityDescriptor*>(issuer->getParent()) : nullptr;
856     const char* relyingParty = application.getRelyingParty(entity)->getString("entityID").second;
857
858     // Check for statements.
859     if (XMLString::equals(xmlObject.getElementQName().getLocalPart(), saml1::AttributeStatement::LOCAL_NAME)) {
860         const saml2::AttributeStatement* statement2 = dynamic_cast<const saml2::AttributeStatement*>(&xmlObject);
861         if (statement2) {
862             auto_ptr_char assertingParty(entity ? entity->getEntityID() : nullptr);
863             m_impl->extractAttributes(application, request, assertingParty.get(), relyingParty, *statement2, attributes);
864             // Handle EncryptedAttributes inline so we have access to the role descriptor.
865             const vector<saml2::EncryptedAttribute*>& encattrs = statement2->getEncryptedAttributes();
866             for_each(
867                 make_indirect_iterator(encattrs.begin()), make_indirect_iterator(encattrs.end()),
868                 boost::bind(extractEncrypted, this, boost::ref(application), request, issuer, _1, boost::ref(attributes))
869                 );
870             return;
871         }
872
873         const saml1::AttributeStatement* statement1 = dynamic_cast<const saml1::AttributeStatement*>(&xmlObject);
874         if (statement1) {
875             auto_ptr_char assertingParty(entity ? entity->getEntityID() : nullptr);
876             m_impl->extractAttributes(application, request, assertingParty.get(), relyingParty, *statement1, attributes);
877             return;
878         }
879
880         throw AttributeExtractionException("Unable to extract attributes, unknown object type.");
881     }
882
883     // Check for assertions.
884     if (XMLString::equals(xmlObject.getElementQName().getLocalPart(), saml1::Assertion::LOCAL_NAME)) {
885         const saml2::Assertion* token2 = dynamic_cast<const saml2::Assertion*>(&xmlObject);
886         if (token2) {
887             auto_ptr_char assertingParty(entity ? entity->getEntityID() : nullptr);
888             const vector<saml2::AttributeStatement*>& statements = token2->getAttributeStatements();
889             for (indirect_iterator<vector<saml2::AttributeStatement*>::const_iterator> s = make_indirect_iterator(statements.begin());
890                     s != make_indirect_iterator(statements.end()); ++s) {
891                 m_impl->extractAttributes(application, request, assertingParty.get(), relyingParty, *s, attributes);
892                 // Handle EncryptedAttributes inline so we have access to the role descriptor.
893                 const vector<saml2::EncryptedAttribute*>& encattrs = const_cast<const saml2::AttributeStatement&>(*s).getEncryptedAttributes();
894                 for_each(
895                     make_indirect_iterator(encattrs.begin()), make_indirect_iterator(encattrs.end()),
896                     boost::bind(extractEncrypted, this, boost::ref(application), request, issuer, _1, boost::ref(attributes))
897                     );
898             }
899             return;
900         }
901
902         const saml1::Assertion* token1 = dynamic_cast<const saml1::Assertion*>(&xmlObject);
903         if (token1) {
904             auto_ptr_char assertingParty(entity ? entity->getEntityID() : nullptr);
905             const vector<saml1::AttributeStatement*>& statements = token1->getAttributeStatements();
906             for_each(make_indirect_iterator(statements.begin()), make_indirect_iterator(statements.end()),
907                 boost::bind(extractV1Statement, m_impl.get(), boost::ref(application), request, assertingParty.get(), relyingParty, _1, boost::ref(attributes))
908                 );
909             return;
910         }
911
912         throw AttributeExtractionException("Unable to extract attributes, unknown object type.");
913     }
914
915     // Check for metadata.
916     if (XMLString::equals(xmlObject.getElementQName().getNamespaceURI(), samlconstants::SAML20MD_NS)) {
917         const RoleDescriptor* roleToExtract = dynamic_cast<const RoleDescriptor*>(&xmlObject);
918         const EntityDescriptor* entityToExtract = roleToExtract ? dynamic_cast<const EntityDescriptor*>(roleToExtract->getParent()) : nullptr;
919         if (!entityToExtract)
920             throw AttributeExtractionException("Unable to extract attributes, unknown metadata object type.");
921         const Extensions* ext = entityToExtract->getExtensions();
922         if (ext) {
923             m_impl->extractAttributes(
924                 application,
925                 request,
926                 dynamic_cast<const ObservableMetadataProvider*>(application.getMetadataProvider(false)),
927                 entityToExtract->getEntityID(),
928                 relyingParty,
929                 *ext,
930                 attributes
931                 );
932         }
933         const EntitiesDescriptor* group = dynamic_cast<const EntitiesDescriptor*>(entityToExtract->getParent());
934         while (group) {
935             ext = group->getExtensions();
936             if (ext) {
937                 m_impl->extractAttributes(
938                     application,
939                     request,
940                     dynamic_cast<const ObservableMetadataProvider*>(application.getMetadataProvider(false)),
941                     nullptr,   // not an entity, so inline assertions won't be processed
942                     relyingParty,
943                     *ext,
944                     attributes
945                     );
946             }
947             group = dynamic_cast<const EntitiesDescriptor*>(group->getParent());
948         }
949         return;
950     }
951
952     // Check for attributes.
953     if (XMLString::equals(xmlObject.getElementQName().getLocalPart(), saml1::Attribute::LOCAL_NAME)) {
954         auto_ptr_char assertingParty(entity ? entity->getEntityID() : nullptr);
955         const saml2::Attribute* attr2 = dynamic_cast<const saml2::Attribute*>(&xmlObject);
956         if (attr2)
957             return m_impl->extractAttributes(application, request, assertingParty.get(), relyingParty, *attr2, attributes);
958
959         const saml1::Attribute* attr1 = dynamic_cast<const saml1::Attribute*>(&xmlObject);
960         if (attr1)
961             return m_impl->extractAttributes(application, request, assertingParty.get(), relyingParty, *attr1, attributes);
962
963         throw AttributeExtractionException("Unable to extract attributes, unknown object type.");
964     }
965
966     if (XMLString::equals(xmlObject.getElementQName().getLocalPart(), EncryptedAttribute::LOCAL_NAME)) {
967         const EncryptedAttribute* encattr = dynamic_cast<const EncryptedAttribute*>(&xmlObject);
968         if (encattr) {
969             const XMLCh* recipient = application.getXMLString("entityID").second;
970             CredentialResolver* cr = application.getCredentialResolver();
971             if (!cr) {
972                 m_log.warn("found encrypted attribute, but no CredentialResolver was available");
973                 return;
974             }
975
976             try {
977                 Locker credlocker(cr);
978                 if (issuer) {
979                     MetadataCredentialCriteria mcc(*issuer);
980                     scoped_ptr<XMLObject> decrypted(encattr->decrypt(*cr, recipient, &mcc));
981                     if (m_log.isDebugEnabled())
982                         m_log.debugStream() << "decrypted Attribute: " << *decrypted << logging::eol;
983                     return extractAttributes(application, request, issuer, *decrypted, attributes);
984                 }
985                 else {
986                     scoped_ptr<XMLObject> decrypted(encattr->decrypt(*cr, recipient));
987                     if (m_log.isDebugEnabled())
988                         m_log.debugStream() << "decrypted Attribute: " << *decrypted << logging::eol;
989                     return extractAttributes(application, request, issuer, *decrypted, attributes);
990                 }
991             }
992             catch (std::exception& ex) {
993                 m_log.error("caught exception decrypting Attribute: %s", ex.what());
994                 return;
995             }
996         }
997     }
998
999     // Check for NameIDs.
1000     const NameID* name2 = dynamic_cast<const NameID*>(&xmlObject);
1001     if (name2) {
1002         auto_ptr_char assertingParty(entity ? entity->getEntityID() : nullptr);
1003         return m_impl->extractAttributes(application, assertingParty.get(), relyingParty, *name2, attributes);
1004     }
1005
1006     const NameIdentifier* name1 = dynamic_cast<const NameIdentifier*>(&xmlObject);
1007     if (name1) {
1008         auto_ptr_char assertingParty(entity ? entity->getEntityID() : nullptr);
1009         return m_impl->extractAttributes(application, assertingParty.get(), relyingParty, *name1, attributes);
1010     }
1011
1012     m_log.debug("unable to extract attributes, unknown XML object type: %s", xmlObject.getElementQName().toString().c_str());
1013 }
1014
1015 pair<bool,DOMElement*> XMLExtractor::background_load()
1016 {
1017     // Load from source using base class.
1018     pair<bool,DOMElement*> raw = ReloadableXMLFile::load();
1019
1020     // If we own it, wrap it.
1021     XercesJanitor<DOMDocument> docjanitor(raw.first ? raw.second->getOwnerDocument() : nullptr);
1022
1023     scoped_ptr<XMLExtractorImpl> impl(new XMLExtractorImpl(raw.second, m_log));
1024
1025     // If we held the document, transfer it to the impl. If we didn't, it's a no-op.
1026     impl->setDocument(docjanitor.release());
1027
1028     // Perform the swap inside a lock.
1029     if (m_lock)
1030         m_lock->wrlock();
1031     SharedLock locker(m_lock, false);
1032     m_impl.swap(impl);
1033
1034     return make_pair(false,(DOMElement*)nullptr);
1035 }