X-Git-Url: http://www.project-moonshot.org/gitweb/?a=blobdiff_plain;f=xmltooling%2Futil%2FParserPool.cpp;h=ee64178e1803f4f748e71aa79289cb76cedc2498;hb=c9a063e554871e212edc46a1fd9ae865cbc9da9e;hp=8a22b6e4f76fe85da18621587014f6222add096a;hpb=dd11b998f20c8d80c9e7ecda0d8983610c7c9589;p=shibboleth%2Fcpp-xmltooling.git

diff --git a/xmltooling/util/ParserPool.cpp b/xmltooling/util/ParserPool.cpp
index 8a22b6e..ee64178 100644
--- a/xmltooling/util/ParserPool.cpp
+++ b/xmltooling/util/ParserPool.cpp
@@ -1,5 +1,5 @@
 /*
- *  Copyright 2001-2006 Internet2
+ *  Copyright 2001-2008 Internet2
  * 
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -17,11 +17,13 @@
 /**
  * ParserPool.cpp
  * 
- * XML parsing
+ * A thread-safe pool of parsers that share characteristics.
  */
 
 #include "internal.h"
 #include "exceptions.h"
+#include "logging.h"
+#include "util/CurlURLInputStream.h"
 #include "util/NDC.h"
 #include "util/ParserPool.h"
 #include "util/XMLHelper.h"
@@ -30,7 +32,6 @@
 #include <functional>
 #include <sys/types.h>
 #include <sys/stat.h>
-#include <log4cpp/Category.hh>
 #include <xercesc/util/PlatformUtils.hpp>
 #include <xercesc/util/XMLUniDefs.hpp>
 #include <xercesc/sax/SAXException.hpp>
@@ -38,12 +39,63 @@
 #include <xercesc/framework/LocalFileInputSource.hpp>
 #include <xercesc/framework/Wrapper4InputSource.hpp>
 
+using namespace xmltooling::logging;
 using namespace xmltooling;
+using namespace xercesc;
 using namespace std;
-using namespace log4cpp;
+
+
+namespace {
+    class MyErrorHandler : public DOMErrorHandler {
+    public:
+        unsigned int errors;
+
+        MyErrorHandler() : errors(0) {}
+
+        bool handleError(const DOMError& e)
+        {
+#ifdef _DEBUG
+            xmltooling::NDC ndc("handleError");
+#endif
+            Category& log=Category::getInstance(XMLTOOLING_LOGCAT".ParserPool");
+
+            DOMLocator* locator=e.getLocation();
+            auto_ptr_char temp(e.getMessage());
+
+            switch (e.getSeverity()) {
+                case DOMError::DOM_SEVERITY_WARNING:
+                    log.warnStream() << "warning on line " << locator->getLineNumber()
+                        << ", column " << locator->getColumnNumber()
+                        << ", message: " << temp.get() << logging::eol;
+                    return true;
+
+                case DOMError::DOM_SEVERITY_ERROR:
+                    ++errors;
+                    log.errorStream() << "error on line " << locator->getLineNumber()
+                        << ", column " << locator->getColumnNumber()
+                        << ", message: " << temp.get() << logging::eol;
+                    return true;
+
+                case DOMError::DOM_SEVERITY_FATAL_ERROR:
+                    ++errors;
+                    log.errorStream() << "fatal error on line " << locator->getLineNumber()
+                        << ", column " << locator->getColumnNumber()
+                        << ", message: " << temp.get() << logging::eol;
+                    return true;
+            }
+
+            ++errors;
+            log.errorStream() << "undefined error type on line " << locator->getLineNumber()
+                << ", column " << locator->getColumnNumber()
+                << ", message: " << temp.get() << logging::eol;
+            return false;
+        }
+    };
+}
+
 
 ParserPool::ParserPool(bool namespaceAware, bool schemaAware)
-    : m_namespaceAware(namespaceAware), m_schemaAware(schemaAware), m_lock(Mutex::create()) {}
+    : m_namespaceAware(namespaceAware), m_schemaAware(schemaAware), m_lock(Mutex::create()), m_security(new SecurityManager()) {}
 
 ParserPool::~ParserPool()
 {
@@ -52,6 +104,7 @@ ParserPool::~ParserPool()
         m_pool.pop();
     }
     delete m_lock;
+    delete m_security;
 }
 
 DOMDocument* ParserPool::newDocument()
@@ -59,26 +112,78 @@ DOMDocument* ParserPool::newDocument()
     return DOMImplementationRegistry::getDOMImplementation(NULL)->createDocument();
 }
 
+#ifdef XMLTOOLING_XERCESC_COMPLIANT_DOMLS
+
+DOMDocument* ParserPool::parse(DOMLSInput& domsrc)
+{
+    DOMLSParser* parser=checkoutBuilder();
+    XercesJanitor<DOMLSParser> janitor(parser);
+    try {
+        MyErrorHandler deh;
+        parser->getDomConfig()->setParameter(XMLUni::fgDOMErrorHandler, dynamic_cast<DOMErrorHandler*>(&deh));
+        DOMDocument* doc=parser->parse(&domsrc);
+        if (deh.errors) {
+            if (doc)
+                doc->release();
+            throw XMLParserException("XML error(s) during parsing, check log for specifics");
+        }
+        parser->getDomConfig()->setParameter(XMLUni::fgDOMErrorHandler, (void*)NULL);
+        parser->getDomConfig()->setParameter(XMLUni::fgXercesUserAdoptsDOMDocument, true);
+        checkinBuilder(janitor.release());
+        return doc;
+    }
+    catch (XMLException& ex) {
+        parser->getDomConfig()->setParameter(XMLUni::fgDOMErrorHandler, (void*)NULL);
+        parser->getDomConfig()->setParameter(XMLUni::fgXercesUserAdoptsDOMDocument, true);
+        checkinBuilder(janitor.release());
+        auto_ptr_char temp(ex.getMessage());
+        throw XMLParserException(string("Xerces error during parsing: ") + (temp.get() ? temp.get() : "no message"));
+    }
+    catch (XMLToolingException&) {
+        parser->getDomConfig()->setParameter(XMLUni::fgDOMErrorHandler, (void*)NULL);
+        parser->getDomConfig()->setParameter(XMLUni::fgXercesUserAdoptsDOMDocument, true);
+        checkinBuilder(janitor.release());
+        throw;
+    }
+}
+
+#else
+
 DOMDocument* ParserPool::parse(DOMInputSource& domsrc)
 {
     DOMBuilder* parser=checkoutBuilder();
     XercesJanitor<DOMBuilder> janitor(parser);
     try {
+        MyErrorHandler deh;
+        parser->setErrorHandler(&deh);
         DOMDocument* doc=parser->parse(domsrc);
-        parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument,true);
+        if (deh.errors) {
+            if (doc)
+                doc->release();
+            throw XMLParserException("XML error(s) during parsing, check log for specifics");
+        }
+        parser->setErrorHandler(NULL);
+        parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument, true);
         checkinBuilder(janitor.release());
         return doc;
     }
-    catch (XMLException&) {
+    catch (XMLException& ex) {
+        parser->setErrorHandler(NULL);
+        parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument, true);
         checkinBuilder(janitor.release());
-        throw;
+        auto_ptr_char temp(ex.getMessage());
+        throw XMLParserException(string("Xerces error during parsing: ") + (temp.get() ? temp.get() : "no message"));
     }
     catch (XMLToolingException&) {
+        parser->setErrorHandler(NULL);
+        parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument, true);
         checkinBuilder(janitor.release());
         throw;
     }
 }
 
+#endif
+
 DOMDocument* ParserPool::parse(istream& is)
 {
     StreamInputSource src(is);
@@ -91,7 +196,7 @@ template <class T> class doubleit
 {
 public:
     doubleit(T& t, const typename T::value_type& s) : temp(t), sep(s) {}
-    void operator() (const pair<T,T>& s) { temp += s.first + sep + s.first + sep; }
+    void operator() (const pair<const T,T>& s) { temp += s.first + sep + s.first + sep; }
     T& temp;
     const typename T::value_type& sep;
 };
@@ -141,9 +246,10 @@ bool ParserPool::loadCatalog(const XMLCh* pathname)
     Category& log=Category::getInstance(XMLTOOLING_LOGCAT".ParserPool");
 
     // XML constants
-    static const XMLCh catalog[] = { chLatin_c, chLatin_a, chLatin_t, chLatin_a, chLatin_l, chLatin_o, chLatin_g, chNull };
-    static const XMLCh uri[] = { chLatin_u, chLatin_r, chLatin_i, chNull };
-    static const XMLCh name[] = { chLatin_n, chLatin_a, chLatin_m, chLatin_e, chNull };
+    static const XMLCh catalog[] =  UNICODE_LITERAL_7(c,a,t,a,l,o,g);
+    static const XMLCh system[] =   UNICODE_LITERAL_6(s,y,s,t,e,m);
+    static const XMLCh systemId[] = UNICODE_LITERAL_8(s,y,s,t,e,m,I,d);
+    static const XMLCh uri[] =      UNICODE_LITERAL_3(u,r,i);
     static const XMLCh CATALOG_NS[] = {
         chLatin_u, chLatin_r, chLatin_n, chColon,
         chLatin_o, chLatin_a, chLatin_s, chLatin_i, chLatin_s, chColon,
@@ -176,12 +282,12 @@ bool ParserPool::loadCatalog(const XMLCh* pathname)
             return false;
         }
         
-        // Fetch all the <uri> elements.
-        DOMNodeList* mappings=root->getElementsByTagNameNS(CATALOG_NS,uri);
+        // Fetch all the <system> elements.
+        DOMNodeList* mappings=root->getElementsByTagNameNS(CATALOG_NS,system);
         Lock lock(m_lock);
         for (XMLSize_t i=0; i<mappings->getLength(); i++) {
             root=static_cast<DOMElement*>(mappings->item(i));
-            const XMLCh* from=root->getAttributeNS(NULL,name);
+            const XMLCh* from=root->getAttributeNS(NULL,systemId);
             const XMLCh* to=root->getAttributeNS(NULL,uri);
 #ifdef HAVE_GOOD_STL
             m_schemaLocMap[from]=to;
@@ -198,15 +304,27 @@ bool ParserPool::loadCatalog(const XMLCh* pathname)
         for_each(m_schemaLocMap.begin(),m_schemaLocMap.end(),doubleit<string>(m_schemaLocations,' '));
 #endif
     }
-    catch (XMLParserException& e) {
-        log.error("catalog loader caught XMLParserException: %s", e.what());
+    catch (exception& e) {
+        log.error("catalog loader caught exception: %s", e.what());
         return false;
     }
 
     return true;
 }
 
-DOMInputSource* ParserPool::resolveEntity(const XMLCh* const publicId, const XMLCh* const systemId, const XMLCh* const baseURI)
+#ifdef XMLTOOLING_XERCESC_COMPLIANT_DOMLS
+DOMLSInput* ParserPool::resolveResource(
+            const XMLCh *const resourceType,
+            const XMLCh *const namespaceUri,
+            const XMLCh *const publicId,
+            const XMLCh *const systemId,
+            const XMLCh *const baseURI
+            )
+#else
+DOMInputSource* ParserPool::resolveEntity(
+    const XMLCh* const publicId, const XMLCh* const systemId, const XMLCh* const baseURI
+    )
+#endif
 {
 #if _DEBUG
     xmltooling::NDC ndc("resolveEntity");
@@ -221,69 +339,122 @@ DOMInputSource* ParserPool::resolveEntity(const XMLCh* const publicId, const XML
         log.debug("asked to resolve %s with baseURI %s",sysId.get(),base.get() ? base.get() : "(null)");
     }
 
-    // Find well-known schemas in the specified location.
 #ifdef HAVE_GOOD_STL
+    // Find well-known schemas in the specified location.
     map<xstring,xstring>::const_iterator i=m_schemaLocMap.find(systemId);
     if (i!=m_schemaLocMap.end())
-        return new Wrapper4InputSource(new LocalFileInputSource(NULL,i->second.c_str()));
+        return new Wrapper4InputSource(new LocalFileInputSource(baseURI,i->second.c_str()));
+
+    // Check for entity as a value in the map.
+    for (i=m_schemaLocMap.begin(); i!=m_schemaLocMap.end(); ++i) {
+        if (XMLString::endsWith(i->second.c_str(), systemId))
+            return new Wrapper4InputSource(new LocalFileInputSource(baseURI,i->second.c_str()));
+    }
+
+    // We'll allow anything without embedded slashes.
+    if (XMLString::indexOf(systemId, chForwardSlash)==-1)
+        return new Wrapper4InputSource(new LocalFileInputSource(baseURI,systemId));
 #else
+    // Find well-known schemas in the specified location.
     auto_ptr_char temp(systemId);
     map<string,string>::const_iterator i=m_schemaLocMap.find(temp.get());
-    auto_ptr_XMLCh temp2(i->second.c_str());
-    if (i!=m_schemaLocMap.end())
-        return new Wrapper4InputSource(new LocalFileInputSource(NULL,temp2.get()));
+    if (i!=m_schemaLocMap.end()) {
+        auto_ptr_XMLCh temp2(i->second.c_str());
+        return new Wrapper4InputSource(new LocalFileInputSource(baseURI,temp2.get()));
+    }
+
+    // Check for entity as a value in the map.
+    for (i=m_schemaLocMap.begin(); i!=m_schemaLocMap.end(); ++i) {
+        auto_ptr_XMLCh temp2(i->second.c_str());
+        if (XMLString::endsWith(temp2.get(), systemId))
+            return new Wrapper4InputSource(new LocalFileInputSource(baseURI,temp2.get()));
+    }
+
+    // We'll allow anything without embedded slashes.
+    if (XMLString::indexOf(systemId, chForwardSlash)==-1)
+        return new Wrapper4InputSource(new LocalFileInputSource(baseURI,systemId));
 #endif    
 
     // Shortcircuit the request.
-    auto_ptr_char sysId(systemId);
-    log.warn("unauthorized entity request (%s), blocking it", sysId.get() ? sysId.get() : "no systemId");
+#ifdef HAVE_GOOD_STL
+    auto_ptr_char temp(systemId);
+#endif
+    log.debug("unauthorized entity request (%s), blocking it", temp.get());
     static const XMLByte nullbuf[] = {0};
     return new Wrapper4InputSource(new MemBufInputSource(nullbuf,0,systemId));
 }
 
-bool ParserPool::handleError(const DOMError& e)
+#ifdef XMLTOOLING_XERCESC_COMPLIANT_DOMLS
+
+DOMLSParser* ParserPool::createBuilder()
 {
-#ifdef _DEBUG
-    xmltooling::NDC ndc("handleError");
+    static const XMLCh impltype[] = { chLatin_L, chLatin_S, chNull };
+    DOMImplementation* impl=DOMImplementationRegistry::getDOMImplementation(impltype);
+    DOMLSParser* parser=static_cast<DOMImplementationLS*>(impl)->createLSParser(DOMImplementationLS::MODE_SYNCHRONOUS,NULL);
+    parser->getDomConfig()->setParameter(XMLUni::fgDOMNamespaces, m_namespaceAware);
+    if (m_schemaAware) {
+        parser->getDomConfig()->setParameter(XMLUni::fgDOMNamespaces, true);
+        parser->getDomConfig()->setParameter(XMLUni::fgXercesSchema, true);
+        parser->getDomConfig()->setParameter(XMLUni::fgDOMValidate, true);
+        parser->getDomConfig()->setParameter(XMLUni::fgXercesCacheGrammarFromParse, true);
+        
+        // We build a "fake" schema location hint that binds each namespace to itself.
+        // This ensures the entity resolver will be given the namespace as a systemId it can check. 
+#ifdef HAVE_GOOD_STL
+        parser->getDomConfig()->setParameter(XMLUni::fgXercesSchemaExternalSchemaLocation, const_cast<XMLCh*>(m_schemaLocations.c_str()));
+#else
+        auto_ptr_XMLCh temp(m_schemaLocations.c_str());
+        parser->getDomConfig()->setParameter(XMLUni::fgXercesSchemaExternalSchemaLocation, const_cast<XMLCh*>(temp.get()));
 #endif
-    Category& log=Category::getInstance(XMLTOOLING_LOGCAT".ParserPool");
-    DOMLocator* locator=e.getLocation();
-    auto_ptr_char temp(e.getMessage());
-
-    switch (e.getSeverity()) {
-        case DOMError::DOM_SEVERITY_WARNING:
-            log.warnStream() << "warning on line " << locator->getLineNumber()
-                << ", column " << locator->getColumnNumber()
-                << ", message: " << temp.get() << CategoryStream::ENDLINE;
-            return true;
+    }
+    parser->getDomConfig()->setParameter(XMLUni::fgXercesUserAdoptsDOMDocument, true);
+    parser->getDomConfig()->setParameter(XMLUni::fgXercesDisableDefaultEntityResolution, true);
+    parser->getDomConfig()->setParameter(XMLUni::fgDOMResourceResolver, dynamic_cast<DOMLSResourceResolver*>(this));
+    parser->getDomConfig()->setParameter(XMLUni::fgXercesSecurityManager, m_security);
+    return parser;
+}
 
-        case DOMError::DOM_SEVERITY_ERROR:
-            log.errorStream() << "error on line " << locator->getLineNumber()
-                << ", column " << locator->getColumnNumber()
-                << ", message: " << temp.get() << CategoryStream::ENDLINE;
-            throw XMLParserException(string("error during XML parsing: ") + (temp.get() ? temp.get() : "no message"));
+DOMLSParser* ParserPool::checkoutBuilder()
+{
+    Lock lock(m_lock);
+    if (m_pool.empty()) {
+        DOMLSParser* builder=createBuilder();
+        return builder;
+    }
+    DOMLSParser* p=m_pool.top();
+    m_pool.pop();
+    if (m_schemaAware) {
+#ifdef HAVE_GOOD_STL
+        p->getDomConfig()->setParameter(XMLUni::fgXercesSchemaExternalSchemaLocation, const_cast<XMLCh*>(m_schemaLocations.c_str()));
+#else
+        auto_ptr_XMLCh temp2(m_schemaLocations.c_str());
+        p->getDomConfig()->setParameter(XMLUni::fgXercesSchemaExternalSchemaLocation, const_cast<XMLCh*>(temp2.get()));
+#endif
+    }
+    return p;
+}
 
-        case DOMError::DOM_SEVERITY_FATAL_ERROR:
-            log.critStream() << "fatal error on line " << locator->getLineNumber()
-                << ", column " << locator->getColumnNumber()
-                << ", message: " << temp.get() << CategoryStream::ENDLINE;
-            throw XMLParserException(string("fatal error during XML parsing: ") + (temp.get() ? temp.get() : "no message"));
+void ParserPool::checkinBuilder(DOMLSParser* builder)
+{
+    if (builder) {
+        Lock lock(m_lock);
+        m_pool.push(builder);
     }
-    throw XMLParserException(string("unclassified error during XML parsing: ") + (temp.get() ? temp.get() : "no message"));
 }
 
+#else
+
 DOMBuilder* ParserPool::createBuilder()
 {
     static const XMLCh impltype[] = { chLatin_L, chLatin_S, chNull };
     DOMImplementation* impl=DOMImplementationRegistry::getDOMImplementation(impltype);
     DOMBuilder* parser=static_cast<DOMImplementationLS*>(impl)->createDOMBuilder(DOMImplementationLS::MODE_SYNCHRONOUS,0);
-    if (m_namespaceAware)
-        parser->setFeature(XMLUni::fgDOMNamespaces,true);
+    parser->setFeature(XMLUni::fgDOMNamespaces, m_namespaceAware);
     if (m_schemaAware) {
-        parser->setFeature(XMLUni::fgXercesSchema,true);
-        parser->setFeature(XMLUni::fgDOMValidation,true);
-        parser->setFeature(XMLUni::fgXercesCacheGrammarFromParse,true);
-        parser->setFeature(XMLUni::fgXercesValidationErrorAsFatal,true);
+        parser->setFeature(XMLUni::fgDOMNamespaces, true);
+        parser->setFeature(XMLUni::fgXercesSchema, true);
+        parser->setFeature(XMLUni::fgDOMValidation, true);
+        parser->setFeature(XMLUni::fgXercesCacheGrammarFromParse, true);
         
         // We build a "fake" schema location hint that binds each namespace to itself.
         // This ensures the entity resolver will be given the namespace as a systemId it can check. 
@@ -294,9 +465,10 @@ DOMBuilder* ParserPool::createBuilder()
         parser->setProperty(XMLUni::fgXercesSchemaExternalSchemaLocation,const_cast<XMLCh*>(temp.get()));
 #endif
     }
-    parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument,true);
+    parser->setProperty(XMLUni::fgXercesSecurityManager, m_security);
+    parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument, true);
+    parser->setFeature(XMLUni::fgXercesDisableDefaultEntityResolution, true);
     parser->setEntityResolver(this);
-    parser->setErrorHandler(this);
     return parser;
 }
 
@@ -328,10 +500,12 @@ void ParserPool::checkinBuilder(DOMBuilder* builder)
     }
 }
 
-unsigned int StreamInputSource::StreamBinInputStream::readBytes(XMLByte* const toFill, const unsigned int maxToRead)
+#endif
+
+xsecsize_t StreamInputSource::StreamBinInputStream::readBytes(XMLByte* const toFill, const xsecsize_t maxToRead)
 {
     XMLByte* target=toFill;
-    unsigned int bytes_read=0,request=maxToRead;
+    xsecsize_t bytes_read=0,request=maxToRead;
 
     // Fulfill the rest by reading from the stream.
     if (request && !m_is.eof() && !m_is.fail()) {
@@ -343,10 +517,56 @@ unsigned int StreamInputSource::StreamBinInputStream::readBytes(XMLByte* const t
         catch(ios_base::failure& e) {
             Category::getInstance(XMLTOOLING_LOGCAT".StreamInputSource").critStream()
                 << "XML::StreamInputSource::StreamBinInputStream::readBytes caught an exception: " << e.what()
-                << CategoryStream::ENDLINE;
+                << logging::eol;
             *toFill=0;
             return 0;
         }
     }
     return bytes_read;
 }
+
+#ifdef XMLTOOLING_LITE
+
+URLInputSource::URLInputSource(const XMLCh* url, const char* systemId) : InputSource(systemId), m_url(url)
+{
+}
+
+URLInputSource::URLInputSource(const DOMElement* e, const char* systemId) : InputSource(systemId)
+{
+    static const XMLCh uri[] = UNICODE_LITERAL_3(u,r,i);
+    static const XMLCh url[] = UNICODE_LITERAL_3(u,r,l);
+
+    const XMLCh* attr = e->getAttributeNS(NULL, url);
+    if (!attr || !*attr) {
+        attr = e->getAttributeNS(NULL, uri);
+        if (!attr || !*attr)
+            throw IOException("No URL supplied via DOM to URLInputSource constructor.");
+    }
+
+    m_url.setURL(attr);
+}
+
+BinInputStream* URLInputSource::makeStream() const
+{
+    // Ask the URL to create us an appropriate input stream
+    return m_url.makeNewStream();
+}
+
+#else
+
+URLInputSource::URLInputSource(const XMLCh* url, const char* systemId)
+    : InputSource(systemId), m_url(url), m_root(NULL)
+{
+}
+
+URLInputSource::URLInputSource(const DOMElement* e, const char* systemId)
+    : InputSource(systemId), m_root(e)
+{
+}
+
+BinInputStream* URLInputSource::makeStream() const
+{
+    return m_root ? new CurlURLInputStream(m_root) : new CurlURLInputStream(m_url.get());
+}
+
+#endif