2 * Copyright 2001-2006 Internet2
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
8 * http://www.apache.org/licenses/LICENSE-2.0
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
24 #include "exceptions.h"
26 #include "util/ParserPool.h"
27 #include "util/XMLHelper.h"
31 #include <sys/types.h>
33 #include <log4cpp/Category.hh>
34 #include <xercesc/util/PlatformUtils.hpp>
35 #include <xercesc/util/XMLUniDefs.hpp>
36 #include <xercesc/sax/SAXException.hpp>
37 #include <xercesc/framework/MemBufInputSource.hpp>
38 #include <xercesc/framework/LocalFileInputSource.hpp>
39 #include <xercesc/framework/Wrapper4InputSource.hpp>
41 using namespace xmltooling;
43 using namespace log4cpp;
45 ParserPool::ParserPool(bool namespaceAware, bool schemaAware)
46 : m_namespaceAware(namespaceAware), m_schemaAware(schemaAware), m_lock(XMLPlatformUtils::makeMutex()) {}
48 ParserPool::~ParserPool()
50 while(!m_pool.empty()) {
51 m_pool.top()->release();
54 XMLPlatformUtils::closeMutex(m_lock);
57 DOMDocument* ParserPool::newDocument()
59 return DOMImplementationRegistry::getDOMImplementation(NULL)->createDocument();
62 DOMDocument* ParserPool::parse(DOMInputSource& domsrc)
64 DOMBuilder* parser=checkoutBuilder();
66 DOMDocument* doc=parser->parse(domsrc);
67 parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument,true);
68 checkinBuilder(parser);
72 checkinBuilder(parser);
77 DOMDocument* ParserPool::parse(istream& is)
79 StreamInputSource src(is);
80 Wrapper4InputSource domsrc(&src,false);
84 // Functor to double its argument separated by a character and append to a buffer
85 template <class T> class doubleit
88 doubleit(T& t, const typename T::value_type& s) : temp(t), sep(s) {}
89 void operator() (const pair<T,T>& s) { temp += s.first + sep + s.first + sep; }
91 const typename T::value_type& sep;
94 bool ParserPool::loadSchema(const XMLCh* nsURI, const XMLCh* pathname)
96 // Just check the pathname and then directly register the pair into the map.
98 auto_ptr_char p(pathname);
100 struct _stat stat_buf;
101 if (_stat(p.get(), &stat_buf) != 0)
103 struct stat stat_buf;
104 if (stat(p.get(), &stat_buf) != 0)
108 xmltooling::NDC ndc("loadSchema");
110 Category& log=Category::getInstance(XMLTOOLING_LOGCAT".ParserPool");
111 auto_ptr_char n(nsURI);
112 log.error("failed to load schema for (%s), file not found (%s)",n.get(),p.get());
116 XMLPlatformUtils::lockMutex(m_lock);
118 m_schemaLocMap[nsURI]=pathname;
119 m_schemaLocations.erase();
120 for_each(m_schemaLocMap.begin(),m_schemaLocMap.end(),doubleit<xstring>(m_schemaLocations,chSpace));
122 auto_ptr_char n(nsURI);
123 m_schemaLocMap[n.get()]=p.get();
124 m_schemaLocations.erase();
125 for_each(m_schemaLocMap.begin(),m_schemaLocMap.end(),doubleit<string>(m_schemaLocations,' '));
127 XMLPlatformUtils::unlockMutex(m_lock);
132 bool ParserPool::loadCatalog(const XMLCh* pathname)
135 xmltooling::NDC ndc("loadCatalog");
137 Category& log=Category::getInstance(XMLTOOLING_LOGCAT".ParserPool");
140 static const XMLCh catalog[] = { chLatin_c, chLatin_a, chLatin_t, chLatin_a, chLatin_l, chLatin_o, chLatin_g, chNull };
141 static const XMLCh uri[] = { chLatin_u, chLatin_r, chLatin_i, chNull };
142 static const XMLCh name[] = { chLatin_n, chLatin_a, chLatin_m, chLatin_e, chNull };
143 static const XMLCh CATALOG_NS[] = {
144 chLatin_u, chLatin_r, chLatin_n, chColon,
145 chLatin_o, chLatin_a, chLatin_s, chLatin_i, chLatin_s, chColon,
146 chLatin_n, chLatin_a, chLatin_m, chLatin_e, chLatin_s, chColon,
147 chLatin_t, chLatin_c, chColon,
148 chLatin_e, chLatin_n, chLatin_t, chLatin_i, chLatin_t, chLatin_y, chColon,
149 chLatin_x, chLatin_m, chLatin_l, chLatin_n, chLatin_s, chColon,
150 chLatin_x, chLatin_m, chLatin_l, chColon,
151 chLatin_c, chLatin_a, chLatin_t, chLatin_a, chLatin_l, chLatin_o, chLatin_g, chNull
154 // Parse the catalog with the internal parser pool.
156 if (log.isDebugEnabled()) {
157 auto_ptr_char temp(pathname);
158 log.debug("loading XML catalog from %s", temp.get());
161 LocalFileInputSource fsrc(NULL,pathname);
162 Wrapper4InputSource domsrc(&fsrc,false);
164 DOMDocument* doc=XMLToolingConfig::getConfig().getParser().parse(domsrc);
166 // Check root element.
167 const DOMElement* root=doc->getDocumentElement();
168 if (!XMLHelper::isNodeNamed(root,CATALOG_NS,catalog)) {
169 auto_ptr_char temp(pathname);
170 log.error("unknown root element, failed to load XML catalog from %s", temp.get());
175 // Fetch all the <uri> elements.
176 DOMNodeList* mappings=root->getElementsByTagNameNS(CATALOG_NS,uri);
177 XMLPlatformUtils::lockMutex(m_lock);
178 for (XMLSize_t i=0; i<mappings->getLength(); i++) {
179 root=static_cast<DOMElement*>(mappings->item(i));
180 const XMLCh* from=root->getAttributeNS(NULL,name);
181 const XMLCh* to=root->getAttributeNS(NULL,uri);
183 m_schemaLocMap[from]=to;
185 auto_ptr_char f(from);
187 m_schemaLocMap[f.get()]=t.get();
190 m_schemaLocations.erase();
192 for_each(m_schemaLocMap.begin(),m_schemaLocMap.end(),doubleit<xstring>(m_schemaLocations,chSpace));
194 for_each(m_schemaLocMap.begin(),m_schemaLocMap.end(),doubleit<string>(m_schemaLocations,' '));
196 XMLPlatformUtils::unlockMutex(m_lock);
199 catch (XMLParserException& e) {
200 log.error("catalog loader caught XMLParserException: %s", e.what());
207 DOMInputSource* ParserPool::resolveEntity(const XMLCh* const publicId, const XMLCh* const systemId, const XMLCh* const baseURI)
210 xmltooling::NDC ndc("resolveEntity");
215 Category& log=Category::getInstance(XMLTOOLING_LOGCAT".ParserPool");
216 if (log.isDebugEnabled()) {
217 auto_ptr_char sysId(systemId);
218 auto_ptr_char base(baseURI);
219 log.debug("asked to resolve %s with baseURI %s",sysId.get(),base.get() ? base.get() : "(null)");
222 // Find well-known schemas in the specified location.
224 map<xstring,xstring>::const_iterator i=m_schemaLocMap.find(systemId);
225 if (i!=m_schemaLocMap.end())
226 return new Wrapper4InputSource(new LocalFileInputSource(NULL,i->second.c_str()));
228 auto_ptr_char temp(systemId);
229 map<string,string>::const_iterator i=m_schemaLocMap.find(temp.get());
230 auto_ptr_XMLCh temp2(i->second.c_str());
231 if (i!=m_schemaLocMap.end())
232 return new Wrapper4InputSource(new LocalFileInputSource(NULL,temp2.get()));
235 // Shortcircuit the request.
236 auto_ptr_char sysId(systemId);
237 log.warn("unauthorized entity request (%s), blocking it", sysId.get() ? sysId.get() : "no systemId");
238 static const XMLByte nullbuf[] = {0};
239 return new Wrapper4InputSource(new MemBufInputSource(nullbuf,0,systemId));
242 bool ParserPool::handleError(const DOMError& e)
245 xmltooling::NDC ndc("handleError");
247 Category& log=Category::getInstance(XMLTOOLING_LOGCAT".ParserPool");
248 DOMLocator* locator=e.getLocation();
249 auto_ptr_char temp(e.getMessage());
251 switch (e.getSeverity()) {
252 case DOMError::DOM_SEVERITY_WARNING:
253 log.warnStream() << "warning on line " << locator->getLineNumber()
254 << ", column " << locator->getColumnNumber()
255 << ", message: " << temp.get() << CategoryStream::ENDLINE;
258 case DOMError::DOM_SEVERITY_ERROR:
259 log.errorStream() << "error on line " << locator->getLineNumber()
260 << ", column " << locator->getColumnNumber()
261 << ", message: " << temp.get() << CategoryStream::ENDLINE;
262 throw XMLParserException(string("error during XML parsing: ") + (temp.get() ? temp.get() : "no message"));
264 case DOMError::DOM_SEVERITY_FATAL_ERROR:
265 log.critStream() << "fatal error on line " << locator->getLineNumber()
266 << ", column " << locator->getColumnNumber()
267 << ", message: " << temp.get() << CategoryStream::ENDLINE;
268 throw XMLParserException(string("fatal error during XML parsing: ") + (temp.get() ? temp.get() : "no message"));
270 throw XMLParserException(string("unclassified error during XML parsing: ") + (temp.get() ? temp.get() : "no message"));
273 DOMBuilder* ParserPool::createBuilder()
275 static const XMLCh impltype[] = { chLatin_L, chLatin_S, chNull };
276 DOMImplementation* impl=DOMImplementationRegistry::getDOMImplementation(impltype);
277 DOMBuilder* parser=static_cast<DOMImplementationLS*>(impl)->createDOMBuilder(DOMImplementationLS::MODE_SYNCHRONOUS,0);
278 if (m_namespaceAware)
279 parser->setFeature(XMLUni::fgDOMNamespaces,true);
281 parser->setFeature(XMLUni::fgXercesSchema,true);
282 parser->setFeature(XMLUni::fgDOMValidation,true);
283 parser->setFeature(XMLUni::fgXercesCacheGrammarFromParse,true);
284 parser->setFeature(XMLUni::fgXercesValidationErrorAsFatal,true);
286 // We build a "fake" schema location hint that binds each namespace to itself.
287 // This ensures the entity resolver will be given the namespace as a systemId it can check.
289 parser->setProperty(XMLUni::fgXercesSchemaExternalSchemaLocation,const_cast<XMLCh*>(m_schemaLocations.c_str()));
291 auto_ptr_XMLCh temp(m_schemaLocations.c_str());
292 parser->setProperty(XMLUni::fgXercesSchemaExternalSchemaLocation,const_cast<XMLCh*>(temp.get()));
295 parser->setFeature(XMLUni::fgXercesUserAdoptsDOMDocument,true);
296 parser->setEntityResolver(this);
297 parser->setErrorHandler(this);
301 DOMBuilder* ParserPool::checkoutBuilder()
303 XMLPlatformUtils::lockMutex(m_lock);
305 if (m_pool.empty()) {
306 DOMBuilder* builder=createBuilder();
307 XMLPlatformUtils::unlockMutex(m_lock);
310 DOMBuilder* p=m_pool.top();
314 p->setProperty(XMLUni::fgXercesSchemaExternalSchemaLocation,const_cast<XMLCh*>(m_schemaLocations.c_str()));
316 auto_ptr_XMLCh temp2(m_schemaLocations.c_str());
317 p->setProperty(XMLUni::fgXercesSchemaExternalSchemaLocation,const_cast<XMLCh*>(temp2.get()));
320 XMLPlatformUtils::unlockMutex(m_lock);
324 XMLPlatformUtils::unlockMutex(m_lock);
329 void ParserPool::checkinBuilder(DOMBuilder* builder)
332 XMLPlatformUtils::lockMutex(m_lock);
333 m_pool.push(builder);
334 XMLPlatformUtils::unlockMutex(m_lock);
338 unsigned int StreamInputSource::StreamBinInputStream::readBytes(XMLByte* const toFill, const unsigned int maxToRead)
340 XMLByte* target=toFill;
341 unsigned int bytes_read=0,request=maxToRead;
343 // Fulfill the rest by reading from the stream.
344 if (request && !m_is.eof()) {
346 m_is.read(reinterpret_cast<char* const>(target),request);
347 m_pos+=m_is.gcount();
348 bytes_read+=m_is.gcount();
351 Category::getInstance(XMLTOOLING_LOGCAT".StreamInputSource").critStream() <<
352 "XML::StreamInputSource::StreamBinInputStream::readBytes caught an exception" << CategoryStream::ENDLINE;