From 80b03acfa067ab2a4e4f724373211ff1320d33f2 Mon Sep 17 00:00:00 2001
From: Khaled Noaman <knoaman@apache.org>
Date: Tue, 7 Dec 2004 19:45:43 +0000
Subject: [PATCH] An option to ignore a cached DTD grammar when a document
contains an internal and external subset.
git-svn-id: https://svn.apache.org/repos/asf/xerces/c/trunk@176232 13f79535-47bb-0310-9956-ffa450edef68
---
src/xercesc/internal/DGXMLScanner.cpp | 5 +++-
src/xercesc/internal/IGXMLScanner.cpp | 7 +++--
src/xercesc/internal/XMLScanner.cpp | 4 ++-
src/xercesc/internal/XMLScanner.hpp | 17 +++++++++++++
src/xercesc/parsers/DOMBuilderImpl.cpp | 12 +++++++--
src/xercesc/parsers/SAX2XMLReaderImpl.cpp | 10 ++++++++
src/xercesc/parsers/SAXParser.cpp | 14 ++++++++++
src/xercesc/parsers/SAXParser.hpp | 31 ++++++++++++++++++++++-
src/xercesc/parsers/XercesDOMParser.cpp | 9 +++++++
src/xercesc/parsers/XercesDOMParser.hpp | 25 ++++++++++++++++++
src/xercesc/util/XMLUni.cpp | 15 +++++++++++
src/xercesc/util/XMLUni.hpp | 1 +
12 files changed, 143 insertions(+), 7 deletions(-)
diff --git a/src/xercesc/internal/DGXMLScanner.cpp b/src/xercesc/internal/DGXMLScanner.cpp
index 60d8b0cc2..f6cedef24 100644
--- a/src/xercesc/internal/DGXMLScanner.cpp
+++ b/src/xercesc/internal/DGXMLScanner.cpp
@@ -972,7 +972,10 @@ void DGXMLScanner::scanDocTypeDecl()
InputSource* srcUsed=0;
Janitor<InputSource> janSrc(srcUsed);
- if (fUseCachedGrammar)
+ // If we had an internal subset and we're using the cached grammar, it
+ // means that the ignoreCachedDTD is set, so we ignore the cached
+ // grammar
+ if (fUseCachedGrammar && !hasIntSubset)
{
srcUsed = resolveSystemId(sysId, pubId);
janSrc.reset(srcUsed);
diff --git a/src/xercesc/internal/IGXMLScanner.cpp b/src/xercesc/internal/IGXMLScanner.cpp
index faf8a25b6..47d09fa12 100644
--- a/src/xercesc/internal/IGXMLScanner.cpp
+++ b/src/xercesc/internal/IGXMLScanner.cpp
@@ -1266,7 +1266,7 @@ void IGXMLScanner::scanDocTypeDecl()
Janitor<DTDElementDecl> rootDeclJanitor(rootDecl);
rootDecl->setCreateReason(DTDElementDecl::AsRootElem);
rootDecl->setExternalElemDeclaration(true);
- if(!fUseCachedGrammar)
+ if(!fUseCachedGrammar)
{
// this will break getRootElemId on DTDGrammar when
// cached grammars are in use, but
@@ -1423,7 +1423,10 @@ void IGXMLScanner::scanDocTypeDecl()
InputSource* srcUsed=0;
Janitor<InputSource> janSrc(srcUsed);
- if (fUseCachedGrammar)
+ // If we had an internal subset and we're using the cached grammar, it
+ // means that the ignoreCachedDTD is set, so we ignore the cached
+ // grammar
+ if (fUseCachedGrammar && !hasIntSubset)
{
srcUsed = resolveSystemId(sysId, pubId);
janSrc.reset(srcUsed);
diff --git a/src/xercesc/internal/XMLScanner.cpp b/src/xercesc/internal/XMLScanner.cpp
index b46f4387c..d0f25980f 100644
--- a/src/xercesc/internal/XMLScanner.cpp
+++ b/src/xercesc/internal/XMLScanner.cpp
@@ -160,6 +160,7 @@ XMLScanner::XMLScanner(XMLValidator* const valToAdopt,
, fNormalizeData(true)
, fGenerateSyntheticAnnotations(false)
, fValidateAnnotations(false)
+ , fIgnoreCachedDTD(false)
, fErrorCount(0)
, fEntityExpansionLimit(0)
, fEntityExpansionCount(0)
@@ -243,6 +244,7 @@ XMLScanner::XMLScanner( XMLDocumentHandler* const docHandler
, fNormalizeData(true)
, fGenerateSyntheticAnnotations(false)
, fValidateAnnotations(false)
+ , fIgnoreCachedDTD(false)
, fErrorCount(0)
, fEntityExpansionLimit(0)
, fEntityExpansionCount(0)
@@ -1752,7 +1754,7 @@ void XMLScanner::checkInternalDTD(bool hasExtSubset
if (fToCacheGrammar)
ThrowXMLwithMemMgr(RuntimeException, XMLExcepts::Val_CantHaveIntSS, fMemoryManager);
- if (fUseCachedGrammar && hasExtSubset )
+ if (fUseCachedGrammar && hasExtSubset && !fIgnoreCachedDTD)
{
InputSource* sysIdSrc = resolveSystemId(sysId, pubId);
Janitor<InputSource> janSysIdSrc(sysIdSrc);
diff --git a/src/xercesc/internal/XMLScanner.hpp b/src/xercesc/internal/XMLScanner.hpp
index 51b5e3d9a..2b55544b2 100644
--- a/src/xercesc/internal/XMLScanner.hpp
+++ b/src/xercesc/internal/XMLScanner.hpp
@@ -16,6 +16,10 @@
/*
* $Log$
+ * Revision 1.45 2004/12/07 19:45:43 knoaman
+ * An option to ignore a cached DTD grammar when a document contains an
+ * internal and external subset.
+ *
* Revision 1.44 2004/12/03 19:40:30 cargilld
* Change call to resolveEntity to pass in public id so that only one call to resolveEntity is needed (a follow-on to Alberto's fix).
*
@@ -546,6 +550,7 @@ public :
bool getGenerateSyntheticAnnotations() const;
bool getValidateAnnotations() const;
+ bool getIgnoreCachedDTD() const;
// -----------------------------------------------------------------------
// Getter methods
@@ -643,6 +648,7 @@ public :
void setGenerateSyntheticAnnotations(const bool newValue);
void setValidateAnnotations(const bool newValue);
+ void setIgnoredCachedDTD(const bool newValue);
// -----------------------------------------------------------------------
// Mutator methods
@@ -1010,6 +1016,7 @@ protected:
bool fNormalizeData;
bool fGenerateSyntheticAnnotations;
bool fValidateAnnotations;
+ bool fIgnoreCachedDTD;
int fErrorCount;
unsigned int fEntityExpansionLimit;
unsigned int fEntityExpansionCount;
@@ -1366,6 +1373,11 @@ inline bool XMLScanner::getValidateAnnotations() const
return fValidateAnnotations;
}
+inline bool XMLScanner::getIgnoreCachedDTD() const
+{
+ return fIgnoreCachedDTD;
+}
+
// ---------------------------------------------------------------------------
// XMLScanner: Setter methods
// ---------------------------------------------------------------------------
@@ -1535,6 +1547,11 @@ inline void XMLScanner::setInputBufferSize(const size_t bufferSize)
fCDataBuf.setFullHandler(this, fBufferSize);
}
+inline void XMLScanner::setIgnoredCachedDTD(const bool newValue)
+{
+ fIgnoreCachedDTD = newValue;
+}
+
// ---------------------------------------------------------------------------
// XMLScanner: Mutator methods
// ---------------------------------------------------------------------------
diff --git a/src/xercesc/parsers/DOMBuilderImpl.cpp b/src/xercesc/parsers/DOMBuilderImpl.cpp
index 7a69254f9..a978290ba 100644
--- a/src/xercesc/parsers/DOMBuilderImpl.cpp
+++ b/src/xercesc/parsers/DOMBuilderImpl.cpp
@@ -243,7 +243,10 @@ void DOMBuilderImpl::setFeature(const XMLCh* const name, const bool state)
{
getScanner()->setIdentityConstraintChecking(state);
}
-
+ else if (XMLString::compareIString(name, XMLUni::fgXercesIgnoreCachedDTD) == 0)
+ {
+ getScanner()->setIgnoredCachedDTD(state);
+ }
else {
throw DOMException(DOMException::NOT_FOUND_ERR, 0, getMemoryManager());
}
@@ -342,6 +345,10 @@ bool DOMBuilderImpl::getFeature(const XMLCh* const name) const
{
return getScanner()->getValidateAnnotations();
}
+ else if (XMLString::compareIString(name, XMLUni::fgXercesIgnoreCachedDTD) == 0)
+ {
+ return getScanner()->getIgnoreCachedDTD();
+ }
else {
throw DOMException(DOMException::NOT_FOUND_ERR, 0, getMemoryManager());
}
@@ -365,7 +372,8 @@ bool DOMBuilderImpl::canSetFeature(const XMLCh* const name, const bool state) co
(XMLString::compareIString(name, XMLUni::fgXercesDOMHasPSVIInfo) == 0) ||
(XMLString::compareIString(name, XMLUni::fgXercesValidateAnnotations) == 0) ||
(XMLString::compareIString(name, XMLUni::fgXercesGenerateSyntheticAnnotations) == 0) ||
- (XMLString::compareIString(name, XMLUni::fgXercesIdentityConstraintChecking) == 0)
+ (XMLString::compareIString(name, XMLUni::fgXercesIdentityConstraintChecking) == 0) ||
+ (XMLString::compareIString(name, XMLUni::fgXercesIgnoreCachedDTD) == 0)
) {
return true;
}
diff --git a/src/xercesc/parsers/SAX2XMLReaderImpl.cpp b/src/xercesc/parsers/SAX2XMLReaderImpl.cpp
index d439c4c93..e0a3709cb 100644
--- a/src/xercesc/parsers/SAX2XMLReaderImpl.cpp
+++ b/src/xercesc/parsers/SAX2XMLReaderImpl.cpp
@@ -16,6 +16,10 @@
/*
* $Log$
+ * Revision 1.41 2004/12/07 19:45:43 knoaman
+ * An option to ignore a cached DTD grammar when a document contains an
+ * internal and external subset.
+ *
* Revision 1.40 2004/10/04 11:30:51 amassari
* As start/endPrefixMapping doesn't use the XMLBufMgr variable, we need only one XMLBuffer
*
@@ -1573,6 +1577,10 @@ void SAX2XMLReaderImpl::setFeature(const XMLCh* const name, const bool value)
{
fScanner->setValidateAnnotations(value);
}
+ else if (XMLString::compareIString(name, XMLUni::fgXercesIgnoreCachedDTD) == 0)
+ {
+ fScanner->setIgnoredCachedDTD(value);
+ }
else
throw SAXNotRecognizedException("Unknown Feature", fMemoryManager);
}
@@ -1611,6 +1619,8 @@ bool SAX2XMLReaderImpl::getFeature(const XMLCh* const name) const
return fScanner->getGenerateSyntheticAnnotations();
else if (XMLString::compareIString(name, XMLUni::fgXercesValidateAnnotations) == 0)
return fScanner->getValidateAnnotations();
+ else if (XMLString::compareIString(name, XMLUni::fgXercesIgnoreCachedDTD) == 0)
+ return fScanner->getIgnoreCachedDTD();
else
throw SAXNotRecognizedException("Unknown Feature", fMemoryManager);
diff --git a/src/xercesc/parsers/SAXParser.cpp b/src/xercesc/parsers/SAXParser.cpp
index 1395d1ef0..e00bff31b 100644
--- a/src/xercesc/parsers/SAXParser.cpp
+++ b/src/xercesc/parsers/SAXParser.cpp
@@ -16,6 +16,10 @@
/*
* $Log$
+ * Revision 1.37 2004/12/07 19:45:43 knoaman
+ * An option to ignore a cached DTD grammar when a document contains an
+ * internal and external subset.
+ *
* Revision 1.36 2004/09/29 19:00:29 peiyongz
* [jira1207] --patch from Dan Rosen
*
@@ -568,6 +572,11 @@ unsigned int SAXParser::getSrcOffset() const
return fScanner->getSrcOffset();
}
+bool SAXParser::getIgnoreCachedDTD() const
+{
+ return fScanner->getIgnoreCachedDTD();
+}
+
// ---------------------------------------------------------------------------
// SAXParser: Setter methods
// ---------------------------------------------------------------------------
@@ -704,6 +713,11 @@ void SAXParser::setInputBufferSize(const size_t bufferSize)
fScanner->setInputBufferSize(bufferSize);
}
+void SAXParser::setIgnoreCachedDTD(const bool newValue)
+{
+ fScanner->setIgnoredCachedDTD(newValue);
+}
+
// ---------------------------------------------------------------------------
// SAXParser: Overrides of the SAX Parser interface
// ---------------------------------------------------------------------------
diff --git a/src/xercesc/parsers/SAXParser.hpp b/src/xercesc/parsers/SAXParser.hpp
index 83f06962f..c194dc776 100644
--- a/src/xercesc/parsers/SAXParser.hpp
+++ b/src/xercesc/parsers/SAXParser.hpp
@@ -16,6 +16,10 @@
/*
* $Log$
+ * Revision 1.37 2004/12/07 19:45:43 knoaman
+ * An option to ignore a cached DTD grammar when a document contains an
+ * internal and external subset.
+ *
* Revision 1.36 2004/09/29 19:28:12 cargilld
* Mark SAXParser as deprecated.
*
@@ -680,6 +684,15 @@ public :
*/
bool getValidateAnnotations() const;
+ /** Get the 'ignore cached DTD grammar' flag
+ *
+ * @return true, if the parser is currently configured to
+ * ignore cached DTD, false otherwise.
+ *
+ * @see #setIgnoreCachedDTD
+ */
+ bool getIgnoreCachedDTD() const;
+
//@}
@@ -948,7 +961,7 @@ public :
* instead of building the grammar from scratch, to validate XML
* documents.
*
- * If the 'Grammar caching' flag is set to true, this mehod ignore the
+ * If the 'Grammar caching' flag is set to true, this method ignores the
* value passed in.
*
* The parser's default state is: false.
@@ -1008,6 +1021,22 @@ public :
*/
void setInputBufferSize(const size_t bufferSize);
+ /** Set the 'ignore cached DTD grammar' flag
+ *
+ * This method gives users the option to ignore a cached DTD grammar, when
+ * an XML document contains both an internal and external DTD, and the use
+ * cached grammar from parse option is enabled. Currently, we do not allow
+ * using cached DTD grammar when an internal subset is present in the
+ * document. This option will only affect the behavior of the parser when
+ * an internal and external DTD both exist in a document (i.e. no effect
+ * if document has no internal subset).
+ *
+ * The parser's default state is false
+ *
+ * @param newValue The state to set
+ */
+ void setIgnoreCachedDTD(const bool newValue);
+
//@}
diff --git a/src/xercesc/parsers/XercesDOMParser.cpp b/src/xercesc/parsers/XercesDOMParser.cpp
index 416e09248..8e54f7cbd 100644
--- a/src/xercesc/parsers/XercesDOMParser.cpp
+++ b/src/xercesc/parsers/XercesDOMParser.cpp
@@ -98,6 +98,11 @@ unsigned int XercesDOMParser::getSrcOffset() const
return getScanner()->getSrcOffset();
}
+bool XercesDOMParser::getIgnoreCachedDTD() const
+{
+ return getScanner()->getIgnoreCachedDTD();
+}
+
// ---------------------------------------------------------------------------
// XercesDOMParser: Setter methods
// ---------------------------------------------------------------------------
@@ -153,6 +158,10 @@ void XercesDOMParser::useCachedGrammarInParse(const bool newState)
getScanner()->useCachedGrammarInParse(newState);
}
+void XercesDOMParser::setIgnoreCachedDTD(const bool newValue)
+{
+ getScanner()->setIgnoredCachedDTD(newValue);
+}
// ---------------------------------------------------------------------------
// XercesDOMParser: Utilities
diff --git a/src/xercesc/parsers/XercesDOMParser.hpp b/src/xercesc/parsers/XercesDOMParser.hpp
index 22f1f2d16..88bbe2310 100644
--- a/src/xercesc/parsers/XercesDOMParser.hpp
+++ b/src/xercesc/parsers/XercesDOMParser.hpp
@@ -200,6 +200,15 @@ public :
*/
unsigned int getSrcOffset() const;
+ /** Get the 'ignore cached DTD grammar' flag
+ *
+ * @return true, if the parser is currently configured to
+ * ignore cached DTD, false otherwise.
+ *
+ * @see #setIgnoreCachedDTD
+ */
+ bool getIgnoreCachedDTD() const;
+
//@}
@@ -301,6 +310,22 @@ public :
*/
void useCachedGrammarInParse(const bool newState);
+ /** Set the 'ignore cached DTD grammar' flag
+ *
+ * This method gives users the option to ignore a cached DTD grammar, when
+ * an XML document contains both an internal and external DTD, and the use
+ * cached grammar from parse option is enabled. Currently, we do not allow
+ * using cached DTD grammar when an internal subset is present in the
+ * document. This option will only affect the behavior of the parser when
+ * an internal and external DTD both exist in a document (i.e. no effect
+ * if document has no internal subset).
+ *
+ * The parser's default state is false
+ *
+ * @param newValue The state to set
+ */
+ void setIgnoreCachedDTD(const bool newValue);
+
//@}
// -----------------------------------------------------------------------
diff --git a/src/xercesc/util/XMLUni.cpp b/src/xercesc/util/XMLUni.cpp
index ab0823015..b2382a368 100644
--- a/src/xercesc/util/XMLUni.cpp
+++ b/src/xercesc/util/XMLUni.cpp
@@ -1258,6 +1258,21 @@ const XMLCh XMLUni::fgXercesUseCachedGrammarInParse[] =
, chLatin_s, chLatin_e, chNull
};
+//Xerces: http://apache.org/xml/features/validation/ignoreCachedDTD
+const XMLCh XMLUni::fgXercesIgnoreCachedDTD[] =
+{
+ chLatin_h, chLatin_t, chLatin_t, chLatin_p, chColon, chForwardSlash
+ , chForwardSlash, chLatin_a, chLatin_p, chLatin_a, chLatin_c, chLatin_h
+ , chLatin_e, chPeriod, chLatin_o, chLatin_r, chLatin_g, chForwardSlash
+ , chLatin_x, chLatin_m, chLatin_l, chForwardSlash, chLatin_f, chLatin_e
+ , chLatin_a, chLatin_t, chLatin_u, chLatin_r, chLatin_e, chLatin_s
+ , chForwardSlash, chLatin_v, chLatin_a, chLatin_l, chLatin_i, chLatin_d
+ , chLatin_a, chLatin_t, chLatin_i, chLatin_o, chLatin_n, chForwardSlash
+ , chLatin_i, chLatin_g, chLatin_n, chLatin_o, chLatin_r, chLatin_e
+ , chLatin_C, chLatin_a, chLatin_c, chLatin_h, chLatin_e, chLatin_d
+ , chLatin_D, chLatin_T, chLatin_D, chNull
+};
+
//Introduced in DOM Level 3
const XMLCh XMLUni::fgDOMCanonicalForm[] =
diff --git a/src/xercesc/util/XMLUni.hpp b/src/xercesc/util/XMLUni.hpp
index db029d422..001d33108 100644
--- a/src/xercesc/util/XMLUni.hpp
+++ b/src/xercesc/util/XMLUni.hpp
@@ -223,6 +223,7 @@ public :
static const XMLCh fgXercesDOMHasPSVIInfo[];
static const XMLCh fgXercesGenerateSyntheticAnnotations[];
static const XMLCh fgXercesValidateAnnotations[];
+ static const XMLCh fgXercesIgnoreCachedDTD[];
// SAX2 features/properties names
--
GitLab