cargilld 2005/04/04 08:11:38
Modified: c/src/xercesc/internal DGXMLScanner.cpp IGXMLScanner.cpp
IGXMLScanner.hpp IGXMLScanner2.cpp ReaderMgr.hpp
SGXMLScanner.cpp SGXMLScanner.hpp WFXMLScanner.cpp
XMLReader.cpp XMLReader.hpp XSAXMLScanner.cpp
Log:
Fix a problem where illegal qualified names were not reported as errors.
Also store the colon position when searching for a qualified name to avoid
looking it up again.
Revision Changes Path
1.62 +19 -30 xml-xerces/c/src/xercesc/internal/DGXMLScanner.cpp
Index: DGXMLScanner.cpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/DGXMLScanner.cpp,v
retrieving revision 1.61
retrieving revision 1.62
diff -u -r1.61 -r1.62
--- DGXMLScanner.cpp 12 Jan 2005 10:36:43 -0000 1.61
+++ DGXMLScanner.cpp 4 Apr 2005 15:11:37 -0000 1.62
@@ -786,8 +786,10 @@
// Get a name from the input, which should be the name of the root
// element of the upcoming content.
- fReaderMgr.getName(bbRootName.getBuffer());
- if (bbRootName.isEmpty())
+ int colonPosition;
+ bool validName = fDoNamespaces ?
fReaderMgr.getQName(bbRootName.getBuffer(), &colonPosition) :
+
fReaderMgr.getName(bbRootName.getBuffer());
+ if (!validName)
{
emitError(XMLErrs::NoRootElemInDOCTYPE);
fReaderMgr.skipPastChar(chCloseAngle);
@@ -1094,7 +1096,11 @@
// Get the QName. In this case, we are not doing namespaces, so we just
// use it as is and don't have to break it into parts.
- if (!fReaderMgr.getName(fQNameBuf))
+
+ int colonPosition;
+ bool validName = fDoNamespaces ? fReaderMgr.getQName(fQNameBuf,
&colonPosition) :
+ fReaderMgr.getName(fQNameBuf);
+ if (!validName)
{
emitError(XMLErrs::ExpectedElementName);
fReaderMgr.skipToChar(chOpenAngle);
@@ -1244,7 +1250,10 @@
{
// Assume its going to be an attribute, so get a name from
// the input.
- if (!fReaderMgr.getName(fAttNameBuf))
+
+ validName = fDoNamespaces ? fReaderMgr.getQName(fAttNameBuf,
&colonPosition) :
+ fReaderMgr.getName(fAttNameBuf);
+ if (!validName)
{
emitError(XMLErrs::ExpectedAttrName);
fReaderMgr.skipPastChar(chCloseAngle);
@@ -1472,30 +1481,6 @@
}
}
- if (fDoNamespaces)
- {
- // Make sure that the name is basically well formed for
namespace
- // enabled rules. It either has no colons, or it has one
which
- // is neither the first or last char.
- const int colonFirst =
XMLString::indexOf(fAttNameBuf.getRawBuffer(), chColon);
- if (colonFirst != -1)
- {
- const int colonLast =
XMLString::lastIndexOf(fAttNameBuf.getRawBuffer(), chColon);
-
- if (colonFirst != colonLast)
- {
- emitError(XMLErrs::TooManyColonsInName);
- continue;
- }
- else if ((colonFirst == 0)
- || (colonLast == (int)fAttNameBuf.getLen() - 1))
- {
- emitError(XMLErrs::InvalidColonPos);
- continue;
- }
- }
- }
-
attCount++;
// And jump back to the top of the loop
@@ -3163,7 +3148,11 @@
// Expand it since its a normal entity ref
XMLBufBid bbName(&fBufMgr);
- if (!fReaderMgr.getName(bbName.getBuffer()))
+
+ int colonPosition;
+ bool validName = fDoNamespaces ? fReaderMgr.getQName(bbName.getBuffer(),
&colonPosition) :
+ fReaderMgr.getName(bbName.getBuffer());
+ if (!validName)
{
emitError(XMLErrs::ExpectedEntityRefName);
return EntityExp_Failed;
1.88 +61 -34 xml-xerces/c/src/xercesc/internal/IGXMLScanner.cpp
Index: IGXMLScanner.cpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/IGXMLScanner.cpp,v
retrieving revision 1.87
retrieving revision 1.88
diff -u -r1.87 -r1.88
--- IGXMLScanner.cpp 21 Feb 2005 18:19:45 -0000 1.87
+++ IGXMLScanner.cpp 4 Apr 2005 15:11:37 -0000 1.88
@@ -61,6 +61,8 @@
, fElemState(0)
, fContent(1023, manager)
, fRawAttrList(0)
+ , fRawAttrColonListSize(32)
+ , fRawAttrColonList(0)
, fDTDValidator(0)
, fSchemaValidator(0)
, fDTDGrammar(0)
@@ -111,6 +113,8 @@
, fElemState(0)
, fContent(1023, manager)
, fRawAttrList(0)
+ , fRawAttrColonListSize(32)
+ , fRawAttrColonList(0)
, fDTDValidator(0)
, fSchemaValidator(0)
, fDTDGrammar(0)
@@ -516,6 +520,10 @@
// And we need one for the raw attribute scan. This just stores key/
// value string pairs (prior to any processing.)
fRawAttrList = new (fMemoryManager) RefVectorOf<KVStringPair>(32, true,
fMemoryManager);
+ fRawAttrColonList = (int*) fMemoryManager->allocate
+ (
+ fRawAttrColonListSize * sizeof(int)
+ );
// Create the Validator and init them
fDTDValidator = new (fMemoryManager) DTDValidator();
@@ -550,6 +558,7 @@
{
fMemoryManager->deallocate(fElemState); //delete [] fElemState;
delete fRawAttrList;
+ fMemoryManager->deallocate(fRawAttrColonList);
delete fDTDValidator;
delete fSchemaValidator;
delete fICHandler;
@@ -625,7 +634,8 @@
{
// Assume its going to be an attribute, so get a name from
// the input.
- if (!fReaderMgr.getName(fAttNameBuf))
+ int colonPosition;
+ if (!fReaderMgr.getQName(fAttNameBuf, &colonPosition))
{
emitError(XMLErrs::ExpectedAttrName);
fReaderMgr.skipPastChar(chCloseAngle);
@@ -710,27 +720,6 @@
}
}
- // Make sure that the name is basically well formed for
namespace
- // enabled rules. It either has no colons, or it has one which
- // is neither the first or last char.
- const int colonFirst = XMLString::indexOf(curAttNameBuf,
chColon);
- if (colonFirst != -1)
- {
- const int colonLast = XMLString::lastIndexOf(chColon,
curAttNameBuf, fAttNameBuf.getLen());
-
- if (colonFirst != colonLast)
- {
- emitError(XMLErrs::TooManyColonsInName);
- continue;
- }
- else if ((colonFirst == 0)
- || (colonLast == (int)fAttNameBuf.getLen() - 1))
- {
- emitError(XMLErrs::InvalidColonPos);
- continue;
- }
- }
-
// And now lets add it to the passed collection. If we have not
// filled it up yet, then we use the next element. Else we add
// a new one.
@@ -745,9 +734,9 @@
, fAttValueBuf.getLen()
, fMemoryManager
);
- toFill.addElement(curPair);
+ toFill.addElement(curPair);
}
- else
+ else
{
curPair = toFill.elementAt(attCount);
curPair->set
@@ -756,8 +745,14 @@
fAttNameBuf.getLen(),
fAttValueBuf.getRawBuffer(),
fAttValueBuf.getLen()
- );
+ );
+ }
+
+ if (attCount >= fRawAttrColonListSize) {
+ resizeRawAttrColonList();
}
+ fRawAttrColonList[attCount] = colonPosition;
+
// And bump the count of attributes we've gotten
attCount++;
@@ -1244,8 +1239,10 @@
// Get a name from the input, which should be the name of the root
// element of the upcoming content.
- fReaderMgr.getName(bbRootName.getBuffer());
- if (bbRootName.isEmpty())
+ int colonPosition;
+ bool validName = fDoNamespaces ?
fReaderMgr.getQName(bbRootName.getBuffer(), &colonPosition) :
+
fReaderMgr.getName(bbRootName.getBuffer());
+ if (!validName)
{
emitError(XMLErrs::NoRootElemInDOCTYPE);
fReaderMgr.skipPastChar(chCloseAngle);
@@ -2125,7 +2122,8 @@
// The current position is after the open bracket, so we need to read in
// in the element name.
- if (!fReaderMgr.getName(fQNameBuf))
+ int prefixColonPos;
+ if (!fReaderMgr.getQName(fQNameBuf, &prefixColonPos))
{
emitError(XMLErrs::ExpectedElementName);
fReaderMgr.skipToChar(chOpenAngle);
@@ -2252,10 +2250,9 @@
// Resolve the qualified name to a URI and name so that we can look up
// the element decl for this element. We have now update the prefix to
- // namespace map so we should get the correct element now.
- int prefixColonPos = -1;
+ // namespace map so we should get the correct element now.
const XMLCh* qnameRawBuf = fQNameBuf.getRawBuffer();
- unsigned int uriId = resolveQName
+ unsigned int uriId = resolveQNameWithColon
(
qnameRawBuf
, fPrefixBuf
@@ -2967,15 +2964,25 @@
}
+
unsigned int
IGXMLScanner::resolveQName(const XMLCh* const qName
, XMLBuffer& prefixBuf
, const short mode
, int& prefixColonPos)
{
- // Lets split out the qName into a URI and name buffer first. The URI
- // can be empty.
prefixColonPos = XMLString::indexOf(qName, chColon);
+ return resolveQNameWithColon(qName, prefixBuf, mode, prefixColonPos);
+}
+
+unsigned int
+IGXMLScanner::resolveQNameWithColon(const XMLCh* const qName
+ , XMLBuffer& prefixBuf
+ , const short mode
+ , const int prefixColonPos)
+{
+ // Lets split out the qName into a URI and name buffer first. The URI
+ // can be empty.
if (prefixColonPos == -1)
{
// Its all name with no prefix, so put the whole thing into the name
@@ -3048,6 +3055,26 @@
fElemStateSize = newSize;
}
+void IGXMLScanner::resizeRawAttrColonList() {
+
+ unsigned int newSize = fRawAttrColonListSize * 2;
+ int* newRawAttrColonList = (int*) fMemoryManager->allocate
+ (
+ newSize * sizeof(int)
+ ); //new int[newSize];
+
+ // Copy the existing values
+ unsigned int index = 0;
+ for (; index < fRawAttrColonListSize; index++)
+ newRawAttrColonList[index] = fRawAttrColonList[index];
+
+ // Delete the old array and udpate our members
+ fMemoryManager->deallocate(fRawAttrColonList); //delete []
fRawAttrColonList;
+ fRawAttrColonList = newRawAttrColonList;
+ fRawAttrColonListSize = newSize;
+}
+
+
//
---------------------------------------------------------------------------
// IGXMLScanner: Grammar preparsing
//
---------------------------------------------------------------------------
1.26 +20 -0 xml-xerces/c/src/xercesc/internal/IGXMLScanner.hpp
Index: IGXMLScanner.hpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/IGXMLScanner.hpp,v
retrieving revision 1.25
retrieving revision 1.26
diff -u -r1.25 -r1.26
--- IGXMLScanner.hpp 14 Dec 2004 16:16:36 -0000 1.25
+++ IGXMLScanner.hpp 4 Apr 2005 15:11:37 -0000 1.26
@@ -16,6 +16,9 @@
/*
* $Log$
+ * Revision 1.26 2005/04/04 15:11:37 cargilld
+ * Fix a problem where illegal qualified names were not reported as errors.
Also store the colon position when searching for a qualified name to avoid
looking it up again.
+ *
* Revision 1.25 2004/12/14 16:16:36 cargilld
* Fix for xercesc-684: Add accessor to XMLScanner to get the current
grammar type.
*
@@ -240,7 +243,13 @@
void updateNSMap
(
const XMLCh* const attrName
+ , const XMLCh* const attrValue
+ );
+ void updateNSMap
+ (
+ const XMLCh* const attrName
, const XMLCh* const attrValue
+ , const int colonPosition
);
void scanRawAttrListforNameSpaces(int attCount);
void parseSchemaLocation(const XMLCh* const schemaLocationStr);
@@ -256,6 +265,15 @@
void resizeElemState();
void processSchemaLocation(XMLCh* const schemaLoc);
+ void resizeRawAttrColonList();
+
+ unsigned int resolveQNameWithColon
+ (
+ const XMLCh* const qName
+ , XMLBuffer& prefixBufToFill
+ , const short mode
+ , const int prefixColonPos
+ );
//
-----------------------------------------------------------------------
// Private scanning methods
//
-----------------------------------------------------------------------
@@ -347,6 +365,8 @@
unsigned int* fElemState;
XMLBuffer fContent;
RefVectorOf<KVStringPair>* fRawAttrList;
+ unsigned int fRawAttrColonListSize;
+ int* fRawAttrColonList;
DTDValidator* fDTDValidator;
SchemaValidator* fSchemaValidator;
DTDGrammar* fDTDGrammar;
1.83 +16 -7 xml-xerces/c/src/xercesc/internal/IGXMLScanner2.cpp
Index: IGXMLScanner2.cpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/IGXMLScanner2.cpp,v
retrieving revision 1.82
retrieving revision 1.83
diff -u -r1.82 -r1.83
--- IGXMLScanner2.cpp 11 Mar 2005 17:06:28 -0000 1.82
+++ IGXMLScanner2.cpp 4 Apr 2005 15:11:37 -0000 1.83
@@ -143,7 +143,7 @@
// use a stack-based buffer when possible.
XMLCh tempBuffer[100];
- const int colonInd = XMLString::indexOf(namePtr, chColon);
+ const int colonInd = fRawAttrColonList[index];
const XMLCh* prefPtr = XMLUni::fgZeroLenString;
const XMLCh* suffPtr = XMLUni::fgZeroLenString;
if (colonInd != -1)
@@ -1488,6 +1488,13 @@
void IGXMLScanner::updateNSMap(const XMLCh* const attrName
, const XMLCh* const attrValue)
{
+ updateNSMap(attrName, attrValue, XMLString::indexOf(attrName, chColon));
+}
+
+void IGXMLScanner::updateNSMap(const XMLCh* const attrName
+ , const XMLCh* const attrValue
+ , const int colonOfs)
+{
// We need a buffer to normalize the attribute value into
XMLBufBid bbNormal(&fBufMgr);
XMLBuffer& normalBuf = bbNormal.getBuffer();
@@ -1507,8 +1514,7 @@
// 2. if xxx is xml, then yyy must match XMLUni::fgXMLURIName,
and vice versa
// 3. yyy is not XMLUni::fgXMLNSURIName
// 4. if xxx is not null, then yyy cannot be an empty string.
- const XMLCh* prefPtr = XMLUni::fgZeroLenString;
- const int colonOfs = XMLString::indexOf(attrName, chColon);
+ const XMLCh* prefPtr = XMLUni::fgZeroLenString;
if (colonOfs != -1) {
prefPtr = &attrName[colonOfs + 1];
@@ -1563,7 +1569,7 @@
{
const XMLCh* valuePtr = curPair->getValue();
- updateNSMap(rawPtr, valuePtr);
+ updateNSMap(rawPtr, valuePtr, fRawAttrColonList[index]);
// if the schema URI is seen in the the valuePtr, set the
boolean seeXsi
if (XMLString::equals(valuePtr, SchemaSymbols::fgURI_XSI)) {
@@ -1585,7 +1591,7 @@
const KVStringPair* curPair = fRawAttrList->elementAt(index);
const XMLCh* rawPtr = curPair->getKey();
const XMLCh* prefPtr = XMLUni::fgZeroLenString;
- int colonInd = XMLString::indexOf(rawPtr, chColon);
+ int colonInd = fRawAttrColonList[index];
if (colonInd != -1) {
@@ -2857,7 +2863,10 @@
// Expand it since its a normal entity ref
XMLBufBid bbName(&fBufMgr);
- if (!fReaderMgr.getName(bbName.getBuffer()))
+ int colonPosition;
+ bool validName = fDoNamespaces ? fReaderMgr.getQName(bbName.getBuffer(),
&colonPosition) :
+ fReaderMgr.getName(bbName.getBuffer());
+ if (!validName)
{
emitError(XMLErrs::ExpectedEntityRefName);
return EntityExp_Failed;
1.17 +10 -0 xml-xerces/c/src/xercesc/internal/ReaderMgr.hpp
Index: ReaderMgr.hpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/ReaderMgr.hpp,v
retrieving revision 1.16
retrieving revision 1.17
diff -u -r1.16 -r1.17
--- ReaderMgr.hpp 9 Mar 2005 16:07:11 -0000 1.16
+++ ReaderMgr.hpp 4 Apr 2005 15:11:38 -0000 1.17
@@ -16,6 +16,9 @@
/*
* $Log$
+ * Revision 1.17 2005/04/04 15:11:38 cargilld
+ * Fix a problem where illegal qualified names were not reported as errors.
Also store the colon position when searching for a qualified name to avoid
looking it up again.
+ *
* Revision 1.16 2005/03/09 16:07:11 amassari
* Protected getSrcOffset to avoid crashing when parsing has finished;
updated documentation
*
@@ -183,6 +186,7 @@
//
-----------------------------------------------------------------------
bool atEOF() const;
bool getName(XMLBuffer& toFill);
+ bool getQName(XMLBuffer& toFill, int* colonPosition);
bool getNameToken(XMLBuffer& toFill);
XMLCh getNextChar();
bool getNextCharIfNot(const XMLCh chNotToGet, XMLCh& chGotten);
@@ -392,6 +396,12 @@
return fCurReader->getName(toFill, false);
}
+inline bool ReaderMgr::getQName(XMLBuffer& toFill, int *colonPosition)
+{
+ toFill.reset();
+ return fCurReader->getQName(toFill, colonPosition);
+}
+
inline bool ReaderMgr::getNameToken(XMLBuffer& toFill)
{
toFill.reset();
1.110 +63 -35 xml-xerces/c/src/xercesc/internal/SGXMLScanner.cpp
Index: SGXMLScanner.cpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/SGXMLScanner.cpp,v
retrieving revision 1.109
retrieving revision 1.110
diff -u -r1.109 -r1.110
--- SGXMLScanner.cpp 21 Feb 2005 18:19:45 -0000 1.109
+++ SGXMLScanner.cpp 4 Apr 2005 15:11:38 -0000 1.110
@@ -66,6 +66,8 @@
, fContent(1023, manager)
, fEntityTable(0)
, fRawAttrList(0)
+ , fRawAttrColonListSize(32)
+ , fRawAttrColonList(0)
, fSchemaGrammar(0)
, fSchemaValidator(0)
, fICHandler(0)
@@ -119,6 +121,8 @@
, fContent(1023, manager)
, fEntityTable(0)
, fRawAttrList(0)
+ , fRawAttrColonListSize(32)
+ , fRawAttrColonList(0)
, fSchemaGrammar(0)
, fSchemaValidator(0)
, fICHandler(0)
@@ -556,7 +560,8 @@
{
// Assume its going to be an attribute, so get a name from
// the input.
- if (!fReaderMgr.getName(fAttNameBuf))
+ int colonPosition;
+ if (!fReaderMgr.getQName(fAttNameBuf, &colonPosition))
{
emitError(XMLErrs::ExpectedAttrName);
fReaderMgr.skipPastChar(chCloseAngle);
@@ -641,27 +646,6 @@
}
}
- // Make sure that the name is basically well formed for
namespace
- // enabled rules. It either has no colons, or it has one which
- // is neither the first or last char.
- const int colonFirst = XMLString::indexOf(curAttNameBuf,
chColon);
- if (colonFirst != -1)
- {
- const int colonLast = XMLString::lastIndexOf(chColon,
curAttNameBuf, fAttNameBuf.getLen());
-
- if (colonFirst != colonLast)
- {
- emitError(XMLErrs::TooManyColonsInName);
- continue;
- }
- else if ((colonFirst == 0)
- || (colonLast == (int)fAttNameBuf.getLen() - 1))
- {
- emitError(XMLErrs::InvalidColonPos);
- continue;
- }
- }
-
// And now lets add it to the passed collection. If we have not
// filled it up yet, then we use the next element. Else we add
// a new one.
@@ -689,6 +673,10 @@
, fAttValueBuf.getLen()
);
}
+ if (attCount >= fRawAttrColonListSize) {
+ resizeRawAttrColonList();
+ }
+ fRawAttrColonList[attCount] = colonPosition;
// And bump the count of attributes we've gotten
attCount++;
@@ -1091,7 +1079,8 @@
// The current position is after the open bracket, so we need to read in
// in the element name.
- if (!fReaderMgr.getName(fQNameBuf))
+ int prefixColonPos;
+ if (!fReaderMgr.getQName(fQNameBuf, &prefixColonPos))
{
emitError(XMLErrs::ExpectedElementName);
fReaderMgr.skipToChar(chOpenAngle);
@@ -1179,9 +1168,8 @@
// Resolve the qualified name to a URI and name so that we can look up
// the element decl for this element. We have now update the prefix to
- // namespace map so we should get the correct element now.
- int prefixColonPos = -1;
- unsigned int uriId = resolveQName
+ // namespace map so we should get the correct element now.
+ unsigned int uriId = resolveQNameWithColon
(
qnameRawBuf
, fPrefixBuf
@@ -1818,9 +1806,18 @@
, const short mode
, int& prefixColonPos)
{
- // Lets split out the qName into a URI and name buffer first. The URI
- // can be empty.
prefixColonPos = XMLString::indexOf(qName, chColon);
+ return resolveQNameWithColon(qName, prefixBuf, mode, prefixColonPos);
+}
+
+unsigned int
+SGXMLScanner::resolveQNameWithColon(const XMLCh* const qName
+ , XMLBuffer& prefixBuf
+ , const short mode
+ , const int prefixColonPos)
+{
+ // Lets split out the qName into a URI and name buffer first. The URI
+ // can be empty.
if (prefixColonPos == -1)
{
// Its all name with no prefix, so put the whole thing into the name
@@ -1992,6 +1989,10 @@
// And we need one for the raw attribute scan. This just stores key/
// value string pairs (prior to any processing.)
fRawAttrList = new (fMemoryManager) RefVectorOf<KVStringPair>(32, true,
fMemoryManager);
+ fRawAttrColonList = (int*) fMemoryManager->allocate
+ (
+ fRawAttrColonListSize * sizeof(int)
+ );
// Create the Validator and init them
fSchemaValidator = new (fMemoryManager) SchemaValidator(0,
fMemoryManager);
@@ -2026,6 +2027,7 @@
delete fSchemaGrammar;
delete fEntityTable;
delete fRawAttrList;
+ fMemoryManager->deallocate(fRawAttrColonList);
delete fSchemaValidator;
delete fICHandler;
delete fElemNonDeclPool;
@@ -2061,6 +2063,25 @@
fElemStateSize = newSize;
}
+void SGXMLScanner::resizeRawAttrColonList() {
+
+ unsigned int newSize = fRawAttrColonListSize * 2;
+ int* newRawAttrColonList = (int*) fMemoryManager->allocate
+ (
+ newSize * sizeof(int)
+ ); //new int[newSize];
+
+ // Copy the existing values
+ unsigned int index = 0;
+ for (; index < fRawAttrColonListSize; index++)
+ newRawAttrColonList[index] = fRawAttrColonList[index];
+
+ // Delete the old array and udpate our members
+ fMemoryManager->deallocate(fRawAttrColonList); //delete []
fRawAttrColonList;
+ fRawAttrColonList = newRawAttrColonList;
+ fRawAttrColonListSize = newSize;
+}
+
// This method is called from scanStartTag() to build up the list of
// XMLAttr objects that will be passed out in the start tag callout. We
// get the key/value pairs from the raw scan of explicitly provided attrs,
@@ -2128,7 +2149,7 @@
// use a stack-based buffer when possible.
XMLCh tempBuffer[100];
- const int colonInd = XMLString::indexOf(namePtr, chColon);
+ const int colonInd = fRawAttrColonList[index];
const XMLCh* prefPtr = XMLUni::fgZeroLenString;
const XMLCh* suffPtr = XMLUni::fgZeroLenString;
if (colonInd != -1)
@@ -3320,6 +3341,13 @@
void SGXMLScanner::updateNSMap(const XMLCh* const attrName
, const XMLCh* const attrValue)
{
+ updateNSMap(attrName, attrValue, XMLString::indexOf(attrName, chColon));
+}
+
+void SGXMLScanner::updateNSMap(const XMLCh* const attrName
+ , const XMLCh* const attrValue
+ , const int colonOfs)
+{
// We need a buffer to normalize the attribute value into
XMLBufBid bbNormal(&fBufMgr);
XMLBuffer& normalBuf = bbNormal.getBuffer();
@@ -3339,8 +3367,7 @@
// 2. if xxx is xml, then yyy must match XMLUni::fgXMLURIName,
and vice versa
// 3. yyy is not XMLUni::fgXMLNSURIName
// 4. if xxx is not null, then yyy cannot be an empty string.
- const XMLCh* prefPtr = XMLUni::fgZeroLenString;
- const int colonOfs = XMLString::indexOf(attrName, chColon);
+ const XMLCh* prefPtr = XMLUni::fgZeroLenString;
if (colonOfs != -1) {
prefPtr = &attrName[colonOfs + 1];
@@ -3381,7 +3408,7 @@
// schema attributes.
// When we find one, send it off to be used to update the element
stack's
// namespace mappings.
- int index = 0;
+ int index;
for (index = 0; index < attCount; index++)
{
// each attribute has the prefix:suffix="value"
@@ -3395,7 +3422,7 @@
{
const XMLCh* valuePtr = curPair->getValue();
- updateNSMap(rawPtr, valuePtr);
+ updateNSMap(rawPtr, valuePtr, fRawAttrColonList[index]);
// if the schema URI is seen in the the valuePtr, set the
boolean seeXsi
if (XMLString::equals(valuePtr, SchemaSymbols::fgURI_XSI)) {
@@ -4411,7 +4438,8 @@
// Expand it since its a normal entity ref
XMLBufBid bbName(&fBufMgr);
- if (!fReaderMgr.getName(bbName.getBuffer()))
+ int colonPosition;
+ if (!fReaderMgr.getQName(bbName.getBuffer(), &colonPosition))
{
emitError(XMLErrs::ExpectedEntityRefName);
return EntityExp_Failed;
1.24 +19 -1 xml-xerces/c/src/xercesc/internal/SGXMLScanner.hpp
Index: SGXMLScanner.hpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/SGXMLScanner.hpp,v
retrieving revision 1.23
retrieving revision 1.24
diff -u -r1.23 -r1.24
--- SGXMLScanner.hpp 14 Dec 2004 16:16:36 -0000 1.23
+++ SGXMLScanner.hpp 4 Apr 2005 15:11:38 -0000 1.24
@@ -16,6 +16,9 @@
/*
* $Log$
+ * Revision 1.24 2005/04/04 15:11:38 cargilld
+ * Fix a problem where illegal qualified names were not reported as errors.
Also store the colon position when searching for a qualified name to avoid
looking it up again.
+ *
* Revision 1.23 2004/12/14 16:16:36 cargilld
* Fix for xercesc-684: Add accessor to XMLScanner to get the current
grammar type.
*
@@ -209,7 +212,20 @@
);
void resizeElemState();
-
+ void updateNSMap
+ (
+ const XMLCh* const attrName
+ , const XMLCh* const attrValue
+ , const int colonPosition
+ );
+ void resizeRawAttrColonList();
+ unsigned int resolveQNameWithColon
+ (
+ const XMLCh* const qName
+ , XMLBuffer& prefixBufToFill
+ , const short mode
+ , const int prefixColonPos
+ );
//
-----------------------------------------------------------------------
// Data members
//
@@ -252,6 +268,8 @@
XMLBuffer fContent;
ValueHashTableOf<XMLCh>* fEntityTable;
RefVectorOf<KVStringPair>* fRawAttrList;
+ unsigned int fRawAttrColonListSize;
+ int* fRawAttrColonList;
SchemaGrammar* fSchemaGrammar;
SchemaValidator* fSchemaValidator;
IdentityConstraintHandler* fICHandler;
1.29 +5 -24 xml-xerces/c/src/xercesc/internal/WFXMLScanner.cpp
Index: WFXMLScanner.cpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/WFXMLScanner.cpp,v
retrieving revision 1.28
retrieving revision 1.29
diff -u -r1.28 -r1.29
--- WFXMLScanner.cpp 12 Jan 2005 10:37:32 -0000 1.28
+++ WFXMLScanner.cpp 4 Apr 2005 15:11:38 -0000 1.29
@@ -1097,7 +1097,8 @@
// The current position is after the open bracket, so we need to read in
// in the element name.
- if (!fReaderMgr.getName(fQNameBuf))
+ int colonPosition;
+ if (!fReaderMgr.getQName(fQNameBuf, &colonPosition))
{
emitError(XMLErrs::ExpectedElementName);
fReaderMgr.skipToChar(chOpenAngle);
@@ -1182,7 +1183,8 @@
{
// Assume its going to be an attribute, so get a name from
// the input.
- if (!fReaderMgr.getName(fAttNameBuf))
+ int colonPosition;
+ if (!fReaderMgr.getQName(fAttNameBuf, &colonPosition))
{
emitError(XMLErrs::ExpectedAttrName);
fReaderMgr.skipPastChar(chCloseAngle);
@@ -1322,27 +1324,6 @@
fAttrNameHashList->setElementAt(attNameHash, attCount);
}
- // Make sure that the name is basically well formed for namespace
- // enabled rules. It either has no colons, or it has one which
- // is neither the first or last char.
- const int colonFirst = XMLString::indexOf(attNameRawBuf,
chColon);
- if (colonFirst != -1)
- {
- const int colonLast = XMLString::lastIndexOf(attNameRawBuf,
chColon);
-
- if (colonFirst != colonLast)
- {
- emitError(XMLErrs::TooManyColonsInName);
- continue;
- }
- else if ((colonFirst == 0)
- || (colonLast == (int)fAttNameBuf.getLen() - 1))
- {
- emitError(XMLErrs::InvalidColonPos);
- continue;
- }
- }
-
// Map prefix to namespace
const XMLCh* attPrefix = curAtt->getPrefix();
const XMLCh* attLocalName = curAtt->getName();
1.30 +164 -6 xml-xerces/c/src/xercesc/internal/XMLReader.cpp
Index: XMLReader.cpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/XMLReader.cpp,v
retrieving revision 1.29
retrieving revision 1.30
diff -u -r1.29 -r1.30
--- XMLReader.cpp 22 Mar 2005 20:43:27 -0000 1.29
+++ XMLReader.cpp 4 Apr 2005 15:11:38 -0000 1.30
@@ -408,6 +408,7 @@
}
if( fCharIndex < fCharsAvail ) {
+
return (fSrcOfsBase + fCharOfsBuf[fCharIndex]);
}
@@ -591,7 +592,7 @@
return false;
}
- unsigned int fCharIndex_start = fCharIndex;
+ unsigned int charIndex_start = fCharIndex;
// Lets check the first char for being a first name char. If not, then
// what's the point in living mannnn? Just give up now. We only do this
@@ -658,10 +659,10 @@
}
// we have to copy the accepted character(s), and update column
- if (fCharIndex != fCharIndex_start)
+ if (fCharIndex != charIndex_start)
{
- fCurCol += fCharIndex - fCharIndex_start;
- toFill.append(&fCharBuf[fCharIndex_start], fCharIndex -
fCharIndex_start);
+ fCurCol += fCharIndex - charIndex_start;
+ toFill.append(&fCharBuf[charIndex_start], fCharIndex -
charIndex_start);
}
// something is wrong if there is still something in the buffer
@@ -670,12 +671,169 @@
!refreshCharBuffer())
break;
- fCharIndex_start = fCharIndex;
+ charIndex_start = fCharIndex;
}
return !toFill.isEmpty();
}
+bool XMLReader::getQName(XMLBuffer& toFill, int* colonPosition)
+{
+ // Ok, first lets see if we have chars in the buffer. If not, then lets
+ // reload.
+ if (fCharIndex == fCharsAvail)
+ {
+ if (!refreshCharBuffer())
+ return false;
+ }
+
+ unsigned int charIndex_start = fCharIndex;
+ *colonPosition = -1;
+ bool checkNextCharacterForFirstNCName = false;
+
+ // Lets check the first char for being a first name char. If not, then
+ // what's the point in living mannnn? Just give up now. We only do this
+ // if its a name and not a name token that they want.
+ if (fXMLVersion == XMLV1_1 && ((fCharBuf[fCharIndex] >= 0xD800) &&
(fCharBuf[fCharIndex] <= 0xDB7F))) {
+ // make sure one more char is in the buffer, the transcoder
+ // should put only a complete surrogate pair into the buffer
+ assert(fCharIndex+1 < fCharsAvail);
+ if ((fCharBuf[fCharIndex+1] < 0xDC00) || (fCharBuf[fCharIndex+1] >
0xDFFF))
+ return false;
+
+ // Looks ok, so lets eat it
+ fCharIndex += 2;
+ }
+ else {
+ if (!isFirstNameChar(fCharBuf[fCharIndex]))
+ return false;
+ if (fCharBuf[fCharIndex] == chColon)
+ return false;
+ // Looks ok, so lets eat it
+ fCharIndex ++;
+ }
+
+ // And now we loop until we run out of data in this reader or we hit
+ // a non-name char.
+ while (true)
+ {
+ if (checkNextCharacterForFirstNCName) {
+ checkNextCharacterForFirstNCName = false;
+ if (fXMLVersion == XMLV1_1) {
+ if ( !((fCharBuf[fCharIndex] >= 0xD800) &&
(fCharBuf[fCharIndex] <= 0xDB7F)) )
+ {
+ if (!isFirstNameChar(fCharBuf[fCharIndex]))
+ return false;
+ if (fCharBuf[fCharIndex] == chColon)
+ return false;
+ fCharIndex++;
+ }
+ }
+ else {
+ if (!isFirstNameChar(fCharBuf[fCharIndex]))
+ return false;
+ if (fCharBuf[fCharIndex] == chColon)
+ return false;
+ fCharIndex++;
+ }
+
+ }
+
+ if (fXMLVersion == XMLV1_1)
+ {
+ while (fCharIndex < fCharsAvail)
+ {
+ // Check the current char and take it if its a name char.
Else
+ // break out.
+ if ( (fCharBuf[fCharIndex] >= 0xD800) &&
(fCharBuf[fCharIndex] <= 0xDB7F) )
+ {
+ // make sure one more char is in the buffer, the
transcoder
+ // should put only a complete surrogate pair into the
buffer
+ assert(fCharIndex+1 < fCharsAvail);
+ if ( (fCharBuf[fCharIndex+1] < 0xDC00) ||
+ (fCharBuf[fCharIndex+1] > 0xDFFF) )
+ break;
+ fCharIndex += 2;
+
+ }
+ else
+ {
+ if (!isNameChar(fCharBuf[fCharIndex]))
+ break;
+ if (fCharBuf[fCharIndex] == chColon) {
+ if (*colonPosition != -1) {
+ return false;
+ }
+ // update the buffer to get the colon Offset position
+ fCurCol += fCharIndex - charIndex_start;
+ toFill.append(&fCharBuf[charIndex_start], fCharIndex
- charIndex_start);
+ charIndex_start = fCharIndex;
+ *colonPosition = toFill.getLen();
+ if (fCharIndex + 1 < fCharsAvail) {
+ fCharIndex++;
+ if (!isFirstNameChar(fCharBuf[fCharIndex]))
+ return false;
+ if (fCharBuf[fCharIndex] == chColon)
+ return false;
+ }
+ else {
+ checkNextCharacterForFirstNCName = true;
+ }
+ }
+ fCharIndex++;
+ }
+ }
+ }
+ else // XMLV1_0
+ {
+ while (fCharIndex < fCharsAvail)
+ {
+ if (!isNameChar(fCharBuf[fCharIndex]))
+ break;
+ if (fCharBuf[fCharIndex] == chColon) {
+ if (*colonPosition != -1) {
+ return false;
+ }
+ // update the buffer to get the colon Offset position
+ fCurCol += fCharIndex - charIndex_start;
+ toFill.append(&fCharBuf[charIndex_start], fCharIndex -
charIndex_start);
+ charIndex_start = fCharIndex;
+ *colonPosition = toFill.getLen();
+ if (fCharIndex + 1 < fCharsAvail) {
+ fCharIndex++;
+ if (!isFirstNameChar(fCharBuf[fCharIndex]))
+ return false;
+ if (fCharBuf[fCharIndex] == chColon)
+ return false;
+ }
+ else {
+ checkNextCharacterForFirstNCName = true;
+ }
+ }
+ fCharIndex++;
+ }
+ }
+
+ // we have to copy the accepted character(s), and update column
+ if (fCharIndex != charIndex_start)
+ {
+ fCurCol += fCharIndex - charIndex_start;
+ toFill.append(&fCharBuf[charIndex_start], fCharIndex -
charIndex_start);
+ }
+
+ // something is wrong if there is still something in the buffer
+ // or if we don't get no more, then break out.
+ if ((fCharIndex < fCharsAvail) ||
+ !refreshCharBuffer())
+ break;
+
+ charIndex_start = fCharIndex;
+ }
+
+ if (checkNextCharacterForFirstNCName)
+ return false;
+ return !toFill.isEmpty();
+}
bool XMLReader::getSpaces(XMLBuffer& toFill)
{
1.22 +5 -1 xml-xerces/c/src/xercesc/internal/XMLReader.hpp
Index: XMLReader.hpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/XMLReader.hpp,v
retrieving revision 1.21
retrieving revision 1.22
diff -u -r1.21 -r1.22
--- XMLReader.hpp 22 Mar 2005 20:43:27 -0000 1.21
+++ XMLReader.hpp 4 Apr 2005 15:11:38 -0000 1.22
@@ -16,6 +16,9 @@
/*
* $Log$
+ * Revision 1.22 2005/04/04 15:11:38 cargilld
+ * Fix a problem where illegal qualified names were not reported as errors.
Also store the colon position when searching for a qualified name to avoid
looking it up again.
+ *
* Revision 1.21 2005/03/22 20:43:27 cargilld
* Check in Christian's patches for Xercesc-1369 and 1370.
*
@@ -233,7 +236,7 @@
bool isXMLLetter(const XMLCh toCheck) const;
bool isFirstNameChar(const XMLCh toCheck) const;
- bool isNameChar(const XMLCh toCheck) const;
+ bool isNameChar(const XMLCh toCheck) const;
bool isPlainContentChar(const XMLCh toCheck) const;
bool isSpecialStartTagChar(const XMLCh toCheck) const;
bool isXMLChar(const XMLCh toCheck) const;
@@ -302,6 +305,7 @@
// Scanning methods
//
-----------------------------------------------------------------------
bool getName(XMLBuffer& toFill, const bool token);
+ bool getQName(XMLBuffer& toFill, int* colonPosition);
bool getNextChar(XMLCh& chGotten);
bool getNextCharIfNot(const XMLCh chNotToGet, XMLCh& chGotten);
void movePlainContentChars(XMLBuffer &dest);
1.7 +6 -6 xml-xerces/c/src/xercesc/internal/XSAXMLScanner.cpp
Index: XSAXMLScanner.cpp
===================================================================
RCS file: /home/cvs/xml-xerces/c/src/xercesc/internal/XSAXMLScanner.cpp,v
retrieving revision 1.6
retrieving revision 1.7
diff -u -r1.6 -r1.7
--- XSAXMLScanner.cpp 19 Feb 2005 23:44:17 -0000 1.6
+++ XSAXMLScanner.cpp 4 Apr 2005 15:11:38 -0000 1.7
@@ -192,7 +192,8 @@
// The current position is after the open bracket, so we need to read in
// in the element name.
- if (!fReaderMgr.getName(fQNameBuf))
+ int prefixColonPos;
+ if (!fReaderMgr.getQName(fQNameBuf, &prefixColonPos))
{
emitError(XMLErrs::ExpectedElementName);
fReaderMgr.skipToChar(chOpenAngle);
@@ -263,9 +264,8 @@
// Resolve the qualified name to a URI and name so that we can look up
// the element decl for this element. We have now update the prefix to
- // namespace map so we should get the correct element now.
- int prefixColonPos = -1;
- unsigned int uriId = resolveQName
+ // namespace map so we should get the correct element now.
+ unsigned int uriId = resolveQNameWithColon
(
qnameRawBuf, fPrefixBuf, ElemStack::Mode_Element, prefixColonPos
);
@@ -612,7 +612,7 @@
{
const XMLCh* valuePtr = curPair->getValue();
- updateNSMap(rawPtr, valuePtr);
+ updateNSMap(rawPtr, valuePtr, fRawAttrColonList[index]);
// if the schema URI is seen in the the valuePtr, set the
boolean seeXsi
if (XMLString::equals(valuePtr, SchemaSymbols::fgURI_XSI)) {
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
