You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@xerces.apache.org by am...@apache.org on 2008/01/04 19:25:55 UTC
svn commit: r608953 - in
/xerces/c/branches/xerces-2/src/xercesc/validators/datatype:
AnyURIDatatypeValidator.cpp AnyURIDatatypeValidator.hpp
Author: amassari
Date: Fri Jan 4 10:25:55 2008
New Revision: 608953
URL: http://svn.apache.org/viewvc?rev=608953&view=rev
Log:
- If the URI contains several non-ASCII characters, the optimistic allocation (len*3) could run out of space (as each non-ASCII char would be replaced by at least 2 UTF-8 bytes, and each of them by 3 characters) -> use an XMLBuffer that can grow past the initial allocation
- the static buffers used by printf were 2 bytes long, but sprintf would also write the NULL terminator
- the converted UTF-8 string was converted partially, as the loop used the length of the original string instead of the length of the UTF-8 string
- each UTF-8 byte was assumed to be a signed byte, but XMLByte is unsigned; so the b<0 test was never true (and why add 256 to the signed value to make it positive? the new number would be different) and the next text was accessing a static array of 128 items with indexes between 0 and 255
Modified:
xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.cpp
xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.hpp
Modified: xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.cpp
URL: http://svn.apache.org/viewvc/xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.cpp?rev=608953&r1=608952&r2=608953&view=diff
==============================================================================
--- xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.cpp (original)
+++ xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.cpp Fri Jan 4 10:25:55 2008
@@ -25,6 +25,7 @@
#include <stdio.h>
#include <xercesc/util/OutOfMemoryException.hpp>
#include <xercesc/util/XMLUTF8Transcoder.hpp>
+#include <xercesc/framework/XMLBuffer.hpp>
#include <xercesc/validators/datatype/AnyURIDatatypeValidator.hpp>
#include <xercesc/validators/datatype/InvalidDatatypeFacetException.hpp>
#include <xercesc/validators/datatype/InvalidDatatypeValueException.hpp>
@@ -82,10 +83,9 @@
if (len)
{
// Encode special characters using XLink 5.4 algorithm
- XMLCh* encoded = (XMLCh*)manager->allocate((len*3+1) * sizeof(XMLCh));
- ArrayJanitor<XMLCh> encodedJan(encoded, manager);
+ XMLBuffer encoded((len*3)+1, manager);
encode(content, len, encoded, manager);
- validURI = XMLUri::isValidURI(true, encoded);
+ validURI = XMLUri::isValidURI(true, encoded.getRawBuffer(), true);
}
}
catch(const OutOfMemoryException&)
@@ -113,7 +113,7 @@
* special ASCII characters: 0x00~0x1F, 0x7F, ' ', '<', '>', etc.
* and non-ASCII characters (whose value >= 128).
***/
-void AnyURIDatatypeValidator::encode(const XMLCh* const content, const unsigned int len, XMLCh* encoded, MemoryManager* const manager)
+void AnyURIDatatypeValidator::encode(const XMLCh* const content, const unsigned int len, XMLBuffer& encoded, MemoryManager* const manager)
{
static const bool needEscapeMap[] = {
true , true , true , true , true , true , true , true , true , true , true , true , true , true , true , true , /* 0x00 to 0x0F need escape */
@@ -139,15 +139,15 @@
if (needEscapeMap[ch])
{
- char tempStr[2] = "\0";
+ char tempStr[3] = "\0";
sprintf(tempStr, "%02X", ch);
- encoded[bufferIndex++] = '%';
- encoded[bufferIndex++] = (XMLCh)tempStr[0];
- encoded[bufferIndex++] = (XMLCh)tempStr[1];
+ encoded.append('%');
+ encoded.append((XMLCh)tempStr[0]);
+ encoded.append((XMLCh)tempStr[1]);
}
else
{
- encoded[bufferIndex++] = (XMLCh)ch;
+ encoded.append((XMLCh)ch);
}
}
@@ -160,38 +160,27 @@
unsigned int charsEaten;
XMLUTF8Transcoder transcoder(XMLUni::fgUTF8EncodingString, remContentLen*4+1, manager);
- transcoder.transcodeTo(remContent, remContentLen, UTF8Byte, remContentLen*4, charsEaten, XMLTranscoder::UnRep_RepChar);
+ unsigned int utf8Len = transcoder.transcodeTo(remContent, remContentLen, UTF8Byte, remContentLen*4, charsEaten, XMLTranscoder::UnRep_RepChar);
assert(charsEaten == remContentLen);
unsigned int j;
- for (j = 0; j < remContentLen; j++) {
+ for (j = 0; j < utf8Len; j++) {
XMLByte b = UTF8Byte[j];
- // for non-ascii character: make it positive, then escape
- if (b < 0) {
- int ch = b + 256;
- char tempStr[2] = "\0";
- sprintf(tempStr, "%02X", ch);
- encoded[bufferIndex++] = '%';
- encoded[bufferIndex++] = (XMLCh)tempStr[0];
- encoded[bufferIndex++] = (XMLCh)tempStr[1];
- }
- else if (needEscapeMap[b])
+ if (b >= 128 || needEscapeMap[b])
{
- char tempStr[2] = "\0";
+ char tempStr[3] = "\0";
sprintf(tempStr, "%02X", b);
- encoded[bufferIndex++] = '%';
- encoded[bufferIndex++] = (XMLCh)tempStr[0];
- encoded[bufferIndex++] = (XMLCh)tempStr[1];
+ encoded.append('%');
+ encoded.append((XMLCh)tempStr[0]);
+ encoded.append((XMLCh)tempStr[1]);
}
else
{
- encoded[bufferIndex++] = (XMLCh)b;
+ encoded.append((XMLCh)b);
}
}
manager->deallocate(UTF8Byte);
}
-
- encoded[bufferIndex] = (XMLCh)0;
}
/***
Modified: xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.hpp
URL: http://svn.apache.org/viewvc/xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.hpp?rev=608953&r1=608952&r2=608953&view=diff
==============================================================================
--- xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.hpp (original)
+++ xerces/c/branches/xerces-2/src/xercesc/validators/datatype/AnyURIDatatypeValidator.hpp Fri Jan 4 10:25:55 2008
@@ -27,6 +27,8 @@
XERCES_CPP_NAMESPACE_BEGIN
+class XMLBuffer;
+
class VALIDATORS_EXPORT AnyURIDatatypeValidator : public AbstractStringValidator
{
public:
@@ -81,7 +83,7 @@
// -----------------------------------------------------------------------
AnyURIDatatypeValidator(const AnyURIDatatypeValidator&);
AnyURIDatatypeValidator& operator=(const AnyURIDatatypeValidator&);
- void encode(const XMLCh* const content, const unsigned int len, XMLCh* encoded, MemoryManager* const manager);
+ void encode(const XMLCh* const content, const unsigned int len, XMLBuffer& encoded, MemoryManager* const manager);
};
XERCES_CPP_NAMESPACE_END
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@xerces.apache.org
For additional commands, e-mail: commits-help@xerces.apache.org