CFX_WideString CFX_WideString::FromUTF8(const char* str, FX_STRSIZE len) {
if (!str || 0 == len) {
return CFX_WideString();
}
CFX_UTF8Decoder decoder;
for (FX_STRSIZE i = 0; i < len; i++) {
decoder.Input(str[i]);
}
return decoder.GetResult();
}
void CXML_Parser::GetAttrValue(CFX_WideString &value)
{
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (IsEOF()) {
return;
}
CFX_UTF8Decoder decoder;
uint8_t mark = 0, ch = 0;
do {
while (m_dwIndex < m_dwBufferSize) {
ch = m_pBuffer[m_dwIndex];
if (mark == 0) {
if (ch != '\'' && ch != '"') {
return;
}
mark = ch;
m_dwIndex ++;
ch = 0;
continue;
}
m_dwIndex ++;
if (ch == mark) {
break;
}
if (ch == '&') {
decoder.AppendChar(GetCharRef());
if (IsEOF()) {
value = decoder.GetResult();
return;
}
} else {
decoder.Input(ch);
}
}
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (ch == mark || m_dwIndex < m_dwBufferSize || IsEOF()) {
break;
}
} while (ReadNextBlock());
value = decoder.GetResult();
}
CXML_Element* CXML_Parser::ParseElement(CXML_Element* pParent, bool bStartTag) {
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (IsEOF()) {
return nullptr;
}
CFX_ByteString tag_name, tag_space;
bool bEndTag;
GetTagName(tag_space, tag_name, bEndTag, bStartTag);
if (tag_name.IsEmpty() || bEndTag) {
return nullptr;
}
CXML_Element* pElement = new CXML_Element;
pElement->m_pParent = pParent;
pElement->SetTag(tag_space.AsStringC(), tag_name.AsStringC());
do {
CFX_ByteString attr_space, attr_name;
while (m_dwIndex < m_dwBufferSize) {
SkipWhiteSpaces();
if (IsEOF()) {
break;
}
if (!g_FXCRT_XML_IsNameIntro(m_pBuffer[m_dwIndex])) {
break;
}
GetName(attr_space, attr_name);
SkipWhiteSpaces();
if (IsEOF()) {
break;
}
if (m_pBuffer[m_dwIndex] != '=') {
break;
}
m_dwIndex++;
SkipWhiteSpaces();
if (IsEOF()) {
break;
}
CFX_WideString attr_value;
GetAttrValue(attr_value);
pElement->m_AttrMap.SetAt(attr_space, attr_name, attr_value);
}
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (m_dwIndex < m_dwBufferSize || IsEOF()) {
break;
}
} while (ReadNextBlock());
SkipWhiteSpaces();
if (IsEOF()) {
return pElement;
}
uint8_t ch = m_pBuffer[m_dwIndex++];
if (ch == '/') {
m_dwIndex++;
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
return pElement;
}
if (ch != '>') {
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
delete pElement;
return nullptr;
}
SkipWhiteSpaces();
if (IsEOF()) {
return pElement;
}
CFX_UTF8Decoder decoder;
CFX_WideTextBuf content;
bool bCDATA = false;
int32_t iState = 0;
do {
while (m_dwIndex < m_dwBufferSize) {
ch = m_pBuffer[m_dwIndex++];
switch (iState) {
case 0:
if (ch == '<') {
iState = 1;
} else if (ch == '&') {
decoder.ClearStatus();
decoder.AppendChar(GetCharRef());
} else {
decoder.Input(ch);
}
break;
case 1:
if (ch == '!') {
iState = 2;
} else if (ch == '?') {
SkipLiterals("?>");
SkipWhiteSpaces();
iState = 0;
} else if (ch == '/') {
CFX_ByteString space, name;
GetName(space, name);
SkipWhiteSpaces();
m_dwIndex++;
iState = 10;
} else {
content << decoder.GetResult();
CFX_WideString dataStr = content.MakeString();
if (!bCDATA && !m_bSaveSpaceChars) {
dataStr.TrimRight(L" \t\r\n");
}
InsertContentSegment(bCDATA, dataStr.AsStringC(), pElement);
content.Clear();
decoder.Clear();
bCDATA = false;
iState = 0;
m_dwIndex--;
CXML_Element* pSubElement = ParseElement(pElement, true);
if (!pSubElement) {
break;
}
pSubElement->m_pParent = pElement;
pElement->m_Children.push_back(
{CXML_Element::Element, pSubElement});
SkipWhiteSpaces();
}
break;
case 2:
if (ch == '[') {
SkipLiterals("]]>");
} else if (ch == '-') {
m_dwIndex++;
SkipLiterals("-->");
} else {
SkipLiterals(">");
}
decoder.Clear();
SkipWhiteSpaces();
iState = 0;
break;
}
if (iState == 10) {
break;
}
}
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (iState == 10 || m_dwIndex < m_dwBufferSize || IsEOF()) {
break;
}
} while (ReadNextBlock());
content << decoder.GetResult();
CFX_WideString dataStr = content.MakeString();
if (!m_bSaveSpaceChars) {
dataStr.TrimRight(L" \t\r\n");
}
InsertContentSegment(bCDATA, dataStr.AsStringC(), pElement);
content.Clear();
decoder.Clear();
bCDATA = false;
return pElement;
}
CXML_Element* CXML_Parser::ParseElement(CXML_Element* pParent, FX_BOOL bStartTag)
{
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (IsEOF()) {
return NULL;
}
CFX_ByteString tag_name, tag_space;
FX_BOOL bEndTag;
GetTagName(tag_space, tag_name, bEndTag, bStartTag);
if (tag_name.IsEmpty() || bEndTag) {
return NULL;
}
CXML_Element* pElement;
pElement = FX_NEW CXML_Element;
if (pElement) {
pElement->m_pParent = pParent;
pElement->SetTag(tag_space, tag_name);
}
if (!pElement) {
return NULL;
}
do {
CFX_ByteString attr_space, attr_name;
while (m_dwIndex < m_dwBufferSize) {
SkipWhiteSpaces();
if (IsEOF()) {
break;
}
if (!g_FXCRT_XML_IsNameIntro(m_pBuffer[m_dwIndex])) {
break;
}
GetName(attr_space, attr_name);
SkipWhiteSpaces();
if (IsEOF()) {
break;
}
if (m_pBuffer[m_dwIndex] != '=') {
break;
}
m_dwIndex ++;
SkipWhiteSpaces();
if (IsEOF()) {
break;
}
CFX_WideString attr_value;
GetAttrValue(attr_value);
pElement->m_AttrMap.SetAt(attr_space, attr_name, attr_value);
}
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (m_dwIndex < m_dwBufferSize || IsEOF()) {
break;
}
} while (ReadNextBlock());
SkipWhiteSpaces();
if (IsEOF()) {
return pElement;
}
FX_BYTE ch = m_pBuffer[m_dwIndex ++];
if (ch == '/') {
m_dwIndex ++;
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
return pElement;
}
if (ch != '>') {
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
delete pElement;
return NULL;
}
SkipWhiteSpaces();
if (IsEOF()) {
return pElement;
}
CFX_UTF8Decoder decoder;
CFX_WideTextBuf content;
FX_BOOL bCDATA = FALSE;
FX_INT32 iState = 0;
do {
while (m_dwIndex < m_dwBufferSize) {
ch = m_pBuffer[m_dwIndex ++];
switch (iState) {
case 0:
if (ch == '<') {
iState = 1;
} else if (ch == '&') {
decoder.ClearStatus();
decoder.AppendChar(GetCharRef());
} else {
decoder.Input(ch);
}
break;
case 1:
if (ch == '!') {
iState = 2;
} else if (ch == '?') {
SkipLiterals(FX_BSTRC("?>"));
SkipWhiteSpaces();
iState = 0;
} else if (ch == '/') {
CFX_ByteString space, name;
GetName(space, name);
SkipWhiteSpaces();
m_dwIndex ++;
iState = 10;
} else {
content << decoder.GetResult();
CFX_WideString dataStr = content.GetWideString();
if (!bCDATA && !m_bSaveSpaceChars) {
dataStr.TrimRight((FX_LPCWSTR)L" \t\r\n");
}
InsertContentSegment(bCDATA, dataStr, pElement);
content.Clear();
decoder.Clear();
bCDATA = FALSE;
iState = 0;
m_dwIndex --;
CXML_Element* pSubElement = ParseElement(pElement, TRUE);
if (pSubElement == NULL) {
break;
}
pSubElement->m_pParent = pElement;
pElement->m_Children.Add((FX_LPVOID)CXML_Element::Element);
pElement->m_Children.Add(pSubElement);
SkipWhiteSpaces();
}
break;
case 2:
if (ch == '[') {
SkipLiterals(FX_BSTRC("]]>"));
} else if (ch == '-') {
m_dwIndex ++;
SkipLiterals(FX_BSTRC("-->"));
} else {
SkipLiterals(FX_BSTRC(">"));
}
decoder.Clear();
SkipWhiteSpaces();
iState = 0;
break;
}
if (iState == 10) {
break;
}
}
m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex;
if (iState == 10 || m_dwIndex < m_dwBufferSize || IsEOF()) {
break;
}
} while (ReadNextBlock());
content << decoder.GetResult();
CFX_WideString dataStr = content.GetWideString();
if (!m_bSaveSpaceChars) {
dataStr.TrimRight((FX_LPCWSTR)L" \t\r\n");
}
InsertContentSegment(bCDATA, dataStr, pElement);
content.Clear();
decoder.Clear();
bCDATA = FALSE;
return pElement;
}