CFX_WideString PDF_GetFirstTextLine_Unicode(CPDF_Document* pDoc, CPDF_Dictionary* pPage)
{
CFX_WideTextBuf buffer;
buffer.EstimateSize(0, 1024);
CPDF_Page page;
page.Load(pDoc, pPage);
CPDF_ParseOptions options;
options.m_bTextOnly = TRUE;
options.m_bSeparateForm = FALSE;
page.ParseContent(&options);
CPDF_TextStream textstream(buffer, FALSE, NULL);
FX_POSITION pos = page.GetFirstObjectPosition();
while (pos) {
CPDF_PageObject* pObject = page.GetNextObject(pos);
if (pObject->m_Type != PDFPAGE_TEXT) {
continue;
}
if (textstream.ProcessObject((CPDF_TextObject*)pObject, TRUE)) {
break;
}
}
return buffer.GetWideString();
}
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 = new CXML_Element;
pElement->m_pParent = pParent;
pElement->SetTag(tag_space, tag_name);
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 NULL;
}
SkipWhiteSpaces();
if (IsEOF()) {
return pElement;
}
CFX_UTF8Decoder decoder;
CFX_WideTextBuf content;
FX_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(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(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((void*)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(L" \t\r\n");
}
InsertContentSegment(bCDATA, dataStr, pElement);
content.Clear();
decoder.Clear();
bCDATA = FALSE;
return pElement;
}