diff options
Diffstat (limited to 'xfa/src/fgas/src/xml')
| -rw-r--r-- | xfa/src/fgas/src/xml/fx_sax_imp.cpp | 1378 | ||||
| -rw-r--r-- | xfa/src/fgas/src/xml/fx_sax_imp.h | 268 |
2 files changed, 823 insertions, 823 deletions
diff --git a/xfa/src/fgas/src/xml/fx_sax_imp.cpp b/xfa/src/fgas/src/xml/fx_sax_imp.cpp index 6b3cc778ab..bb1602ca72 100644 --- a/xfa/src/fgas/src/xml/fx_sax_imp.cpp +++ b/xfa/src/fgas/src/xml/fx_sax_imp.cpp @@ -1,689 +1,689 @@ -// Copyright 2014 PDFium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
-
-#include <algorithm>
-
-#include "xfa/src/fgas/src/fgas_base.h"
-#include "fx_sax_imp.h"
-
-namespace {
-
-const FX_DWORD kSaxFileBufSize = 32768;
-
-} // namespace
-
-IFX_SAXReader* FX_SAXReader_Create() {
- return new CFX_SAXReader;
-}
-CFX_SAXFile::CFX_SAXFile()
- : m_pFile(NULL),
- m_dwStart(0),
- m_dwEnd(0),
- m_dwCur(0),
- m_pBuf(NULL),
- m_dwBufSize(0),
- m_dwBufIndex(0) {}
-FX_BOOL CFX_SAXFile::StartFile(IFX_FileRead* pFile,
- FX_DWORD dwStart,
- FX_DWORD dwLen) {
- FXSYS_assert(m_pFile == NULL && pFile != NULL);
- FX_DWORD dwSize = pFile->GetSize();
- if (dwStart >= dwSize) {
- return FALSE;
- }
- if (dwLen == -1 || dwStart + dwLen > dwSize) {
- dwLen = dwSize - dwStart;
- }
- if (dwLen == 0) {
- return FALSE;
- }
- m_dwBufSize = std::min(dwLen, kSaxFileBufSize);
- m_pBuf = FX_Alloc(uint8_t, m_dwBufSize);
- if (!pFile->ReadBlock(m_pBuf, dwStart, m_dwBufSize)) {
- return FALSE;
- }
- m_dwStart = dwStart;
- m_dwEnd = dwStart + dwLen;
- m_dwCur = dwStart;
- m_pFile = pFile;
- m_dwBufIndex = 0;
- return TRUE;
-}
-FX_BOOL CFX_SAXFile::ReadNextBlock() {
- FXSYS_assert(m_pFile != NULL);
- FX_DWORD dwSize = m_dwEnd - m_dwCur;
- if (dwSize == 0) {
- return FALSE;
- }
- m_dwBufSize = std::min(dwSize, kSaxFileBufSize);
- if (!m_pFile->ReadBlock(m_pBuf, m_dwCur, m_dwBufSize)) {
- return FALSE;
- }
- m_dwBufIndex = 0;
- return TRUE;
-}
-void CFX_SAXFile::Reset() {
- if (m_pBuf) {
- FX_Free(m_pBuf);
- m_pBuf = NULL;
- }
- m_pFile = NULL;
-}
-CFX_SAXReader::CFX_SAXReader()
- : m_File(),
- m_pHandler(nullptr),
- m_iState(-1),
- m_pRoot(nullptr),
- m_pCurItem(nullptr),
- m_dwItemID(0),
- m_iDataSize(256),
- m_iNameSize(256),
- m_dwParseMode(0),
- m_pCommentContext(nullptr) {
- m_pszData = FX_Alloc(uint8_t, m_iDataSize);
- m_pszName = FX_Alloc(uint8_t, m_iNameSize);
-}
-CFX_SAXReader::~CFX_SAXReader() {
- Reset();
- if (m_pszData) {
- FX_Free(m_pszData);
- m_pszData = NULL;
- }
- if (m_pszName) {
- FX_Free(m_pszName);
- m_pszName = NULL;
- }
-}
-void CFX_SAXReader::Reset() {
- m_File.Reset();
- CFX_SAXItem* pItem = m_pRoot;
- while (pItem) {
- CFX_SAXItem* pNext = pItem->m_pNext;
- delete pItem;
- pItem = pNext;
- }
- m_pRoot = NULL;
- m_pCurItem = NULL;
- m_dwItemID = 0;
- m_SkipStack.RemoveAll();
- m_SkipChar = 0;
- m_iDataLength = 0;
- m_iEntityStart = -1;
- m_iNameLength = 0;
- m_iDataPos = 0;
- if (m_pCommentContext) {
- delete m_pCommentContext;
- m_pCommentContext = NULL;
- }
-}
-inline void CFX_SAXReader::Push() {
- CFX_SAXItem* pNew = new CFX_SAXItem;
- pNew->m_dwID = ++m_dwItemID;
- pNew->m_bSkip = m_pCurItem->m_bSkip;
- pNew->m_pPrev = m_pCurItem;
- m_pCurItem->m_pNext = pNew;
- m_pCurItem = pNew;
-}
-inline void CFX_SAXReader::Pop() {
- if (!m_pCurItem) {
- return;
- }
- CFX_SAXItem* pPrev = m_pCurItem->m_pPrev;
- pPrev->m_pNext = NULL;
- delete m_pCurItem;
- m_pCurItem = pPrev;
-}
-inline void CFX_SAXReader::AppendData(uint8_t ch) {
- ReallocDataBuffer();
- m_pszData[m_iDataPos++] = ch;
-}
-inline void CFX_SAXReader::AppendName(uint8_t ch) {
- ReallocNameBuffer();
- m_pszName[m_iDataPos++] = ch;
-}
-void CFX_SAXReader::ReallocDataBuffer() {
- if (m_iDataPos < m_iDataSize) {
- return;
- }
- if (m_iDataSize <= 1024 * 1024) {
- m_iDataSize *= 2;
- } else {
- m_iDataSize += 1024 * 1024;
- }
- m_pszData = (uint8_t*)FX_Realloc(uint8_t, m_pszData, m_iDataSize);
-}
-void CFX_SAXReader::ReallocNameBuffer() {
- if (m_iDataPos < m_iNameSize) {
- return;
- }
- if (m_iNameSize <= 1024 * 1024) {
- m_iNameSize *= 2;
- } else {
- m_iNameSize += 1024 * 1024;
- }
- m_pszName = (uint8_t*)FX_Realloc(uint8_t, m_pszName, m_iNameSize);
-}
-inline FX_BOOL CFX_SAXReader::SkipSpace(uint8_t ch) {
- return (m_dwParseMode & FX_SAXPARSEMODE_NotSkipSpace) == 0 && ch < 0x21;
-}
-int32_t CFX_SAXReader::StartParse(IFX_FileRead* pFile,
- FX_DWORD dwStart,
- FX_DWORD dwLen,
- FX_DWORD dwParseMode) {
- m_iState = -1;
- Reset();
- if (!m_File.StartFile(pFile, dwStart, dwLen)) {
- return -1;
- }
- m_iState = 0;
- m_eMode = FX_SAXMODE_Text;
- m_ePrevMode = FX_SAXMODE_Text;
- m_bCharData = FALSE;
- m_dwDataOffset = 0;
- m_pRoot = m_pCurItem = new CFX_SAXItem;
- m_pCurItem->m_dwID = ++m_dwItemID;
- m_dwParseMode = dwParseMode;
- return 0;
-}
-typedef void (CFX_SAXReader::*FX_SAXReader_LPFParse)();
-static const FX_SAXReader_LPFParse g_FX_SAXReader_LPFParse[FX_SAXMODE_MAX] = {
- &CFX_SAXReader::ParseText,
- &CFX_SAXReader::ParseNodeStart,
- &CFX_SAXReader::ParseDeclOrComment,
- &CFX_SAXReader::ParseDeclNode,
- &CFX_SAXReader::ParseComment,
- &CFX_SAXReader::ParseCommentContent,
- &CFX_SAXReader::ParseTagName,
- &CFX_SAXReader::ParseTagAttributeName,
- &CFX_SAXReader::ParseTagAttributeEqual,
- &CFX_SAXReader::ParseTagAttributeValue,
- &CFX_SAXReader::ParseMaybeClose,
- &CFX_SAXReader::ParseTagClose,
- &CFX_SAXReader::ParseTagEnd,
- &CFX_SAXReader::ParseTargetData,
-};
-int32_t CFX_SAXReader::ContinueParse(IFX_Pause* pPause) {
- if (m_iState < 0 || m_iState > 99) {
- return m_iState;
- }
- while (m_File.m_dwCur < m_File.m_dwEnd) {
- FX_DWORD& index = m_File.m_dwBufIndex;
- FX_DWORD size = m_File.m_dwBufSize;
- const uint8_t* pBuf = m_File.m_pBuf;
- while (index < size) {
- m_CurByte = pBuf[index];
- (this->*g_FX_SAXReader_LPFParse[m_eMode])();
- index++;
- }
- m_File.m_dwCur += index;
- m_iState = (m_File.m_dwCur - m_File.m_dwStart) * 100 /
- (m_File.m_dwEnd - m_File.m_dwStart);
- if (m_File.m_dwCur >= m_File.m_dwEnd) {
- break;
- }
- if (!m_File.ReadNextBlock()) {
- m_iState = -2;
- break;
- }
- m_dwDataOffset = 0;
- if (pPause && pPause->NeedToPauseNow()) {
- break;
- }
- }
- return m_iState;
-}
-void CFX_SAXReader::ParseChar(uint8_t ch) {
- ReallocDataBuffer();
- m_pszData[m_iDataPos] = ch;
- if (m_iEntityStart > -1 && ch == ';') {
- int32_t iSaveEntityStart = m_iEntityStart;
- CFX_ByteString csEntity(m_pszData + m_iEntityStart + 1,
- m_iDataPos - m_iEntityStart - 1);
- int32_t iLen = csEntity.GetLength();
- if (iLen > 0) {
- if (csEntity[0] == '#') {
- if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_sharp) == 0) {
- ch = 0;
- uint8_t w;
- if (iLen > 1 && csEntity[1] == 'x') {
- for (int32_t i = 2; i < iLen; i++) {
- w = csEntity[i];
- if (w >= '0' && w <= '9') {
- ch = (ch << 4) + w - '0';
- } else if (w >= 'A' && w <= 'F') {
- ch = (ch << 4) + w - 55;
- } else if (w >= 'a' && w <= 'f') {
- ch = (ch << 4) + w - 87;
- } else {
- break;
- }
- }
- } else {
- for (int32_t i = 1; i < iLen; i++) {
- w = csEntity[i];
- if (w < '0' || w > '9') {
- break;
- }
- ch = ch * 10 + w - '0';
- }
- }
- if (ch != 0) {
- m_pszData[m_iEntityStart++] = ch;
- }
- }
- } else {
- if (csEntity.Compare("amp") == 0) {
- if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_amp) == 0) {
- m_pszData[m_iEntityStart++] = '&';
- }
- } else if (csEntity.Compare("lt") == 0) {
- if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_lt) == 0) {
- m_pszData[m_iEntityStart++] = '<';
- }
- } else if (csEntity.Compare("gt") == 0) {
- if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_gt) == 0) {
- m_pszData[m_iEntityStart++] = '>';
- }
- } else if (csEntity.Compare("apos") == 0) {
- if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_apos) == 0) {
- m_pszData[m_iEntityStart++] = '\'';
- }
- } else if (csEntity.Compare("quot") == 0) {
- if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_quot) == 0) {
- m_pszData[m_iEntityStart++] = '\"';
- }
- }
- }
- }
- if (iSaveEntityStart != m_iEntityStart) {
- m_iDataPos = m_iEntityStart;
- m_iEntityStart = -1;
- } else {
- m_iDataPos++;
- m_iEntityStart = -1;
- }
- } else {
- if (m_iEntityStart < 0 && ch == '&') {
- m_iEntityStart = m_iDataPos;
- }
- m_iDataPos++;
- }
-}
-void CFX_SAXReader::ParseText() {
- if (m_CurByte == '<') {
- if (m_iDataPos > 0) {
- m_iDataLength = m_iDataPos;
- m_iDataPos = 0;
- if (m_pHandler) {
- NotifyData();
- }
- }
- Push();
- m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex;
- m_eMode = FX_SAXMODE_NodeStart;
- return;
- }
- if (m_iDataPos < 1 && SkipSpace(m_CurByte)) {
- return;
- }
- ParseChar(m_CurByte);
-}
-void CFX_SAXReader::ParseNodeStart() {
- if (m_CurByte == '?') {
- m_pCurItem->m_eNode = FX_SAXNODE_Instruction;
- m_eMode = FX_SAXMODE_TagName;
- return;
- }
- if (m_CurByte == '!') {
- m_eMode = FX_SAXMODE_DeclOrComment;
- return;
- }
- if (m_CurByte == '/') {
- m_eMode = FX_SAXMODE_TagEnd;
- return;
- }
- if (m_CurByte == '>') {
- Pop();
- m_eMode = FX_SAXMODE_Text;
- return;
- }
- if (m_CurByte > 0x20) {
- m_dwDataOffset = m_File.m_dwBufIndex;
- m_pCurItem->m_eNode = FX_SAXNODE_Tag;
- m_eMode = FX_SAXMODE_TagName;
- AppendData(m_CurByte);
- }
-}
-void CFX_SAXReader::ParseDeclOrComment() {
- if (m_CurByte == '-') {
- m_eMode = FX_SAXMODE_Comment;
- m_pCurItem->m_eNode = FX_SAXNODE_Comment;
- if (m_pCommentContext == NULL) {
- m_pCommentContext = new CFX_SAXCommentContext;
- }
- m_pCommentContext->m_iHeaderCount = 1;
- m_pCommentContext->m_iTailCount = 0;
- } else {
- m_eMode = FX_SAXMODE_DeclNode;
- m_dwDataOffset = m_File.m_dwBufIndex;
- m_SkipChar = '>';
- m_SkipStack.Add('>');
- SkipNode();
- }
-}
-void CFX_SAXReader::ParseComment() {
- m_pCommentContext->m_iHeaderCount = 2;
- m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex;
- m_eMode = FX_SAXMODE_CommentContent;
-}
-void CFX_SAXReader::ParseCommentContent() {
- if (m_CurByte == '-') {
- m_pCommentContext->m_iTailCount++;
- } else if (m_CurByte == '>' && m_pCommentContext->m_iTailCount == 2) {
- m_iDataLength = m_iDataPos;
- m_iDataPos = 0;
- if (m_pHandler) {
- NotifyTargetData();
- }
- Pop();
- m_eMode = FX_SAXMODE_Text;
- } else {
- while (m_pCommentContext->m_iTailCount > 0) {
- AppendData('-');
- m_pCommentContext->m_iTailCount--;
- }
- AppendData(m_CurByte);
- }
-}
-void CFX_SAXReader::ParseDeclNode() {
- SkipNode();
-}
-void CFX_SAXReader::ParseTagName() {
- if (m_CurByte < 0x21 || m_CurByte == '/' || m_CurByte == '>' ||
- m_CurByte == '?') {
- m_iDataLength = m_iDataPos;
- m_iDataPos = 0;
- if (m_pHandler) {
- NotifyEnter();
- }
- if (m_CurByte < 0x21) {
- m_eMode = FX_SAXMODE_TagAttributeName;
- } else if (m_CurByte == '/' || m_CurByte == '?') {
- m_ePrevMode = m_eMode;
- m_eMode = FX_SAXMODE_TagMaybeClose;
- } else {
- if (m_pHandler) {
- NotifyBreak();
- }
- m_eMode = FX_SAXMODE_Text;
- }
- } else {
- AppendData(m_CurByte);
- }
-}
-void CFX_SAXReader::ParseTagAttributeName() {
- if (m_CurByte < 0x21 || m_CurByte == '=') {
- if (m_iDataPos < 1 && m_CurByte < 0x21) {
- return;
- }
- m_iNameLength = m_iDataPos;
- m_iDataPos = 0;
- m_SkipChar = 0;
- m_eMode = m_CurByte == '=' ? FX_SAXMODE_TagAttributeValue
- : FX_SAXMODE_TagAttributeEqual;
- return;
- }
- if (m_CurByte == '/' || m_CurByte == '>' || m_CurByte == '?') {
- if (m_CurByte == '/' || m_CurByte == '?') {
- m_ePrevMode = m_eMode;
- m_eMode = FX_SAXMODE_TagMaybeClose;
- } else {
- if (m_pHandler) {
- NotifyBreak();
- }
- m_eMode = FX_SAXMODE_Text;
- }
- return;
- }
- if (m_iDataPos < 1) {
- m_dwDataOffset = m_File.m_dwBufIndex;
- }
- AppendName(m_CurByte);
-}
-void CFX_SAXReader::ParseTagAttributeEqual() {
- if (m_CurByte == '=') {
- m_SkipChar = 0;
- m_eMode = FX_SAXMODE_TagAttributeValue;
- return;
- } else if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) {
- m_iDataPos = m_iNameLength;
- AppendName(0x20);
- m_eMode = FX_SAXMODE_TargetData;
- ParseTargetData();
- }
-}
-void CFX_SAXReader::ParseTagAttributeValue() {
- if (m_SkipChar) {
- if (m_SkipChar == m_CurByte) {
- {
- m_iDataLength = m_iDataPos;
- m_iDataPos = 0;
- if (m_pHandler) {
- NotifyAttribute();
- }
- }
- m_SkipChar = 0;
- m_eMode = FX_SAXMODE_TagAttributeName;
- return;
- }
- ParseChar(m_CurByte);
- return;
- }
- if (m_CurByte < 0x21) {
- return;
- }
- if (m_iDataPos < 1) {
- if (m_CurByte == '\'' || m_CurByte == '\"') {
- m_SkipChar = m_CurByte;
- }
- }
-}
-void CFX_SAXReader::ParseMaybeClose() {
- if (m_CurByte == '>') {
- if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) {
- m_iNameLength = m_iDataPos;
- m_iDataPos = 0;
- if (m_pHandler) {
- NotifyTargetData();
- }
- }
- ParseTagClose();
- m_eMode = FX_SAXMODE_Text;
- } else if (m_ePrevMode == FX_SAXMODE_TagName) {
- AppendData('/');
- m_eMode = FX_SAXMODE_TagName;
- m_ePrevMode = FX_SAXMODE_Text;
- ParseTagName();
- } else if (m_ePrevMode == FX_SAXMODE_TagAttributeName) {
- AppendName('/');
- m_eMode = FX_SAXMODE_TagAttributeName;
- m_ePrevMode = FX_SAXMODE_Text;
- ParseTagAttributeName();
- } else if (m_ePrevMode == FX_SAXMODE_TargetData) {
- AppendName('?');
- m_eMode = FX_SAXMODE_TargetData;
- m_ePrevMode = FX_SAXMODE_Text;
- ParseTargetData();
- }
-}
-void CFX_SAXReader::ParseTagClose() {
- m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex;
- if (m_pHandler) {
- NotifyClose();
- }
- Pop();
-}
-void CFX_SAXReader::ParseTagEnd() {
- if (m_CurByte < 0x21) {
- return;
- }
- if (m_CurByte == '>') {
- Pop();
- m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex;
- m_iDataLength = m_iDataPos;
- m_iDataPos = 0;
- if (m_pHandler) {
- NotifyEnd();
- }
- Pop();
- m_eMode = FX_SAXMODE_Text;
- } else {
- ParseChar(m_CurByte);
- }
-}
-void CFX_SAXReader::ParseTargetData() {
- if (m_CurByte == '?') {
- m_ePrevMode = m_eMode;
- m_eMode = FX_SAXMODE_TagMaybeClose;
- } else {
- AppendName(m_CurByte);
- }
-}
-void CFX_SAXReader::SkipNode() {
- int32_t iLen = m_SkipStack.GetSize();
- if (m_SkipChar == '\'' || m_SkipChar == '\"') {
- if (m_CurByte != m_SkipChar) {
- return;
- }
- iLen--;
- FXSYS_assert(iLen > -1);
- m_SkipStack.RemoveAt(iLen, 1);
- m_SkipChar = iLen ? m_SkipStack[iLen - 1] : 0;
- return;
- }
- switch (m_CurByte) {
- case '<':
- m_SkipChar = '>';
- m_SkipStack.Add('>');
- break;
- case '[':
- m_SkipChar = ']';
- m_SkipStack.Add(']');
- break;
- case '(':
- m_SkipChar = ')';
- m_SkipStack.Add(')');
- break;
- case '\'':
- m_SkipChar = '\'';
- m_SkipStack.Add('\'');
- break;
- case '\"':
- m_SkipChar = '\"';
- m_SkipStack.Add('\"');
- break;
- default:
- if (m_CurByte == m_SkipChar) {
- iLen--;
- m_SkipStack.RemoveAt(iLen, 1);
- m_SkipChar = iLen ? m_SkipStack[iLen - 1] : 0;
- if (iLen == 0 && m_CurByte == '>') {
- m_iDataLength = m_iDataPos;
- m_iDataPos = 0;
- if (m_iDataLength >= 9 &&
- FXSYS_memcmp(m_pszData, "[CDATA[", 7 * sizeof(uint8_t)) == 0 &&
- FXSYS_memcmp(m_pszData + m_iDataLength - 2, "]]",
- 2 * sizeof(uint8_t)) == 0) {
- Pop();
- m_iDataLength -= 9;
- m_dwDataOffset += 7;
- FXSYS_memmove(m_pszData, m_pszData + 7,
- m_iDataLength * sizeof(uint8_t));
- m_bCharData = TRUE;
- if (m_pHandler) {
- NotifyData();
- }
- m_bCharData = FALSE;
- } else {
- Pop();
- }
- m_eMode = FX_SAXMODE_Text;
- }
- }
- break;
- }
- if (iLen > 0) {
- ParseChar(m_CurByte);
- }
-}
-void CFX_SAXReader::NotifyData() {
- FXSYS_assert(m_pHandler != NULL);
- if (m_pCurItem->m_eNode == FX_SAXNODE_Tag)
- m_pHandler->OnTagData(m_pCurItem->m_pNode,
- m_bCharData ? FX_SAXNODE_CharData : FX_SAXNODE_Text,
- CFX_ByteStringC(m_pszData, m_iDataLength),
- m_File.m_dwCur + m_dwDataOffset);
-}
-void CFX_SAXReader::NotifyEnter() {
- FXSYS_assert(m_pHandler != NULL);
- if (m_pCurItem->m_eNode == FX_SAXNODE_Tag ||
- m_pCurItem->m_eNode == FX_SAXNODE_Instruction) {
- m_pCurItem->m_pNode =
- m_pHandler->OnTagEnter(CFX_ByteStringC(m_pszData, m_iDataLength),
- m_pCurItem->m_eNode, m_dwNodePos);
- }
-}
-void CFX_SAXReader::NotifyAttribute() {
- FXSYS_assert(m_pHandler != NULL);
- if (m_pCurItem->m_eNode == FX_SAXNODE_Tag ||
- m_pCurItem->m_eNode == FX_SAXNODE_Instruction) {
- m_pHandler->OnTagAttribute(m_pCurItem->m_pNode,
- CFX_ByteStringC(m_pszName, m_iNameLength),
- CFX_ByteStringC(m_pszData, m_iDataLength));
- }
-}
-void CFX_SAXReader::NotifyBreak() {
- FXSYS_assert(m_pHandler != NULL);
- if (m_pCurItem->m_eNode == FX_SAXNODE_Tag) {
- m_pHandler->OnTagBreak(m_pCurItem->m_pNode);
- }
-}
-void CFX_SAXReader::NotifyClose() {
- FXSYS_assert(m_pHandler != NULL);
- if (m_pCurItem->m_eNode == FX_SAXNODE_Tag ||
- m_pCurItem->m_eNode == FX_SAXNODE_Instruction) {
- m_pHandler->OnTagClose(m_pCurItem->m_pNode, m_dwNodePos);
- }
-}
-void CFX_SAXReader::NotifyEnd() {
- FXSYS_assert(m_pHandler != NULL);
- if (m_pCurItem->m_eNode == FX_SAXNODE_Tag) {
- m_pHandler->OnTagEnd(m_pCurItem->m_pNode,
- CFX_ByteStringC(m_pszData, m_iDataLength),
- m_dwNodePos);
- }
-}
-void CFX_SAXReader::NotifyTargetData() {
- FXSYS_assert(m_pHandler != NULL);
- if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) {
- m_pHandler->OnTargetData(m_pCurItem->m_pNode, m_pCurItem->m_eNode,
- CFX_ByteStringC(m_pszName, m_iNameLength),
- m_dwNodePos);
- } else if (m_pCurItem->m_eNode == FX_SAXNODE_Comment) {
- m_pHandler->OnTargetData(m_pCurItem->m_pNode, m_pCurItem->m_eNode,
- CFX_ByteStringC(m_pszData, m_iDataLength),
- m_dwNodePos);
- }
-}
-void CFX_SAXReader::SkipCurrentNode() {
- if (!m_pCurItem) {
- return;
- }
- m_pCurItem->m_bSkip = TRUE;
-}
-void CFX_SAXReader::SetHandler(IFX_SAXReaderHandler* pHandler) {
- m_pHandler = pHandler;
-}
+// Copyright 2014 PDFium Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. + +// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com + +#include <algorithm> + +#include "xfa/src/fgas/src/fgas_base.h" +#include "fx_sax_imp.h" + +namespace { + +const FX_DWORD kSaxFileBufSize = 32768; + +} // namespace + +IFX_SAXReader* FX_SAXReader_Create() { + return new CFX_SAXReader; +} +CFX_SAXFile::CFX_SAXFile() + : m_pFile(NULL), + m_dwStart(0), + m_dwEnd(0), + m_dwCur(0), + m_pBuf(NULL), + m_dwBufSize(0), + m_dwBufIndex(0) {} +FX_BOOL CFX_SAXFile::StartFile(IFX_FileRead* pFile, + FX_DWORD dwStart, + FX_DWORD dwLen) { + FXSYS_assert(m_pFile == NULL && pFile != NULL); + FX_DWORD dwSize = pFile->GetSize(); + if (dwStart >= dwSize) { + return FALSE; + } + if (dwLen == -1 || dwStart + dwLen > dwSize) { + dwLen = dwSize - dwStart; + } + if (dwLen == 0) { + return FALSE; + } + m_dwBufSize = std::min(dwLen, kSaxFileBufSize); + m_pBuf = FX_Alloc(uint8_t, m_dwBufSize); + if (!pFile->ReadBlock(m_pBuf, dwStart, m_dwBufSize)) { + return FALSE; + } + m_dwStart = dwStart; + m_dwEnd = dwStart + dwLen; + m_dwCur = dwStart; + m_pFile = pFile; + m_dwBufIndex = 0; + return TRUE; +} +FX_BOOL CFX_SAXFile::ReadNextBlock() { + FXSYS_assert(m_pFile != NULL); + FX_DWORD dwSize = m_dwEnd - m_dwCur; + if (dwSize == 0) { + return FALSE; + } + m_dwBufSize = std::min(dwSize, kSaxFileBufSize); + if (!m_pFile->ReadBlock(m_pBuf, m_dwCur, m_dwBufSize)) { + return FALSE; + } + m_dwBufIndex = 0; + return TRUE; +} +void CFX_SAXFile::Reset() { + if (m_pBuf) { + FX_Free(m_pBuf); + m_pBuf = NULL; + } + m_pFile = NULL; +} +CFX_SAXReader::CFX_SAXReader() + : m_File(), + m_pHandler(nullptr), + m_iState(-1), + m_pRoot(nullptr), + m_pCurItem(nullptr), + m_dwItemID(0), + m_iDataSize(256), + m_iNameSize(256), + m_dwParseMode(0), + m_pCommentContext(nullptr) { + m_pszData = FX_Alloc(uint8_t, m_iDataSize); + m_pszName = FX_Alloc(uint8_t, m_iNameSize); +} +CFX_SAXReader::~CFX_SAXReader() { + Reset(); + if (m_pszData) { + FX_Free(m_pszData); + m_pszData = NULL; + } + if (m_pszName) { + FX_Free(m_pszName); + m_pszName = NULL; + } +} +void CFX_SAXReader::Reset() { + m_File.Reset(); + CFX_SAXItem* pItem = m_pRoot; + while (pItem) { + CFX_SAXItem* pNext = pItem->m_pNext; + delete pItem; + pItem = pNext; + } + m_pRoot = NULL; + m_pCurItem = NULL; + m_dwItemID = 0; + m_SkipStack.RemoveAll(); + m_SkipChar = 0; + m_iDataLength = 0; + m_iEntityStart = -1; + m_iNameLength = 0; + m_iDataPos = 0; + if (m_pCommentContext) { + delete m_pCommentContext; + m_pCommentContext = NULL; + } +} +inline void CFX_SAXReader::Push() { + CFX_SAXItem* pNew = new CFX_SAXItem; + pNew->m_dwID = ++m_dwItemID; + pNew->m_bSkip = m_pCurItem->m_bSkip; + pNew->m_pPrev = m_pCurItem; + m_pCurItem->m_pNext = pNew; + m_pCurItem = pNew; +} +inline void CFX_SAXReader::Pop() { + if (!m_pCurItem) { + return; + } + CFX_SAXItem* pPrev = m_pCurItem->m_pPrev; + pPrev->m_pNext = NULL; + delete m_pCurItem; + m_pCurItem = pPrev; +} +inline void CFX_SAXReader::AppendData(uint8_t ch) { + ReallocDataBuffer(); + m_pszData[m_iDataPos++] = ch; +} +inline void CFX_SAXReader::AppendName(uint8_t ch) { + ReallocNameBuffer(); + m_pszName[m_iDataPos++] = ch; +} +void CFX_SAXReader::ReallocDataBuffer() { + if (m_iDataPos < m_iDataSize) { + return; + } + if (m_iDataSize <= 1024 * 1024) { + m_iDataSize *= 2; + } else { + m_iDataSize += 1024 * 1024; + } + m_pszData = (uint8_t*)FX_Realloc(uint8_t, m_pszData, m_iDataSize); +} +void CFX_SAXReader::ReallocNameBuffer() { + if (m_iDataPos < m_iNameSize) { + return; + } + if (m_iNameSize <= 1024 * 1024) { + m_iNameSize *= 2; + } else { + m_iNameSize += 1024 * 1024; + } + m_pszName = (uint8_t*)FX_Realloc(uint8_t, m_pszName, m_iNameSize); +} +inline FX_BOOL CFX_SAXReader::SkipSpace(uint8_t ch) { + return (m_dwParseMode & FX_SAXPARSEMODE_NotSkipSpace) == 0 && ch < 0x21; +} +int32_t CFX_SAXReader::StartParse(IFX_FileRead* pFile, + FX_DWORD dwStart, + FX_DWORD dwLen, + FX_DWORD dwParseMode) { + m_iState = -1; + Reset(); + if (!m_File.StartFile(pFile, dwStart, dwLen)) { + return -1; + } + m_iState = 0; + m_eMode = FX_SAXMODE_Text; + m_ePrevMode = FX_SAXMODE_Text; + m_bCharData = FALSE; + m_dwDataOffset = 0; + m_pRoot = m_pCurItem = new CFX_SAXItem; + m_pCurItem->m_dwID = ++m_dwItemID; + m_dwParseMode = dwParseMode; + return 0; +} +typedef void (CFX_SAXReader::*FX_SAXReader_LPFParse)(); +static const FX_SAXReader_LPFParse g_FX_SAXReader_LPFParse[FX_SAXMODE_MAX] = { + &CFX_SAXReader::ParseText, + &CFX_SAXReader::ParseNodeStart, + &CFX_SAXReader::ParseDeclOrComment, + &CFX_SAXReader::ParseDeclNode, + &CFX_SAXReader::ParseComment, + &CFX_SAXReader::ParseCommentContent, + &CFX_SAXReader::ParseTagName, + &CFX_SAXReader::ParseTagAttributeName, + &CFX_SAXReader::ParseTagAttributeEqual, + &CFX_SAXReader::ParseTagAttributeValue, + &CFX_SAXReader::ParseMaybeClose, + &CFX_SAXReader::ParseTagClose, + &CFX_SAXReader::ParseTagEnd, + &CFX_SAXReader::ParseTargetData, +}; +int32_t CFX_SAXReader::ContinueParse(IFX_Pause* pPause) { + if (m_iState < 0 || m_iState > 99) { + return m_iState; + } + while (m_File.m_dwCur < m_File.m_dwEnd) { + FX_DWORD& index = m_File.m_dwBufIndex; + FX_DWORD size = m_File.m_dwBufSize; + const uint8_t* pBuf = m_File.m_pBuf; + while (index < size) { + m_CurByte = pBuf[index]; + (this->*g_FX_SAXReader_LPFParse[m_eMode])(); + index++; + } + m_File.m_dwCur += index; + m_iState = (m_File.m_dwCur - m_File.m_dwStart) * 100 / + (m_File.m_dwEnd - m_File.m_dwStart); + if (m_File.m_dwCur >= m_File.m_dwEnd) { + break; + } + if (!m_File.ReadNextBlock()) { + m_iState = -2; + break; + } + m_dwDataOffset = 0; + if (pPause && pPause->NeedToPauseNow()) { + break; + } + } + return m_iState; +} +void CFX_SAXReader::ParseChar(uint8_t ch) { + ReallocDataBuffer(); + m_pszData[m_iDataPos] = ch; + if (m_iEntityStart > -1 && ch == ';') { + int32_t iSaveEntityStart = m_iEntityStart; + CFX_ByteString csEntity(m_pszData + m_iEntityStart + 1, + m_iDataPos - m_iEntityStart - 1); + int32_t iLen = csEntity.GetLength(); + if (iLen > 0) { + if (csEntity[0] == '#') { + if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_sharp) == 0) { + ch = 0; + uint8_t w; + if (iLen > 1 && csEntity[1] == 'x') { + for (int32_t i = 2; i < iLen; i++) { + w = csEntity[i]; + if (w >= '0' && w <= '9') { + ch = (ch << 4) + w - '0'; + } else if (w >= 'A' && w <= 'F') { + ch = (ch << 4) + w - 55; + } else if (w >= 'a' && w <= 'f') { + ch = (ch << 4) + w - 87; + } else { + break; + } + } + } else { + for (int32_t i = 1; i < iLen; i++) { + w = csEntity[i]; + if (w < '0' || w > '9') { + break; + } + ch = ch * 10 + w - '0'; + } + } + if (ch != 0) { + m_pszData[m_iEntityStart++] = ch; + } + } + } else { + if (csEntity.Compare("amp") == 0) { + if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_amp) == 0) { + m_pszData[m_iEntityStart++] = '&'; + } + } else if (csEntity.Compare("lt") == 0) { + if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_lt) == 0) { + m_pszData[m_iEntityStart++] = '<'; + } + } else if (csEntity.Compare("gt") == 0) { + if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_gt) == 0) { + m_pszData[m_iEntityStart++] = '>'; + } + } else if (csEntity.Compare("apos") == 0) { + if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_apos) == 0) { + m_pszData[m_iEntityStart++] = '\''; + } + } else if (csEntity.Compare("quot") == 0) { + if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_quot) == 0) { + m_pszData[m_iEntityStart++] = '\"'; + } + } + } + } + if (iSaveEntityStart != m_iEntityStart) { + m_iDataPos = m_iEntityStart; + m_iEntityStart = -1; + } else { + m_iDataPos++; + m_iEntityStart = -1; + } + } else { + if (m_iEntityStart < 0 && ch == '&') { + m_iEntityStart = m_iDataPos; + } + m_iDataPos++; + } +} +void CFX_SAXReader::ParseText() { + if (m_CurByte == '<') { + if (m_iDataPos > 0) { + m_iDataLength = m_iDataPos; + m_iDataPos = 0; + if (m_pHandler) { + NotifyData(); + } + } + Push(); + m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; + m_eMode = FX_SAXMODE_NodeStart; + return; + } + if (m_iDataPos < 1 && SkipSpace(m_CurByte)) { + return; + } + ParseChar(m_CurByte); +} +void CFX_SAXReader::ParseNodeStart() { + if (m_CurByte == '?') { + m_pCurItem->m_eNode = FX_SAXNODE_Instruction; + m_eMode = FX_SAXMODE_TagName; + return; + } + if (m_CurByte == '!') { + m_eMode = FX_SAXMODE_DeclOrComment; + return; + } + if (m_CurByte == '/') { + m_eMode = FX_SAXMODE_TagEnd; + return; + } + if (m_CurByte == '>') { + Pop(); + m_eMode = FX_SAXMODE_Text; + return; + } + if (m_CurByte > 0x20) { + m_dwDataOffset = m_File.m_dwBufIndex; + m_pCurItem->m_eNode = FX_SAXNODE_Tag; + m_eMode = FX_SAXMODE_TagName; + AppendData(m_CurByte); + } +} +void CFX_SAXReader::ParseDeclOrComment() { + if (m_CurByte == '-') { + m_eMode = FX_SAXMODE_Comment; + m_pCurItem->m_eNode = FX_SAXNODE_Comment; + if (m_pCommentContext == NULL) { + m_pCommentContext = new CFX_SAXCommentContext; + } + m_pCommentContext->m_iHeaderCount = 1; + m_pCommentContext->m_iTailCount = 0; + } else { + m_eMode = FX_SAXMODE_DeclNode; + m_dwDataOffset = m_File.m_dwBufIndex; + m_SkipChar = '>'; + m_SkipStack.Add('>'); + SkipNode(); + } +} +void CFX_SAXReader::ParseComment() { + m_pCommentContext->m_iHeaderCount = 2; + m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; + m_eMode = FX_SAXMODE_CommentContent; +} +void CFX_SAXReader::ParseCommentContent() { + if (m_CurByte == '-') { + m_pCommentContext->m_iTailCount++; + } else if (m_CurByte == '>' && m_pCommentContext->m_iTailCount == 2) { + m_iDataLength = m_iDataPos; + m_iDataPos = 0; + if (m_pHandler) { + NotifyTargetData(); + } + Pop(); + m_eMode = FX_SAXMODE_Text; + } else { + while (m_pCommentContext->m_iTailCount > 0) { + AppendData('-'); + m_pCommentContext->m_iTailCount--; + } + AppendData(m_CurByte); + } +} +void CFX_SAXReader::ParseDeclNode() { + SkipNode(); +} +void CFX_SAXReader::ParseTagName() { + if (m_CurByte < 0x21 || m_CurByte == '/' || m_CurByte == '>' || + m_CurByte == '?') { + m_iDataLength = m_iDataPos; + m_iDataPos = 0; + if (m_pHandler) { + NotifyEnter(); + } + if (m_CurByte < 0x21) { + m_eMode = FX_SAXMODE_TagAttributeName; + } else if (m_CurByte == '/' || m_CurByte == '?') { + m_ePrevMode = m_eMode; + m_eMode = FX_SAXMODE_TagMaybeClose; + } else { + if (m_pHandler) { + NotifyBreak(); + } + m_eMode = FX_SAXMODE_Text; + } + } else { + AppendData(m_CurByte); + } +} +void CFX_SAXReader::ParseTagAttributeName() { + if (m_CurByte < 0x21 || m_CurByte == '=') { + if (m_iDataPos < 1 && m_CurByte < 0x21) { + return; + } + m_iNameLength = m_iDataPos; + m_iDataPos = 0; + m_SkipChar = 0; + m_eMode = m_CurByte == '=' ? FX_SAXMODE_TagAttributeValue + : FX_SAXMODE_TagAttributeEqual; + return; + } + if (m_CurByte == '/' || m_CurByte == '>' || m_CurByte == '?') { + if (m_CurByte == '/' || m_CurByte == '?') { + m_ePrevMode = m_eMode; + m_eMode = FX_SAXMODE_TagMaybeClose; + } else { + if (m_pHandler) { + NotifyBreak(); + } + m_eMode = FX_SAXMODE_Text; + } + return; + } + if (m_iDataPos < 1) { + m_dwDataOffset = m_File.m_dwBufIndex; + } + AppendName(m_CurByte); +} +void CFX_SAXReader::ParseTagAttributeEqual() { + if (m_CurByte == '=') { + m_SkipChar = 0; + m_eMode = FX_SAXMODE_TagAttributeValue; + return; + } else if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { + m_iDataPos = m_iNameLength; + AppendName(0x20); + m_eMode = FX_SAXMODE_TargetData; + ParseTargetData(); + } +} +void CFX_SAXReader::ParseTagAttributeValue() { + if (m_SkipChar) { + if (m_SkipChar == m_CurByte) { + { + m_iDataLength = m_iDataPos; + m_iDataPos = 0; + if (m_pHandler) { + NotifyAttribute(); + } + } + m_SkipChar = 0; + m_eMode = FX_SAXMODE_TagAttributeName; + return; + } + ParseChar(m_CurByte); + return; + } + if (m_CurByte < 0x21) { + return; + } + if (m_iDataPos < 1) { + if (m_CurByte == '\'' || m_CurByte == '\"') { + m_SkipChar = m_CurByte; + } + } +} +void CFX_SAXReader::ParseMaybeClose() { + if (m_CurByte == '>') { + if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { + m_iNameLength = m_iDataPos; + m_iDataPos = 0; + if (m_pHandler) { + NotifyTargetData(); + } + } + ParseTagClose(); + m_eMode = FX_SAXMODE_Text; + } else if (m_ePrevMode == FX_SAXMODE_TagName) { + AppendData('/'); + m_eMode = FX_SAXMODE_TagName; + m_ePrevMode = FX_SAXMODE_Text; + ParseTagName(); + } else if (m_ePrevMode == FX_SAXMODE_TagAttributeName) { + AppendName('/'); + m_eMode = FX_SAXMODE_TagAttributeName; + m_ePrevMode = FX_SAXMODE_Text; + ParseTagAttributeName(); + } else if (m_ePrevMode == FX_SAXMODE_TargetData) { + AppendName('?'); + m_eMode = FX_SAXMODE_TargetData; + m_ePrevMode = FX_SAXMODE_Text; + ParseTargetData(); + } +} +void CFX_SAXReader::ParseTagClose() { + m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; + if (m_pHandler) { + NotifyClose(); + } + Pop(); +} +void CFX_SAXReader::ParseTagEnd() { + if (m_CurByte < 0x21) { + return; + } + if (m_CurByte == '>') { + Pop(); + m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; + m_iDataLength = m_iDataPos; + m_iDataPos = 0; + if (m_pHandler) { + NotifyEnd(); + } + Pop(); + m_eMode = FX_SAXMODE_Text; + } else { + ParseChar(m_CurByte); + } +} +void CFX_SAXReader::ParseTargetData() { + if (m_CurByte == '?') { + m_ePrevMode = m_eMode; + m_eMode = FX_SAXMODE_TagMaybeClose; + } else { + AppendName(m_CurByte); + } +} +void CFX_SAXReader::SkipNode() { + int32_t iLen = m_SkipStack.GetSize(); + if (m_SkipChar == '\'' || m_SkipChar == '\"') { + if (m_CurByte != m_SkipChar) { + return; + } + iLen--; + FXSYS_assert(iLen > -1); + m_SkipStack.RemoveAt(iLen, 1); + m_SkipChar = iLen ? m_SkipStack[iLen - 1] : 0; + return; + } + switch (m_CurByte) { + case '<': + m_SkipChar = '>'; + m_SkipStack.Add('>'); + break; + case '[': + m_SkipChar = ']'; + m_SkipStack.Add(']'); + break; + case '(': + m_SkipChar = ')'; + m_SkipStack.Add(')'); + break; + case '\'': + m_SkipChar = '\''; + m_SkipStack.Add('\''); + break; + case '\"': + m_SkipChar = '\"'; + m_SkipStack.Add('\"'); + break; + default: + if (m_CurByte == m_SkipChar) { + iLen--; + m_SkipStack.RemoveAt(iLen, 1); + m_SkipChar = iLen ? m_SkipStack[iLen - 1] : 0; + if (iLen == 0 && m_CurByte == '>') { + m_iDataLength = m_iDataPos; + m_iDataPos = 0; + if (m_iDataLength >= 9 && + FXSYS_memcmp(m_pszData, "[CDATA[", 7 * sizeof(uint8_t)) == 0 && + FXSYS_memcmp(m_pszData + m_iDataLength - 2, "]]", + 2 * sizeof(uint8_t)) == 0) { + Pop(); + m_iDataLength -= 9; + m_dwDataOffset += 7; + FXSYS_memmove(m_pszData, m_pszData + 7, + m_iDataLength * sizeof(uint8_t)); + m_bCharData = TRUE; + if (m_pHandler) { + NotifyData(); + } + m_bCharData = FALSE; + } else { + Pop(); + } + m_eMode = FX_SAXMODE_Text; + } + } + break; + } + if (iLen > 0) { + ParseChar(m_CurByte); + } +} +void CFX_SAXReader::NotifyData() { + FXSYS_assert(m_pHandler != NULL); + if (m_pCurItem->m_eNode == FX_SAXNODE_Tag) + m_pHandler->OnTagData(m_pCurItem->m_pNode, + m_bCharData ? FX_SAXNODE_CharData : FX_SAXNODE_Text, + CFX_ByteStringC(m_pszData, m_iDataLength), + m_File.m_dwCur + m_dwDataOffset); +} +void CFX_SAXReader::NotifyEnter() { + FXSYS_assert(m_pHandler != NULL); + if (m_pCurItem->m_eNode == FX_SAXNODE_Tag || + m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { + m_pCurItem->m_pNode = + m_pHandler->OnTagEnter(CFX_ByteStringC(m_pszData, m_iDataLength), + m_pCurItem->m_eNode, m_dwNodePos); + } +} +void CFX_SAXReader::NotifyAttribute() { + FXSYS_assert(m_pHandler != NULL); + if (m_pCurItem->m_eNode == FX_SAXNODE_Tag || + m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { + m_pHandler->OnTagAttribute(m_pCurItem->m_pNode, + CFX_ByteStringC(m_pszName, m_iNameLength), + CFX_ByteStringC(m_pszData, m_iDataLength)); + } +} +void CFX_SAXReader::NotifyBreak() { + FXSYS_assert(m_pHandler != NULL); + if (m_pCurItem->m_eNode == FX_SAXNODE_Tag) { + m_pHandler->OnTagBreak(m_pCurItem->m_pNode); + } +} +void CFX_SAXReader::NotifyClose() { + FXSYS_assert(m_pHandler != NULL); + if (m_pCurItem->m_eNode == FX_SAXNODE_Tag || + m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { + m_pHandler->OnTagClose(m_pCurItem->m_pNode, m_dwNodePos); + } +} +void CFX_SAXReader::NotifyEnd() { + FXSYS_assert(m_pHandler != NULL); + if (m_pCurItem->m_eNode == FX_SAXNODE_Tag) { + m_pHandler->OnTagEnd(m_pCurItem->m_pNode, + CFX_ByteStringC(m_pszData, m_iDataLength), + m_dwNodePos); + } +} +void CFX_SAXReader::NotifyTargetData() { + FXSYS_assert(m_pHandler != NULL); + if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { + m_pHandler->OnTargetData(m_pCurItem->m_pNode, m_pCurItem->m_eNode, + CFX_ByteStringC(m_pszName, m_iNameLength), + m_dwNodePos); + } else if (m_pCurItem->m_eNode == FX_SAXNODE_Comment) { + m_pHandler->OnTargetData(m_pCurItem->m_pNode, m_pCurItem->m_eNode, + CFX_ByteStringC(m_pszData, m_iDataLength), + m_dwNodePos); + } +} +void CFX_SAXReader::SkipCurrentNode() { + if (!m_pCurItem) { + return; + } + m_pCurItem->m_bSkip = TRUE; +} +void CFX_SAXReader::SetHandler(IFX_SAXReaderHandler* pHandler) { + m_pHandler = pHandler; +} diff --git a/xfa/src/fgas/src/xml/fx_sax_imp.h b/xfa/src/fgas/src/xml/fx_sax_imp.h index 71be388efd..76809b9503 100644 --- a/xfa/src/fgas/src/xml/fx_sax_imp.h +++ b/xfa/src/fgas/src/xml/fx_sax_imp.h @@ -1,134 +1,134 @@ -// Copyright 2014 PDFium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
-
-#ifndef _FX_SAX_IMP_
-#define _FX_SAX_IMP_
-
-class CFX_SAXFile {
- public:
- CFX_SAXFile();
- FX_BOOL StartFile(IFX_FileRead* pFile, FX_DWORD dwStart, FX_DWORD dwLen);
- FX_BOOL ReadNextBlock();
- void Reset();
- IFX_FileRead* m_pFile;
- FX_DWORD m_dwStart;
- FX_DWORD m_dwEnd;
- FX_DWORD m_dwCur;
- uint8_t* m_pBuf;
- FX_DWORD m_dwBufSize;
- FX_DWORD m_dwBufIndex;
-};
-enum FX_SAXMODE {
- FX_SAXMODE_Text = 0,
- FX_SAXMODE_NodeStart,
- FX_SAXMODE_DeclOrComment,
- FX_SAXMODE_DeclNode,
- FX_SAXMODE_Comment,
- FX_SAXMODE_CommentContent,
- FX_SAXMODE_TagName,
- FX_SAXMODE_TagAttributeName,
- FX_SAXMODE_TagAttributeEqual,
- FX_SAXMODE_TagAttributeValue,
- FX_SAXMODE_TagMaybeClose,
- FX_SAXMODE_TagClose,
- FX_SAXMODE_TagEnd,
- FX_SAXMODE_TargetData,
- FX_SAXMODE_MAX,
-};
-class CFX_SAXItem {
- public:
- CFX_SAXItem()
- : m_pNode(NULL),
- m_eNode(FX_SAXNODE_Unknown),
- m_dwID(0),
- m_bSkip(FALSE),
- m_pPrev(NULL),
- m_pNext(NULL) {}
- void* m_pNode;
- FX_SAXNODE m_eNode;
- FX_DWORD m_dwID;
- FX_BOOL m_bSkip;
- CFX_SAXItem* m_pPrev;
- CFX_SAXItem* m_pNext;
-};
-class CFX_SAXCommentContext {
- public:
- CFX_SAXCommentContext() : m_iHeaderCount(0), m_iTailCount(0) {}
- int32_t m_iHeaderCount;
- int32_t m_iTailCount;
-};
-class CFX_SAXReader : public IFX_SAXReader {
- public:
- CFX_SAXReader();
- ~CFX_SAXReader();
- virtual void Release() { delete this; }
- virtual int32_t StartParse(IFX_FileRead* pFile,
- FX_DWORD dwStart = 0,
- FX_DWORD dwLen = -1,
- FX_DWORD dwParseMode = 0);
- virtual int32_t ContinueParse(IFX_Pause* pPause = NULL);
- virtual void SkipCurrentNode();
- virtual void SetHandler(IFX_SAXReaderHandler* pHandler);
- void AppendData(uint8_t ch);
- void AppendName(uint8_t ch);
- void ParseText();
- void ParseNodeStart();
- void ParseInstruction();
- void ParseDeclOrComment();
- void ParseDeclNode();
- void ParseComment();
- void ParseCommentContent();
- void ParseTagName();
- void ParseTagAttributeName();
- void ParseTagAttributeEqual();
- void ParseTagAttributeValue();
- void ParseMaybeClose();
- void ParseTagClose();
- void ParseTagEnd();
- void ParseTargetData();
-
- protected:
- CFX_SAXFile m_File;
- IFX_SAXReaderHandler* m_pHandler;
- int32_t m_iState;
- CFX_SAXItem* m_pRoot;
- CFX_SAXItem* m_pCurItem;
- FX_DWORD m_dwItemID;
- FX_SAXMODE m_eMode;
- FX_SAXMODE m_ePrevMode;
- FX_BOOL m_bCharData;
- uint8_t m_CurByte;
- FX_DWORD m_dwDataOffset;
- CFX_ByteArray m_SkipStack;
- uint8_t m_SkipChar;
- FX_DWORD m_dwNodePos;
- uint8_t* m_pszData;
- int32_t m_iDataSize;
- int32_t m_iDataLength;
- int32_t m_iEntityStart;
- int32_t m_iDataPos;
- uint8_t* m_pszName;
- int32_t m_iNameSize;
- int32_t m_iNameLength;
- FX_DWORD m_dwParseMode;
- CFX_SAXCommentContext* m_pCommentContext;
- void Reset();
- void Push();
- void Pop();
- FX_BOOL SkipSpace(uint8_t ch);
- void SkipNode();
- void NotifyData();
- void NotifyEnter();
- void NotifyAttribute();
- void NotifyBreak();
- void NotifyClose();
- void NotifyEnd();
- void NotifyTargetData();
- void ReallocDataBuffer();
- void ReallocNameBuffer();
- void ParseChar(uint8_t ch);
-};
-#endif
+// Copyright 2014 PDFium Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. + +// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com + +#ifndef _FX_SAX_IMP_ +#define _FX_SAX_IMP_ + +class CFX_SAXFile { + public: + CFX_SAXFile(); + FX_BOOL StartFile(IFX_FileRead* pFile, FX_DWORD dwStart, FX_DWORD dwLen); + FX_BOOL ReadNextBlock(); + void Reset(); + IFX_FileRead* m_pFile; + FX_DWORD m_dwStart; + FX_DWORD m_dwEnd; + FX_DWORD m_dwCur; + uint8_t* m_pBuf; + FX_DWORD m_dwBufSize; + FX_DWORD m_dwBufIndex; +}; +enum FX_SAXMODE { + FX_SAXMODE_Text = 0, + FX_SAXMODE_NodeStart, + FX_SAXMODE_DeclOrComment, + FX_SAXMODE_DeclNode, + FX_SAXMODE_Comment, + FX_SAXMODE_CommentContent, + FX_SAXMODE_TagName, + FX_SAXMODE_TagAttributeName, + FX_SAXMODE_TagAttributeEqual, + FX_SAXMODE_TagAttributeValue, + FX_SAXMODE_TagMaybeClose, + FX_SAXMODE_TagClose, + FX_SAXMODE_TagEnd, + FX_SAXMODE_TargetData, + FX_SAXMODE_MAX, +}; +class CFX_SAXItem { + public: + CFX_SAXItem() + : m_pNode(NULL), + m_eNode(FX_SAXNODE_Unknown), + m_dwID(0), + m_bSkip(FALSE), + m_pPrev(NULL), + m_pNext(NULL) {} + void* m_pNode; + FX_SAXNODE m_eNode; + FX_DWORD m_dwID; + FX_BOOL m_bSkip; + CFX_SAXItem* m_pPrev; + CFX_SAXItem* m_pNext; +}; +class CFX_SAXCommentContext { + public: + CFX_SAXCommentContext() : m_iHeaderCount(0), m_iTailCount(0) {} + int32_t m_iHeaderCount; + int32_t m_iTailCount; +}; +class CFX_SAXReader : public IFX_SAXReader { + public: + CFX_SAXReader(); + ~CFX_SAXReader(); + virtual void Release() { delete this; } + virtual int32_t StartParse(IFX_FileRead* pFile, + FX_DWORD dwStart = 0, + FX_DWORD dwLen = -1, + FX_DWORD dwParseMode = 0); + virtual int32_t ContinueParse(IFX_Pause* pPause = NULL); + virtual void SkipCurrentNode(); + virtual void SetHandler(IFX_SAXReaderHandler* pHandler); + void AppendData(uint8_t ch); + void AppendName(uint8_t ch); + void ParseText(); + void ParseNodeStart(); + void ParseInstruction(); + void ParseDeclOrComment(); + void ParseDeclNode(); + void ParseComment(); + void ParseCommentContent(); + void ParseTagName(); + void ParseTagAttributeName(); + void ParseTagAttributeEqual(); + void ParseTagAttributeValue(); + void ParseMaybeClose(); + void ParseTagClose(); + void ParseTagEnd(); + void ParseTargetData(); + + protected: + CFX_SAXFile m_File; + IFX_SAXReaderHandler* m_pHandler; + int32_t m_iState; + CFX_SAXItem* m_pRoot; + CFX_SAXItem* m_pCurItem; + FX_DWORD m_dwItemID; + FX_SAXMODE m_eMode; + FX_SAXMODE m_ePrevMode; + FX_BOOL m_bCharData; + uint8_t m_CurByte; + FX_DWORD m_dwDataOffset; + CFX_ByteArray m_SkipStack; + uint8_t m_SkipChar; + FX_DWORD m_dwNodePos; + uint8_t* m_pszData; + int32_t m_iDataSize; + int32_t m_iDataLength; + int32_t m_iEntityStart; + int32_t m_iDataPos; + uint8_t* m_pszName; + int32_t m_iNameSize; + int32_t m_iNameLength; + FX_DWORD m_dwParseMode; + CFX_SAXCommentContext* m_pCommentContext; + void Reset(); + void Push(); + void Pop(); + FX_BOOL SkipSpace(uint8_t ch); + void SkipNode(); + void NotifyData(); + void NotifyEnter(); + void NotifyAttribute(); + void NotifyBreak(); + void NotifyClose(); + void NotifyEnd(); + void NotifyTargetData(); + void ReallocDataBuffer(); + void ReallocNameBuffer(); + void ParseChar(uint8_t ch); +}; +#endif |