#include "config.h"
#include "HTMLTokenizer.h"
#include "CSSHelper.h"
#include "Cache.h"
#include "CachedScript.h"
#include "DocLoader.h"
#include "DocumentFragment.h"
#include "Event.h"
#include "EventNames.h"
#include "Frame.h"
#include "FrameLoader.h"
#include "FrameView.h"
#include "HTMLElement.h"
#include "HTMLNames.h"
#include "HTMLParser.h"
#include "HTMLScriptElement.h"
#include "HTMLViewSourceDocument.h"
#include "ImageLoader.h"
#include "InspectorTimelineAgent.h"
#include "MappedAttribute.h"
#include "Page.h"
#include "PreloadScanner.h"
#include "ScriptController.h"
#include "ScriptSourceCode.h"
#include "ScriptValue.h"
#include "XSSAuditor.h"
#include <wtf/ASCIICType.h>
#include <wtf/CurrentTime.h>
#include "HTMLEntityNames.c"
#define PRELOAD_SCANNER_ENABLED 1
using namespace WTF;
using namespace std;
namespace WebCore {
using namespace HTMLNames;
static const int defaultTokenizerChunkSize = 4096;
static const double defaultTokenizerTimeDelay = 0.500;
static const char commentStart [] = "<!--";
static const char doctypeStart [] = "<!doctype";
static const char publicStart [] = "public";
static const char systemStart [] = "system";
static const char scriptEnd [] = "</script";
static const char xmpEnd [] = "</xmp";
static const char styleEnd [] = "</style";
static const char textareaEnd [] = "</textarea";
static const char titleEnd [] = "</title";
static const char iframeEnd [] = "</iframe";
static const UChar windowsLatin1ExtensionArray[32] = {
0x20AC, 0x0081, 0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021, 0x02C6, 0x2030, 0x0160, 0x2039, 0x0152, 0x008D, 0x017D, 0x008F, 0x0090, 0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014, 0x02DC, 0x2122, 0x0161, 0x203A, 0x0153, 0x009D, 0x017E, 0x0178 };
static inline UChar fixUpChar(UChar c)
{
if ((c & ~0x1F) != 0x0080)
return c;
return windowsLatin1ExtensionArray[c - 0x80];
}
static inline bool tagMatch(const char* s1, const UChar* s2, unsigned length)
{
for (unsigned i = 0; i != length; ++i) {
unsigned char c1 = s1[i];
unsigned char uc1 = toASCIIUpper(static_cast<char>(c1));
UChar c2 = s2[i];
if (c1 != c2 && uc1 != c2)
return false;
}
return true;
}
inline void Token::addAttribute(AtomicString& attrName, const AtomicString& attributeValue, bool viewSourceMode)
{
if (!attrName.isEmpty()) {
ASSERT(!attrName.contains('/'));
RefPtr<MappedAttribute> a = MappedAttribute::create(attrName, attributeValue);
if (!attrs) {
attrs = NamedMappedAttrMap::create();
attrs->reserveInitialCapacity(10);
}
attrs->insertAttribute(a.release(), viewSourceMode);
}
attrName = emptyAtom;
}
HTMLTokenizer::HTMLTokenizer(HTMLDocument* doc, bool reportErrors)
: Tokenizer()
, m_buffer(0)
, m_scriptCode(0)
, m_scriptCodeSize(0)
, m_scriptCodeCapacity(0)
, m_scriptCodeResync(0)
, m_executingScript(0)
, m_requestingScript(false)
, m_hasScriptsWaitingForStylesheets(false)
, m_timer(this, &HTMLTokenizer::timerFired)
, m_externalScriptsTimer(this, &HTMLTokenizer::executeExternalScriptsTimerFired)
, m_doc(doc)
, m_parser(new HTMLParser(doc, reportErrors))
, m_inWrite(false)
, m_fragment(false)
, m_scriptingPermission(FragmentScriptingAllowed)
{
begin();
}
HTMLTokenizer::HTMLTokenizer(HTMLViewSourceDocument* doc)
: Tokenizer(true)
, m_buffer(0)
, m_scriptCode(0)
, m_scriptCodeSize(0)
, m_scriptCodeCapacity(0)
, m_scriptCodeResync(0)
, m_executingScript(0)
, m_requestingScript(false)
, m_hasScriptsWaitingForStylesheets(false)
, m_timer(this, &HTMLTokenizer::timerFired)
, m_externalScriptsTimer(this, &HTMLTokenizer::executeExternalScriptsTimerFired)
, m_doc(doc)
, m_parser(0)
, m_inWrite(false)
, m_fragment(false)
, m_scriptingPermission(FragmentScriptingAllowed)
{
begin();
}
HTMLTokenizer::HTMLTokenizer(DocumentFragment* frag, FragmentScriptingPermission scriptingPermission)
: m_buffer(0)
, m_scriptCode(0)
, m_scriptCodeSize(0)
, m_scriptCodeCapacity(0)
, m_scriptCodeResync(0)
, m_executingScript(0)
, m_requestingScript(false)
, m_hasScriptsWaitingForStylesheets(false)
, m_timer(this, &HTMLTokenizer::timerFired)
, m_externalScriptsTimer(this, &HTMLTokenizer::executeExternalScriptsTimerFired)
, m_doc(frag->document())
, m_parser(new HTMLParser(frag, scriptingPermission))
, m_inWrite(false)
, m_fragment(true)
, m_scriptingPermission(scriptingPermission)
{
begin();
}
void HTMLTokenizer::reset()
{
ASSERT(m_executingScript == 0);
while (!m_pendingScripts.isEmpty()) {
CachedScript* cs = m_pendingScripts.first().get();
m_pendingScripts.removeFirst();
ASSERT(cache()->disabled() || cs->accessCount() > 0);
cs->removeClient(this);
}
fastFree(m_buffer);
m_buffer = m_dest = 0;
m_bufferSize = 0;
fastFree(m_scriptCode);
m_scriptCode = 0;
m_scriptCodeSize = m_scriptCodeCapacity = m_scriptCodeResync = 0;
m_timer.stop();
m_externalScriptsTimer.stop();
m_state.setAllowYield(false);
m_state.setForceSynchronous(false);
m_currentToken.reset();
m_doctypeToken.reset();
m_doctypeSearchCount = 0;
m_doctypeSecondarySearchCount = 0;
m_hasScriptsWaitingForStylesheets = false;
}
void HTMLTokenizer::begin()
{
m_executingScript = 0;
m_requestingScript = false;
m_hasScriptsWaitingForStylesheets = false;
m_state.setLoadingExtScript(false);
reset();
m_bufferSize = 254;
m_buffer = static_cast<UChar*>(fastMalloc(sizeof(UChar) * 254));
m_dest = m_buffer;
tquote = NoQuote;
searchCount = 0;
m_state.setEntityState(NoEntity);
m_scriptTagSrcAttrValue = String();
m_pendingSrc.clear();
m_currentPrependingSrc = 0;
m_noMoreData = false;
m_brokenComments = false;
m_brokenServer = false;
m_lineNumber = 0;
m_currentScriptTagStartLineNumber = 0;
m_currentTagStartLineNumber = 0;
m_state.setForceSynchronous(false);
Page* page = m_doc->page();
if (page && page->hasCustomHTMLTokenizerTimeDelay())
m_tokenizerTimeDelay = page->customHTMLTokenizerTimeDelay();
else
m_tokenizerTimeDelay = defaultTokenizerTimeDelay;
if (page && page->hasCustomHTMLTokenizerChunkSize())
m_tokenizerChunkSize = page->customHTMLTokenizerChunkSize();
else
m_tokenizerChunkSize = defaultTokenizerChunkSize;
}
void HTMLTokenizer::setForceSynchronous(bool force)
{
m_state.setForceSynchronous(force);
}
HTMLTokenizer::State HTMLTokenizer::processListing(SegmentedString list, State state)
{
while (!list.isEmpty()) {
if (state.skipLF()) {
state.setSkipLF(false);
if (*list == '\n') {
list.advance();
continue;
}
}
checkBuffer();
if (*list == '\n' || *list == '\r') {
if (state.discardLF())
state.setDiscardLF(false); else
*m_dest++ = '\n';
if (*list == '\r')
state.setSkipLF(true);
list.advance();
} else {
state.setDiscardLF(false);
*m_dest++ = *list;
list.advance();
}
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseNonHTMLText(SegmentedString& src, State state)
{
ASSERT(state.inTextArea() || state.inTitle() || state.inIFrame() || !state.hasEntityState());
ASSERT(!state.hasTagState());
ASSERT(state.inXmp() + state.inTextArea() + state.inTitle() + state.inStyle() + state.inScript() + state.inIFrame() == 1);
if (state.inScript() && !m_currentScriptTagStartLineNumber)
m_currentScriptTagStartLineNumber = m_lineNumber;
if (state.inComment())
state = parseComment(src, state);
int lastDecodedEntityPosition = -1;
while (!src.isEmpty()) {
checkScriptBuffer();
UChar ch = *src;
if (!m_scriptCodeResync && !m_brokenComments &&
!state.inXmp() && ch == '-' && m_scriptCodeSize >= 3 && !src.escaped() &&
m_scriptCode[m_scriptCodeSize - 3] == '<' && m_scriptCode[m_scriptCodeSize - 2] == '!' && m_scriptCode[m_scriptCodeSize - 1] == '-' &&
(lastDecodedEntityPosition < m_scriptCodeSize - 3)) {
state.setInComment(true);
state = parseComment(src, state);
continue;
}
if (m_scriptCodeResync && !tquote && ch == '>') {
src.advancePastNonNewline();
m_scriptCodeSize = m_scriptCodeResync - 1;
m_scriptCodeResync = 0;
m_scriptCode[m_scriptCodeSize] = m_scriptCode[m_scriptCodeSize + 1] = 0;
if (state.inScript())
state = scriptHandler(state);
else {
state = processListing(SegmentedString(m_scriptCode, m_scriptCodeSize), state);
processToken();
if (state.inStyle()) {
m_currentToken.tagName = styleTag.localName();
m_currentToken.beginTag = false;
} else if (state.inTextArea()) {
m_currentToken.tagName = textareaTag.localName();
m_currentToken.beginTag = false;
} else if (state.inTitle()) {
m_currentToken.tagName = titleTag.localName();
m_currentToken.beginTag = false;
} else if (state.inXmp()) {
m_currentToken.tagName = xmpTag.localName();
m_currentToken.beginTag = false;
} else if (state.inIFrame()) {
m_currentToken.tagName = iframeTag.localName();
m_currentToken.beginTag = false;
}
processToken();
state.setInStyle(false);
state.setInScript(false);
state.setInTextArea(false);
state.setInTitle(false);
state.setInXmp(false);
state.setInIFrame(false);
tquote = NoQuote;
m_scriptCodeSize = m_scriptCodeResync = 0;
}
return state;
}
if (!m_scriptCodeResync && !state.escaped() && !src.escaped() && (ch == '>' || ch == '/' || isASCIISpace(ch)) &&
m_scriptCodeSize >= m_searchStopperLength &&
tagMatch(m_searchStopper, m_scriptCode + m_scriptCodeSize - m_searchStopperLength, m_searchStopperLength) &&
(lastDecodedEntityPosition < m_scriptCodeSize - m_searchStopperLength)) {
m_scriptCodeResync = m_scriptCodeSize-m_searchStopperLength+1;
tquote = NoQuote;
continue;
}
if (m_scriptCodeResync && !state.escaped()) {
if (ch == '\"')
tquote = (tquote == NoQuote) ? DoubleQuote : ((tquote == SingleQuote) ? SingleQuote : NoQuote);
else if (ch == '\'')
tquote = (tquote == NoQuote) ? SingleQuote : (tquote == DoubleQuote) ? DoubleQuote : NoQuote;
else if (tquote != NoQuote && (ch == '\r' || ch == '\n'))
tquote = NoQuote;
}
state.setEscaped(!state.escaped() && ch == '\\');
if (!m_scriptCodeResync && (state.inTextArea() || state.inTitle() || state.inIFrame()) && !src.escaped() && ch == '&') {
UChar* scriptCodeDest = m_scriptCode + m_scriptCodeSize;
src.advancePastNonNewline();
state = parseEntity(src, scriptCodeDest, state, m_cBufferPos, true, false);
if (scriptCodeDest == m_scriptCode + m_scriptCodeSize)
lastDecodedEntityPosition = m_scriptCodeSize;
else
m_scriptCodeSize = scriptCodeDest - m_scriptCode;
} else {
m_scriptCode[m_scriptCodeSize++] = ch;
src.advance(m_lineNumber);
}
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::scriptHandler(State state)
{
bool doScriptExec = false;
int startLine = m_currentScriptTagStartLineNumber + 1;
m_currentScriptTagStartLineNumber = 0;
bool followingFrameset = (m_doc->body() && m_doc->body()->hasTagName(framesetTag));
CachedScript* cs = 0;
if (!inViewSourceMode()) {
if (!m_scriptTagSrcAttrValue.isEmpty() && m_doc->frame()) {
if (!m_parser->skipMode() && !followingFrameset) {
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("Requesting script at time %d\n", m_doc->elapsedTime());
#endif
if (!m_parserStopped && m_scriptNode->dispatchBeforeLoadEvent(m_scriptTagSrcAttrValue) &&
(cs = m_doc->docLoader()->requestScript(m_scriptTagSrcAttrValue, m_scriptTagCharsetAttrValue)))
m_pendingScripts.append(cs);
else
m_scriptNode = 0;
} else
m_scriptNode = 0;
m_scriptTagSrcAttrValue = String();
} else {
doScriptExec = m_scriptNode->shouldExecuteAsJavaScript();
#if ENABLE(XHTMLMP)
if (!doScriptExec)
m_doc->setShouldProcessNoscriptElement(true);
#endif
m_scriptNode = 0;
}
}
state = processListing(SegmentedString(m_scriptCode, m_scriptCodeSize), state);
RefPtr<Node> node = processToken();
if (node && m_scriptingPermission == FragmentScriptingNotAllowed) {
ExceptionCode ec;
node->remove(ec);
node = 0;
}
String scriptString = node ? node->textContent() : "";
m_currentToken.tagName = scriptTag.localName();
m_currentToken.beginTag = false;
processToken();
state.setInScript(false);
m_scriptCodeSize = m_scriptCodeResync = 0;
if (inViewSourceMode())
return state;
SegmentedString* savedPrependingSrc = m_currentPrependingSrc;
SegmentedString prependingSrc;
m_currentPrependingSrc = &prependingSrc;
if (!m_parser->skipMode() && !followingFrameset) {
if (cs) {
if (savedPrependingSrc)
savedPrependingSrc->append(m_src);
else
m_pendingSrc.prepend(m_src);
setSrc(SegmentedString());
m_state = state;
bool savedRequestingScript = m_requestingScript;
m_requestingScript = true;
cs->addClient(this);
m_requestingScript = savedRequestingScript;
state = m_state;
if (!m_pendingScripts.isEmpty())
state.setLoadingExtScript(true);
} else if (!m_fragment && doScriptExec) {
if (!m_executingScript)
m_pendingSrc.prepend(m_src);
else
prependingSrc = m_src;
setSrc(SegmentedString());
state = scriptExecution(ScriptSourceCode(scriptString, m_doc->frame() ? m_doc->frame()->document()->url() : KURL(), startLine), state);
}
}
if (!m_executingScript && !state.loadingExtScript()) {
m_src.append(m_pendingSrc);
m_pendingSrc.clear();
} else if (!prependingSrc.isEmpty()) {
m_currentPrependingSrc = savedPrependingSrc;
if (!m_pendingScripts.isEmpty()) {
if (m_currentPrependingSrc)
m_currentPrependingSrc->append(prependingSrc);
else
m_pendingSrc.prepend(prependingSrc);
} else {
m_state = state;
write(prependingSrc, false);
state = m_state;
}
}
#if PRELOAD_SCANNER_ENABLED
if (!m_pendingScripts.isEmpty() && !m_executingScript) {
if (!m_preloadScanner)
m_preloadScanner.set(new PreloadScanner(m_doc));
if (!m_preloadScanner->inProgress()) {
m_preloadScanner->begin();
m_preloadScanner->write(m_pendingSrc);
}
}
#endif
m_currentPrependingSrc = savedPrependingSrc;
return state;
}
HTMLTokenizer::State HTMLTokenizer::scriptExecution(const ScriptSourceCode& sourceCode, State state)
{
if (m_fragment || !m_doc->frame())
return state;
m_executingScript++;
SegmentedString* savedPrependingSrc = m_currentPrependingSrc;
SegmentedString prependingSrc;
m_currentPrependingSrc = &prependingSrc;
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("beginning script execution at %d\n", m_doc->elapsedTime());
#endif
m_state = state;
m_doc->frame()->script()->executeScript(sourceCode);
state = m_state;
state.setAllowYield(true);
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("ending script execution at %d\n", m_doc->elapsedTime());
#endif
m_executingScript--;
if (!m_executingScript && !state.loadingExtScript()) {
m_pendingSrc.prepend(prependingSrc);
m_src.append(m_pendingSrc);
m_pendingSrc.clear();
} else if (!prependingSrc.isEmpty()) {
m_currentPrependingSrc = savedPrependingSrc;
if (!m_pendingScripts.isEmpty()) {
if (m_currentPrependingSrc)
m_currentPrependingSrc->append(prependingSrc);
else
m_pendingSrc.prepend(prependingSrc);
#if PRELOAD_SCANNER_ENABLED
PreloadScanner documentWritePreloadScanner(m_doc);
documentWritePreloadScanner.begin();
documentWritePreloadScanner.write(prependingSrc);
documentWritePreloadScanner.end();
#endif
} else {
m_state = state;
write(prependingSrc, false);
state = m_state;
}
}
m_currentPrependingSrc = savedPrependingSrc;
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseComment(SegmentedString& src, State state)
{
checkScriptBuffer(src.length());
while (!src.isEmpty()) {
UChar ch = *src;
m_scriptCode[m_scriptCodeSize++] = ch;
if (ch == '>') {
bool handleBrokenComments = m_brokenComments && !(state.inScript() || state.inStyle());
int endCharsCount = 1; if (m_scriptCodeSize > 2 && m_scriptCode[m_scriptCodeSize-3] == '-' && m_scriptCode[m_scriptCodeSize-2] == '-') {
endCharsCount = 3;
} else if (m_scriptCodeSize > 3 && m_scriptCode[m_scriptCodeSize-4] == '-' && m_scriptCode[m_scriptCodeSize-3] == '-' &&
m_scriptCode[m_scriptCodeSize-2] == '!') {
endCharsCount = 4;
}
if (handleBrokenComments || endCharsCount > 1) {
src.advancePastNonNewline();
if (!(state.inTitle() || state.inScript() || state.inXmp() || state.inTextArea() || state.inStyle() || state.inIFrame())) {
checkScriptBuffer();
m_scriptCode[m_scriptCodeSize] = 0;
m_scriptCode[m_scriptCodeSize + 1] = 0;
m_currentToken.tagName = commentAtom;
m_currentToken.beginTag = true;
state = processListing(SegmentedString(m_scriptCode, m_scriptCodeSize - endCharsCount), state);
processToken();
m_currentToken.tagName = commentAtom;
m_currentToken.beginTag = false;
processToken();
m_scriptCodeSize = 0;
}
state.setInComment(false);
return state; }
}
src.advance(m_lineNumber);
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseServer(SegmentedString& src, State state)
{
checkScriptBuffer(src.length());
while (!src.isEmpty()) {
UChar ch = *src;
m_scriptCode[m_scriptCodeSize++] = ch;
if (ch == '>' && m_scriptCodeSize > 1 && m_scriptCode[m_scriptCodeSize - 2] == '%') {
src.advancePastNonNewline();
state.setInServer(false);
m_scriptCodeSize = 0;
return state; }
src.advance(m_lineNumber);
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseProcessingInstruction(SegmentedString& src, State state)
{
UChar oldchar = 0;
while (!src.isEmpty()) {
UChar chbegin = *src;
if (chbegin == '\'')
tquote = tquote == SingleQuote ? NoQuote : SingleQuote;
else if (chbegin == '\"')
tquote = tquote == DoubleQuote ? NoQuote : DoubleQuote;
else if (chbegin == '>' && (!tquote || oldchar == '?')) {
state.setInProcessingInstruction(false);
src.advancePastNonNewline();
state.setDiscardLF(true);
return state; }
src.advance(m_lineNumber);
oldchar = chbegin;
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseText(SegmentedString& src, State state)
{
while (!src.isEmpty()) {
UChar cc = *src;
if (state.skipLF()) {
state.setSkipLF(false);
if (cc == '\n') {
src.advancePastNewline(m_lineNumber);
continue;
}
}
checkBuffer();
if (cc == '\r') {
state.setSkipLF(true);
*m_dest++ = '\n';
} else
*m_dest++ = cc;
src.advance(m_lineNumber);
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseEntity(SegmentedString& src, UChar*& dest, State state, unsigned& cBufferPos, bool start, bool parsingTag)
{
if (start) {
cBufferPos = 0;
state.setEntityState(SearchEntity);
EntityUnicodeValue = 0;
}
while (!src.isEmpty()) {
UChar cc = *src;
switch (state.entityState()) {
case NoEntity:
ASSERT(state.entityState() != NoEntity);
return state;
case SearchEntity:
if (cc == '#') {
m_cBuffer[cBufferPos++] = cc;
src.advancePastNonNewline();
state.setEntityState(NumericSearch);
} else
state.setEntityState(EntityName);
break;
case NumericSearch:
if (cc == 'x' || cc == 'X') {
m_cBuffer[cBufferPos++] = cc;
src.advancePastNonNewline();
state.setEntityState(Hexadecimal);
} else if (cc >= '0' && cc <= '9')
state.setEntityState(Decimal);
else
state.setEntityState(SearchSemicolon);
break;
case Hexadecimal: {
int ll = min(src.length(), 10 - cBufferPos);
while (ll--) {
cc = *src;
if (!((cc >= '0' && cc <= '9') || (cc >= 'a' && cc <= 'f') || (cc >= 'A' && cc <= 'F'))) {
state.setEntityState(SearchSemicolon);
break;
}
int digit;
if (cc < 'A')
digit = cc - '0';
else
digit = (cc - 'A' + 10) & 0xF; EntityUnicodeValue = EntityUnicodeValue * 16 + digit;
m_cBuffer[cBufferPos++] = cc;
src.advancePastNonNewline();
}
if (cBufferPos == 10)
state.setEntityState(SearchSemicolon);
break;
}
case Decimal:
{
int ll = min(src.length(), 9-cBufferPos);
while (ll--) {
cc = *src;
if (!(cc >= '0' && cc <= '9')) {
state.setEntityState(SearchSemicolon);
break;
}
EntityUnicodeValue = EntityUnicodeValue * 10 + (cc - '0');
m_cBuffer[cBufferPos++] = cc;
src.advancePastNonNewline();
}
if (cBufferPos == 9)
state.setEntityState(SearchSemicolon);
break;
}
case EntityName:
{
int ll = min(src.length(), 9-cBufferPos);
while (ll--) {
cc = *src;
if (!((cc >= 'a' && cc <= 'z') || (cc >= '0' && cc <= '9') || (cc >= 'A' && cc <= 'Z'))) {
state.setEntityState(SearchSemicolon);
break;
}
m_cBuffer[cBufferPos++] = cc;
src.advancePastNonNewline();
}
if (cBufferPos == 9)
state.setEntityState(SearchSemicolon);
if (state.entityState() == SearchSemicolon) {
if (cBufferPos > 1) {
unsigned testedEntityNameLen = 0;
char tmpEntityNameBuffer[10];
ASSERT(cBufferPos < 10);
for (; testedEntityNameLen < cBufferPos; ++testedEntityNameLen) {
if (m_cBuffer[testedEntityNameLen] > 0x7e)
break;
tmpEntityNameBuffer[testedEntityNameLen] = m_cBuffer[testedEntityNameLen];
}
const Entity *e;
if (testedEntityNameLen == cBufferPos)
e = findEntity(tmpEntityNameBuffer, cBufferPos);
else
e = 0;
if (e)
EntityUnicodeValue = e->code;
if (parsingTag && EntityUnicodeValue > 255 && *src != ';')
EntityUnicodeValue = 0;
}
}
else
break;
}
case SearchSemicolon:
if (EntityUnicodeValue > 0 && EntityUnicodeValue <= 0x10FFFF) {
if (!inViewSourceMode()) {
if (*src == ';')
src.advancePastNonNewline();
if (EntityUnicodeValue <= 0xFFFF) {
checkBuffer();
src.push(fixUpChar(EntityUnicodeValue));
} else {
checkBuffer(2);
src.push(U16_LEAD(EntityUnicodeValue));
src.push(U16_TRAIL(EntityUnicodeValue));
}
} else {
checkBuffer(12);
*dest++ = '&';
for (unsigned i = 0; i < cBufferPos; i++)
dest[i] = m_cBuffer[i];
dest += cBufferPos;
if (*src == ';') {
*dest++ = ';';
src.advancePastNonNewline();
}
}
} else {
checkBuffer(11);
*dest++ = '&';
for (unsigned i = 0; i < cBufferPos; i++)
dest[i] = m_cBuffer[i];
dest += cBufferPos;
}
state.setEntityState(NoEntity);
return state;
}
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseDoctype(SegmentedString& src, State state)
{
ASSERT(state.inDoctype());
while (!src.isEmpty() && state.inDoctype()) {
UChar c = *src;
bool isWhitespace = c == '\r' || c == '\n' || c == '\t' || c == ' ';
switch (m_doctypeToken.state()) {
case DoctypeBegin: {
m_doctypeToken.setState(DoctypeBeforeName);
if (isWhitespace) {
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
}
break;
}
case DoctypeBeforeName: {
if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
if (inViewSourceMode())
processDoctypeToken();
} else if (isWhitespace) {
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else
m_doctypeToken.setState(DoctypeName);
break;
}
case DoctypeName: {
if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
processDoctypeToken();
} else if (isWhitespace) {
m_doctypeSearchCount = 0; m_doctypeSecondarySearchCount = 0; m_doctypeToken.setState(DoctypeAfterName);
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else {
src.advancePastNonNewline();
m_doctypeToken.m_name.append(c);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
}
break;
}
case DoctypeAfterName: {
if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
processDoctypeToken();
} else if (!isWhitespace) {
src.advancePastNonNewline();
if (toASCIILower(c) == publicStart[m_doctypeSearchCount]) {
m_doctypeSearchCount++;
if (m_doctypeSearchCount == 6)
m_doctypeToken.setState(DoctypeBeforePublicID);
} else if (m_doctypeSearchCount > 0) {
m_doctypeSearchCount = 0;
m_doctypeToken.setState(DoctypeBogus);
} else if (toASCIILower(c) == systemStart[m_doctypeSecondarySearchCount]) {
m_doctypeSecondarySearchCount++;
if (m_doctypeSecondarySearchCount == 6)
m_doctypeToken.setState(DoctypeBeforeSystemID);
} else {
m_doctypeSecondarySearchCount = 0;
m_doctypeToken.setState(DoctypeBogus);
}
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else {
src.advance(m_lineNumber); if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
}
break;
}
case DoctypeBeforePublicID: {
if (c == '\"' || c == '\'') {
tquote = c == '\"' ? DoubleQuote : SingleQuote;
m_doctypeToken.setState(DoctypePublicID);
src.advancePastNonNewline();
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
if (inViewSourceMode())
processDoctypeToken();
} else if (isWhitespace) {
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else
m_doctypeToken.setState(DoctypeBogus);
break;
}
case DoctypePublicID: {
if ((c == '\"' && tquote == DoubleQuote) || (c == '\'' && tquote == SingleQuote)) {
src.advancePastNonNewline();
m_doctypeToken.setState(DoctypeAfterPublicID);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
if (inViewSourceMode())
processDoctypeToken();
} else {
m_doctypeToken.m_publicID.append(c);
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
}
break;
}
case DoctypeAfterPublicID:
if (c == '\"' || c == '\'') {
tquote = c == '\"' ? DoubleQuote : SingleQuote;
m_doctypeToken.setState(DoctypeSystemID);
src.advancePastNonNewline();
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
processDoctypeToken();
} else if (isWhitespace) {
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else
m_doctypeToken.setState(DoctypeBogus);
break;
case DoctypeBeforeSystemID:
if (c == '\"' || c == '\'') {
tquote = c == '\"' ? DoubleQuote : SingleQuote;
m_doctypeToken.setState(DoctypeSystemID);
src.advancePastNonNewline();
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
} else if (isWhitespace) {
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else
m_doctypeToken.setState(DoctypeBogus);
break;
case DoctypeSystemID:
if ((c == '\"' && tquote == DoubleQuote) || (c == '\'' && tquote == SingleQuote)) {
src.advancePastNonNewline();
m_doctypeToken.setState(DoctypeAfterSystemID);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
if (inViewSourceMode())
processDoctypeToken();
} else {
m_doctypeToken.m_systemID.append(c);
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
}
break;
case DoctypeAfterSystemID:
if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
processDoctypeToken();
} else if (isWhitespace) {
src.advance(m_lineNumber);
if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
} else
m_doctypeToken.setState(DoctypeBogus);
break;
case DoctypeBogus:
if (c == '>') {
src.advancePastNonNewline();
state.setInDoctype(false);
if (inViewSourceMode())
processDoctypeToken();
} else {
src.advance(m_lineNumber); if (inViewSourceMode())
m_doctypeToken.m_source.append(c);
}
break;
default:
break;
}
}
return state;
}
HTMLTokenizer::State HTMLTokenizer::parseTag(SegmentedString& src, State state)
{
ASSERT(!state.hasEntityState());
unsigned cBufferPos = m_cBufferPos;
bool lastIsSlash = false;
while (!src.isEmpty()) {
checkBuffer();
switch (state.tagState()) {
case NoTag:
{
m_cBufferPos = cBufferPos;
return state;
}
case TagName:
{
if (searchCount > 0) {
if (*src == commentStart[searchCount]) {
searchCount++;
if (searchCount == 2)
m_doctypeSearchCount++; else
m_doctypeSearchCount = 0;
if (searchCount == 4) {
src.advancePastNonNewline();
m_dest = m_buffer; state.setInComment(true);
state.setTagState(NoTag);
if (!src.isEmpty() && *src == '>' && m_doc->inCompatMode()) {
state.setInComment(false);
src.advancePastNonNewline();
if (!src.isEmpty())
m_cBuffer[cBufferPos++] = *src;
} else
state = parseComment(src, state);
m_cBufferPos = cBufferPos;
return state; }
m_cBuffer[cBufferPos++] = *src;
src.advancePastNonNewline();
break;
} else
searchCount = 0; }
if (m_doctypeSearchCount > 0) {
if (toASCIILower(*src) == doctypeStart[m_doctypeSearchCount]) {
m_doctypeSearchCount++;
m_cBuffer[cBufferPos++] = *src;
src.advancePastNonNewline();
if (m_doctypeSearchCount == 9) {
state.setInDoctype(true);
state.setTagState(NoTag);
m_doctypeToken.reset();
if (inViewSourceMode())
m_doctypeToken.m_source.append(m_cBuffer, cBufferPos);
state = parseDoctype(src, state);
m_cBufferPos = cBufferPos;
return state;
}
break;
} else
m_doctypeSearchCount = 0; }
bool finish = false;
unsigned int ll = min(src.length(), CBUFLEN - cBufferPos);
while (ll--) {
UChar curchar = *src;
if (isASCIISpace(curchar) || curchar == '>' || curchar == '<') {
finish = true;
break;
}
if (curchar >= 'A' && curchar <= 'Z' && !inViewSourceMode())
m_cBuffer[cBufferPos++] = curchar + ('a' - 'A');
else
m_cBuffer[cBufferPos++] = curchar;
src.advancePastNonNewline();
}
if (finish || CBUFLEN == cBufferPos) {
bool beginTag;
UChar* ptr = m_cBuffer;
unsigned int len = cBufferPos;
m_cBuffer[cBufferPos] = '\0';
if ((cBufferPos > 0) && (*ptr == '/')) {
beginTag = false;
ptr++;
len--;
}
else
beginTag = true;
if (len > 1 && ptr[len-1] == '/' && !inViewSourceMode())
ptr[--len] = '\0';
if (ptr[0] != '!' || inViewSourceMode()) {
m_currentToken.tagName = AtomicString(ptr);
m_currentToken.beginTag = beginTag;
}
m_dest = m_buffer;
state.setTagState(SearchAttribute);
cBufferPos = 0;
}
break;
}
case SearchAttribute:
while (!src.isEmpty()) {
UChar curchar = *src;
if (!isASCIISpace(curchar) && curchar != '\'' && curchar != '"') {
if (curchar == '<' || curchar == '>')
state.setTagState(SearchEnd);
else
state.setTagState(AttributeName);
cBufferPos = 0;
break;
}
if (inViewSourceMode())
m_currentToken.addViewSourceChar(curchar);
src.advance(m_lineNumber);
}
break;
case AttributeName:
{
m_rawAttributeBeforeValue.clear();
int ll = min(src.length(), CBUFLEN - cBufferPos);
while (ll--) {
UChar curchar = *src;
if (curchar <= '>' && (curchar >= '<' || isASCIISpace(curchar) || curchar == '/')) {
m_cBuffer[cBufferPos] = '\0';
m_attrName = AtomicString(m_cBuffer);
m_dest = m_buffer;
*m_dest++ = 0;
state.setTagState(SearchEqual);
if (inViewSourceMode())
m_currentToken.addViewSourceChar('a');
break;
}
if (curchar >= 'A' && curchar <= 'Z' && !inViewSourceMode())
m_cBuffer[cBufferPos++] = curchar + ('a' - 'A');
else
m_cBuffer[cBufferPos++] = curchar;
m_rawAttributeBeforeValue.append(curchar);
src.advance(m_lineNumber);
}
if (cBufferPos == CBUFLEN) {
m_cBuffer[cBufferPos] = '\0';
m_attrName = AtomicString(m_cBuffer);
m_dest = m_buffer;
*m_dest++ = 0;
state.setTagState(SearchEqual);
if (inViewSourceMode())
m_currentToken.addViewSourceChar('a');
}
break;
}
case SearchEqual:
while (!src.isEmpty()) {
UChar curchar = *src;
if (lastIsSlash && curchar == '>') {
if (m_currentToken.tagName == scriptTag)
m_currentToken.selfClosingTag = true;
m_currentToken.brokenXMLStyle = true;
}
if (!isASCIISpace(curchar) && curchar != '\'' && curchar != '"' && curchar != '/') {
if (curchar == '=') {
state.setTagState(SearchValue);
if (inViewSourceMode())
m_currentToken.addViewSourceChar(curchar);
m_rawAttributeBeforeValue.append(curchar);
src.advancePastNonNewline();
} else {
m_currentToken.addAttribute(m_attrName, emptyAtom, inViewSourceMode());
m_dest = m_buffer;
state.setTagState(SearchAttribute);
lastIsSlash = false;
}
break;
}
lastIsSlash = curchar == '/';
if (inViewSourceMode())
m_currentToken.addViewSourceChar(curchar);
m_rawAttributeBeforeValue.append(curchar);
src.advance(m_lineNumber);
}
break;
case SearchValue:
while (!src.isEmpty()) {
UChar curchar = *src;
if (!isASCIISpace(curchar)) {
if (curchar == '\'' || curchar == '\"') {
tquote = curchar == '\"' ? DoubleQuote : SingleQuote;
state.setTagState(QuotedValue);
if (inViewSourceMode())
m_currentToken.addViewSourceChar(curchar);
m_rawAttributeBeforeValue.append(curchar);
src.advancePastNonNewline();
} else
state.setTagState(Value);
break;
}
if (inViewSourceMode())
m_currentToken.addViewSourceChar(curchar);
m_rawAttributeBeforeValue.append(curchar);
src.advance(m_lineNumber);
}
break;
case QuotedValue:
while (!src.isEmpty()) {
checkBuffer();
UChar curchar = *src;
if (curchar <= '>' && !src.escaped()) {
if (curchar == '>' && m_attrName.isEmpty()) {
while (m_dest > m_buffer + 1 && (m_dest[-1] == '\n' || m_dest[-1] == '\r'))
m_dest--; AtomicString attributeValue(m_buffer + 1, m_dest - m_buffer - 1);
if (!attributeValue.contains('/'))
m_attrName = attributeValue; m_currentToken.addAttribute(m_attrName, attributeValue, inViewSourceMode());
if (inViewSourceMode())
m_currentToken.addViewSourceChar('x');
state.setTagState(SearchAttribute);
m_dest = m_buffer;
tquote = NoQuote;
break;
}
if (curchar == '&') {
src.advancePastNonNewline();
state = parseEntity(src, m_dest, state, cBufferPos, true, true);
break;
}
if ((tquote == SingleQuote && curchar == '\'') || (tquote == DoubleQuote && curchar == '\"')) {
while (m_dest > m_buffer + 1 && (m_dest[-1] == '\n' || m_dest[-1] == '\r'))
m_dest--; AtomicString attributeValue(m_buffer + 1, m_dest - m_buffer - 1);
if (m_attrName.isEmpty() && !attributeValue.contains('/')) {
m_attrName = attributeValue; if (inViewSourceMode())
m_currentToken.addViewSourceChar('x');
} else if (inViewSourceMode())
m_currentToken.addViewSourceChar('v');
if (m_currentToken.beginTag && m_currentToken.tagName == scriptTag && !inViewSourceMode() && !m_parser->skipMode() && m_attrName == srcAttr) {
String context(m_rawAttributeBeforeValue.data(), m_rawAttributeBeforeValue.size());
if (m_XSSAuditor && !m_XSSAuditor->canLoadExternalScriptFromSrc(context, attributeValue))
attributeValue = blankURL().string();
}
m_currentToken.addAttribute(m_attrName, attributeValue, inViewSourceMode());
m_dest = m_buffer;
state.setTagState(SearchAttribute);
tquote = NoQuote;
if (inViewSourceMode())
m_currentToken.addViewSourceChar(curchar);
src.advancePastNonNewline();
break;
}
}
*m_dest++ = curchar;
src.advance(m_lineNumber);
}
break;
case Value:
while (!src.isEmpty()) {
checkBuffer();
UChar curchar = *src;
if (curchar <= '>' && !src.escaped()) {
if (curchar == '&') {
src.advancePastNonNewline();
state = parseEntity(src, m_dest, state, cBufferPos, true, true);
break;
}
if (isASCIISpace(curchar) || curchar == '>') {
AtomicString attributeValue(m_buffer + 1, m_dest - m_buffer - 1);
if (m_currentToken.beginTag && m_currentToken.tagName == scriptTag && !inViewSourceMode() && !m_parser->skipMode() && m_attrName == srcAttr) {
String context(m_rawAttributeBeforeValue.data(), m_rawAttributeBeforeValue.size());
if (m_XSSAuditor && !m_XSSAuditor->canLoadExternalScriptFromSrc(context, attributeValue))
attributeValue = blankURL().string();
}
m_currentToken.addAttribute(m_attrName, attributeValue, inViewSourceMode());
if (inViewSourceMode())
m_currentToken.addViewSourceChar('v');
m_dest = m_buffer;
state.setTagState(SearchAttribute);
break;
}
}
*m_dest++ = curchar;
src.advance(m_lineNumber);
}
break;
case SearchEnd:
{
while (!src.isEmpty()) {
UChar ch = *src;
if (ch == '>' || ch == '<')
break;
if (ch == '/')
m_currentToken.selfClosingTag = true;
if (inViewSourceMode())
m_currentToken.addViewSourceChar(ch);
src.advance(m_lineNumber);
}
if (src.isEmpty())
break;
searchCount = 0; state.setTagState(NoTag);
tquote = NoQuote;
if (*src != '<')
src.advance(m_lineNumber);
if (m_currentToken.tagName == nullAtom) { m_cBufferPos = cBufferPos;
return state;
}
AtomicString tagName = m_currentToken.tagName;
bool isSelfClosingScript = m_currentToken.selfClosingTag && m_currentToken.beginTag && m_currentToken.tagName == scriptTag;
bool beginTag = !m_currentToken.selfClosingTag && m_currentToken.beginTag;
if (m_currentToken.beginTag && m_currentToken.tagName == scriptTag && !inViewSourceMode() && !m_parser->skipMode()) {
Attribute* a = 0;
m_scriptTagSrcAttrValue = String();
m_scriptTagCharsetAttrValue = String();
if (m_currentToken.attrs && !m_fragment) {
if (m_doc->frame() && m_doc->frame()->script()->canExecuteScripts(NotAboutToExecuteScript)) {
if ((a = m_currentToken.attrs->getAttributeItem(srcAttr)))
m_scriptTagSrcAttrValue = m_doc->completeURL(deprecatedParseURL(a->value())).string();
}
}
}
RefPtr<Node> n = processToken();
m_cBufferPos = cBufferPos;
if (n || inViewSourceMode()) {
State savedState = state;
SegmentedString savedSrc = src;
long savedLineno = m_lineNumber;
if ((tagName == preTag || tagName == listingTag) && !inViewSourceMode()) {
if (beginTag)
state.setDiscardLF(true); } else if (tagName == scriptTag) {
ASSERT(!m_scriptNode);
m_scriptNode = static_pointer_cast<HTMLScriptElement>(n);
if (m_scriptNode)
m_scriptTagCharsetAttrValue = m_scriptNode->scriptCharset();
if (beginTag) {
m_searchStopper = scriptEnd;
m_searchStopperLength = 8;
state.setInScript(true);
state = parseNonHTMLText(src, state);
} else if (isSelfClosingScript) { state.setInScript(true);
state = scriptHandler(state);
}
} else if (tagName == styleTag) {
if (beginTag) {
m_searchStopper = styleEnd;
m_searchStopperLength = 7;
state.setInStyle(true);
state = parseNonHTMLText(src, state);
}
} else if (tagName == textareaTag) {
if (beginTag) {
m_searchStopper = textareaEnd;
m_searchStopperLength = 10;
state.setInTextArea(true);
state = parseNonHTMLText(src, state);
}
} else if (tagName == titleTag) {
if (beginTag) {
m_searchStopper = titleEnd;
m_searchStopperLength = 7;
state.setInTitle(true);
state = parseNonHTMLText(src, state);
}
} else if (tagName == xmpTag) {
if (beginTag) {
m_searchStopper = xmpEnd;
m_searchStopperLength = 5;
state.setInXmp(true);
state = parseNonHTMLText(src, state);
}
} else if (tagName == iframeTag) {
if (beginTag) {
m_searchStopper = iframeEnd;
m_searchStopperLength = 8;
state.setInIFrame(true);
state = parseNonHTMLText(src, state);
}
}
if (src.isEmpty() && (state.inTitle() || inViewSourceMode()) && !state.inComment() && !(state.inScript() && m_currentScriptTagStartLineNumber)) {
state = savedState;
src = savedSrc;
m_lineNumber = savedLineno;
m_scriptCodeSize = 0;
}
}
if (tagName == plaintextTag)
state.setInPlainText(beginTag);
return state; }
} }
m_cBufferPos = cBufferPos;
return state;
}
inline bool HTMLTokenizer::continueProcessing(int& processedCount, double startTime, State &state)
{
bool allowedYield = state.allowYield();
state.setAllowYield(false);
if (!state.loadingExtScript() && !state.forceSynchronous() && !m_executingScript && (processedCount > m_tokenizerChunkSize || allowedYield)) {
processedCount = 0;
if (currentTime() - startTime > m_tokenizerTimeDelay) {
m_timer.startOneShot(0);
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (currentTime() - startTime > m_tokenizerTimeDelay)
printf("Deferring processing of data because 500ms elapsed away from event loop.\n");
#endif
return false;
}
}
processedCount++;
return true;
}
void HTMLTokenizer::write(const SegmentedString& str, bool appendData)
{
if (!m_buffer)
return;
if (m_parserStopped)
return;
SegmentedString source(str);
if (m_executingScript)
source.setExcludeLineNumbers();
if ((m_executingScript && appendData) || !m_pendingScripts.isEmpty()) {
if (m_currentPrependingSrc)
m_currentPrependingSrc->append(source);
else {
m_pendingSrc.append(source);
#if PRELOAD_SCANNER_ENABLED
if (m_preloadScanner && m_preloadScanner->inProgress() && appendData)
m_preloadScanner->write(source);
#endif
}
return;
}
#if PRELOAD_SCANNER_ENABLED
if (m_preloadScanner && m_preloadScanner->inProgress() && appendData)
m_preloadScanner->end();
#endif
if (!m_src.isEmpty())
m_src.append(source);
else
setSrc(source);
if (m_timer.isActive())
return;
bool wasInWrite = m_inWrite;
m_inWrite = true;
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("Beginning write at time %d\n", m_doc->elapsedTime());
#endif
int processedCount = 0;
double startTime = currentTime();
#if ENABLE(INSPECTOR)
if (InspectorTimelineAgent* timelineAgent = m_doc->inspectorTimelineAgent())
timelineAgent->willWriteHTML(source.length(), m_lineNumber);
#endif
Frame* frame = m_doc->frame();
State state = m_state;
while (!m_src.isEmpty() && (!frame || !frame->redirectScheduler()->locationChangePending())) {
if (!continueProcessing(processedCount, startTime, state))
break;
checkBuffer();
UChar cc = *m_src;
bool wasSkipLF = state.skipLF();
if (wasSkipLF)
state.setSkipLF(false);
if (wasSkipLF && (cc == '\n'))
m_src.advance();
else if (state.needsSpecialWriteHandling()) {
if (state.hasEntityState())
state = parseEntity(m_src, m_dest, state, m_cBufferPos, false, state.hasTagState());
else if (state.inPlainText())
state = parseText(m_src, state);
else if (state.inAnyNonHTMLText())
state = parseNonHTMLText(m_src, state);
else if (state.inComment())
state = parseComment(m_src, state);
else if (state.inDoctype())
state = parseDoctype(m_src, state);
else if (state.inServer())
state = parseServer(m_src, state);
else if (state.inProcessingInstruction())
state = parseProcessingInstruction(m_src, state);
else if (state.hasTagState())
state = parseTag(m_src, state);
else if (state.startTag()) {
state.setStartTag(false);
switch (cc) {
case '/':
break;
case '!': {
searchCount = 1; m_doctypeSearchCount = 1;
break;
}
case '?': {
state.setInProcessingInstruction(true);
tquote = NoQuote;
state = parseProcessingInstruction(m_src, state);
continue;
break;
}
case '%':
if (!m_brokenServer) {
state.setInServer(true);
tquote = NoQuote;
state = parseServer(m_src, state);
continue;
}
default: {
if ( ((cc >= 'a') && (cc <= 'z')) || ((cc >= 'A') && (cc <= 'Z'))) {
} else {
*m_dest = '<';
m_dest++;
continue;
}
}
};
processToken();
m_cBufferPos = 0;
state.setTagState(TagName);
state = parseTag(m_src, state);
}
} else if (cc == '&' && !m_src.escaped()) {
m_src.advancePastNonNewline();
state = parseEntity(m_src, m_dest, state, m_cBufferPos, true, state.hasTagState());
} else if (cc == '<' && !m_src.escaped()) {
m_currentTagStartLineNumber = m_lineNumber;
m_src.advancePastNonNewline();
state.setStartTag(true);
state.setDiscardLF(false);
} else if (cc == '\n' || cc == '\r') {
if (state.discardLF())
state.setDiscardLF(false); else {
*m_dest++ = '\n';
if (cc == '\r' && !m_src.excludeLineNumbers())
m_lineNumber++;
}
if (cc == '\r')
state.setSkipLF(true);
m_src.advance(m_lineNumber);
} else {
state.setDiscardLF(false);
*m_dest++ = cc;
m_src.advancePastNonNewline();
}
}
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("Ending write at time %d\n", m_doc->elapsedTime());
#endif
#if ENABLE(INSPECTOR)
if (InspectorTimelineAgent* timelineAgent = m_doc->inspectorTimelineAgent())
timelineAgent->didWriteHTML(m_lineNumber);
#endif
m_inWrite = wasInWrite;
m_state = state;
if (m_noMoreData && !m_inWrite && !state.loadingExtScript() && !m_executingScript && !m_timer.isActive())
end();
if (!m_fragment)
ImageLoader::dispatchPendingBeforeLoadEvents();
}
void HTMLTokenizer::stopParsing()
{
Tokenizer::stopParsing();
m_timer.stop();
if (!m_fragment && m_doc->frame())
m_doc->frame()->loader()->tokenizerProcessedData();
}
bool HTMLTokenizer::processingData() const
{
return m_timer.isActive() || m_inWrite;
}
void HTMLTokenizer::timerFired(Timer<HTMLTokenizer>*)
{
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("Beginning timer write at time %d\n", m_doc->elapsedTime());
#endif
if (m_doc->view() && m_doc->view()->layoutPending() && !m_doc->minimumLayoutDelay()) {
m_timer.startOneShot(0);
return;
}
write(SegmentedString(), true);
}
void HTMLTokenizer::end()
{
ASSERT(!m_timer.isActive());
m_timer.stop();
if (m_buffer) {
if (!m_state.hasTagState())
processToken();
fastFree(m_scriptCode);
m_scriptCode = 0;
m_scriptCodeSize = m_scriptCodeCapacity = m_scriptCodeResync = 0;
fastFree(m_buffer);
m_buffer = 0;
}
if (!inViewSourceMode())
m_parser->finished();
else
m_doc->finishedParsing();
}
void HTMLTokenizer::finish()
{
while ((m_state.inComment() || m_state.inServer()) && m_scriptCode && m_scriptCodeSize) {
if (m_state.inComment())
m_brokenComments = true;
else
m_brokenServer = true;
checkScriptBuffer();
m_scriptCode[m_scriptCodeSize] = 0;
m_scriptCode[m_scriptCodeSize + 1] = 0;
int pos;
String food;
if (m_state.inScript() || m_state.inStyle() || m_state.inTextArea())
food = String(m_scriptCode, m_scriptCodeSize);
else if (m_state.inServer()) {
food = "<";
food.append(m_scriptCode, m_scriptCodeSize);
} else {
pos = find(m_scriptCode, m_scriptCodeSize, '>');
food = String(m_scriptCode + pos + 1, m_scriptCodeSize - pos - 1);
}
fastFree(m_scriptCode);
m_scriptCode = 0;
m_scriptCodeSize = m_scriptCodeCapacity = m_scriptCodeResync = 0;
m_state.setInComment(false);
m_state.setInServer(false);
if (!food.isEmpty())
write(food, true);
}
m_noMoreData = true;
if (!m_inWrite && !m_state.loadingExtScript() && !m_executingScript && !m_timer.isActive())
end(); }
PassRefPtr<Node> HTMLTokenizer::processToken()
{
ScriptController* scriptController = (!m_fragment && m_doc->frame()) ? m_doc->frame()->script() : 0;
if (scriptController && scriptController->canExecuteScripts(NotAboutToExecuteScript))
scriptController->setEventHandlerLineNumber(m_currentTagStartLineNumber + 1); if (m_dest > m_buffer) {
m_currentToken.text = StringImpl::createStrippingNullCharacters(m_buffer, m_dest - m_buffer);
if (m_currentToken.tagName != commentAtom)
m_currentToken.tagName = textAtom;
} else if (m_currentToken.tagName == nullAtom) {
m_currentToken.reset();
if (scriptController)
scriptController->setEventHandlerLineNumber(m_lineNumber + 1); return 0;
}
m_dest = m_buffer;
RefPtr<Node> n;
if (!m_parserStopped) {
if (NamedMappedAttrMap* map = m_currentToken.attrs.get())
map->shrinkToLength();
if (inViewSourceMode())
static_cast<HTMLViewSourceDocument*>(m_doc)->addViewSourceToken(&m_currentToken);
else
n = m_parser->parseToken(&m_currentToken);
}
m_currentToken.reset();
if (scriptController)
scriptController->setEventHandlerLineNumber(0);
return n.release();
}
void HTMLTokenizer::processDoctypeToken()
{
if (inViewSourceMode())
static_cast<HTMLViewSourceDocument*>(m_doc)->addViewSourceDoctypeToken(&m_doctypeToken);
else
m_parser->parseDoctypeToken(&m_doctypeToken);
}
HTMLTokenizer::~HTMLTokenizer()
{
ASSERT(!m_inWrite);
reset();
}
void HTMLTokenizer::enlargeBuffer(int len)
{
int delta = max(len, m_bufferSize);
static const int maxSize = INT_MAX / sizeof(UChar);
if (delta > maxSize - m_bufferSize)
CRASH();
int newSize = m_bufferSize + delta;
int oldOffset = m_dest - m_buffer;
m_buffer = static_cast<UChar*>(fastRealloc(m_buffer, newSize * sizeof(UChar)));
m_dest = m_buffer + oldOffset;
m_bufferSize = newSize;
}
void HTMLTokenizer::enlargeScriptBuffer(int len)
{
int delta = max(len, m_scriptCodeCapacity);
static const int maxSize = INT_MAX / sizeof(UChar);
if (delta > maxSize - m_scriptCodeCapacity)
CRASH();
int newSize = m_scriptCodeCapacity + delta;
if (!newSize) {
ASSERT(!m_scriptCode);
return;
}
m_scriptCode = static_cast<UChar*>(fastRealloc(m_scriptCode, newSize * sizeof(UChar)));
m_scriptCodeCapacity = newSize;
}
void HTMLTokenizer::executeScriptsWaitingForStylesheets()
{
ASSERT(m_doc->haveStylesheetsLoaded());
if (m_hasScriptsWaitingForStylesheets)
notifyFinished(0);
}
void HTMLTokenizer::notifyFinished(CachedResource*)
{
executeExternalScriptsIfReady();
}
void HTMLTokenizer::executeExternalScriptsIfReady()
{
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("script loaded at %d\n", m_doc->elapsedTime());
#endif
ASSERT(!m_pendingScripts.isEmpty());
m_hasScriptsWaitingForStylesheets = !m_doc->haveStylesheetsLoaded();
if (m_hasScriptsWaitingForStylesheets)
return;
bool finished = false;
double startTime = currentTime();
while (!finished && m_pendingScripts.first()->isLoaded()) {
if (!continueExecutingExternalScripts(startTime))
break;
CachedScript* cs = m_pendingScripts.first().get();
m_pendingScripts.removeFirst();
ASSERT(cache()->disabled() || cs->accessCount() > 0);
setSrc(SegmentedString());
ScriptSourceCode sourceCode(cs);
bool errorOccurred = cs->errorOccurred();
cs->removeClient(this);
RefPtr<Node> n = m_scriptNode.release();
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("external script beginning execution at %d\n", m_doc->elapsedTime());
#endif
if (errorOccurred)
n->dispatchEvent(Event::create(eventNames().errorEvent, true, false));
else {
if (static_cast<HTMLScriptElement*>(n.get())->shouldExecuteAsJavaScript())
m_state = scriptExecution(sourceCode, m_state);
#if ENABLE(XHTMLMP)
else
m_doc->setShouldProcessNoscriptElement(true);
#endif
n->dispatchEvent(Event::create(eventNames().loadEvent, false, false));
}
finished = m_pendingScripts.isEmpty();
if (finished) {
ASSERT(!m_hasScriptsWaitingForStylesheets);
m_state.setLoadingExtScript(false);
#ifdef INSTRUMENT_LAYOUT_SCHEDULING
if (!m_doc->ownerElement())
printf("external script finished execution at %d\n", m_doc->elapsedTime());
#endif
} else if (m_hasScriptsWaitingForStylesheets) {
finished = true;
}
if (!m_requestingScript) {
SegmentedString rest = m_pendingSrc;
m_pendingSrc.clear();
write(rest, false);
}
}
}
void HTMLTokenizer::executeExternalScriptsTimerFired(Timer<HTMLTokenizer>*)
{
if (m_doc->view() && m_doc->view()->layoutPending() && !m_doc->minimumLayoutDelay()) {
m_externalScriptsTimer.startOneShot(0);
return;
}
executeExternalScriptsIfReady();
}
bool HTMLTokenizer::continueExecutingExternalScripts(double startTime)
{
if (m_externalScriptsTimer.isActive())
return false;
if (currentTime() - startTime > m_tokenizerTimeDelay) {
m_externalScriptsTimer.startOneShot(0);
return false;
}
return true;
}
bool HTMLTokenizer::isWaitingForScripts() const
{
return m_state.loadingExtScript();
}
void HTMLTokenizer::setSrc(const SegmentedString& source)
{
m_src = source;
}
void parseHTMLDocumentFragment(const String& source, DocumentFragment* fragment, FragmentScriptingPermission scriptingPermission)
{
HTMLTokenizer tok(fragment, scriptingPermission);
tok.setForceSynchronous(true);
tok.write(source, true);
tok.finish();
ASSERT(!tok.processingData()); }
UChar decodeNamedEntity(const char* name)
{
const Entity* e = findEntity(name, strlen(name));
return e ? e->code : 0;
}
}