diff options
Diffstat (limited to 'WebCore/html/HTMLDocument.cpp')
| -rw-r--r-- | WebCore/html/HTMLDocument.cpp | 254 |
1 files changed, 216 insertions, 38 deletions
diff --git a/WebCore/html/HTMLDocument.cpp b/WebCore/html/HTMLDocument.cpp index eb9eb4d..8db8422 100644 --- a/WebCore/html/HTMLDocument.cpp +++ b/WebCore/html/HTMLDocument.cpp @@ -74,6 +74,11 @@ #include "DocTypeStrings.cpp" +#ifdef ANDROID_META_SUPPORT +#include "FrameTree.h" +#include "Settings.h" +#endif + namespace WebCore { using namespace HTMLNames; @@ -82,7 +87,6 @@ HTMLDocument::HTMLDocument(DOMImplementation* implementation, Frame* frame) : Document(implementation, frame) { clearXMLVersion(); - setParseMode(Compat); } HTMLDocument::~HTMLDocument() @@ -274,7 +278,7 @@ Tokenizer *HTMLDocument::createTokenizer() bool HTMLDocument::childAllowed(Node *newChild) { - return newChild->hasTagName(htmlTag) || newChild->isCommentNode() || (newChild->nodeType() == DOCUMENT_TYPE_NODE && !doctype()); + return newChild->hasTagName(htmlTag) || newChild->isCommentNode(); } PassRefPtr<Element> HTMLDocument::createElement(const String &name, ExceptionCode& ec) @@ -347,62 +351,236 @@ bool HTMLDocument::hasDocExtraNamedItem(const String& name) return docExtraNamedItemCounts.get(name.impl()) != 0; } -void HTMLDocument::determineParseMode() +const int PARSEMODE_HAVE_DOCTYPE = (1<<0); +const int PARSEMODE_HAVE_PUBLIC_ID = (1<<1); +const int PARSEMODE_HAVE_SYSTEM_ID = (1<<2); +const int PARSEMODE_HAVE_INTERNAL = (1<<3); + +static int parseDocTypePart(const String& buffer, int index) +{ + while (true) { + UChar ch = buffer[index]; + if (ch == ' ' || ch == '\t' || ch == '\n' || ch == '\r') + ++index; + else if (ch == '-') { + int tmpIndex=index; + if (buffer[index+1] == '-' && + ((tmpIndex=buffer.find("--", index+2)) != -1)) + index = tmpIndex+2; + else + return index; + } + else + return index; + } +} + +static bool containsString(const char* str, const String& buffer, int offset) +{ + String startString(str); + if (offset + startString.length() > buffer.length()) + return false; + + String bufferString = buffer.substring(offset, startString.length()).lower(); + String lowerStart = startString.lower(); + + return bufferString.startsWith(lowerStart); +} + +static bool parseDocTypeDeclaration(const String& buffer, + int* resultFlags, + String& name, + String& publicID, + String& systemID) { - // FIXME: It's terrible that this code runs separately and isn't just built in to the - // HTML tokenizer/parser. + bool haveDocType = false; + *resultFlags = 0; + + // Skip through any comments and processing instructions. + int index = 0; + do { + index = buffer.find('<', index); + if (index == -1) break; + UChar nextChar = buffer[index+1]; + if (nextChar == '!') { + if (containsString("doctype", buffer, index+2)) { + haveDocType = true; + index += 9; // Skip "<!DOCTYPE" + break; + } + index = parseDocTypePart(buffer,index); + index = buffer.find('>', index); + } + else if (nextChar == '?') + index = buffer.find('>', index); + else + break; + } while (index != -1); + + if (!haveDocType) + return true; + *resultFlags |= PARSEMODE_HAVE_DOCTYPE; + + index = parseDocTypePart(buffer, index); + if (!containsString("html", buffer, index)) + return false; + + name = buffer.substring(index, 4); + index = parseDocTypePart(buffer, index+4); + bool hasPublic = containsString("public", buffer, index); + if (hasPublic) { + index = parseDocTypePart(buffer, index+6); + + // We've read <!DOCTYPE HTML PUBLIC (not case sensitive). + // Now we find the beginning and end of the public identifers + // and system identifiers (assuming they're even present). + UChar theChar = buffer[index]; + if (theChar != '\"' && theChar != '\'') + return false; + + // |start| is the first character (after the quote) and |end| + // is the final quote, so there are |end|-|start| characters. + int publicIDStart = index+1; + int publicIDEnd = buffer.find(theChar, publicIDStart); + if (publicIDEnd == -1) + return false; + index = parseDocTypePart(buffer, publicIDEnd+1); + UChar next = buffer[index]; + if (next == '>') { + // Public identifier present, but no system identifier. + // Do nothing. Note that this is the most common + // case. + } + else if (next == '\"' || next == '\'') { + // We have a system identifier. + *resultFlags |= PARSEMODE_HAVE_SYSTEM_ID; + int systemIDStart = index+1; + int systemIDEnd = buffer.find(next, systemIDStart); + if (systemIDEnd == -1) + return false; + systemID = buffer.substring(systemIDStart, systemIDEnd - systemIDStart); + } + else if (next == '[') { + // We found an internal subset. + *resultFlags |= PARSEMODE_HAVE_INTERNAL; + } + else + return false; // Something's wrong. + + // We need to trim whitespace off the public identifier. + publicID = buffer.substring(publicIDStart, publicIDEnd - publicIDStart); + publicID = publicID.stripWhiteSpace(); + *resultFlags |= PARSEMODE_HAVE_PUBLIC_ID; + } else { + if (containsString("system", buffer, index)) { + // Doctype has a system ID but no public ID + *resultFlags |= PARSEMODE_HAVE_SYSTEM_ID; + index = parseDocTypePart(buffer, index+6); + UChar next = buffer[index]; + if (next != '\"' && next != '\'') + return false; + int systemIDStart = index+1; + int systemIDEnd = buffer.find(next, systemIDStart); + if (systemIDEnd == -1) + return false; + systemID = buffer.substring(systemIDStart, systemIDEnd - systemIDStart); + index = parseDocTypePart(buffer, systemIDEnd+1); + } + + UChar nextChar = buffer[index]; + if (nextChar == '[') + *resultFlags |= PARSEMODE_HAVE_INTERNAL; + else if (nextChar != '>') + return false; + } + + return true; +} +void HTMLDocument::determineParseMode(const String& str) +{ // This code more or less mimics Mozilla's implementation (specifically the // doctype parsing implemented by David Baron in Mozilla's nsParser.cpp). // // There are three possible parse modes: - // COMPAT - quirks mode emulates WinIE and NS4. CSS parsing is also relaxed in this mode, e.g., unit types can + // COMPAT - quirks mode emulates WinIE + // and NS4. CSS parsing is also relaxed in this mode, e.g., unit types can // be omitted from numbers. - // ALMOST STRICT - This mode is identical to strict mode except for its treatment of line-height in the inline box model. For - // now (until the inline box model is re-written), this mode is identical to STANDARDS mode. - // STRICT - no quirks apply. Web pages will obey the specifications to the letter. - bool wasInCompatMode = inCompatMode(); - DocumentType* docType = doctype(); - if (!docType || !equalIgnoringCase(docType->name(), "html")) - // No doctype found at all or the doctype is not HTML. Default to quirks mode and Html4. - setParseMode(Compat); - else if (!doctype()->systemId().isEmpty() && equalIgnoringCase(docType->systemId(), "http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd")) - // Assume quirks mode for this particular system ID. In the HTML5 spec, this is the only - // system identifier that is examined. - setParseMode(Compat); - else if (docType->publicId().isEmpty()) - // A doctype without a public ID means use strict mode. - setParseMode(Strict); - else { - // We have to check a list of public IDs to see what we - // should do. - String lowerPubID = docType->publicId().lower(); - CString pubIDStr = lowerPubID.latin1(); - - // Look up the entry in our gperf-generated table. - const PubIDInfo* doctypeEntry = findDoctypeEntry(pubIDStr.data(), pubIDStr.length()); - if (!doctypeEntry) - // The DOCTYPE is not in the list. Assume strict mode. + // ALMOST STRICT - This mode is identical to strict mode + // except for its treatment of line-height in the inline box model. For + // now (until the inline box model is re-written), this mode is identical + // to STANDARDS mode. + // STRICT - no quirks apply. Web pages will obey the specifications to + // the letter. + + String name, systemID, publicID; + int resultFlags = 0; + if (parseDocTypeDeclaration(str, &resultFlags, name, publicID, systemID)) { + if (resultFlags & PARSEMODE_HAVE_DOCTYPE) + setDocType(new DocumentType(this, name, publicID, systemID)); + if (!(resultFlags & PARSEMODE_HAVE_DOCTYPE)) { + // No doctype found at all. Default to quirks mode and Html4. + setParseMode(Compat); + setHTMLMode(Html4); + } + else if ((resultFlags & PARSEMODE_HAVE_INTERNAL) || + !(resultFlags & PARSEMODE_HAVE_PUBLIC_ID)) { + // Internal subsets always denote full standards, as does + // a doctype without a public ID. setParseMode(Strict); + setHTMLMode(Html4); + } else { - switch (docType->systemId().isEmpty() ? - doctypeEntry->mode_if_no_sysid : - doctypeEntry->mode_if_sysid) { + // We have to check a list of public IDs to see what we + // should do. + String lowerPubID = publicID.lower(); + CString pubIDStr = lowerPubID.latin1(); + +#ifdef ANDROID_META_SUPPORT + if ((!frame()->tree() || !frame()->tree()->parent()) && + strstr(pubIDStr.data(), "-//wapforum//dtd xhtml mobile 1.") == pubIDStr.data()) { + // fit mobile sites directly in the screen + frame()->settings()->setMetadataSettings("width", "device-width"); + } +#endif + // Look up the entry in our gperf-generated table. + const PubIDInfo* doctypeEntry = findDoctypeEntry(pubIDStr.data(), pubIDStr.length()); + if (!doctypeEntry) { + // The DOCTYPE is not in the list. Assume strict mode. + setParseMode(Strict); + setHTMLMode(Html4); + return; + } + + switch ((resultFlags & PARSEMODE_HAVE_SYSTEM_ID) ? + doctypeEntry->mode_if_sysid : + doctypeEntry->mode_if_no_sysid) + { case PubIDInfo::eQuirks3: + setParseMode(Compat); + setHTMLMode(Html3); + break; case PubIDInfo::eQuirks: setParseMode(Compat); + setHTMLMode(Html4); break; case PubIDInfo::eAlmostStandards: setParseMode(AlmostStrict); + setHTMLMode(Html4); break; default: ASSERT(false); } - } + } } - - if (inCompatMode() != wasInCompatMode) - updateStyleSelector(); + else { + // Malformed doctype implies quirks mode. + setParseMode(Compat); + setHTMLMode(Html3); + } + + styleSelector()->strictParsing = !inCompatMode(); + } } |