qxmlstream.g

奇趣公司比较新的qt/emd版本

    bool scanUntil(const char *str, short tokenToInject = -1);
    bool scanString(const char *str, short tokenToInject, bool requireSpace = true);
    inline void injectToken(ushort tokenToInject) {
        putChar(int(tokenToInject) << 16);
    }

    static bool validateName(const QStringRef &name);

    void parseEntity(const QString &value);
    QXmlStreamReaderPrivate *entityParser;

    bool scanAfterLangleBang();
    bool scanPublicOrSystem();
    bool scanNData();
    bool scanAfterDefaultDecl();
    bool scanAttType();


    // scan optimization functions. Not strictly necessary but LALR is
    // not very well suited for scanning fast
    int fastScanLiteralContent();
    int fastScanSpace();
    int fastScanContentCharList();
    int fastScanName(int *prefix = 0);
    inline int fastScanNMTOKEN();


    bool parse();
    inline void consumeRule(int);

    void raiseError(QXmlStreamReader::Error error, const QString& message = QString());
    void raiseWellFormedError(const QString &message);

private:
    /*! \internal
       Never assign to variable type directly. Instead use this function.

       This prevents errors from being ignored.
     */
    inline void setType(const QXmlStreamReader::TokenType t)
    {
        if(type != QXmlStreamReader::Invalid)
            type = t;
    }
};

bool QXmlStreamReaderPrivate::parse()
{
    // cleanup currently reported token

    switch (type) {
    case QXmlStreamReader::StartElement:
        name.clear();
        prefix.clear();
	qualifiedName.clear();
        namespaceUri.clear();
        if (publicNamespaceDeclarations.size())
            publicNamespaceDeclarations.clear();
        if (attributes.size())
            attributes.resize(0);
        if (isEmptyElement) {
            setType(QXmlStreamReader::EndElement);
            Tag &tag = tagStack_pop();
            namespaceUri = tag.namespaceDeclaration.namespaceUri;
            name = tag.name;
	    qualifiedName = tag.qualifiedName;
            isEmptyElement = false;
            return true;
        }
        clearTextBuffer();
        break;
    case QXmlStreamReader::EndElement:
        name.clear();
        prefix.clear();
	qualifiedName.clear();
        namespaceUri.clear();
        clearTextBuffer();
        break;
    case QXmlStreamReader::DTD:
        publicNotationDeclarations.clear();
        publicEntityDeclarations.clear();
        // fall through
    case QXmlStreamReader::Comment:
    case QXmlStreamReader::Characters:
        isCDATA = isWhitespace = false;
        text.clear();
        clearTextBuffer();
        break;
    case QXmlStreamReader::EntityReference:
        text.clear();
        name.clear();
        clearTextBuffer();
        break;
    case QXmlStreamReader::ProcessingInstruction:
        processingInstructionTarget.clear();
        processingInstructionData.clear();
	clearTextBuffer();
        break;
    case QXmlStreamReader::NoToken:
    case QXmlStreamReader::Invalid:
        break;
    case QXmlStreamReader::StartDocument:
	lockEncoding = true;
#ifndef QT_NO_TEXTCODEC
	if(decoder->hasFailure()) {
	    raiseWellFormedError(QXmlStream::tr("Encountered incorrectly encoded content."));
	    readBuffer.clear();
	    return false;
	}
#endif
        // fall through
    default:
        clearTextBuffer();
        ;
    }

    setType(QXmlStreamReader::NoToken);


    // the main parse loop
    int act, r;

    if (resumeReduction) {
        act = state_stack[tos-1];
        r = resumeReduction;
        resumeReduction = 0;
        goto ResumeReduction;
    }

    act = state_stack[tos];

    forever {
        if (token == -1 && - TERMINAL_COUNT != action_index[act]) {
            uint cu = getChar();
            token = NOTOKEN;
            token_char = cu;
            if (cu & 0xff0000) {
                token = cu >> 16;
            } else switch (token_char) {
            case 0xfffe:
            case 0xffff:
                token = ERROR;
                break;
            case '\r':
                token = SPACE;
                if (cu == '\r') {
                    if ((token_char = filterCarriageReturn())) {
                        ++lineNumber;
                        lastLineStart = characterOffset + readBufferPos;
                        break;
                    }
                } else {
                    break;
                }
                // fall through
            case '\0': {
                token = EOF_SYMBOL;
                if (!tagsDone && !inParseEntity) {
                    int a = t_action(act, token);
                    if (a < 0) {
                        raiseError(QXmlStreamReader::PrematureEndOfDocumentError);
                        return false;
                    }
                }

            } break;
            case '\n':
                ++lineNumber;
                lastLineStart = characterOffset + readBufferPos;
            case ' ':
            case '\t':
                token = SPACE;
                break;
            case '&':
                token = AMPERSAND;
                break;
            case '#':
                token = HASH;
                break;
            case '\'':
                token = QUOTE;
                break;
            case '\"':
                token = DBLQUOTE;
                break;
            case '<':
                token = LANGLE;
                break;
            case '>':
                token = RANGLE;
                break;
            case '[':
                token = LBRACK;
                break;
            case ']':
                token = RBRACK;
                break;
            case '(':
                token = LPAREN;
                break;
            case ')':
                token = RPAREN;
                break;
            case '|':
                token = PIPE;
                break;
            case '=':
                token = EQ;
                break;
            case '%':
                token = PERCENT;
                break;
            case '/':
                token = SLASH;
                break;
            case ':':
                token = COLON;
                break;
            case ';':
                token = SEMICOLON;
                break;
            case ',':
                token = COMMA;
                break;
            case '-':
                token = DASH;
                break;
            case '+':
                token = PLUS;
                break;
            case '*':
                token = STAR;
                break;
            case '.':
                token = DOT;
                break;
            case '?':
                token = QUESTIONMARK;
                break;
            case '!':
                token = BANG;
                break;
            case '0':
            case '1':
            case '2':
            case '3':
            case '4':
            case '5':
            case '6':
            case '7':
            case '8':
            case '9':
                token = DIGIT;
                break;
            default:
                if (cu < 0x20)
                    token = NOTOKEN;
                else
                    token = LETTER;
                break;
            }
        }

        act = t_action (act, token);
        if (act == ACCEPT_STATE) {
            // reset the parser in case someone resumes (process instructions can follow a valid document)
            tos = 0;
            state_stack[tos++] = 0;
            state_stack[tos] = 0;
            return true;
        } else if (act > 0) {
            if (++tos == stack_size)
                reallocateStack();

            Value &val = sym_stack[tos];
            val.c = token_char;
            val.pos = textBuffer.size();
            val.prefix = 0;
            val.len = 1;
            if (token_char)
                textBuffer.inline_append(token_char);

            state_stack[tos] = act;
            token = -1;


        } else if (act < 0) {
            r = - act - 1;

#if defined (QLALR_DEBUG)
            int ridx = rule_index[r];
            printf ("%3d) %s ::=", r + 1, spell[rule_info[ridx]]);
            ++ridx;
            for (int i = ridx; i < ridx + rhs[r]; ++i) {
                int symbol = rule_info[i];
                if (const char *name = spell[symbol])
                    printf (" %s", name);
                else
                    printf (" #%d", symbol);
            }
            printf ("\n");
#endif

            tos -= rhs[r];
            act = state_stack[tos++];
        ResumeReduction:
            switch (r) {
./

document ::= PARSE_ENTITY content;
/.
        case $rule_number:
            setType(QXmlStreamReader::EndDocument);
        break;
./

document ::= prolog;
/.
        case $rule_number:
            if (type != QXmlStreamReader::Invalid) {
                if (hasSeenTag || inParseEntity) {
                    setType(QXmlStreamReader::EndDocument);
                } else {
                    raiseError(QXmlStreamReader::PrematureEndOfDocumentError, QXmlStream::tr("Start tag expected."));
                    // reset the parser
                    tos = 0;
                    state_stack[tos++] = 0;
                    state_stack[tos] = 0;
                    return false;
                }
            }
        break;
./


prolog ::= prolog stag content etag;
prolog ::= prolog empty_element_tag;
prolog ::= prolog comment;
prolog ::= prolog xml_decl;
prolog ::= prolog processing_instruction;
prolog ::= prolog doctype_decl;
prolog ::= prolog SPACE;
prolog ::=;

entity_done ::= ENTITY_DONE;
/.
        case $rule_number:
            entityReferenceStack.pop()->isCurrentlyReferenced = false;
            clearSym();
        break;
./


xml_decl_start ::= XML;
/.
        case $rule_number:
            if (!scanString(spell[VERSION], VERSION, false) && atEnd) {
                resume($rule_number);
                return false;
            }
        break;
./

xml_decl ::= xml_decl_start VERSION space_opt EQ space_opt literal attribute_list_opt QUESTIONMARK RANGLE;
/.
        case $rule_number:
            setType(QXmlStreamReader::StartDocument);
            startDocument(symString(6));
        break;
./

external_id ::= SYSTEM literal;
/.
        case $rule_number:
            hasExternalDtdSubset = true;
        break;
./
external_id ::= PUBLIC public_literal space literal;
/.
        case $rule_number:
            checkPublicLiteral(symString(2));
            hasExternalDtdSubset = true;
        break;
./
external_id ::=;

doctype_decl_start ::= langle_bang DOCTYPE name space;
/.
        case $rule_number:
            if (!scanPublicOrSystem() && atEnd) {
                resume($rule_number);
                return false;
            }
        break;
./

doctype_decl ::= doctype_decl_start external_id space_opt markup space_opt RANGLE;
/.
        case $rule_number:./
doctype_decl ::= doctype_decl_start external_id space_opt RANGLE;
/.
        case $rule_number:./
doctype_decl ::= langle_bang DOCTYPE name RANGLE;
/.
        case $rule_number:./
doctype_decl ::= langle_bang DOCTYPE name markup space_opt RANGLE;
/.
        case $rule_number:
            setType(QXmlStreamReader::DTD);
            text = &textBuffer;
        break;
./

markup_start ::= LBRACK;
/.
        case $rule_number:
            scanDtd = true;
        break;
./

markup ::= markup_start markup_list RBRACK;
/.
        case $rule_number:
            scanDtd = false;
        break;
./


markup_list ::= markup_decl | space | pereference;
markup_list ::= markup_list markup_decl | markup_list space | markup_list pereference;

markup_decl ::= element_decl | attlist_decl | entity_decl | entity_done | notation_decl | processing_instruction | comment;


element_decl_start ::= langle_bang ELEMENT name space;
/.
        case $rule_number:
            if (!scanString(spell[EMPTY], EMPTY, false)
                && !scanString(spell[ANY], ANY, false)
                && atEnd) {