xml11entityscanner.java

来自「JAVA 所有包」· Java 代码 · 共 1,525 行 · 第 1/5 页

        }

        // scan qualified name
        int offset = fCurrentEntity.position;
        char ch = fCurrentEntity.ch[offset];
        
        if (XML11Char.isXML11NCNameStart(ch)) {
            if (++fCurrentEntity.position == fCurrentEntity.count) {
                fCurrentEntity.ch[0] = ch;
                offset = 0;
                if (load(1, false)) {
                    fCurrentEntity.columnNumber++;
                    String name = fSymbolTable.addSymbol(fCurrentEntity.ch, 0, 1);
                    qname.setValues(null, name, name, null);
                    return true;
                }
            }
        }
        else if (XML11Char.isXML11NameHighSurrogate(ch)) {
            if (++fCurrentEntity.position == fCurrentEntity.count) {
                fCurrentEntity.ch[0] = ch;
                offset = 0;
                if (load(1, false)) {
                    --fCurrentEntity.startPosition;
                    --fCurrentEntity.position;
                    return false;
                }
            }
            char ch2 = fCurrentEntity.ch[fCurrentEntity.position];
            if ( !XMLChar.isLowSurrogate(ch2) ||
                 !XML11Char.isXML11NCNameStart(XMLChar.supplemental(ch, ch2)) ) {
                --fCurrentEntity.position;
                return false;
            }
            if (++fCurrentEntity.position == fCurrentEntity.count) {
                fCurrentEntity.ch[0] = ch;
                fCurrentEntity.ch[1] = ch2;
                offset = 0;
                if (load(2, false)) {
                    fCurrentEntity.columnNumber += 2;
                    String name = fSymbolTable.addSymbol(fCurrentEntity.ch, 0, 2);
                    qname.setValues(null, name, name, null);
                    return true;
                }
            }
        }
        else {
            return false;
        }
        
        int index = -1;
        boolean sawIncompleteSurrogatePair = false;
        do {
            ch = fCurrentEntity.ch[fCurrentEntity.position];
            if (XML11Char.isXML11Name(ch)) {
                if (ch == ':') {
                    if (index != -1) {
                        break;
                    }
                    index = fCurrentEntity.position;
                }
                if (++fCurrentEntity.position == fCurrentEntity.count) {
                    int length = fCurrentEntity.position - offset;
                    if (length == fCurrentEntity.ch.length) {
                        // bad luck we have to resize our buffer
                        char[] tmp = new char[fCurrentEntity.ch.length << 1];
                        System.arraycopy(fCurrentEntity.ch, offset,
                                         tmp, 0, length);
                        fCurrentEntity.ch = tmp;
                    }
                    else {
                        System.arraycopy(fCurrentEntity.ch, offset,
                                         fCurrentEntity.ch, 0, length);
                    }
                    if (index != -1) {
                        index = index - offset;
                    }
                    offset = 0;
                    if (load(length, false)) {
                        break;
                    }
                }
            }
            else if (XML11Char.isXML11NameHighSurrogate(ch)) {
                if (++fCurrentEntity.position == fCurrentEntity.count) {
                    int length = fCurrentEntity.position - offset;
                    if (length == fCurrentEntity.ch.length) {
                        // bad luck we have to resize our buffer
                        char[] tmp = new char[fCurrentEntity.ch.length << 1];
                        System.arraycopy(fCurrentEntity.ch, offset,
                                         tmp, 0, length);
                        fCurrentEntity.ch = tmp;
                    }
                    else {
                        System.arraycopy(fCurrentEntity.ch, offset,
                                         fCurrentEntity.ch, 0, length);
                    }
                    if (index != -1) {
                        index = index - offset;
                    }
                    offset = 0;
                    if (load(length, false)) {
                        sawIncompleteSurrogatePair = true;
                        --fCurrentEntity.startPosition;
                        --fCurrentEntity.position;
                        break;
                    }
                }
                char ch2 = fCurrentEntity.ch[fCurrentEntity.position];
                if ( !XMLChar.isLowSurrogate(ch2) ||
                     !XML11Char.isXML11Name(XMLChar.supplemental(ch, ch2)) ) {
                    sawIncompleteSurrogatePair = true;
                    --fCurrentEntity.position;
                    break;
                }
                if (++fCurrentEntity.position == fCurrentEntity.count) {
                    int length = fCurrentEntity.position - offset;
                    if (length == fCurrentEntity.ch.length) {
                        // bad luck we have to resize our buffer
                        char[] tmp = new char[fCurrentEntity.ch.length << 1];
                        System.arraycopy(fCurrentEntity.ch, offset,
                                         tmp, 0, length);
                        fCurrentEntity.ch = tmp;
                    }
                    else {
                        System.arraycopy(fCurrentEntity.ch, offset,
                                         fCurrentEntity.ch, 0, length);
                    }
                    if (index != -1) {
                        index = index - offset;
                    }
                    offset = 0;
                    if (load(length, false)) {
                        break;
                    }
                }
            }
            else {
                break;
            }
        }
        while (true);
        
        int length = fCurrentEntity.position - offset;
        fCurrentEntity.columnNumber += length;
        
        if (length > 0) {
            String prefix = null;
            String localpart = null;
            String rawname = fSymbolTable.addSymbol(fCurrentEntity.ch,
                                                    offset, length);
            if (index != -1) {
                int prefixLength = index - offset;
                prefix = fSymbolTable.addSymbol(fCurrentEntity.ch,
                                                    offset, prefixLength);
                int len = length - prefixLength - 1;
                int startLocal = index +1;
                if (!XML11Char.isXML11NCNameStart(fCurrentEntity.ch[startLocal]) &&
                    (!XML11Char.isXML11NameHighSurrogate(fCurrentEntity.ch[startLocal]) ||
                    sawIncompleteSurrogatePair)){
                    fErrorReporter.reportError(XMLMessageFormatter.XML_DOMAIN,
                                               "IllegalQName",
                                               null,
                                               XMLErrorReporter.SEVERITY_FATAL_ERROR);
                }
                localpart = fSymbolTable.addSymbol(fCurrentEntity.ch,
                                                   index + 1, len);

            }
            else {
                localpart = rawname;
            }
            qname.setValues(prefix, localpart, rawname, null);
            return true;
        }
        return false;

    } // scanQName(QName):boolean

    /**
     * Scans a range of parsed character data, setting the fields of the
     * XMLString structure, appropriately.
     * <p>
     * <strong>Note:</strong> The characters are consumed.
     * <p>
     * <strong>Note:</strong> This method does not guarantee to return
     * the longest run of parsed character data. This method may return
     * before markup due to reaching the end of the input buffer or any
     * other reason.
     * <p>
     * <strong>Note:</strong> The fields contained in the XMLString
     * structure are not guaranteed to remain valid upon subsequent calls
     * to the entity scanner. Therefore, the caller is responsible for
     * immediately using the returned character data or making a copy of
     * the character data.
     *
     * @param content The content structure to fill.
     *
     * @return Returns the next character on the input, if known. This
     *         value may be -1 but this does <em>note</em> designate
     *         end of file.
     *
     * @throws IOException  Thrown if i/o error occurs.
     * @throws EOFException Thrown on end of file.
     */
    public int scanContent(XMLString content) throws IOException {

        // load more characters, if needed
        if (fCurrentEntity.position == fCurrentEntity.count) {
            load(0, true);
        }
        else if (fCurrentEntity.position == fCurrentEntity.count - 1) {
            fCurrentEntity.ch[0] = fCurrentEntity.ch[fCurrentEntity.count - 1];
            load(1, false);
            fCurrentEntity.position = 0;
            fCurrentEntity.startPosition = 0;
        }

        // normalize newlines
        int offset = fCurrentEntity.position;
        int c = fCurrentEntity.ch[offset];
        int newlines = 0;
        boolean external = fCurrentEntity.isExternal();
        if (c == '\n' || ((c == '\r' || c == 0x85 || c == 0x2028) && external)) {
            do {
                c = fCurrentEntity.ch[fCurrentEntity.position++];
                if ((c == '\r' ) && external) {
                    newlines++;
                    fCurrentEntity.lineNumber++;
                    fCurrentEntity.columnNumber = 1;
                    if (fCurrentEntity.position == fCurrentEntity.count) {
                        offset = 0;
                        fCurrentEntity.baseCharOffset += (fCurrentEntity.position - fCurrentEntity.startPosition);
                        fCurrentEntity.position = newlines;
                        fCurrentEntity.startPosition = newlines;
                        if (load(newlines, false)) {
                            break;
                        }
                    }
                    int cc = fCurrentEntity.ch[fCurrentEntity.position]; 
                    if (cc == '\n' || cc == 0x85) {
                        fCurrentEntity.position++;
                        offset++;
                    }
                    /*** NEWLINE NORMALIZATION ***/
                    else {
                        newlines++;
                    }
                }
                else if (c == '\n' || ((c == 0x85 || c == 0x2028) && external)) {
                    newlines++;
                    fCurrentEntity.lineNumber++;
                    fCurrentEntity.columnNumber = 1;
                    if (fCurrentEntity.position == fCurrentEntity.count) {
                        offset = 0;
                        fCurrentEntity.baseCharOffset += (fCurrentEntity.position - fCurrentEntity.startPosition);
                        fCurrentEntity.position = newlines;
                        fCurrentEntity.startPosition = newlines;
                        if (load(newlines, false)) {
                            break;
                        }
                    }
                }
                else {
                    fCurrentEntity.position--;
                    break;
                }
            } while (fCurrentEntity.position < fCurrentEntity.count - 1);
            for (int i = offset; i < fCurrentEntity.position; i++) {
                fCurrentEntity.ch[i] = '\n';
            }
            int length = fCurrentEntity.position - offset;
            if (fCurrentEntity.position == fCurrentEntity.count - 1) {
                content.setValues(fCurrentEntity.ch, offset, length);
                return -1;
            }
        }

        // inner loop, scanning for content
        if (external) {
            while (fCurrentEntity.position < fCurrentEntity.count) {
                c = fCurrentEntity.ch[fCurrentEntity.position++];
                if (!XML11Char.isXML11Content(c) || c == 0x85 || c == 0x2028) {
                    fCurrentEntity.position--;
                    break;
                }
            }
        }
        else {
            while (fCurrentEntity.position < fCurrentEntity.count) {
                c = fCurrentEntity.ch[fCurrentEntity.position++];
                // In internal entities control characters are allowed to appear unescaped.
                if (!XML11Char.isXML11InternalEntityContent(c)) {
                    fCurrentEntity.position--;
                    break;
                }
            }
        }
        int length = fCurrentEntity.position - offset;
        fCurrentEntity.columnNumber += length - newlines;
        content.setValues(fCurrentEntity.ch, offset, length);

        // return next character
        if (fCurrentEntity.position != fCurrentEntity.count) {
            c = fCurrentEntity.ch[fCurrentEntity.position];