Subversion Repositories general

Compare Revisions

Ignore whitespace Rev 1352 → Rev 1353

/xmlparser_java/branches/001_buffer_move/XmlParser.java
19,9 → 19,11
FileInputStream in = new FileInputStream(args[0]);
try {
XmlDocument doc = new XmlDocument();
XmlListenerImpl listener = new XmlListenerImpl();
XmlDocument doc = new XmlDocument(listener);
doc.parse(in);
System.out.println(listener.count + " elements found");
}
catch(XmlException ex) {
System.out.println(ex);
30,11 → 32,47
in.close();
}
 
System.err.println("Elapsed: " + (System.nanoTime() - startTime) / 1000000 + "ms");
System.out.println("Elapsed: " + (System.nanoTime() - startTime) / 1000000 + "ms");
}
}
 
// --------------------------------------------------------------------------------------------------------------------
class XmlListenerImpl
implements XmlListener
{
public long count = 0;
private XmlDocument document;
public void init(XmlDocument document)
throws XmlException, IOException
{
this.document = document;
}
public void processElementBegin(XmlElement element)
throws XmlException, IOException
{
++count;
}
public void processElementEnd(XmlElement element)
throws XmlException, IOException
{
}
public void processCharData(XmlSelection sel)
throws XmlException, IOException
{
}
public void processCData(XmlSelection sel)
throws XmlException, IOException
{
}
}
 
// --------------------------------------------------------------------------------------------------------------------
class XmlException
extends Exception
{
71,22 → 109,48
}
 
// --------------------------------------------------------------------------------------------------------------------
interface XmlListener
{
public void init(XmlDocument document)
throws XmlException, IOException;
public void processElementBegin(XmlElement element)
throws XmlException, IOException;
public void processElementEnd(XmlElement element)
throws XmlException, IOException;
public void processCharData(XmlSelection sel)
throws XmlException, IOException;
public void processCData(XmlSelection sel)
throws XmlException, IOException;
}
 
// --------------------------------------------------------------------------------------------------------------------
class XmlElement
{
public boolean isEmpty = false;
public boolean isEmpty = false;
public XmlSelection elementSel;
public XmlSelection nameSel;
}
 
// --------------------------------------------------------------------------------------------------------------------
class XmlDocument
{
private XmlListener listener;
private InputStream in;
private byte[] buf = new byte[2048]; // (buf.length % 16 == 0)
private int bufLen;
private int bufPos;
private long bufOffset;
private long line;
private long linePos;
private int level;
 
public XmlDocument(XmlListener listener)
{
this.listener = listener;
}
public void parse(InputStream in)
throws XmlException, IOException
{
98,16 → 162,16
this.bufLen = 0;
this.bufPos = 0;
this.bufOffset = 0;
this.line = 1;
this.linePos = 0;
this.level = 0;
 
if(listener != null) listener.init(this);
if(parseProlog()) {
parseElement();
while(parseMisc());
/*
if(this.pos < this.len) throwException("trash at end of text");
*/
 
if(bufPos < bufLen) throwException("EoF expected");
}
}
 
114,7 → 178,7
private void throwException(String message)
throws XmlException, IOException
{
throw new XmlException(message, line, /*linePos*/ bufPos);
throw new XmlException(message, 1, bufPos);
}
 
private void log(String message)
122,12 → 186,17
System.out.println(message);
}
 
private String toString(XmlSelection sel)
public String selectionToString(XmlSelection sel)
throws IOException
{
return new String(buf, (int)(sel.begin - bufOffset), (int)sel.getLength(), "UTF-8");
}
public int getLevel()
{
return level;
}
private void saveSelBegin(XmlSelection sel)
{
sel.begin = bufOffset + bufPos;
142,7 → 211,6
throws XmlException, IOException
{
if(bufPos + count >= bufLen) {
//log("ensureNext start " + bufPos + " " + count);
if(bufLen == 0) {
// read full buffer at begin
bufLen = in.read(buf);
183,7 → 251,6
private boolean skipSpaces()
throws XmlException, IOException
{
//log("skipSpaces begin " + bufPos);
boolean found = false;
for(;;) {
199,7 → 266,6
}
}
//log("skipSpaces " + found + " " + bufPos);
return found;
}
 
206,13 → 272,10
private boolean parseDecl()
throws XmlException, IOException
{
//log("parseDecl begin " + bufPos);
// begin
if(!testChar('<') || !testChar('?') || !testChar('x')
|| !testChar('m') || !testChar('l'))
{
//log("parseDecl no 'xml' " + bufPos);
return false;
}
229,7 → 292,6
throwException("end of XML declaration expected");
}
//log("parseDecl ok " + bufPos);
return true;
}
 
244,7 → 306,6
private boolean parseName(XmlSelection sel)
throws XmlException, IOException
{
//log("parseName begin " + bufPos);
saveSelBegin(sel);
int start = bufPos;
257,7 → 318,6
++bufPos;
}
else {
//log("parseName not a name " + bufPos);
bufPos = start;
return false;
}
279,7 → 339,6
}
saveSelEnd(sel);
//log("parseName ok " + bufPos);
return true;
}
 
302,10 → 361,8
private boolean parseAttribute(XmlSelection selName, XmlSelection selValue)
throws XmlException, IOException
{
//log("parseAttribute begin " + bufPos);
// name
if(!parseName(selName)) {
//log("parseAttribute no name " + bufPos);
return false;
}
342,7 → 399,6
// FIXME check '[WFC: No External Entity References]'
//log("parseAttribute ok " + bufPos);
return true;
}
 
349,21 → 405,16
private boolean parseMisc()
throws XmlException, IOException
{
//log("parseMisc begin " + bufPos);
if(parseComment()) {
//log("parseMisc comment ok " + bufPos);
return true;
}
else if(parseProcessInstruction()) {
//log("parseMisc pi ok " + bufPos);
return true;
}
else if(skipSpaces()) {
//log("parseMisc spaces ok " + bufPos);
return true;
}
else {
//log("parseMisc false " + bufPos);
return false;
}
}
371,13 → 422,10
private boolean parseComment()
throws XmlException, IOException
{
//log("parseComment begin " + bufPos);
if(bufPos+3 >= bufLen && !ensureNext(4)) {
//log("parseComment no data " + bufPos);
return false;
}
if(buf[bufPos] != '<' || buf[bufPos+1] != '!' || buf[bufPos+2] != '-' || buf[bufPos+3] != '-') {
//log("parseComment no signature " + bufPos);
return false;
}
390,7 → 438,7
if(buf[bufPos] == '-' && buf[bufPos+1] == '-') {
if(buf[bufPos+2] == '>') {
bufPos += 3;
//log("parseComment ok " + bufPos);
return true;
}
else {
418,6 → 466,7
 
if(buf[bufPos] == '?' && buf[bufPos+1] == '>') {
bufPos += 2;
return true;
}
}
426,15 → 475,18
private XmlElement parseElement()
throws XmlException, IOException
{
//log("parseElement begin " + bufPos);
 
XmlElement element = new XmlElement();
element.elementSel = new XmlSelection();
saveSelBegin(element.elementSel);
if(!parseStartTag(element)) {
//log("parseElement no start tag " + bufPos);
return null;
}
 
++level;
if(listener != null) listener.processElementBegin(element);
if(!element.isEmpty) {
if(!parseTagContent(element)) {
throwException("cannot parse tag content");
443,11 → 495,16
XmlSelection selEndName = new XmlSelection();
parseEndTag(selEndName);
//if(element.name != selEndName
//if(!selectionToString(element.nameSel).equals(selectionToString(selEndName))) {
// throwException("tag names do not match");
//}
}
//log("parseElement ok " + bufPos);
saveSelEnd(element.elementSel);
if(listener != null) listener.processElementEnd(element);
--level;
return element;
}
 
454,23 → 511,20
private boolean parseStartTag(XmlElement element)
throws XmlException, IOException
{
//log("parseStartTag begin " + bufPos);
int start = bufPos;
 
// begin
if(!testChar('<')) {
//log("parseStartTag no signature " + bufPos);
return false;
}
// name
XmlSelection sel = new XmlSelection();
if(!parseName(sel)) {
//log("parseStartTag no name " + bufPos);
element.nameSel = new XmlSelection();
if(!parseName(element.nameSel)) {
bufPos = start;
return false;
}
//System.out.print("[" + toString(sel) + "]");
// attributes
XmlSelection selName = new XmlSelection();
478,9 → 532,7
for(;;) {
if(!skipSpaces()) break;
if(!parseAttribute(selName, selValue)) break;
//System.out.print(" [" + toString(selName) + "]=[" + toString(selValue) + "]");
}
//System.out.println();
// end
element.isEmpty = testChar('/');
488,7 → 540,6
throwException("end of tag expected");
}
//log("parseStartTag ok " + bufPos);
return true;
}
 
495,7 → 546,6
private void parseEndTag(XmlSelection sel)
throws XmlException, IOException
{
//log("parseEndTag begin " + bufPos);
// begin
if(!testChar('<') || !testChar('/')) {
throwException("cannot find tag end");
513,14 → 563,11
if(!testChar('>')) {
throwException("end of tag expected");
}
//log("parseEndTag ok " + bufPos);
}
 
private boolean parseTagContent(XmlElement element)
throws XmlException, IOException
{
//log("parseTagContent begin " + bufPos);
XmlSelection sel = new XmlSelection();
for(;;) {
if(bufPos+1 >= bufLen && !ensureNext(2)) {
538,29 → 585,31
}
else if(parseCharData(sel)) {
}
else {
throwException("unexpected tag content");
}
// FIXME allow 'Reference' here
}
//log("parseTagContent ok " + bufPos);
return true;
}
 
// FIXME not fully conform the standard
private boolean parseCharData(XmlSelection sel)
throws XmlException, IOException
{
//log("parseCharData begin " + bufPos);
// FIXME allow 'Reference' here
boolean found = false;
saveSelBegin(sel);
for(;; ++bufPos) {
if(bufPos >= bufLen && !ensureNext(1)) {
return true;
throwException("unexpected EoF");
}
byte c = buf[bufPos];
if(c == '<') {
if(c == '<' || c == '&') {
break;
}
567,8 → 616,9
found = true;
}
//log("parseCharData " + found + " " + bufPos);
saveSelEnd(sel);
if(listener != null) listener.processCharData(sel);
 
return found;
}
 
598,6 → 648,8
if(buf[bufPos] == ']' && buf[bufPos+1] == ']' && buf[bufPos+2] == '>') {
saveSelEnd(sel);
bufPos += 3;
if(listener != null) listener.processCData(sel);
 
return true;
}
}