0,0 → 1,679 |
/* |
* input 2072160977 bytes; AMD Athlon 3100+, disk speed 38.0 MB/s |
* 137.05s user 10.79s system 89% cpu 2:45.85 total; RES 9420K (empty java - 8200K) |
*/ |
import java.io.*; |
|
public class XmlParser |
{ |
public static void main(String[] args) |
throws Exception |
{ |
if(args.length < 1) { |
System.err.println("Need file name as parameter"); |
return; |
} |
|
long startTime = System.nanoTime(); |
|
FileInputStream in = new FileInputStream(args[0]); |
|
try { |
XmlDocument doc = new XmlDocument(); |
|
doc.parse(in); |
} |
catch(XmlException ex) { |
System.out.println(ex); |
} |
finally { |
in.close(); |
} |
|
System.err.println("Elapsed: " + (System.nanoTime() - startTime) / 1000000 + "ms"); |
} |
} |
|
// -------------------------------------------------------------------------------------------------------------------- |
class XmlException |
extends Exception |
{ |
static final long serialVersionUID = 1; |
|
private long line; |
private long linePos; |
|
public XmlException(String message, long line, long linePos) |
{ |
super(message); |
|
this.line = line; |
this.linePos = linePos; |
} |
|
public long getLine() { return line; } |
|
public long getLinePos() { return linePos; } |
|
public String toString() |
{ |
return "Error: " + getMessage() + " at " + line + ":" + linePos; |
} |
} |
|
// -------------------------------------------------------------------------------------------------------------------- |
class XmlSelection |
{ |
public long begin; |
public long end; |
|
public long getLength() { return end - begin; } |
} |
|
// -------------------------------------------------------------------------------------------------------------------- |
class XmlElement |
{ |
public boolean isEmpty = false; |
} |
|
// -------------------------------------------------------------------------------------------------------------------- |
class XmlInputStream |
extends InputStream |
{ |
private InputStream origin; |
private long pos = -1; |
private long marked = 0; |
private int markedRead = -1; |
private int lastRead = -1; |
private boolean useLast = false; |
|
public XmlInputStream(InputStream origin) |
{ |
this.origin = new BufferedInputStream(origin, 1024); |
} |
|
public long getPosition() |
{ |
return pos; |
} |
|
public long getNextPosition() |
{ |
return (useLast ? pos : pos+1); |
} |
|
public boolean eof() |
{ |
return (pos > 0 && lastRead < 0); |
} |
|
public void back() |
{ |
useLast = true; |
} |
|
public int read() |
throws IOException |
{ |
if(useLast) { |
useLast = false; |
} |
else { |
lastRead = origin.read(); |
++pos; |
} |
|
return lastRead; |
} |
|
public int read(byte[] b, int off, int len) |
throws IOException |
{ |
throw new IOException("not supported"); |
} |
|
public long skip(long n) |
throws IOException |
{ |
long res = origin.skip(n); |
|
if(res > 0) { |
pos += res; |
} |
|
return res; |
} |
|
public int available() |
throws IOException |
{ |
return origin.available(); |
} |
|
public void close() |
throws IOException |
{ |
origin.close(); |
} |
|
public boolean markSupported() |
{ |
return origin.markSupported(); |
} |
|
public void mark() |
{ |
mark(1024); |
} |
|
public void mark(int readlimit) |
{ |
marked = pos; |
if(useLast) { |
markedRead = lastRead; |
} |
else { |
markedRead = -1; |
} |
origin.mark(readlimit); |
} |
|
public void reset() |
throws IOException |
{ |
origin.reset(); |
pos = marked; |
if(markedRead >= 0) { |
useLast = true; |
lastRead = markedRead; |
} |
else { |
useLast = false; |
} |
} |
} |
|
// -------------------------------------------------------------------------------------------------------------------- |
class XmlDocument |
{ |
private XmlInputStream in; |
private long line; |
private long linePos; |
|
public void parse(InputStream in) |
throws XmlException, IOException |
{ |
this.in = new XmlInputStream(in); |
this.line = 1; |
this.linePos = 0; |
|
if(parseProlog()) { |
parseElement(); |
|
while(parseMisc()); |
/* |
if(this.pos < this.len) throwException("trash at end of text"); |
*/ |
} |
} |
|
private void throwException(String message) |
throws XmlException, IOException |
{ |
throw new XmlException(message, line, /*linePos*/ in.getPosition()); |
} |
|
private void log(String message) |
{ |
System.out.println(message); |
} |
|
private String toString(XmlSelection sel) |
throws IOException |
{ |
//return new String(buf, (int)(sel.begin - bufOffset), (int)sel.getLength(), "UTF-8"); |
return (sel.begin + "," + (int)sel.getLength()); |
} |
|
private void saveSelBegin(XmlSelection sel) |
{ |
sel.begin = in.getNextPosition(); |
} |
|
private void saveSelEnd(XmlSelection sel) |
{ |
sel.end = in.getPosition(); |
} |
|
private boolean parseProlog() |
throws XmlException, IOException |
{ |
parseDecl(); |
while(parseMisc()); |
if(parseDoctype()) { |
while(parseMisc()); |
} |
|
return true; |
} |
|
private boolean skipSpaces() |
throws XmlException, IOException |
{ |
//log("skipSpaces begin " + in.getPosition()); |
boolean found = false; |
|
for(;;) { |
int c = in.read(); |
if(c < 0) { |
break; |
} |
else if(c == ' ' || c == '\t' || c == '\n' || c == '\r') { |
found = true; |
} |
else { |
in.back(); |
break; |
} |
} |
|
//log("skipSpaces " + found + " " + in.getPosition()); |
return found; |
} |
|
private boolean parseDecl() |
throws XmlException, IOException |
{ |
//log("parseDecl begin " + in.getPosition()); |
|
// begin |
if(!testChar('<') || !testChar('?') || !testChar('x') |
|| !testChar('m') || !testChar('l')) |
{ |
//log("parseDecl no 'xml' " + in.getPosition()); |
return false; |
} |
|
// attributes |
XmlSelection selName = new XmlSelection(); |
XmlSelection selValue = new XmlSelection(); |
for(;;) { |
if(!skipSpaces()) break; |
if(!parseAttribute(selName, selValue)) break; |
} |
|
// end |
if(!testChar('?') || !testChar('>')) { |
throwException("end of XML declaration expected"); |
} |
|
//log("parseDecl ok " + in.getPosition()); |
return true; |
} |
|
private boolean parseDoctype() |
throws XmlException, IOException |
{ |
// FIXME not implemented |
|
return true; |
} |
|
private boolean parseName(XmlSelection sel) |
throws XmlException, IOException |
{ |
//log("parseName begin " + in.getPosition()); |
saveSelBegin(sel); |
in.mark(); |
|
int c = in.read(); |
if(c < 0) { |
return false; |
} |
else if(('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z') || c == '_' || c == ':') { |
} |
else { |
//log("parseName not a name " + in.getPosition()); |
in.reset(); |
return false; |
} |
|
for(;;) { |
c = in.read(); |
if(c < 0) { |
throwException("unexpected EoF"); |
} |
else if(('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z') || ('0' <= c && c <= '9') |
|| c == '.' || c == '-' || c == '_' || c == ':') |
{ |
} |
else { |
in.back(); |
break; |
} |
} |
saveSelEnd(sel); |
|
//log("parseName ok " + in.getPosition()); |
return true; |
} |
|
private boolean testChar(char c) |
throws XmlException, IOException |
{ |
int cc = in.read(); |
|
if(cc < 0) { |
//log("testChar eof " + in.getPosition() + " [" + c + "]"); |
return false; |
} |
else if(cc != c) { |
//log("testChar false " + in.getPosition() + " [" + c + "] [" + (char)cc + "]"); |
in.back(); |
return false; |
} |
else { |
//log("testChar true " + in.getPosition() + " [" + c + "] [" + (char)cc + "]"); |
return true; |
} |
} |
|
private boolean parseAttribute(XmlSelection selName, XmlSelection selValue) |
throws XmlException, IOException |
{ |
//log("parseAttribute begin " + in.getPosition()); |
// name |
if(!parseName(selName)) { |
//log("parseAttribute no name " + in.getPosition()); |
return false; |
} |
|
// eq |
skipSpaces(); |
if(!testChar('=')) { |
throwException("equal sign expected"); |
} |
skipSpaces(); |
|
// FIXME allow 'Reference' here |
|
// value |
if(!testChar('"')) { |
throwException("quoted string expected"); |
} |
|
saveSelBegin(selValue); |
for(;;) { |
int c = in.read(); |
if(c < 0) { |
throwException("unexpected EoF"); |
} |
else if(c == '<' || c == '&' || c == '"') { |
in.back(); |
break; |
} |
} |
saveSelEnd(selValue); |
|
if(!testChar('"')) { |
throwException("end of quoted string expected"); |
} |
|
// FIXME check '[WFC: No External Entity References]' |
|
//log("parseAttribute ok " + in.getPosition()); |
return true; |
} |
|
private boolean parseMisc() |
throws XmlException, IOException |
{ |
//log("parseMisc begin " + in.getPosition()); |
if(parseComment()) { |
//log("parseMisc comment ok " + in.getPosition()); |
return true; |
} |
else if(parseProcessInstruction()) { |
//log("parseMisc pi ok " + in.getPosition()); |
return true; |
} |
else if(skipSpaces()) { |
//log("parseMisc spaces ok " + in.getPosition()); |
return true; |
} |
else { |
//log("parseMisc false " + in.getPosition()); |
return false; |
} |
} |
|
private boolean parseComment() |
throws XmlException, IOException |
{ |
//log("parseComment begin " + in.getPosition()); |
in.mark(); |
if(!testChar('<') || !testChar('!') || !testChar('-') || !testChar('-')) { |
in.reset(); |
//log("parseComment no signature " + in.getPosition()); |
return false; |
} |
|
for(; !in.eof(); in.read()) { |
if(testChar('-') && testChar('-')) { |
if(testChar('>')) { |
//log("parseComment ok " + in.getPosition()); |
return true; |
} |
else { |
throwException("Sequence '--' is not allowed in comment"); |
} |
} |
} |
|
throwException("unexpected EoF"); |
return false; |
} |
|
private boolean parseProcessInstruction() |
throws XmlException, IOException |
{ |
//log("parseProcessInstruction begin " + in.getPosition()); |
in.mark(); |
if(!testChar('<') || !testChar('?')) { |
in.reset(); |
//log("parseProcessInstruction no signature " + in.getPosition()); |
return false; |
} |
|
for(; !in.eof(); in.read()) { |
if(testChar('?') && testChar('>')) { |
//log("parseProcessInstruction ok " + in.getPosition()); |
return true; |
} |
} |
|
throwException("unexpected EoF"); |
return false; |
} |
|
private XmlElement parseElement() |
throws XmlException, IOException |
{ |
//log("parseElement begin " + in.getPosition()); |
|
XmlElement element = new XmlElement(); |
|
if(!parseStartTag(element)) { |
//log("parseElement no start tag " + in.getPosition()); |
return null; |
} |
|
if(!element.isEmpty) { |
if(!parseTagContent(element)) { |
throwException("cannot parse tag content"); |
} |
|
XmlSelection selEndName = new XmlSelection(); |
parseEndTag(selEndName); |
|
//if(element.name != selEndName |
// throwException("tag names do not match"); |
} |
|
//log("parseElement ok " + in.getPosition()); |
return element; |
} |
|
private boolean parseStartTag(XmlElement element) |
throws XmlException, IOException |
{ |
//log("parseStartTag begin " + in.getPosition()); |
in.mark(); |
|
// begin |
if(!testChar('<')) { |
//log("parseStartTag no signature " + in.getPosition()); |
return false; |
} |
|
// name |
XmlSelection sel = new XmlSelection(); |
if(!parseName(sel)) { |
//log("parseStartTag no name " + in.getPosition()); |
in.reset(); |
return false; |
} |
//System.out.print("[" + toString(sel) + "]"); |
|
// attributes |
XmlSelection selName = new XmlSelection(); |
XmlSelection selValue = new XmlSelection(); |
for(;;) { |
if(!skipSpaces()) break; |
if(!parseAttribute(selName, selValue)) break; |
//System.out.print(" [" + toString(selName) + "]=[" + toString(selValue) + "]"); |
} |
//System.out.println(); |
|
// end |
element.isEmpty = testChar('/'); |
if(!testChar('>')) { |
throwException("end of tag expected"); |
} |
|
//log("parseStartTag ok " + in.getPosition()); |
return true; |
} |
|
private void parseEndTag(XmlSelection sel) |
throws XmlException, IOException |
{ |
//log("parseEndTag begin " + in.getPosition()); |
// begin |
if(!testChar('<') || !testChar('/')) { |
throwException("cannot find tag end"); |
} |
|
// name |
if(!parseName(sel)) { |
throwException("tag name expected"); |
} |
|
// spaces |
skipSpaces(); |
|
// end |
if(!testChar('>')) { |
throwException("end of tag expected"); |
} |
|
//log("parseEndTag ok " + in.getPosition()); |
} |
|
private boolean parseTagContent(XmlElement element) |
throws XmlException, IOException |
{ |
//log("parseTagContent begin " + in.getPosition()); |
XmlSelection sel = new XmlSelection(); |
for(;;) { |
if(in.eof()) { |
throwException("unexpected EoF"); |
} |
in.mark(); |
if(testChar('<') && testChar('/')) { |
in.reset(); |
break; |
} |
in.reset(); |
|
if(parseElement() != null) { |
} |
else if(parseComment()) { |
} |
else if(parseCData(sel)) { |
} |
else if(parseProcessInstruction()) { |
} |
else if(parseCharData(sel)) { |
} |
|
// FIXME allow 'Reference' here |
} |
|
//log("parseTagContent ok " + in.getPosition()); |
return true; |
} |
|
// FIXME not fully conform the standard |
private boolean parseCharData(XmlSelection sel) |
throws XmlException, IOException |
{ |
//log("parseCharData begin " + in.getPosition()); |
boolean found = false; |
saveSelBegin(sel); |
|
for(;;) { |
int c = in.read(); |
if(c < 0) { |
return false; |
} |
else if(c == '<') { |
in.back(); |
break; |
} |
|
found = true; |
} |
|
//log("parseCharData " + found + " " + in.getPosition()); |
saveSelEnd(sel); |
return found; |
} |
|
private boolean parseCData(XmlSelection sel) |
throws XmlException, IOException |
{ |
in.mark(); |
if(!testChar('<') || !testChar('!') |
|| !testChar('[') || !testChar('C') |
|| !testChar('D') || !testChar('A') |
|| !testChar('T') || !testChar('A') |
|| !testChar('[')) |
{ |
in.reset(); |
return false; |
} |
|
saveSelBegin(sel); |
|
for(; !in.eof(); in.read()) { |
if(testChar(']') && testChar(']') && testChar('>')) { |
saveSelEnd(sel); |
return true; |
} |
} |
|
throwException("unexpected EoF"); |
return false; |
} |
} |