1,6 → 1,6 |
/* |
* input 2072160977 bytes; AMD Athlon 3100+, disk speed 38.0 MB/s |
* 49.18s user 10.07s system 89% cpu 1:06.32 total; RES 9384K (empty java - 8200K) |
* 59.18s user 9.77s system 92% cpu 1:14.72 total; RES 9420K (empty java - 8200K) |
*/ |
import java.io.*; |
|
77,13 → 77,139 |
} |
|
// -------------------------------------------------------------------------------------------------------------------- |
class XmlDocument |
class XmlBuffer |
{ |
private InputStream in; |
private byte[] buf = new byte[2048]; // (buf.length % 16 == 0) |
private int bufLen; |
private int bufPos; |
private long bufOffset; |
private int len; |
private int pos; |
private long offset; |
private int marked = -1; |
|
public XmlBuffer(InputStream in) |
{ |
if(buf.length % 16 != 0) { |
throw new RuntimeException("wrong buffer size: " + buf.length); |
} |
|
this.in = in; |
this.len = 0; |
this.pos = 0; |
this.offset = 0; |
} |
|
public byte cur() |
throws XmlException, IOException |
{ |
if(pos >= len) { |
if(!ensureNext(1)) return 0; |
} |
|
return buf[pos]; |
} |
|
public byte at(int n) |
throws XmlException, IOException |
{ |
if(pos + n >= len) { |
if(!ensureNext(n+1)) return 0; |
} |
|
return buf[pos + n]; |
} |
|
public void toNext() |
{ |
++pos; |
} |
|
public boolean isEnd() |
throws XmlException, IOException |
{ |
return (pos >= len && !ensureNext(1)); |
} |
|
public void skip(int n) |
{ |
pos += n; |
} |
|
public long getOffset() |
{ |
return offset; |
} |
|
public int getPosition() |
{ |
return pos; |
} |
|
public long getAbsPosition() |
{ |
return (offset + pos); |
} |
|
public void mark() |
{ |
marked = pos; |
} |
|
public void reset() |
{ |
if(marked < 0) { |
throw new RuntimeException("no position saved"); |
} |
else { |
pos = marked; |
marked = -1; |
} |
} |
|
public boolean ensureNext(int count) |
throws IOException |
{ |
if(pos + count >= len) { |
//log("ensureNext start " + pos + " " + count); |
if(len == 0) { |
// read full buffer at begin |
len = in.read(buf); |
pos = 0; |
} |
else if(len < buf.length) { |
// we could not fill full buffer last time - no more data |
return false; |
} |
else { |
// move last 1/16 of data to begin, fill rest with new data |
System.arraycopy(buf, buf.length / 16 * 15, buf, 0, buf.length / 16); |
int read = in.read(buf, buf.length / 16, buf.length / 16 * 15); |
len = buf.length / 16 + read; |
pos -= buf.length / 16 * 15; |
if(marked >= 0) { |
marked -= buf.length / 16 * 15; |
if(marked < 0) marked = -1; |
} |
offset += buf.length / 16 * 15; |
} |
|
return (pos + count < len); |
} |
else { |
return true; |
} |
} |
|
public String toString(long begin, int length) |
throws IOException |
{ |
return new String(buf, (int)(begin - offset), length, "UTF-8"); |
} |
} |
|
// -------------------------------------------------------------------------------------------------------------------- |
class XmlDocument |
{ |
private XmlBuffer buf; |
private long line; |
private long linePos; |
|
90,14 → 216,7 |
public void parse(InputStream in) |
throws XmlException, IOException |
{ |
if(buf.length % 16 != 0) { |
throwException("wrong buffer size: " + buf.length); |
} |
|
this.in = in; |
this.bufLen = 0; |
this.bufPos = 0; |
this.bufOffset = 0; |
this.buf = new XmlBuffer(in); |
this.line = 1; |
this.linePos = 0; |
|
114,7 → 233,7 |
private void throwException(String message) |
throws XmlException, IOException |
{ |
throw new XmlException(message, line, /*linePos*/ bufPos); |
throw new XmlException(message, line, /*linePos*/ buf.getAbsPosition()); |
} |
|
private void log(String message) |
125,49 → 244,19 |
private String toString(XmlSelection sel) |
throws IOException |
{ |
return new String(buf, (int)(sel.begin - bufOffset), (int)sel.getLength(), "UTF-8"); |
return buf.toString(sel.begin, (int)sel.getLength()); |
} |
|
private void saveSelBegin(XmlSelection sel) |
{ |
sel.begin = bufOffset + bufPos; |
sel.begin = buf.getAbsPosition(); |
} |
|
private void saveSelEnd(XmlSelection sel) |
{ |
sel.end = bufOffset + bufPos; |
sel.end = buf.getAbsPosition(); |
} |
|
private boolean ensureNext(int count) |
throws XmlException, IOException |
{ |
if(bufPos + count >= bufLen) { |
//log("ensureNext start " + bufPos + " " + count); |
if(bufLen == 0) { |
// read full buffer at begin |
bufLen = in.read(buf); |
bufPos = 0; |
} |
else if(bufLen < buf.length) { |
// we could not fill full buffer last time - no more data |
return false; |
} |
else { |
// move last 1/16 of data to begin, fill rest with new data |
System.arraycopy(buf, buf.length / 16 * 15, buf, 0, buf.length / 16); |
int read = in.read(buf, buf.length / 16, buf.length / 16 * 15); |
bufLen = buf.length / 16 + read; |
bufPos -= buf.length / 16 * 15; |
bufOffset += buf.length / 16 * 15; |
} |
|
return (bufPos + count < bufLen); |
} |
else { |
return true; |
} |
} |
|
private boolean parseProlog() |
throws XmlException, IOException |
{ |
187,12 → 276,13 |
boolean found = false; |
|
for(;;) { |
if(bufPos >= bufLen && !ensureNext(1)) break; |
|
byte c = buf[bufPos]; |
if(c == ' ' || c == '\t' || c == '\n' || c == '\r') { |
byte c = buf.cur(); |
if(c == 0) { |
break; |
} |
else if(c == ' ' || c == '\t' || c == '\n' || c == '\r') { |
found = true; |
++bufPos; |
buf.toNext(); |
} |
else { |
break; |
209,8 → 299,8 |
//log("parseDecl begin " + bufPos); |
|
// begin |
if(!testChar('<', bufPos) || !testChar('?', bufPos) || !testChar('x', bufPos) |
|| !testChar('m', bufPos) || !testChar('l', bufPos)) |
if(!testChar('<') || !testChar('?') || !testChar('x') |
|| !testChar('m') || !testChar('l')) |
{ |
//log("parseDecl no 'xml' " + bufPos); |
return false; |
225,7 → 315,7 |
} |
|
// end |
if(!testChar('?', bufPos) || !testChar('>', bufPos)) { |
if(!testChar('?') || !testChar('>')) { |
throwException("end of XML declaration expected"); |
} |
|
246,32 → 336,26 |
{ |
//log("parseName begin " + bufPos); |
saveSelBegin(sel); |
int start = bufPos; |
|
if(bufPos >= bufLen && !ensureNext(1)) { |
return false; |
} |
|
byte c = buf[bufPos]; |
|
byte c = buf.cur(); |
if(('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z') || c == '_' || c == ':') { |
++bufPos; |
buf.toNext(); |
} |
else { |
//log("parseName not a name " + bufPos); |
bufPos = start; |
return false; |
} |
|
for(;;) { |
if(bufPos >= bufLen && !ensureNext(1)) { |
if(buf.isEnd()) { |
throwException("unexpected EoF"); |
} |
|
c = buf[bufPos]; |
c = buf.cur(); |
if(('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z') || ('0' <= c && c <= '9') |
|| c == '.' || c == '-' || c == '_' || c == ':') |
{ |
++bufPos; |
buf.toNext(); |
} |
else { |
break; |
283,19 → 367,14 |
return true; |
} |
|
private boolean testChar(char c, int rollback) |
private boolean testChar(char c) |
throws XmlException, IOException |
{ |
if(bufPos >= bufLen && !ensureNext(1)) { |
if(buf.cur() != c) { |
return false; |
} |
|
if(buf[bufPos] != c) { |
bufPos = rollback; |
return false; |
} |
else { |
++bufPos; |
buf.toNext(); |
return true; |
} |
} |
305,14 → 384,16 |
{ |
//log("parseAttribute begin " + bufPos); |
// name |
buf.mark(); |
if(!parseName(selName)) { |
//log("parseAttribute no name " + bufPos); |
buf.reset(); |
return false; |
} |
|
// eq |
skipSpaces(); |
if(!testChar('=', bufPos)) { |
if(!testChar('=')) { |
throwException("equal sign expected"); |
} |
skipSpaces(); |
320,17 → 401,17 |
// FIXME allow 'Reference' here |
|
// value |
if(!testChar('"', bufPos)) { |
if(!testChar('"')) { |
throwException("quoted string expected"); |
} |
|
saveSelBegin(selValue); |
for(;; ++bufPos) { |
if(bufPos >= bufLen && !ensureNext(1)) { |
for(;; buf.toNext()) { |
if(buf.isEnd()) { |
throwException("unexpected EoF"); |
} |
|
byte c = buf[bufPos]; |
byte c = buf.cur(); |
if(c == '<' || c == '&' || c == '"') { |
break; |
} |
337,7 → 418,7 |
} |
saveSelEnd(selValue); |
|
if(!testChar('"', bufPos)) { |
if(!testChar('"')) { |
throwException("end of quoted string expected"); |
} |
|
373,24 → 454,20 |
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] != '-') { |
if(buf.at(0) != '<' || buf.at(1) != '!' || buf.at(2) != '-' || buf.at(3) != '-') { |
//log("parseComment no signature " + bufPos); |
return false; |
} |
|
bufPos += 4; |
for(;; ++bufPos) { |
if(bufPos+2 >= bufLen && !ensureNext(3)) { |
buf.skip(4); |
for(;; buf.toNext()) { |
if(buf.isEnd()) { |
throwException("unexpected EoF"); |
} |
|
if(buf[bufPos] == '-' && buf[bufPos+1] == '-') { |
if(buf[bufPos+2] == '>') { |
bufPos += 3; |
if(buf.at(0) == '-' && buf.at(1) == '-') { |
if(buf.at(2) == '>') { |
buf.skip(3); |
//log("parseComment ok " + bufPos); |
return true; |
} |
404,21 → 481,18 |
private boolean parseProcessInstruction() |
throws XmlException, IOException |
{ |
if(bufPos+1 >= bufLen && !ensureNext(2)) { |
if(buf.at(0) != '<' || buf.at(1) != '?') { |
return false; |
} |
if(buf[bufPos] != '<' || buf[bufPos+1] != '?') { |
return false; |
} |
|
bufPos += 2; |
for(;; ++bufPos) { |
if(bufPos+1 >= bufLen && !ensureNext(2)) { |
buf.skip(2); |
for(;; buf.toNext()) { |
if(buf.isEnd()) { |
throwException("unexpected EoF"); |
} |
|
if(buf[bufPos] == '?' && buf[bufPos+1] == '>') { |
bufPos += 3; |
if(buf.at(0) == '?' && buf.at(1) == '>') { |
buf.skip(2); |
return true; |
} |
} |
456,19 → 530,17 |
throws XmlException, IOException |
{ |
//log("parseStartTag begin " + bufPos); |
int start = bufPos; |
|
// begin |
if(!testChar('<', bufPos)) { |
if(!testChar('<')) { |
//log("parseStartTag no signature " + bufPos); |
return false; |
} |
|
// name |
buf.mark(); |
XmlSelection sel = new XmlSelection(); |
if(!parseName(sel)) { |
//log("parseStartTag no name " + bufPos); |
bufPos = start; |
buf.reset(); |
return false; |
} |
//System.out.print("[" + toString(sel) + "]"); |
484,8 → 556,8 |
//System.out.println(); |
|
// end |
element.isEmpty = testChar('/', bufPos); |
if(!testChar('>', bufPos)) { |
element.isEmpty = testChar('/'); |
if(!testChar('>')) { |
throwException("end of tag expected"); |
} |
|
498,7 → 570,7 |
{ |
//log("parseEndTag begin " + bufPos); |
// begin |
if(!testChar('<', bufPos) || !testChar('/', bufPos)) { |
if(!testChar('<') || !testChar('/')) { |
throwException("cannot find tag end"); |
} |
|
511,7 → 583,7 |
skipSpaces(); |
|
// end |
if(!testChar('>', bufPos)) { |
if(!testChar('>')) { |
throwException("end of tag expected"); |
} |
|
524,10 → 596,10 |
//log("parseTagContent begin " + bufPos); |
XmlSelection sel = new XmlSelection(); |
for(;;) { |
if(bufPos+1 >= bufLen && !ensureNext(2)) { |
if(buf.isEnd()) { |
throwException("unexpected EoF"); |
} |
if(buf[bufPos] == '<' && buf[bufPos+1] == '/') break; |
if(buf.at(0) == '<' && buf.at(1) == '/') break; |
|
if(parseElement() != null) { |
} |
555,13 → 627,9 |
boolean found = false; |
saveSelBegin(sel); |
|
for(;; ++bufPos) { |
if(bufPos >= bufLen && !ensureNext(1)) { |
return false; |
} |
|
byte c = buf[bufPos]; |
if(c == '<') { |
for(;; buf.toNext()) { |
byte c = buf.cur(); |
if(c == 0 || c == '<') { |
break; |
} |
|
576,29 → 644,26 |
private boolean parseCData(XmlSelection sel) |
throws XmlException, IOException |
{ |
if(bufPos+8 >= bufLen && !ensureNext(9)) { |
return false; |
} |
if(buf[bufPos] != '<' || buf[bufPos+1] != '!' |
|| buf[bufPos+2] != '[' || buf[bufPos+3] != 'C' |
|| buf[bufPos+4] != 'D' || buf[bufPos+5] != 'A' |
|| buf[bufPos+6] != 'T' || buf[bufPos+7] != 'A' |
|| buf[bufPos+8] != '[') |
if(buf.at(0) != '<' || buf.at(1) != '!' |
|| buf.at(2) != '[' || buf.at(3) != 'C' |
|| buf.at(4) != 'D' || buf.at(5) != 'A' |
|| buf.at(6) != 'T' || buf.at(7) != 'A' |
|| buf.at(8) != '[') |
{ |
return false; |
} |
|
bufPos += 9; |
buf.skip(9); |
saveSelBegin(sel); |
|
for(;; ++bufPos) { |
if(bufPos+2 >= bufLen && !ensureNext(3)) { |
for(;; buf.toNext()) { |
if(buf.isEnd()) { |
throwException("unexpected EoF"); |
} |
|
if(buf[bufPos] == ']' && buf[bufPos+1] == ']' && buf[bufPos+2] == '>') { |
if(buf.at(0) == ']' && buf.at(1) == ']' && buf.at(2) == '>') { |
saveSelEnd(sel); |
bufPos += 3; |
buf.skip(3); |
return true; |
} |
} |