@Override
public String det() {
final String path = input.io().path();
return path.isEmpty()
? Util.info(LINE_X, input.line())
: Util.info(SCANPOS_X_X, input.io().path(), input.line());
}
/**
* Scans an external ID.
*
* @param f full flag
* @param r root flag
* @return id
* @throws IOException I/O exception
*/
private byte[] externalID(final boolean f, final boolean r) throws IOException {
byte[] cont = null;
final boolean pub = consume(PUBLIC);
if (pub || consume(SYSTEM)) {
checkS();
if (pub) {
pubidLit();
if (f) checkS();
}
final int qu = consume(); // [11]
if (qu == '\'' || qu == '"') {
int ch;
final TokenBuilder tok = new TokenBuilder();
while ((ch = nextChar()) != qu) tok.add(ch);
if (!f) return null;
final String name = string(tok.finish());
if (!dtd && r) return cont;
final XMLInput tin = input;
try {
final IO file = input.io().merge(name);
cont = file.read();
} catch (final IOException ex) {
Util.debug(ex);
// skip unknown DTDs/entities
cont = new byte[] {'?'};
}
input = new XMLInput(new IOContent(cont, name));
if (consume(XDECL)) {
check(XML);
s();
if (version()) checkS();
s();
if (encoding() == null) error(TEXTENC);
ch = nextChar();
if (s(ch)) ch = nextChar();
if (ch != '?') error(WRONGCHAR, '?', ch);
ch = nextChar();
if (ch != '>') error(WRONGCHAR, '>', ch);
cont = Arrays.copyOfRange(cont, input.pos(), cont.length);
}
s();
if (r) {
extSubsetDecl();
if (!consume((char) 0)) error(INVEND);
}
input = tin;
} else {
if (f) error(SCANQUOTE, (char) qu);
prev(1);
}
}
return cont;
}
/**
* Reads next character or throws an exception if all bytes have been read.
*
* @return next character
* @throws IOException I/O exception
*/
private int consume() throws IOException {
while (true) {
final int ch = input.read();
if (ch == -1) return 0;
if (ch == '%' && pe) { // [69]
final byte[] key = name(true);
final byte[] val = pents.get(key);
if (val == null) error(UNKNOWNPE, key);
check(';');
input.add(val, true);
} else {
return ch;
}
}
}
/**
* Scans an attribute value. [10]
*
* @param ch current character
* @throws IOException I/O exception
*/
private void attValue(final int ch) throws IOException {
boolean wrong = false;
int c = ch;
do {
if (c == 0) error(ATTCLOSE, (char) c);
wrong |= c == '\'' || c == '"';
if (c == '<') error(wrong ? ATTCLOSE : ATTCHAR, (char) c);
if (c == 0x0A) c = ' ';
if (c == '&') {
// verify...
final byte[] r = ref(true);
if (r.length == 1) token.add(r);
else if (!input.add(r, false)) error(RECENT);
} else {
token.add(c);
}
} while ((c = consume()) != quote);
}
/**
* Initializes the scanner.
*
* @param f input file
* @param pr database properties
* @param frag allow parsing of document fragment
* @throws IOException I/O exception
*/
XMLScanner(final IO f, final Prop pr, final boolean frag) throws IOException {
input = new XMLInput(f);
fragment = frag;
try {
for (int e = 0; e < ENTITIES.length; e += 2) {
ents.add(token(ENTITIES[e]), token(ENTITIES[e + 1]));
}
dtd = pr.is(Prop.DTD);
chop = pr.is(Prop.CHOP);
String enc = null;
// process document declaration...
if (consume(DOCDECL)) {
if (s()) {
if (!version()) error(DECLSTART);
boolean s = s();
enc = encoding();
if (enc != null) {
if (!s) error(WSERROR);
s = s();
}
if (sddecl() != null && !s) error(WSERROR);
s();
final int ch = nextChar();
if (ch != '?' || nextChar() != '>') error(DECLWRONG);
} else {
prev(5);
}
}
encoding = enc == null ? UTF8 : enc;
if (!fragment) {
final int n = consume();
if (!s(n)) {
if (n != '<') error(BEFOREROOT);
prev(1);
}
}
} catch (final IOException ex) {
input.close();
throw ex;
}
}
/**
* Scans XML text.
*
* @param ch current character
* @throws IOException I/O exception
*/
private void content(final int ch) throws IOException {
type = Type.TEXT;
boolean f = true;
int c = ch;
while (c != 0) {
if (c != '<') {
if (c == '&') {
// scan entity
final byte[] r = ref(true);
if (r.length == 1) token.add(r);
else if (!input.add(r, false)) error(RECENT);
} else {
if (c == ']') {
// ']]>' not allowed in content
if (consume() == ']') {
if (consume() == '>') error(CONTCDATA);
prev(1);
}
prev(1);
}
// add character to cached content
token.add(c);
}
} else {
if (!f && !isCDATA()) {
text = false;
prev(1);
if (chop) token.trim();
return;
}
cDATA();
}
c = consume();
f = false;
}
// end of file
if (!fragment) {
if (!ws(token.finish())) error(AFTERROOT);
type = Type.EOF;
}
}
/**
* Scans a document encoding.
*
* @return encoding
* @throws IOException I/O exception
*/
private String encoding() throws IOException {
if (!consume(ENCOD)) {
if (fragment) error(TEXTENC);
return null;
}
s();
check('=');
s();
final TokenBuilder enc = new TokenBuilder();
final int d = qu();
int ch = nextChar();
if (letter(ch) && ch != '_') {
while (letterOrDigit(ch) || ch == '.' || ch == '-') {
enc.add(ch);
ch = nextChar();
}
prev(1);
}
check((char) d);
if (enc.isEmpty()) error(DECLENCODE, enc);
final String e = string(enc.finish());
input.encoding(e);
return e;
}
/**
* Jumps the specified number of characters back.
*
* @param p number of characters
*/
private void prev(final int p) {
input.prev(p);
}
/**
* Finishes file scanning.
*
* @throws IOException I/O exception
*/
void close() throws IOException {
input.close();
if (!fragment && prolog) error(DOCEMPTY);
}
@Override
public double prog() {
final double l = input.length();
return l <= 0 ? 0 : input.pos() / l;
}