static {
RSHIPS.put("NT", "BT");
RSHIPS.put("BT", "BT");
RSHIPS.put("BTG", "NTG");
RSHIPS.put("NTG", "BTG");
RSHIPS.put("BTP", "NTP");
RSHIPS.put("NTP", "BTP");
RSHIPS.put("USE", "UF");
RSHIPS.put("UF", "USE");
RSHIPS.put("RT", "RT");
}
/**
* Scans a PEReference. [69]
*
* @return entity
* @throws IOException I/O exception
*/
private byte[] peRef() throws IOException {
// scans predefined entities
final byte[] name = name(true);
consume(';');
final byte[] en = pents.get(name);
if (en != null) return en;
return name;
}
/**
* Scans a reference. [67]
*
* @param f dissolve entities
* @return entity
* @throws IOException I/O exception
*/
private byte[] ref(final boolean f) throws IOException {
// scans numeric entities
if (consume('#')) { // [66]
final TokenBuilder ent = new TokenBuilder();
int b = 10;
int ch = nextChar();
ent.add(ch);
if (ch == 'x') {
b = 16;
ent.add(ch = nextChar());
}
int n = 0;
do {
final boolean m = ch >= '0' && ch <= '9';
final boolean h = b == 16 && (ch >= 'a' && ch <= 'f' || ch >= 'A' && ch <= 'F');
if (!m && !h) {
completeRef(ent);
return QUESTION;
}
n *= b;
n += ch & 15;
if (!m) n += 9;
ent.add(ch = nextChar());
} while (ch != ';');
if (!valid(n)) return QUESTION;
ent.reset();
ent.add(n);
return ent.finish();
}
// scans predefined entities [68]
final byte[] name = name(false);
if (!consume(';')) return QUESTION;
if (!f) return concat(AMPER, name, SEMI);
byte[] en = ents.get(name);
if (en == null) {
// unknown entity: try HTML entities (lazy initialization)
if (HTMLENTS.size() == 0) {
for (int s = 0; s < HTMLENTITIES.length; s += 2) {
HTMLENTS.add(token(HTMLENTITIES[s]), token(HTMLENTITIES[s + 1]));
}
}
en = HTMLENTS.get(name);
}
return en == null ? QUESTION : en;
}
/**
* 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;
}
}
}
/**
* Builds the thesaurus.
*
* @param value input nodes
* @throws QueryException query exception
*/
private void build(final Value value) throws QueryException {
final Value synonyms = nodes("*:synonym", value);
if (synonyms.isEmpty()) return;
final ThesNode term = node(text("*:term", value));
for (final Item synonym : synonyms) {
final ThesNode sterm = node(text("*:term", synonym));
final byte[] rs = text("*:relationship", synonym);
term.add(sterm, rs);
final byte[] srs = RSHIPS.get(rs);
if (srs != null) sterm.add(term, srs);
build(synonyms);
}
}
/**
* 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;
}
}
/**
* Inserts a data instance at the specified pre value. Note that the specified data instance must
* differ from this instance.
*
* @param ipre value at which to insert new data
* @param ipar parent pre value of node
* @param clip data clip
*/
public final void insert(final int ipre, final int ipar, final DataClip clip) {
meta.update();
// update value and document indexes
if (meta.updindex) indexBegin();
resources.insert(ipre, clip);
final int dsize = clip.size();
final int buf = Math.min(dsize, IO.BLOCKSIZE >> IO.NODEPOWER);
// resize buffer to cache more entries
buffer(buf);
// find all namespaces in scope to avoid duplicate declarations
final TokenMap nsScope = nspaces.scope(ipar, this);
// loop through all entries
final IntList preStack = new IntList();
final NSNode nsRoot = nspaces.current();
final HashSet<NSNode> newNodes = new HashSet<NSNode>();
final IntList flagPres = new IntList();
// indicates if database only contains a dummy node
final Data data = clip.data;
int c = 0;
for (int dpre = clip.start; dpre < clip.end; ++dpre, ++c) {
if (c != 0 && c % buf == 0) insert(ipre + c - buf);
final int pre = ipre + c;
final int dkind = data.kind(dpre);
final int dpar = data.parent(dpre, dkind);
// ipar < 0 if document nodes on top level are added
final int dis = dpar >= 0 ? dpre - dpar : ipar >= 0 ? pre - ipar : 0;
final int par = dis == 0 ? -1 : pre - dis;
if (c == 0) nspaces.root(par, this);
while (!preStack.isEmpty() && preStack.peek() > par) nspaces.close(preStack.pop());
switch (dkind) {
case DOC:
// add document
nspaces.prepare();
final int s = data.size(dpre, dkind);
doc(pre, s, data.text(dpre, true));
meta.ndocs++;
preStack.push(pre);
break;
case ELEM:
// add element
nspaces.prepare();
boolean ne = false;
if (data.nsFlag(dpre)) {
final Atts at = data.ns(dpre);
for (int a = 0; a < at.size(); ++a) {
// see if prefix has been declared/ is part of current ns scope
final byte[] old = nsScope.get(at.name(a));
if (old == null || !eq(old, at.value(a))) {
// we have to keep track of all new NSNodes that are added
// to the Namespace structure, as their pre values must not
// be updated. I.e. if an NSNode N with pre value 3 existed
// prior to inserting and two new nodes are inserted at
// location pre == 3 we have to make sure N and only N gets
// updated.
newNodes.add(nspaces.add(at.name(a), at.value(a), pre));
ne = true;
}
}
}
byte[] nm = data.name(dpre, dkind);
elem(
dis,
tagindex.index(nm, null, false),
data.attSize(dpre, dkind),
data.size(dpre, dkind),
nspaces.uri(nm, true),
ne);
preStack.push(pre);
break;
case TEXT:
case COMM:
case PI:
// add text
text(pre, dis, data.text(dpre, true), dkind);
break;
case ATTR:
// add attribute
nm = data.name(dpre, dkind);
// check if prefix already in nsScope or not
final byte[] attPref = prefix(nm);
// check if prefix of attribute has already been declared, otherwise
// add declaration to parent node
if (data.nsFlag(dpre) && nsScope.get(attPref) == null) {
nspaces.add(
par,
preStack.isEmpty() ? -1 : preStack.peek(),
attPref,
data.nspaces.uri(data.uri(dpre, dkind)),
this);
// save pre value to set ns flag later for this node. can't be done
// here as direct table access would interfere with the buffer
flagPres.add(par);
}
attr(
pre,
dis,
atnindex.index(nm, null, false),
data.text(dpre, false),
nspaces.uri(nm, false),
false);
break;
}
}
// finalize and update namespace structure
while (!preStack.isEmpty()) nspaces.close(preStack.pop());
nspaces.root(nsRoot);
if (bp != 0) insert(ipre + c - 1 - (c - 1) % buf);
// reset buffer to old size
buffer(1);
// set ns flags
for (int f = 0; f < flagPres.size(); f++) {
final int fl = flagPres.get(f);
table.write2(fl, 1, name(fl) | 1 << 15);
}
// increase size of ancestors
int p = ipar;
while (p >= 0) {
final int k = kind(p);
size(p, k, size(p, k) + dsize);
p = parent(p, k);
}
if (meta.updindex) {
// add the entries to the ID -> PRE mapping:
idmap.insert(ipre, id(ipre), dsize);
indexEnd();
}
if (!cache) updateDist(ipre + dsize, dsize);
// propagate PRE value shifts to namespaces
if (ipar != -1) nspaces.insert(ipre, dsize, newNodes);
}
/**
* Scans a markup declaration. [29]
*
* @return true if a declaration was found
* @throws IOException I/O exception
*/
private boolean markupDecl() throws IOException {
if (consume(ENT)) { // [70]
checkS();
if (consume('%')) { // [72] PEDecl
checkS();
final byte[] key = name(true);
checkS();
byte[] val = entityValue(true); // [74]
if (val == null) {
val = externalID(true, false);
if (val == null) error(INVEND);
}
s();
pents.add(key, val);
} else { // [71] GEDecl
final byte[] key = name(true);
checkS();
byte[] val = entityValue(false); // [73] EntityDef
if (val == null) {
val = externalID(true, false);
if (val == null) error(INVEND);
if (s()) {
check(ND);
checkS();
name(true);
}
}
s();
ents.add(key, val);
}
check('>');
pe = true;
} else if (consume(ELEM)) { // [45]
checkS();
name(true);
checkS();
pe = true;
if (!consume(EMP) && !consume(ANY)) { // [46]
if (consume('(')) {
s();
if (consume(PC)) { // [51]
s();
boolean alt = false;
while (consume('|')) {
s();
name(true);
s();
alt = true;
}
check(')');
if (!consume('*') && alt) error(INVEND);
} else {
cp();
s();
// check(')'); // to be fixed...
while (!consume(')')) consume();
// input.prev(1);
occ();
}
} else {
error(INVEND);
}
}
s();
check('>');
} else if (consume(ATTL)) { // [52]
pe = true;
checkS();
name(true);
s();
while (name(false) != null) { // [53]
checkS();
if (!consume(CD)
&& !consume(IDRS)
&& !consume(IDR)
&& !consume(ID)
&& !consume(ENTS)
&& !consume(ENT1)
&& !consume(NMTS)
&& !consume(NMT)) { // [56]
if (consume(NOT)) { // [57,58]
checkS();
check('(');
s();
name(true);
s();
while (consume('|')) {
s();
name(true);
s();
}
check(')');
} else { // [59]
check('(');
s();
nmtoken();
s();
while (consume('|')) {
s();
nmtoken();
s();
}
check(')');
}
}
// [54]
pe = true;
checkS();
if (!consume(REQ) && !consume(IMP)) { // [60]
if (consume(FIX)) checkS();
quote = qu();
attValue(consume());
}
s();
}
check('>');
} else if (consume(NOTA)) { // [82]
checkS();
name(true);
s();
externalID(false, false);
s();
check('>');
} else if (consume(COMS)) {
comment();
} else if (consume(XML)) {
pi();
} else {
return false;
}
s();
pe = false;
return true;
}