/**
* Searches a string in a file.
*
* @param path file path
* @param search codepoints of search string
* @return success flag
*/
private static boolean filterContent(final String path, final int[] search) {
final int cl = search.length;
if (cl == 0) return true;
try (final TextInput ti = new TextInput(new IOFile(path))) {
final IntList il = new IntList(cl - 1);
int c = 0;
while (true) {
if (!il.isEmpty()) {
if (il.remove(0) == search[c++]) continue;
c = 0;
}
while (true) {
final int cp = ti.read();
if (cp == -1 || !XMLToken.valid(cp)) return false;
final int lc = Token.lc(cp);
if (c > 0) il.add(lc);
if (lc == search[c]) {
if (++c == cl) return true;
} else {
c = 0;
break;
}
}
}
} catch (final IOException ex) {
// file may not be accessible
Util.debug(ex);
return false;
}
}
@Override
public void add(final ValueCache cache) {
for (final byte[] key : cache) {
final IntList vals = cache.ids(key);
if (!vals.isEmpty()) add(key, vals.sort().finish());
}
finish();
}
/**
* Optimizes the structures of a database.
*
* @param data data
* @param enforceText enforce creation or deletion of text index
* @param enforceAttr enforce creation or deletion of attribute index
* @param enforceToken enforce creation or deletion of token index
* @param enforceFt enforce creation or deletion of full-text index
* @param cmd calling command instance (may be {@code null})
* @throws IOException I/O Exception during index rebuild
*/
public static void optimize(
final Data data,
final boolean enforceText,
final boolean enforceAttr,
final boolean enforceToken,
final boolean enforceFt,
final Optimize cmd)
throws IOException {
// initialize structural indexes
final MetaData md = data.meta;
if (!md.uptodate) {
data.paths.init();
data.elemNames.init();
data.attrNames.init();
md.dirty = true;
final IntList pars = new IntList(), elms = new IntList();
int n = 0;
for (int pre = 0; pre < md.size; ++pre) {
final byte kind = (byte) data.kind(pre);
final int par = data.parent(pre, kind);
while (!pars.isEmpty() && pars.peek() > par) {
pars.pop();
elms.pop();
}
final int level = pars.size();
if (kind == Data.DOC) {
data.paths.put(0, Data.DOC, level);
pars.push(pre);
elms.push(0);
++n;
} else if (kind == Data.ELEM) {
final int id = data.nameId(pre);
data.elemNames.index(data.elemNames.key(id), null, true);
data.paths.put(id, Data.ELEM, level);
pars.push(pre);
elms.push(id);
} else if (kind == Data.ATTR) {
final int id = data.nameId(pre);
final byte[] val = data.text(pre, false);
data.attrNames.index(data.attrNames.key(id), val, true);
data.paths.put(id, Data.ATTR, level, val, md);
} else {
final byte[] val = data.text(pre, true);
if (kind == Data.TEXT && level > 1) data.elemNames.index(elms.peek(), val);
data.paths.put(0, kind, level, val, md);
}
if (cmd != null) cmd.pre = pre;
}
md.ndocs = n;
md.uptodate = true;
}
// rebuild value indexes
optimize(IndexType.TEXT, data, md.createtext, md.textindex, enforceText, cmd);
optimize(IndexType.ATTRIBUTE, data, md.createattr, md.attrindex, enforceAttr, cmd);
optimize(IndexType.TOKEN, data, md.createtoken, md.tokenindex, enforceToken, cmd);
optimize(IndexType.FULLTEXT, data, md.createft, md.ftindex, enforceFt, cmd);
}