@Override
public void execute(final GUI gui) {
final DialogExport dialog = new DialogExport(gui);
if (!dialog.ok()) return;
final IOFile root = new IOFile(dialog.path());
// check if existing files will be overwritten
if (root.exists()) {
IO file = null;
boolean overwrite = false;
final Data d = gui.context.data();
final IntList il = d.resources.docs();
final int is = il.size();
for (int i = 0; i < is; i++) {
file = root.merge(Token.string(d.text(il.get(i), true)));
if (file.exists()) {
if (overwrite) {
// more than one file will be overwritten; check remaining tests
file = null;
break;
}
overwrite = true;
}
}
if (overwrite) {
// show message for overwriting files or directories
final String msg = file == null ? FILES_REPLACE_X : FILE_EXISTS_X;
if (file == null) file = root;
if (!BaseXDialog.confirm(gui, Util.info(msg, file))) return;
}
}
DialogProgress.execute(gui, new Export(root.path()));
}
/**
* Creates new node ids and recreates updatable index structures.
*
* @param data data
* @throws IOException I/O Exception during index rebuild
*/
public static void ids(final Data data) throws IOException {
final MetaData md = data.meta;
final int size = md.size;
for (int pre = 0; pre < size; ++pre) data.id(pre, pre);
md.lastid = size - 1;
md.dirty = true;
if (data.meta.updindex) {
data.idmap = new IdPreMap(md.lastid);
if (data.meta.textindex) optimize(IndexType.TEXT, data, true, true, true, null);
if (data.meta.attrindex) optimize(IndexType.ATTRIBUTE, data, true, true, true, null);
}
}
/**
* Lists resources of the specified database.
*
* @return success flag
* @throws IOException I/O exception
*/
private boolean listDB() throws IOException {
final String db = args[0];
final String path = args[1] != null ? args[1] : "";
if (!Databases.validName(db)) return error(NAME_INVALID_X, db);
final Table table = new Table();
table.description = RESOURCES;
table.header.add(INPUT_PATH);
table.header.add(TYPE);
table.header.add(MimeTypes.CONTENT_TYPE);
table.header.add(SIZE);
try {
// add xml documents
final Data data = Open.open(db, context);
final Resources res = data.resources;
final IntList il = res.docs(path);
final int ds = il.size();
for (int i = 0; i < ds; i++) {
final int pre = il.get(i);
final TokenList tl = new TokenList(3);
final byte[] file = data.text(pre, true);
tl.add(file);
tl.add(DataText.M_XML);
tl.add(MimeTypes.APP_XML);
tl.add(data.size(pre, Data.DOC));
table.contents.add(tl);
}
// add binary resources
for (final byte[] file : res.binaries(path)) {
final String f = string(file);
final TokenList tl = new TokenList(3);
tl.add(file);
tl.add(DataText.M_RAW);
tl.add(MimeTypes.get(f));
tl.add(data.meta.binary(f).length());
table.contents.add(tl);
}
Close.close(data, context);
} catch (final IOException ex) {
return error(Util.message(ex));
}
out.println(table.sort().finish());
return true;
}
@Override
public void execute(final GUI gui) {
// skip operation for root context
final Context ctx = gui.context;
if (ctx.root()) return;
// check if all nodes are document nodes
boolean doc = true;
final Data data = ctx.data();
for (final int pre : ctx.current().pres) doc &= data.kind(pre) == Data.DOC;
if (doc) {
// if yes, jump to database root
ctx.update();
gui.notify.context(ctx.current(), false, null);
} else {
// otherwise, jump to parent nodes
gui.execute(new Cs(".."));
}
}
@Override
public boolean indexAccessible(final IndexInfo ii) {
/* If the following conditions yield true, the index is accessed:
* - all query terms are statically available
* - no FTTimes option is specified
* - explicitly set case, diacritics and stemming match options do not
* conflict with index options. */
data = ii.ic.data;
final MetaData md = data.meta;
final FTOpt fto = ftt.opt;
/* Index will be applied if no explicit match options have been set
* that conflict with the index options. As a consequence, though, index-
* based querying might yield other results than sequential scanning. */
if (occ != null
|| fto.cs != null && md.casesens == (fto.cs == FTCase.INSENSITIVE)
|| fto.isSet(DC) && md.diacritics != fto.is(DC)
|| fto.isSet(ST) && md.stemming != fto.is(ST)
|| fto.ln != null && !fto.ln.equals(md.language)) return false;
// adopt database options to tokenizer
fto.copy(md);
// estimate costs if text is not known at compile time
if (tokens == null) {
ii.costs = Math.max(2, data.meta.size / 30);
return true;
}
// summarize number of hits; break loop if no hits are expected
final FTLexer ft = new FTLexer(fto);
ii.costs = 0;
for (byte[] t : tokens) {
ft.init(t);
while (ft.hasNext()) {
final byte[] tok = ft.nextToken();
if (fto.sw != null && fto.sw.contains(tok)) continue;
if (fto.is(WC)) {
// don't use index if one of the terms starts with a wildcard
t = ft.get();
if (t[0] == '.') return false;
// don't use index if certain characters or more than 1 dot are found
int d = 0;
for (final byte w : t) {
if (w == '{' || w == '\\' || w == '.' && ++d > 1) return false;
}
}
// favor full-text index requests over exact queries
final int costs = data.costs(ft);
if (costs != 0) ii.costs += Math.max(2, costs / 100);
}
}
return true;
}
/**
* 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);
}
/**
* Replaces parts of the database with the specified data instance.
*
* @param rpre pre value to be replaced
* @param clip data clip
*/
public final void replace(final int rpre, final DataClip clip) {
meta.update();
final int dsize = clip.size();
final Data data = clip.data;
final int rkind = kind(rpre);
final int rsize = size(rpre, rkind);
final int rpar = parent(rpre, rkind);
final int diff = dsize - rsize;
buffer(dsize);
resources.replace(rpre, rsize, clip);
if (meta.updindex) {
// update index
indexDelete(rpre, rsize);
indexBegin();
}
for (int dpre = clip.start; dpre < clip.end; ++dpre) {
final int dkind = data.kind(dpre);
final int dpar = data.parent(dpre, dkind);
final int pre = rpre + dpre - clip.start;
final int dis = dpar >= 0 ? dpre - dpar : pre - rpar;
switch (dkind) {
case DOC:
// add document
doc(pre, data.size(dpre, dkind), data.text(dpre, true));
meta.ndocs++;
break;
case ELEM:
// add element
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),
false);
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);
attr(
pre,
dis,
atnindex.index(nm, null, false),
data.text(dpre, false),
nspaces.uri(nm, false),
false);
break;
}
}
if (meta.updindex) {
indexEnd();
// update ID -> PRE map:
idmap.delete(rpre, id(rpre), -rsize);
idmap.insert(rpre, meta.lastid - dsize + 1, dsize);
}
// update table:
table.replace(rpre, buffer(), rsize);
buffer(1);
// no distance/size update if the two subtrees are of equal size
if (diff == 0) return;
// increase/decrease size of ancestors, adjust distances of siblings
int p = rpar;
while (p >= 0) {
final int k = kind(p);
size(p, k, size(p, k) + diff);
p = parent(p, k);
}
if (!cache) updateDist(rpre + dsize, diff);
// adjust attribute size of parent if attributes inserted. attribute size
// of parent cannot be reduced via a replace expression.
int dpre = clip.start;
if (data.kind(dpre) == ATTR) {
int d = 0;
while (dpre < clip.end && data.kind(dpre++) == ATTR) d++;
if (d > 1) attSize(rpar, kind(rpar), d + 1);
}
}
/**
* 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);
}
