@Override
public Item item(final QueryContext qc, final InputInfo ii) throws QueryException {
checkCreate(qc);
final String driver = string(toToken(exprs[0], qc));
if (Reflect.find(driver) == null) throw BXSQ_DRIVER_X.get(info, driver);
return null;
}
static {
IOFile dic = null;
if (Reflect.available(PATTERN)) {
dic = new IOFile(LANG);
if (!dic.exists()) {
dic = new IOFile(Prop.HOME, "etc/" + LANG);
if (!dic.exists()) {
available = false;
}
}
} else {
available = false;
}
if (available) {
Class<?> clz = Reflect.find(PATTERN);
if (clz == null) {
Util.debug("Could not initialize Igo Japanese lexer.");
} else {
/* Igo constructor. */
final Constructor<?> tgr = Reflect.find(clz, String.class);
tagger = Reflect.get(tgr, dic.path());
if (tagger == null) {
available = false;
Util.debug("Could not initialize Igo Japanese lexer.");
} else {
parse = Reflect.method(clz, "parse", CharSequence.class);
if (parse == null) {
Util.debug("Could not initialize Igo lexer method.");
}
clz = Reflect.find("net.reduls.igo.Morpheme");
surface = Reflect.field(clz, "surface");
feature = Reflect.field(clz, "feature");
start = Reflect.field(clz, "start");
}
}
}
}
static {
if (Reflect.available(PATTERN, "German")) {
for (final Language l : Language.ALL.values()) {
final Class<?> clz = Reflect.find(PATTERN, l);
if (clz == null) continue;
final Method m1 = Reflect.method(clz, "setCurrent", String.class);
final Method m2 = Reflect.method(clz, "stem");
final Method m3 = Reflect.method(clz, "getCurrent");
if (m1 == null || m2 == null || m3 == null) {
Util.errln("Could not initialize \"%\" Snowball stemmer.", l);
} else {
CLASSES.put(l, new StemmerClass(clz, m1, m2, m3));
}
}
}
}
/**
* Creates a new instance of the function.
*
* @param ii input info
* @param arg arguments
* @return function
*/
public FuncCall get(final InputInfo ii, final Expr... arg) {
return (FuncCall)
Reflect.get(
Reflect.find(func, InputInfo.class, Function.class, Expr[].class), ii, this, arg);
}
/**
* This class uses TagSoup to convert HTML input to well-formed XML. If TagSoup is not found in the
* classpath, the original document is passed on.
*
* <p>TagSoup was written by John Cowan and is based on the Apache 2.0 License: {@code
* http://home.ccil.org/~cowan/XML/tagsoup/}.
*
* @author BaseX Team 2005-14, BSD License
* @author Christian Gruen
*/
public final class HtmlParser extends XMLParser {
/** Name of HTML Parser. */
private static final String NAME = "TagSoup";
/** TagSoup URL. */
private static final String FEATURES = "http://www.ccil.org/~cowan/tagsoup/features/";
/** XML parser class string. */
private static final String PCLASS = "org.ccil.cowan.tagsoup.Parser";
/** XML writer class string. */
private static final String WCLASS = "org.ccil.cowan.tagsoup.XMLWriter";
/** HTML reader. */
private static final Class<?> READER = Reflect.find(PCLASS);
/** HTML writer. */
private static final Constructor<?> WRITER = Reflect.find(Reflect.find(WCLASS), Writer.class);
/** XML writer output property method. */
private static final Method METHOD =
Reflect.method(
Reflect.find(WCLASS), "setOutputProperty", new Class[] {String.class, String.class});
/**
* Checks if a CatalogResolver is available.
*
* @return result of check
*/
public static boolean available() {
return READER != null;
}
/**
* Returns the name of the parser, or an empty string.
*
* @return name of parser
*/
public static String parser() {
return available() ? NAME : "";
}
/**
* Constructor.
*
* @param source document source
* @param opts database options
* @throws IOException I/O exception
*/
public HtmlParser(final IO source, final MainOptions opts) throws IOException {
this(source, opts, opts.get(MainOptions.HTMLPARSER));
}
/**
* Constructor.
*
* @param source document source
* @param opts database options
* @param hopts html options
* @throws IOException I/O exception
*/
public HtmlParser(final IO source, final MainOptions opts, final HtmlOptions hopts)
throws IOException {
super(toXML(source, hopts), opts);
}
/**
* Converts an HTML document to XML.
*
* @param io io reference
* @param opts html options
* @return parser
* @throws IOException I/O exception
*/
private static IO toXML(final IO io, final HtmlOptions opts) throws IOException {
// reader could not be initialized; fall back to XML
if (READER == null) return io;
try {
// tries to extract the encoding from the input
final TextInput ti = new TextInput(io);
String enc = ti.encoding();
final byte[] content = ti.content();
// looks for a charset definition
final byte[] encoding = token("charset=");
int cs = indexOf(content, encoding);
if (cs > 0) {
// extracts the encoding string
cs += encoding.length;
int ce = cs;
final int cl = content.length;
while (++ce < cl && content[ce] > 0x28) ;
enc = string(substring(content, cs, ce));
}
// define input
final InputSource is = new InputSource(new ArrayInput(content));
is.setEncoding(supported(enc) ? normEncoding(enc) : UTF8);
// define output
final StringWriter sw = new StringWriter();
final XMLReader reader = (XMLReader) Reflect.get(READER);
final Object writer = Reflect.get(WRITER, sw);
// set TagSoup options
if (opts.get(HtmlOptions.HTML)) {
reader.setFeature("http://xml.org/sax/features/namespaces", false);
opt("method", "html");
opt("omit-xml-declaration", "yes");
}
if (opts.get(HtmlOptions.NONS))
reader.setFeature("http://xml.org/sax/features/namespaces", false);
if (opts.get(HtmlOptions.OMITXML)) opt("omit-xml-declaration", "yes");
if (opts.get(HtmlOptions.NOBOGONS)) reader.setFeature(FEATURES + "ignore-bogons", true);
if (opts.get(HtmlOptions.NODEFAULTS))
reader.setFeature(FEATURES + "default-attributes", false);
if (opts.get(HtmlOptions.NOCOLONS)) reader.setFeature(FEATURES + "translate-colons", true);
if (opts.get(HtmlOptions.NORESTART)) reader.setFeature(FEATURES + "restart-elements", false);
if (opts.get(HtmlOptions.IGNORABLE))
reader.setFeature(FEATURES + "ignorable-whitespace", true);
if (opts.get(HtmlOptions.EMPTYBOGONS)) reader.setFeature(FEATURES + "bogons-empty", true);
if (opts.get(HtmlOptions.ANY)) reader.setFeature(FEATURES + "bogons-empty", false);
if (opts.get(HtmlOptions.NOROOTBOGONS)) reader.setFeature(FEATURES + "root-bogons", false);
if (opts.get(HtmlOptions.NOCDATA)) reader.setFeature(FEATURES + "cdata-elements", false);
if (opts.get(HtmlOptions.LEXICAL))
reader.setProperty("http://xml.org/sax/properties/lexical-handler", writer);
if (opts.contains(HtmlOptions.METHOD)) opt("method", opts.get(HtmlOptions.METHOD));
if (opts.contains(HtmlOptions.DOCTYPESYS))
opt("doctype-system", opts.get(HtmlOptions.DOCTYPESYS));
if (opts.contains(HtmlOptions.DOCTYPEPUB))
opt("doctype-public", opts.get(HtmlOptions.DOCTYPEPUB));
if (opts.contains(HtmlOptions.ENCODING)) is.setEncoding(opts.get(HtmlOptions.ENCODING));
// end TagSoup options
reader.setContentHandler((ContentHandler) writer);
reader.parse(is);
return new IOContent(token(sw.toString()), io.name());
} catch (final SAXException ex) {
Util.errln(ex);
return io;
}
}
/**
* Reflection invoke XMLWriter.setOutputProperty().
*
* @param name property
* @param value value
*/
private static void opt(final String name, final String value) {
Reflect.invoke(METHOD, WRITER, name, value);
}
}
/**
* Initializes JDBC with the specified driver.
*
* @param ctx query context
* @return {@code null}
* @throws QueryException query exception
*/
private Item init(final QueryContext ctx) throws QueryException {
final String driver = string(checkStr(expr[0], ctx));
if (Reflect.find(driver) == null) SQLINIT.thrw(input, driver);
return null;
}
/**
* Returns the specified function.
*
* @param name name of the function
* @param args optional arguments
* @param dyn compile-/run-time flag
* @param ctx query context
* @param ii input info
* @return function instance
* @throws QueryException query exception
*/
public TypedFunc get(
final QNm name,
final Expr[] args,
final boolean dyn,
final QueryContext ctx,
final InputInfo ii)
throws QueryException {
// find function
final byte[] uri = name.uri();
final byte[] ln = name.local();
// parse data type constructors
if (eq(uri, XSURI)) {
final Type type = AtomType.find(name, true);
if (type == null || type == AtomType.NOT || type == AtomType.AAT) {
final Levenshtein ls = new Levenshtein();
for (final AtomType t : AtomType.values()) {
if (t.par != null && ls.similar(lc(ln), lc(t.string()), 0))
FUNSIMILAR.thrw(ii, name.string(), t.string());
}
FUNCUNKNOWN.thrw(ii, name.string());
}
if (args.length != 1) FUNCTYPE.thrw(ii, name.string());
final SeqType to = SeqType.get(type, SeqType.Occ.ZO);
return TypedFunc.constr(new Cast(ii, args[0], to), to);
}
// check Java functions - only allowed with administrator permissions
if (startsWith(uri, JAVAPRE) && ctx.context.user.perm(User.ADMIN)) {
final String c = string(substring(uri, JAVAPRE.length));
// convert dashes to upper-case initials
final TokenBuilder tb = new TokenBuilder().add(c).add('.');
boolean dash = false;
for (int p = 0; p < ln.length; p += cl(ln, p)) {
final int ch = cp(ln, p);
if (dash) {
tb.add(Character.toUpperCase(ch));
dash = false;
} else {
dash = ch == '-';
if (!dash) tb.add(ch);
}
}
final String java = tb.toString();
final int i = java.lastIndexOf(".");
final Class<?> cls = Reflect.find(java.substring(0, i));
if (cls == null) FUNCJAVA.thrw(ii, java);
final String mth = java.substring(i + 1);
return TypedFunc.java(new JavaFunc(ii, cls, mth, args));
}
// check predefined functions
final FuncCall fun = Functions.get().get(ln, uri, args, ctx, ii);
if (fun != null) {
for (final Function f : Function.UPDATING) {
if (fun.def == f) {
ctx.updating = true;
break;
}
}
return new TypedFunc(fun, fun.def.type(args.length));
}
// find local function
for (int l = 0; l < func.length; ++l) {
final QNm qn = func[l].name;
if (eq(ln, qn.local()) && eq(uri, qn.uri()) && args.length == func[l].args.length)
return new TypedFunc(add(ii, qn, l, args), FuncType.get(func[l]));
}
// add function call for function that has not been defined yet
if (!dyn && Types.find(name, false) == null) {
return new TypedFunc(
add(ii, name, add(new UserFunc(ii, name, new Var[args.length], null, false), ii), args),
FuncType.arity(args.length));
}
return null;
}
