/**
* Construct a new instance and configure it.
*
* @param broker
* @param conf
*/
public TextSearchEngine(DBBroker broker, Configuration conf) {
this.broker = broker;
this.config = conf;
String stopword, tokenizerClass;
Boolean num, stemming, termFrequencies;
if ((num = (Boolean) config.getProperty(PROPERTY_INDEX_NUMBERS)) != null)
indexNumbers = num.booleanValue();
if ((stemming = (Boolean) config.getProperty(PROPERTY_STEM)) != null)
stem = stemming.booleanValue();
if ((termFrequencies = (Boolean) config.getProperty(PROPERTY_STORE_TERM_FREQUENCY)) != null)
termFreq = termFrequencies.booleanValue();
String track = (String) config.getProperty(Serializer.PROPERTY_TAG_MATCHING_ELEMENTS);
if (track != null)
trackMatches =
track.equalsIgnoreCase("yes") ? Serializer.TAG_ELEMENT_MATCHES : Serializer.TAG_NONE;
track = (String) config.getProperty(Serializer.PROPERTY_TAG_MATCHING_ATTRIBUTES);
if (track != null && track.equalsIgnoreCase("yes"))
trackMatches = trackMatches | Serializer.TAG_ATTRIBUTE_MATCHES;
if ((tokenizerClass = (String) config.getProperty(PROPERTY_TOKENIZER)) != null) {
try {
Class tokClass = Class.forName(tokenizerClass);
tokenizer = (Tokenizer) tokClass.newInstance();
LOG.debug("using tokenizer: " + tokenizerClass);
} catch (ClassNotFoundException e) {
LOG.debug(e);
} catch (InstantiationException e) {
LOG.debug(e);
} catch (IllegalAccessException e) {
LOG.debug(e);
}
}
if (tokenizer == null) {
LOG.debug("using simple tokenizer");
tokenizer = new SimpleTokenizer();
}
if (stem) stemmer = new PorterStemmer();
tokenizer.setStemming(stem);
if ((stopword = (String) config.getProperty(PROPERTY_STOPWORD_FILE)) != null) {
try {
FileReader in = new FileReader(stopword);
StreamTokenizer tok = new StreamTokenizer(in);
int next = tok.nextToken();
while (next != StreamTokenizer.TT_EOF) {
if (next != StreamTokenizer.TT_WORD) continue;
stoplist.add(tok.sval);
next = tok.nextToken();
}
} catch (FileNotFoundException e) {
LOG.debug(e);
} catch (IOException e) {
LOG.debug(e);
}
}
}
/** Constructor for XUpdateProcessor. */
public XUpdateProcessor(DBBroker broker, DocumentSet docs, AccessContext accessCtx)
throws ParserConfigurationException {
if (accessCtx == null) {
throw new NullAccessContextException();
}
this.accessCtx = accessCtx;
final DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
factory.setNamespaceAware(true);
factory.setValidating(false);
this.builder = factory.newDocumentBuilder();
this.broker = broker;
this.documentSet = docs;
// namespaces.put("xml", Namespaces.XML_NS);
// TODO : move this to a dedicated configure() method.
if (broker != null) {
final Configuration config = broker.getConfiguration();
Boolean temp;
if ((temp = (Boolean) config.getProperty("indexer.preserve-whitespace-mixed-content"))
!= null) {
preserveWhitespaceTemp = temp.booleanValue();
}
}
}
/**
* Setup Validator object with brokerpool as db connection.
*
* @param pool Brokerpool
*/
public Validator(BrokerPool pool) {
logger.info("Initializing Validator.");
if (brokerPool == null) {
this.brokerPool = pool;
}
// Get configuration
config = brokerPool.getConfiguration();
// Check xerces version
StringBuilder xmlLibMessage = new StringBuilder();
if (!XmlLibraryChecker.hasValidParser(xmlLibMessage)) {
logger.error(xmlLibMessage);
}
// setup grammar brokerPool
grammarPool = (GrammarPool) config.getProperty(XMLReaderObjectFactory.GRAMMER_POOL);
// setup system wide catalog resolver
systemCatalogResolver =
(eXistXMLCatalogResolver) config.getProperty(XMLReaderObjectFactory.CATALOG_RESOLVER);
}
public void configure(Configuration config, Properties properties) throws EXistException {
exportDir = properties.getProperty(OUTPUT_PROP_NAME, "export");
File dir = new File(exportDir);
if (!dir.isAbsolute()) {
dir = new File((String) config.getProperty(BrokerPool.PROPERTY_DATA_DIR), exportDir);
}
dir.mkdirs();
exportDir = dir.getAbsolutePath();
if (LOG.isDebugEnabled()) {
LOG.debug("Using output directory " + exportDir);
}
final String backup = properties.getProperty(BACKUP_PROP_NAME, "no");
createBackup = backup.equalsIgnoreCase("YES");
final String zip = properties.getProperty(ZIP_PROP_NAME, "yes");
createZip = zip.equalsIgnoreCase("YES");
final String inc = properties.getProperty(INCREMENTAL_PROP_NAME, "no");
incremental = inc.equalsIgnoreCase("YES");
final String incCheck = properties.getProperty(INCREMENTAL_CHECK_PROP_NAME, "yes");
incrementalCheck = incCheck.equalsIgnoreCase("YES");
final String max = properties.getProperty(MAX_PROP_NAME, "5");
try {
maxInc = Integer.parseInt(max);
} catch (final NumberFormatException e) {
throw new EXistException("Parameter 'max' has to be an integer");
}
final String check = properties.getProperty(CHECK_DOCS_PROP_NAME, "no");
checkDocs = check.equalsIgnoreCase("YES");
}