//Die Variable filename wird der Klasse vorher übergeben.
try {
SAXBuilder builder = new SAXBuilder( false );
Document docXml = builder.build( new File( filename ) );
Element discourse = docXml.getRootElement();
Element version = (Element) discourse.getChildren().get(0);
String v = version.getText();
System.out.println(v);
for (int i = 1; i < discourse.getChildren().size(); i++) {
//create article with title, date and text
Element article = (Element)discourse.getChildren().get(i);
Element text = (Element) article.getChild("text");
String articleText = new String("");
for (Iterator<?> j = text.getContent().iterator(); j.hasNext( ); ) {
Object o = j.next( );
if (o instanceof Text) {
articleText = articleText + ((Text)o).getText();
} else if (o instanceof CDATA) {
articleText = articleText + ((CDATA)o).getText();
} else if (o instanceof Comment) {
articleText = articleText + ((Comment)o).getText();
} else if (o instanceof ProcessingInstruction) {
articleText = articleText + (ProcessingInstruction)o;
} else if (o instanceof EntityRef) {
articleText = articleText + (EntityRef)o;
} else if (o instanceof Element) {
articleText = articleText + "<" + ((Element)o).getName() + "/>";
}
}
articleText = articleText.replaceAll("\n", "<br />");
String dateString = article.getChild("date").getText();
String title = article.getChild("title").getText();
Date date = new Date();
try {
SimpleDateFormat sdfToDate = new SimpleDateFormat("dd.MM.yyyy");
date = sdfToDate.parse(dateString);
//Hier werden dem Hauptprogramm die Variablen title, date
//und articleText übergeben. Ich habe es der Übersicht halber entfernt.
} catch (ParseException pe) {
pe.printStackTrace();
}
//create statements
Element statementsXml = (Element) article.getChild("statements");
for (int j = 0; j < statementsXml.getChildren().size(); j++) {
Element statement = (Element) statementsXml.getChildren().get(j);
int id = Integer.parseInt(statement.getChild("id").getText());
String start = statement.getChild("start").getText();
String end = statement.getChild("end").getText();
String person = statement.getChild("person").getText();
String organization = statement.getChild("organization").getText();
String category = statement.getChild("category").getText();
String agreement = statement.getChild("agreement").getText();
int startInt = Integer.valueOf( start ).intValue();
int endInt = Integer.valueOf( end ).intValue();
String selection = articleText.replaceAll("<br />", "\n");
try {
if (endInt > selection.length() + 1) {
endInt = selection.length() + 1;
System.out.println("End position of statement " + id + " was corrected.");
}
if (startInt > selection.length()) {
startInt = selection.length();
System.out.println("Start position of statement " + id + " was corrected.");
}
selection = selection.substring(startInt-1, endInt-1);
} catch (StringIndexOutOfBoundsException e) {
System.out.println("Statement text of statement " + id + " could not be identified.");
}
//An dieser Stelle werden dann die Variablen id, startInt, endInt, date, selection,
//title, person, organization, category und agreement dem Hauptprogramm übergeben.
//Ich habe es der Übersicht halber entfernt.
}
}
} catch (OutOfMemoryError ome) {
System.out.println("Out of memory.");
} catch (IOException e) {
System.out.println("Error while reading the file \"" + filename + "\": " + e.getMessage());
} catch (org.jdom.JDOMException e) {
System.out.println("Error while opening the file \"" + filename + "\": " + e.getMessage());
}
