/** Generates array of XContour from local contours and modules. Used for TTF building. */
private XContour[] toContours() {
XContour[] retval;
ArrayList<XContour> list = new ArrayList<>();
XContour[] contours = m_glyph.getBody().getContour();
for (int i = 0; i < contours.length; i++) {
EContour contour = (EContour) contours[i];
list.add(contour.toQuadratic());
} // for i
XModule[] modules = m_glyph.getBody().getModule();
for (int i = 0; i < modules.length; i++) {
EModuleInvoke module = (EModuleInvoke) modules[i];
// push and pop happens inside toContour
list.add(module.toContour(new AffineTransform()));
} // for i
if (list.size() == 0) return null;
retval = new XContour[list.size()];
for (int i = 0; i < list.size(); i++) {
retval[i] = list.get(i);
} // for i
return retval;
}
public static void parseDocumentFragment(Reader reader, XMLReceiver xmlReceiver)
throws SAXException {
try {
final XMLReader xmlReader = newSAXParser(XMLUtils.ParserConfiguration.PLAIN).getXMLReader();
xmlReader.setContentHandler(new XMLFragmentReceiver(xmlReceiver));
final ArrayList<Reader> readers = new ArrayList<Reader>(3);
readers.add(new StringReader("<root>"));
readers.add(reader);
readers.add(new StringReader("</root>"));
xmlReader.parse(new InputSource(new SequenceReader(readers.iterator())));
} catch (IOException e) {
throw new OXFException(e);
}
}
public ArrayList<String> parseXML() throws Exception {
ArrayList<String> ret = new ArrayList<String>();
handshake();
URL url =
new URL(
"http://mangaonweb.com/page.do?cdn="
+ cdn
+ "&cpn=book.xml&crcod="
+ crcod
+ "&rid="
+ (int) (Math.random() * 10000));
String page = DownloaderUtils.getPage(url.toString(), "UTF-8", cookies);
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
DocumentBuilder builder = factory.newDocumentBuilder();
InputSource is = new InputSource(new StringReader(page));
Document d = builder.parse(is);
Element doc = d.getDocumentElement();
NodeList pages = doc.getElementsByTagName("page");
total = pages.getLength();
for (int i = 0; i < pages.getLength(); i++) {
Element e = (Element) pages.item(i);
ret.add(e.getAttribute("path"));
}
return (ret);
}
public TTGlyph toSimpleGlyph() {
// convert the file into array of contours
XContour[] contours = toContours();
if ((contours == null) && (!isRequiredGlyph())) {
return null;
} // if
TTGlyph retval = new TTGlyph();
retval.setSimple(true);
retval.setAdvanceWidth(getAdvanceWidth());
if (contours == null) {
return retval;
} // if
ArrayList<EContourPoint> points = new ArrayList<>();
for (int i = 0; i < contours.length; i++) {
XContour contour = contours[i];
XContourPoint[] contourPoints = contour.getContourPoint();
for (int j = 0; j < contourPoints.length; j++) {
points.add((EContourPoint) contourPoints[j]);
} // for j
retval.addEndPoint(points.size() - 1);
} // for i
for (EContourPoint point : points) {
loadContourPoint(retval, point);
} // for point
boolean hasGridfit = false;
// I need int i here.
for (int i = 0; i < points.size(); i++) {
EContourPoint point = points.get(i);
if (!point.isRounded()) {
continue;
} // if
hasGridfit = true;
loadGridfit(retval, point, i);
} // for i
if (hasGridfit) {
retval.addInstruction(TTGlyph.IUP1);
retval.addInstruction(TTGlyph.IUP0);
} // if
// I need int i here.
for (int i = 0; i < points.size(); i++) {
EContourPoint point = points.get(i);
if (point.getHint().length == 0) {
continue;
} // if
loadHint(retval, point, i);
} // for i
return retval;
}
private QueryResult gatherResultInfoForSelectQuery(
String queryString, int queryNr, boolean sorted, Document doc, String[] rows) {
Element root = doc.getRootElement();
// Get head information
Element child =
root.getChild("head", Namespace.getNamespace("http://www.w3.org/2005/sparql-results#"));
// Get result rows (<head>)
List headChildren =
child.getChildren(
"variable", Namespace.getNamespace("http://www.w3.org/2005/sparql-results#"));
Iterator it = headChildren.iterator();
ArrayList<String> headList = new ArrayList<String>();
while (it.hasNext()) {
headList.add(((Element) it.next()).getAttributeValue("name"));
}
List resultChildren =
root.getChild("results", Namespace.getNamespace("http://www.w3.org/2005/sparql-results#"))
.getChildren(
"result", Namespace.getNamespace("http://www.w3.org/2005/sparql-results#"));
int nrResults = resultChildren.size();
QueryResult queryResult = new QueryResult(queryNr, queryString, nrResults, sorted, headList);
it = resultChildren.iterator();
while (it.hasNext()) {
Element resultElement = (Element) it.next();
String result = "";
// get the row values and paste it together to one String
for (int i = 0; i < rows.length; i++) {
List bindings =
resultElement.getChildren(
"binding", Namespace.getNamespace("http://www.w3.org/2005/sparql-results#"));
String rowName = rows[i];
for (int j = 0; j < bindings.size(); j++) {
Element binding = (Element) bindings.get(j);
if (binding.getAttributeValue("name").equals(rowName))
if (result.equals(""))
result +=
rowName + ": " + ((Element) binding.getChildren().get(0)).getTextNormalize();
else
result +=
"\n"
+ rowName
+ ": "
+ ((Element) binding.getChildren().get(0)).getTextNormalize();
}
}
queryResult.addResult(result);
}
return queryResult;
}
public void endElement(String uri, String sName, String qName) {
if (qName.equals("PROPSTABLE")) {
} else if (qName.equals("PROPSROW")) {
if (PROPSROW_DESC == null
|| (curPROPSROW_DESC != null && curPROPSROW_DESC.equals(PROPSROW_DESC)))
propsAl.add(curProps);
} else {
// LogUtil.fine("/"+qName);
textFlag = false;
}
}
/**
* Populates LOCALES list with contents of xml.
*
* @param list the configuration list
*/
private static void parseLocales(NodeList list) {
for (int i = 0; i < list.getLength(); ++i) {
Node node = list.item(i);
NamedNodeMap attributes = node.getAttributes();
String label = ((Attr) attributes.getNamedItem("label")).getValue();
String code = ((Attr) attributes.getNamedItem("isoCode")).getValue();
String dictLocation = ((Attr) attributes.getNamedItem("dictionaryUrl")).getValue();
try {
LOCALES.add(new Locale(label, code, new URL(dictLocation)));
} catch (MalformedURLException exc) {
logger.warn(
"Unable to parse dictionary location of " + label + " (" + dictLocation + ")", exc);
}
}
}
// Add/Insert
public int add(Script script) {
// Find the correct spot to add it alphabetically
int i, limit;
for (i = 0, limit = scripts.size(); i < limit; i++) {
Script scriptTemp = (Script) scripts.get(i);
if (scriptTemp.getName().compareTo(script.getName()) >= 0) {
break;
}
}
scripts.add(i, script);
// Update the table
fireTableRowsInserted(i, i);
return i;
}
public void endElement(String uri, String localName, String qname) {
super.endElement(uri, localName, qname);
if (record != null && localName.equals(DataRecord.ENCODING_RECORD)) {
// end of record
if (dataImport.importRecord(record, fieldsInImport)) {
count++;
}
record = null;
fieldsInImport = null;
}
String fieldValue = getFieldValue();
if (record != null && fieldValue != null) {
// end of field
try {
if (textImport) {
if (!record.setFromText(fieldName, fieldValue.trim())) {
dataImport.logImportWarning(
record,
"Value '"
+ fieldValue
+ "' for Field '"
+ fieldName
+ "' corrected to '"
+ record.getAsText(fieldName)
+ "'");
}
} else {
record.set(fieldName, fieldValue.trim());
}
String[] equivFields = record.getEquivalentFields(fieldName);
for (String f : equivFields) {
fieldsInImport.add(f);
}
} catch (Exception esetvalue) {
dataImport.logImportWarning(
record,
"Cannot set value '"
+ fieldValue
+ "' for Field '"
+ fieldName
+ "': "
+ esetvalue.toString());
}
}
}
/**
* this is fired when a tag start event is found on an xml document
*
* @param uri namespace for tag being processed
* @param localName tag name
* @param qName fully qualified name for tag
* @param attributes tag attributes
* @throws SAXException if parsing fails
*/
@Override
public void startElement(String uri, String localName, String qName, Attributes attributes)
throws SAXException {
// if a class is being parsed
if (qName.equalsIgnoreCase(getTagName())) {
// set class name from xml attributes
mClassName = attributes.getValue(TAG_ATTRIBUTES.name.toString());
// if an attribute is being parsed
} else if (qName.equalsIgnoreCase(AttributeDescriptor.TAG_NAME)) {
// create an attribute descriptor
AttributeDescriptor attribute = new AttributeDescriptor();
// add an attribute to this class
mAttributes.add(attribute);
// let the attribute parse itself
mCtx.pushHandler(attribute);
// forward event to new handler
attribute.startElement(uri, localName, qName, attributes);
}
}
public int runCsvImport() {
int count = 0;
try {
int linecnt = 0;
String[] header = null;
ArrayList<String> fieldsInImport = new ArrayList<String>();
BufferedReader f =
new BufferedReader(new InputStreamReader(new FileInputStream(filename), encoding));
String s;
DataRecord dummyRecord = storageObject.createNewRecord();
while ((s = f.readLine()) != null) {
s = s.trim();
if (s.length() == 0) {
continue;
}
Vector<String> fields = splitFields(s);
if (fields.size() > 0) {
if (linecnt == 0) {
// header
header = new String[fields.size()];
for (int i = 0; i < fields.size(); i++) {
header[i] = fields.get(i);
if (header[i].startsWith("#") && header[i].endsWith("#") && header.length > 2) {
header[i] = header[i].substring(1, header[i].length() - 1).trim();
overrideKeyField = header[i];
}
String[] equivFields = dummyRecord.getEquivalentFields(header[i]);
for (String ef : equivFields) {
fieldsInImport.add(ef);
}
}
} else {
// fields
DataRecord r = storageObject.createNewRecord();
for (int i = 0; i < header.length; i++) {
String value = (fields.size() > i ? fields.get(i) : null);
if (value != null && value.length() > 0) {
try {
if (!r.setFromText(header[i], value.trim())) {
logImportWarning(
r,
"Value '"
+ value
+ "' for Field '"
+ header[i]
+ "' corrected to '"
+ r.getAsText(header[i])
+ "'");
}
} catch (Exception esetvalue) {
logImportWarning(
r,
"Cannot set value '"
+ value
+ "' for Field '"
+ header[i]
+ "': "
+ esetvalue.toString());
}
}
}
if (importRecord(r, fieldsInImport)) {
count++;
}
}
}
linecnt++;
}
f.close();
} catch (Exception e) {
logInfo(e.toString());
errorCount++;
Logger.log(e);
if (Daten.isGuiAppl()) {
Dialog.error(e.toString());
}
}
return count;
}
public void endElement(java.lang.String uri, java.lang.String localName, java.lang.String qName)
throws SAXException {
if (qName.equals("date")) {
accumulator.setLength(0);
} else if (qName.equals("ref") || qName.equals("bibl")) {
String refString = getRefText();
refString = refString.replace("\n", " ");
refString = refString.replace("\t", " ");
refString = refString.replace(" ", " ");
if (npl && ref) {
if (referencesNPL == null) referencesNPL = new ArrayList<String>();
referencesNPL.add(refString);
refFound = true;
if (nplReferences) nbNPLRef++;
} else if (ref) {
if (referencesPatent == null) {
referencesPatent = new HashMap<String, ArrayList<String>>();
}
ArrayList<String> refss = referencesPatent.get(currentFileName);
if (refss == null) {
refss = new ArrayList<String>();
}
refss.add(refString);
referencesPatent.put(currentFileName, refss);
refFound = true;
if (patentReferences) {
nbPatentRef++;
}
}
if (refFound) {
// we tokenize the text
// ArrayList<String> tokens = TextUtilities.segment(refString,
// "[("+TextUtilities.punctuations);
// StringTokenizer st = new StringTokenizer(refString, delimiters, true);
List<String> tokenizations = new ArrayList<String>();
try {
// TBD: pass a language object to the tokenize method call
tokenizations = analyzer.tokenize(refString);
} catch (Exception e) {
LOGGER.debug("Tokenization for XML patent document has failed.");
}
int i = 0;
// String token = null;
// for(String token : tokens) {
// while (st.hasMoreTokens()) {
for (String token : tokenizations) {
// token = st.nextToken().trim();
if ((token.trim().length() == 0)
|| (token.equals(" "))
|| (token.equals("\t"))
|| (token.equals("\n"))
|| (token.equals("\r"))) {
continue;
}
try {
accumulatedText.append(token + "\t");
allContent.append(token + " ");
if (npl) {
if (nplReferences) {
if (i == 0) {
// accumulatedText.append("refNPLBegin\n");
accumulatedText.append("I-<refNPL>\n");
} else if (token == null) {
// accumulatedText.append("refNPLEnd\n");
accumulatedText.append("E-<refNPL>\n");
} else {
accumulatedText.append("<refNPL>\n");
}
} else accumulatedText.append("<other>\n");
} else {
if (patentReferences) {
if (i == 0) accumulatedText.append("I-<refPatent>\n");
else if (token == null) accumulatedText.append("E-<refPatent>\n");
else accumulatedText.append("<refPatent>\n");
} else accumulatedText.append("<other>\n");
}
} catch (Exception e) {
// e.printStackTrace();
throw new GrobidException("An exception occured while running Grobid.", e);
}
i++;
}
}
ref = false;
} else if (qName.equals("classification-ipcr")) {
accumulator.setLength(0);
} else if (qName.equals("classification-symbol")) {
accumulator.setLength(0);
} else if (qName.equals("abstract")) {
accumulator.setLength(0);
} else if (qName.equals("heading")) {
accumulator.append(" ");
} else if (qName.equals("description")) {
if (refFound) {
String content = getText();
// we tokenize the text
// ArrayList<String> tokens = TextUtilities.segment(content,
// "[("+TextUtilities.punctuations);
// StringTokenizer st = new StringTokenizer(content, delimiters, true);
List<String> tokenizations = new ArrayList<String>();
try {
// TBD: pass a language object to the tokenize method call
tokenizations = analyzer.tokenize(content);
} catch (Exception e) {
LOGGER.debug("Tokenization for XML patent document has failed.");
}
int i = 0;
// String token = null;
// for(String token : tokens) {
// while (st.hasMoreTokens()) {
for (String token : tokenizations) {
// token = st.nextToken().trim();
if ((token.trim().length() == 0)
|| (token.equals(" "))
|| (token.equals("\t"))
|| (token.equals("\n"))
|| (token.equals("\r"))) {
continue;
}
// we print only a window of N words
if ((i > N) && (N != -1)) {
// break;
token = token.trim();
if (token.length() > 0) {
accumulatedText.append(token + "\t" + "<ignore>\n");
allContent.append(token + " ");
}
} else {
try {
token = token.trim();
if (token.length() > 0) {
accumulatedText.append(token + "\t" + "<other>\n");
allContent.append(token + " ");
}
} catch (Exception e) {
throw new GrobidException("An exception occured while running Grobid.", e);
}
}
i++;
}
accumulator.setLength(0);
refFound = false;
}
} else if (qName.equals("patcit")) {
// we register the citation, the citation context will be marked in a later stage
if (citations == null) citations = new ArrayList<String>();
citations.add(cited_number);
accumulator.setLength(0);
} else if (qName.equals("invention-title")) {
accumulator.setLength(0);
} else if (qName.equals("applicants")) {
accumulator.setLength(0);
} else if (qName.equals("inventors")) {
accumulator.setLength(0);
} else if (qName.equals("document-id")) {
accumulator.setLength(0);
} else if (qName.equals("legal-status")) {
accumulator.setLength(0);
} else if (qName.equals("bibliographic-data")) {
accumulator.setLength(0);
} else if (qName.equals("doc-number")) {
accumulator.setLength(0);
} else if (qName.equals("country")) {
accumulator.setLength(0);
} else if (qName.equals("kind")) {
accumulator.setLength(0);
} else if (qName.equals("classification-symbol")) {
accumulator.setLength(0);
} else if (qName.equals("classification-ecla")) {
accumulator.setLength(0);
} else if (qName.equals("patent-document") || qName.equals("fulltext-document")) {
String allString = allContent.toString();
journalsPositions = lexicon.inJournalNames(allString);
abbrevJournalsPositions = lexicon.inAbbrevJournalNames(allString);
conferencesPositions = lexicon.inConferenceNames(allString);
publishersPositions = lexicon.inPublisherNames(allString);
allContent = null;
allString = null;
} else if (qName.equals("row")) {
accumulator.append(" ");
} else if (qName.equals("p")) {
accumulator.append("\n");
}
}
public void startElement(String namespaceURI, String localName, String qName, Attributes atts)
throws SAXException {
if (qName.equals("PAGE")) {
int length = atts.getLength();
currentPage++;
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("id")) {;
} else if (name.equals("number")) {
} else if (name.equals("width")) {
} else if (name.equals("height")) {
}
}
}
/*
* if (block != null) { blabla.append("\n");
* tokenizations.add("\n"); block.setText(blabla.toString());
* block.setNbTokens(nbTokens); doc.addBlock(block); } Block block0
* = new Block(); block0.setText("@PAGE\n"); block0.setNbTokens(0);
* doc.addBlock(block0);
*/
/*
* block = new Block(); blabla = new StringBuffer(); nbTokens = 0;
* //blabla.append("\n@block\n"); tokenizations.add("\n");
*/
} else if (qName.equals("BLOCK")) {
block = new Block();
blabla = new StringBuffer();
nbTokens = 0;
block.setPage(currentPage);
// blabla.append("\n@block\n");
} else if (qName.equals("IMAGE")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("href")) {
// if (images == null)
// images = new ArrayList<String>();
images.add(value);
} else if (name.equals("x")) {
double x = Double.parseDouble(value);
if (x != currentX) {
currentX = x;
}
} else if (name.equals("y")) {
double y = Double.parseDouble(value);
if (y != currentY) {
currentY = y;
}
} else if (name.equals("width")) {
double width = Double.parseDouble(value);
if (width != currentWidth) {
currentWidth = width;
}
} else if (name.equals("height")) {
double height = Double.parseDouble(value);
if (height != currentHeight) {
currentHeight = height;
}
}
}
}
} else if (qName.equals("TEXT")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("id")) {
} else if (name.equals("x")) {
} else if (name.equals("y")) {
} else if (name.equals("width")) {
} else if (name.equals("height")) {
}
}
}
} else if (qName.equals("TOKEN")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("id")) {;
} else if (name.equals("font-name")) {
if (!value.equals(currentFont)) {
currentFont = value;
blabla.append(" ");
}
} else if (name.equals("font-size")) {
double fontSize = Double.parseDouble(value);
if (fontSize != currentFontSize) {
currentFontSize = fontSize;
blabla.append(" ");
}
} else if (name.equals("bold")) {
if (value.equals("yes")) {
currentBold = true;
} else {
currentBold = false;
}
} else if (name.equals("italic")) {
if (value.equals("yes")) {
currentItalic = true;
} else {
currentItalic = false;
}
} else if (name.equals("font-color")) {
if (!value.equals(colorFont)) {
colorFont = value;
}
} else if (name.equals("rotation")) {
if (value.equals("0")) currentRotation = false;
else currentRotation = true;
} else if (name.equals("x")) {
double x = Double.parseDouble(value);
if (x != currentX) {
currentX = x;
}
} else if (name.equals("y")) {
double y = Double.parseDouble(value);
if (y != currentY) {
currentY = y;
}
} else if (name.equals("base")) {
double base = Double.parseDouble(value);
} else if (name.equals("width")) {
double width = Double.parseDouble(value);
if (width != currentWidth) {
currentWidth = width;
}
} else if (name.equals("height")) {
double height = Double.parseDouble(value);
if (height != currentHeight) {
currentHeight = height;
}
}
}
}
} else if (qName.equals("xi:include")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("href")) {
// if (images == null)
// images = new ArrayList<String>();
images.add(value);
}
}
}
}
// accumulator.setLength(0);
}
public void endElement(java.lang.String uri, java.lang.String localName, java.lang.String qName)
throws SAXException {
// if (!qName.equals("TOKEN") && !qName.equals("BLOCK") &&
// !qName.equals("TEXT"))
// System.out.println(qName);
if (qName.equals("TEXT")) {
blabla.append("\n");
LayoutToken token = new LayoutToken();
token.setText("\n");
block.addToken(token);
accumulator.setLength(0);
tokenizations.add("\n");
} else if (qName.equals("METADATA")) {
accumulator.setLength(0);
} else if (qName.equals("TOKEN")) {
String tok0 = TextUtilities.clean(getText());
if (block.getStartToken() == -1) {
block.setStartToken(tokenizations.size());
}
if (tok0.length() > 0) {
StringTokenizer st = new StringTokenizer(tok0, TextUtilities.fullPunctuations, true);
boolean diaresis = false;
boolean accent = false;
boolean keepLast = false;
while (st.hasMoreTokens()) {
diaresis = false;
accent = false;
keepLast = false;
String tok = st.nextToken();
if (tok.length() > 0) {
LayoutToken token = new LayoutToken();
if ((previousToken != null)
&& (tok != null)
&& (previousToken.length() > 0)
&& (tok.length() > 0)
&& blabla.length() > 0) {
Character leftChar = previousTok.getText().charAt(previousTok.getText().length() - 1);
Character rightChar = tok.charAt(0);
ModifierClass leftClass = classifyChar(leftChar);
ModifierClass rightClass = classifyChar(rightChar);
ModifierClass modifierClass = ModifierClass.NOT_A_MODIFIER;
if (leftClass != ModifierClass.NOT_A_MODIFIER
|| rightClass != ModifierClass.NOT_A_MODIFIER) {
Character baseChar = null;
Character modifierChar = null;
if (leftClass != ModifierClass.NOT_A_MODIFIER) {
if (rightClass != ModifierClass.NOT_A_MODIFIER) {
// assert false;
// keeping characters, but setting class
// to not a modifier
baseChar = leftChar;
modifierChar = rightChar;
modifierClass = ModifierClass.NOT_A_MODIFIER;
} else {
baseChar = rightChar;
modifierChar = leftChar;
modifierClass = leftClass;
}
} else {
baseChar = leftChar;
modifierChar = rightChar;
modifierClass = rightClass;
}
String updatedChar = modifyCharacter(baseChar, modifierChar);
tokenizations.remove(tokenizations.size() - 1);
if (tokenizations.size() > 0) {
tokenizations.remove(tokenizations.size() - 1);
}
blabla.deleteCharAt(blabla.length() - 1);
if (blabla.length() > 0) {
blabla.deleteCharAt(blabla.length() - 1);
}
removeLastCharacterIfPresent(previousTok);
if (updatedChar != null) {
blabla.append(updatedChar);
previousTok.setText(previousTok.getText() + updatedChar);
}
blabla.append(tok.substring(1, tok.length()));
previousTok.setText(previousTok.getText() + tok.substring(1, tok.length()));
tokenizations.add(previousTok.getText());
diaresis =
(modifierClass == ModifierClass.DIAERESIS
|| modifierClass == ModifierClass.NORDIC_RING
|| modifierClass == ModifierClass.CZECH_CARON
|| modifierClass == ModifierClass.TILDE
|| modifierClass == ModifierClass.CEDILLA);
accent =
(modifierClass == ModifierClass.ACUTE_ACCENT
|| modifierClass == ModifierClass.CIRCUMFLEX
|| modifierClass == ModifierClass.GRAVE_ACCENT);
if (rightClass != ModifierClass.NOT_A_MODIFIER) {
tok = ""; // resetting current token as it
// is a single-item
}
}
}
if (tok != null) {
// actually in certain cases, the extracted string under token can be a chunk of text
// with separators that need to be preserved
// tok = tok.replace(" ", "");
}
if ((!diaresis) && (!accent)) {
// blabla.append(" ");
blabla.append(tok);
token.setText(tok);
tokenizations.add(tok);
} else {
tok = "";
keepLast = true;
}
/*
* StringTokenizer st0 = new StringTokenizer(tok0,
* TextUtilities.fullPunctuations, true);
* while(st0.hasMoreTokens()) { String tok =
* st0.nextToken(); tokenizations.add(tok); }
* tokenizations.add(" ");
*/
/*
* boolean punct1 = false; boolean punct2 = false;
* boolean punct3 = false; String content = null; int i
* = 0; for(; i<TextUtilities.punctuations.length();
* i++) { if (tok.length() > 0) { if
* (tok.charAt(tok.length()-1) ==
* TextUtilities.punctuations.charAt(i)) { punct1 =
* true; content = tok.substring(0, tok.length()-1); if
* (tok.length() > 1) { int j = 0; for(;
* j<TextUtilities.punctuations.length(); j++) { if
* (tok.charAt(tok.length()-2) ==
* TextUtilities.punctuations.charAt(j)) { punct3 =
* true; content = tok.substring(0, tok.length()-2); } }
* } break; } } } if (tok.length() > 0) { if (
* (tok.startsWith("(")) && (tok.length() > 1) ) { if
* ((punct3) && (tok.length() > 2)) content =
* tok.substring(1, tok.length()-2); else if (punct1)
* content = tok.substring(1, tok.length()-1); else
* content = tok.substring(1, tok.length()); punct2 =
* true; token.setText("("); } else if (
* (tok.startsWith("[")) && (tok.length() > 1) ) { if
* ((punct3) && (tok.length() > 2)) content =
* tok.substring(1, tok.length()-2); else if (punct1)
* content = tok.substring(1, tok.length()-1); else
* content = tok.substring(1, tok.length()); punct2 =
* true; token.setText("["); } else if (
* (tok.startsWith("\"")) && (tok.length() > 1) ) { if
* ((punct3) && (tok.length() > 2)) content =
* tok.substring(1, tok.length()-2); else if (punct1)
* content = tok.substring(1, tok.length()-1); else
* content = tok.substring(1, tok.length()); punct2 =
* true; token.setText("\""); } }
*/
if (currentRotation) currentFontSize = currentFontSize / 2;
/*
* if (punct2) { if (currentFont != null)
* token.setFont(currentFont.toLowerCase()); else
* token.setFont("default");
* token.setItalic(currentItalic);
* token.setBold(currentBold);
* token.setRotation(currentRotation);
* token.setColorFont(colorFont); token.setX(currentX);
* token.setY(currentY); token.setWidth(currentWidth);
* token.setHeight(currentHeight);
* token.setFontSize(currentFontSize);
* block.addToken(token);
*
* token = new LayoutToken(); token.setText(content); }
* if (punct1) { token.setText(content); if (currentFont
* != null) token.setFont(currentFont.toLowerCase());
* else token.setFont("default");
* token.setItalic(currentItalic);
* token.setBold(currentBold);
* token.setRotation(currentRotation);
* token.setColorFont(colorFont); token.setX(currentX);
* token.setY(currentY); token.setWidth(currentWidth);
* token.setHeight(currentHeight);
* token.setFontSize(currentFontSize);
* block.addToken(token);
*
* if (punct3) { token = new LayoutToken();
* token.setText(""+tok.charAt(tok.length()-2)); if
* (currentFont != null)
* token.setFont(currentFont.toLowerCase()); else
* token.setFont("default");
* token.setItalic(currentItalic);
* token.setBold(currentBold);
* token.setRotation(currentRotation);
* token.setColorFont(colorFont); token.setX(currentX);
* token.setY(currentY); token.setWidth(currentWidth);
* token.setHeight(currentHeight);
* token.setFontSize(currentFontSize);
* block.addToken(token); }
*
* token = new LayoutToken();
* token.setText(""+tok.charAt(tok.length()-1)); }
*/
if (currentFont != null) token.setFont(currentFont.toLowerCase());
else token.setFont("default");
token.setItalic(currentItalic);
token.setBold(currentBold);
token.setRotation(currentRotation);
token.setColorFont(colorFont);
token.setX(currentX);
token.setY(currentY);
token.setWidth(currentWidth);
token.setHeight(currentHeight);
token.setFontSize(currentFontSize);
if (!diaresis && !accent) {
block.addToken(token);
}
if (block.getFont() == null) {
if (currentFont != null) block.setFont(currentFont.toLowerCase());
else token.setFont("default");
}
if (nbTokens == 0) {
block.setItalic(currentItalic);
block.setBold(currentBold);
}
if (block.getColorFont() == null) block.setColorFont(colorFont);
if (block.getX() == 0.0) block.setX(currentX);
if (block.getY() == 0.0) block.setY(currentY);
if (block.getWidth() == 0.0) block.setWidth(currentWidth);
if (block.getHeight() == 0.0) block.setHeight(currentHeight);
if (block.getFontSize() == 0.0) block.setFontSize(currentFontSize);
if (!diaresis && !accent) {
previousToken = tok;
previousTok = token;
} else {
previousToken = previousTok.getText();
}
nbTokens++;
accumulator.setLength(0);
}
}
if (tokenizations.size() > 0) {
String justBefore = tokenizations.get(tokenizations.size() - 1);
if (!justBefore.endsWith("-")) {
tokenizations.add(" ");
blabla.append(" ");
}
}
}
block.setEndToken(tokenizations.size());
} else if (qName.equals("PAGE")) {
// page marker are usefull to detect headers (same first line(s)
// appearing on each page)
if (block != null) {
blabla.append("\n");
tokenizations.add("\n");
block.setText(blabla.toString());
block.setNbTokens(nbTokens);
doc.addBlock(block);
}
Block block0 = new Block();
block0.setText("@PAGE\n");
block0.setNbTokens(0);
block0.setPage(currentPage);
doc.addBlock(block0);
block = new Block();
block.setPage(currentPage);
blabla = new StringBuffer();
nbTokens = 0;
// blabla.append("\n@block\n");
tokenizations.add("\n");
} else if (qName.equals("IMAGE")) {
if (block != null) {
blabla.append("\n");
block.setText(blabla.toString());
block.setNbTokens(nbTokens);
doc.addBlock(block);
}
block = new Block();
block.setPage(currentPage);
blabla = new StringBuffer();
if (images.size() > 0) {
blabla.append("@IMAGE " + images.get(images.size() - 1) + "\n");
}
block.setText(blabla.toString());
block.setNbTokens(nbTokens);
if (block.getX() == 0.0) block.setX(currentX);
if (block.getY() == 0.0) block.setY(currentY);
if (block.getWidth() == 0.0) block.setWidth(currentWidth);
if (block.getHeight() == 0.0) block.setHeight(currentHeight);
doc.addBlock(block);
blabla = new StringBuffer();
nbTokens = 0;
block = new Block();
block.setPage(currentPage);
}
/*
* else if (qName.equals("VECTORIALIMAGES")) { if (block != null) {
* blabla.append("\n"); block.setText(blabla.toString());
* block.setNbTokens(nbTokens); doc.addBlock(block); } block = new
* Block(); block.setPage(currentPage); blabla = new StringBuffer();
* blabla.append("@IMAGE " + "vectorial \n");
* block.setText(blabla.toString()); block.setNbTokens(nbTokens); if
* (block.getX() == 0.0) block.setX(currentX); if (block.getY() == 0.0)
* block.setY(currentY); if (block.getWidth() == 0.0)
* block.setWidth(currentWidth); if (block.getHeight() == 0.0)
* block.setHeight(currentHeight); doc.addBlock(block); blabla = new
* StringBuffer(); nbTokens = 0; block = new Block();
* block.setPage(currentPage); }
*/
else if (qName.equals("BLOCK")) {
blabla.append("\n");
tokenizations.add("\n");
block.setText(blabla.toString());
block.setNbTokens(nbTokens);
block.setWidth(currentX - block.getX() + currentWidth);
block.setHeight(currentY - block.getY() + currentHeight);
doc.addBlock(block);
// blabla = new StringBuffer();
nbTokens = 0;
block = null;
} else if (qName.equals("xi:include")) {
if (block != null) {
blabla.append("\n");
block.setText(blabla.toString());
block.setNbTokens(nbTokens);
doc.addBlock(block);
}
block = new Block();
block.setPage(currentPage);
blabla = new StringBuffer();
blabla.append("@IMAGE " + images.get(images.size() - 1) + "\n");
block.setText(blabla.toString());
block.setNbTokens(nbTokens);
doc.addBlock(block);
blabla = new StringBuffer();
nbTokens = 0;
block = new Block();
block.setPage(currentPage);
}
/*
* else if (qName.equals("DOCUMENT")) {
* System.out.println(blabla.toString()); }
*/
}