/** External TMX - level 1. Skip text inside inline tags. */
protected void parseSegExtLevel1() throws Exception {
segContent.setLength(0);
int inlineLevel = 0;
while (true) {
XMLEvent e = xml.nextEvent();
switch (e.getEventType()) {
case XMLEvent.START_ELEMENT:
inlineLevel++;
break;
case XMLEvent.END_ELEMENT:
inlineLevel--;
EndElement eEnd = (EndElement) e;
if ("seg".equals(eEnd.getName().getLocalPart())) {
return;
}
break;
case XMLEvent.CHARACTERS:
if (inlineLevel == 0) {
Characters c = (Characters) e;
segContent.append(c.getData());
}
break;
}
}
}
/**
* This test was inspired by an actual bug in one of implementations: initial state was not
* properly set if nextTag() was called (instead of nextEvent()), and subsequent peek() failed.
*/
public void testPeek() throws XMLStreamException {
String XML = "<root>text</root>";
for (int i = 0; i < 4; ++i) {
boolean ns = (i & 1) != 0;
boolean coal = (i & 2) != 0;
XMLEventReader er = getReader(XML, ns, coal);
assertTokenType(START_DOCUMENT, er.nextEvent().getEventType());
XMLEvent tag = er.nextTag();
assertTokenType(START_ELEMENT, tag.getEventType());
// Now, peek() should produce text..
XMLEvent text = er.peek();
assertTokenType(CHARACTERS, text.getEventType());
Characters chars = text.asCharacters();
assertNotNull(chars);
assertEquals("text", chars.getData());
// and need nextEvent() to get rid of it, too:
text = er.nextEvent();
// Let's verify it again:
assertTokenType(CHARACTERS, text.getEventType());
chars = text.asCharacters();
assertNotNull(chars);
assertEquals("text", chars.getData());
assertTokenType(END_ELEMENT, er.nextTag().getEventType());
assertTokenType(END_DOCUMENT, er.nextEvent().getEventType());
// And at the end, peek() should return null
assertNull(er.peek());
}
}
public DataMetadata parse(XMLEventReader xmlEventReader, StartElement start)
throws GraphIOException {
try {
// Create the new port.
DataMetadata data = new DataMetadata();
// Parse the attributes.
@SuppressWarnings("unchecked")
Iterator<Attribute> iterator = start.getAttributes();
while (iterator.hasNext()) {
Attribute attribute = (Attribute) iterator.next();
String name = attribute.getName().getLocalPart();
String value = attribute.getValue();
if (data.getKey() == null && GraphMLConstants.KEY_NAME.equals(name)) {
data.setKey(value);
}
}
// Make sure the key has been set.
if (data.getKey() == null) {
throw new GraphIOException("Element 'data' is missing attribute 'key'");
}
while (xmlEventReader.hasNext()) {
XMLEvent event = xmlEventReader.nextEvent();
if (event.isStartElement()) {
StartElement element = (StartElement) event;
// Treat any child elements as unknown
getUnknownParser().parse(xmlEventReader, element);
}
if (event.isCharacters()) {
Characters characters = (Characters) event;
data.setValue(characters.getData());
}
if (event.isEndElement()) {
EndElement end = (EndElement) event;
verifyMatch(start, end);
break;
}
}
return data;
} catch (Exception e) {
ExceptionConverter.convert(e);
}
return null;
}
/** OmegaT TMX - just read full text. */
protected void parseSegOmegaT() throws Exception {
segContent.setLength(0);
while (true) {
XMLEvent e = xml.nextEvent();
switch (e.getEventType()) {
case XMLEvent.END_ELEMENT:
EndElement eEnd = (EndElement) e;
if ("seg".equals(eEnd.getName().getLocalPart())) {
return;
}
break;
case XMLEvent.CHARACTERS:
Characters c = (Characters) e;
segContent.append(c.getData());
break;
}
}
}
protected void parseNote(StartElement element) throws Exception {
noteContent.setLength(0);
while (true) {
XMLEvent e = xml.nextEvent();
switch (e.getEventType()) {
case XMLEvent.END_ELEMENT:
EndElement eEnd = (EndElement) e;
if ("note".equals(eEnd.getName().getLocalPart())) {
currentTu.note = noteContent.toString();
return;
}
break;
case XMLEvent.CHARACTERS:
Characters c = (Characters) e;
noteContent.append(c.getData());
break;
}
}
}
protected void parseProp(StartElement element) throws Exception {
String propType = getAttributeValue(element, "type");
propContent.setLength(0);
while (true) {
XMLEvent e = xml.nextEvent();
switch (e.getEventType()) {
case XMLEvent.END_ELEMENT:
EndElement eEnd = (EndElement) e;
if ("prop".equals(eEnd.getName().getLocalPart())) {
currentTu.props.add(new TMXProp(propType, propContent.toString()));
return;
}
break;
case XMLEvent.CHARACTERS:
Characters c = (Characters) e;
propContent.append(c.getData());
break;
}
}
}
private boolean isEqual(Characters e1, Characters e2) {
if (bopt) return isEqual(e1.getData().trim(), e2.getData().trim());
else return isEqual(e1.getData(), e2.getData());
}
private boolean isBlank(Characters e) {
if (e.isIgnorableWhiteSpace()) return true;
return Util.isBlank(e.getData());
}
/** External TMX - level 2. Replace all tags into shortcuts. */
protected void parseSegExtLevel2() throws Exception {
segContent.setLength(0);
segInlineTag.setLength(0);
inlineTagHandler.reset();
int inlineLevel = 0;
while (true) {
XMLEvent e = xml.nextEvent();
switch (e.getEventType()) {
case XMLEvent.START_ELEMENT:
StartElement eStart = e.asStartElement();
if ("hi".equals(eStart.getName().getLocalPart())) {
// tag should be skipped
break;
}
inlineLevel++;
segInlineTag.setLength(0);
if ("bpt".equals(eStart.getName().getLocalPart())) {
inlineTagHandler.startBPT(
getAttributeValue(eStart, "i"), getAttributeValue(eStart, "x"));
inlineTagHandler.setTagShortcutLetter(
StringUtil.getFirstLetterLowercase(getAttributeValue(eStart, "type")));
} else if ("ept".equals(eStart.getName().getLocalPart())) {
inlineTagHandler.startEPT(getAttributeValue(eStart, "i"));
} else if ("it".equals(eStart.getName().getLocalPart())) {
inlineTagHandler.startOTHER();
inlineTagHandler.setOtherTagShortcutLetter(
StringUtil.getFirstLetterLowercase(getAttributeValue(eStart, "type")));
inlineTagHandler.setCurrentPos(getAttributeValue(eStart, "pos"));
} else if ("ph".equals(eStart.getName().getLocalPart())) {
inlineTagHandler.startOTHER();
inlineTagHandler.setOtherTagShortcutLetter(
StringUtil.getFirstLetterLowercase(getAttributeValue(eStart, "type")));
} else {
inlineTagHandler.startOTHER();
}
break;
case XMLEvent.END_ELEMENT:
EndElement eEnd = e.asEndElement();
if ("hi".equals(eEnd.getName().getLocalPart())) {
// tag should be skipped
break;
}
inlineLevel--;
if ("seg".equals(eEnd.getName().getLocalPart())) {
return;
}
boolean slashBefore = false;
boolean slashAfter = false;
char tagName = StringUtil.getFirstLetterLowercase(segInlineTag);
Integer tagN;
if ("bpt".equals(eEnd.getName().getLocalPart())) {
if (tagName != 0) {
inlineTagHandler.setTagShortcutLetter(tagName);
} else {
tagName = inlineTagHandler.getTagShortcutLetter();
}
tagN = inlineTagHandler.endBPT();
} else if ("ept".equals(eEnd.getName().getLocalPart())) {
slashBefore = true;
tagName = inlineTagHandler.getTagShortcutLetter();
tagN = inlineTagHandler.endEPT();
} else if ("it".equals(eEnd.getName().getLocalPart())) {
if (tagName != 0) {
inlineTagHandler.setOtherTagShortcutLetter(tagName);
} else {
tagName = inlineTagHandler.getOtherTagShortcutLetter();
}
tagN = inlineTagHandler.endOTHER();
if ("end".equals(inlineTagHandler.getCurrentPos())) {
slashBefore = true;
}
} else if ("ph".equals(eEnd.getName().getLocalPart())) {
if (tagName != 0) {
inlineTagHandler.setOtherTagShortcutLetter(tagName);
} else {
tagName = inlineTagHandler.getOtherTagShortcutLetter();
}
tagN = inlineTagHandler.endOTHER();
if (useSlash) {
slashAfter = true;
}
} else {
tagN = inlineTagHandler.endOTHER();
if (useSlash) {
slashAfter = true;
}
}
if (tagName == 0) {
tagName = 'f';
}
if (tagN == null) {
// check error of TMX reading
Log.logErrorRB(
"TMX_ERROR_READING_LEVEL2",
e.getLocation().getLineNumber(),
e.getLocation().getColumnNumber());
errorsCount++;
segContent.setLength(0);
// wait for end seg
while (true) {
XMLEvent ev = xml.nextEvent();
switch (ev.getEventType()) {
case XMLEvent.END_ELEMENT:
EndElement evEnd = (EndElement) ev;
if ("seg".equals(evEnd.getName().getLocalPart())) {
return;
}
}
}
}
segContent.append('<');
if (slashBefore) {
segContent.append('/');
}
segContent.append(tagName);
segContent.append(Integer.toString(tagN));
if (slashAfter) {
segContent.append('/');
}
segContent.append('>');
break;
case XMLEvent.CHARACTERS:
Characters c = (Characters) e;
if (inlineLevel == 0) {
segContent.append(c.getData());
} else {
segInlineTag.append(c.getData());
}
break;
}
}
}
@Override
public boolean isIgnorableWhiteSpace() {
return delegate.isIgnorableWhiteSpace();
}
@Override
public boolean isCData() {
return delegate.isCData();
}
/**
* Handles a Stream event.
*
* @param event
* @throws SAXException
*/
private void handleEvent(XMLEvent event) throws SAXException {
if (event.getEventType() == XMLStreamConstants.CHARACTERS) {
Characters c = event.asCharacters();
lastContents += c.getData();
} else if (event.getEventType() == XMLStreamConstants.START_ELEMENT) {
StartElement startElement = event.asStartElement();
String tagLocalName = startElement.getName().getLocalPart();
if ("row".equals(tagLocalName)) {
Attribute rowIndex = startElement.getAttributeByName(new QName("r"));
if (firstRowIndex == -1) {
firstRowIndex = Integer.parseInt(rowIndex.getValue());
}
currentRow = new StreamingRow(Integer.parseInt(rowIndex.getValue()) - 1);
} else if ("cols".equals(tagLocalName)) {
parsingCols = true;
} else if ("col".equals(tagLocalName) && parsingCols) {
colNumber = colNumber + 1;
} else if ("c".equals(tagLocalName)) {
Attribute ref = startElement.getAttributeByName(new QName("r"));
String[] coord = ref.getValue().split("(?<=\\D)(?=\\d)|(?<=\\d)(?=\\D)");
currentCell =
new StreamingCell(
CellReference.convertColStringToIndex(coord[0]), Integer.parseInt(coord[1]) - 1);
setFormatString(startElement, currentCell);
Attribute type = startElement.getAttributeByName(new QName("t"));
if (type != null) {
currentCell.setType(type.getValue());
} else {
currentCell.setType("n");
}
Attribute style = startElement.getAttributeByName(new QName("s"));
if (style != null) {
String indexStr = style.getValue();
try {
int index = Integer.parseInt(indexStr);
currentCell.setCellStyle(stylesTable.getStyleAt(index));
} catch (NumberFormatException nfe) {
LOGGER.warn("Ignoring invalid style index {}", indexStr);
}
}
// we store the dimension as well to revert with this method when cols not found
// can happen see xlsx attached here https://jira.talendforge.org/browse/TDP-1957
// <dimension ref="A1:B60"/>
} else if ("dimension".equals(tagLocalName)) {
Attribute attribute = startElement.getAttributeByName(new QName("ref"));
if (attribute != null) {
this.dimension = attribute.getValue();
}
}
// Clear contents cache
lastContents = "";
} else if (event.getEventType() == XMLStreamConstants.END_ELEMENT) {
EndElement endElement = event.asEndElement();
String tagLocalName = endElement.getName().getLocalPart();
if ("v".equals(tagLocalName) || "t".equals(tagLocalName)) {
currentCell.setRawContents(unformattedContents());
currentCell.setContents(formattedContents());
} else if ("row".equals(tagLocalName) && currentRow != null) {
rowCache.add(currentRow);
} else if ("c".equals(tagLocalName)) {
currentRow.getCellMap().put(currentCell.getColumnIndex(), currentCell);
} else if ("cols".equals(tagLocalName)) {
parsingCols = false;
}
}
}
@SuppressWarnings("null")
public Feed readFeed() {
Feed feed = null;
try {
boolean isFeedHeader = true;
// Set header values intial to the empty string
String description = "";
String title = "";
String link = "";
String language = "";
String copyright = "";
String author = "";
String pubdate = "";
String guid = "";
// First create a new XMLInputFactory
XMLInputFactory inputFactory = XMLInputFactory.newInstance();
// Setup a new eventReader
InputStream in = read();
XMLEventReader eventReader = inputFactory.createXMLEventReader(in);
// Read the XML document
while (eventReader.hasNext()) {
XMLEvent event = eventReader.nextEvent();
if (event.isStartElement()) {
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(ITEM)) {
if (isFeedHeader) {
isFeedHeader = false;
feed = new Feed(title, link, description, language, copyright, pubdate);
}
event = eventReader.nextEvent();
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(TITLE)) {
event = eventReader.nextEvent();
title = event.asCharacters().getData();
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(DESCRIPTION)) {
event = eventReader.nextEvent();
description = event.asCharacters().getData();
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(LINK)) {
event = eventReader.nextEvent();
// System.out.println("Teh hack: " + event.toString() + event.getClass());
Object chars = event.asCharacters();
if (chars instanceof javax.xml.stream.events.Characters) {
javax.xml.stream.events.Characters jchars =
(javax.xml.stream.events.Characters) chars;
link = jchars.getData();
} else {
link = event.asCharacters().getData();
}
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(GUID)) {
event = eventReader.nextEvent();
guid = event.asCharacters().getData();
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(LANGUAGE)) {
event = eventReader.nextEvent();
language = event.asCharacters().getData();
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(AUTHOR)) {
event = eventReader.nextEvent();
author = event.asCharacters().getData();
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(PUB_DATE)) {
event = eventReader.nextEvent();
pubdate = event.asCharacters().getData();
continue;
}
if (event.asStartElement().getName().getLocalPart() != null
&& event.asStartElement().getName().getLocalPart().equals(COPYRIGHT)) {
event = eventReader.nextEvent();
copyright = event.asCharacters().getData();
continue;
}
} else if (event.isEndElement()) {
if (event.asEndElement().getName().getLocalPart() != null
&& event.asEndElement().getName().getLocalPart().equals(ITEM)) {
FeedMessage message = new FeedMessage();
message.setAuthor(author);
message.setDescription(description);
message.setGuid(guid);
message.setLink(link);
message.setTitle(title);
feed.getMessages().add(message);
event = eventReader.nextEvent();
continue;
}
}
}
} catch (XMLStreamException e) {
throw new RuntimeException(e);
}
return feed;
}