protected void parseHeader(StartElement element, final Language sourceLanguage) {
isParagraphSegtype = SEG_PARAGRAPH.equals(getAttributeValue(element, "segtype"));
isOmegaT = CT_OMEGAT.equals(getAttributeValue(element, "creationtool"));
// log some details
Log.logRB("TMXR_INFO_CREATION_TOOL", new Object[] {getAttributeValue(element, "creationtool")});
Log.logRB(
"TMXR_INFO_CREATION_TOOL_VERSION",
new Object[] {getAttributeValue(element, "creationtoolversion")});
Log.logRB("TMXR_INFO_SEG_TYPE", new Object[] {getAttributeValue(element, "segtype")});
Log.logRB("TMXR_INFO_SOURCE_LANG", new Object[] {getAttributeValue(element, "srclang")});
// give a warning if the TMX source language is
// different from the project source language
String tmxSourceLanguage = getAttributeValue(element, "srclang");
if (!tmxSourceLanguage.equalsIgnoreCase(sourceLanguage.getLanguage())) {
Log.logWarningRB(
"TMXR_WARNING_INCORRECT_SOURCE_LANG", new Object[] {tmxSourceLanguage, sourceLanguage});
}
// give a warning that TMX file will be upgraded to sentence segmentation
if (isSegmentingEnabled && isParagraphSegtype) {
Log.logWarningRB("TMXR_WARNING_UPGRADE_SENTSEG");
}
}
/** Read TMX file. */
public void readTMX(
File file,
final Language sourceLanguage,
final Language targetLanguage,
boolean isSegmentingEnabled,
final boolean forceOmegaTMX,
final boolean extTmxLevel2,
final boolean useSlash,
final LoadCallback callback)
throws Exception {
this.extTmxLevel2 = extTmxLevel2;
this.useSlash = useSlash;
this.isSegmentingEnabled = isSegmentingEnabled;
// log the parsing attempt
Log.logRB("TMXR_INFO_READING_FILE", new Object[] {file.getAbsolutePath()});
boolean allFound = true;
InputStream in;
if (file.getName().endsWith(".gz")) {
in = new BufferedInputStream(new GZIPInputStream(new FileInputStream(file)));
} else {
in = new BufferedInputStream(new FileInputStream(file));
}
xml = factory.createXMLEventReader(in);
try {
while (xml.hasNext()) {
XMLEvent e = xml.nextEvent();
switch (e.getEventType()) {
case XMLEvent.START_ELEMENT:
StartElement eStart = (StartElement) e;
if ("tu".equals(eStart.getName().getLocalPart())) {
parseTu(eStart);
ParsedTuv origTuv = getTuvByLang(sourceLanguage);
ParsedTuv targetTuv = getTuvByLang(targetLanguage);
allFound &= callback.onEntry(currentTu, origTuv, targetTuv, isParagraphSegtype);
} else if ("header".equals(eStart.getName().getLocalPart())) {
parseHeader(eStart, sourceLanguage);
}
break;
}
}
} finally {
xml.close();
in.close();
}
if (!allFound) {
Log.logWarningRB("TMXR_WARNING_SOURCE_NOT_FOUND");
warningsCount++;
}
Log.logRB("TMXR_INFO_READING_COMPLETE");
Log.log("");
}