private void readTuElementAttribute(TmxTU tu) throws VTDException {
VTDNav vn = vu.getVTDNav();
vn.push();
AutoPilot apAttributes = new AutoPilot(vu.getVTDNav());
apAttributes.selectXPath("@*");
int inx = -1;
while ((inx = apAttributes.evalXPath()) != -1) {
String name = vn.toString(inx);
inx = vn.getAttrVal(name);
String value = inx != -1 ? vn.toString(inx) : "";
// tuid, o-encoding, datatype, usagecount, lastusagedate, creationtool, creationtoolversion,
// creationdate,
// creationid, changedate, segtype, changeid, o-tmf, srclang.
if (name.equals("tuid")) {
tu.setTuId(value);
} else if (name.equals("creationtool")) {
tu.setCreationTool(value);
} else if (name.equals("creationtoolversion")) {
tu.setCreationToolVersion(value);
} else if (name.equals("creationdate")) {
tu.setCreationDate(value);
} else if (name.equals("creationid")) {
tu.setCreationUser(value);
} else if (name.equals("changedate")) {
tu.setChangeDate(value);
} else if (name.equals("changeid")) {
tu.setChangeUser(value);
} else {
tu.appendAttribute(name, value);
}
}
vn.pop();
}
/**
* Validate TMX Format,and pilot to Body XMLElement
*
* @param vg
* @throws TmxReadException ;
*/
private void validateTmxAndParseHeader(VTDGen vg) throws TmxReadException {
VTDNav vn = vg.getNav();
AutoPilot ap = new AutoPilot(vn);
String rootPath = "/tmx";
vu = new VTDUtils();
try {
vu.bind(vn);
ap.selectXPath(rootPath);
if (ap.evalXPath() == -1) {
throw new TmxReadException(Messages.getString("document.TmxReader.validateTmxFileError"));
}
ap.resetXPath();
ap.selectXPath("/tmx/header");
if (ap.evalXPath() == -1) {
throw new TmxReadException(Messages.getString("document.TmxReader.validateTmxFileError"));
}
int id = vu.getVTDNav().getAttrVal("srclang");
if (id == -1) {
throw new TmxReadException(Messages.getString("document.TmxReader.validateTmxFileError"));
}
header.setSrclang(vu.getVTDNav().toString(id).trim());
if (vu.pilot("/tmx/body") == -1) {
throw new TmxReadException(Messages.getString("document.TmxReader.validateTmxFileError"));
}
// compute total tu number
this.totalTu = vu.getChildElementsCount();
} catch (VTDException e) {
logger.error("", e);
throw new TmxReadException(
Messages.getString("document.TmxReader.parseTmxFileError") + e.getMessage());
} finally {
vg.clear();
}
}
/**
* 获取 tmxfile 中的所有语言
*
* @return
*/
public List<String> getLangs() {
List<String> langs = new LinkedList<String>();
langs.add(LanguageUtils.convertLangCode(header.getSrclang()));
VTDNav vn = vu.getVTDNav();
vn.push();
AutoPilot ap = new AutoPilot(vn);
try {
ap.selectXPath("/tmx/body/tu/tuv");
String lang;
int index = -1;
while (ap.evalXPath() != -1) {
index = vn.getAttrVal("xml:lang");
if (index == -1) {
index = vn.getAttrVal("lang"); // version 1.1
if (index == -1) {
continue;
}
}
lang = LanguageUtils.convertLangCode(vn.toRawString(index));
if (!langs.contains(lang)) {
langs.add(lang);
}
}
} catch (Exception e) {
e.printStackTrace();
}
vn.pop();
return langs;
}
private void readTuPropElement(TmxTU tu) throws VTDException {
VTDNav vn = vu.getVTDNav();
vn.push();
AutoPilot ap = new AutoPilot(vn);
ap.selectXPath("./prop");
while (ap.evalXPath() != -1) {
String content = vu.getElementContent();
if (content == null) {
continue;
}
int inx = vn.getAttrVal("type");
String typeValue = inx != -1 ? vn.toString(inx) : null;
if (typeValue == null) {
continue;
}
if (typeValue.equals(TmxContexts.PRE_CONTEXT_NAME)) {
tu.appendContext(TmxContexts.PRE_CONTEXT_NAME, content.trim());
} else if (typeValue.equals(TmxContexts.NEXT_CONTEXT_NAME)) {
tu.appendContext(TmxContexts.NEXT_CONTEXT_NAME, content.trim());
} else if (typeValue.equals("x-Context")) {
// Trados TMX file
String[] contexts = content.split(",");
if (contexts.length == 2) {
tu.appendContext(TmxContexts.PRE_CONTEXT_NAME, contexts[0].trim());
tu.appendContext(TmxContexts.NEXT_CONTEXT_NAME, contexts[1].trim());
}
} else {
TmxProp p = new TmxProp(typeValue, content);
tu.appendProp(p);
}
}
vn.pop();
}
public TmxReader(String tmxContent) throws TmxReadException {
// 解析文件
VTDGen vg = new VTDGen();
vg.setDoc(tmxContent.getBytes());
String message = "";
try {
vg.parse(true);
} catch (EncodingException e) {
logger.error(Messages.getString("document.ImportAbstract.logger1"), e);
message = Messages.getString("document.ImportAbstract.msg1");
throw new TmxReadException(message + e.getMessage());
} catch (ParseException e) {
logger.error(Messages.getString("document.ImportAbstract.logger3"), e);
String errMsg = e.getMessage();
if (errMsg.indexOf("invalid encoding") != -1) { // 编码异常
message = Messages.getString("document.ImportAbstract.msg1");
} else {
message = Messages.getString("document.ImportAbstract.msg3");
}
throw new TmxReadException(message + e.getMessage());
}
header = new TmxHeader();
validateTmxAndParseHeader(vg);
tuAp = new AutoPilot(vu.getVTDNav());
try {
tuAp.selectXPath("./tu");
} catch (XPathParseException e) {
throw new TmxReadException(Messages.getString("document.TmxReader.parseTmxFileError"));
}
}
private void readTuTuvElement(TmxTU tu) throws VTDException {
VTDNav vn = vu.getVTDNav();
vn.push();
AutoPilot ap = new AutoPilot(vn);
ap.selectXPath("./tuv");
// TUV 节点下的Note,Prop节点暂时不处理，所以此处暂时不解析
while (ap.evalXPath() != -1) {
int inx = vn.getAttrVal("xml:lang");
inx = inx == -1 ? vn.getAttrVal("lang") : inx;
String lang = inx != -1 ? vn.toString(inx) : null;
if (lang == null) {
continue;
}
vn.push();
if (vu.pilot("./seg") != -1) {
String fullText = vu.getElementContent().trim();
String pureText = DocUtils.getTmxTbxPureText(vu).trim();
if (fullText == null || pureText == null || fullText.equals("") || pureText.equals("")) {
// fix Bug #2928 by Jason SQLite--导入TMX异常, 导入程序正常退出，但是未完全导入所有内容，此处在continue时应该先调用vn.pop()
vn.pop();
continue;
}
TmxSegement segment = new TmxSegement();
segment.setLangCode(Utils.convertLangCode(lang));
if (tmxFilter == null) segment.setFullText(fullText);
else {
String text = tmxFilter.clearString(fullText);
segment.setFullText(text);
}
segment.setPureText(pureText);
if (lang.equalsIgnoreCase(header.getSrclang())) {
tu.setSource(segment);
} else {
tu.appendSegement(segment);
}
}
vn.pop();
}
vn.pop();
}
public TmxReader(File file) throws TmxReadException {
// 解析文件
VTDGen vg = null;
try {
vg = VTDLoader.loadVTDGen(file, FileEncodingDetector.detectFileEncoding(file));
} catch (IOException e) {
logger.error(Messages.getString("document.DocUtils.logger1"), e);
throw new TmxReadException(Messages.getString("document.TmxReader.parseTmxFileError"));
} catch (EncodingException e) {
logger.error(Messages.getString("document.ImportAbstract.logger1"), e);
String message = Messages.getString("document.ImportAbstract.msg1");
throw new TmxReadException(message + e.getMessage());
} catch (ParseException e) {
logger.error(Messages.getString("document.ImportAbstract.logger3"), e);
String errMsg = e.getMessage();
String message;
if (errMsg.indexOf("invalid encoding") != -1) { // 编码异常
message = Messages.getString("document.ImportAbstract.msg1");
} else {
message = Messages.getString("document.ImportAbstract.msg3");
}
throw new TmxReadException(message + e.getMessage());
} catch (EmptyFileException e) {
logger.error(Messages.getString("document.DocUtils.logger1"), e);
throw new TmxReadException(Messages.getString("document.TmxReader.EmptyTmxFileError"));
}
if (vg == null) {
throw new TmxReadException(Messages.getString("document.TmxReader.parseTmxFileError"));
}
// 验证TMX ,解析Header XMLElement，将节点导航到Body XMLElement
header = new TmxHeader();
validateTmxAndParseHeader(vg);
tuAp = new AutoPilot(vu.getVTDNav());
try {
tuAp.selectXPath("./tu");
} catch (XPathParseException e) {
throw new TmxReadException(Messages.getString("document.TmxReader.parseTmxFileError"));
}
}
private void readTuNoteElement(TmxTU tu) throws VTDException {
VTDNav vn = vu.getVTDNav();
vn.push();
AutoPilot ap = new AutoPilot(vn);
ap.selectXPath("./note");
while (ap.evalXPath() != -1) {
String fragment = vu.getElementFragment();
TmxNote note = new TmxNote();
note.setContent(fragment);
int inx = vn.getAttrVal("xml:lang");
String value = inx != -1 ? vn.toString(inx) : null;
if (value != null) {
note.setXmlLang(value);
}
inx = vn.getAttrVal("o-encoding");
value = inx != -1 ? vn.toString(inx) : null;
if (value != null) {
note.setXmlLang(value);
}
tu.appendNote(note);
}
vn.pop();
}