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"));
}
}
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"));
}
}
/**
* Parse file with VTD-XML
*
* @param file
* @return
* @throws TmxReadException All Exception come from VTDExcetpion;
*/
private VTDGen paseFile(File file) throws TmxReadException {
String encoding = FileEncodingDetector.detectFileEncoding(file);
VTDGen vg = new VTDGen();
FileInputStream fis = null;
String message = "";
try {
fis = new FileInputStream(file);
byte[] bArr = new byte[(int) file.length()];
int offset = 0;
int numRead = 0;
int numOfBytes = 1048576; // I choose this value randomally,
// any other (not too big) value also can be here.
if (bArr.length - offset < numOfBytes) {
numOfBytes = bArr.length - offset;
}
while (offset < bArr.length && (numRead = fis.read(bArr, offset, numOfBytes)) >= 0) {
offset += numRead;
if (bArr.length - offset < numOfBytes) {
numOfBytes = bArr.length - offset;
}
}
// clean invalid XML character
byte[] _bArr = new byte[bArr.length];
int _bArrIndx = 0;
int type = 0;
if (encoding.equalsIgnoreCase("UTF-16LE") || encoding.equalsIgnoreCase("UTF-16BE")) {
type = 1;
}
for (int i = 0; i < bArr.length; i++) {
byte b = bArr[i];
if ((b >= type && b <= 8) || b == 11 || b == 12 || (b >= 14 && b <= 31)) {
continue;
} else if (b == 38
&& i + 1 < bArr.length
&& bArr[i + 1] == 35
&& i + 2 < bArr.length) { // &#
List<Byte> entis = new ArrayList<Byte>();
entis.add((byte) 38);
entis.add((byte) 35);
int j = i + 2;
if (bArr[j] == 120) { // x
entis.add((byte) 120);
while (true) {
j++;
if (j >= bArr.length) {
entis.clear();
b = bArr[i];
break;
}
b = bArr[j];
if ((b >= 48 && b <= 57) || (b >= 97 && b <= 102) || (b >= 65 && b <= 70)) {
entis.add(b);
} else if (b == 59) {
entis.add(b);
i = j;
break;
} else if (j - i > 10) {
entis.clear();
b = bArr[i];
break;
} else {
entis.clear();
b = bArr[i];
break;
}
}
} else {
while (true) {
b = bArr[j];
if ((b >= 48 && b <= 57)) {
entis.add(b);
} else if (b == 59) {
entis.add(b);
i = j;
break;
} else if (j - i > 10) {
entis.clear();
b = bArr[i];
break;
} else {
entis.clear();
b = bArr[i];
break;
}
j++;
if (j >= bArr.length) {
entis.clear();
b = bArr[i];
break;
}
}
}
if (!entis.isEmpty()) {
byte[] t = new byte[entis.size()];
for (int ti = 0; ti < entis.size(); ti++) {
t[ti] = entis.get(ti);
}
String s = new String(t);
if (s.matches("((&#[x]?)(([0]?([0-8]|[BbCcEe]))|(1[0-9])|(1[a-fA-F]));)")) {
continue;
}
}
}
_bArr[_bArrIndx++] = b;
}
bArr = null;
bArr = Arrays.copyOf(_bArr, _bArrIndx);
// use vtd parse
vg.setDoc(bArr);
vg.parse(true);
} 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);
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());
} finally {
if (fis != null) {
try {
fis.close();
} catch (Exception e) {
}
}
}
return vg;
}