/**
* The primary method which contains the highest level rules for how to decide appropriate
* decoding rules: 1. first check for unicode stream 2. then looked for encoding specified in
* content (according to the type of content that is it ... xml, html, jsp, etc. 3. then check for
* various settings: file settings first, if null check project settings, if null, check user
* preferences. 4. lastly (or, what is the last user preference) is to use "workbench defaults".
*/
private void analyze() throws CoreException, IOException {
Reader resettableReader = getResettableReader();
try {
if (fCurrentEncodingMemento == null) {
resettableReader.reset();
fCurrentEncodingMemento = checkForEncodingInContents();
}
// if encoding stratagy doesn't provide answer,
// then try file settings, project settings,
// user preferences, and
// finally workbench default.
//
if (fCurrentEncodingMemento == null
|| fCurrentEncodingMemento.getDetectedCharsetName() == null) {
resettableReader.reset();
fCurrentEncodingMemento = getEncodingMementoFromResourceAndPreference();
}
// use DefaultNameRules from NonContentBasedEncodingRules as the
// final default
if (fEncodingMemento == null) {
handleNotProvidedFromContentCase();
}
fHasBeenAnalyzed = true;
} finally {
if (resettableReader != null) {
resettableReader.reset();
}
}
}
/**
* Need to check conversion early on. There's some danger than old contents of a file are set to
* empty, if an exception occurs.
*
* @param allText
* @param encoding
* @param encodingRule
* @throws java.io.UnsupportedEncodingException
* @throws MalformedOutputExceptionWithDetail
* @deprecated - we need to find "cheaper" way to to this functionality so likely to go away in
* future
*/
private void checkConversion(EncodingMemento memento, EncodingRule encodingRule)
throws IOException {
String javaEncoding = memento.getJavaCharsetName();
String detectedEncoding = memento.getDetectedCharsetName();
Charset charset = Charset.forName(javaEncoding);
CharsetEncoder charsetEncoder = charset.newEncoder();
charsetEncoder.onMalformedInput(CodingErrorAction.REPORT);
charsetEncoder.onUnmappableCharacter(CodingErrorAction.REPORT);
Reader reader = getResettableReader();
reader.reset();
int currentChar = reader.read();
int currentPos = 1;
try {
while (currentChar != -1) {
// note: this can probably be made more
// efficient later to
// check buffer by buffer, instead of
// character by character.
try {
boolean canConvert = charsetEncoder.canEncode((char) currentChar);
if (!canConvert) {
if (encodingRule == EncodingRule.IGNORE_CONVERSION_ERROR) {
// if we're told to ignore the
// encoding conversion
// error,
// notice we still want to detect
// and log it. We simply
// don't throw the exception, and
// we do continue with
// the
// save.
Logger.log(Logger.ERROR, "Encoding Conversion Error during save"); // $NON-NLS-1$
} else {
throw new MalformedOutputExceptionWithDetail(
javaEncoding, detectedEncoding, currentPos);
}
}
currentChar = reader.read();
currentPos++;
}
// IBM's JRE seems to throw NPE when DBCS char is given to
// SBCS charsetEncoder
catch (NullPointerException e) {
throw new CharConversionErrorWithDetail(javaEncoding); // $NON-NLS-1$
}
}
// if we get all the way through loop without throwing exception,
// then there must
// be an error not detectable when going character by character.
throw new CharConversionErrorWithDetail(javaEncoding); // $NON-NLS-1$
} finally {
reader.reset();
}
}
private void dump(OutputStream outputStream, EncodingRule encodingRule, boolean use3ByteBOMifUTF8)
throws CoreException, IOException {
getCurrentEncodingMemento();
String javaEncodingName = null;
if (encodingRule == EncodingRule.CONTENT_BASED) {
if (fCurrentEncodingMemento.isValid()) {
javaEncodingName = fCurrentEncodingMemento.getJavaCharsetName();
} else {
throw new UnsupportedCharsetExceptionWithDetail(fCurrentEncodingMemento);
}
} else if (encodingRule == EncodingRule.IGNORE_CONVERSION_ERROR)
javaEncodingName = fCurrentEncodingMemento.getJavaCharsetName();
else if (encodingRule == EncodingRule.FORCE_DEFAULT)
javaEncodingName = fCurrentEncodingMemento.getAppropriateDefault();
// write appropriate "header" unicode BOM bytes
// Note: Java seems to write appropriate header for
// UTF-16, but not
// UTF-8 nor UTF-16BE. This
// may vary by JRE version, so need to test well.
// Note: javaEncodingName can be null in invalid
// cases, so we no hard
// to skip whole check if that's the case.
if (javaEncodingName != null) {
if ((javaEncodingName.equals(UTF_8_CHARSET_NAME) && use3ByteBOMifUTF8)
|| (javaEncodingName.equals(UTF_8_CHARSET_NAME)
&& fCurrentEncodingMemento.isUTF83ByteBOMUsed())) {
outputStream.write(UTF3BYTEBOM);
} else if (javaEncodingName.equals(UTF_16LE_CHARSET_NAME)) {
outputStream.write(UTF16LEBOM);
} else if (javaEncodingName.equals(UTF_16BE_CHARSET_NAME)) {
outputStream.write(UTF16BEBOM);
}
}
// TODO add back in line delimiter handling the
// "right" way (updating
// markers, not requiring string, etc. .. may need
// to move to document
// level)
// allTextBuffer =
// handleLineDelimiter(allTextBuffer, document);
Reader reader = getResettableReader();
// be sure to test large "readers" ... we'll need
// to make sure they all
// can reset to initial position (StringReader,
// CharArrayReader, and
// DocumentReader should all work ok).
reader.reset();
// There must be cleaner logic somehow, but the
// idea is that
// javaEncodingName can be null
// if original detected encoding is not valid (and
// if FORCE_DEFAULT was
// not specified). Hence, we WANT the first
// Charset.forName to
// throw appropriate exception.
Charset charset = null;
// this call checks "override" properties file
javaEncodingName = CodedIO.getAppropriateJavaCharset(javaEncodingName);
if (javaEncodingName == null) {
charset = Charset.forName(fCurrentEncodingMemento.getDetectedCharsetName());
} else {
charset = Charset.forName(javaEncodingName);
}
CharsetEncoder charsetEncoder = charset.newEncoder();
if (!(encodingRule == EncodingRule.IGNORE_CONVERSION_ERROR)) {
charsetEncoder.onMalformedInput(CodingErrorAction.REPORT);
charsetEncoder.onUnmappableCharacter(CodingErrorAction.REPORT);
} else {
charsetEncoder.onMalformedInput(CodingErrorAction.REPLACE);
charsetEncoder.onUnmappableCharacter(CodingErrorAction.REPLACE);
}
OutputStreamWriter outputStreamWriter = new OutputStreamWriter(outputStream, charsetEncoder);
// TODO: this may no longer be needed (and is at
// least wrong spot for
// it).
// if (checkConversion && (!(encodingRule ==
// EncodingRule.IGNORE_CONVERSION_ERROR))) {
// checkConversion(fCurrentEncodingMemento,
// encodingRule);
// }
char[] charbuf = new char[CodedIO.MAX_BUF_SIZE];
int nRead = 0;
try {
while (nRead != -1) {
nRead = reader.read(charbuf, 0, MAX_BUF_SIZE);
if (nRead > 0) {
outputStreamWriter.flush();
outputStreamWriter.write(charbuf, 0, nRead);
}
}
} catch (UnmappableCharacterException e) {
checkConversion(fCurrentEncodingMemento, encodingRule);
} finally {
// since we don't own the original output stream, we
// won't close it ours.
// the caller who passed it to us must close original one
// when appropriate.
// (but we do flush to be sure all up-to-date)
outputStreamWriter.flush();
}
}