@Nullable(
"null means no luck, otherwise it's tuple(guessed encoding, hint about content if was unable to guess, BOM)")
public static Trinity<Charset, CharsetToolkit.GuessedEncoding, byte[]> guessFromContent(
@NotNull VirtualFile virtualFile, @NotNull byte[] content, int length) {
Charset defaultCharset =
ObjectUtils.notNull(
EncodingManager.getInstance().getEncoding(virtualFile, true),
CharsetToolkit.getDefaultSystemCharset());
CharsetToolkit toolkit = GUESS_UTF ? new CharsetToolkit(content, defaultCharset) : null;
String detectedFromBytes = null;
try {
if (GUESS_UTF) {
toolkit.setEnforce8Bit(true);
Charset charset = toolkit.guessFromBOM();
if (charset != null) {
detectedFromBytes = AUTO_DETECTED_FROM_BOM;
byte[] bom =
ObjectUtils.notNull(CharsetToolkit.getMandatoryBom(charset), CharsetToolkit.UTF8_BOM);
return Trinity.create(charset, null, bom);
}
CharsetToolkit.GuessedEncoding guessed = toolkit.guessFromContent(length);
if (guessed == CharsetToolkit.GuessedEncoding.VALID_UTF8) {
detectedFromBytes = "auto-detected from bytes";
return Trinity.create(
CharsetToolkit.UTF8_CHARSET, guessed, null); // UTF detected, ignore all directives
}
if (guessed == CharsetToolkit.GuessedEncoding.SEVEN_BIT) {
return Trinity.create(null, guessed, null);
}
}
return null;
} finally {
setCharsetWasDetectedFromBytes(virtualFile, detectedFromBytes);
}
}
public static void setDetectedFromBytesFlagBack(
@NotNull VirtualFile virtualFile, @NotNull byte[] content) {
if (virtualFile.getBOM() == null) {
guessFromContent(virtualFile, content, content.length);
} else {
// prevent file to be reloaded in other encoding after save with BOM
setCharsetWasDetectedFromBytes(virtualFile, AUTO_DETECTED_FROM_BOM);
}
}
@NotNull
private static Pair.NonNull<Charset, byte[]> doDetectCharsetAndSetBOM(
@NotNull VirtualFile virtualFile,
@NotNull byte[] content,
boolean saveBOM,
@NotNull FileType fileType) {
@NotNull
Charset charset =
virtualFile.isCharsetSet()
? virtualFile.getCharset()
: detectCharset(virtualFile, content, fileType);
Pair.NonNull<Charset, byte[]> bomAndCharset = getCharsetAndBOM(content, charset);
final byte[] bom = bomAndCharset.second;
if (saveBOM && bom.length != 0) {
virtualFile.setBOM(bom);
setCharsetWasDetectedFromBytes(virtualFile, AUTO_DETECTED_FROM_BOM);
}
return bomAndCharset;
}
