Non-auto-detected encodings not taken in account in XLIFFFilter

[Former user]

Original issue 99 created by @ysavourel on 2009-07-26T22:18:06.000Z:

When an XLIFF document has an encoding different from one auto-detected by
the BomandEncoding detector, it is not set properly. The XML stream reader
uses the one define for the Reader used in open(), but that one is the one
from the auto-detector and may be wrong.

It seems we need to pass the rawdocument.stream to the open method, not
use getReader(), but it seems we cannot because it has been used by the
auto-detector.

Simple test: use a Windows-1252 xliff with some extended chars. they will
come as Windows-1252 read as UTF-8.

[Former user]

Comment [1.](https://code.google.com/p/okapi/issues/detail?id=99#c1) originally posted by @ysavourel on 2009-07-27T17:49:31.000Z:

Its OK to use getStream after auto-detection because it resets the stream. Its one of the exception cases where the caller can reset the stream properly for reuse.

[Jim Hargrave (OLD)]

‌

// Determine encoding based on BOM, if any
input.setEncoding(ENCODING.name()); // Default for XML, other should be auto-detected
BOMNewlineEncodingDetector detector = new BOMNewlineEncodingDetector(input.getStream(), input.getEncoding());
detector.detectBom();

String inStreamCharset = ENCODING.name();
if ( detector.isAutodetected() ) {
    inStreamCharset = detector.getEncoding();
}

We shouldn’t override the passed in encoding (input.setEncoding(ENCODING.name());). Code should look like this:

‌

BOMNewlineEncodingDetector detector = new BOMNewlineEncodingDetector(input.getStream(), input.getEncoding());
detector.detectBom();
String encoding = detector.getEncoding(); // return the passed in encoding - or the one detected using the BOM.
// should log warning if they differ

‌