#1579 Detect TSV glossary encoding via magic comment

[Aaron Madlon-Kay]

This is a mitigation for [bugs:#1046]

Detecting the encoding of TSV glossary files is fallible due to the nature of encoding detection algorithms.

To help avoid misdetections, OmegaT now attempts to determine the encoding of a TSV glossary file (except with extension .utf8) by inspecting the first line of the file for a "magic comment".

A magic comment is a comment line with content formatted like

-*- foo: bar; biz: baz -*-

which represents instructions to set foo to bar and biz to baz.

OmegaT recognizes only the setting coding: <charset> where <charset> is a charset recognized by Java's Charset#forName, such as utf-8.

A magic comment setting the coding to utf-8 will be automatically included as the first line of a writable glossary file created by OmegaT. Since the recognized comment marker for glossary files is #, the magic comment is:

# -*- coding: utf-8 -*-

Note that you can include arbitrary content between the # and the first -*-.

Existing glossary files are not modified with respect to the magic comment, and still suffer from [bugs:#1046]. Users should add an appropriate magic comment if desired.

[Aaron Madlon-Kay]

This is implemented in [85bd5c].

[Aaron Madlon-Kay]

Released in OmegaT 5.6.0.