Ewts Converter

This Java package implements the conversion between Unicode Tibetan text, and Extended Wylie transliteration (EWTS). It also has convenience conversion methods from Diacritics Transliteration Scheme (DTS) and ALA-LC romanization to EWTS, and from EWTS to ALA-LC romanization.

It is based on the equivalent Perl module, Lingua::BO::Converter.

See Change log for version notes.

Installation

Using maven:

    <dependency>
      <groupId>io.bdrc.ewtsconverter</groupId>
      <artifactId>ewts-converter</artifactId>
      <version>1.4.2</version>
    </dependency>

Building

We provide one maven option: -DperformRelease=true, which will make the jar file gpg-signed.

Use

import io.bdrc.ewtsconverter.EwtsConverter;

EwtsConverter wl = new EwtsConverter();
System.out.println(wl.toUnicode("sems can thams cad"));
System.out.println(wl.toWylie("\u0f66\u0f44\u0f66\u0f0b\u0f62\u0f92\u0fb1\u0f66\u000a"));

Options

You can pass some options to the constructor:

EwtsConverter(boolean check, boolean check_strict, boolean print_warnings, boolean fix_spacing, Mode mode)

• check: generate warnings for illegal consonant sequences; default is true.
• check_strict: stricter checking, examine the whole stack; default is true.
• print_warnings: print generated warnings to System.out; default is false.
• fix_spacing: remove spaces after newlines, collapse multiple tseks into one, fix case, etc; default is true.
• mode: an EwtsConverter.Mode value, one of EWTS (default), ALALC (alalc transliteration scheme) or DTS (close to alalc, not publicly documented).

API

Functions of the EwtsConverter object

String toUnicode(String wylie_string)

Converts from Converter (EWTS) to Unicode.

String toUnicode(String wylie_string, ArrayList warns)

Converts from Converter (EWTS) to Unicode; puts the generated warnings in the list.

String toWylie(String unicode_string)

Converts from Unicode to Converter. Anything that is not Tibetan Unicode is converted to EWTS comment blocks [between brackets].

String toWylie(String unicode_string, ArrayList warns, boolean escape)

Converts from Unicode to Converter. Puts the generated warnings in the list. If escape is false, anything that is not Tibetan Unicode is just passed through as it is.

Static functions of the EwtsConverter class

String normalizeSloppyWylie(String str)

Returns a string normalizing common errors in EWTS.

boolean isCombining(char c)

Returns true if the character is a Tibetan combining character.

Static functions of the TransConverter class

String dtsToEwts(String dtsString)

Converts a string from DTS to EWTS.

String alalcToEwts(String alalcStr)

Converts a string from ALA-LC to EWTS.

String ewtsToAlalc(String ewtsStr, boolean sloppy)

Converts a string from EWTS to ALA-LC (in NFKD, lower-case). If sloppy is true, also normalizes common errors in EWTS.

Performance and Concurrency

This code should perform quite decently. When converting from Ewts to Unicode, the entire string is split into tokens, which are themselves strings. If this takes too much memory, consider converting your text in smaller chunks. With today's computers, it should not be a problem to convert several megabytes of tibetan text in one call. Otherwise, it could be worthwhile to tokenize the input on the fly, rather than all at once.

This class is entirely thread-safe. In a multi-threaded environment, multiple threads can share the same instance without any problems.

License

For simplicity reasons, we distribute our modifications only under the Apache 2.0 License, but the original version had this statement:

This library is Free Software.  You can redistribute it or modify it, under
the terms of, at your choice, the GNU General Public License (version 2 or
higher), the GNU Lesser General Public License (version 2 or higher), the
Mozilla Public License (any version) or the Apache License version 2 or
higher.

Please contact the author if you wish to use it under some terms not covered
here.