From 303e99050b18bd1d0ecea1fe269c9907a1201ff1 Mon Sep 17 00:00:00 2001
From: Rainer Kottenhoff <rainer.kottenhoff@inform-software.com>
Date: Thu, 5 Mar 2026 08:52:25 +0100
Subject: [PATCH] fix: utf-8 affinity for uchardet uncertainty (lower than 66%)

---
 src/EncodingDetection.cpp                     |  23 ++++++++++++------
 src/uchardet/uchardet/test_moved.readme       |   1 +
 .../encoding/uchardet}/CMakeLists.txt         |   0
 .../encoding/uchardet}/ar/iso-8859-6.txt      |   0
 .../encoding/uchardet}/ar/utf-8.txt           |   0
 .../encoding/uchardet}/ar/windows-1256.txt    |   0
 .../encoding/uchardet}/bg/windows-1251.txt    |   0
 .../encoding/uchardet}/cs/ibm852.txt          |   0
 .../encoding/uchardet}/cs/iso-8859-2.txt      |   0
 .../uchardet}/cs/mac-centraleurope.txt        |   0
 .../encoding/uchardet}/cs/utf-8.txt           |   0
 .../encoding/uchardet}/cs/windows-1250.txt    |   0
 .../encoding/uchardet}/da/ibm865.txt          |   0
 .../encoding/uchardet}/da/iso-8859-1.txt      |   0
 .../encoding/uchardet}/da/iso-8859-15.txt     |   0
 .../encoding/uchardet}/da/utf-8.txt           |   0
 .../encoding/uchardet}/da/windows-1252.txt    |   0
 .../encoding/uchardet}/de/iso-8859-1.txt      |   0
 .../encoding/uchardet}/de/windows-1252.txt    |   0
 .../encoding/uchardet}/el/iso-8859-7.txt      |   0
 .../encoding/uchardet}/el/utf-8.txt           |   0
 .../encoding/uchardet}/el/windows-1253.txt    |   0
 .../encoding/uchardet}/en/ascii.txt           |   0
 .../encoding/uchardet}/eo/iso-8859-3.txt      |   0
 .../encoding/uchardet}/es/iso-8859-1.txt      |   0
 .../encoding/uchardet}/es/iso-8859-15.txt     |   0
 .../encoding/uchardet}/es/utf-8.txt           |   0
 .../encoding/uchardet}/es/windows-1252.txt    |   0
 .../encoding/uchardet}/et/iso-8859-13.txt     |   0
 .../encoding/uchardet}/et/iso-8859-15.txt     |   0
 .../encoding/uchardet}/et/iso-8859-4.txt      |   0
 .../encoding/uchardet}/et/utf-8.txt           |   0
 .../encoding/uchardet}/et/windows-1252.txt    |   0
 .../encoding/uchardet}/et/windows-1257.txt    |   0
 .../encoding/uchardet}/fi/iso-8859-1.txt      |   0
 .../encoding/uchardet}/fi/utf-8.txt           |   0
 .../encoding/uchardet}/fr/iso-8859-1.txt      |   0
 .../encoding/uchardet}/fr/iso-8859-15.txt     |   0
 .../encoding/uchardet}/fr/utf-16.be           | Bin
 .../encoding/uchardet}/fr/utf-32.le           | Bin
 .../encoding/uchardet}/fr/utf-8.txt           |   0
 .../encoding/uchardet}/fr/windows-1252.txt    |   0
 .../encoding/uchardet}/ga/iso-8859-1.txt      |   0
 .../encoding/uchardet}/ga/utf-8.txt           |   0
 .../encoding/uchardet}/ga/windows-1252.txt    |   0
 .../encoding/uchardet}/he/iso-8859-8.txt      |   0
 .../encoding/uchardet}/he/utf-8.txt           |   0
 .../encoding/uchardet}/he/windows-1255.txt    |   0
 .../encoding/uchardet}/hr/ibm852.txt          |   0
 .../encoding/uchardet}/hr/iso-8859-13.txt     |   0
 .../encoding/uchardet}/hr/iso-8859-16.txt     |   0
 .../encoding/uchardet}/hr/iso-8859-2.txt      |   0
 .../uchardet}/hr/mac-centraleurope.txt        |   0
 .../encoding/uchardet}/hr/utf-8.txt           |   0
 .../encoding/uchardet}/hr/windows-1250.txt    |   0
 .../encoding/uchardet}/hu/iso-8859-2.txt      |   0
 .../encoding/uchardet}/hu/windows-1250.txt    |   0
 .../encoding/uchardet}/it/iso-8859-1.txt      |   0
 .../encoding/uchardet}/it/utf-8.txt           |   0
 .../encoding/uchardet}/ja/euc-jp.txt          |   0
 .../encoding/uchardet}/ja/iso-2022-jp.txt     |   0
 .../encoding/uchardet}/ja/shift_jis.txt       |   0
 .../encoding/uchardet}/ja/utf-16be.txt        | Bin
 .../encoding/uchardet}/ja/utf-16le.txt        | Bin
 .../encoding/uchardet}/ja/utf-8.txt           |   0
 .../encoding/uchardet}/ko/iso-2022-kr.txt     |   0
 .../test_files/encoding/uchardet}/ko/uhc.smi  |   0
 .../encoding/uchardet}/ko/utf-16.le           | Bin
 .../encoding/uchardet}/ko/utf-32.be           | Bin
 .../encoding/uchardet}/ko/utf-8.txt           |   0
 .../encoding/uchardet}/lt/iso-8859-10.txt     |   0
 .../encoding/uchardet}/lt/iso-8859-13.txt     |   0
 .../encoding/uchardet}/lt/iso-8859-4.txt      |   0
 .../encoding/uchardet}/lt/utf-8.txt           |   0
 .../encoding/uchardet}/lv/iso-8859-10.txt     |   0
 .../encoding/uchardet}/lv/iso-8859-13.txt     |   0
 .../encoding/uchardet}/lv/iso-8859-4.txt      |   0
 .../encoding/uchardet}/lv/utf-8.txt           |   0
 .../encoding/uchardet}/mt/iso-8859-3.txt      |   0
 .../encoding/uchardet}/mt/utf-8.txt           |   0
 .../encoding/uchardet}/no/ibm865.txt          |   0
 .../encoding/uchardet}/no/iso-8859-1.txt      |   0
 .../encoding/uchardet}/no/iso-8859-15.txt     |   0
 .../encoding/uchardet}/no/utf-8.txt           |   0
 .../encoding/uchardet}/no/windows-1252.txt    |   0
 .../encoding/uchardet}/pl/ibm852.txt          |   0
 .../encoding/uchardet}/pl/iso-8859-13.txt     |   0
 .../encoding/uchardet}/pl/iso-8859-16.txt     |   0
 .../encoding/uchardet}/pl/iso-8859-2.txt      |   0
 .../uchardet}/pl/mac-centraleurope.txt        |   0
 .../encoding/uchardet}/pl/utf-8.txt           |   0
 .../encoding/uchardet}/pl/windows-1250.txt    |   0
 .../encoding/uchardet}/pt/iso-8859-1.txt      |   0
 .../encoding/uchardet}/pt/utf-8.txt           |   0
 .../encoding/uchardet}/ro/ibm852.txt          |   0
 .../encoding/uchardet}/ro/iso-8859-16.txt     |   0
 .../encoding/uchardet}/ro/utf-8.txt           |   0
 .../encoding/uchardet}/ro/windows-1250.txt    |   0
 .../encoding/uchardet}/ru/ibm855.txt          |   0
 .../encoding/uchardet}/ru/ibm866.txt          |   0
 .../encoding/uchardet}/ru/iso-8859-5.txt      |   0
 .../encoding/uchardet}/ru/koi8-r.txt          |   0
 .../encoding/uchardet}/ru/mac-cyrillic.txt    |   0
 .../encoding/uchardet}/ru/windows-1251.txt    |   0
 .../encoding/uchardet}/sk/ibm852.txt          |   0
 .../encoding/uchardet}/sk/iso-8859-2.txt      |   0
 .../uchardet}/sk/mac-centraleurope.txt        |   0
 .../encoding/uchardet}/sk/utf-8.txt           |   0
 .../encoding/uchardet}/sk/windows-1250.txt    |   0
 .../encoding/uchardet}/sl/ibm852.txt          |   0
 .../encoding/uchardet}/sl/iso-8859-16.txt     |   0
 .../encoding/uchardet}/sl/iso-8859-2.txt      |   0
 .../uchardet}/sl/mac-centraleurope.txt        |   0
 .../encoding/uchardet}/sl/utf-8.txt           |   0
 .../encoding/uchardet}/sl/windows-1250.txt    |   0
 .../encoding/uchardet}/sv/iso-8859-1.txt      |   0
 .../encoding/uchardet}/sv/utf-8.txt           |   0
 .../encoding/uchardet}/sv/windows-1252.txt    |   0
 .../encoding/uchardet}/th/iso-8859-11.txt     |   0
 .../encoding/uchardet}/th/tis-620.txt         |   0
 .../encoding/uchardet}/th/utf-8.txt           |   0
 .../encoding/uchardet}/tr/iso-8859-3.txt      |   0
 .../encoding/uchardet}/tr/iso-8859-9.txt      |   0
 .../encoding/uchardet}/uchardet-tests.c       |   0
 .../encoding/uchardet}/vi/utf-8.txt           |   0
 .../encoding/uchardet}/vi/viscii.txt          |   0
 .../encoding/uchardet}/vi/windows-1258.txt    |   0
 .../test_files/encoding/uchardet}/zh/big5.txt |   0
 .../encoding/uchardet}/zh/euc-tw.txt          |   0
 .../encoding/uchardet}/zh/gb18030.txt         |   0
 .../encoding/uchardet}/zh/utf-8.txt           |   0
 131 files changed, 17 insertions(+), 7 deletions(-)
 create mode 100644 src/uchardet/uchardet/test_moved.readme
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/CMakeLists.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ar/iso-8859-6.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ar/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ar/windows-1256.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/bg/windows-1251.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/cs/ibm852.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/cs/iso-8859-2.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/cs/mac-centraleurope.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/cs/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/cs/windows-1250.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/da/ibm865.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/da/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/da/iso-8859-15.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/da/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/da/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/de/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/de/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/el/iso-8859-7.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/el/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/el/windows-1253.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/en/ascii.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/eo/iso-8859-3.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/es/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/es/iso-8859-15.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/es/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/es/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/et/iso-8859-13.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/et/iso-8859-15.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/et/iso-8859-4.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/et/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/et/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/et/windows-1257.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fi/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fi/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fr/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fr/iso-8859-15.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fr/utf-16.be (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fr/utf-32.le (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fr/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/fr/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ga/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ga/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ga/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/he/iso-8859-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/he/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/he/windows-1255.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hr/ibm852.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hr/iso-8859-13.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hr/iso-8859-16.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hr/iso-8859-2.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hr/mac-centraleurope.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hr/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hr/windows-1250.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hu/iso-8859-2.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/hu/windows-1250.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/it/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/it/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ja/euc-jp.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ja/iso-2022-jp.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ja/shift_jis.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ja/utf-16be.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ja/utf-16le.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ja/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ko/iso-2022-kr.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ko/uhc.smi (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ko/utf-16.le (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ko/utf-32.be (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ko/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lt/iso-8859-10.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lt/iso-8859-13.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lt/iso-8859-4.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lt/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lv/iso-8859-10.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lv/iso-8859-13.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lv/iso-8859-4.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/lv/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/mt/iso-8859-3.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/mt/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/no/ibm865.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/no/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/no/iso-8859-15.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/no/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/no/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pl/ibm852.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pl/iso-8859-13.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pl/iso-8859-16.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pl/iso-8859-2.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pl/mac-centraleurope.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pl/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pl/windows-1250.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pt/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/pt/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ro/ibm852.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ro/iso-8859-16.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ro/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ro/windows-1250.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ru/ibm855.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ru/ibm866.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ru/iso-8859-5.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ru/koi8-r.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ru/mac-cyrillic.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/ru/windows-1251.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sk/ibm852.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sk/iso-8859-2.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sk/mac-centraleurope.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sk/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sk/windows-1250.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sl/ibm852.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sl/iso-8859-16.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sl/iso-8859-2.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sl/mac-centraleurope.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sl/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sl/windows-1250.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sv/iso-8859-1.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sv/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/sv/windows-1252.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/th/iso-8859-11.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/th/tis-620.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/th/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/tr/iso-8859-3.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/tr/iso-8859-9.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/uchardet-tests.c (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/vi/utf-8.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/vi/viscii.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/vi/windows-1258.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/zh/big5.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/zh/euc-tw.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/zh/gb18030.txt (100%)
 rename {src/uchardet/uchardet/test => test/test_files/encoding/uchardet}/zh/utf-8.txt (100%)

diff --git a/src/EncodingDetection.cpp b/src/EncodingDetection.cpp
index 3bb5e3009..05eb4eed7 100644
--- a/src/EncodingDetection.cpp
+++ b/src/EncodingDetection.cpp
@@ -647,11 +647,7 @@ constexpr cpi_enc_t _MapStdEncodingString2CPI(const char* encStrg, float* pConfi
             bool bBOM = false;
             bool bReverse = false;
             cpi_enc_t const cpi = GetUnicodeEncoding(text, len, &bBOM, &bReverse);
-            if (!Encoding_IsNONE(cpiEncoding)) {
-                cpiEncoding = cpi;
-            } else {
-                cpiEncoding = bBOM ? (bReverse ? CPI_UNICODEBE : CPI_UNICODE) : (bReverse ? CPI_UNICODEBE : CPI_UNICODE);
-            }
+            cpiEncoding = cpi;
         }
 
         // check for default ANSI
@@ -814,6 +810,7 @@ void Encoding_AnalyzeText(const char* const text, const size_t len,
     // ---  re-mapping UCD ----
 
     switch (Encoding_GetCodePage(cpiEncoding_UCD)) {
+
     case 28591:  // ISO 8859 - 1  mapped to  Windows - 1252  (HTML5 Standard advice)
         cpiEncoding_UCD = Encoding_GetByCodePage(1252); // auto detect default ANSI (!)
         break;
@@ -1303,6 +1300,7 @@ extern "C" ENC_DET_T Encoding_DetectEncoding(const HPATHL hpath, const char* lpD
     }
 
     if (!IS_ENC_ENFORCED() || bForceEncDetection) {
+
         if (!bSkipANSICPDetection) {
             // ---------------------------------------------------------------------------
             Encoding_AnalyzeText(lpData, cbNbytes4Analysis, &encDetRes, iAnalyzeHint);
@@ -1375,11 +1373,22 @@ extern "C" ENC_DET_T Encoding_DetectEncoding(const HPATHL hpath, const char* lpD
     }
     else if (Encoding_IsValid(encDetRes.analyzedEncoding) && (encDetRes.bIsAnalysisReliable || !Settings.UseReliableCEDonly))
     {
-        encDetRes.Encoding = encDetRes.analyzedEncoding;
+        if (!encDetRes.bIsAnalysisReliable && !Encoding_IsUTF8(encDetRes.analyzedEncoding) && encDetRes.bValidUTF8) {
+            encDetRes.Encoding = CPI_UTF8; // unreliable non-UTF-8 guess, but data is valid UTF-8
+        } else {
+            encDetRes.Encoding = encDetRes.analyzedEncoding;
+        }
+    }
+    else if (!encDetRes.bIsAnalysisReliable && Encoding_IsValid(encDetRes.analyzedEncoding))
+    {
+        // UCHARDET below confidence threshold (UseReliableCEDonly is true)
+        encDetRes.Encoding = encDetRes.bValidUTF8 ? CPI_UTF8 : CPI_ANSI_DEFAULT;
     }
     else if (Encoding_IsUNICODE(encDetRes.unicodeAnalysis) && (iConfidence > 66))
     {
-        encDetRes.Encoding = encDetRes.analyzedEncoding; // (1) rely on analyzed encoding
+        // unicodeAnalysis (IsTextUnicode) confirms Unicode structure,
+        // iConfidence is from UCHARDET analysis — use analyzedEncoding (intentional)
+        encDetRes.Encoding = encDetRes.analyzedEncoding;
     }
     else if (Encoding_IsValid(Encoding_SrcWeak(CPI_GET)))
     {
diff --git a/src/uchardet/uchardet/test_moved.readme b/src/uchardet/uchardet/test_moved.readme
new file mode 100644
index 000000000..7e5c8f255
--- /dev/null
+++ b/src/uchardet/uchardet/test_moved.readme
@@ -0,0 +1 @@
+tests are moved to Notepad3\test\test_files\encoding\uchardet\
diff --git a/src/uchardet/uchardet/test/CMakeLists.txt b/test/test_files/encoding/uchardet/CMakeLists.txt
similarity index 100%
rename from src/uchardet/uchardet/test/CMakeLists.txt
rename to test/test_files/encoding/uchardet/CMakeLists.txt
diff --git a/src/uchardet/uchardet/test/ar/iso-8859-6.txt b/test/test_files/encoding/uchardet/ar/iso-8859-6.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ar/iso-8859-6.txt
rename to test/test_files/encoding/uchardet/ar/iso-8859-6.txt
diff --git a/src/uchardet/uchardet/test/ar/utf-8.txt b/test/test_files/encoding/uchardet/ar/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ar/utf-8.txt
rename to test/test_files/encoding/uchardet/ar/utf-8.txt
diff --git a/src/uchardet/uchardet/test/ar/windows-1256.txt b/test/test_files/encoding/uchardet/ar/windows-1256.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ar/windows-1256.txt
rename to test/test_files/encoding/uchardet/ar/windows-1256.txt
diff --git a/src/uchardet/uchardet/test/bg/windows-1251.txt b/test/test_files/encoding/uchardet/bg/windows-1251.txt
similarity index 100%
rename from src/uchardet/uchardet/test/bg/windows-1251.txt
rename to test/test_files/encoding/uchardet/bg/windows-1251.txt
diff --git a/src/uchardet/uchardet/test/cs/ibm852.txt b/test/test_files/encoding/uchardet/cs/ibm852.txt
similarity index 100%
rename from src/uchardet/uchardet/test/cs/ibm852.txt
rename to test/test_files/encoding/uchardet/cs/ibm852.txt
diff --git a/src/uchardet/uchardet/test/cs/iso-8859-2.txt b/test/test_files/encoding/uchardet/cs/iso-8859-2.txt
similarity index 100%
rename from src/uchardet/uchardet/test/cs/iso-8859-2.txt
rename to test/test_files/encoding/uchardet/cs/iso-8859-2.txt
diff --git a/src/uchardet/uchardet/test/cs/mac-centraleurope.txt b/test/test_files/encoding/uchardet/cs/mac-centraleurope.txt
similarity index 100%
rename from src/uchardet/uchardet/test/cs/mac-centraleurope.txt
rename to test/test_files/encoding/uchardet/cs/mac-centraleurope.txt
diff --git a/src/uchardet/uchardet/test/cs/utf-8.txt b/test/test_files/encoding/uchardet/cs/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/cs/utf-8.txt
rename to test/test_files/encoding/uchardet/cs/utf-8.txt
diff --git a/src/uchardet/uchardet/test/cs/windows-1250.txt b/test/test_files/encoding/uchardet/cs/windows-1250.txt
similarity index 100%
rename from src/uchardet/uchardet/test/cs/windows-1250.txt
rename to test/test_files/encoding/uchardet/cs/windows-1250.txt
diff --git a/src/uchardet/uchardet/test/da/ibm865.txt b/test/test_files/encoding/uchardet/da/ibm865.txt
similarity index 100%
rename from src/uchardet/uchardet/test/da/ibm865.txt
rename to test/test_files/encoding/uchardet/da/ibm865.txt
diff --git a/src/uchardet/uchardet/test/da/iso-8859-1.txt b/test/test_files/encoding/uchardet/da/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/da/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/da/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/da/iso-8859-15.txt b/test/test_files/encoding/uchardet/da/iso-8859-15.txt
similarity index 100%
rename from src/uchardet/uchardet/test/da/iso-8859-15.txt
rename to test/test_files/encoding/uchardet/da/iso-8859-15.txt
diff --git a/src/uchardet/uchardet/test/da/utf-8.txt b/test/test_files/encoding/uchardet/da/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/da/utf-8.txt
rename to test/test_files/encoding/uchardet/da/utf-8.txt
diff --git a/src/uchardet/uchardet/test/da/windows-1252.txt b/test/test_files/encoding/uchardet/da/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/da/windows-1252.txt
rename to test/test_files/encoding/uchardet/da/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/de/iso-8859-1.txt b/test/test_files/encoding/uchardet/de/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/de/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/de/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/de/windows-1252.txt b/test/test_files/encoding/uchardet/de/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/de/windows-1252.txt
rename to test/test_files/encoding/uchardet/de/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/el/iso-8859-7.txt b/test/test_files/encoding/uchardet/el/iso-8859-7.txt
similarity index 100%
rename from src/uchardet/uchardet/test/el/iso-8859-7.txt
rename to test/test_files/encoding/uchardet/el/iso-8859-7.txt
diff --git a/src/uchardet/uchardet/test/el/utf-8.txt b/test/test_files/encoding/uchardet/el/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/el/utf-8.txt
rename to test/test_files/encoding/uchardet/el/utf-8.txt
diff --git a/src/uchardet/uchardet/test/el/windows-1253.txt b/test/test_files/encoding/uchardet/el/windows-1253.txt
similarity index 100%
rename from src/uchardet/uchardet/test/el/windows-1253.txt
rename to test/test_files/encoding/uchardet/el/windows-1253.txt
diff --git a/src/uchardet/uchardet/test/en/ascii.txt b/test/test_files/encoding/uchardet/en/ascii.txt
similarity index 100%
rename from src/uchardet/uchardet/test/en/ascii.txt
rename to test/test_files/encoding/uchardet/en/ascii.txt
diff --git a/src/uchardet/uchardet/test/eo/iso-8859-3.txt b/test/test_files/encoding/uchardet/eo/iso-8859-3.txt
similarity index 100%
rename from src/uchardet/uchardet/test/eo/iso-8859-3.txt
rename to test/test_files/encoding/uchardet/eo/iso-8859-3.txt
diff --git a/src/uchardet/uchardet/test/es/iso-8859-1.txt b/test/test_files/encoding/uchardet/es/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/es/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/es/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/es/iso-8859-15.txt b/test/test_files/encoding/uchardet/es/iso-8859-15.txt
similarity index 100%
rename from src/uchardet/uchardet/test/es/iso-8859-15.txt
rename to test/test_files/encoding/uchardet/es/iso-8859-15.txt
diff --git a/src/uchardet/uchardet/test/es/utf-8.txt b/test/test_files/encoding/uchardet/es/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/es/utf-8.txt
rename to test/test_files/encoding/uchardet/es/utf-8.txt
diff --git a/src/uchardet/uchardet/test/es/windows-1252.txt b/test/test_files/encoding/uchardet/es/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/es/windows-1252.txt
rename to test/test_files/encoding/uchardet/es/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/et/iso-8859-13.txt b/test/test_files/encoding/uchardet/et/iso-8859-13.txt
similarity index 100%
rename from src/uchardet/uchardet/test/et/iso-8859-13.txt
rename to test/test_files/encoding/uchardet/et/iso-8859-13.txt
diff --git a/src/uchardet/uchardet/test/et/iso-8859-15.txt b/test/test_files/encoding/uchardet/et/iso-8859-15.txt
similarity index 100%
rename from src/uchardet/uchardet/test/et/iso-8859-15.txt
rename to test/test_files/encoding/uchardet/et/iso-8859-15.txt
diff --git a/src/uchardet/uchardet/test/et/iso-8859-4.txt b/test/test_files/encoding/uchardet/et/iso-8859-4.txt
similarity index 100%
rename from src/uchardet/uchardet/test/et/iso-8859-4.txt
rename to test/test_files/encoding/uchardet/et/iso-8859-4.txt
diff --git a/src/uchardet/uchardet/test/et/utf-8.txt b/test/test_files/encoding/uchardet/et/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/et/utf-8.txt
rename to test/test_files/encoding/uchardet/et/utf-8.txt
diff --git a/src/uchardet/uchardet/test/et/windows-1252.txt b/test/test_files/encoding/uchardet/et/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/et/windows-1252.txt
rename to test/test_files/encoding/uchardet/et/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/et/windows-1257.txt b/test/test_files/encoding/uchardet/et/windows-1257.txt
similarity index 100%
rename from src/uchardet/uchardet/test/et/windows-1257.txt
rename to test/test_files/encoding/uchardet/et/windows-1257.txt
diff --git a/src/uchardet/uchardet/test/fi/iso-8859-1.txt b/test/test_files/encoding/uchardet/fi/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/fi/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/fi/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/fi/utf-8.txt b/test/test_files/encoding/uchardet/fi/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/fi/utf-8.txt
rename to test/test_files/encoding/uchardet/fi/utf-8.txt
diff --git a/src/uchardet/uchardet/test/fr/iso-8859-1.txt b/test/test_files/encoding/uchardet/fr/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/fr/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/fr/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/fr/iso-8859-15.txt b/test/test_files/encoding/uchardet/fr/iso-8859-15.txt
similarity index 100%
rename from src/uchardet/uchardet/test/fr/iso-8859-15.txt
rename to test/test_files/encoding/uchardet/fr/iso-8859-15.txt
diff --git a/src/uchardet/uchardet/test/fr/utf-16.be b/test/test_files/encoding/uchardet/fr/utf-16.be
similarity index 100%
rename from src/uchardet/uchardet/test/fr/utf-16.be
rename to test/test_files/encoding/uchardet/fr/utf-16.be
diff --git a/src/uchardet/uchardet/test/fr/utf-32.le b/test/test_files/encoding/uchardet/fr/utf-32.le
similarity index 100%
rename from src/uchardet/uchardet/test/fr/utf-32.le
rename to test/test_files/encoding/uchardet/fr/utf-32.le
diff --git a/src/uchardet/uchardet/test/fr/utf-8.txt b/test/test_files/encoding/uchardet/fr/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/fr/utf-8.txt
rename to test/test_files/encoding/uchardet/fr/utf-8.txt
diff --git a/src/uchardet/uchardet/test/fr/windows-1252.txt b/test/test_files/encoding/uchardet/fr/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/fr/windows-1252.txt
rename to test/test_files/encoding/uchardet/fr/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/ga/iso-8859-1.txt b/test/test_files/encoding/uchardet/ga/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ga/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/ga/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/ga/utf-8.txt b/test/test_files/encoding/uchardet/ga/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ga/utf-8.txt
rename to test/test_files/encoding/uchardet/ga/utf-8.txt
diff --git a/src/uchardet/uchardet/test/ga/windows-1252.txt b/test/test_files/encoding/uchardet/ga/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ga/windows-1252.txt
rename to test/test_files/encoding/uchardet/ga/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/he/iso-8859-8.txt b/test/test_files/encoding/uchardet/he/iso-8859-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/he/iso-8859-8.txt
rename to test/test_files/encoding/uchardet/he/iso-8859-8.txt
diff --git a/src/uchardet/uchardet/test/he/utf-8.txt b/test/test_files/encoding/uchardet/he/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/he/utf-8.txt
rename to test/test_files/encoding/uchardet/he/utf-8.txt
diff --git a/src/uchardet/uchardet/test/he/windows-1255.txt b/test/test_files/encoding/uchardet/he/windows-1255.txt
similarity index 100%
rename from src/uchardet/uchardet/test/he/windows-1255.txt
rename to test/test_files/encoding/uchardet/he/windows-1255.txt
diff --git a/src/uchardet/uchardet/test/hr/ibm852.txt b/test/test_files/encoding/uchardet/hr/ibm852.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hr/ibm852.txt
rename to test/test_files/encoding/uchardet/hr/ibm852.txt
diff --git a/src/uchardet/uchardet/test/hr/iso-8859-13.txt b/test/test_files/encoding/uchardet/hr/iso-8859-13.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hr/iso-8859-13.txt
rename to test/test_files/encoding/uchardet/hr/iso-8859-13.txt
diff --git a/src/uchardet/uchardet/test/hr/iso-8859-16.txt b/test/test_files/encoding/uchardet/hr/iso-8859-16.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hr/iso-8859-16.txt
rename to test/test_files/encoding/uchardet/hr/iso-8859-16.txt
diff --git a/src/uchardet/uchardet/test/hr/iso-8859-2.txt b/test/test_files/encoding/uchardet/hr/iso-8859-2.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hr/iso-8859-2.txt
rename to test/test_files/encoding/uchardet/hr/iso-8859-2.txt
diff --git a/src/uchardet/uchardet/test/hr/mac-centraleurope.txt b/test/test_files/encoding/uchardet/hr/mac-centraleurope.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hr/mac-centraleurope.txt
rename to test/test_files/encoding/uchardet/hr/mac-centraleurope.txt
diff --git a/src/uchardet/uchardet/test/hr/utf-8.txt b/test/test_files/encoding/uchardet/hr/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hr/utf-8.txt
rename to test/test_files/encoding/uchardet/hr/utf-8.txt
diff --git a/src/uchardet/uchardet/test/hr/windows-1250.txt b/test/test_files/encoding/uchardet/hr/windows-1250.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hr/windows-1250.txt
rename to test/test_files/encoding/uchardet/hr/windows-1250.txt
diff --git a/src/uchardet/uchardet/test/hu/iso-8859-2.txt b/test/test_files/encoding/uchardet/hu/iso-8859-2.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hu/iso-8859-2.txt
rename to test/test_files/encoding/uchardet/hu/iso-8859-2.txt
diff --git a/src/uchardet/uchardet/test/hu/windows-1250.txt b/test/test_files/encoding/uchardet/hu/windows-1250.txt
similarity index 100%
rename from src/uchardet/uchardet/test/hu/windows-1250.txt
rename to test/test_files/encoding/uchardet/hu/windows-1250.txt
diff --git a/src/uchardet/uchardet/test/it/iso-8859-1.txt b/test/test_files/encoding/uchardet/it/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/it/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/it/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/it/utf-8.txt b/test/test_files/encoding/uchardet/it/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/it/utf-8.txt
rename to test/test_files/encoding/uchardet/it/utf-8.txt
diff --git a/src/uchardet/uchardet/test/ja/euc-jp.txt b/test/test_files/encoding/uchardet/ja/euc-jp.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ja/euc-jp.txt
rename to test/test_files/encoding/uchardet/ja/euc-jp.txt
diff --git a/src/uchardet/uchardet/test/ja/iso-2022-jp.txt b/test/test_files/encoding/uchardet/ja/iso-2022-jp.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ja/iso-2022-jp.txt
rename to test/test_files/encoding/uchardet/ja/iso-2022-jp.txt
diff --git a/src/uchardet/uchardet/test/ja/shift_jis.txt b/test/test_files/encoding/uchardet/ja/shift_jis.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ja/shift_jis.txt
rename to test/test_files/encoding/uchardet/ja/shift_jis.txt
diff --git a/src/uchardet/uchardet/test/ja/utf-16be.txt b/test/test_files/encoding/uchardet/ja/utf-16be.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ja/utf-16be.txt
rename to test/test_files/encoding/uchardet/ja/utf-16be.txt
diff --git a/src/uchardet/uchardet/test/ja/utf-16le.txt b/test/test_files/encoding/uchardet/ja/utf-16le.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ja/utf-16le.txt
rename to test/test_files/encoding/uchardet/ja/utf-16le.txt
diff --git a/src/uchardet/uchardet/test/ja/utf-8.txt b/test/test_files/encoding/uchardet/ja/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ja/utf-8.txt
rename to test/test_files/encoding/uchardet/ja/utf-8.txt
diff --git a/src/uchardet/uchardet/test/ko/iso-2022-kr.txt b/test/test_files/encoding/uchardet/ko/iso-2022-kr.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ko/iso-2022-kr.txt
rename to test/test_files/encoding/uchardet/ko/iso-2022-kr.txt
diff --git a/src/uchardet/uchardet/test/ko/uhc.smi b/test/test_files/encoding/uchardet/ko/uhc.smi
similarity index 100%
rename from src/uchardet/uchardet/test/ko/uhc.smi
rename to test/test_files/encoding/uchardet/ko/uhc.smi
diff --git a/src/uchardet/uchardet/test/ko/utf-16.le b/test/test_files/encoding/uchardet/ko/utf-16.le
similarity index 100%
rename from src/uchardet/uchardet/test/ko/utf-16.le
rename to test/test_files/encoding/uchardet/ko/utf-16.le
diff --git a/src/uchardet/uchardet/test/ko/utf-32.be b/test/test_files/encoding/uchardet/ko/utf-32.be
similarity index 100%
rename from src/uchardet/uchardet/test/ko/utf-32.be
rename to test/test_files/encoding/uchardet/ko/utf-32.be
diff --git a/src/uchardet/uchardet/test/ko/utf-8.txt b/test/test_files/encoding/uchardet/ko/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ko/utf-8.txt
rename to test/test_files/encoding/uchardet/ko/utf-8.txt
diff --git a/src/uchardet/uchardet/test/lt/iso-8859-10.txt b/test/test_files/encoding/uchardet/lt/iso-8859-10.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lt/iso-8859-10.txt
rename to test/test_files/encoding/uchardet/lt/iso-8859-10.txt
diff --git a/src/uchardet/uchardet/test/lt/iso-8859-13.txt b/test/test_files/encoding/uchardet/lt/iso-8859-13.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lt/iso-8859-13.txt
rename to test/test_files/encoding/uchardet/lt/iso-8859-13.txt
diff --git a/src/uchardet/uchardet/test/lt/iso-8859-4.txt b/test/test_files/encoding/uchardet/lt/iso-8859-4.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lt/iso-8859-4.txt
rename to test/test_files/encoding/uchardet/lt/iso-8859-4.txt
diff --git a/src/uchardet/uchardet/test/lt/utf-8.txt b/test/test_files/encoding/uchardet/lt/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lt/utf-8.txt
rename to test/test_files/encoding/uchardet/lt/utf-8.txt
diff --git a/src/uchardet/uchardet/test/lv/iso-8859-10.txt b/test/test_files/encoding/uchardet/lv/iso-8859-10.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lv/iso-8859-10.txt
rename to test/test_files/encoding/uchardet/lv/iso-8859-10.txt
diff --git a/src/uchardet/uchardet/test/lv/iso-8859-13.txt b/test/test_files/encoding/uchardet/lv/iso-8859-13.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lv/iso-8859-13.txt
rename to test/test_files/encoding/uchardet/lv/iso-8859-13.txt
diff --git a/src/uchardet/uchardet/test/lv/iso-8859-4.txt b/test/test_files/encoding/uchardet/lv/iso-8859-4.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lv/iso-8859-4.txt
rename to test/test_files/encoding/uchardet/lv/iso-8859-4.txt
diff --git a/src/uchardet/uchardet/test/lv/utf-8.txt b/test/test_files/encoding/uchardet/lv/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/lv/utf-8.txt
rename to test/test_files/encoding/uchardet/lv/utf-8.txt
diff --git a/src/uchardet/uchardet/test/mt/iso-8859-3.txt b/test/test_files/encoding/uchardet/mt/iso-8859-3.txt
similarity index 100%
rename from src/uchardet/uchardet/test/mt/iso-8859-3.txt
rename to test/test_files/encoding/uchardet/mt/iso-8859-3.txt
diff --git a/src/uchardet/uchardet/test/mt/utf-8.txt b/test/test_files/encoding/uchardet/mt/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/mt/utf-8.txt
rename to test/test_files/encoding/uchardet/mt/utf-8.txt
diff --git a/src/uchardet/uchardet/test/no/ibm865.txt b/test/test_files/encoding/uchardet/no/ibm865.txt
similarity index 100%
rename from src/uchardet/uchardet/test/no/ibm865.txt
rename to test/test_files/encoding/uchardet/no/ibm865.txt
diff --git a/src/uchardet/uchardet/test/no/iso-8859-1.txt b/test/test_files/encoding/uchardet/no/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/no/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/no/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/no/iso-8859-15.txt b/test/test_files/encoding/uchardet/no/iso-8859-15.txt
similarity index 100%
rename from src/uchardet/uchardet/test/no/iso-8859-15.txt
rename to test/test_files/encoding/uchardet/no/iso-8859-15.txt
diff --git a/src/uchardet/uchardet/test/no/utf-8.txt b/test/test_files/encoding/uchardet/no/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/no/utf-8.txt
rename to test/test_files/encoding/uchardet/no/utf-8.txt
diff --git a/src/uchardet/uchardet/test/no/windows-1252.txt b/test/test_files/encoding/uchardet/no/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/no/windows-1252.txt
rename to test/test_files/encoding/uchardet/no/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/pl/ibm852.txt b/test/test_files/encoding/uchardet/pl/ibm852.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pl/ibm852.txt
rename to test/test_files/encoding/uchardet/pl/ibm852.txt
diff --git a/src/uchardet/uchardet/test/pl/iso-8859-13.txt b/test/test_files/encoding/uchardet/pl/iso-8859-13.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pl/iso-8859-13.txt
rename to test/test_files/encoding/uchardet/pl/iso-8859-13.txt
diff --git a/src/uchardet/uchardet/test/pl/iso-8859-16.txt b/test/test_files/encoding/uchardet/pl/iso-8859-16.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pl/iso-8859-16.txt
rename to test/test_files/encoding/uchardet/pl/iso-8859-16.txt
diff --git a/src/uchardet/uchardet/test/pl/iso-8859-2.txt b/test/test_files/encoding/uchardet/pl/iso-8859-2.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pl/iso-8859-2.txt
rename to test/test_files/encoding/uchardet/pl/iso-8859-2.txt
diff --git a/src/uchardet/uchardet/test/pl/mac-centraleurope.txt b/test/test_files/encoding/uchardet/pl/mac-centraleurope.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pl/mac-centraleurope.txt
rename to test/test_files/encoding/uchardet/pl/mac-centraleurope.txt
diff --git a/src/uchardet/uchardet/test/pl/utf-8.txt b/test/test_files/encoding/uchardet/pl/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pl/utf-8.txt
rename to test/test_files/encoding/uchardet/pl/utf-8.txt
diff --git a/src/uchardet/uchardet/test/pl/windows-1250.txt b/test/test_files/encoding/uchardet/pl/windows-1250.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pl/windows-1250.txt
rename to test/test_files/encoding/uchardet/pl/windows-1250.txt
diff --git a/src/uchardet/uchardet/test/pt/iso-8859-1.txt b/test/test_files/encoding/uchardet/pt/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pt/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/pt/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/pt/utf-8.txt b/test/test_files/encoding/uchardet/pt/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/pt/utf-8.txt
rename to test/test_files/encoding/uchardet/pt/utf-8.txt
diff --git a/src/uchardet/uchardet/test/ro/ibm852.txt b/test/test_files/encoding/uchardet/ro/ibm852.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ro/ibm852.txt
rename to test/test_files/encoding/uchardet/ro/ibm852.txt
diff --git a/src/uchardet/uchardet/test/ro/iso-8859-16.txt b/test/test_files/encoding/uchardet/ro/iso-8859-16.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ro/iso-8859-16.txt
rename to test/test_files/encoding/uchardet/ro/iso-8859-16.txt
diff --git a/src/uchardet/uchardet/test/ro/utf-8.txt b/test/test_files/encoding/uchardet/ro/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ro/utf-8.txt
rename to test/test_files/encoding/uchardet/ro/utf-8.txt
diff --git a/src/uchardet/uchardet/test/ro/windows-1250.txt b/test/test_files/encoding/uchardet/ro/windows-1250.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ro/windows-1250.txt
rename to test/test_files/encoding/uchardet/ro/windows-1250.txt
diff --git a/src/uchardet/uchardet/test/ru/ibm855.txt b/test/test_files/encoding/uchardet/ru/ibm855.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ru/ibm855.txt
rename to test/test_files/encoding/uchardet/ru/ibm855.txt
diff --git a/src/uchardet/uchardet/test/ru/ibm866.txt b/test/test_files/encoding/uchardet/ru/ibm866.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ru/ibm866.txt
rename to test/test_files/encoding/uchardet/ru/ibm866.txt
diff --git a/src/uchardet/uchardet/test/ru/iso-8859-5.txt b/test/test_files/encoding/uchardet/ru/iso-8859-5.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ru/iso-8859-5.txt
rename to test/test_files/encoding/uchardet/ru/iso-8859-5.txt
diff --git a/src/uchardet/uchardet/test/ru/koi8-r.txt b/test/test_files/encoding/uchardet/ru/koi8-r.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ru/koi8-r.txt
rename to test/test_files/encoding/uchardet/ru/koi8-r.txt
diff --git a/src/uchardet/uchardet/test/ru/mac-cyrillic.txt b/test/test_files/encoding/uchardet/ru/mac-cyrillic.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ru/mac-cyrillic.txt
rename to test/test_files/encoding/uchardet/ru/mac-cyrillic.txt
diff --git a/src/uchardet/uchardet/test/ru/windows-1251.txt b/test/test_files/encoding/uchardet/ru/windows-1251.txt
similarity index 100%
rename from src/uchardet/uchardet/test/ru/windows-1251.txt
rename to test/test_files/encoding/uchardet/ru/windows-1251.txt
diff --git a/src/uchardet/uchardet/test/sk/ibm852.txt b/test/test_files/encoding/uchardet/sk/ibm852.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sk/ibm852.txt
rename to test/test_files/encoding/uchardet/sk/ibm852.txt
diff --git a/src/uchardet/uchardet/test/sk/iso-8859-2.txt b/test/test_files/encoding/uchardet/sk/iso-8859-2.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sk/iso-8859-2.txt
rename to test/test_files/encoding/uchardet/sk/iso-8859-2.txt
diff --git a/src/uchardet/uchardet/test/sk/mac-centraleurope.txt b/test/test_files/encoding/uchardet/sk/mac-centraleurope.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sk/mac-centraleurope.txt
rename to test/test_files/encoding/uchardet/sk/mac-centraleurope.txt
diff --git a/src/uchardet/uchardet/test/sk/utf-8.txt b/test/test_files/encoding/uchardet/sk/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sk/utf-8.txt
rename to test/test_files/encoding/uchardet/sk/utf-8.txt
diff --git a/src/uchardet/uchardet/test/sk/windows-1250.txt b/test/test_files/encoding/uchardet/sk/windows-1250.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sk/windows-1250.txt
rename to test/test_files/encoding/uchardet/sk/windows-1250.txt
diff --git a/src/uchardet/uchardet/test/sl/ibm852.txt b/test/test_files/encoding/uchardet/sl/ibm852.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sl/ibm852.txt
rename to test/test_files/encoding/uchardet/sl/ibm852.txt
diff --git a/src/uchardet/uchardet/test/sl/iso-8859-16.txt b/test/test_files/encoding/uchardet/sl/iso-8859-16.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sl/iso-8859-16.txt
rename to test/test_files/encoding/uchardet/sl/iso-8859-16.txt
diff --git a/src/uchardet/uchardet/test/sl/iso-8859-2.txt b/test/test_files/encoding/uchardet/sl/iso-8859-2.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sl/iso-8859-2.txt
rename to test/test_files/encoding/uchardet/sl/iso-8859-2.txt
diff --git a/src/uchardet/uchardet/test/sl/mac-centraleurope.txt b/test/test_files/encoding/uchardet/sl/mac-centraleurope.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sl/mac-centraleurope.txt
rename to test/test_files/encoding/uchardet/sl/mac-centraleurope.txt
diff --git a/src/uchardet/uchardet/test/sl/utf-8.txt b/test/test_files/encoding/uchardet/sl/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sl/utf-8.txt
rename to test/test_files/encoding/uchardet/sl/utf-8.txt
diff --git a/src/uchardet/uchardet/test/sl/windows-1250.txt b/test/test_files/encoding/uchardet/sl/windows-1250.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sl/windows-1250.txt
rename to test/test_files/encoding/uchardet/sl/windows-1250.txt
diff --git a/src/uchardet/uchardet/test/sv/iso-8859-1.txt b/test/test_files/encoding/uchardet/sv/iso-8859-1.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sv/iso-8859-1.txt
rename to test/test_files/encoding/uchardet/sv/iso-8859-1.txt
diff --git a/src/uchardet/uchardet/test/sv/utf-8.txt b/test/test_files/encoding/uchardet/sv/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sv/utf-8.txt
rename to test/test_files/encoding/uchardet/sv/utf-8.txt
diff --git a/src/uchardet/uchardet/test/sv/windows-1252.txt b/test/test_files/encoding/uchardet/sv/windows-1252.txt
similarity index 100%
rename from src/uchardet/uchardet/test/sv/windows-1252.txt
rename to test/test_files/encoding/uchardet/sv/windows-1252.txt
diff --git a/src/uchardet/uchardet/test/th/iso-8859-11.txt b/test/test_files/encoding/uchardet/th/iso-8859-11.txt
similarity index 100%
rename from src/uchardet/uchardet/test/th/iso-8859-11.txt
rename to test/test_files/encoding/uchardet/th/iso-8859-11.txt
diff --git a/src/uchardet/uchardet/test/th/tis-620.txt b/test/test_files/encoding/uchardet/th/tis-620.txt
similarity index 100%
rename from src/uchardet/uchardet/test/th/tis-620.txt
rename to test/test_files/encoding/uchardet/th/tis-620.txt
diff --git a/src/uchardet/uchardet/test/th/utf-8.txt b/test/test_files/encoding/uchardet/th/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/th/utf-8.txt
rename to test/test_files/encoding/uchardet/th/utf-8.txt
diff --git a/src/uchardet/uchardet/test/tr/iso-8859-3.txt b/test/test_files/encoding/uchardet/tr/iso-8859-3.txt
similarity index 100%
rename from src/uchardet/uchardet/test/tr/iso-8859-3.txt
rename to test/test_files/encoding/uchardet/tr/iso-8859-3.txt
diff --git a/src/uchardet/uchardet/test/tr/iso-8859-9.txt b/test/test_files/encoding/uchardet/tr/iso-8859-9.txt
similarity index 100%
rename from src/uchardet/uchardet/test/tr/iso-8859-9.txt
rename to test/test_files/encoding/uchardet/tr/iso-8859-9.txt
diff --git a/src/uchardet/uchardet/test/uchardet-tests.c b/test/test_files/encoding/uchardet/uchardet-tests.c
similarity index 100%
rename from src/uchardet/uchardet/test/uchardet-tests.c
rename to test/test_files/encoding/uchardet/uchardet-tests.c
diff --git a/src/uchardet/uchardet/test/vi/utf-8.txt b/test/test_files/encoding/uchardet/vi/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/vi/utf-8.txt
rename to test/test_files/encoding/uchardet/vi/utf-8.txt
diff --git a/src/uchardet/uchardet/test/vi/viscii.txt b/test/test_files/encoding/uchardet/vi/viscii.txt
similarity index 100%
rename from src/uchardet/uchardet/test/vi/viscii.txt
rename to test/test_files/encoding/uchardet/vi/viscii.txt
diff --git a/src/uchardet/uchardet/test/vi/windows-1258.txt b/test/test_files/encoding/uchardet/vi/windows-1258.txt
similarity index 100%
rename from src/uchardet/uchardet/test/vi/windows-1258.txt
rename to test/test_files/encoding/uchardet/vi/windows-1258.txt
diff --git a/src/uchardet/uchardet/test/zh/big5.txt b/test/test_files/encoding/uchardet/zh/big5.txt
similarity index 100%
rename from src/uchardet/uchardet/test/zh/big5.txt
rename to test/test_files/encoding/uchardet/zh/big5.txt
diff --git a/src/uchardet/uchardet/test/zh/euc-tw.txt b/test/test_files/encoding/uchardet/zh/euc-tw.txt
similarity index 100%
rename from src/uchardet/uchardet/test/zh/euc-tw.txt
rename to test/test_files/encoding/uchardet/zh/euc-tw.txt
diff --git a/src/uchardet/uchardet/test/zh/gb18030.txt b/test/test_files/encoding/uchardet/zh/gb18030.txt
similarity index 100%
rename from src/uchardet/uchardet/test/zh/gb18030.txt
rename to test/test_files/encoding/uchardet/zh/gb18030.txt
diff --git a/src/uchardet/uchardet/test/zh/utf-8.txt b/test/test_files/encoding/uchardet/zh/utf-8.txt
similarity index 100%
rename from src/uchardet/uchardet/test/zh/utf-8.txt
rename to test/test_files/encoding/uchardet/zh/utf-8.txt