From 75931feabe99595e9659a423e299c4229d3c02ba Mon Sep 17 00:00:00 2001 From: Alexander Barkov Date: Tue, 14 Jul 2015 12:00:05 +0400 Subject: MDEV-8362 dash '-' is not recognized in charset armscii8 on select where query --- strings/ctype-simple.c | 23 ++++++++++++++++++++++- 1 file changed, 22 insertions(+), 1 deletion(-) (limited to 'strings') diff --git a/strings/ctype-simple.c b/strings/ctype-simple.c index d7a1b3f33b4..394924c8209 100644 --- a/strings/ctype-simple.c +++ b/strings/ctype-simple.c @@ -1303,7 +1303,28 @@ create_fromuni(struct charset_info_st *cs, if (wc >= idx[i].uidx.from && wc <= idx[i].uidx.to && wc) { int ofs= wc - idx[i].uidx.from; - tab[ofs]= ch; + if (!tab[ofs] || tab[ofs] > 0x7F) /* Prefer ASCII*/ + { + /* + Some character sets can have double encoding. For example, + in ARMSCII8, the following characters are encoded twice: + + Encoding#1 Encoding#2 Unicode Character Name + ---------- ---------- ------- -------------- + 0x27 0xFF U+0027 APOSTROPHE + 0x28 0xA5 U+0028 LEFT PARENTHESIS + 0x29 0xA4 U+0029 RIGHT PARENTHESIS + 0x2C 0xAB U+002C COMMA + 0x2D 0xAC U+002D HYPHEN-MINUS + 0x2E 0xA9 U+002E FULL STOP + + That is, both 0x27 and 0xFF convert to Unicode U+0027. + When converting back from Unicode to ARMSCII, + we prefer the ASCII range, that is we want U+0027 + to convert to 0x27 rather than to 0xFF. + */ + tab[ofs]= ch; + } } } } -- cgit v1.2.1