From 4e02a284ebbf190275d6e3f5e8cf4dc08511e5e1 Mon Sep 17 00:00:00 2001
From: Vincent Bernat <vincent@bernat.im>
Date: Tue, 15 Dec 2015 07:47:35 +0100
Subject: [PATCH] Ensure LC_CTYPE uses UTF-8

See #2012 for some context. Many unicode-related tasks need that
coaelescing to handle unicode correctly. If for some reason, the user is
using a not using a non-unicode locale, we want to ensure that at least
LC_CTYPE is set to an unicode locale.

It's difficult to know if a locale is an unicode one, so we assume that
we want UTF-8 as an unicode codec. Aliases are not handled, so LC_CTYPE
may be redefined even if it was already unicode aware.

The first available UTF-8 locale is used because there is no guarantee
that C.UTF-8 or en_US.UTF-8 are available. For LC_CTYPE, there is no
difference between those locales as coaelescing is not dependant on the
language.

LC_CTYPE is exported as oh-my-zsh may use some shell tools at some
point (sed, awk, ...).
---
 lib/misc.zsh | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

diff --git a/lib/misc.zsh b/lib/misc.zsh
index c81dab41..782f8708 100644
--- a/lib/misc.zsh
+++ b/lib/misc.zsh
@@ -34,9 +34,20 @@ else
 fi
 
 # only define LC_CTYPE if undefined
-if [[ -z "$LC_CTYPE" && -z "$LC_ALL" ]]; then
-	export LC_CTYPE=${LANG%%:*} # pick the first entry from LANG
-fi
+LC_CTYPE=${LC_CTYPE:-${LC_ALL:-${LANG:-C}}}
+case $LC_CTYPE in
+   *.utf8|*.UTF-8)
+    # All is correct
+    ;;
+   *)
+    # Need to select an UTF-8 locale
+    local -a available
+    available=("${(f)$(locale -a 2> /dev/null)}")
+    export LC_CTYPE=${${${${(@M)available:#*.UTF-8}[1]}:-${${(@M)available:#*.utf8}[1]}}:-C.UTF-8}
+    unset available
+   ;;
+esac
+export LC_CTYPE
 
 # recognize comments
 setopt interactivecomments