From 45f3df375950ac7366c15258684e98a01ca5de6a Mon Sep 17 00:00:00 2001
From: Koichi Murase <myoga.murase@gmail.com>
Date: Sat, 24 Sep 2022 11:25:35 +0900
Subject: [PATCH] encoding: add minor fixes and workarounds

* util (ble/encoding:UTF-8/b2c): fix interpretation of leading byte
* util (ble/util/s2c): work around intermediate mbstate of bash <= 5.2
---
 src/util.sh | 35 +++++++++++++++++++++++------------
 1 file changed, 23 insertions(+), 12 deletions(-)

diff --git a/src/util.sh b/src/util.sh
index 705690ea..59a73417 100644
--- a/src/util.sh
+++ b/src/util.sh
@@ -3563,11 +3563,22 @@ _ble_util_s2c_table_enabled=
 ##   @param[in] text
 ##   @param[in,opt] index
 ##   @var[out] ret
-if ((_ble_bash>=40100)); then
-  # - printf "'c" で Unicode が読める (どの LC_CTYPE でも Unicode になる)
+if ((_ble_bash>=50300)); then
+  # printf "'c" で Unicode が読める (どの LC_CTYPE でも Unicode になる)
   function ble/util/s2c {
     builtin printf -v ret '%d' "'${1:$2:1}"
   }
+elif ((_ble_bash>=40100)); then
+  function ble/util/s2c {
+    # Note #D1881: bash-5.2 以前では printf %d "'x" に対して mbstate_t 状態が
+    # 残ってしまう。なので一旦 clear を試みる。
+    if ble/util/is-unicode-output; then
+      builtin printf -v ret %d "'μ"
+    else
+      builtin printf -v ret %d "'x"
+    fi
+    builtin printf -v ret %d "'$1"
+  }
 elif ((_ble_bash>=40000&&!_ble_bash_loaded_in_function)); then
   # - 連想配列にキャッシュできる
   # - printf "'c" で unicode が読める
@@ -3627,7 +3638,7 @@ if ((_ble_bash>=40200)); then
   # workarounds of bashbug that printf '\uFFFF' results in a broken surrogate
   # pair in systems where sizeof(wchar_t) == 2.
   function ble/util/.has-bashbug-printf-uffff {
-    ((40200<=_ble_bash&&_ble_bash<40500)) || return 1
+    ((40200<=_ble_bash&&_ble_bash<50000)) || return 1
     local LC_ALL=C.UTF-8 2>/dev/null # Workaround: CentOS 7 に C.UTF-8 がなかった
     local ret
     builtin printf -v ret '\uFFFF'
@@ -3732,6 +3743,12 @@ function ble/util/.cache/update-locale {
   fi
 }
 
+function ble/util/is-unicode-output {
+  [[ $_ble_util_locale_triple != "$LC_ALL:$LC_CTYPE:$LANG" ]] &&
+    ble/util/.cache/update-locale
+  [[ $_ble_util_locale_encoding == UTF-8 ]]
+}
+
 #------------------------------------------------------------------------------
 
 ## 関数 ble/util/s2chars text
@@ -3849,9 +3866,9 @@ function ble/encoding:UTF-8/b2c {
   bytes=("$@")
   ret=0
   ((b0=bytes[0]&0xFF))
-  ((n=b0>0xF0
-    ?(b0>0xFC?5:(b0>0xF8?4:3))
-    :(b0>0xE0?2:(b0>0xC0?1:0)),
+  ((n=b0>=0xF0
+    ?(b0>=0xFC?5:(b0>=0xF8?4:3))
+    :(b0>=0xE0?2:(b0>=0xC0?1:0)),
     ret=n?b0&0x7F>>n:b0))
   for ((i=1;i<=n;i++)); do
     ((ret=ret<<6|0x3F&bytes[i]))
@@ -3892,9 +3909,3 @@ function ble/encoding:C/c2b {
   local code=$1
   bytes=($((code&0xFF)))
 }
-
-function ble/util/is-unicode-output {
-  [[ $_ble_util_locale_triple != "$LC_ALL:$LC_CTYPE:$LANG" ]] &&
-    ble/util/.cache/update-locale
-  [[ $_ble_util_locale_encoding == UTF-8 ]]
-}