コミット: d53c4f3350f9e85274596ad07229a0ffe8d3130c - pukiwiki15_devel (git)

コミットメタ情報

• リビジョン: d53c4f3350f9e85274596ad07229a0ffe8d3130c (tree)
• 日時: 2014-12-22 23:41:59
• 作者: umorigu <umorigu@gmai...>
• コミッター: umorigu

ログメッセージ

Implement converting UTF-8 source code to EUC-JP dist files

Convert option sample:

for UTF-8 to UTF-8:

$ ./release.sh --utf8

for UTF-8 to EUC-JP:

$ ./release.sh --eucjp

for EUC-JP to UTF-8:

$ ./release.sh --utf8 --fromeucjp

for EUC-JP to EUC-JP:

$ ./release.sh --eucjp --fromeucjp

変更サマリ

• modified: release.sh

差分

--- a/release.sh
+++ b/release.sh
@@ -18,7 +18,9 @@ usage(){
   warn  "    --nopkg     Suppress creating archive (Extract and chmod only)"
   warn  "    --norm      --nopkg, and remove nothing (.cvsignore etc)"
   warn  "    --co        --norm, and use 'checkout' command instead of 'export'"
-  warn  "    --utf8      Create UTF-8 converted archive (EXPERIMENTAL)"
+  warn  "    --utf8      Create UTF-8 converted archive"
+  warn  "    --eucjp     Create EUC_JP converted archive"
+  warn  "    --fromeucjp Indicate source charset is EUC_JP"
   warn  "    -z|--zip    Create *.zip archive"
   warn  "    --move-dist Move *.ini.php => *.ini-dist.php"
   warn  "    --copy-dist Move, and Copy *.ini.php <= *.ini-dist.php"
@@ -88,6 +90,8 @@ getopt(){ _arg=noarg
   --co|--checkout   ) echo _nopkg _noremove _checkout 1 ;;
   -z|--zip     ) echo _zip 1      ;;
   --ut|--utf|--utf8|--utf-8 ) echo _utf8 1  ;;
+  --eucjp ) echo _eucjp 1 ;;
+  --fromeucjp ) echo _fromeucjp 1 ;;
   --copy-dist  ) echo _copy_dist 1 ;;
   --move-dist  ) echo _move_dist 1 ;;
   -d  ) echo _CVSROOT 2 ; _arg="$2" ;;
@@ -134,7 +138,12 @@ if [ $# -eq 0 ] ; then usage ; exit ; fi
 
 # Utility check ---------------------------------------------
 
-if [ "$__utf8" ] ; then
+# Convert UTF-8 -> EUC_JP or EUC_JP -> UTF-8
+if [ \( "$__eucjp" -a ! "$__fromeucjp" \) -o \( "$__utf8" -a "$__fromeucjp" \) ] ; then
+  __charconv=1
+fi
+
+if [ "$__charconv" ] ; then
 
   # nkf
   which nkf || err "nkf version 2.0 or later (UTF-8 enabled) not found"
@@ -153,9 +162,13 @@ if [ "$__utf8" ] ; then
   fi
 
   convert(){
+    __nkfopt="--ic=eucJP-ms -w"
+    if [ "$__eucjp" ] ; then
+      __nkfopt="--ic=UTF8 --oc=eucJP-ms"
+    fi
     for list in "$@" ; do
       # NOTE: Specify '-E'(From EUC-JP) otherwise skin file will be collapsed
-      nkf --ic=eucJP-ms -w "$list" > "$list.$$.tmp" && mv "$list.$$.tmp" "$list" && echo "  $list"
+      nkf $__nkfopt "$list" > "$list.$$.tmp" && mv "$list.$$.tmp" "$list" && echo "  $list"
     done
   }
   convert_EUCJP2UTF8(){
@@ -165,6 +178,13 @@ if [ "$__utf8" ] ; then
       sed 's#^//UTF-8:##' "$list" > "$list.$$.tmp" && mv "$list.$$.tmp" "$list"
     done
   }
+  convert_UTF82EUCJP(){
+    for list in "$@" ; do
+      # Very rough conversion!
+      #sed 's/EUC-JP/UTF-8/g' "$list" > "$list.$$.tmp" && mv "$list.$$.tmp" "$list"
+      sed 's#^\(.*//UTF-8 only\)$#//UTF-8:\0#' "$list" > "$list.$$.tmp" && mv "$list.$$.tmp" "$list"
+    done
+  }
 fi > /dev/null
 
 if [ -z "$__zip" ]
@@ -241,8 +261,12 @@ mv $pkg_dir/cache/_recent.dat $pkg_dir/cache/recent.dat
 
 # Conversion ------------------------------------------------
 
-if [ "$__utf8" ] ; then
-  echo "Converting EUC-JP => UTF-8 ..."
+if [ "$__charconv" ] ; then
+  if [ "$__utf8" ] ; then
+    echo "Converting EUC-JP => UTF-8 ..."
+  else
+    echo "Converting UTF-8 => EUC-JP ..."
+  fi
   find "$pkg_dir" -type f \( -name "*.txt" -or -name "*.php" -or -name "*.lng"  -or -name "*.dat" -or -name "*.ref" \) |
   while read line; do
     case "$line" in
@@ -251,10 +275,17 @@ if [ "$__utf8" ] ; then
     esac
   done
 
-  # Replace 'EUC-JP' => 'UTF-8'
-  ( cd "$pkg_dir" &&
-    convert_EUCJP2UTF8 lib/init.php skin/pukiwiki.skin*.php
-  )
+  if [ "$__utf8" ] ; then
+    # Replace 'EUC-JP' => 'UTF-8'
+    ( cd "$pkg_dir" &&
+      convert_EUCJP2UTF8 lib/init.php skin/pukiwiki.skin*.php
+    )
+  else
+    # Replace 'UTF-8' => 'EUC-JP'
+    ( cd "$pkg_dir" &&
+      convert_UTF82EUCJP lib/init.php skin/pukiwiki.skin*.php
+    )
+  fi
 
   # Filename encoded 'encoded-EUC-JP' to 'encoded-UTF-8'
   echo "Renaming encoded-EUC-JP => encoded-UTF-8 ..."