git-wad

commit a9dbcb373c61451ff85af5672bf4621f9cf461e8
parent f8d1a1a287194e97a95e4c8cd4feb475c67d4c0e
Author: Vincent Forest <vincent.forest@meso-star.com>
Date:   Sun, 12 Jan 2025 15:14:42 +0100

Force default character type when using sed

Avoid the illegal byte sequence problem reported by some sed
implementations when using system locale settings.

Diffstat:
M
git-wad
 | 
21
+++++++++++++--------

1 file changed, 13 insertions(+), 8 deletions(-)
diff --git a/git-wad b/git-wad
@@ -30,6 +30,11 @@ if [ "${is_inside_git_dir}" = "true" ]; then
   die
 fi
 
+# Force default locale when using sed, i.e. treat all characters as
+# individual bytes. Otherwise, some sed implementations return
+# multi-byte encoding errors.
+alias sed__='LC_CTYPE=C sed'
+
 # shellcheck disable=SC2310
 working_tree="$(git rev-parse --show-toplevel)" || die "$?"
 git_wad_tmpdir="$(mktemp -d "${TMPDIR:-/tmp}/git_wad_XXXXXX")" || die "$?"
@@ -104,9 +109,9 @@ wad_hashes()
    git rev-list --objects HEAD -1 \
  | cut -d' ' -f1 \
  | git cat-file --batch-check \
- | sed -n 's/\(^[a-z0-9]\{40\}\) blob [0-9]\{1,\}$/\1/p' \
+ | sed__ -n 's/\(^[a-z0-9]\{40\}\) blob [0-9]\{1,\}$/\1/p' \
  | git cat-file --batch='%(objectname)' \
- | sed -n "/^[0-9a-z]\{40\}$/{N;s/^\([0-9a-z]\{40\}\)\n${wad_re}$/\1/p}"
+ | sed__ -n "/^[0-9a-z]\{40\}$/{N;s/^\([0-9a-z]\{40\}\)\n${wad_re}$/\1/p}"
 }
 
 # List paths of WAD file from the current commit
@@ -156,16 +161,16 @@ wad_objects() # [-1a]
    git rev-list --objects ${rev} \
  | cut -d' ' -f1 \
  | git cat-file --batch-check \
- | sed -n 's/\(^[a-z0-9]\{40\}\) blob [0-9]\{1,\}$/\1/p' \
+ | sed__ -n 's/\(^[a-z0-9]\{40\}\) blob [0-9]\{1,\}$/\1/p' \
  | git cat-file --batch \
- | sed -n "s/^${GIT_WAD_HEADER} \([0-9a-z]\{64\}\) [0-9]\{1,\}$/\1/p"
+ | sed__ -n "s/^${GIT_WAD_HEADER} \([0-9a-z]\{64\}\) [0-9]\{1,\}$/\1/p"
 }
 
 # List WAD objects staged for the next commit
 wad_objects_staged()
 {
     git diff --cached \
-  | sed -n "s/^+${GIT_WAD_HEADER} \([0-9a-z]\{64\}\) [0-9]\{1,\}$/\1/p"
+  | sed__ -n "s/^+${GIT_WAD_HEADER} \([0-9a-z]\{64\}\) [0-9]\{1,\}$/\1/p"
 }
 
 # List all stored WAD objects of the working tree. These may be WAD
@@ -230,7 +235,7 @@ next_timestamp()
 restore() # WAD file
 {
   wad="$1"
-  digest=$(sed "s/^${GIT_WAD_HEADER} \([0-9a-z]\{64\}\) [0-9]\{1,\}$/\1/" "${wad}")
+  digest=$(sed__ "s/^${GIT_WAD_HEADER} \([0-9a-z]\{64\}\) [0-9]\{1,\}$/\1/" "${wad}")
 
   if [ -z "${digest}" ]; then
     >&2 printf "Invalid WAD file %s\n" "$1"
@@ -319,7 +324,7 @@ smudge() # stdin
 
   else  # It is a WAD
     # The sed directive remove the space before the digest
-    digest_size="$(cat - | sed "1s/^ //")"
+    digest_size="$(cat - | sed__ "1s/^ //")"
 
     digest="$(echo "${digest_size}" | cut -d' ' -f1)"
     size="$(echo "${digest_size}" | cut -d' ' -f2)"
@@ -496,7 +501,7 @@ status() # [-1a]
 
     else
       # Read the WAD digest from its header
-      digest=$(sed \
+      digest=$(sed__ \
         -e "1s/^${GIT_WAD_HEADER} \([0-9a-z]\{64\}\) [0-9]\{1,\}$/\1/" \
         "${wad}")