From 59e70ab7c156fb22855d91d7541db307749ff259 Mon Sep 17 00:00:00 2001
From: Miao Wang <shankerwangmiao@gmail.com>
Date: Wed, 11 Oct 2023 22:34:39 +0800
Subject: [PATCH] git-recursive: change the way to iterate submoudles

---
 git-recursive.sh | 107 ++++++++++++++++++++++++-----------------------
 1 file changed, 55 insertions(+), 52 deletions(-)

diff --git a/git-recursive.sh b/git-recursive.sh
index ff6ff6d..42d7ea8 100755
--- a/git-recursive.sh
+++ b/git-recursive.sh
@@ -27,7 +27,7 @@ fi
 depth=0
 
 function echon() {
-  echo depth $depth $@
+  echo depth "$depth" "$@"
 }
 
 function repo_init() {
@@ -42,7 +42,7 @@ function update_linux_git() {
   cd "$working_dir"
   echon "==== SYNC $upstream START ===="
   git remote set-url origin "$upstream"
-  /usr/bin/timeout -s INT 3600 git remote -v update -p
+  "timeout" -s INT 3600 git remote -v update -p
   local ret=$?
   [[ $ret -ne 0 ]] && echon "git update failed with rc=$ret"
   local head=$(git remote show origin | awk '/HEAD branch:/ {print $NF}')
@@ -60,60 +60,63 @@ function git_sync() {
   local working_dir=$2
   if [[ ! -f "$working_dir/HEAD" ]]; then
     echon "Initializing $upstream mirror"
-    repo_init $1 $2
+    repo_init "$upstream" "$working_dir"
     return $?
   fi
-  update_linux_git $1 $2
+  update_linux_git "$upstream" "$working_dir"
 }
 
 function checkout_repo() {
   local repo_dir="$1"
   local work_tree="$2"
   local commit="$3"
-  echon "Checkout $repo_dir to $work_tree"
-  rm -rf $work_tree
-  git clone "$repo_dir" "$work_tree"
-  local ret=$?
-  if [[ $ret -ne 0 ]]; then
-    echon "git clone failed with rc=$ret"
-    return $ret
-  fi
-  if [[ ! -z "$commit" ]]; then
-    echon "Worktree $work_tree switch to commit $commit"
-    git -C $work_tree checkout $commit
-  fi
 
+  if [[ -z "$commit" ]]; then
+    commit="HEAD"
+  fi
+  echon "Considering $repo_dir on commit $commit"
+  local commit_obj_type=$(git -C "$repo_dir" cat-file -t "$commit")
+  local rc=$?
+  if [[ $rc -ne 0 ]] || [[ "$commit_obj_type" != "commit" ]]; then
+    echon "Commit $commit is not a valid commit object"
+    return 1
+  fi
   local repo_dir_no_git=${repo_dir%%.git}
-  local gitmodules="$work_tree/.gitmodules"
-  if [[ -f "$gitmodules" ]]; then
-    local paths_str=$(cat $gitmodules | grep path | cut -d '=' -f 2 | sed 's/^[[:blank:]]*//')
-    local urls_str=$(cat $gitmodules | grep url | cut -d '=' -f 2 | sed 's/^[[:blank:]]*//')
-    local -a paths
-    local -a urls
-    readarray -t paths <<<"$paths_str"
-    readarray -t urls <<<"$urls_str"
-    local -i i
-    for ((i=0;i<${#paths[@]};i++)); do
-      local path=${paths[$i]}
-      # ignore empty .gitmodules
-      if [[ "$path" == "" ]]; then
+  if git -C "$repo_dir" cat-file -e "$commit:.gitmodules" 2>/dev/null; then
+    echon "Find submodules for $repo_dir"
+    local -a submodules
+    IFS= readarray -d '' submodules < <(
+      git -C "$repo_dir" config --null --blob "$commit:.gitmodules" --name-only --get-regexp "^submodule\." | sed --null-data 's/\.[^.]*$//' | sort --zero-terminated --unique
+    )
+    for submoudle in "${submodules[@]}"; do
+      local submodule_path=$(git -C "$repo_dir" config --blob "$commit:.gitmodules" --get "$submoudle.path")
+      local submodule_url=$(git -C "$repo_dir" config --blob "$commit:.gitmodules" --get "$submoudle.url")
+      if [[ -z "$submodule_path" ]] || [[ -z "$submodule_url" ]]; then
         continue
       fi
-      local commit=$(git -C $work_tree submodule status $path | cut -d ' ' -f 1 | cut -d '-' -f 2)
-      local git_path=$repo_dir_no_git/$path.git
-      mkdir -p $git_path
-      local mirror_url=$(echo $git_path | sed "s#$WORKING_DIR_BASE#$MIRROR_BASE_URL#")
+      local submodule_path_parent=$(dirname -- "$submodule_path")
+      if [[ "$submodule_path_parent" = "." ]]; then
+        submodule_path_parent=""
+      fi
+      local submodule_commit=$(git -C "$repo_dir" ls-tree -d -z --format="%(path)/%(objectname)" "$commit:$submodule_path_parent" | fgrep --null-data "$(basename -- "$submodule_path")/" | head --zero-terminated --lines 1 | cut -d '/' -f 2 | tr -d '\0' )
+      if [[ -z "$submodule_commit" ]]; then
+        echon "Cannot find submodule commit for $submodule_path"
+        continue
+      fi
+      local submodule_git_path="$repo_dir_no_git/$submodule_path.git"
+      mkdir -p -- "$submodule_git_path"
+      local submodule_mirror_url=$(echo "$submodule_git_path" | sed "s#$WORKING_DIR_BASE#$MIRROR_BASE_URL#")
       script_append "cat >>.git/config <<EOF"
-      script_append "[submodule \"$path\"]"
+      script_append "[submodule \"$submodule_path\"]"
       script_append "	active = true"
-      script_append "	url = $mirror_url"
+      script_append "	url = $submodule_mirror_url"
       script_append "EOF"
-      script_append "mkdir -p $path"
-      script_append "git clone $mirror_url $path"
-      script_append "git submodule init $path"
-      script_append "git submodule update $path"
-      script_append "pushd $path"
-      git_sync_recursive ${urls[$i]} $git_path $commit
+      script_append "mkdir -p ${submodule_path@Q}"
+      script_append "git clone ${submodule_mirror_url@Q} ${submodule_path@Q}"
+      script_append "git submodule init ${submodule_path@Q}"
+      script_append "git submodule update ${submodule_path@Q}"
+      script_append "pushd ${submodule_path@Q}"
+      git_sync_recursive "$submodule_url" "$submodule_git_path" "$submodule_commit"
       script_append "popd"
     done
   fi
@@ -125,7 +128,7 @@ function git_sync_recursive() {
   local working_dir=$2
   # it is ok that commit=""
   local commit=$3
-  git_sync $1 $2
+  git_sync "$upstream" "$working_dir"
   local ret=$?
   if [[ $ret -ne 0 ]]; then
     echon "git sync failed with rc=$ret"
@@ -133,25 +136,25 @@ function git_sync_recursive() {
   fi
 
   if [[ ! -z "$RECURSIVE" ]]; then
-    working_dir_name=$(basename $2)
+    working_dir_name=$(basename -- "$working_dir")
     working_dir_name_no_git=${working_dir_name%%.git}
-    checkout_repo $working_dir $TMPDIR/$working_dir_name_no_git $commit
+    checkout_repo "$working_dir" "$TMPDIR/$working_dir_name_no_git" "$commit"
   fi
   depth=$(($depth-1))
 }
 
-path=$(basename $TUNASYNC_WORKING_DIR)
+path=$(basename -- "$TUNASYNC_WORKING_DIR")
 path_no_git=${path%%.git}
-mirror_url=$(echo $TUNASYNC_WORKING_DIR | sed "s#$WORKING_DIR_BASE#$MIRROR_BASE_URL#")
-script_append "mkdir -p $path_no_git"
-script_append "git clone $mirror_url $path_no_git"
-script_append "pushd $path_no_git"
-git_sync_recursive $TUNASYNC_UPSTREAM_URL $TUNASYNC_WORKING_DIR
+mirror_url=$(echo "$TUNASYNC_WORKING_DIR" | sed "s#$WORKING_DIR_BASE#$MIRROR_BASE_URL#")
+script_append "mkdir -p ${path_no_git@Q}"
+script_append "git clone ${mirror_url@Q} ${path_no_git@Q}"
+script_append "pushd ${path_no_git@Q}"
+git_sync_recursive "$TUNASYNC_UPSTREAM_URL" "$TUNASYNC_WORKING_DIR"
 script_append "git submodule absorbgitdirs"
 script_append "popd"
 
 if [[ ! -z "$RECURSIVE" ]]; then
-  mv $GENERATED_SCRIPT.tmp $GENERATED_SCRIPT
+  mv -- "$GENERATED_SCRIPT.tmp" "$GENERATED_SCRIPT"
 fi
 
-echo "Total size is" $(numfmt --to=iec $total_size)
+echo "Total size is" $(numfmt --to=iec "$total_size")