wsl_scripts/svn_interact/svn_git_sync

#!/usr/bin/env bash
# svn_git_sync — sync current SVN working copy into a Git mirror
# Usage:
#   svn_git_sync init [--history=N] [mirror_name]

set -euo pipefail

show_usage() {
  echo "Usage: $0 init [--history=N] [mirror_name]" >&2
  exit 1
}

subcommand="${1:-}"
[[ -n "$subcommand" ]] || show_usage
shift

history_count=0
mirror_arg=""

parse_opts() {
  while [[ $# -gt 0 ]]; do
    case "$1" in
      --history=*)
        history_count="${1#*=}"
        shift
        ;;
      --history)
        history_count="${2:-}"
        shift 2
        ;;
      -h|--help)
        show_usage
        ;;
      *)
        mirror_arg="$1"
        shift
        break
        ;;
    esac
  done
  if [[ -n "$mirror_arg" && $# -gt 0 ]]; then
    echo "Unexpected extra arguments: $*" >&2
    show_usage
  fi
}

case "$subcommand" in
  init)
    parse_opts "$@"
    ;;
  -h|--help|help)
    show_usage
    ;;
  *)
    echo "Unknown command: $subcommand" >&2
    show_usage
    ;;
esac

SRC_DIR="$(pwd)"
MIRROR_ROOT="${GIT_SVN_MIRROR_ROOT:-$HOME/git_svn_mirror}"

# Derive a collision-resistant mirror name from the full path (unless overridden).
raw_name="${mirror_arg:-$(pwd -P)}"
BASE_NAME="$(printf "%s" "$raw_name" | sed 's#^[\\/]*##; s#[\\/]#_#g')"
DEST="${MIRROR_ROOT}/${BASE_NAME}"

echo "Exporting SVN working copy to $DEST (versioned files only)..."
rm -rf -- "$DEST"
mkdir -p "$MIRROR_ROOT"

start_ts="$(date +%s.%N)"
elapsed() {
  local now
  now="${1:-$(date +%s.%N)}"
  awk -v s="$start_ts" -v n="$now" 'BEGIN{split(s,sa,".");split(n,na,".");printf "%.1f", (na[1]-sa[1]) + (na[2]-sa[2])/1e9}'
}

simple_export() {
  # Optionally compute total items upfront for a rough percentage
  local total_items=""
  if total=$(svn list -R . 2>/dev/null | wc -l || true); then
    total_items="$total"
  fi

  svn export --quiet . "$DEST" &
  export_pid=$!

  local last_reported=-1
  while kill -0 "$export_pid" 2>/dev/null; do
    if [[ -d "$DEST" ]]; then
      current=$(find "$DEST" -mindepth 1 -print | wc -l)
    else
      current=0
    fi

    if [[ "$current" -ne "$last_reported" ]]; then
      elapsed_val="$(elapsed)"
      pct=""
      if [[ -n "$total_items" && "$total_items" -gt 0 ]]; then
        pct=$(awk -v c="$current" -v t="$total_items" 'BEGIN{printf(" (%.1f%%)", (c*100)/t)}')
      fi
      printf "[%ss] Exported %d items%s\n" "$elapsed_val" "$current" "$pct"
      last_reported="$current"
    fi

    sleep 1
  done

  wait "$export_pid"

  final_count=$(find "$DEST" -mindepth 1 -print | wc -l 2>/dev/null || echo 0)
  printf "[%ss] Exported %d items (done)\n" "$(elapsed)" "$final_count"
}

init_git_repo() {
  echo "[$(elapsed) s] Initializing git repository..."
  GIT_QUIET=1 git init >/dev/null
  git config init.defaultBranch trunk || true
  git config core.fileMode false || true
  git config core.autocrlf true || true
  git config core.safecrlf false || true

  # Seed ignores for common noise (pyc/cache)
  if [[ ! -e .gitignore ]]; then
    echo "[$(elapsed) s] Adding default ignores (.pyc/__pycache__)..."
    printf "*.pyc\n__pycache__/\n" > .gitignore
  fi
}

link_git_to_source() {
  echo "[$(elapsed) s] Linking git metadata to source working tree..."
  DEST_ABS="$(pwd -P)"
  cd "$SRC_DIR"
  if [[ -e .git || -L .git ]]; then
    echo "[$(elapsed)s] Replacing existing .git in $SRC_DIR..."
    rm -rf -- .git
  fi
  printf "gitdir: %s/.git\n" "$DEST_ABS" > .git
  echo "[$(elapsed)s] Linked .git to $DEST_ABS/.git"

  echo "[$(elapsed) s] Syncing working tree to repo content..."
  GIT_QUIET=1 git checkout -q -- .
}

stage_and_commit() {
  local subject="${1:-}"
  local body="${2:-}"
  local allow_empty="${3:-0}"
  local display="${4:-$subject}"
  [[ -z "$subject" ]] && subject="SVN snapshot"
  git add -A

  if (( allow_empty )); then
    if [[ -n "$body" ]]; then
      GIT_QUIET=1 git commit -q --allow-empty -m "$subject" -m "$body"
    else
      GIT_QUIET=1 git commit -q --allow-empty -m "$subject"
    fi
  else
    if [[ -n "$body" ]]; then
      GIT_QUIET=1 git commit -q -m "$subject" -m "$body"
    else
      GIT_QUIET=1 git commit -q -m "$subject"
    fi
  fi
}

# Validate history_count is numeric (default 0)
if [[ -z "${history_count:-}" ]]; then
  history_count=0
elif ! [[ "$history_count" =~ ^[0-9]+$ ]]; then
  echo "--history must be a non-negative integer" >&2
  exit 1
fi

if (( history_count == 0 )); then
  simple_export
  cd "$DEST"
  init_git_repo
  stage_and_commit "Import SVN snapshot"
  link_git_to_source
  exit 0
fi

# History mode: pull last N revisions into Git.
SRC_URL="$(svn info --show-item url 2>/dev/null || svn info | awk -F': ' '/^URL:/ {print $2; exit}')"
if [[ -z "$SRC_URL" ]]; then
  echo "Unable to determine SVN URL" >&2
  exit 1
fi
REPO_ROOT="$(svn info --show-item repos-root-url 2>/dev/null || svn info | awk -F': ' '/^Repository Root:/ {print $2; exit}')"

mapfile -t REV_LIST < <(svn log -q -r HEAD:1 --limit "$history_count" "$SRC_URL" | awk '/^r[0-9]+/ {sub(/^r/,"",$1); print $1}' | tac)
if (( ${#REV_LIST[@]} == 0 )); then
  echo "No SVN revisions found for path $SRC_URL" >&2
  exit 1
fi

start_rev="${REV_LIST[0]}"
head_rev="${REV_LIST[-1]}"
echo "Including last ${#REV_LIST[@]} revisions affecting this path (r${start_rev}..r${head_rev})..."

WC_DIR="$(mktemp -d "${MIRROR_ROOT}/.svn_git_sync_wc.XXXXXX")"
mkdir -p "$DEST"

sync_wc_to_dest() {
  rsync -a --delete --exclude '.svn' --exclude '.git' "$WC_DIR"/ "$DEST"/
  if [[ ! -e "$DEST/.gitignore" ]]; then
    printf "*.pyc\n__pycache__/\n" > "$DEST/.gitignore"
  fi
}

echo "[$(elapsed) s] Checking out r${start_rev}..."
svn checkout -q -r "$start_rev" "$SRC_URL" "$WC_DIR"

sync_wc_to_dest

cd "$DEST"
init_git_repo
build_commit_msg() {
  local rev="$1"
  local svn_log author msg_body
  svn_log="$(svn log -r "$rev" -l 1 "$SRC_URL")"
  author="$(printf "%s\n" "$svn_log" | awk 'NR==2{split($0,a,"|"); gsub(/^ +| +$/,"",a[2]); print a[2]}')"
  msg_body="$(printf "%s\n" "$svn_log" | sed '1,2d;/^---/,$d')"
  if { [[ -z "$author" ]] || [[ -z "$msg_body" ]]; } && [[ -n "$REPO_ROOT" ]]; then
    svn_log="$(svn log -r "$rev" -l 1 "$REPO_ROOT" 2>/dev/null || true)"
    author="$(printf "%s\n" "$svn_log" | awk 'NR==2{split($0,a,"|"); gsub(/^ +| +$/,"",a[2]); print a[2]}')"
    msg_body="$(printf "%s\n" "$svn_log" | sed '1,2d;/^---/,$d')"
  fi
  msg_body="$(printf "%s" "$msg_body" | perl -0777 -pe 's/\A\s+//; s/\s+\z//; s/\n{2,}/\n\n/g')"
  COMMIT_BODY="$msg_body"
  COMMIT_BODY_FIRST="$(printf "%s" "$msg_body" | sed -n '1p')"
  COMMIT_SUBJECT="SVN r${rev}"
  [[ -n "$author" ]] && COMMIT_SUBJECT+=" by $author"
  COMMIT_PROGRESS="$COMMIT_SUBJECT"
  [[ -n "$COMMIT_BODY_FIRST" ]] && COMMIT_PROGRESS+=" — $COMMIT_BODY_FIRST"
}

build_commit_msg "$start_rev"
echo "[$(elapsed) s] Preparing commit: $COMMIT_PROGRESS"
stage_and_commit "$COMMIT_SUBJECT" "$COMMIT_BODY" 0 "$COMMIT_PROGRESS"

if (( ${#REV_LIST[@]} > 1 )); then
  for rev in "${REV_LIST[@]:1}"; do
    svn update -q -r "$rev" "$WC_DIR"
    sync_wc_to_dest
    build_commit_msg "$rev"
    echo "[$(elapsed) s] Preparing commit: $COMMIT_PROGRESS"
    stage_and_commit "$COMMIT_SUBJECT" "$COMMIT_BODY" 1 "$COMMIT_PROGRESS"
  done
fi

rm -rf -- "$WC_DIR"

link_git_to_source