y-cluster-blobs

#!/usr/bin/env bash
[ -z "$DEBUG" ] || set -x
set -eo pipefail

YHELP='y-cluster-blobs - List and read objects in cluster blob storage

Usage: y-cluster-blobs SUBCOMMAND [options] [CONTEXT://BUCKET/PATH]

Subcommands:
  ls [CONTEXT://BUCKET[/PREFIX]]    List buckets or objects
  cat CONTEXT://BUCKET/OBJECT       Print object contents to stdout
  help                              Show this help

Options:
  -l          Long listing with size and timestamp (ls only)
  -r          Recursive listing (ls only)

URL format:
  local://                  List all buckets
  local://BUCKET            List objects in bucket
  local://BUCKET/PREFIX     List objects matching prefix
  local://BUCKET/KEY        Read object (cat)

The CONTEXT part maps to --context for kubectl (currently only local).
Modeled after gcloud storage ls / gcloud storage cat.
Reads VersityGW filesystem directly via kubectl exec for speed.

Environment:
  NAMESPACE    Override blob store namespace (default: blobs)

Exit codes:
  0          Success
  1          Usage error
  2          Bucket not found
'

case "${1:-}" in
  help|--help) echo "$YHELP"; exit 0 ;;
esac

SUBCMD="${1:-}"
shift || true

case "$SUBCMD" in
  ls|cat) ;;
  "") echo "ERROR: subcommand required (ls, cat, help)" >&2; exit 1 ;;
  *) echo "ERROR: unknown subcommand '$SUBCMD'" >&2; exit 1 ;;
esac

# Parse flags
LONG=false
RECURSIVE=false
while [ $# -gt 0 ]; do
  case "$1" in
    -l) LONG=true; shift ;;
    -r) RECURSIVE=true; shift ;;
    -lr|-rl) LONG=true; RECURSIVE=true; shift ;;
    -*) echo "ERROR: unknown flag '$1'" >&2; exit 1 ;;
    *) break ;;
  esac
done

URL="${1:-}"
[ -z "$NAMESPACE" ] && NAMESPACE=blobs

# Parse URL: context://bucket/path
CTX=""
BUCKET=""
OBJPATH=""

if [ -n "$URL" ]; then
  case "$URL" in
    *://*)
      CTX_NAME="${URL%%://*}"
      CTX="--context=$CTX_NAME"
      rest="${URL#*://}"
      BUCKET="${rest%%/*}"
      if [ "$rest" = "$BUCKET" ]; then
        OBJPATH=""
      else
        OBJPATH="${rest#*/}"
      fi
      ;;
    *)
      echo "ERROR: URL must use context:// format, got '$URL'" >&2
      exit 1
      ;;
  esac
fi

if [ -z "$CTX" ]; then
  echo "ERROR: URL required, e.g. local://" >&2
  exit 1
fi

vgw() {
  kubectl $CTX -n "$NAMESPACE" exec deploy/versitygw -- "$@"
}

DATA_ROOT="/data"

case "$SUBCMD" in
  ls)
    PREFIX="$CTX_NAME://"
    if [ -z "$BUCKET" ]; then
      # List buckets
      if [ "$LONG" = "true" ]; then
        vgw find "$DATA_ROOT" -mindepth 1 -maxdepth 1 -type d -exec stat -c '%Y %n' {} \; 2>/dev/null \
          | while read -r epoch name; do
              ts=$(date -d "@$epoch" '+%Y-%m-%d %H:%M:%S' 2>/dev/null || echo "$epoch")
              echo "[$ts]     0B ${PREFIX}$(basename "$name")/"
            done
      else
        vgw ls "$DATA_ROOT" 2>/dev/null | while read -r name; do
          echo "${PREFIX}$name/"
        done
      fi
    else
      PREFIX="$CTX_NAME://$BUCKET/"
      BUCKET_PATH="$DATA_ROOT/$BUCKET"
      if ! vgw test -d "$BUCKET_PATH" 2>/dev/null; then
        echo "ERROR: bucket '$BUCKET' not found" >&2
        exit 2
      fi
      TARGET="$BUCKET_PATH${OBJPATH:+/$OBJPATH}"
      DEPTH_ARGS=""
      if [ "$RECURSIVE" = "false" ]; then
        DEPTH_ARGS="-mindepth 1 -maxdepth 1"
      fi
      if [ "$LONG" = "true" ]; then
        vgw sh -c "find '$TARGET' $DEPTH_ARGS -type d ! -path '$TARGET' -exec stat -c 'd %s %Y %n' {} \\; -o -type f -exec stat -c 'f %s %Y %n' {} \\;" 2>/dev/null \
          | while read -r type size epoch name; do
              ts=$(date -d "@$epoch" '+%Y-%m-%d %H:%M:%S' 2>/dev/null || echo "$epoch")
              rel="${name#$BUCKET_PATH/}"
              if [ "$type" = "d" ]; then
                echo "[$ts]     0B ${PREFIX}$rel/"
              else
                echo "[$ts] ${size}B ${PREFIX}$rel"
              fi
            done
      else
        vgw sh -c "find '$TARGET' $DEPTH_ARGS -type d ! -path '$TARGET' -exec echo d {} \\; -o -type f -exec echo f {} \\;" 2>/dev/null \
          | while read -r type name; do
              rel="${name#$BUCKET_PATH/}"
              if [ "$type" = "d" ]; then
                echo "${PREFIX}$rel/"
              else
                echo "${PREFIX}$rel"
              fi
            done
      fi
    fi
    ;;
  cat)
    if [ -z "$BUCKET" ] || [ -z "$OBJPATH" ]; then
      echo "ERROR: cat requires a full object path, e.g. local://bucket/key" >&2
      exit 1
    fi
    OBJ_FILE="$DATA_ROOT/$BUCKET/$OBJPATH"
    vgw cat "$OBJ_FILE"
    ;;
esac