#!/usr/bin/env bash
# Read-only. Establishes portable paths that work on BOTH hosts.
#
# Your storage is one NFS export mounted at different points:
#   jupyter-pod    ->  /workspace            (has the MIG GPU)
#   kls-headnode   ->  /home/dgx-i-carbine   (this shell)
#
# So we never hardcode absolutes — everything is derived from where this repo
# actually sits.  Run on either host:  bash scripts/whereami.sh
set -uo pipefail

REPO="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
ANURAG="$(dirname "$REPO")"            # .../carbine/anurag
CARBINE="$(dirname "$ANURAG")"         # .../carbine
BASE="$(dirname "$CARBINE")"           # /workspace  OR  /home/dgx-i-carbine
ROOT="$ANURAG/voicerag_data"           # data root, INSIDE anurag

echo "════════ HOST ════════"
echo "  hostname : $(hostname -s 2>/dev/null || hostname)"
echo "  user     : $(id -un 2>/dev/null || echo "uid $(id -u)")"

echo
echo "════════ DERIVED PATHS (portable) ════════"
printf '  %-14s %s\n' repo    "$REPO"
printf '  %-14s %s\n' anurag  "$ANURAG"
printf '  %-14s %s\n' base    "$BASE"
printf '  %-14s %s\n' "DATA ROOT" "$ROOT"

echo
echo "════════ STORAGE ════════"
printf '  %-14s ' "anurag";  df -h "$ANURAG" 2>/dev/null | awk 'NR==2{printf "%s free of %s  (%s)\n",$4,$2,$1}' || echo "?"
for extra in /data "$BASE/../data"; do
  [ -d "$extra" ] && { printf '  %-14s ' "$extra"; df -h "$extra" 2>/dev/null | awk 'NR==2{printf "%s free of %s\n",$4,$2}'; }
done

echo
echo "  -- stray data dirs from the earlier default (should be empty/absent) --"
_seen=""
for cand in "$BASE/voicerag_data" "$HOME/voicerag_data" "/workspace/voicerag_data"; do
  _real="$(readlink -f "$cand" 2>/dev/null || echo "$cand")"
  case " $_seen " in *" $_real "*) continue ;; esac
  _seen="$_seen $_real"
  cand="$_real"
  if [ -d "$cand" ]; then
    echo "    FOUND  $cand   ($(du -sh "$cand" 2>/dev/null | cut -f1))"
  else
    echo "    absent $cand"
  fi
done

echo
echo "  -- current data root --"
if [ -d "$ROOT" ]; then
  echo "    exists: $ROOT  ($(du -sh "$ROOT" 2>/dev/null | cut -f1))"
  [ -d "$ROOT/hf_cache" ] && echo "    hf_cache: $(du -sh "$ROOT/hf_cache" 2>/dev/null | cut -f1)"
else
  echo "    not created yet"
fi

echo
echo "════════ GPU ON THIS HOST ════════"
if command -v nvidia-smi >/dev/null 2>&1; then
  nvidia-smi -L 2>&1 | sed 's/^/  /'
else
  echo "  no nvidia-smi -> no GPU here. Run GPU work on jupyter-pod."
fi

echo
echo "════════ RUNNING JOBS (this host only) ════════"
pgrep -af '01_download.py|pseudo_docs.py|fertility.py' 2>/dev/null | sed 's/^/  /' || echo "  none"
echo "  NOTE: processes on jupyter-pod are invisible from here and vice versa."

echo
echo "════════ USE THIS ════════"
cat <<EOF
  export VOICERAG_ROOT="$ROOT"

  Or portably, from anywhere in the repo:
    export VOICERAG_ROOT="\$(cd "\$(git rev-parse --show-toplevel 2>/dev/null || echo "$REPO")/.." && pwd)/voicerag_data"
EOF