#!/usr/bin/env bash
# run-task.sh <task-name> <prompt-file> [cwd]
# Quota-aware worker runner: survives 5-hour usage-limit caps by sleeping until
# quota refreshes and resuming the SAME session instead of restarting from zero.
set -u
NAME="$1"; PROMPT_FILE="$2"; CWD="${3:-$HOME/bookshelf}"
L="$HOME/bookshelf/logs"; mkdir -p "$L"
LOG="$L/${NAME}.json"; ERR="$L/${NAME}.err"; SIDF="$L/${NAME}.sid"; ST="$L/${NAME}.state"

MAX_WALL="${MAX_WALL:-86400}"   # 24h total patience
POLL="${POLL:-600}"             # 10 min between quota probes
MAX_HARD_FAILS="${MAX_HARD_FAILS:-3}"

export JAVA_HOME="$HOME/toolchain/jdk21"
export ANDROID_HOME="$HOME/toolchain/android-sdk"
export ANDROID_SDK_ROOT="$ANDROID_HOME"
export PATH="$JAVA_HOME/bin:$ANDROID_HOME/platform-tools:$PATH"
export GRADLE_USER_HOME="$HOME/.gradle"
cd "$CWD" || exit 1

# Stable session id so a killed run can be resumed rather than restarted.
if [ ! -s "$SIDF" ]; then
  python3 -c "import uuid;print(uuid.uuid4())" > "$SIDF"
  FRESH=1
else
  FRESH=0   # sid pre-seeded (recovered) or left by an earlier attempt
fi
SID="$(cat "$SIDF")"

say(){ echo "[$(date -Is)] $NAME: $*" >> "$ST"; }
say "start sid=$SID fresh=$FRESH wall=${MAX_WALL}s poll=${POLL}s"

CONT_PROMPT="Continue the task you were working on, from wherever you left off. \
Your original instructions are earlier in this conversation; re-read them and any \
files you already wrote before doing more work. Do not restart from scratch and do \
not redo completed work. Finish the task and give the final report."

deadline=$(( $(date +%s) + MAX_WALL ))
attempt=0; hard=0; quota_waits=0

while [ "$(date +%s)" -lt "$deadline" ]; do
  attempt=$((attempt+1))
  if [ "$attempt" -eq 1 ] && [ "$FRESH" -eq 1 ]; then
    say "attempt $attempt: fresh (--session-id)"
    claude -p --model sonnet --permission-mode bypassPermissions \
      --output-format json --add-dir "$HOME/bookshelf" \
      --session-id "$SID" < "$PROMPT_FILE" > "$LOG" 2>"$ERR"
  else
    say "attempt $attempt: resume $SID"
    printf '%s' "$CONT_PROMPT" | claude -p --model sonnet \
      --permission-mode bypassPermissions --output-format json \
      --add-dir "$HOME/bookshelf" --resume "$SID" > "$LOG" 2>"$ERR"
  fi
  rc=$?

  blob="$(cat "$LOG" "$ERR" 2>/dev/null | head -c 20000)"

  # 1) quota / rate limit -> wait it out, do NOT burn a hard-fail
  if printf '%s' "$blob" | grep -qiE 'usage limit|limit will reset|limit resets|rate_limit_error|rate limit exceeded|429|too many requests|overloaded_error'; then
    quota_waits=$((quota_waits+1))
    say "QUOTA hit (wait #$quota_waits). sleeping ${POLL}s then probing again."
    sleep "$POLL"
    continue
  fi

  # 2) session vanished -> start clean once
  if printf '%s' "$blob" | grep -qiE 'no conversation found|session not found|could not resume'; then
    say "session $SID unresumable; starting fresh"
    python3 -c "import uuid;print(uuid.uuid4())" > "$SIDF"; SID="$(cat "$SIDF")"; FRESH=1; attempt=0
    continue
  fi

  # 3) success
  isErr="$(jq -r '.is_error // false' "$LOG" 2>/dev/null)"
  if [ "$rc" -eq 0 ] && [ "$isErr" != "true" ]; then
    say "SUCCESS after $attempt attempt(s), $quota_waits quota wait(s)"
    break
  fi

  # 4) genuine failure
  hard=$((hard+1))
  say "hard failure #$hard (rc=$rc is_error=$isErr)"
  if [ "$hard" -ge "$MAX_HARD_FAILS" ]; then say "GIVING UP after $hard hard failures"; break; fi
  sleep 60
done

[ "$(date +%s)" -ge "$deadline" ] && say "WALL CLOCK EXCEEDED"

{
  echo "=== $NAME  attempts=$attempt quota_waits=$quota_waits hard_fails=$hard ==="
  jq -r '"cost=$" + ((.total_cost_usd//0)|tostring) + " turns=" + ((.num_turns//0)|tostring) + " err=" + ((.is_error//"?")|tostring)' "$LOG" 2>/dev/null
  echo "--- result (tail) ---"
  jq -r '.result // "no result"' "$LOG" 2>/dev/null | tail -c 1800
} > "$L/${NAME}.summary"