scripts/podcast/podcast-pipeline.sh

#!/usr/bin/env bash
set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
source "${SCRIPT_DIR}/scripts/lib.sh"

usage() {
  cat << EOF
Podcast Descomplicar Digital — Pipeline Automatizado

Usage: $0 <command> [args]

Commands:
  batch N         Process next N episodes (content generation)
  status          Show pipeline state
  retry NNN       Re-run failed stage for episode NNN
  csv [start] [N] Generate Canva CSV (default: next 7 episodes)
  publish NNN D   Generate publish commands for episode NNN on date D (YYYY-MM-DD)
  init NNN        Add episode NNN to pipeline state
  produce NNN F   Post-produce episode NNN from raw audio file F

Examples:
  $0 batch 5
  $0 status
  $0 csv 20 10
  $0 produce 20 /path/to/raw.wav
  $0 publish 20 2026-04-14
EOF
  exit 1
}

[[ $# -lt 1 ]] && usage

CMD="$1"; shift

ensure_state_file
mkdir -p "$LOG_DIR"

case "$CMD" in
  batch)
    COUNT="${1:-5}"
    START="$(jq -r '.next_episode' "$STATE_FILE")"
    END=$((START + COUNT - 1))
    log_info "=== BATCH: Processing episodes ${START}-${END} ==="

    CURRENT_DATE="$(jq -r '.next_publish_date' "$STATE_FILE")"
    [[ -z "$CURRENT_DATE" || "$CURRENT_DATE" == "null" ]] && CURRENT_DATE="$(date '+%Y-%m-%d')"

    for ((ep=START; ep<=END; ep++)); do
      log_info "--- Episode ${ep} ---"

      STATUS="$(get_episode_status "$ep")"
      if [[ "$STATUS" == "not_found" || "$STATUS" == "pending" ]]; then
        "${SCRIPT_DIR}/scripts/generate-content.sh" "$ep" || {
          log_error "EP$(pad_number "$ep"): Content generation failed, stopping batch"
          exit 1
        }
      else
        log_info "EP$(pad_number "$ep"): Already at status '${STATUS}', skipping content generation"
      fi

      STATUS="$(get_episode_status "$ep")"
      if [[ "$STATUS" == "script_done" ]]; then
        log_warn "EP$(pad_number "$ep"): Audio TTS needed (manual step via AI Studio)"
        log_warn "  -> Generate audio, save to Episodios/Audios/raw/ep_$(pad_number "$ep")_raw.wav"
        log_warn "  -> Then run: $0 produce ${ep} <raw_audio_path>"
      fi

      CURRENT_DATE="$(next_weekday "$CURRENT_DATE")"
    done

    jq --argjson n "$((END + 1))" --arg d "$CURRENT_DATE" \
      '.next_episode = $n | .next_publish_date = $d' \
      "$STATE_FILE" > "${STATE_FILE}.tmp" && mv "${STATE_FILE}.tmp" "$STATE_FILE"

    log_info "=== BATCH COMPLETE. Next episode: $((END + 1)), next date: ${CURRENT_DATE} ==="
    ;;

  status)
    echo "=== Pipeline Status ==="
    echo "Next episode: $(jq -r '.next_episode' "$STATE_FILE")"
    echo "Next publish date: $(jq -r '.next_publish_date' "$STATE_FILE")"
    echo ""
    EPISODE_COUNT="$(jq '.episodes | length' "$STATE_FILE")"
    if [[ "$EPISODE_COUNT" -gt 0 ]]; then
      echo "Episodes in pipeline:"
      jq -r '.episodes[] | "  EP\(.number | tostring | if length < 3 then "0" * (3 - length) + . else . end): \(.status) - \(.title)"' "$STATE_FILE"
      echo ""
      echo "Counts:"
      jq -r '.episodes | group_by(.status) | map({status: .[0].status, count: length}) | .[] | "  \(.status): \(.count)"' "$STATE_FILE"
    else
      echo "No episodes in pipeline yet. Run 'batch' to start."
    fi
    ;;

  retry)
    EP="${1:?Episode number required}"
    STATUS="$(get_episode_status "$EP")"
    log_info "Retrying EP$(pad_number "$EP") (current status: ${STATUS})"

    case "$STATUS" in
      pending|not_found)
        "${SCRIPT_DIR}/scripts/generate-content.sh" "$EP"
        ;;
      script_done)
        log_warn "Audio TTS needed — manual step via AI Studio"
        ;;
      audio_done)
        RAW=""
        for f in "${PROJECT_ROOT}/Episodios/Audios/raw/ep_$(pad_number "$EP")_"*; do
          [[ -f "$f" ]] && RAW="$f" && break
        done
        if [[ -n "$RAW" ]]; then
          "${SCRIPT_DIR}/scripts/post-produce.sh" "$EP" "$RAW"
        else
          log_error "Raw audio not found. Use: $0 produce ${EP} <path>"
        fi
        ;;
      produced)
        log_info "Ready to publish. Use: $0 publish ${EP} YYYY-MM-DD"
        ;;
      *)
        log_warn "Unknown status: ${STATUS}"
        ;;
    esac
    ;;

  csv)
    "${SCRIPT_DIR}/scripts/generate-csv.sh" "${@}"
    ;;

  produce)
    EP="${1:?Episode number required}"
    RAW="${2:?Raw audio path required}"
    "${SCRIPT_DIR}/scripts/post-produce.sh" "$EP" "$RAW"
    ;;

  publish)
    EP="${1:?Episode number required}"
    SCHED_DATE="${2:-$(jq -r '.next_publish_date' "$STATE_FILE")}"
    "${SCRIPT_DIR}/scripts/publish-episode.sh" "$EP" "$SCHED_DATE"
    ;;

  init)
    EP="${1:?Episode number required}"
    MAP_FILE="${PROJECT_ROOT}/config/episode-guide-map.json"
    EP_DATA="$(jq -r --arg n "$EP" '.[$n] // empty' "$MAP_FILE")"
    if [[ -n "$EP_DATA" ]]; then
      TITLE="$(echo "$EP_DATA" | jq -r '.podcast_title')"
      URL="$(echo "$EP_DATA" | jq -r '.guide_url')"
      add_episode "$EP" "$TITLE" "$URL"
      log_info "Added EP$(pad_number "$EP"): ${TITLE}"
    else
      log_error "Episode $EP not found in episode-guide-map.json"
    fi
    ;;

  *)
    usage
    ;;
esac