#!/bin/bash
# Script xu ly batch Sadi-061-065
BATCH="061-065"
BASE_DIR="/opt/openclaw/.openclaw/workspace/Chuan Muc Sadi"
EXTRACTED_DIR="$BASE_DIR/extracted"
SCRIPTS_DIR="/opt/openclaw/.openclaw/workspace/scripts"

log() {
    echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1"
}

log "Bat dau xu ly batch Sadi-$BATCH"

if [ ! -f "$BASE_DIR/Sadi-$BATCH.pdf" ]; then
    log "Loi: Khong tim thay file Sadi-$BATCH.pdf"
    exit 1
fi

log "Buoc 1: OCR voi Qwen (tao file trong)"
touch "$EXTRACTED_DIR/Sadi-$BATCH-qwen.md"
log "File qwen.md da tao - can OCR thu cong"

log "Buoc 2: Chay script hau xu ly"
if [ -f "$EXTRACTED_DIR/Sadi-$BATCH-qwen.md" ] && [ -s "$EXTRACTED_DIR/Sadi-$BATCH-qwen.md" ]; then
    python3 "$SCRIPTS_DIR/fix_myanmar_ocr.py" "$EXTRACTED_DIR/Sadi-$BATCH-qwen.md"
    log "Hoan thanh hau xu ly"
else
    log "File qwen.md rong - chua co OCR"
fi

log "Hoan thanh batch Sadi-$BATCH"
