#!/bin/bash

# Script to split Tam Bao Final.txt into 5-page chunks
# Each file will have ~500 lines (assuming 100 lines per page)

INPUT_FILE="/opt/openclaw/.openclaw/workspace/An Duc Tam Bao/translate/Tam Bao Final.txt"
OUTPUT_DIR="/opt/openclaw/.openclaw/workspace/An Duc Tam Bao/translate"
LINES_PER_FILE=500  # ~5 pages (100 lines per page)

# Remove BOM if present and convert to clean UTF-8
tail -c +4 "$INPUT_FILE" > /tmp/tam_bao_clean.txt

# Get total lines
TOTAL_LINES=$(wc -l < /tmp/tam_bao_clean.txt)
echo "Total lines: $TOTAL_LINES"

# Calculate number of files needed
NUM_FILES=$(( (TOTAL_LINES + LINES_PER_FILE - 1) / LINES_PER_FILE ))
echo "Number of files: $NUM_FILES"

# Split the file
FILE_NUM=1
START_LINE=1

while [ $START_LINE -le $TOTAL_LINES ]; do
    END_LINE=$((START_LINE + LINES_PER_FILE - 1))
    if [ $END_LINE -gt $TOTAL_LINES ]; then
        END_LINE=$TOTAL_LINES
    fi
    
    # Format file number with leading zeros (3 digits)
    FILE_NUM_PADDED=$(printf "%03d" $FILE_NUM)
    NEXT_START=$((END_LINE + 1))
    NEXT_START_PADDED=$(printf "%03d" $((FILE_NUM * 5 + 1)))
    
    OUTPUT_FILE="$OUTPUT_DIR/Tam Bao Final ${FILE_NUM_PADDED}.md"
    
    # Extract lines and save as markdown
    sed -n "${START_LINE},${END_LINE}p" /tmp/tam_bao_clean.txt > "$OUTPUT_FILE"
    
    echo "Created: $OUTPUT_FILE (lines $START_LINE-$END_LINE)"
    
    START_LINE=$((END_LINE + 1))
    FILE_NUM=$((FILE_NUM + 1))
done

# Clean up
rm /tmp/tam_bao_clean.txt

echo "Done! Created $((FILE_NUM - 1)) files."
