generate_reports.sh

#!/bin/bash
set -euo pipefail

# ================================================
#  Generate HTML and Excel reports from quality dirs
#  and from top-level results (contamination/svevrste)
# ================================================

# Base results directory (relative to where this script runs)
BASE_DIR="$(pwd)/results"

# Path to your Python reporting script (html.py)
SCRIPT_PATH="$(pwd)/python/html.py"

# Log file for tracking actions
LOG_FILE="${BASE_DIR}/report_generation.log"
echo "=== Report generation started at $(date) ===" > "$LOG_FILE"

# --------------------------------------------
# Function: Generate and rename reports
# --------------------------------------------
generate_and_rename_report() {
    local quality_dir="$1"
    local report_prefix="$2"

    echo "[$(date)] Generating report for $quality_dir" >> "$LOG_FILE"

    # Copy the Python script into the directory
    cp "$SCRIPT_PATH" "$quality_dir"
    cd "$quality_dir"

    # Run the report generator
    if python3 html.py; then
        # Rename HTML report
        if [ -f "final_report.html" ]; then
            mv "final_report.html" "${report_prefix}_final_report.html"
            echo "Renamed final_report.html to ${report_prefix}_final_report.html" >> "$LOG_FILE"
        else
            echo "final_report.html not found in $quality_dir" >> "$LOG_FILE"
        fi

        # Rename Excel report
        if [ -f "final_report.xlsx" ]; then
            mv "final_report.xlsx" "${report_prefix}_final_report.xlsx"
            echo "Renamed final_report.xlsx to ${report_prefix}_final_report.xlsx" >> "$LOG_FILE"
        else
            echo "final_report.xlsx not found in $quality_dir" >> "$LOG_FILE"
        fi
    else
        echo "Error running html.py in $quality_dir" >> "$LOG_FILE"
    fi

    # Remove the copied script to keep directories clean
    rm -f "html.py"
    echo "[$(date)] Report generation completed for $quality_dir" >> "$LOG_FILE"
}

# --------------------------------------------
# Find all quality directories automatically
# (exactly three levels deep, avoids nested dups)
# --------------------------------------------
mapfile -t quality_dirs < <(find "$BASE_DIR" -mindepth 3 -maxdepth 3 -type d -name quality)

# --------------------------------------------
# Process each quality directory
# --------------------------------------------
for quality_dir in "${quality_dirs[@]}"; do
    if [ "$(ls -A "$quality_dir")" ]; then
        report_prefix=$(basename "$(dirname "$quality_dir")")
        generate_and_rename_report "$quality_dir" "$report_prefix"
    else
        echo "[$(date)] $quality_dir is empty or does not exist, skipping" >> "$LOG_FILE"
    fi
done

# --------------------------------------------
# Generate top-level results summary (for contamination/svevrste)
# --------------------------------------------
if compgen -G "${BASE_DIR}/contamination*.txt" > /dev/null || \
   compgen -G "${BASE_DIR}/svevrste*.txt" > /dev/null; then
    echo "[$(date)] Generating top-level results report for $BASE_DIR" >> "$LOG_FILE"
    cp "$SCRIPT_PATH" "$BASE_DIR"
    cd "$BASE_DIR"
    if python3 html.py; then
        mv "final_report.html" "results_summary_final_report.html"
        mv "final_report.xlsx" "results_summary_final_report.xlsx"
        echo "Top-level results summary created in $BASE_DIR" >> "$LOG_FILE"
    else
        echo "Error running html.py in $BASE_DIR" >> "$LOG_FILE"
    fi
    rm -f "html.py"
fi

echo "=== All reports generated successfully at $(date) ===" >> "$LOG_FILE"