#!/bin/bash
# PHP Domain Analyzer - Collect real usage data for true optimization
# Run this before optimization to build accurate domain profiles
# Uses actual logs and process data instead of thresholds

# Source required libraries
PHP_TOOLKIT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && cd ../.. && pwd)"
source "$PHP_TOOLKIT_DIR/lib/common-functions.sh" 2>/dev/null || { echo "ERROR: common-functions.sh not found"; exit 1; }
source "$PHP_TOOLKIT_DIR/lib/system-detect.sh" 2>/dev/null || { echo "ERROR: system-detect.sh not found"; exit 1; }
source "$PHP_TOOLKIT_DIR/lib/user-manager.sh" 2>/dev/null || { echo "ERROR: user-manager.sh not found"; exit 1; }
source "$PHP_TOOLKIT_DIR/lib/php-detector.sh" 2>/dev/null || { echo "ERROR: php-detector.sh not found"; exit 1; }
source "$PHP_TOOLKIT_DIR/lib/php-analytics.sh" 2>/dev/null || { echo "ERROR: php-analytics.sh not found"; exit 1; }
source "$PHP_TOOLKIT_DIR/lib/php-scanner.sh" 2>/dev/null || { echo "ERROR: php-scanner.sh not found"; exit 1; }

# Color codes
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
CYAN='\033[0;36m'
WHITE='\033[1;37m'
BOLD='\033[1m'
NC='\033[0m'

cecho() {
    echo -e "$@"
}

# ============================================================================
# MAIN ANALYSIS
# ============================================================================

initialize_system_detection

if [ "$EUID" -ne 0 ]; then
    cecho "${RED}ERROR: This script must be run as root${NC}"
    exit 1
fi

cecho "${CYAN}╔════════════════════════════════════════════════════════════════════════╗${NC}"
cecho "${CYAN}║${WHITE}         PHP DOMAIN ANALYZER - REAL USAGE DATA COLLECTION          ${CYAN}║${NC}"
cecho "${CYAN}╚════════════════════════════════════════════════════════════════════════╝${NC}"
echo ""

cecho "${WHITE}${BOLD}Starting Analysis...${NC}"
cecho "${CYAN}This may take a few minutes. Analyzing logs, processes, and traffic patterns.${NC}"
echo ""

# Get all users and domains
users=$(list_all_users)

declare -a all_profiles
total_domains=0
analyzed=0

cecho "${YELLOW}Analyzing domains...${NC}"
echo ""

while IFS= read -r username; do
    [ -z "$username" ] && continue

    user_domains=$(get_user_domains "$username")

    while IFS= read -r domain; do
        [ -z "$domain" ] && continue

        total_domains=$((total_domains + 1))

        cecho -n "  [$total_domains] Analyzing $domain..."

        # Build profile with actual data
        profile=$(build_domain_profile "$username" "$domain")

        if [ -n "$profile" ]; then
            all_profiles+=("$profile")
            store_domain_profile "$profile"
            analyzed=$((analyzed + 1))

            # Extract key metrics
            peak_concurrent=$(echo "$profile" | cut -d'|' -f3)
            peak_mem=$(echo "$profile" | cut -d'|' -f11)
            leak_status=$(echo "$profile" | cut -d'|' -f12)

            cecho "${GREEN} ✓${NC}"
            cecho "      Peak: $peak_concurrent concurrent | Memory: ${peak_mem}MB | Leak: $leak_status"
        else
            cecho "${YELLOW} ✗${NC} (could not collect data)"
        fi
    done <<< "$user_domains"
done <<< "$users"

echo ""
cecho "${CYAN}═══════════════════════════════════════════════════════════════════════════${NC}"
cecho "${WHITE}${BOLD}ANALYSIS COMPLETE${NC}"
cecho "${CYAN}═══════════════════════════════════════════════════════════════════════════${NC}"
echo ""

cecho "  Total domains analyzed: ${WHITE}${analyzed}${NC} / ${total_domains}"
cecho "  Profiles stored in: ${WHITE}/tmp/php-domain-profiles/${NC}"
echo ""

# Display summary
if [ "${#all_profiles[@]}" -gt 0 ]; then
    cecho "${CYAN}ANALYSIS SUMMARY:${NC}"
    echo ""

    # Find domains with highest memory usage
    cecho "${WHITE}${BOLD}Top Memory Usage:${NC}"
    printf '%s\n' "${all_profiles[@]}" | sort -t'|' -k11 -rn | head -5 | while IFS='|' read -r domain username peak_concurrent avg_concurrent total_hits min_mem max_mem avg_mem proc_count mem_exhausted peak_mem leak rest; do
        cecho "  • $domain: ${peak_mem}MB peak memory"
    done

    echo ""

    # Find domains with highest traffic
    cecho "${WHITE}${BOLD}Top Traffic:${NC}"
    printf '%s\n' "${all_profiles[@]}" | sort -t'|' -k3 -rn | head -5 | while IFS='|' read -r domain username peak_concurrent avg_concurrent total_hits rest; do
        cecho "  • $domain: $peak_concurrent concurrent requests"
    done

    echo ""

    # Find domains with potential leaks
    leak_count=$(printf '%s\n' "${all_profiles[@]}" | \grep -c "LIKELY_LEAK")
    if [ "$leak_count" -gt 0 ]; then
        cecho "${RED}${BOLD}⚠ Domains with potential memory leaks:${NC} $leak_count"
        printf '%s\n' "${all_profiles[@]}" | \grep "LIKELY_LEAK" | while IFS='|' read -r domain username rest; do
            cecho "  • $domain"
        done
        echo ""
    fi

    # Show high memory usage domains
    high_mem=$(printf '%s\n' "${all_profiles[@]}" | awk -F'|' '$11 > 200 {print}' | wc -l)
    if [ "$high_mem" -gt 0 ]; then
        cecho "${YELLOW}${BOLD}Domains using >200MB:${NC} $high_mem"
    fi
fi

echo ""
cecho "${CYAN}═══════════════════════════════════════════════════════════════════════════${NC}"
echo ""
cecho "${GREEN}${BOLD}✓ Domain profiles ready for optimization!${NC}"
echo ""
cecho "Next step: Run php-optimizer.sh → Option 5 → Select optimization level"
cecho "The optimizer will now use REAL usage data for accurate recommendations."
echo ""