#!/bin/bash
#############################################################
#### Nginx Metrics Exporter for Prometheus               ####
#### Comprehensive nginx monitoring via stub_status,     ####
#### logs, SSL, process, and config metrics              ####
####                                                     ####
#### Author: Phil Connor                                 ####
#### Contact: contact@mylinux.work                       ####
#### License: MIT                                        ####
#### Version: 1.1                                        ####
####                                                     ####
#### Usage: ./nginx-metrics-exporter.sh [OPTIONS]        ####
#############################################################
#
# Metrics collected:
#   - stub_status: connections, accepts, handled, requests, reading, writing, waiting
#   - Process: worker count, memory usage, CPU usage, open files
#   - Access logs: requests by status code, response times, bytes transferred
#   - SSL: certificate expiry days for configured domains
#   - Config: worker_processes, worker_connections, keepalive_timeout
#   - Upstream: health status (if configured)
#
# Requirements:
#   - nginx with stub_status module enabled
#   - socat (for HTTP server)
#   - curl (for stub_status fetching)
#
set -euo pipefail

#########################
### Configuration ###
#########################

LISTEN_PORT="${NGINX_EXPORTER_PORT:-9113}"
STUB_STATUS_URL="${NGINX_STUB_URL:-http://127.0.0.1/nginx_status}"
ACCESS_LOG="${NGINX_ACCESS_LOG:-/var/log/nginx/access.log}"
ERROR_LOG="${NGINX_ERROR_LOG:-/var/log/nginx/error.log}"
NGINX_CONF="${NGINX_CONF:-/etc/nginx/nginx.conf}"
SITES_DIR="${NGINX_SITES_DIR:-/etc/nginx/sites-enabled}"
CONF_D_DIR="${NGINX_CONF_D:-/etc/nginx/conf.d}"
SCRAPE_INTERVAL="${SCRAPE_INTERVAL:-15}"
SSL_CHECK_DOMAINS="${SSL_CHECK_DOMAINS:-}"  # Comma-separated list of domains to check SSL

# Log parsing settings
LOG_TAIL_LINES="${LOG_TAIL_LINES:-10000}"  # Number of lines to parse from access log
LOG_PARSE_INTERVAL="${LOG_PARSE_INTERVAL:-60}"  # How often to parse logs (seconds)

# State files for log metrics
STATE_DIR="/tmp/nginx-metrics"
LAST_LOG_PARSE=0

# Output mode
TEXTFILE_DIR="/var/lib/node_exporter"
OUTPUT_FILE=""
HTTP_MODE=false

#########################
### Logging ###
#########################

log() {
    echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*" >&2
}

#########################
### Parse Arguments ###
#########################

parse_args() {
    while [[ $# -gt 0 ]]; do
        case "$1" in
            --textfile)
                OUTPUT_FILE="$TEXTFILE_DIR/nginx.prom"
                shift
                ;;
            --http)
                HTTP_MODE=true
                shift
                ;;
            --output|-o)
                OUTPUT_FILE="$2"
                shift 2
                ;;
            --port)
                LISTEN_PORT="$2"
                shift 2
                ;;
            --stub-url)
                STUB_STATUS_URL="$2"
                shift 2
                ;;
            --access-log)
                ACCESS_LOG="$2"
                shift 2
                ;;
            --error-log)
                ERROR_LOG="$2"
                shift 2
                ;;
            --nginx-conf)
                NGINX_CONF="$2"
                shift 2
                ;;
            --ssl-domains)
                SSL_CHECK_DOMAINS="$2"
                shift 2
                ;;
            --help)
                cat <<EOF
Nginx Metrics Exporter for Prometheus

Usage: $0 [OPTIONS]

MODES:
  --textfile             Write to node_exporter textfile collector
  --http                 Run HTTP server on port $LISTEN_PORT (default mode with no other output)
  (no flag)              Output to stdout

OPTIONS:
  --port PORT            HTTP port (default: 9113)
  -o, --output PATH      Output file path
  --stub-url URL         Nginx stub_status URL (default: http://127.0.0.1/nginx_status)
  --access-log PATH      Path to access log (default: /var/log/nginx/access.log)
  --error-log PATH       Path to error log (default: /var/log/nginx/error.log)
  --nginx-conf PATH      Path to nginx.conf (default: /etc/nginx/nginx.conf)
  --ssl-domains LIST     Comma-separated domains to check SSL expiry
  --help                 Show this help

EXAMPLES:
  $0 --textfile                    # Write to textfile collector
  $0 --http --port 9113            # Run HTTP server
  $0 -o /tmp/nginx.prom            # Write to custom file
  $0                               # Output to stdout

EOF
                exit 0
                ;;
            *)
                log "Unknown option: $1"
                exit 1
                ;;
        esac
    done
}

#########################
### Setup ###
#########################

detect_package_manager() {
    if command -v apt-get &>/dev/null; then
        echo "apt"
    elif command -v dnf &>/dev/null; then
        echo "dnf"
    elif command -v yum &>/dev/null; then
        echo "yum"
    elif command -v zypper &>/dev/null; then
        echo "zypper"
    elif command -v pacman &>/dev/null; then
        echo "pacman"
    elif command -v apk &>/dev/null; then
        echo "apk"
    else
        echo ""
    fi
}

install_package() {
    local pkg="$1"
    local pkgmgr
    pkgmgr=$(detect_package_manager)
    
    log "Installing $pkg..."
    
    case "$pkgmgr" in
        apt)
            apt-get update -qq && apt-get install -y -qq "$pkg"
            ;;
        dnf)
            dnf install -y -q "$pkg"
            ;;
        yum)
            yum install -y -q "$pkg"
            ;;
        zypper)
            zypper install -y -q "$pkg"
            ;;
        pacman)
            pacman -S --noconfirm "$pkg"
            ;;
        apk)
            apk add --quiet "$pkg"
            ;;
        *)
            log "ERROR: Unknown package manager. Please install $pkg manually."
            return 1
            ;;
    esac
}

setup() {
    mkdir -p "$STATE_DIR"
    
    # Check for required tools and install if missing
    if ! command -v socat &>/dev/null; then
        log "socat not found, attempting to install..."
        if [[ $EUID -eq 0 ]]; then
            if ! install_package socat; then
                log "ERROR: Failed to install socat"
                exit 1
            fi
            log "socat installed successfully"
        else
            log "ERROR: socat is required. Run as root to auto-install, or install manually:"
            log "  Debian/Ubuntu: apt install socat"
            log "  RHEL/CentOS:   yum install socat"
            log "  Fedora:        dnf install socat"
            log "  Alpine:        apk add socat"
            exit 1
        fi
    fi
    
    if ! command -v curl &>/dev/null; then
        log "curl not found, attempting to install..."
        if [[ $EUID -eq 0 ]]; then
            if ! install_package curl; then
                log "ERROR: Failed to install curl"
                exit 1
            fi
            log "curl installed successfully"
        else
            log "ERROR: curl is required. Run as root to auto-install, or install manually."
            exit 1
        fi
    fi
    
    # Check if nginx is running
    if ! pgrep -x nginx &>/dev/null && ! pidof nginx &>/dev/null; then
        log "WARNING: nginx process not found - process metrics will show nginx_process_running=0"
    fi
    
    # Check if stub_status is accessible
    check_stub_status
}

check_stub_status() {
    log "Checking stub_status at $STUB_STATUS_URL..."
    
    local response http_code
    response=$(curl -s -o /dev/null -w "%{http_code}" --max-time 5 "$STUB_STATUS_URL" 2>/dev/null)
    
    if [[ "$response" == "200" ]]; then
        # Verify it's actually stub_status output
        local content
        content=$(curl -s --max-time 5 "$STUB_STATUS_URL" 2>/dev/null)
        if echo "$content" | grep -q "Active connections:"; then
            log "✓ stub_status is working correctly"
            return 0
        else
            log "WARNING: $STUB_STATUS_URL returned 200 but doesn't look like stub_status output"
            log "  Expected 'Active connections:' in response"
            show_stub_status_help
            return 1
        fi
    elif [[ "$response" == "000" ]]; then
        log "WARNING: Cannot connect to $STUB_STATUS_URL (connection refused/timeout)"
        log "  stub_status metrics will show nginx_up=0"
        show_stub_status_help
        return 1
    elif [[ "$response" == "403" ]]; then
        log "WARNING: Access denied to $STUB_STATUS_URL (HTTP 403)"
        log "  Check 'allow' directives in stub_status location block"
        show_stub_status_help
        return 1
    elif [[ "$response" == "404" ]]; then
        log "WARNING: stub_status endpoint not found at $STUB_STATUS_URL (HTTP 404)"
        log "  stub_status may not be configured"
        show_stub_status_help
        return 1
    else
        log "WARNING: Unexpected response from $STUB_STATUS_URL (HTTP $response)"
        show_stub_status_help
        return 1
    fi
}

show_stub_status_help() {
    log ""
    log "To enable stub_status, add this to your nginx config:"
    log ""
    log "  server {"
    log "      listen 127.0.0.1:80;"
    log "      server_name localhost;"
    log "      "
    log "      location /nginx_status {"
    log "          stub_status on;"
    log "          access_log off;"
    log "          allow 127.0.0.1;"
    log "          deny all;"
    log "      }"
    log "  }"
    log ""
    log "Then reload nginx: nginx -t && systemctl reload nginx"
    log ""
    log "Or specify a different URL with: --stub-url <url>"
    log ""
}

#########################
### Stub Status Metrics ###
#########################

collect_stub_status() {
    local stub_output
    
    echo "# HELP nginx_up Whether nginx stub_status is reachable"
    echo "# TYPE nginx_up gauge"
    
    if ! stub_output=$(curl -s --max-time 5 "$STUB_STATUS_URL" 2>/dev/null); then
        echo "nginx_up 0"
        return
    fi
    
    echo "nginx_up 1"
    
    # Parse stub_status output
    # Format:
    # Active connections: 43
    # server accepts handled requests
    #  7368 7368 10993
    # Reading: 0 Writing: 5 Waiting: 38
    
    local active_connections accepts handled requests reading writing waiting
    
    active_connections=$(echo "$stub_output" | grep -oP 'Active connections:\s*\K\d+' || echo "0")
    
    # Parse the accepts/handled/requests line
    local stats_line
    stats_line=$(echo "$stub_output" | grep -E '^\s*[0-9]+\s+[0-9]+\s+[0-9]+' || echo "0 0 0")
    accepts=$(echo "$stats_line" | awk '{print $1}')
    handled=$(echo "$stats_line" | awk '{print $2}')
    requests=$(echo "$stats_line" | awk '{print $3}')
    
    # Parse reading/writing/waiting
    reading=$(echo "$stub_output" | grep -oP 'Reading:\s*\K\d+' || echo "0")
    writing=$(echo "$stub_output" | grep -oP 'Writing:\s*\K\d+' || echo "0")
    waiting=$(echo "$stub_output" | grep -oP 'Waiting:\s*\K\d+' || echo "0")
    
    cat <<EOF
# HELP nginx_connections_active Current number of active client connections
# TYPE nginx_connections_active gauge
nginx_connections_active $active_connections

# HELP nginx_connections_accepted Total number of accepted client connections
# TYPE nginx_connections_accepted counter
nginx_connections_accepted $accepts

# HELP nginx_connections_handled Total number of handled connections
# TYPE nginx_connections_handled counter
nginx_connections_handled $handled

# HELP nginx_http_requests_total Total number of client requests
# TYPE nginx_http_requests_total counter
nginx_http_requests_total $requests

# HELP nginx_connections_reading Current number of connections where nginx is reading request header
# TYPE nginx_connections_reading gauge
nginx_connections_reading $reading

# HELP nginx_connections_writing Current number of connections where nginx is writing response
# TYPE nginx_connections_writing gauge
nginx_connections_writing $writing

# HELP nginx_connections_waiting Current number of idle client connections waiting for request
# TYPE nginx_connections_waiting gauge
nginx_connections_waiting $waiting
EOF
}

#########################
### Process Metrics ###
#########################

collect_process_metrics() {
    local nginx_master_pid nginx_pids worker_count total_memory total_cpu total_fds
    local total_connections total_threads
    
    # Find nginx master process
    nginx_master_pid=$(pgrep -x nginx -o 2>/dev/null || pidof nginx 2>/dev/null | awk '{print $1}' || echo "")
    
    if [[ -z "$nginx_master_pid" ]]; then
        echo "# HELP nginx_process_running Whether nginx process is running"
        echo "# TYPE nginx_process_running gauge"
        echo "nginx_process_running 0"
        return
    fi
    
    echo "# HELP nginx_process_running Whether nginx process is running"
    echo "# TYPE nginx_process_running gauge"
    echo "nginx_process_running 1"
    
    # Get all nginx PIDs
    nginx_pids=$(pgrep -x nginx 2>/dev/null || pidof nginx 2>/dev/null || echo "")
    
    # Count workers (total processes minus master)
    worker_count=$(echo "$nginx_pids" | wc -w)
    if [[ $worker_count -gt 0 ]]; then
        worker_count=$((worker_count - 1))  # Subtract master
    fi
    
    echo "# HELP nginx_workers_count Number of nginx worker processes"
    echo "# TYPE nginx_workers_count gauge"
    echo "nginx_workers_count $worker_count"
    
    # Calculate total memory usage (RSS in bytes)
    total_memory=0
    total_cpu=0
    total_fds=0
    total_threads=0
    
    for pid in $nginx_pids; do
        if [[ -d "/proc/$pid" ]]; then
            # Memory (RSS in KB from /proc/pid/status, convert to bytes)
            local rss
            rss=$(grep -m1 'VmRSS:' "/proc/$pid/status" 2>/dev/null | awk '{print $2}' || echo "0")
            total_memory=$((total_memory + rss * 1024))
            
            # CPU time (from /proc/pid/stat - utime + stime in jiffies)
            local stat_line utime stime
            if stat_line=$(cat "/proc/$pid/stat" 2>/dev/null); then
                utime=$(echo "$stat_line" | awk '{print $14}')
                stime=$(echo "$stat_line" | awk '{print $15}')
                total_cpu=$((total_cpu + utime + stime))
            fi
            
            # Open file descriptors
            local fds
            fds=$(ls -1 "/proc/$pid/fd" 2>/dev/null | wc -l || echo "0")
            total_fds=$((total_fds + fds))
            
            # Threads
            local threads
            threads=$(grep -c '^Threads:' "/proc/$pid/status" 2>/dev/null || echo "0")
            if [[ "$threads" -eq 0 ]]; then
                threads=$(grep 'Threads:' "/proc/$pid/status" 2>/dev/null | awk '{print $2}' || echo "1")
            fi
            total_threads=$((total_threads + threads))
        fi
    done
    
    # Convert CPU jiffies to seconds (assuming 100 Hz)
    local cpu_seconds
    cpu_seconds=$(echo "scale=2; $total_cpu / 100" | bc 2>/dev/null || echo "$total_cpu")
    
    cat <<EOF
# HELP nginx_process_memory_bytes Total memory used by nginx processes (RSS)
# TYPE nginx_process_memory_bytes gauge
nginx_process_memory_bytes $total_memory

# HELP nginx_process_cpu_seconds_total Total CPU time consumed by nginx processes
# TYPE nginx_process_cpu_seconds_total counter
nginx_process_cpu_seconds_total $cpu_seconds

# HELP nginx_process_open_fds Total number of open file descriptors
# TYPE nginx_process_open_fds gauge
nginx_process_open_fds $total_fds

# HELP nginx_process_threads_total Total number of threads
# TYPE nginx_process_threads_total gauge
nginx_process_threads_total $total_threads

# HELP nginx_process_start_time_seconds Start time of nginx master process
# TYPE nginx_process_start_time_seconds gauge
EOF
    
    # Get start time of master process
    if [[ -f "/proc/$nginx_master_pid/stat" ]]; then
        local starttime boot_time start_seconds
        starttime=$(awk '{print $22}' "/proc/$nginx_master_pid/stat" 2>/dev/null || echo "0")
        boot_time=$(awk '{print $1}' /proc/stat 2>/dev/null | head -1 || echo "0")
        # starttime is in jiffies since boot
        start_seconds=$(awk "BEGIN {printf \"%.0f\", $(cat /proc/uptime | awk '{print $1}') - ($starttime / 100)}")
        local now_epoch
        now_epoch=$(date +%s)
        local process_start=$((now_epoch - start_seconds))
        echo "nginx_process_start_time_seconds $process_start"
    else
        echo "nginx_process_start_time_seconds 0"
    fi
    
    # Get max open files limit
    if [[ -f "/proc/$nginx_master_pid/limits" ]]; then
        local max_fds
        max_fds=$(grep 'Max open files' "/proc/$nginx_master_pid/limits" 2>/dev/null | awk '{print $4}' || echo "0")
        echo ""
        echo "# HELP nginx_process_max_fds Maximum number of open file descriptors"
        echo "# TYPE nginx_process_max_fds gauge"
        echo "nginx_process_max_fds $max_fds"
    fi
}

#########################
### Config Metrics ###
#########################

collect_config_metrics() {
    if [[ ! -f "$NGINX_CONF" ]]; then
        echo "# nginx.conf not found at $NGINX_CONF"
        return
    fi
    
    local worker_processes worker_connections keepalive_timeout
    local multi_accept use_epoll sendfile tcp_nopush tcp_nodelay gzip_enabled
    
    # Parse worker_processes (can be 'auto' or a number)
    worker_processes=$(grep -E '^\s*worker_processes' "$NGINX_CONF" 2>/dev/null | head -1 | awk '{print $2}' | tr -d ';' || echo "auto")
    if [[ "$worker_processes" == "auto" ]]; then
        worker_processes=$(nproc 2>/dev/null || echo "1")
    fi
    
    # Parse worker_connections
    worker_connections=$(grep -E '^\s*worker_connections' "$NGINX_CONF" 2>/dev/null | head -1 | awk '{print $2}' | tr -d ';' || echo "0")
    
    # Parse keepalive_timeout
    keepalive_timeout=$(grep -E '^\s*keepalive_timeout' "$NGINX_CONF" 2>/dev/null | head -1 | awk '{print $2}' | tr -d ';s' || echo "0")
    
    # Check various settings
    multi_accept=$(grep -qE '^\s*multi_accept\s+on' "$NGINX_CONF" 2>/dev/null && echo "1" || echo "0")
    use_epoll=$(grep -qE '^\s*use\s+epoll' "$NGINX_CONF" 2>/dev/null && echo "1" || echo "0")
    sendfile=$(grep -qE '^\s*sendfile\s+on' "$NGINX_CONF" 2>/dev/null && echo "1" || echo "0")
    tcp_nopush=$(grep -qE '^\s*tcp_nopush\s+on' "$NGINX_CONF" 2>/dev/null && echo "1" || echo "0")
    tcp_nodelay=$(grep -qE '^\s*tcp_nodelay\s+on' "$NGINX_CONF" 2>/dev/null && echo "1" || echo "0")
    gzip_enabled=$(grep -qE '^\s*gzip\s+on' "$NGINX_CONF" 2>/dev/null && echo "1" || echo "0")
    
    cat <<EOF
# HELP nginx_config_worker_processes Configured number of worker processes
# TYPE nginx_config_worker_processes gauge
nginx_config_worker_processes $worker_processes

# HELP nginx_config_worker_connections Maximum connections per worker
# TYPE nginx_config_worker_connections gauge
nginx_config_worker_connections $worker_connections

# HELP nginx_config_keepalive_timeout_seconds Keepalive timeout in seconds
# TYPE nginx_config_keepalive_timeout_seconds gauge
nginx_config_keepalive_timeout_seconds $keepalive_timeout

# HELP nginx_config_multi_accept Whether multi_accept is enabled
# TYPE nginx_config_multi_accept gauge
nginx_config_multi_accept $multi_accept

# HELP nginx_config_use_epoll Whether epoll is configured
# TYPE nginx_config_use_epoll gauge
nginx_config_use_epoll $use_epoll

# HELP nginx_config_sendfile Whether sendfile is enabled
# TYPE nginx_config_sendfile gauge
nginx_config_sendfile $sendfile

# HELP nginx_config_tcp_nopush Whether tcp_nopush is enabled
# TYPE nginx_config_tcp_nopush gauge
nginx_config_tcp_nopush $tcp_nopush

# HELP nginx_config_tcp_nodelay Whether tcp_nodelay is enabled
# TYPE nginx_config_tcp_nodelay gauge
nginx_config_tcp_nodelay $tcp_nodelay

# HELP nginx_config_gzip Whether gzip is enabled
# TYPE nginx_config_gzip gauge
nginx_config_gzip $gzip_enabled
EOF

    # Count virtual hosts
    local vhost_count=0
    if [[ -d "$SITES_DIR" ]]; then
        vhost_count=$(find "$SITES_DIR" -type f -o -type l 2>/dev/null | wc -l)
    elif [[ -d "$CONF_D_DIR" ]]; then
        vhost_count=$(find "$CONF_D_DIR" -name "*.conf" -type f 2>/dev/null | wc -l)
    fi
    
    echo ""
    echo "# HELP nginx_config_vhosts_total Number of configured virtual hosts"
    echo "# TYPE nginx_config_vhosts_total gauge"
    echo "nginx_config_vhosts_total $vhost_count"
    
    # Calculate max possible connections
    local max_connections=$((worker_processes * worker_connections))
    echo ""
    echo "# HELP nginx_config_max_connections Maximum theoretical connections (workers * connections)"
    echo "# TYPE nginx_config_max_connections gauge"
    echo "nginx_config_max_connections $max_connections"
}

#########################
### Access Log Metrics ###
#########################

collect_access_log_metrics() {
    if [[ ! -f "$ACCESS_LOG" ]] || [[ ! -r "$ACCESS_LOG" ]]; then
        echo "# Access log not readable at $ACCESS_LOG"
        return
    fi
    
    local now
    now=$(date +%s)
    
    # Only parse logs every LOG_PARSE_INTERVAL seconds
    if [[ -f "$STATE_DIR/last_parse" ]]; then
        LAST_LOG_PARSE=$(cat "$STATE_DIR/last_parse")
    fi
    
    if [[ $((now - LAST_LOG_PARSE)) -lt $LOG_PARSE_INTERVAL ]] && [[ -f "$STATE_DIR/log_metrics" ]]; then
        cat "$STATE_DIR/log_metrics"
        return
    fi
    
    echo "$now" > "$STATE_DIR/last_parse"
    
    # Parse access log for status codes and other metrics
    # Assuming combined log format: $remote_addr - $remote_user [$time_local] "$request" $status $body_bytes_sent "$http_referer" "$http_user_agent"
    
    local log_data
    log_data=$(tail -n "$LOG_TAIL_LINES" "$ACCESS_LOG" 2>/dev/null || echo "")
    
    if [[ -z "$log_data" ]]; then
        echo "# No log data available"
        return
    fi
    
    local metrics_output=""
    
    # Count by status code
    local status_counts
    status_counts=$(echo "$log_data" | awk '{print $9}' | grep -E '^[0-9]{3}$' | sort | uniq -c | sort -rn)
    
    metrics_output+="# HELP nginx_http_requests_by_status_total HTTP requests by status code (from last $LOG_TAIL_LINES log lines)
# TYPE nginx_http_requests_by_status_total gauge
"
    
    # Initialize counters for status code groups
    local count_1xx=0 count_2xx=0 count_3xx=0 count_4xx=0 count_5xx=0
    
    while read -r count status; do
        if [[ -n "$status" ]] && [[ -n "$count" ]]; then
            metrics_output+="nginx_http_requests_by_status_total{status=\"$status\"} $count
"
            # Aggregate by category
            case "${status:0:1}" in
                1) count_1xx=$((count_1xx + count)) ;;
                2) count_2xx=$((count_2xx + count)) ;;
                3) count_3xx=$((count_3xx + count)) ;;
                4) count_4xx=$((count_4xx + count)) ;;
                5) count_5xx=$((count_5xx + count)) ;;
            esac
        fi
    done <<< "$status_counts"
    
    metrics_output+="
# HELP nginx_http_requests_by_status_class_total HTTP requests by status class
# TYPE nginx_http_requests_by_status_class_total gauge
nginx_http_requests_by_status_class_total{class=\"1xx\"} $count_1xx
nginx_http_requests_by_status_class_total{class=\"2xx\"} $count_2xx
nginx_http_requests_by_status_class_total{class=\"3xx\"} $count_3xx
nginx_http_requests_by_status_class_total{class=\"4xx\"} $count_4xx
nginx_http_requests_by_status_class_total{class=\"5xx\"} $count_5xx
"
    
    # Calculate total bytes sent
    local total_bytes
    total_bytes=$(echo "$log_data" | awk '{sum += $10} END {print sum+0}')
    
    metrics_output+="
# HELP nginx_http_response_bytes_total Total bytes sent in responses (from last $LOG_TAIL_LINES log lines)
# TYPE nginx_http_response_bytes_total gauge
nginx_http_response_bytes_total $total_bytes
"
    
    # Count requests by method
    local method_counts
    method_counts=$(echo "$log_data" | awk -F'"' '{print $2}' | awk '{print $1}' | grep -E '^(GET|POST|PUT|DELETE|PATCH|HEAD|OPTIONS)$' | sort | uniq -c)
    
    metrics_output+="
# HELP nginx_http_requests_by_method_total HTTP requests by method (from last $LOG_TAIL_LINES log lines)
# TYPE nginx_http_requests_by_method_total gauge
"
    
    while read -r count method; do
        if [[ -n "$method" ]] && [[ -n "$count" ]]; then
            metrics_output+="nginx_http_requests_by_method_total{method=\"$method\"} $count
"
        fi
    done <<< "$method_counts"
    
    # Count unique IPs
    local unique_ips
    unique_ips=$(echo "$log_data" | awk '{print $1}' | sort -u | wc -l)
    
    metrics_output+="
# HELP nginx_http_unique_clients Unique client IPs (from last $LOG_TAIL_LINES log lines)
# TYPE nginx_http_unique_clients gauge
nginx_http_unique_clients $unique_ips
"
    
    # Top URIs (for potential abuse detection)
    local top_uris
    top_uris=$(echo "$log_data" | awk -F'"' '{print $2}' | awk '{print $2}' | grep -v '^-$' | sort | uniq -c | sort -rn | head -5)
    
    metrics_output+="
# HELP nginx_http_top_uri_requests_total Top requested URIs (from last $LOG_TAIL_LINES log lines)
# TYPE nginx_http_top_uri_requests_total gauge
"
    
    local rank=1
    while read -r count uri; do
        if [[ -n "$uri" ]] && [[ -n "$count" ]]; then
            # Truncate URI and escape quotes
            uri="${uri:0:100}"
            uri="${uri//\"/\\\"}"
            metrics_output+="nginx_http_top_uri_requests_total{uri=\"$uri\",rank=\"$rank\"} $count
"
            rank=$((rank + 1))
        fi
    done <<< "$top_uris"
    
    # Count requests in time windows (if log has parseable timestamps)
    local recent_requests
    recent_requests=$(echo "$log_data" | wc -l)
    
    metrics_output+="
# HELP nginx_http_requests_in_sample Total requests in sample window
# TYPE nginx_http_requests_in_sample gauge
nginx_http_requests_in_sample $recent_requests
"
    
    # Save metrics for caching
    echo "$metrics_output" > "$STATE_DIR/log_metrics"
    echo "$metrics_output"
}

#########################
### Error Log Metrics ###
#########################

collect_error_log_metrics() {
    if [[ ! -f "$ERROR_LOG" ]] || [[ ! -r "$ERROR_LOG" ]]; then
        echo "# Error log not readable at $ERROR_LOG"
        return
    fi
    
    # Count errors by level from last 1000 lines
    local log_data
    log_data=$(tail -n 1000 "$ERROR_LOG" 2>/dev/null || echo "")
    
    if [[ -z "$log_data" ]]; then
        return
    fi
    
    local emerg_count alert_count crit_count error_count warn_count notice_count info_count
    
    emerg_count=$(echo "$log_data" | grep -c '\[emerg\]' 2>/dev/null) || emerg_count=0
    alert_count=$(echo "$log_data" | grep -c '\[alert\]' 2>/dev/null) || alert_count=0
    crit_count=$(echo "$log_data" | grep -c '\[crit\]' 2>/dev/null) || crit_count=0
    error_count=$(echo "$log_data" | grep -c '\[error\]' 2>/dev/null) || error_count=0
    warn_count=$(echo "$log_data" | grep -c '\[warn\]' 2>/dev/null) || warn_count=0
    notice_count=$(echo "$log_data" | grep -c '\[notice\]' 2>/dev/null) || notice_count=0
    info_count=$(echo "$log_data" | grep -c '\[info\]' 2>/dev/null) || info_count=0
    
    cat <<EOF
# HELP nginx_error_log_messages_total Error log messages by level (from last 1000 lines)
# TYPE nginx_error_log_messages_total gauge
nginx_error_log_messages_total{level="emerg"} $emerg_count
nginx_error_log_messages_total{level="alert"} $alert_count
nginx_error_log_messages_total{level="crit"} $crit_count
nginx_error_log_messages_total{level="error"} $error_count
nginx_error_log_messages_total{level="warn"} $warn_count
nginx_error_log_messages_total{level="notice"} $notice_count
nginx_error_log_messages_total{level="info"} $info_count
EOF

    # Check error log file size and age
    local log_size log_mtime now log_age
    log_size=$(stat -c %s "$ERROR_LOG" 2>/dev/null || echo "0")
    log_mtime=$(stat -c %Y "$ERROR_LOG" 2>/dev/null || echo "0")
    now=$(date +%s)
    log_age=$((now - log_mtime))
    
    cat <<EOF

# HELP nginx_error_log_size_bytes Size of error log file
# TYPE nginx_error_log_size_bytes gauge
nginx_error_log_size_bytes $log_size

# HELP nginx_error_log_last_modified_seconds Seconds since error log was last modified
# TYPE nginx_error_log_last_modified_seconds gauge
nginx_error_log_last_modified_seconds $log_age
EOF
}

#########################
### SSL Certificate Metrics ###
#########################

collect_ssl_metrics() {
    local domains="$SSL_CHECK_DOMAINS"
    
    # If no domains specified, try to find them from nginx config
    if [[ -z "$domains" ]]; then
        # Look for ssl_certificate directives
        local cert_files
        cert_files=$(grep -rh 'ssl_certificate\s' "$SITES_DIR" "$CONF_D_DIR" "$NGINX_CONF" 2>/dev/null | grep -v '#' | grep -v 'ssl_certificate_key' | awk '{print $2}' | tr -d ';' | sort -u || echo "")
        
        if [[ -z "$cert_files" ]]; then
            echo "# No SSL certificates found in nginx config"
            return
        fi
        
        echo "# HELP nginx_ssl_certificate_expiry_days Days until SSL certificate expires"
        echo "# TYPE nginx_ssl_certificate_expiry_days gauge"
        echo "# HELP nginx_ssl_certificate_expiry_timestamp Unix timestamp when certificate expires"
        echo "# TYPE nginx_ssl_certificate_expiry_timestamp gauge"
        
        while read -r cert_file; do
            if [[ -f "$cert_file" ]]; then
                local expiry_date expiry_epoch now_epoch days_left cn
                
                expiry_date=$(openssl x509 -enddate -noout -in "$cert_file" 2>/dev/null | cut -d= -f2 || echo "")
                if [[ -n "$expiry_date" ]]; then
                    expiry_epoch=$(date -d "$expiry_date" +%s 2>/dev/null || echo "0")
                    now_epoch=$(date +%s)
                    days_left=$(( (expiry_epoch - now_epoch) / 86400 ))
                    
                    # Get CN from certificate
                    cn=$(openssl x509 -subject -noout -in "$cert_file" 2>/dev/null | grep -oP 'CN\s*=\s*\K[^,/]+' || basename "$cert_file")
                    cn="${cn// /_}"
                    
                    echo "nginx_ssl_certificate_expiry_days{certificate=\"$cn\",file=\"$cert_file\"} $days_left"
                    echo "nginx_ssl_certificate_expiry_timestamp{certificate=\"$cn\",file=\"$cert_file\"} $expiry_epoch"
                fi
            fi
        done <<< "$cert_files"
        return
    fi
    
    # Check specified domains via network
    echo "# HELP nginx_ssl_certificate_expiry_days Days until SSL certificate expires"
    echo "# TYPE nginx_ssl_certificate_expiry_days gauge"
    echo "# HELP nginx_ssl_certificate_expiry_timestamp Unix timestamp when certificate expires"  
    echo "# TYPE nginx_ssl_certificate_expiry_timestamp gauge"
    
    IFS=',' read -ra domain_array <<< "$domains"
    for domain in "${domain_array[@]}"; do
        domain=$(echo "$domain" | tr -d ' ')
        if [[ -n "$domain" ]]; then
            local expiry_date expiry_epoch now_epoch days_left
            
            expiry_date=$(echo | openssl s_client -servername "$domain" -connect "$domain:443" 2>/dev/null | openssl x509 -noout -enddate 2>/dev/null | cut -d= -f2 || echo "")
            
            if [[ -n "$expiry_date" ]]; then
                expiry_epoch=$(date -d "$expiry_date" +%s 2>/dev/null || echo "0")
                now_epoch=$(date +%s)
                days_left=$(( (expiry_epoch - now_epoch) / 86400 ))
                
                echo "nginx_ssl_certificate_expiry_days{domain=\"$domain\"} $days_left"
                echo "nginx_ssl_certificate_expiry_timestamp{domain=\"$domain\"} $expiry_epoch"
            else
                echo "nginx_ssl_certificate_expiry_days{domain=\"$domain\"} -1"
            fi
        fi
    done
}

#########################
### Upstream Metrics ###
#########################

collect_upstream_metrics() {
    # Check for upstream configurations
    local upstreams
    upstreams=$(grep -rh 'upstream\s' "$SITES_DIR" "$CONF_D_DIR" "$NGINX_CONF" 2>/dev/null | grep -v '#' | awk '{print $2}' | tr -d '{' | sort -u || echo "")
    
    if [[ -z "$upstreams" ]]; then
        return
    fi
    
    echo "# HELP nginx_upstream_configured Number of configured upstreams"
    echo "# TYPE nginx_upstream_configured gauge"
    echo "nginx_upstream_configured $(echo "$upstreams" | wc -w)"
    
    echo ""
    echo "# HELP nginx_upstream_servers_total Servers configured per upstream"
    echo "# TYPE nginx_upstream_servers_total gauge"
    
    # This is a simplified count - for real upstream health you'd need nginx-plus or lua module
    while read -r upstream; do
        if [[ -n "$upstream" ]]; then
            # Try to count servers in this upstream block
            local server_count
            server_count=$(grep -A 20 "upstream $upstream" "$SITES_DIR"/* "$CONF_D_DIR"/* "$NGINX_CONF" 2>/dev/null | grep -c 'server\s' 2>/dev/null) || server_count=0
            echo "nginx_upstream_servers_total{upstream=\"$upstream\"} $server_count"
        fi
    done <<< "$upstreams"
}

#########################
### Version Metrics ###
#########################

collect_version_metrics() {
    local version
    version=$(nginx -v 2>&1 | grep -oP 'nginx/\K[0-9.]+' || echo "unknown")
    
    echo "# HELP nginx_version_info Nginx version information"
    echo "# TYPE nginx_version_info gauge"
    echo "nginx_version_info{version=\"$version\"} 1"
    
    # Check if nginx is compiled with certain modules
    local modules_output
    modules_output=$(nginx -V 2>&1 || echo "")
    
    local has_ssl has_http2 has_gzip has_stub_status has_realip has_geoip has_lua
    has_ssl=$(echo "$modules_output" | grep -qE 'with-http_ssl_module|--with-openssl' && echo "1" || echo "0")
    has_http2=$(echo "$modules_output" | grep -q 'http_v2_module' && echo "1" || echo "0")
    has_gzip=$(echo "$modules_output" | grep -q 'http_gzip' && echo "1" || echo "0")
    has_stub_status=$(echo "$modules_output" | grep -q 'http_stub_status_module' && echo "1" || echo "0")
    has_realip=$(echo "$modules_output" | grep -q 'http_realip_module' && echo "1" || echo "0")
    has_geoip=$(echo "$modules_output" | grep -q 'http_geoip_module' && echo "1" || echo "0")
    has_lua=$(echo "$modules_output" | grep -qE 'lua|ndk_http_module' && echo "1" || echo "0")
    
    cat <<EOF

# HELP nginx_module_enabled Whether nginx module is compiled in
# TYPE nginx_module_enabled gauge
nginx_module_enabled{module="ssl"} $has_ssl
nginx_module_enabled{module="http2"} $has_http2
nginx_module_enabled{module="gzip"} $has_gzip
nginx_module_enabled{module="stub_status"} $has_stub_status
nginx_module_enabled{module="realip"} $has_realip
nginx_module_enabled{module="geoip"} $has_geoip
nginx_module_enabled{module="lua"} $has_lua
EOF
}

#########################
### File Descriptor Metrics ###
#########################

collect_system_metrics() {
    # System-wide limits that affect nginx
    local max_files ulimit_n
    max_files=$(cat /proc/sys/fs/file-max 2>/dev/null || echo "0")
    ulimit_n=$(ulimit -n 2>/dev/null || echo "0")
    
    cat <<EOF
# HELP nginx_system_file_max System-wide maximum file descriptors
# TYPE nginx_system_file_max gauge
nginx_system_file_max $max_files

# HELP nginx_system_ulimit_n Current shell ulimit for open files
# TYPE nginx_system_ulimit_n gauge
nginx_system_ulimit_n $ulimit_n
EOF

    # Current system-wide open files
    local open_files
    open_files=$(cat /proc/sys/fs/file-nr 2>/dev/null | awk '{print $1}' || echo "0")
    
    echo ""
    echo "# HELP nginx_system_open_files Current system-wide open files"
    echo "# TYPE nginx_system_open_files gauge"
    echo "nginx_system_open_files $open_files"
}

#########################
### Collect All Metrics ###
#########################

collect_all_metrics() {
    local hostname
    hostname=$(hostname -f 2>/dev/null || hostname)
    
    cat <<EOF
# Nginx Metrics Exporter
# Host: $hostname
# Collected at: $(date -Iseconds)

EOF
    
    collect_version_metrics
    echo ""
    collect_stub_status
    echo ""
    collect_process_metrics
    echo ""
    collect_config_metrics
    echo ""
    collect_access_log_metrics
    echo ""
    collect_error_log_metrics
    echo ""
    collect_ssl_metrics
    echo ""
    collect_upstream_metrics
    echo ""
    collect_system_metrics
}

#########################
### HTTP Server ###
#########################

handle_request() {
    local request_line=""
    local content_length=0
    
    # Read request
    while IFS= read -r line; do
        line="${line%%$'\r'}"
        [[ -z "$line" ]] && break
        [[ -z "$request_line" ]] && request_line="$line"
    done
    
    local method path
    method=$(echo "$request_line" | awk '{print $1}')
    path=$(echo "$request_line" | awk '{print $2}')
    
    case "$path" in
        /metrics|/)
            local metrics
            metrics=$(collect_all_metrics)
            local body_length=${#metrics}
            
            cat <<EOF
HTTP/1.1 200 OK
Content-Type: text/plain; charset=utf-8
Content-Length: $body_length
Connection: close

$metrics
EOF
            ;;
        /health|/healthz)
            local health_body="OK"
            cat <<EOF
HTTP/1.1 200 OK
Content-Type: text/plain
Content-Length: 2
Connection: close

OK
EOF
            ;;
        *)
            cat <<EOF
HTTP/1.1 404 Not Found
Content-Type: text/plain
Content-Length: 9
Connection: close

Not Found
EOF
            ;;
    esac
}

start_server() {
    log "Starting Nginx Metrics Exporter on port $LISTEN_PORT"
    log "Metrics available at http://localhost:$LISTEN_PORT/metrics"
    log "Stub status URL: $STUB_STATUS_URL"
    
    while true; do
        socat TCP-LISTEN:"$LISTEN_PORT",reuseaddr,fork EXEC:"$0 --handle-request" 2>/dev/null || {
            log "Server error, restarting in 5 seconds..."
            sleep 5
        }
    done
}

#########################
### Output ###
#########################

write_output() {
    local metrics
    metrics=$(collect_all_metrics)
    
    if [[ -n "$OUTPUT_FILE" ]]; then
        local tmp_file="${OUTPUT_FILE}.$$"
        echo "$metrics" > "$tmp_file"
        mv "$tmp_file" "$OUTPUT_FILE"
    else
        echo "$metrics"
    fi
}

#########################
### Main ###
#########################

main() {
    if [[ "${1:-}" == "--handle-request" ]]; then
        handle_request
        exit 0
    fi
    
    parse_args "$@"
    setup
    
    if [[ "$HTTP_MODE" == true ]]; then
        start_server
    elif [[ -n "$OUTPUT_FILE" ]]; then
        write_output
    else
        collect_all_metrics
    fi
}

main "$@"