#!/usr/bin/env bash

#########################################################################################
#### logrotate-check-exporter.sh — Logrotate health metrics for Prometheus           ####
#### Tracks rotation timestamps, log file sizes/ages, and stale log detection        ####
#### Requires: bash 4+, coreutils                                                    ####
####                                                                                 ####
#### Author: Phil Connor                                                             ####
#### Contact: contact@mylinux.work                                                   ####
#### License: MIT                                                                    ####
#### Version: 1.2                                                                    ####
####                                                                                 ####
#### Usage:                                                                          ####
####   ./logrotate-check-exporter.sh                    # stdout                     ####
####   ./logrotate-check-exporter.sh --textfile         # node_exporter textfile     ####
####   ./logrotate-check-exporter.sh --daemon           # continuous collection      ####
####                                                                                 ####
#### See --help for all options.                                                     ####
#########################################################################################

set -euo pipefail

SCRIPT_NAME=$(basename "$0")
readonly SCRIPT_NAME

# Default configuration
TEXTFILE_DIR="/var/lib/node_exporter"
OUTPUT_FILE=""
LOG_DIR="${LOG_DIR:-/var/log}"
WATCH_PATHS=""
readonly DEFAULT_STALE_THRESHOLD=172800
readonly DEFAULT_COLLECTION_INTERVAL=60

# Configuration variables (can be overridden by environment)
STALE_THRESHOLD=${STALE_THRESHOLD:-$DEFAULT_STALE_THRESHOLD}
COLLECTION_INTERVAL=${COLLECTION_INTERVAL:-$DEFAULT_COLLECTION_INTERVAL}
DEBUG=${DEBUG:-}

# Runtime flags
RUN_MODE="once"

debug_echo() {
    if [[ -n "$DEBUG" ]]; then
        echo "[DEBUG] $*" >&2
    fi
}

show_help() {
    cat << EOF
Usage: $SCRIPT_NAME [OPTIONS]

Logrotate health metrics for Prometheus (v1.2).

Parses the logrotate status file to track rotation timestamps, monitors log file
sizes and ages, and detects stale logs that haven't been rotated within a
configurable threshold.

By default, discovers all log files in $LOG_DIR (excluding compressed archives).
Use --watch to monitor specific files instead.

MODES:
    --textfile      Write to node_exporter textfile collector
    --daemon        Run continuously at COLLECTION_INTERVAL
    (default)       Output metrics to stdout

OPTIONS:
    --watch PATHS   Comma-separated log file paths to monitor instead of auto-discovery
    --log-dir DIR   Base directory for auto-discovery (default: /var/log)
    -o, --output    Output file path
    -h, --help      Show this help message

ENVIRONMENT VARIABLES:
    LOG_DIR                 Base directory for log file discovery (default: /var/log)
    STALE_THRESHOLD         Seconds before a file is considered stale (default: $DEFAULT_STALE_THRESHOLD = 48h)
    COLLECTION_INTERVAL     Seconds between collections in daemon mode (default: $DEFAULT_COLLECTION_INTERVAL)
    DEBUG                   Enable debug output

EXAMPLES:
    $SCRIPT_NAME                                        # Auto-discover all logs in /var/log
    $SCRIPT_NAME --textfile                              # Write to textfile collector
    $SCRIPT_NAME --watch /var/log/syslog,/var/log/auth.log  # Monitor specific files
    $SCRIPT_NAME --log-dir /opt/app/logs                # Scan a custom log directory
    $SCRIPT_NAME --daemon                                # Continuous collection

METRICS:
    - logrotate_last_run_timestamp     Unix timestamp of last logrotate run
    - logrotate_status                 Whether logrotate has run recently (1/0)
    - logrotate_files_total            Files tracked in logrotate status
    - logrotate_stale_files_total      Files not rotated within threshold
    - log_file_size_bytes              Size of monitored log files
    - log_file_age_seconds             Age of monitored log files

EOF
    exit 0
}

# Parse arguments
while [[ $# -gt 0 ]]; do
    case "$1" in
        --textfile) OUTPUT_FILE="$TEXTFILE_DIR/logrotate_check.prom"; shift ;;
        --daemon)   RUN_MODE="daemon"; shift ;;
        --watch)    WATCH_PATHS="${2:?--watch requires paths}"; shift 2 ;;
        --log-dir)  LOG_DIR="${2:?--log-dir requires a path}"; shift 2 ;;
        -o|--output) OUTPUT_FILE="${2:?--output requires a path}"; shift 2 ;;
        --help|-h)  show_help ;;
        *) echo "Unknown option: $1" >&2; show_help ;;
    esac
done

# Locate the logrotate status file
find_status_file() {
    if [[ -f /var/lib/logrotate/status ]]; then
        echo "/var/lib/logrotate/status"
    elif [[ -f /var/lib/logrotate.status ]]; then
        echo "/var/lib/logrotate.status"
    else
        debug_echo "No logrotate status file found"
        echo ""
    fi
}

# Validate output directory exists when writing to file
validate_output() {
    if [[ -n "$OUTPUT_FILE" ]]; then
        local output_dir
        output_dir="$(dirname "$OUTPUT_FILE")"
        if [[ ! -d "$output_dir" ]]; then
            echo "Error: Output directory not found: $output_dir" >&2
            echo "Create it: sudo mkdir -p $output_dir" >&2
            exit 1
        fi
    fi
}

# Collect logrotate status metrics
collect_logrotate_status() {
    local status_file
    status_file=$(find_status_file)

    if [[ -z "$status_file" ]]; then
        cat <<EOF
# HELP logrotate_last_run_timestamp Unix timestamp of last logrotate run.
# TYPE logrotate_last_run_timestamp gauge
logrotate_last_run_timestamp 0

# HELP logrotate_status Whether logrotate has run recently (1=recent, 0=stale).
# TYPE logrotate_status gauge
logrotate_status 0

# HELP logrotate_files_total Total number of files tracked in logrotate status.
# TYPE logrotate_files_total gauge
logrotate_files_total 0

# HELP logrotate_stale_files_total Number of files not rotated within threshold.
# TYPE logrotate_stale_files_total gauge
logrotate_stale_files_total 0

EOF
        return
    fi

    debug_echo "Using status file: $status_file"

    # Last run timestamp from status file mtime
    local last_run_ts
    last_run_ts=$(stat -c %Y "$status_file" 2>/dev/null || echo 0)

    # Determine if logrotate is stale
    local now
    now=$(date +%s)
    local age=$(( now - last_run_ts ))
    local status=1
    if [[ $age -gt $STALE_THRESHOLD ]]; then
        status=0
    fi

    # Count total tracked files and stale files in status
    local files_total=0
    local stale_files=0

    while IFS= read -r line; do
        # Skip header line and empty lines
        [[ "$line" =~ ^\".*\" ]] || continue

        files_total=$(( files_total + 1 ))

        # Extract the date portion — format: "filename" date
        local date_str
        date_str=$(echo "$line" | sed 's/^"[^"]*"[[:space:]]*//')

        if [[ -n "$date_str" ]]; then
            local file_ts
            file_ts=$(date -d "$date_str" +%s 2>/dev/null || echo 0)
            if [[ $file_ts -gt 0 ]]; then
                local file_age=$(( now - file_ts ))
                if [[ $file_age -gt $STALE_THRESHOLD ]]; then
                    stale_files=$(( stale_files + 1 ))
                fi
            fi
        fi
    done < "$status_file"

    debug_echo "Status file: $files_total files tracked, $stale_files stale"

    cat <<EOF
# HELP logrotate_last_run_timestamp Unix timestamp of last logrotate run.
# TYPE logrotate_last_run_timestamp gauge
logrotate_last_run_timestamp $last_run_ts

# HELP logrotate_status Whether logrotate has run recently (1=recent, 0=stale).
# TYPE logrotate_status gauge
logrotate_status $status

# HELP logrotate_files_total Total number of files tracked in logrotate status.
# TYPE logrotate_files_total gauge
logrotate_files_total $files_total

# HELP logrotate_stale_files_total Number of files not rotated within threshold.
# TYPE logrotate_stale_files_total gauge
logrotate_stale_files_total $stale_files

EOF
}

# Discover log files in LOG_DIR (excludes compressed archives)
discover_log_files() {
    find "$LOG_DIR" -maxdepth 2 -type f \
        -not -name '*.gz' -not -name '*.xz' -not -name '*.bz2' \
        -not -name '*.zst' -not -name '*.zip' \
        -not -name '*.prom' -not -name '*.json' \
        -not -name 'wtmp' -not -name 'btmp' -not -name 'lastlog' \
        2>/dev/null | sort
}

# Collect log file size and age metrics
collect_log_files() {
    local now
    now=$(date +%s)

    local size_lines=""
    local age_lines=""
    local has_entries=0

    if [[ -n "$WATCH_PATHS" ]]; then
        # Explicit watch list
        IFS=',' read -ra paths <<< "$WATCH_PATHS"
        for path in "${paths[@]}"; do
            path=$(echo "$path" | xargs)
            [[ -f "$path" ]] || continue

            has_entries=1

            local size
            size=$(stat -c %s "$path" 2>/dev/null || echo 0)
            local mtime
            mtime=$(stat -c %Y "$path" 2>/dev/null || echo 0)
            local age=$(( now - mtime ))

            size_lines+="log_file_size_bytes{path=\"${path}\"} ${size}\n"
            age_lines+="log_file_age_seconds{path=\"${path}\"} ${age}\n"
            debug_echo "Log file: $path size=$size age=${age}s"
        done
    else
        # Auto-discover log files
        debug_echo "Auto-discovering log files in $LOG_DIR"
        while IFS= read -r path; do
            [[ -f "$path" ]] || continue

            has_entries=1

            local size
            size=$(stat -c %s "$path" 2>/dev/null || echo 0)
            local mtime
            mtime=$(stat -c %Y "$path" 2>/dev/null || echo 0)
            local age=$(( now - mtime ))

            size_lines+="log_file_size_bytes{path=\"${path}\"} ${size}\n"
            age_lines+="log_file_age_seconds{path=\"${path}\"} ${age}\n"
            debug_echo "Log file: $path size=$size age=${age}s"
        done < <(discover_log_files)
    fi

    if [[ $has_entries -eq 1 ]]; then
        echo "# HELP log_file_size_bytes Size of monitored log file in bytes."
        echo "# TYPE log_file_size_bytes gauge"
        echo -e "$size_lines"

        echo "# HELP log_file_age_seconds Seconds since last modification of monitored log file."
        echo "# TYPE log_file_age_seconds gauge"
        echo -e "$age_lines"
    fi
}

# Collect exporter metadata
collect_metadata() {
    local duration="$1"
    local success="$2"

    cat <<EOF
# HELP logrotate_exporter_duration_seconds Time taken for the collection run.
# TYPE logrotate_exporter_duration_seconds gauge
logrotate_exporter_duration_seconds $duration

# HELP logrotate_exporter_last_run_timestamp Unix timestamp of last exporter run.
# TYPE logrotate_exporter_last_run_timestamp gauge
logrotate_exporter_last_run_timestamp $(date +%s)

# HELP logrotate_exporter_success Whether the last collection completed successfully.
# TYPE logrotate_exporter_success gauge
logrotate_exporter_success $success

EOF
}

# Generate all metrics to stdout
generate_metrics() {
    local start_time end_time duration
    start_time=$(date +%s%N)
    local success=1

    {
        collect_logrotate_status
        collect_log_files
    } || success=0

    end_time=$(date +%s%N)
    duration=$(awk "BEGIN {printf \"%.3f\", ($end_time - $start_time) / 1000000000}")

    collect_metadata "$duration" "$success"
}

# Write metrics to file atomically
write_metrics() {
    local output_dir
    output_dir="$(dirname "$OUTPUT_FILE")"

    local temp_file
    temp_file=$(mktemp "${output_dir}/.logrotate_check.XXXXXX")

    if ! generate_metrics > "$temp_file" 2>/dev/null; then
        rm -f "$temp_file"
        echo "ERROR: Failed to generate metrics" >&2
        exit 1
    fi

    local file_lines
    file_lines=$(wc -l < "$temp_file" 2>/dev/null || echo 0)

    if [[ "$file_lines" -lt 5 ]]; then
        rm -f "$temp_file"
        echo "ERROR: Metrics file too small ($file_lines lines), keeping previous" >&2
        exit 1
    fi

    chmod 644 "$temp_file"
    mv -f "$temp_file" "$OUTPUT_FILE"

    debug_echo "Metrics written to $OUTPUT_FILE ($file_lines lines)"
}

# Main
main() {
    validate_output

    case "$RUN_MODE" in
        once)
            if [[ -n "$OUTPUT_FILE" ]]; then
                write_metrics
            else
                generate_metrics
            fi
            ;;
        daemon)
            [[ -z "$OUTPUT_FILE" ]] && OUTPUT_FILE="$TEXTFILE_DIR/logrotate_check.prom"
            validate_output
            echo "$SCRIPT_NAME running in daemon mode (interval: ${COLLECTION_INTERVAL}s)" >&2
            while true; do
                write_metrics
                sleep "$COLLECTION_INTERVAL"
            done
            ;;
    esac
}

main