<?php

/**
 * Dedup hash for tickets created through create_ticket_api.php.
 *
 * Two reports with the same hash land on the same ticket (updated while open,
 * reopened when closed), so the components below decide what counts as "the
 * same problem". Changing them re-keys existing tickets: see
 * scripts/rehash_network_tickets.php for how #116 handled that.
 */
class TicketHash
{
    /**
     * The stable components the hash is built from.
     *
     * @param array $data Request payload: title, plus optional serial / check_code.
     */
    public static function components(array $data): array
    {
        $title = (string)($data['title'] ?? '');

        // Prefer explicit serial from payload; fall back to extracting device path from title
        // for backwards compatibility with older hwmonDaemon versions.
        $serial = isset($data['serial']) && $data['serial'] !== null && $data['serial'] !== ''
            ? (string)$data['serial']
            : null;

        // Extract device name if present (matches /dev/sdX, /dev/nvmeXnY patterns)
        preg_match('/\/dev\/(sd[a-z]+|nvme\d+n\d+)/', $title, $deviceMatches);
        $isDriveTicket = !empty($deviceMatches) || $serial !== null;

        // Extract first bracketed tag as hostname/source
        preg_match('/^\[([^\]]+)\]/', $title, $hostMatches);
        $hostname = $hostMatches[1] ?? '';

        // Detect issue category and optional sub-type
        $issueCategory = '';
        $issueSubtype  = '';
        $isClusterWide = false;

        if (stripos($title, 'SMART issues') !== false) {
            $issueCategory = 'smart';
        } elseif (stripos($title, 'ZFS pool') !== false) {
            $issueCategory = 'zfs';
            // Extract pool name so each pool gets its own ticket
            if (preg_match("/ZFS pool '([^']+)'/i", $title, $poolMatch)) {
                $poolName = strtolower(preg_replace('/[^a-z0-9_]/i', '_', $poolMatch[1]));
                if (stripos($title, 'state:') !== false || preg_match('/DEGRADED|FAULTED|UNAVAIL|OFFLINE/i', $title)) {
                    $issueSubtype = 'pool_state_' . $poolName;
                } elseif (stripos($title, 'usage') !== false) {
                    $issueSubtype = 'pool_usage_' . $poolName;
                } elseif (stripos($title, 'errors') !== false) {
                    $issueSubtype = 'pool_errors_' . $poolName;
                } else {
                    $issueSubtype = 'pool_' . $poolName;
                }
            }
        } elseif (stripos($title, 'LXC') !== false || stripos($title, 'storage usage') !== false) {
            $issueCategory = 'storage';
            // Include the LXC container ID so each container gets its own ticket
            if (preg_match('/LXC\s+(\d+)/i', $title, $lxcMatch)) {
                $issueSubtype = 'lxc_' . $lxcMatch[1];
            }
        } elseif (stripos($title, 'memory') !== false) {
            $issueCategory = 'memory';
        } elseif (stripos($title, 'cpu') !== false) {
            $issueCategory = 'cpu';
        } elseif (stripos($title, 'network') !== false) {
            $issueCategory = 'network';
            // One ticket per kind of network problem per host, not one per
            // host. With an empty subtype every network alert for a host
            // (reboot, NIC link-down, cable mismatch, ...) hashed alike, so an
            // interface alert landed on an unrelated closed reboot ticket
            // (#116). Prefer the reporter's check_code when sent; it must name
            // the instance too (e.g. "iface_down:enp3s0"), not just the alert
            // type. Otherwise key on the title text outside the [tags]; the
            // reporters' headings are stable per problem, and digits are kept
            // because they identify it (enp3s0, cable C07). Manual tickets
            // are already keyed on their full title, so they are left as-is.
            $checkCode = isset($data['check_code']) && is_string($data['check_code']) ? $data['check_code'] : '';
            if (stripos($title, '[auto]') === false) {
                // manual: unchanged hash
            } elseif ($checkCode !== '' && preg_match('/^[A-Za-z0-9_.:-]{1,80}$/', $checkCode)) {
                $issueSubtype = 'net_check_' . strtolower($checkCode);
            } else {
                $heading = trim(preg_replace('/\[[^\]]*\]/', '', $title), " -\t");
                $slug = trim(preg_replace('/[^a-z0-9]+/', '_', strtolower($heading)), '_');
                if ($slug !== '') {
                    $issueSubtype = 'net_' . substr($slug, 0, 80);
                }
            }
        } elseif (stripos($title, 'Ceph') !== false || stripos($title, '[ceph]') !== false) {
            $issueCategory = 'ceph';
            if (
                stripos($title, '[cluster-wide]') !== false ||
                stripos($title, 'HEALTH_ERR') !== false ||
                stripos($title, 'HEALTH_WARN') !== false ||
                stripos($title, 'cluster usage') !== false
            ) {
                $isClusterWide = true;
            }
            // Normalize the specific Ceph warning type so different warnings get distinct tickets
            if (stripos($title, 'slow') !== false && stripos($title, 'BlueStore') !== false) {
                $issueSubtype = 'bluestore_slow';
            } elseif (stripos($title, 'clock skew') !== false) {
                $issueSubtype = 'clock_skew';
            } elseif (stripos($title, 'cluster usage') !== false) {
                $issueSubtype = 'usage';
            } elseif (stripos($title, 'OSD down') !== false || preg_match('/osd\.\d+\s+is\s+DOWN/i', $title)) {
                // Include the specific OSD ID so each individual OSD gets its own ticket
                if (preg_match('/osd\.(\d+)/i', $title, $osdMatch)) {
                    $issueSubtype = 'osd_down_' . $osdMatch[1];
                } else {
                    $issueSubtype = 'osd_down';
                }
            } elseif (stripos($title, 'HEALTH_ERR') !== false) {
                $issueSubtype = 'health_err';
            } else {
                // Any other warning gets its own ticket. Previously every other
                // HEALTH_WARN shared one empty subtype, so unrelated warnings
                // (noout, nearfull, insecure keys, ...) collapsed onto a single
                // ticket and overwrote its title in turn every run (#112).
                // Prefer Ceph's own health-check code when the reporter sends
                // it; otherwise key on the warning text with counts stripped.
                $checkCode = isset($data['check_code']) && is_string($data['check_code']) ? $data['check_code'] : '';
                if ($checkCode !== '' && preg_match('/^[A-Z0-9_]{1,64}$/', $checkCode)) {
                    $issueSubtype = 'check_' . strtolower($checkCode);
                } elseif (preg_match('/HEALTH_WARN:\s*([^\[]+)/i', $title, $warnMatch)) {
                    $warnText = strtolower(explode(':', $warnMatch[1])[0]);
                    $warnText = trim(preg_replace('/[^a-z]+/', '_', $warnText), '_');
                    if ($warnText !== '') {
                        $issueSubtype = 'warn_' . $warnText;
                    }
                }
            }
        }

        // Include source type so automated tickets never collide with manual ones
        $sourceType = stripos($title, '[auto]') !== false ? 'auto' : 'manual';

        // Build stable components
        $stableComponents = [
            'source_type'    => $sourceType,
            'issue_category' => $issueCategory,
            'issue_subtype'  => $issueSubtype,
            'environment_tags' => (function () use ($title) {
                // Extract each [bracketed] tag, then keep the known environment ones.
                // (explode('][') leaves brackets stuck to the first/last tag, so e.g.
                // "[production] ..." never matched and the env tag was dropped from the
                // dedup hash — letting prod and staging issues collide onto one ticket.)
                preg_match_all('/\[([^\]]+)\]/', $title, $m);
                return array_values(array_filter(
                    $m[1],
                    fn($tag) => in_array($tag, ['production', 'development', 'staging', 'single-node', 'cluster-wide'], true)
                ));
            })(),
        ];

        // Manual tickets should be unique by title (so different software installs don't collide)
        if ($sourceType === 'manual') {
            $stableComponents['title'] = $title;
        }

        // Include hostname for node-specific issues
        if (!$isClusterWide) {
            $stableComponents['hostname'] = $hostname;
        }

        // Include drive identifier for drive-specific tickets.
        // Use serial when available (stable across reboots/reshuffles); fall back to
        // device path for tickets created before serial was added to the payload.
        if ($isDriveTicket) {
            $stableComponents['drive'] = $serial ?? ($deviceMatches[0] ?? '');
        }

        sort($stableComponents['environment_tags']);

        return $stableComponents;
    }

    /** SHA-256 dedup hash of the payload's stable components. */
    public static function generate(array $data): string
    {
        return hash('sha256', json_encode(self::components($data), JSON_UNESCAPED_SLASHES));
    }
}