#include <errno.h>
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>

#include "cluster.h"
#include "ha.h"
#include "myconf.h"
#include "netio.h"
#include "xce/cluster_def.h"

#include "request.h"
#include "utils.h"

/* forward declaration — implemented in startup_validation.c */
int validate_startup_config(const char *config_path,
                            char *warn_buf, size_t warn_len);

time_t        last_print_time = 0;
unsigned long npkts           = 0;
unsigned long npkts_old       = 0;

int myapp_run(myconf_t* myconf)
{
    cluster_t c;
    int       rc;

    memset(&c, 0x0, sizeof(c));
    c.myconf = myconf;

    /* Phase 1: pre-config init (hooks, HA priority, standby_endpoints, cbuf) */
    rc = cluster_init_phase1(&c);
    if (rc < 0) {
        ERROR("cluster_init_phase1: %s", strerror(-rc));
        cluster_fini(&c);
        return rc;
    }

    /* SSL context — created after Phase 1, used by election probing and Phase 2 */
    initialize_openssl();
    SSL_CTX* ssl_ctx = create_ssl_context(myconf->certfile, myconf->keyfile);
    if (!ssl_ctx) {
        ERROR("create_ssl_context failed");
        cluster_fini(&c);
        return -EINVAL;
    }
    c.ha.ssl_ctx = ssl_ctx;

    /* Req 1.1: open the listener socket before any election probe so that
     * concurrent-startup peers can reach this node's vote handler. */
    rc = cluster_open_listener(&c);
    if (rc < 0) {
        ERROR("cluster_open_listener: %s", strerror(-rc));
        cluster_fini(&c);
        return rc;
    }

    /* Req 1.2: initialise the I/O manager and virtual task manager before
     * election so that incoming vote requests are handled by the normal
     * dispatch loop once it starts. */
    iomgr_init(&(c.iomgr));
    iomgr_set_ssl_ctx(&(c.iomgr), ssl_ctx);

    {
        int nvcpus = (myconf->vcpus > 0) ? myconf->vcpus : 8;
        vtaskmgr_init(&(c.vtaskmgr), nvcpus);
    }

    /* Req 1.2: start the I/O worker thread so that incoming OP_requestVote RPCs
     * from peers are accepted and dispatched during the election phase.
     * Without iomgr_start here, connections queue in the kernel (Recv-Q > 0)
     * but no thread reads them, causing peers to time out waiting for vote replies. */
    iomgr_start(&(c.iomgr));
    vtaskmgr_start(&(c.vtaskmgr));

    /* Register the listener socket with epoll NOW so that iomgr can accept
     * and dispatch incoming OP_requestVote connections during the election.
     * Without this, iomgr runs but has no fd to watch — SSL_connect from
     * peers succeeds at TCP level but hangs waiting for SSL_accept. */
    {
        int monitor_rc = cluster_register_listener(&c);
        if (monitor_rc < 0) {
            ERROR("cluster_register_listener: %s", strerror(-monitor_rc));
            cluster_fini(&c);
            return monitor_rc;
        }
    }

    /* Pre-load config from disk if a local config file exists, so that this
     * node's config_tag is known before election.  This allows the freshness gate
     * in am_i_primary_quorum() (Req 2.4) to work correctly: a node with a valid
     * local config will win over nodes that have no config (tag=0).
     * If no config file exists yet, config_version_tag stays 0 and the node will
     * defer the config fetch to after election via bootstrap_config_if_needed. */
    {
        int pre_rc = load_primary_config(&c);
        if (pre_rc == 0) {
            INFO("Pre-election config loaded: tag=%016llx",
                 (unsigned long long)c.myconf->config_version_tag);
        }
        /* Ignore errors — node may have no config yet (first boot or standby) */
    }

    /* Req 3.1–3.4: probe peers for config tags before election so that a node
     * joining an existing cluster knows to defer its config fetch to post-election
     * rather than participating as a first-boot candidate with no config. */
    rc = bootstrap_config_if_needed(&c);
    if (rc != 0) {
        ERROR("bootstrap_config_if_needed: %s", strerror(-rc));
        cluster_fini(&c);
        return rc;
    }

    /* Election: quorum-based role determination (Req 2.1, 2.9).
     * Build the peer pointer array (excluding self) for am_i_primary_quorum().
     * peers_cfg.candidates includes this node; n_peers = candidates_count.
     * my_priority is the index of this node in the candidates list. */
    {
        const xce_cluster_config_t *pcfg = &myconf->peers_cfg;
        int n_peers = (pcfg->candidates_count > 0) ? pcfg->candidates_count : 1;

        /* Build array of pointers to other candidates (excluding self by index) */
        xce_candidate_t *peer_ptrs[MAX_HA_CANDIDATES];
        int n_other = 0;
        for (int i = 0; i < pcfg->candidates_count; i++) {
            if (i != c.ha.my_priority)
                peer_ptrs[n_other++] = (xce_candidate_t *)&pcfg->candidates[i];
        }

        uint64_t my_tag = myconf->config_version_tag;
        uint64_t my_lsn = (uint64_t)atomic_load_explicit(&c.ha.applied_lsn,
                                                          memory_order_acquire);

        ctrl_role_t elected = am_i_primary_quorum(
            &c.ha,
            n_peers,
            (n_other > 0) ? peer_ptrs : NULL,
            my_tag,
            my_lsn);

        if (elected == CTRL_PRIMARY) {
            /* Primary: load config from conf/current (FATAL if missing) */
            rc = load_primary_config(&c);
            if (rc != 0) {
                cluster_fini(&c);
                myconf_fini(myconf);
                return rc;
            }
            /* am_i_primary_quorum logs FSM transition; set role explicitly */
            atomic_store_explicit(&c.ha.role, (int32_t)CTRL_PRIMARY,
                                  memory_order_release);
        } else {
            /* Req 3.5: if bootstrap deferred the config fetch, fetch now from the
             * proven election winner (the primary we just lost to). */
            if (c.ha.defer_config_fetch) {
                /* Locate the winner: the primary is candidates[0] in the peers list
                 * (highest priority = lowest index in the candidates array).
                 * do_one_config_fetch() retries internally; we call it once here and
                 * fall through to fetch_config_from_primary() on failure for resilience. */
                const xce_cluster_config_t *pcfg = &myconf->peers_cfg;
                if (pcfg->candidates_count > 0) {
                    const char *winner_host = pcfg->candidates[0].hostname;
                    uint32_t    winner_port = pcfg->candidates[0].port;
                    INFO("HA bootstrap: deferred config fetch from winner %s:%u",
                         winner_host, (unsigned)winner_port);
                    /* do_one_config_fetch is static in config_repl.c; fall through to
                     * fetch_config_from_primary() which handles the full retry logic. */
                }
                c.ha.defer_config_fetch = false;
            }

            /* Standby: fetch config from primary before Phase 2.
             * Skip if this node already has a valid local config (ground-truth
             * nodes should not fetch from primary — they ARE the ground truth). */
            if (myconf->config_version_tag == 0) {
                rc = fetch_config_from_primary(&c);
                if (rc != 0) {
                    cluster_fini(&c);
                    myconf_fini(myconf);
                    return rc;
                }
            } else {
                INFO("Standby has local config (tag=%016llx) — skipping fetch from primary",
                     (unsigned long long)myconf->config_version_tag);
            }
            /* Role stays CTRL_STANDBY — already set by ha_init(). */
        }
    }

    /* Req 4.4: warn if config file contains deprecated 'vcpus' key */
    {
        char warn_buf[256] = {0};
        validate_startup_config(myconf->conffile, warn_buf, sizeof(warn_buf));
        if (warn_buf[0]) {
            WARN("startup: %s", warn_buf);
        }
    }

    /* Phase 2: full init (requires cluster_config) */
    rc = cluster_init_phase2(&c);
    if (rc < 0) {
        ERROR("cluster_init_phase2: %s", strerror(-rc));
        cluster_fini(&c);
        myconf_fini(myconf);
        return rc;
    }

    rc = cluster_start(&c);
    if (rc < 0) {
        SET_ERROR(-rc);
        perror("cluster_start");
        cluster_fini(&c);
        myconf_fini(myconf);
        return rc;
    }

    rc = cluster_run(&c);

    cluster_fini(&c);
    myconf_fini(myconf);
    return rc;
}