Export directory records

[citadel.git] / citadel / utils / ctdldump.c

// Don't run this.  It doesn't work and if you try to run it you will immediately die.
//
// Copyright (c) 2023 by Art Cancro citadel.org
//
// This program is open source software.  Use, duplication, or disclosure
// is subject to the terms of the GNU General Public License, version 3.

#include <stdlib.h>
#include <unistd.h>
#include <ctype.h>
#include <stdio.h>
#include <signal.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/un.h>
#include <netdb.h>
#include <string.h>
#include <pwd.h>
#include <errno.h>
#include <stdarg.h>
#include <limits.h>
#include <libcitadel.h>
#include <zlib.h>
#include <db.h>
#include "../server/sysdep.h"
#include "../server/citadel_defs.h"
#include "../server/server.h"
#include "../server/citadel_dirs.h"


// Wrapper for realloc() that crashes and burns if the call fails.
void *reallok(void *ptr, size_t size) {
        void *p = realloc(ptr, size);
        if (!p) {
                fprintf(stderr, "realloc() failed to resize %p to %ld bytes, error: %m\n", ptr, size);
                exit(1);
        }
        return p;
}
#define realloc reallok


// convert a binary blob to hex (non-reentrant!!!)
char *hexout(void *data, size_t len) {
        static char *outbuf = NULL;
        static size_t outlen = 0;
        int i;
        char ch;

        if ((outbuf == NULL) || (outlen < (len * 2))) {
                outbuf = reallok(outbuf, (len * 2));
                outlen = len * 2;
        }

        for (i=0; i<len; ++i) {
                ch = 0;
                memcpy(&ch, data+i, 1);
                sprintf((outbuf + (i * 2)), "%02X", (int) ch);
        }

        return(outbuf);
}


// Open a database environment
DB_ENV *open_dbenv(char *dirname) {

        DB_ENV *dbenv = NULL;

        int ret;
        int i;
        u_int32_t flags = 0;
        int dbversion_major, dbversion_minor, dbversion_patch;

        fprintf(stderr,
                "db: open_dbenv() starting\n"
                "db:    Linked zlib: %s\n"
                "db: Compiled libdb: %s\n"
                "db:   Linked libdb: %s\n",
                zlibVersion(),
                DB_VERSION_STRING,
                db_version(&dbversion_major, &dbversion_minor, &dbversion_patch)
        );

        // Create synthetic integer version numbers and compare them.
        // Never run with a libdb older than the one with which it was compiled.
        int compiled_db_version = ( (DB_VERSION_MAJOR * 1000000) + (DB_VERSION_MINOR * 1000) + (DB_VERSION_PATCH) );
        int linked_db_version = ( (dbversion_major * 1000000) + (dbversion_minor * 1000) + (dbversion_patch) );
        if (compiled_db_version > linked_db_version) {
                fprintf(stderr, "db: ctdldump is running with a version of libdb older than the one with which it was compiled.\n"
                        "db: This is an invalid configuration.  ctdldump will now exit to prevent data loss.");
                exit(CTDLEXIT_DB);
        }

        fprintf(stderr, "db: Setting up DB environment\n");
        ret = db_env_create(&dbenv, 0);
        if (ret) {
                fprintf(stderr, "db: db_env_create: %s\n", db_strerror(ret));
                fprintf(stderr, "db: exit code %d\n", ret);
                exit(CTDLEXIT_DB);
        }

        // We want to specify the shared memory buffer pool cachesize, but everything else is the default.
        ret = dbenv->set_cachesize(dbenv, 0, 64 * 1024, 0);
        if (ret) {
                fprintf(stderr, "db: set_cachesize: %s\n", db_strerror(ret));
                dbenv->close(dbenv, 0);
                fprintf(stderr, "db: exit code %d\n", ret);
                exit(CTDLEXIT_DB);
        }

        if ((ret = dbenv->set_lk_detect(dbenv, DB_LOCK_DEFAULT))) {
                fprintf(stderr, "db: set_lk_detect: %s\n", db_strerror(ret));
                dbenv->close(dbenv, 0);
                fprintf(stderr, "db: exit code %d\n", ret);
                exit(CTDLEXIT_DB);
        }

        flags = DB_CREATE | DB_INIT_MPOOL | DB_PRIVATE | DB_INIT_LOG;
        fprintf(stderr, "db: dbenv open(dir=%s, flags=%d)\n", dirname, flags);
        ret = dbenv->open(dbenv, dirname, flags, 0);
        if (ret) {
                fprintf(stderr, "db: dbenv->open: %s\n", db_strerror(ret));
                dbenv->close(dbenv, 0);
                fprintf(stderr, "db: exit code %d\n", ret);
                exit(CTDLEXIT_DB);
        }

        return(dbenv);
}


void close_dbenv(DB_ENV *dbenv) {
        int ret = dbenv->close(dbenv, 0);
        if (ret) {
                fprintf(stderr, "db: dbenv->close: %s\n", db_strerror(ret));
        }
}


// export function for a message in msgmain
void export_msgmain(int which_cdb, DBT *in_key, DBT *in_data) {
        long in_msgnum;

        if (in_key->size != sizeof(long)) {
                fprintf(stderr, "\033[31m\033[1m *** BAD DATA *** ABORTING *** \033[0m\n");
                abort();
        }

        memcpy(&in_msgnum, in_key->data, sizeof(in_msgnum));

        // If the msgnum is negative, we are looking at METADATA
        if (in_msgnum < 0) {
                struct MetaData *meta = (struct MetaData *)in_data->data;
                printf("msgmeta|%ld|%d|%s|%ld\n", meta->meta_msgnum, meta->meta_refcount, meta->meta_content_type, meta->meta_rfc822_length);
        }

        // If the msgnum is positive, we are looking at a MESSAGE
        else if (in_msgnum > 0) {
                printf("msgtext|%s\n", hexout(in_data->data, in_data->size));
        }

        // If the msgnum is 0 it's probably not a valid record.
}


// export function for a user record
void export_user(int which_cdb, DBT *in_key, DBT *in_data) {

        struct ctdluser *user = (struct ctdluser *)in_data->data;

        printf("user|%d|%d|%s|%u|%d|%ld|%ld|%d|%s|%ld|%ld|%s|%ld|%ld\n",
                user->version,                  // Citadel version which created this record
                user->uid,                      // Associate with a unix account?
                user->password,                 // password
                user->flags,                    // See US_ flags
                user->axlevel,                  // Access level
                user->usernum,                  // User number (never recycled)
                user->lastcall,                 // Date/time of most recent login
                user->USuserpurge,              // Purge time (in days) for user
                user->fullname,                 // Display name (primary identifier)
                user->msgnum_bio,               // msgnum of user's profile (bio)
                user->msgnum_pic,               // msgnum of user's avatar (photo)
                hexout(user->emailaddrs, strlen(user->emailaddrs)),     // Internet email addresses
                user->msgnum_inboxrules,        // msgnum of user's inbox filtering rules
                user->lastproc_inboxrules       // msgnum of last message filtered
        );
}


// export function for a room record
void export_room(int which_cdb, DBT *in_key, DBT *in_data) {

        struct ctdlroom *room = (struct ctdlroom *)in_data->data;

        printf("room|%s|%s|%ld|%ld|%ld|%u|%s|%ld|%d|%ld|%d|%d|%ld|%d|%u|%d|%ld\n",
                room->QRname,
                room->QRpasswd,
                room->QRroomaide,
                room->QRhighest,
                room->QRgen,
                room->QRflags,
                room->QRdirname,
                room->msgnum_info,
                room->QRfloor,
                room->QRmtime,
                room->QRep.expire_mode,
                room->QRep.expire_value,
                room->QRnumber,
                room->QRorder,
                room->QRflags2,
                room->QRdefaultview,
                room->msgnum_pic
        );
}


// export function for a floor record
void export_floor(int which_cdb, DBT *in_key, DBT *in_data) {

        int floor_num;
        memcpy(&floor_num, in_key->data, sizeof(int));

        struct floor *floor = (struct floor *)in_data->data;

        printf("floor|%d|%u|%s|%d|%d|%d\n",
                floor_num,
                floor->f_flags,
                floor->f_name,
                floor->f_ref_count,
                floor->f_ep.expire_mode,
                floor->f_ep.expire_value
        );
}


// export function for a msglist or a fulltext index record
// (both are indexed by a long and the data is arrays of longs)
void export_msglist(int which_cdb, DBT *in_key, DBT *in_data) {
        int i;
        int num_msgs;
        long msg;

        // records are indexed by a single "long" and contains an array of zero or more "long"s
        long roomnum;
        memcpy(&roomnum, in_key->data, sizeof(long));

        printf("msglist|%ld|", roomnum);

        if (in_data->size > 0) {
                num_msgs = in_data->size / sizeof(long);
                for (i=0; i<num_msgs; ++i) {
                        memcpy(&msg, (in_data->data + (i * sizeof(long))), sizeof(long));
                        if (i != 0) {
                                printf(",");
                        }
                        printf("%ld", msg);
                }
        }
        printf("\n");
}


// export function for a visit record
void export_visit(int which_cdb, DBT *in_key, DBT *in_data) {

        struct visit *visit = (struct visit *)in_data->data;

        printf("visit|%ld|%ld|%ld|%ld|%u|%s|%s|%d\n",
                visit->v_roomnum,
                visit->v_roomgen,
                visit->v_usernum,
                visit->v_lastseen,
                visit->v_flags,
                visit->v_seen,
                visit->v_answered,
                visit->v_view
        );
}


// export function for a directory record
// (This is a secondary index -- should we just regenerate the data after import?)
void export_dir(int which_cdb, DBT *in_key, DBT *in_data) {
        printf("dir|");
        fwrite(in_key->data, in_key->size, 1, stdout);
        printf("|");
        fwrite(in_data->data, in_data->size, 1, stdout);
        printf("\n");
}


#if 0
// export function for a use table record
void export_usetable(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {

        // the key is an int, which is the same size (32 bits) on both 32 and 64 bit systems
        out_key->size = in_key->size;
        out_key->data = realloc(out_key->data, out_key->size);
        memcpy(out_key->data, in_key->data, in_key->size);

        // the data is a "struct UseTable"
        struct UseTable_32 *use32 = (struct UseTable_32 *)in_data->data;
        out_data->size = sizeof(struct UseTable);
        out_data->data = realloc(out_data->data, out_data->size);
        memset(out_data->data, 0, out_data->size);
        struct UseTable *use64 = (struct UseTable *)out_data->data;

        //  the data
        use64->hash                     =               use32->hash;
        use64->timestamp                = (time_t)      use32->timestamp;

        // printf("\033[32m\033[1muse table: %d , %s\033[0m\n", use64->hash, asctime(localtime(&use64->timestamp)));
}


// export function for large message texts
void export_bigmsgs(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {

        // The key is a packed long
        int32_t in_msgnum;
        long out_msgnum;
        memcpy(&in_msgnum, in_key->data, sizeof(in_msgnum));
        out_msgnum = (long)in_msgnum;

        if (in_key->size != 4) {
                fprintf(stderr, "\033[31m\033[1m *** SOURCE DATABASE IS NOT 32-BIT *** ABORTING *** \033[0m\n");
                abort();
        }

        out_key->size = sizeof(long);
        out_key->data = realloc(out_key->data, out_key->size);
        memcpy(out_key->data, &out_msgnum, sizeof(long));

        // the data is binary-ish but has no packed integers
        out_data->size = in_data->size;
        out_data->data = realloc(out_data->data, out_data->size);
        memcpy(out_data->data, in_data->data, in_data->size);

        // printf("\033[32m\033[1mBigmsg %ld , length %d\033[0m\n", out_msgnum, out_data->size);
}


// export function for EUID Index records
void export_euidindex(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {

        // The structure of an euidindex record *key* is:
        // |----room_number----|----------EUID-------------|
        //    (sizeof long)       (actual length of euid)

        // The structure of an euidindex record *value* is:
        // |-----msg_number----|----room_number----|----------EUID-------------|
        //    (sizeof long)       (sizeof long)       (actual length of euid)

        int32_t in_msgnum = 0;
        int32_t in_roomnum = 0;
        char euid[SIZ];
        long out_msgnum = 0;
        long out_roomnum = 0;

        memcpy(&in_msgnum, in_data->data, sizeof(in_msgnum));
        memcpy(&in_roomnum, in_data->data+sizeof(int32_t), sizeof(in_msgnum));
        strcpy(euid, in_data->data+(sizeof(int32_t)*2));

        out_msgnum = (long) in_msgnum;
        out_roomnum = (long) in_roomnum;
        // printf("euidindex: msgnum=%ld, roomnum=%ld, euid=\"%s\"\n", out_msgnum, out_roomnum, euid);

        out_key->size = sizeof(long) + strlen(euid) + 1;
        out_key->data = realloc(out_key->data, out_key->size);
        memcpy(out_key->data, &out_roomnum, sizeof(out_roomnum));
        strcpy(out_key->data+sizeof(out_roomnum), euid);

        out_data->size = sizeof(long) + sizeof(long) + strlen(euid) + 1;
        out_data->data = realloc(out_data->data, out_data->size);
        memcpy(out_data->data, &out_msgnum, sizeof(out_msgnum));
        memcpy(out_data->data+sizeof(out_msgnum), &out_roomnum, sizeof(out_roomnum));
        strcpy(out_data->data+sizeof(out_msgnum)+sizeof(out_roomnum), euid);
}


// convert users-by-number records
void export_usersbynumber(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {

        // key is a long
        // and remember ... "long" is int32_t on the source system
        int32_t in_usernum;
        long out_usernum;
        memcpy(&in_usernum, in_key->data, sizeof(in_usernum));
        out_usernum = (long) in_usernum;

        if (in_key->size != 4) {
                fprintf(stderr, "\033[31m\033[1m *** SOURCE DATABASE IS NOT 32-BIT *** ABORTING *** \033[0m\n");
                abort();
        }

        out_key->size = sizeof(out_usernum);
        out_key->data = realloc(out_key->data, out_key->size);
        memcpy(out_key->data, &out_usernum, sizeof(out_usernum));

        // value is a string
        out_data->size = in_data->size;
        out_data->data = realloc(out_data->data, out_data->size);
        memcpy(out_data->data, in_data->data, in_data->size);

        // printf("usersbynumber: %ld --> %s\n", out_usernum, (char *)out_data->data);
}


// export function for a config record
void export_config(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {

        // the key is a string
        out_key->size = in_key->size;
        out_key->data = realloc(out_key->data, out_key->size + 1);
        memcpy(out_key->data, in_key->data, in_key->size);
        char *k = (char *)out_key->data;
        k[out_key->size] = 0;

        // the data is a pair of strings
        out_data->size = in_data->size;
        out_data->data = realloc(out_data->data, out_data->size + 1);
        memcpy(out_data->data, in_data->data, in_data->size);
        char *d = (char *)out_data->data;
        d[out_data->size] = 0;

        // please excuse my friend, he isn't null terminated
        // printf("\033[32m\033[1mConfig entry: %s -> %s\033[0m\n", (char *)out_key->data, (char *)out_data->data+strlen(out_data->data)+1);
}

#endif

// For obsolete databases, zero all the output
void zero_function(int which_cdb, DBT *in_key, DBT *in_data) {
        //printf("Table %02x, keylen=%d, datalen=%d\n", which_cdb, in_key->size, in_data->size);
}


void (*export_functions[])(int which_cdb, DBT *in_key, DBT *in_data) = {
        export_msgmain,         // CDB_MSGMAIN
        export_user,            // CDB_USERS
        export_room,            // CDB_ROOMS
        export_floor,           // CDB_FLOORTAB
        export_msglist,         // CDB_MSGLISTS
        export_visit,           // CDB_VISIT
        export_dir,             // CDB_DIRECTORY
        zero_function,          // CDB_USETABLE
        zero_function,          // CDB_BIGMSGS
        zero_function,          // CDB_FULLTEXT
        zero_function,          // CDB_EUIDINDEX
        zero_function,          // CDB_USERSBYNUMBER
        zero_function,          // CDB_UNUSED1 (obsolete)
        zero_function           // CDB_CONFIG
};


void export_table(int which_cdb, DB_ENV *src_dbenv) {
        int ret;
        int compressed;
        char dbfilename[32];
        uLongf destLen = 0;

        // shamelessly swiped from https://docs.oracle.com/database/bdb181/html/programmer_reference/am_cursor.html
        DB *src_dbp;
        DBC *src_dbcp;
        DBT in_key, in_data, uncomp_data;
        int num_good_rows = 0;
        int num_bad_rows = 0;

        snprintf(dbfilename, sizeof dbfilename, "cdb.%02x", which_cdb);

        // create a database handle for the source table
        ret = db_create(&src_dbp, src_dbenv, 0);
        if (ret) {
                fprintf(stderr, "db: db_create: %s\n", db_strerror(ret));
                fprintf(stderr, "db: exit code %d\n", ret);
                exit(CTDLEXIT_DB);
        }

        // open the file containing the source table
        ret = src_dbp->open(src_dbp, NULL, dbfilename, NULL, DB_BTREE, 0, 0600);
        if (ret) {
                fprintf(stderr, "db: db_open: %s\n", db_strerror(ret));
                fprintf(stderr, "db: exit code %d\n", ret);
                exit(CTDLEXIT_DB);
        }

        // Acquire a cursor to read the source table
        if ((ret = src_dbp->cursor(src_dbp, NULL, &src_dbcp, 0)) != 0) {
                fprintf(stderr, "db: db_cursor: %s\n", db_strerror(ret));
                fprintf(stderr, "db: exit code %d\n", ret);
                exit(CTDLEXIT_DB);
        }

        // Zero out these database keys
        memset(&in_key,         0, sizeof(DBT));        // input
        memset(&in_data,        0, sizeof(DBT));
        memset(&uncomp_data,    0, sizeof(DBT));        // decompressed input (the key doesn't change)

        // Walk through the database, calling export functions as we go and clearing buffers before each call.
        while (ret = src_dbcp->get(src_dbcp, &in_key, &in_data, DB_NEXT) == 0) {
        
                // If either the key or data are zero length, skip this record
                if ((in_key.size == 0) || (in_data.size == 0)) {
                        ++num_bad_rows;
                }

                else {  // Both key and data are >0 length so we're good to go

                        // Do we need to decompress?
                        static int32_t magic = COMPRESS_MAGIC;
                        compressed = 0;
                        if ( (in_data.size >= sizeof(struct CtdlCompressHeader)) && (!memcmp(in_data.data, &magic, sizeof(magic))) ) {
        
                                // yes, we need to decompress
                                compressed = 1;
                                struct CtdlCompressHeader comp;
                                memcpy(&comp, in_data.data, sizeof(struct CtdlCompressHeader));
                                uncomp_data.size = comp.uncompressed_len;
                                uncomp_data.data = realloc(uncomp_data.data, uncomp_data.size);
                                destLen = (uLongf)comp.uncompressed_len;
        
                                ret = uncompress((Bytef *)uncomp_data.data, (uLongf *)&destLen, (const Bytef *)in_data.data+sizeof(struct CtdlCompressHeader), (uLong)comp.compressed_len);
                                if (ret != Z_OK) {
                                        fprintf(stderr, "db: uncompress() error %d\n", ret);
                                        exit(CTDLEXIT_DB);
                                }
                        }
        
                        // Call the export function registered to this table
                        export_functions[which_cdb](which_cdb, &in_key, (compressed ? &uncomp_data : &in_data));
        
                        // Knowing the total number of rows isn't critical to the program.  It's just for the user to know.
                        fflush(stdout);
                }
        }

        // free any leftover out_data pointers
        free(uncomp_data.data);

        // ...and close the database (table)
        ret = src_dbp->close(src_dbp, 0);
        if (ret) {
                fprintf(stderr, "db: db_close: %s\n", db_strerror(ret));
        }


}


int main(int argc, char **argv) {
        int i = 0;
        char *src_dir = NULL;
        char *dst_dir = NULL;
        int confirmed = 0;
        static DB_ENV *src_dbenv;               // Source DB environment (global)

        // Parse command line
        int a;
        while ((a = getopt(argc, argv, "h:d:y")) != EOF) {
                switch (a) {
                case 'h':
                        src_dir = optarg;
                        break;
                case 'd':
                        dst_dir = optarg;
                        break;
                case 'y':
                        confirmed = 1;
                        break;
                default:
                        fprintf(stderr, "%s: usage: %s -s source_dir -d dest_dir\n", argv[0], argv[0]);
                        exit(2);
                }
        }

        // Warn the user
        fprintf(stderr, "------------------------------------------------------------------------\n");
        fprintf(stderr, "This utility must be run while the server is OFFLINE.                   \n");
        fprintf(stderr, "We \033[1mguarantee\033[0m data corruption if you do not                              \n");
        fprintf(stderr, "observe this warning!  The source [-s] directory should contain a copy  \n");
        fprintf(stderr, "of the database from your source system.  The dump [-d] directory       \n");
        fprintf(stderr, "should be empty and will receive your dump file.                        \n");
        fprintf(stderr, "------------------------------------------------------------------------\n");
        fprintf(stderr, " Source (database) directory: %s\n", src_dir);
        fprintf(stderr, "------------------------------------------------------------------------\n");

        if (confirmed == 1) {
                fprintf(stderr, "You have specified the [-y] flag, so processing will continue.\n");
        }
        else {
                fprintf(stderr, "Please read [ https://www.citadel.org/ctdldump.html ] to learn how to proceed.\n");
                exit(0);
        }

        src_dbenv = open_dbenv(src_dir);
        for (i = 0; i < MAXCDB; ++i) {
                export_table(i, src_dbenv);
        }
        close_dbenv(src_dbenv);

        exit(0);
}