1 // Don't run this. It doesn't work and if you try to run it you will immediately die.
3 // Copyright (c) 2023 by Art Cancro citadel.org
5 // This program is open source software. Use, duplication, or disclosure
6 // is subject to the terms of the GNU General Public License, version 3.
13 #include <sys/types.h>
14 #include <sys/socket.h>
22 #include <libcitadel.h>
25 #include "../server/sysdep.h"
26 #include "../server/citadel_defs.h"
27 #include "../server/server.h"
28 #include "../server/citadel_dirs.h"
31 // Wrapper for realloc() that crashes and burns if the call fails.
32 void *reallok(void *ptr, size_t size) {
33 void *p = realloc(ptr, size);
35 fprintf(stderr, "realloc() failed to resize %p to %ld bytes, error: %m\n", ptr, size);
40 #define realloc reallok
43 // convert a binary blob to hex (non-reentrant!!!)
44 char *hexout(void *data, size_t len) {
45 static char *outbuf = NULL;
46 static size_t outlen = 0;
50 if ((outbuf == NULL) || (outlen < (len * 2))) {
51 outbuf = reallok(outbuf, (len * 2));
55 for (i=0; i<len; ++i) {
57 memcpy(&ch, data+i, 1);
58 sprintf((outbuf + (i * 2)), "%02X", (int) ch);
65 // Open a database environment
66 DB_ENV *open_dbenv(char *dirname) {
73 int dbversion_major, dbversion_minor, dbversion_patch;
76 "db: open_dbenv() starting\n"
77 "db: Linked zlib: %s\n"
78 "db: Compiled libdb: %s\n"
79 "db: Linked libdb: %s\n",
82 db_version(&dbversion_major, &dbversion_minor, &dbversion_patch)
85 // Create synthetic integer version numbers and compare them.
86 // Never run with a libdb older than the one with which it was compiled.
87 int compiled_db_version = ( (DB_VERSION_MAJOR * 1000000) + (DB_VERSION_MINOR * 1000) + (DB_VERSION_PATCH) );
88 int linked_db_version = ( (dbversion_major * 1000000) + (dbversion_minor * 1000) + (dbversion_patch) );
89 if (compiled_db_version > linked_db_version) {
90 fprintf(stderr, "db: ctdldump is running with a version of libdb older than the one with which it was compiled.\n"
91 "db: This is an invalid configuration. ctdldump will now exit to prevent data loss.");
95 fprintf(stderr, "db: Setting up DB environment\n");
96 ret = db_env_create(&dbenv, 0);
98 fprintf(stderr, "db: db_env_create: %s\n", db_strerror(ret));
99 fprintf(stderr, "db: exit code %d\n", ret);
103 // We want to specify the shared memory buffer pool cachesize, but everything else is the default.
104 ret = dbenv->set_cachesize(dbenv, 0, 64 * 1024, 0);
106 fprintf(stderr, "db: set_cachesize: %s\n", db_strerror(ret));
107 dbenv->close(dbenv, 0);
108 fprintf(stderr, "db: exit code %d\n", ret);
112 if ((ret = dbenv->set_lk_detect(dbenv, DB_LOCK_DEFAULT))) {
113 fprintf(stderr, "db: set_lk_detect: %s\n", db_strerror(ret));
114 dbenv->close(dbenv, 0);
115 fprintf(stderr, "db: exit code %d\n", ret);
119 flags = DB_CREATE | DB_INIT_MPOOL | DB_PRIVATE | DB_INIT_LOG;
120 fprintf(stderr, "db: dbenv open(dir=%s, flags=%d)\n", dirname, flags);
121 ret = dbenv->open(dbenv, dirname, flags, 0);
123 fprintf(stderr, "db: dbenv->open: %s\n", db_strerror(ret));
124 dbenv->close(dbenv, 0);
125 fprintf(stderr, "db: exit code %d\n", ret);
133 void close_dbenv(DB_ENV *dbenv) {
134 int ret = dbenv->close(dbenv, 0);
136 fprintf(stderr, "db: dbenv->close: %s\n", db_strerror(ret));
141 // convert function for a message in msgmain
142 void export_msgmain(int which_cdb, DBT *in_key, DBT *in_data) {
145 if (in_key->size != sizeof(long)) {
146 fprintf(stderr, "\033[31m\033[1m *** BAD DATA *** ABORTING *** \033[0m\n");
150 memcpy(&in_msgnum, in_key->data, sizeof(in_msgnum));
152 // If the msgnum is negative, we are looking at METADATA
154 struct MetaData *meta = (struct MetaData *)in_data->data;
155 printf("msgmeta|%ld|%d|%s|%ld\n", meta->meta_msgnum, meta->meta_refcount, meta->meta_content_type, meta->meta_rfc822_length);
158 // If the msgnum is positive, we are looking at a MESSAGE
159 else if (in_msgnum > 0) {
160 printf("msgtext|%s\n", hexout(in_data->data, in_data->size));
163 // If the msgnum is 0 it's probably not a valid record.
167 // convert function for a user record
168 void export_user(int which_cdb, DBT *in_key, DBT *in_data) {
170 struct ctdluser *user = (struct ctdluser *)in_data->data;
172 printf("user|%d|%d|%s|%u|%d|%ld|%ld|%d|%s|%ld|%ld|%s|%ld|%ld\n",
173 user->version, // Citadel version which created this record
174 user->uid, // Associate with a unix account?
175 user->password, // password
176 user->flags, // See US_ flags
177 user->axlevel, // Access level
178 user->usernum, // User number (never recycled)
179 user->lastcall, // Date/time of most recent login
180 user->USuserpurge, // Purge time (in days) for user
181 user->fullname, // Display name (primary identifier)
182 user->msgnum_bio, // msgnum of user's profile (bio)
183 user->msgnum_pic, // msgnum of user's avatar (photo)
184 hexout(user->emailaddrs, strlen(user->emailaddrs)), // Internet email addresses
185 user->msgnum_inboxrules, // msgnum of user's inbox filtering rules
186 user->lastproc_inboxrules // msgnum of last message filtered
191 // convert function for a room record
192 void export_room(int which_cdb, DBT *in_key, DBT *in_data) {
194 struct ctdlroom *room = (struct ctdlroom *)in_data->data;
196 printf("room|%s|%s|%ld|%ld|%ld|%u|%s|%ld|%d|%ld|%d|%d|%ld|%d|%u|%d|%ld\n",
207 room->QRep.expire_mode,
208 room->QRep.expire_value,
219 // convert function for a floor record
220 void export_floors(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
222 // the key is an "int", and "int" is 32-bits on both 32 and 64 bit platforms.
223 out_key->size = in_key->size;
224 out_key->data = realloc(out_key->data, out_key->size);
225 memcpy(out_key->data, in_key->data, in_key->size);
228 struct floor_32 *floor32 = (struct floor_32 *)in_data->data;
229 out_data->size = sizeof(struct floor);
230 out_data->data = realloc(out_data->data, out_data->size);
231 struct floor *floor64 = (struct floor *)out_data->data;
233 // these are probably bit-for-bit identical, actually ... but we do it the "right" way anyway
234 floor64->f_flags = (unsigned short) floor32->f_flags;
235 strcpy(floor64->f_name, floor32->f_name);
236 floor64->f_ref_count = (int) floor32->f_ref_count;
237 floor64->f_ep.expire_mode = (int) floor32->f_ep.expire_mode;
238 floor64->f_ep.expire_value = (int) floor32->f_ep.expire_value;
240 // printf("\033[32m\033[1mFloor: %s\033[0m\n", floor64->f_name);
244 // convert function for a msglist or a fulltext index record
245 // (both are indexed by a long and the data is arrays of longs)
246 void export_msglists(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
249 char *table = (which_cdb == CDB_FULLTEXT) ? "FullText" : "Msglist";
251 // records are indexed by a single "long" and contains an array of zero or more "long"s
252 // and remember ... "long" is int32_t on the source system
255 memcpy(&in_roomnum, in_key->data, sizeof(in_roomnum));
256 out_roomnum = (long) in_roomnum;
258 if (in_key->size != 4) {
259 fprintf(stderr, "\033[31m\033[1m *** SOURCE DATABASE IS NOT 32-BIT *** ABORTING *** \033[0m\n");
263 int num_msgs = in_data->size / sizeof(int32_t);
264 // printf("\033[32m\033[1m%s: key %ld (%d messages)\033[0m\n", table, out_roomnum, num_msgs);
266 // the key is a "long"
267 out_key->size = sizeof(out_roomnum);
268 out_key->data = realloc(out_key->data, out_key->size);
269 memcpy(out_key->data, &out_roomnum, sizeof(out_roomnum));
271 // the data is another array, but a wider type
272 out_data->size = sizeof(long) * num_msgs;
273 out_data->data = realloc(out_data->data, out_data->size);
277 for (i=0; i<num_msgs; ++i) {
278 memcpy(&in_msg, (in_data->data + (i * sizeof(int32_t))), sizeof(int32_t));
279 out_msg = (long) in_msg;
280 memcpy((out_data->data + (i * sizeof(long))), &out_msg, sizeof(long));
281 // printf("msg#%ld\n", out_msg);
286 // convert function for a visit record
287 void export_visits(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
290 struct visit_32 *visit32 = (struct visit_32 *)in_data->data;
291 out_data->size = sizeof(struct visit);
292 out_data->data = realloc(out_data->data, out_data->size);
293 struct visit *visit64 = (struct visit *)out_data->data;
295 // the data (zero it out so it will compress well)
296 memset(visit64, 0, sizeof(struct visit));
297 visit64->v_roomnum = (long) visit32->v_roomnum;
298 visit64->v_roomgen = (long) visit32->v_roomgen;
299 visit64->v_usernum = (long) visit32->v_usernum;
300 visit64->v_lastseen = (long) visit32->v_lastseen;
301 visit64->v_flags = (unsigned) visit32->v_flags;
302 strcpy(visit64->v_seen, visit32->v_seen);
303 strcpy(visit64->v_answered, visit32->v_answered);
304 visit64->v_view = (int) visit32->v_view;
306 // printf("\033[32m\033[1mVisit: room %10ld, gen %10ld, user %10ld\033[0m\n", visit64->v_roomnum, visit64->v_roomgen, visit64->v_usernum);
308 // create the key (which is based on the data, so there is no need to convert the old key)
309 out_key->size = sizeof(struct visit_index);
310 out_key->data = realloc(out_key->data, out_key->size);
311 struct visit_index *newvisitindex = (struct visit_index *) out_key->data;
312 newvisitindex->iRoomID = visit64->v_roomnum;
313 newvisitindex->iRoomGen = visit64->v_roomgen;
314 newvisitindex->iUserID = visit64->v_usernum;
318 // convert function for a directory record
319 void export_dir(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
321 // the key is a string
322 out_key->size = in_key->size;
323 out_key->data = realloc(out_key->data, out_key->size + 1);
324 memcpy(out_key->data, in_key->data, in_key->size);
325 char *k = (char *)out_key->data;
326 k[out_key->size] = 0;
328 // the data is also a string
329 out_data->size = in_data->size;
330 out_data->data = realloc(out_data->data, out_data->size + 1);
331 memcpy(out_data->data, in_data->data, in_data->size);
332 char *d = (char *)out_data->data;
333 d[out_data->size] = 0;
335 // please excuse my friend, he isn't null terminated
336 // printf("\033[32m\033[1mDirectory entry: %s -> %s\033[0m\n", (char *)out_key->data, (char *)out_data->data);
340 // convert function for a use table record
341 void export_usetable(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
343 // the key is an int, which is the same size (32 bits) on both 32 and 64 bit systems
344 out_key->size = in_key->size;
345 out_key->data = realloc(out_key->data, out_key->size);
346 memcpy(out_key->data, in_key->data, in_key->size);
348 // the data is a "struct UseTable"
349 struct UseTable_32 *use32 = (struct UseTable_32 *)in_data->data;
350 out_data->size = sizeof(struct UseTable);
351 out_data->data = realloc(out_data->data, out_data->size);
352 memset(out_data->data, 0, out_data->size);
353 struct UseTable *use64 = (struct UseTable *)out_data->data;
356 use64->hash = use32->hash;
357 use64->timestamp = (time_t) use32->timestamp;
359 // printf("\033[32m\033[1muse table: %d , %s\033[0m\n", use64->hash, asctime(localtime(&use64->timestamp)));
363 // convert function for large message texts
364 void export_bigmsgs(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
366 // The key is a packed long
369 memcpy(&in_msgnum, in_key->data, sizeof(in_msgnum));
370 out_msgnum = (long)in_msgnum;
372 if (in_key->size != 4) {
373 fprintf(stderr, "\033[31m\033[1m *** SOURCE DATABASE IS NOT 32-BIT *** ABORTING *** \033[0m\n");
377 out_key->size = sizeof(long);
378 out_key->data = realloc(out_key->data, out_key->size);
379 memcpy(out_key->data, &out_msgnum, sizeof(long));
381 // the data is binary-ish but has no packed integers
382 out_data->size = in_data->size;
383 out_data->data = realloc(out_data->data, out_data->size);
384 memcpy(out_data->data, in_data->data, in_data->size);
386 // printf("\033[32m\033[1mBigmsg %ld , length %d\033[0m\n", out_msgnum, out_data->size);
390 // convert function for EUID Index records
391 void export_euidindex(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
393 // The structure of an euidindex record *key* is:
394 // |----room_number----|----------EUID-------------|
395 // (sizeof long) (actual length of euid)
397 // The structure of an euidindex record *value* is:
398 // |-----msg_number----|----room_number----|----------EUID-------------|
399 // (sizeof long) (sizeof long) (actual length of euid)
401 int32_t in_msgnum = 0;
402 int32_t in_roomnum = 0;
405 long out_roomnum = 0;
407 memcpy(&in_msgnum, in_data->data, sizeof(in_msgnum));
408 memcpy(&in_roomnum, in_data->data+sizeof(int32_t), sizeof(in_msgnum));
409 strcpy(euid, in_data->data+(sizeof(int32_t)*2));
411 out_msgnum = (long) in_msgnum;
412 out_roomnum = (long) in_roomnum;
413 // printf("euidindex: msgnum=%ld, roomnum=%ld, euid=\"%s\"\n", out_msgnum, out_roomnum, euid);
415 out_key->size = sizeof(long) + strlen(euid) + 1;
416 out_key->data = realloc(out_key->data, out_key->size);
417 memcpy(out_key->data, &out_roomnum, sizeof(out_roomnum));
418 strcpy(out_key->data+sizeof(out_roomnum), euid);
420 out_data->size = sizeof(long) + sizeof(long) + strlen(euid) + 1;
421 out_data->data = realloc(out_data->data, out_data->size);
422 memcpy(out_data->data, &out_msgnum, sizeof(out_msgnum));
423 memcpy(out_data->data+sizeof(out_msgnum), &out_roomnum, sizeof(out_roomnum));
424 strcpy(out_data->data+sizeof(out_msgnum)+sizeof(out_roomnum), euid);
428 // convert users-by-number records
429 void export_usersbynumber(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
432 // and remember ... "long" is int32_t on the source system
435 memcpy(&in_usernum, in_key->data, sizeof(in_usernum));
436 out_usernum = (long) in_usernum;
438 if (in_key->size != 4) {
439 fprintf(stderr, "\033[31m\033[1m *** SOURCE DATABASE IS NOT 32-BIT *** ABORTING *** \033[0m\n");
443 out_key->size = sizeof(out_usernum);
444 out_key->data = realloc(out_key->data, out_key->size);
445 memcpy(out_key->data, &out_usernum, sizeof(out_usernum));
448 out_data->size = in_data->size;
449 out_data->data = realloc(out_data->data, out_data->size);
450 memcpy(out_data->data, in_data->data, in_data->size);
452 // printf("usersbynumber: %ld --> %s\n", out_usernum, (char *)out_data->data);
456 // convert function for a config record
457 void export_config(int which_cdb, DBT *in_key, DBT *in_data, DBT *out_key, DBT *out_data) {
459 // the key is a string
460 out_key->size = in_key->size;
461 out_key->data = realloc(out_key->data, out_key->size + 1);
462 memcpy(out_key->data, in_key->data, in_key->size);
463 char *k = (char *)out_key->data;
464 k[out_key->size] = 0;
466 // the data is a pair of strings
467 out_data->size = in_data->size;
468 out_data->data = realloc(out_data->data, out_data->size + 1);
469 memcpy(out_data->data, in_data->data, in_data->size);
470 char *d = (char *)out_data->data;
471 d[out_data->size] = 0;
473 // please excuse my friend, he isn't null terminated
474 // printf("\033[32m\033[1mConfig entry: %s -> %s\033[0m\n", (char *)out_key->data, (char *)out_data->data+strlen(out_data->data)+1);
479 // For obsolete databases, zero all the output
480 void zero_function(int which_cdb, DBT *in_key, DBT *in_data) {
481 //printf("Table %02x, keylen=%d, datalen=%d\n", which_cdb, in_key->size, in_data->size);
485 void (*export_functions[])(int which_cdb, DBT *in_key, DBT *in_data) = {
486 export_msgmain, // CDB_MSGMAIN
487 export_user, // CDB_USERS
488 export_room, // CDB_ROOMS
489 zero_function, // CDB_FLOORTAB
490 zero_function, // CDB_MSGLISTS
491 zero_function, // CDB_VISIT
492 zero_function, // CDB_DIRECTORY
493 zero_function, // CDB_USETABLE
494 zero_function, // CDB_BIGMSGS
495 zero_function, // CDB_FULLTEXT
496 zero_function, // CDB_EUIDINDEX
497 zero_function, // CDB_USERSBYNUMBER
498 zero_function, // CDB_UNUSED1 (obsolete)
499 zero_function // CDB_CONFIG
503 void export_table(int which_cdb, DB_ENV *src_dbenv) {
509 // shamelessly swiped from https://docs.oracle.com/database/bdb181/html/programmer_reference/am_cursor.html
512 DBT in_key, in_data, uncomp_data;
513 int num_good_rows = 0;
514 int num_bad_rows = 0;
516 snprintf(dbfilename, sizeof dbfilename, "cdb.%02x", which_cdb);
518 // create a database handle for the source table
519 ret = db_create(&src_dbp, src_dbenv, 0);
521 fprintf(stderr, "db: db_create: %s\n", db_strerror(ret));
522 fprintf(stderr, "db: exit code %d\n", ret);
526 // open the file containing the source table
527 ret = src_dbp->open(src_dbp, NULL, dbfilename, NULL, DB_BTREE, 0, 0600);
529 fprintf(stderr, "db: db_open: %s\n", db_strerror(ret));
530 fprintf(stderr, "db: exit code %d\n", ret);
534 // Acquire a cursor to read the source table
535 if ((ret = src_dbp->cursor(src_dbp, NULL, &src_dbcp, 0)) != 0) {
536 fprintf(stderr, "db: db_cursor: %s\n", db_strerror(ret));
537 fprintf(stderr, "db: exit code %d\n", ret);
541 // Zero out these database keys
542 memset(&in_key, 0, sizeof(DBT)); // input
543 memset(&in_data, 0, sizeof(DBT));
544 memset(&uncomp_data, 0, sizeof(DBT)); // decompressed input (the key doesn't change)
546 // Walk through the database, calling convert functions as we go and clearing buffers before each call.
547 while (ret = src_dbcp->get(src_dbcp, &in_key, &in_data, DB_NEXT) == 0) {
549 // If either the key or data are zero length, skip this record
550 if ((in_key.size == 0) || (in_data.size == 0)) {
554 else { // Both key and data are >0 length so we're good to go
556 // Do we need to decompress?
557 static int32_t magic = COMPRESS_MAGIC;
559 if ( (in_data.size >= sizeof(struct CtdlCompressHeader)) && (!memcmp(in_data.data, &magic, sizeof(magic))) ) {
561 // yes, we need to decompress
563 struct CtdlCompressHeader comp;
564 memcpy(&comp, in_data.data, sizeof(struct CtdlCompressHeader));
565 uncomp_data.size = comp.uncompressed_len;
566 uncomp_data.data = realloc(uncomp_data.data, uncomp_data.size);
567 destLen = (uLongf)comp.uncompressed_len;
569 ret = uncompress((Bytef *)uncomp_data.data, (uLongf *)&destLen, (const Bytef *)in_data.data+sizeof(struct CtdlCompressHeader), (uLong)comp.compressed_len);
571 fprintf(stderr, "db: uncompress() error %d\n", ret);
576 // Call the convert function registered to this table
577 export_functions[which_cdb](which_cdb, &in_key, (compressed ? &uncomp_data : &in_data));
579 // Knowing the total number of rows isn't critical to the program. It's just for the user to know.
584 // free any leftover out_data pointers
585 free(uncomp_data.data);
587 // ...and close the database (table)
588 ret = src_dbp->close(src_dbp, 0);
590 fprintf(stderr, "db: db_close: %s\n", db_strerror(ret));
597 int main(int argc, char **argv) {
599 char *src_dir = NULL;
600 char *dst_dir = NULL;
602 static DB_ENV *src_dbenv; // Source DB environment (global)
604 // Parse command line
606 while ((a = getopt(argc, argv, "h:d:y")) != EOF) {
618 fprintf(stderr, "%s: usage: %s -s source_dir -d dest_dir\n", argv[0], argv[0]);
624 fprintf(stderr, "------------------------------------------------------------------------\n");
625 fprintf(stderr, "This utility must be run while the server is OFFLINE. \n");
626 fprintf(stderr, "We \033[1mguarantee\033[0m data corruption if you do not \n");
627 fprintf(stderr, "observe this warning! The source [-s] directory should contain a copy \n");
628 fprintf(stderr, "of the database from your source system. The dump [-d] directory \n");
629 fprintf(stderr, "should be empty and will receive your dump file. \n");
630 fprintf(stderr, "------------------------------------------------------------------------\n");
631 fprintf(stderr, " Source (database) directory: %s\n", src_dir);
632 fprintf(stderr, "------------------------------------------------------------------------\n");
634 if (confirmed == 1) {
635 fprintf(stderr, "You have specified the [-y] flag, so processing will continue.\n");
638 fprintf(stderr, "Please read [ https://www.citadel.org/ctdldump.html ] to learn how to proceed.\n");
642 src_dbenv = open_dbenv(src_dir);
643 for (i = 0; i < MAXCDB; ++i) {
644 export_table(i, src_dbenv);
646 close_dbenv(src_dbenv);