2 Bacula® - The Network Backup Solution
4 Copyright (C) 2009-2009 Free Software Foundation Europe e.V.
6 The main author of Bacula is Kern Sibbald, with contributions from
7 many others, a complete list can be found in the file AUTHORS.
8 This program is Free Software; you can redistribute it and/or
9 modify it under the terms of version two of the GNU General Public
10 License as published by the Free Software Foundation, which is
11 listed in the file LICENSE.
13 This program is distributed in the hope that it will be useful, but
14 WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 Bacula® is a registered trademark of Kern Sibbald.
24 The licensor of Bacula is the Free Software Foundation Europe
25 (FSFE), Fiduciary Program, Sumatrastrasse 25, 8006 Zürich,
26 Switzerland, email:ftf@fsfeurope.org.
29 #define __SQL_C /* indicate that this is sql.c */
32 #include "cats/cats.h"
33 #include "lib/htable.h"
37 #define dbglevel_sql 15
39 static int result_handler(void *ctx, int fields, char **row)
42 Pmsg4(0, "%s\t%s\t%s\t%s\n",
43 row[0], row[1], row[2], row[3]);
44 } else if (fields == 5) {
45 Pmsg5(0, "%s\t%s\t%s\t%s\t%s\n",
46 row[0], row[1], row[2], row[3], row[4]);
47 } else if (fields == 6) {
48 Pmsg6(0, "%s\t%s\t%s\t%s\t%s\t%s\n",
49 row[0], row[1], row[2], row[3], row[4], row[5]);
50 } else if (fields == 7) {
51 Pmsg7(0, "%s\t%s\t%s\t%s\t%s\t%s\t%s\n",
52 row[0], row[1], row[2], row[3], row[4], row[5], row[6]);
57 Bvfs::Bvfs(JCR *j, B_DB *mdb) {
60 db = mdb; /* need to inc ref count */
61 jobids = get_pool_memory(PM_NAME);
62 prev_dir = get_pool_memory(PM_NAME);
63 pattern = get_pool_memory(PM_NAME);
64 *jobids = *prev_dir = *pattern = 0;
65 dir_filenameid = pwd_id = offset = 0;
66 see_copies = see_all_version = false;
69 list_entries = result_handler;
74 free_pool_memory(jobids);
75 free_pool_memory(pattern);
76 free_pool_memory(prev_dir);
82 * TODO: Find a way to let the user choose how he wants to display
83 * files and directories
88 * Working Object to store PathId already seen (avoid
89 * database queries), equivalent to %cache_ppathid in perl
101 htable *cache_ppathid;
106 cache_ppathid = (htable *)malloc(sizeof(htable));
107 cache_ppathid->init(&link, &link, NITEMS);
109 nodes = (hlink *) malloc(max_node * sizeof (hlink));
111 table_node = New(alist(5, owned_by_alist));
112 table_node->append(nodes);
116 if (++nb_node >= max_node) {
118 nodes = (hlink *)malloc(max_node * sizeof(hlink));
119 table_node->append(nodes);
121 return nodes + nb_node;
124 bool lookup(char *pathid) {
125 bool ret = cache_ppathid->lookup(pathid) != NULL;
129 void insert(char *pathid) {
130 hlink *h = get_hlink();
131 cache_ppathid->insert(pathid, h);
135 cache_ppathid->destroy();
140 pathid_cache(const pathid_cache &); /* prohibit pass by value */
141 pathid_cache &operator= (const pathid_cache &);/* prohibit class assignment*/
144 /* Return the parent_dir with the trailing / (update the given string)
145 * TODO: see in the rest of bacula if we don't have already this function
151 char *bvfs_parent_dir(char *path)
154 int len = strlen(path) - 1;
156 if (len >= 0 && path[len] == '/') { /* if directory, skip last / */
162 while (p > path && !IsPathSeparator(*p)) {
170 /* Return the basename of the with the trailing /
171 * TODO: see in the rest of bacula if we don't have
172 * this function already
174 char *bvfs_basename_dir(char *path)
177 int len = strlen(path) - 1;
179 if (path[len] == '/') { /* if directory, skip last / */
185 while (p > path && !IsPathSeparator(*p)) {
189 p++; /* skip first / */
195 static void build_path_hierarchy(JCR *jcr, B_DB *mdb,
196 pathid_cache &ppathid_cache,
197 char *org_pathid, char *path)
199 Dmsg1(dbglevel, "build_path_hierarchy(%s)\n", path);
202 char *bkp = mdb->path;
203 strncpy(pathid, org_pathid, sizeof(pathid));
205 /* Does the ppathid exist for this ? we use a memory cache... In order to
206 * avoid the full loop, we consider that if a dir is allready in the
207 * PathHierarchy table, then there is no need to calculate all the
210 while (path && *path)
212 if (!ppathid_cache.lookup(pathid))
215 "SELECT PPathId FROM PathHierarchy WHERE PathId = %s",
218 QUERY_DB(jcr, mdb, mdb->cmd);
219 /* Do we have a result ? */
220 if (sql_num_rows(mdb) > 0) {
221 ppathid_cache.insert(pathid);
222 /* This dir was in the db ...
223 * It means we can leave, the tree has allready been built for
228 /* search or create parent PathId in Path table */
229 mdb->path = bvfs_parent_dir(path);
230 mdb->pnl = strlen(mdb->path);
231 if (!db_create_path_record(jcr, mdb, &parent)) {
234 ppathid_cache.insert(pathid);
237 "INSERT INTO PathHierarchy (PathId, PPathId) "
239 pathid, (uint64_t) parent.PathId);
241 INSERT_DB(jcr, mdb, mdb->cmd);
243 edit_uint64(parent.PathId, pathid);
244 path = mdb->path; /* already done */
247 /* It's already in the cache. We can leave, no time to waste here,
248 * all the parent dirs have allready been done
260 * Internal function to update path_hierarchy cache with a shared pathid cache
262 static void update_path_hierarchy_cache(JCR *jcr,
264 pathid_cache &ppathid_cache,
267 Dmsg0(dbglevel, "update_path_hierarchy_cache()\n");
271 edit_uint64(JobId, jobid);
274 db_start_transaction(jcr, mdb);
276 Mmsg(mdb->cmd, "SELECT 1 FROM Job WHERE JobId = %s AND HasCache=1", jobid);
278 if (!QUERY_DB(jcr, mdb, mdb->cmd) || sql_num_rows(mdb) > 0) {
279 Dmsg1(dbglevel, "already computed %d\n", (uint32_t)JobId );
283 /* Inserting path records for JobId */
284 Mmsg(mdb->cmd, "INSERT INTO PathVisibility (PathId, JobId) "
285 "SELECT DISTINCT PathId, JobId FROM File WHERE JobId = %s",
287 QUERY_DB(jcr, mdb, mdb->cmd);
290 /* Now we have to do the directory recursion stuff to determine missing
291 * visibility We try to avoid recursion, to be as fast as possible We also
292 * only work on not allready hierarchised directories...
295 "SELECT PathVisibility.PathId, Path "
296 "FROM PathVisibility "
297 "JOIN Path ON( PathVisibility.PathId = Path.PathId) "
298 "LEFT JOIN PathHierarchy "
299 "ON (PathVisibility.PathId = PathHierarchy.PathId) "
300 "WHERE PathVisibility.JobId = %s "
301 "AND PathHierarchy.PathId IS NULL "
302 "ORDER BY Path", jobid);
303 Dmsg1(dbglevel_sql, "q=%s\n", mdb->cmd);
304 QUERY_DB(jcr, mdb, mdb->cmd);
306 /* TODO: I need to reuse the DB connection without emptying the result
307 * So, now i'm copying the result in memory to be able to query the
308 * catalog descriptor again.
310 num = sql_num_rows(mdb);
312 char **result = (char **)malloc (num * 2 * sizeof(char *));
316 while((row = sql_fetch_row(mdb))) {
317 result[i++] = bstrdup(row[0]);
318 result[i++] = bstrdup(row[1]);
323 build_path_hierarchy(jcr, mdb, ppathid_cache, result[i], result[i+1]);
332 "INSERT INTO PathVisibility (PathId, JobId) "
333 "SELECT a.PathId,%s "
335 "SELECT DISTINCT h.PPathId AS PathId "
336 "FROM PathHierarchy AS h "
337 "JOIN PathVisibility AS p ON (h.PathId=p.PathId) "
338 "WHERE p.JobId=%s) AS a LEFT JOIN "
340 "FROM PathVisibility "
341 "WHERE JobId=%s) AS b ON (a.PathId = b.PathId) "
342 "WHERE b.PathId IS NULL", jobid, jobid, jobid);
345 QUERY_DB(jcr, mdb, mdb->cmd);
346 } while (sql_affected_rows(mdb) > 0);
348 Mmsg(mdb->cmd, "UPDATE Job SET HasCache=1 WHERE JobId=%s", jobid);
349 UPDATE_DB(jcr, mdb, mdb->cmd);
352 db_end_transaction(jcr, mdb);
357 * Find an store the filename descriptor for empty directories Filename.Name=''
359 DBId_t Bvfs::get_dir_filenameid()
362 if (dir_filenameid) {
363 return dir_filenameid;
366 Mmsg(q, "SELECT FilenameId FROM Filename WHERE Name = ''");
367 db_sql_query(db, q.c_str(), db_int_handler, &id);
369 return dir_filenameid;
372 void bvfs_update_cache(JCR *jcr, B_DB *mdb)
375 db_list_ctx jobids_list;
378 db_start_transaction(jcr, mdb);
381 /* TODO: Remove this code when updating make_bacula_table script */
382 Mmsg(mdb->cmd, "SELECT 1 FROM Job WHERE HasCache<>2 LIMIT 1");
383 if (!QUERY_DB(jcr, mdb, mdb->cmd)) {
384 Dmsg0(dbglevel, "Creating cache table\n");
385 Mmsg(mdb->cmd, "ALTER TABLE Job ADD HasCache int DEFAULT 0");
386 QUERY_DB(jcr, mdb, mdb->cmd);
389 "CREATE TABLE PathHierarchy ( "
390 "PathId integer NOT NULL, "
391 "PPathId integer NOT NULL, "
392 "CONSTRAINT pathhierarchy_pkey "
393 "PRIMARY KEY (PathId))");
394 QUERY_DB(jcr, mdb, mdb->cmd);
397 "CREATE INDEX pathhierarchy_ppathid "
398 "ON PathHierarchy (PPathId)");
399 QUERY_DB(jcr, mdb, mdb->cmd);
402 "CREATE TABLE PathVisibility ("
403 "PathId integer NOT NULL, "
404 "JobId integer NOT NULL, "
405 "Size int8 DEFAULT 0, "
406 "Files int4 DEFAULT 0, "
407 "CONSTRAINT pathvisibility_pkey "
408 "PRIMARY KEY (JobId, PathId))");
409 QUERY_DB(jcr, mdb, mdb->cmd);
412 "CREATE INDEX pathvisibility_jobid "
413 "ON PathVisibility (JobId)");
414 QUERY_DB(jcr, mdb, mdb->cmd);
420 "SELECT JobId from Job "
421 "WHERE HasCache = 0 "
422 "AND Type IN ('B') AND JobStatus IN ('T', 'f', 'A') "
425 db_sql_query(mdb, mdb->cmd, db_list_handler, &jobids_list);
427 bvfs_update_path_hierarchy_cache(jcr, mdb, jobids_list.list);
429 db_end_transaction(jcr, mdb);
430 db_start_transaction(jcr, mdb);
431 Dmsg0(dbglevel, "Cleaning pathvisibility\n");
433 "DELETE FROM PathVisibility "
435 "(SELECT 1 FROM Job WHERE JobId=PathVisibility.JobId)");
436 nb = DELETE_DB(jcr, mdb, mdb->cmd);
437 Dmsg1(dbglevel, "Affected row(s) = %d\n", nb);
439 db_end_transaction(jcr, mdb);
444 * Update the bvfs cache for given jobids (1,2,3,4)
447 bvfs_update_path_hierarchy_cache(JCR *jcr, B_DB *mdb, char *jobids)
449 pathid_cache ppathid_cache;
454 int stat = get_next_jobid_from_list(&p, &JobId);
461 Dmsg1(dbglevel, "Updating cache for %lld\n", (uint64_t)JobId);
462 update_path_hierarchy_cache(jcr, mdb, ppathid_cache, JobId);
467 * Update the bvfs cache for current jobids
469 void Bvfs::update_cache()
471 bvfs_update_path_hierarchy_cache(jcr, db, jobids);
474 /* Change the current directory, returns true if the path exists */
475 bool Bvfs::ch_dir(const char *path)
477 pm_strcpy(db->path, path);
478 db->pnl = strlen(db->path);
479 ch_dir(db_get_path_record(jcr, db));
484 * Get all file versions for a specified client
486 void Bvfs::get_all_file_versions(DBId_t pathid, DBId_t fnid, const char *client)
488 Dmsg3(dbglevel, "get_all_file_versions(%lld, %lld, %s)\n", (uint64_t)pathid,
489 (uint64_t)fnid, client);
490 char ed1[50], ed2[50];
493 Mmsg(q, " AND Job.Type IN ('C', 'B') ");
495 Mmsg(q, " AND Job.Type = 'B' ");
500 Mmsg(query,// 1 2 3 4
501 "SELECT 'V', File.FileId, File.Md5, File.JobId, File.LStat, "
503 "Media.VolumeName, Media.InChanger "
504 "FROM File, Job, Client, JobMedia, Media "
505 "WHERE File.FilenameId = %s "
506 "AND File.PathId=%s "
507 "AND File.JobId = Job.JobId "
508 "AND Job.ClientId = Client.ClientId "
509 "AND Job.JobId = JobMedia.JobId "
510 "AND File.FileIndex >= JobMedia.FirstIndex "
511 "AND File.FileIndex <= JobMedia.LastIndex "
512 "AND JobMedia.MediaId = Media.MediaId "
513 "AND Client.Name = '%s' "
514 "%s ORDER BY FileId LIMIT %d OFFSET %d"
515 ,edit_uint64(fnid, ed1), edit_uint64(pathid, ed2), client, q.c_str(),
517 Dmsg1(dbglevel_sql, "q=%s\n", query.c_str());
518 db_sql_query(db, query.c_str(), list_entries, user_data);
521 DBId_t Bvfs::get_root()
524 return db_get_path_record(jcr, db);
527 static int path_handler(void *ctx, int fields, char **row)
529 Bvfs *fs = (Bvfs *) ctx;
530 return fs->_handle_path(ctx, fields, row);
533 int Bvfs::_handle_path(void *ctx, int fields, char **row)
535 if (bvfs_is_dir(row)) {
536 /* can have the same path 2 times */
537 if (strcmp(row[BVFS_Name], prev_dir)) {
538 pm_strcpy(prev_dir, row[BVFS_Name]);
539 return list_entries(user_data, fields, row);
546 * Retrieve . and .. information
548 void Bvfs::ls_special_dirs()
550 Dmsg1(dbglevel, "ls_special_dirs(%lld)\n", (uint64_t)pwd_id);
551 char ed1[50], ed2[50];
555 if (!dir_filenameid) {
556 get_dir_filenameid();
559 /* Will fetch directories */
564 "((SELECT PPathId AS PathId, '..' AS Path "
565 "FROM PathHierarchy "
566 "WHERE PathId = %s) "
568 "(SELECT %s AS PathId, '.' AS Path))",
569 edit_uint64(pwd_id, ed1), ed1);
572 Mmsg(query2,// 1 2 3 4 5 6
573 "SELECT 'D', tmp.PathId, 0, tmp.Path, JobId, LStat, FileId "
574 "FROM %s AS tmp LEFT JOIN ( " // get attributes if any
575 "SELECT File1.PathId AS PathId, File1.JobId AS JobId, "
576 "File1.LStat AS LStat, File1.FileId AS FileId FROM File AS File1 "
577 "WHERE File1.FilenameId = %s "
578 "AND File1.JobId IN (%s)) AS listfile1 "
579 "ON (tmp.PathId = listfile1.PathId) "
580 "ORDER BY tmp.Path, JobId DESC ",
581 query.c_str(), edit_uint64(dir_filenameid, ed2), jobids);
583 Dmsg1(dbglevel_sql, "q=%s\n", query2.c_str());
584 db_sql_query(db, query2.c_str(), path_handler, this);
587 /* Returns true if we have dirs to read */
590 Dmsg1(dbglevel, "ls_dirs(%lld)\n", (uint64_t)pwd_id);
591 char ed1[50], ed2[50];
598 Mmsg(filter, " AND Path2.Path %s '%s' ", SQL_MATCH, pattern);
601 if (!dir_filenameid) {
602 get_dir_filenameid();
605 /* the sql query displays same directory multiple time, take the first one */
608 /* Let's retrieve the list of the visible dirs in this dir ...
609 * First, I need the empty filenameid to locate efficiently
610 * the dirs in the file table
611 * my $dir_filenameid = $self->get_dir_filenameid();
613 /* Then we get all the dir entries from File ... */
617 "SELECT 'D', PathId, 0, Path, JobId, LStat, FileId FROM ( "
618 "SELECT Path1.PathId AS PathId, Path1.Path AS Path, "
619 "lower(Path1.Path) AS lpath, "
620 "listfile1.JobId AS JobId, listfile1.LStat AS LStat, "
621 "listfile1.FileId AS FileId "
623 "SELECT DISTINCT PathHierarchy1.PathId AS PathId "
624 "FROM PathHierarchy AS PathHierarchy1 "
625 "JOIN Path AS Path2 "
626 "ON (PathHierarchy1.PathId = Path2.PathId) "
627 "JOIN PathVisibility AS PathVisibility1 "
628 "ON (PathHierarchy1.PathId = PathVisibility1.PathId) "
629 "WHERE PathHierarchy1.PPathId = %s "
630 "AND PathVisibility1.jobid IN (%s) "
633 "JOIN Path AS Path1 ON (listpath1.PathId = Path1.PathId) "
635 "LEFT JOIN ( " /* get attributes if any */
636 "SELECT File1.PathId AS PathId, File1.JobId AS JobId, "
637 "File1.LStat AS LStat, File1.FileId AS FileId FROM File AS File1 "
638 "WHERE File1.FilenameId = %s "
639 "AND File1.JobId IN (%s)) AS listfile1 "
640 "ON (listpath1.PathId = listfile1.PathId) "
641 ") AS A ORDER BY 2,3 DESC LIMIT %d OFFSET %d",
642 edit_uint64(pwd_id, ed1),
645 edit_uint64(dir_filenameid, ed2),
649 Dmsg1(dbglevel_sql, "q=%s\n", query.c_str());
652 db_sql_query(db, query.c_str(), path_handler, this);
653 nb_record = db->num_rows;
656 return nb_record == limit;
659 /* Returns true if we have files to read */
660 bool Bvfs::ls_files()
662 Dmsg1(dbglevel, "ls_files(%lld)\n", (uint64_t)pwd_id);
674 Mmsg(filter, " AND Filename.Name %s '%s' ", SQL_MATCH, pattern);
676 /* TODO: Use JobTDate instead of FileId to determine the latest version */
678 Mmsg(query, // 1 2 3 4
679 "SELECT 'F', File.PathId, File.FilenameId, listfiles.Name, File.JobId, "
680 "File.LStat, listfiles.id "
682 "SELECT Filename.Name as Name, max(File.FileId) as id "
683 "FROM File, Filename "
684 "WHERE File.FilenameId = Filename.FilenameId "
685 "AND Filename.Name != '' "
686 "AND File.PathId = %s "
687 "AND File.JobId IN (%s) "
689 "GROUP BY Filename.Name "
690 "ORDER BY Filename.Name LIMIT %d OFFSET %d "
692 "WHERE File.FileId = listfiles.id",
693 edit_uint64(pwd_id, ed1),
698 Dmsg1(dbglevel_sql, "q=%s\n", query.c_str());
701 db_sql_query(db, query.c_str(), list_entries, user_data);
702 nb_record = db->num_rows;
705 return nb_record == limit;