GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/catalog_sql.h
Date: 2025-02-09 02:34:19
Exec Total Coverage
Lines: 19 21 90.5%
Branches: 11 18 61.1%
Line Branch Exec Source
1 /**
2 * This file is part of the CernVM file system.
3 *
4 * This file provides classes to wrap often used catalog SQL statements.
5 * In particular, it wraps around sqlite3 prepared statement syntax.
6 *
7 * Usage example:
8 * SqlLookup statement(<database>);
9 * statement.BindPathHash(<hash>);
10 * if (statement.FetchRow()) {
11 * statement.GetDirectoryEntry(<catalog>);
12 * }
13 * statement.Reset();
14 */
15
16 #ifndef CVMFS_CATALOG_SQL_H_
17 #define CVMFS_CATALOG_SQL_H_
18
19 #ifndef __STDC_FORMAT_MACROS
20 #define __STDC_FORMAT_MACROS
21 #endif
22
23 #include <inttypes.h>
24
25 #include <string>
26
27 #include "compression/compression.h"
28 #include "crypto/hash.h"
29 #include "directory_entry.h"
30 #include "file_chunk.h"
31 #include "shortstring.h"
32 #include "sql.h"
33
34 class XattrList;
35
36 namespace catalog {
37
38 class Catalog;
39
40
41 class CatalogDatabase : public sqlite::Database<CatalogDatabase> {
42 public:
43 static const float kLatestSchema;
44 static const float kLatestSupportedSchema; // + 1.X catalogs (r/o)
45 // Backwards-compatible schema changes
46 static const unsigned kLatestSchemaRevision;
47
48 bool CreateEmptyDatabase();
49 bool InsertInitialValues(const std::string &root_path,
50 const bool volatile_content,
51 const std::string &voms_authz,
52 const DirectoryEntry &root_entry
53 = DirectoryEntry(kDirentNegative));
54
55 bool CheckSchemaCompatibility();
56 bool LiveSchemaUpgradeIfNecessary();
57 bool CompactDatabase() const;
58
59 double GetRowIdWasteRatio() const;
60 bool SetVOMSAuthz(const std::string&);
61
62 protected:
63 // TODO(rmeusel): C++11 - constructor inheritance
64 friend class sqlite::Database<CatalogDatabase>;
65 241 CatalogDatabase(const std::string &filename,
66 const OpenMode open_mode)
67 241 : sqlite::Database<CatalogDatabase>(filename, open_mode) { }
68 };
69
70
71 //------------------------------------------------------------------------------
72
73
74 /**
75 * Base class for all SQL statement classes. It wraps a single SQL statement
76 * and all necessary calls of the sqlite3 API to deal with this statement.
77 */
78 class SqlCatalog : public sqlite::Sql {
79 public:
80 /**
81 * Basic constructor to use this class for a specific statement.
82 * @param database the database to use the query on
83 * @param statement the statement to prepare
84 */
85 1007 SqlCatalog(const CatalogDatabase &database, const std::string &statement) {
86
2/4
✓ Branch 1 taken 1007 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 1007 times.
✗ Branch 5 not taken.
 1007 Init(database.sqlite_db(), statement);
87 1007 }
88
89 /**
90 * Wrapper for retrieving MD5-ified path names.
91 * @param idx_high offset of most significant bits in database query
92 * @param idx_low offset of least significant bits in database query
93 * @result the retrieved MD5 hash
94 */
95 ✗ inline shash::Md5 RetrieveMd5(const int idx_high, const int idx_low) const {
96 ✗ return shash::Md5(RetrieveInt64(idx_high), RetrieveInt64(idx_low));
97 }
98
99 /**
100 * Wrapper for retrieving a cryptographic hash from a blob field.
101 */
102 733 inline shash::Any RetrieveHashBlob(
103 const int idx_column,
104 const shash::Algorithms hash_algo,
105 const char hash_suffix = shash::kSuffixNone) const
106 {
107 // Note: SQLite documentation advises to first define the data type of BLOB
108 // by calling sqlite3_column_XXX() on the column and _afterwards_ get
109 // the number of bytes using sqlite3_column_bytes().
110 //
111 // See: https://www.sqlite.org/c3ref/column_blob.html
112 const unsigned char *buffer = static_cast<const unsigned char *>(
113 733 RetrieveBlob(idx_column));
114 733 const int byte_count = RetrieveBytes(idx_column);
115 return (byte_count > 0) ? shash::Any(hash_algo, buffer, hash_suffix)
116
2/2
✓ Branch 0 taken 228 times.
✓ Branch 1 taken 505 times.
 733 : shash::Any(hash_algo);
117 }
118
119 /**
120 * Wrapper for retrieving a cryptographic hash from a text field.
121 */
122 inline shash::Any RetrieveHashHex(
123 const int idx_column,
124 const char hash_suffix = shash::kSuffixNone) const
125 {
126 const std::string hash_string = std::string(reinterpret_cast<const char *>(
127 RetrieveText(idx_column)));
128 return shash::MkFromHexPtr(shash::HexPtr(hash_string), hash_suffix);
129 }
130
131 /**
132 * Wrapper for binding a MD5-ified path name.
133 * @param idx_high offset of most significant bits in database query
134 * @param idx_low offset of least significant bits in database query
135 * @param hash the hash to bind in the query
136 * @result true on success, false otherwise
137 */
138 2022 inline bool BindMd5(const int idx_high, const int idx_low,
139 const shash::Md5 &hash)
140 {
141 uint64_t high, low;
142
1/2
✓ Branch 1 taken 2022 times.
✗ Branch 2 not taken.
 2022 hash.ToIntPair(&high, &low);
143
2/4
✓ Branch 1 taken 2022 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 2022 times.
✗ Branch 4 not taken.
 4044 const bool retval = BindInt64(idx_high, static_cast<int64_t>(high)) &&
144
2/4
✓ Branch 1 taken 2022 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 2022 times.
✗ Branch 4 not taken.
 2022 BindInt64(idx_low, static_cast<int64_t>(low));
145 2022 return retval;
146 }
147
148 /**
149 * Wrapper for binding a cryptographic hash. Algorithm of hash has to be
150 * elsewhere (in flags).
151 * @param idx_column offset of the blob field in database query
152 * @param hash the hash to bind in the query
153 * @result true on success, false otherwise
154 */
155 723 inline bool BindHashBlob(const int idx_column, const shash::Any &hash) {
156
2/2
✓ Branch 1 taken 446 times.
✓ Branch 2 taken 277 times.
 723 if (hash.IsNull()) {
157 446 return BindNull(idx_column);
158 } else {
159 277 return BindBlob(idx_column, hash.digest, hash.GetDigestSize());
160 }
161 }
162
163 protected:
164 2412 SqlCatalog() : sqlite::Sql() {}
165 };
166
167
168 //------------------------------------------------------------------------------
169
170
171 /**
172 * Common ancestor of SQL statements that deal with directory entries.
173 */
174 class SqlDirent : public SqlCatalog {
175 public:
176 // Definition of bit positions for the flags field of a DirectoryEntry
177 // All other bit positions are unused
178 static const int kFlagDir = 1;
179 // Link in the parent catalog
180 static const int kFlagDirNestedMountpoint = 2;
181 // Link in the child catalog
182 static const int kFlagDirNestedRoot = 32;
183 static const int kFlagFile = 4;
184 static const int kFlagLink = 8;
185 static const int kFlagFileSpecial = 16;
186 static const int kFlagFileChunk = 64;
187 /**
188 * The file is not natively stored in cvmfs but on a different storage system,
189 * for instance on HTTPS data federation services.
190 * NOTE: used as magic number in SqlListContentHashes::SqlListContentHashes
191 */
192 static const int kFlagFileExternal = 128;
193 // as of 2^8: 3 bit for hashes
194 // - 0: SHA-1
195 // - 1: RIPEMD-160
196 // - ...
197 // Corresponds to shash::algorithms with offset in order to support future
198 // hashes
199 static const int kFlagPosHash = 8;
200 // Compression methods, 3 bits starting at 2^11
201 // Corresponds to zlib::Algorithms
202 static const int kFlagPosCompression = 11;
203 /**
204 * A transition point to a root catalog (instead of a nested catalog). Used
205 * to link previous snapshots into the catalog structure.
206 */
207 static const int kFlagDirBindMountpoint = 0x4000; // 2^14
208 /**
209 * An entry that should not appear in listings. Used for the /.cvmfs
210 * directory.
211 */
212 static const int kFlagHidden = 0x8000; // 2^15
213 /**
214 * For regular files, indicates that the file should be opened with direct I/O
215 */
216 static const int kFlagDirectIo = 0x10000; // 2^16
217
218
219 protected:
220 /**
221 * Take the meta data from the DirectoryEntry and transform it
222 * into a valid flags field ready to be saved in the database.
223 * @param entry the DirectoryEntry to encode
224 * @return an integer containing the bitmap of the flags field
225 */
226 unsigned CreateDatabaseFlags(const DirectoryEntry &entry) const;
227 void StoreHashAlgorithm(const shash::Algorithms algo, unsigned *flags) const;
228 shash::Algorithms RetrieveHashAlgorithm(const unsigned flags) const;
229 zlib::Algorithms RetrieveCompressionAlgorithm(const unsigned flags) const;
230
231 /**
232 * The hardlink information (hardlink group ID and linkcount) is saved in one
233 * uint_64t field in the CVMFS Catalogs. Therefore we need to do bitshifting
234 * in these helper methods.
235 */
236 uint32_t Hardlinks2Linkcount(const uint64_t hardlinks) const;
237 uint32_t Hardlinks2HardlinkGroup(const uint64_t hardlinks) const;
238 uint64_t MakeHardlinks(const uint32_t hardlink_group,
239 const uint32_t linkcount) const;
240
241 /**
242 * Replaces place holder variables in a symbolic link by actual path elements.
243 * @param raw_symlink the raw symlink path (may) containing place holders
244 * @return the expanded symlink
245 */
246 void ExpandSymlink(LinkString *raw_symlink) const;
247 };
248
249
250 //------------------------------------------------------------------------------
251
252
253 class SqlDirentWrite : public SqlDirent {
254 public:
255 /**
256 * To bind an entire DirectoryEntry
257 * @param entry the DirectoryEntry to bind in the SQL statement
258 * @return true on success, false otherwise
259 */
260 virtual bool BindDirent(const DirectoryEntry &entry) = 0;
261
262 protected:
263 bool BindDirentFields(const int hash_idx,
264 const int hardlinks_idx,
265 const int size_idx,
266 const int mode_idx,
267 const int mtime_idx,
268 const int mtimens_idx,
269 const int flags_idx,
270 const int name_idx,
271 const int symlink_idx,
272 const int uid_idx,
273 const int gid_idx,
274 const DirectoryEntry &entry);
275 };
276
277
278 //------------------------------------------------------------------------------
279
280
281 class SqlListContentHashes : public SqlDirent {
282 public:
283 explicit SqlListContentHashes(const CatalogDatabase &database);
284 shash::Any GetHash() const;
285 };
286
287
288 //------------------------------------------------------------------------------
289
290
291 class SqlLookup : public SqlDirent {
292 public:
293 /**
294 * Retrieves a DirectoryEntry from a freshly performed SqlLookup statement.
295 * @param catalog the catalog in which the DirectoryEntry resides
296 * @return the retrieved DirectoryEntry
297 */
298 DirectoryEntry GetDirent(const Catalog *catalog,
299 const bool expand_symlink = true) const;
300
301 /**
302 * DirectoryEntrys do not contain their path hash.
303 * This method retrieves the saved path hash from the database
304 * @return the MD5 path hash of a freshly performed lookup
305 */
306 shash::Md5 GetPathHash() const;
307
308 /**
309 * DirectoryEntries do not contain their parent path hash.
310 * This method retrieves the saved parent path hash from the database
311 * @return the MD5 parent path hash of a freshly performed lookup
312 */
313 shash::Md5 GetParentPathHash() const;
314 };
315
316
317 //------------------------------------------------------------------------------
318
319
320 class SqlListing : public SqlLookup {
321 public:
322 explicit SqlListing(const CatalogDatabase &database);
323 bool BindPathHash(const struct shash::Md5 &hash);
324 };
325
326
327 //------------------------------------------------------------------------------
328
329
330 class SqlLookupPathHash : public SqlLookup {
331 public:
332 explicit SqlLookupPathHash(const CatalogDatabase &database);
333 bool BindPathHash(const struct shash::Md5 &hash);
334 };
335
336
337 //------------------------------------------------------------------------------
338
339
340 class SqlLookupInode : public SqlLookup {
341 public:
342 explicit SqlLookupInode(const CatalogDatabase &database);
343 bool BindRowId(const uint64_t inode);
344 };
345
346
347 //------------------------------------------------------------------------------
348
349
350 /**
351 * This SQL statement is only used for legacy catalog migrations and has been
352 * moved here as it needs to use a locally defined macro inside catalog_sql.cc
353 *
354 * Queries a single catalog and looks for DirectoryEntrys that have direct
355 * children in the same catalog but are marked as 'nested catalog mountpoints'.
356 * This is an inconsistent situation, since a mountpoint is supposed to be empty
357 * and it's children are stored in the corresponding referenced nested catalog.
358 *
359 * Note: the user code needs to check if there is a corresponding nested catalog
360 * reference for the found dangling mountpoints. If so, we also have a
361 * bogus state, but it is not reliably fixable automatically. The child-
362 * DirectoryEntrys would be masked by the mounting nested catalog but it
363 * is not clear if we can simply delete them or if this would destroy data.
364 */
365 class SqlLookupDanglingMountpoints : public catalog::SqlLookup {
366 public:
367 explicit SqlLookupDanglingMountpoints(const CatalogDatabase &database);
368 };
369
370
371 //------------------------------------------------------------------------------
372
373
374 /**
375 * Filesystem like _touch_ of a DirectoryEntry. Only file system specific meta
376 * data will be modified. All CVMFS-specific administrative data stays
377 * unchanged.
378 * NOTE: This is not a subclass of SqlDirent since it works on
379 * DirectoryEntryBase objects, which are restricted to file system meta
380 * data.
381 */
382 class SqlDirentTouch : public SqlCatalog {
383 public:
384 explicit SqlDirentTouch(const CatalogDatabase &database);
385
386 bool BindDirentBase(const DirectoryEntryBase &entry);
387 bool BindPathHash(const shash::Md5 &hash);
388 bool BindXattr(const XattrList &xattrs);
389 bool BindXattrEmpty();
390 };
391
392
393 //------------------------------------------------------------------------------
394
395
396 /**
397 * Nested catalogs and bind mountpoints.
398 */
399 class SqlNestedCatalogLookup : public SqlCatalog {
400 public:
401 explicit SqlNestedCatalogLookup(const CatalogDatabase &database);
402 bool BindSearchPath(const PathString &path);
403 shash::Any GetContentHash() const;
404 uint64_t GetSize() const;
405 };
406
407
408 //------------------------------------------------------------------------------
409
410
411 /**
412 * Nested catalogs and bind mountpoints.
413 */
414 class SqlNestedCatalogListing : public SqlCatalog {
415 public:
416 explicit SqlNestedCatalogListing(const CatalogDatabase &database);
417 PathString GetPath() const;
418 shash::Any GetContentHash() const;
419 uint64_t GetSize() const;
420 };
421
422
423 //------------------------------------------------------------------------------
424
425
426 /**
427 * Only nested catalogs, no bind mountpoints. Used for replication and GC.
428 */
429 class SqlOwnNestedCatalogListing : public SqlCatalog {
430 public:
431 explicit SqlOwnNestedCatalogListing(const CatalogDatabase &database);
432 PathString GetPath() const;
433 shash::Any GetContentHash() const;
434 uint64_t GetSize() const;
435 };
436
437
438 //------------------------------------------------------------------------------
439
440
441 class SqlDirentInsert : public SqlDirentWrite {
442 public:
443 explicit SqlDirentInsert(const CatalogDatabase &database);
444 bool BindPathHash(const shash::Md5 &hash);
445 bool BindParentPathHash(const shash::Md5 &hash);
446 bool BindDirent(const DirectoryEntry &entry);
447 bool BindXattr(const XattrList &xattrs);
448 bool BindXattrEmpty();
449 };
450
451
452 //------------------------------------------------------------------------------
453
454
455 class SqlDirentUpdate : public SqlDirentWrite {
456 public:
457 explicit SqlDirentUpdate(const CatalogDatabase &database);
458 bool BindPathHash(const shash::Md5 &hash);
459 bool BindDirent(const DirectoryEntry &entry);
460 };
461
462
463 //------------------------------------------------------------------------------
464
465
466 class SqlDirentUnlink : public SqlCatalog {
467 public:
468 explicit SqlDirentUnlink(const CatalogDatabase &database);
469 bool BindPathHash(const shash::Md5 &hash);
470 };
471
472
473 //------------------------------------------------------------------------------
474
475
476 /**
477 * Changes the linkcount for all files in a hardlink group.
478 */
479 class SqlIncLinkcount : public SqlCatalog {
480 public:
481 explicit SqlIncLinkcount(const CatalogDatabase &database);
482 bool BindPathHash(const shash::Md5 &hash);
483 bool BindDelta(const int delta);
484 };
485
486
487 //------------------------------------------------------------------------------
488
489
490 class SqlChunkInsert : public SqlCatalog {
491 public:
492 explicit SqlChunkInsert(const CatalogDatabase &database);
493 bool BindPathHash(const shash::Md5 &hash);
494 bool BindFileChunk(const FileChunk &chunk);
495 };
496
497
498 //------------------------------------------------------------------------------
499
500
501 class SqlChunksRemove : public SqlCatalog {
502 public:
503 explicit SqlChunksRemove(const CatalogDatabase &database);
504 bool BindPathHash(const shash::Md5 &hash);
505 };
506
507
508 //------------------------------------------------------------------------------
509
510
511 class SqlChunksListing : public SqlCatalog {
512 public:
513 explicit SqlChunksListing(const CatalogDatabase &database);
514 bool BindPathHash(const shash::Md5 &hash);
515 FileChunk GetFileChunk(const shash::Algorithms interpret_hash_as) const;
516 };
517
518
519 //------------------------------------------------------------------------------
520
521
522 class SqlChunksCount : public SqlCatalog {
523 public:
524 explicit SqlChunksCount(const CatalogDatabase &database);
525 bool BindPathHash(const shash::Md5 &hash);
526 int GetChunkCount() const;
527 };
528
529
530 //------------------------------------------------------------------------------
531
532
533 class SqlMaxHardlinkGroup : public SqlCatalog {
534 public:
535 explicit SqlMaxHardlinkGroup(const CatalogDatabase &database);
536 uint32_t GetMaxGroupId() const;
537 };
538
539
540 //------------------------------------------------------------------------------
541
542
543 class SqlGetCounter : public SqlCatalog {
544 public:
545 explicit SqlGetCounter(const CatalogDatabase &database);
546 bool BindCounter(const std::string &counter);
547 uint64_t GetCounter() const;
548 private:
549 bool compat_;
550 };
551
552
553 //------------------------------------------------------------------------------
554
555
556 class SqlUpdateCounter : public SqlCatalog {
557 public:
558 explicit SqlUpdateCounter(const CatalogDatabase &database);
559 bool BindCounter(const std::string &counter);
560 bool BindDelta(const int64_t delta);
561 };
562
563
564 //------------------------------------------------------------------------------
565
566
567 class SqlCreateCounter : public SqlCatalog {
568 public:
569 explicit SqlCreateCounter(const CatalogDatabase &database);
570 bool BindCounter(const std::string &counter);
571 bool BindInitialValue(const int64_t value);
572 };
573
574
575 //------------------------------------------------------------------------------
576
577
578 class SqlAllChunks : public SqlCatalog {
579 public:
580 explicit SqlAllChunks(const CatalogDatabase &database);
581 bool Open();
582 bool Next(shash::Any *hash, zlib::Algorithms *compression_alg);
583 bool Close();
584 };
585
586
587 //------------------------------------------------------------------------------
588
589
590 class SqlLookupXattrs : public SqlCatalog {
591 public:
592 explicit SqlLookupXattrs(const CatalogDatabase &database);
593 bool BindPathHash(const shash::Md5 &hash);
594 XattrList GetXattrs();
595 };
596
597 } // namespace catalog
598
599 #endif // CVMFS_CATALOG_SQL_H_
600