• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // Copyright 2006 Google Inc.
2 // Authors: Sanjay Ghemawat, Jeff Dean, Chandra Chereddi, Lincoln Smith
3 //
4 // Licensed under the Apache License, Version 2.0 (the "License");
5 // you may not use this file except in compliance with the License.
6 // You may obtain a copy of the License at
7 //
8 //      http://www.apache.org/licenses/LICENSE-2.0
9 //
10 // Unless required by applicable law or agreed to in writing, software
11 // distributed under the License is distributed on an "AS IS" BASIS,
12 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 // See the License for the specific language governing permissions and
14 // limitations under the License.
15 //
16 // Implementation of the Bentley/McIlroy algorithm for finding differences.
17 // Bentley, McIlroy.  DCC 1999.  Data Compression Using Long Common Strings.
18 // http://citeseer.ist.psu.edu/555557.html
19 
20 #ifndef OPEN_VCDIFF_BLOCKHASH_H_
21 #define OPEN_VCDIFF_BLOCKHASH_H_
22 
23 #include <config.h>
24 #include <stddef.h>  // size_t
25 #include <stdint.h>  // uint32_t
26 #include <vector>
27 
28 namespace open_vcdiff {
29 
30 // A generic hash table which will be used to keep track of byte runs
31 // of size kBlockSize in both the incrementally processed target data
32 // and the preprocessed source dictionary.
33 //
34 // A custom hash table implementation is used instead of the standard
35 // hash_map template because we know that there will be exactly one
36 // entry in the BlockHash corresponding to each kBlockSize bytes
37 // in the source data, which makes certain optimizations possible:
38 // * The memory for the hash table and for all hash entries can be allocated
39 //   in one step rather than incrementally for each insert operation.
40 // * A single integer can be used to represent both
41 //   the index of the next hash entry in the chain
42 //   and the position of the entry within the source data
43 //   (== kBlockSize * block_number).  This greatly reduces the size
44 //   of a hash entry.
45 //
46 class BlockHash {
47  public:
48   // Block size as per Bentley/McIlroy; must be a power of two.
49   //
50   // Using (for example) kBlockSize = 4 guarantees that no match smaller
51   // than size 4 will be identified, that some matches having sizes
52   // 4, 5, or 6 may be identified, and that all matches
53   // having size 7 or greater will be identified (because any string of
54   // 7 bytes must contain a complete aligned block of 4 bytes.)
55   //
56   // Increasing kBlockSize by a factor of two will halve the amount of
57   // memory needed for the next block table, and will halve the setup time
58   // for a new BlockHash.  However, it also doubles the minimum
59   // match length that is guaranteed to be found in FindBestMatch(),
60   // so that function will be less effective in finding matches.
61   //
62   // Computational effort in FindBestMatch (which is the inner loop of
63   // the encoding algorithm) will be proportional to the number of
64   // matches found, and a low value of kBlockSize will waste time
65   // tracking down small matches.  On the other hand, if this value
66   // is set too high, no matches will be found at all.
67   //
68   // It is suggested that different values of kBlockSize be tried against
69   // a representative data set to find the best tradeoff between
70   // memory/CPU and the effectiveness of FindBestMatch().
71   //
72   // If you change kBlockSize to a smaller value, please increase
73   // kMaxMatchesToCheck accordingly.
74   static const int kBlockSize = 16;
75 
76   // This class is used to store the best match found by FindBestMatch()
77   // and return it to the caller.
78   class Match {
79    public:
Match()80     Match() : size_(0), source_offset_(-1), target_offset_(-1) { }
81 
ReplaceIfBetterMatch(size_t candidate_size,int candidate_source_offset,int candidate_target_offset)82     void ReplaceIfBetterMatch(size_t candidate_size,
83                               int candidate_source_offset,
84                               int candidate_target_offset) {
85       if (candidate_size > size_) {
86         size_ = candidate_size;
87         source_offset_ = candidate_source_offset;
88         target_offset_ = candidate_target_offset;
89       }
90     }
91 
size()92     size_t size() const { return size_; }
source_offset()93     int source_offset() const { return source_offset_; }
target_offset()94     int target_offset() const { return target_offset_; }
95 
96    private:
97      // The size of the best (longest) match passed to ReplaceIfBetterMatch().
98     size_t size_;
99 
100     // The source offset of the match, including the starting_offset_
101     // of the BlockHash for which the match was found.
102     int source_offset_;
103 
104     // The target offset of the match.  An offset of 0 corresponds to the
105     // data at target_start, which is an argument of FindBestMatch().
106     int target_offset_;
107 
108     // Making these private avoids implicit copy constructor
109     // & assignment operator
110     Match(const Match&);  // NOLINT
111     void operator=(const Match&);
112   };
113 
114   // A BlockHash is created using a buffer of source data.  The hash table
115   // will contain one entry for each kBlockSize-byte block in the
116   // source data.
117   //
118   // See the comments for starting_offset_, below, for a description of
119   // the starting_offset argument.  For a hash of source (dictionary) data,
120   // starting_offset_ will be zero; for a hash of previously encoded
121   // target data, starting_offset_ will be equal to the dictionary size.
122   //
123   BlockHash(const char* source_data, size_t source_size, int starting_offset);
124 
125   ~BlockHash();
126 
127   // Initializes the object before use.
128   // This method must be called after constructing a BlockHash object,
129   // and before any other method may be called.  This is because
130   // Init() dynamically allocates hash_table_ and next_block_table_.
131   // Returns true if initialization succeeded, or false if an error occurred,
132   // in which case no other method except the destructor may then be used
133   // on the object.
134   //
135   // If populate_hash_table is true, then AddAllBlocks() will be called
136   // to populate the hash table.  If populate_hash_table is false, then
137   // classes that inherit from BlockHash are expected to call AddBlock()
138   // to incrementally populate individual blocks of data.
139   //
140   bool Init(bool populate_hash_table);
141 
142   // In the context of the open-vcdiff encoder, BlockHash is used for two
143   // purposes: to hash the source (dictionary) data, and to hash
144   // the previously encoded target data.  The main differences between
145   // a dictionary BlockHash and a target BlockHash are as follows:
146   //
147   //   1. The best_match->source_offset() returned from FindBestMatch()
148   //      for a target BlockHash is computed in the following manner:
149   //      the starting offset of the first byte in the target data
150   //      is equal to the dictionary size.  FindBestMatch() will add
151   //      starting_offset_ to any best_match->source_offset() value it returns,
152   //      in order to produce the correct offset value for a target BlockHash.
153   //   2. For a dictionary BlockHash, the entire data set is hashed at once
154   //      when Init() is called with the parameter populate_hash_table = true.
155   //      For a target BlockHash, because the previously encoded target data
156   //      includes only the data seen up to the current encoding position,
157   //      the data blocks are hashed incrementally as the encoding position
158   //      advances, using AddOneIndexHash() and AddAllBlocksThroughIndex().
159   //
160   // The following two factory functions can be used to create BlockHash
161   // objects for each of these two purposes.  Each factory function calls
162   // the object constructor and also calls Init().  If an error occurs,
163   // NULL is returned; otherwise a valid BlockHash object is returned.
164   // Since a dictionary BlockHash is not expected to be modified after
165   // initialization, a const object is returned.
166   // The caller is responsible for deleting the returned object
167   // (using the C++ delete operator) once it is no longer needed.
168   static const BlockHash* CreateDictionaryHash(const char* dictionary_data,
169                                                size_t dictionary_size);
170   static BlockHash* CreateTargetHash(const char* target_data,
171                                      size_t target_size,
172                                      size_t dictionary_size);
173 
174   // This function will be called to add blocks incrementally to the target hash
175   // as the encoding position advances through the target data.  It will be
176   // called for every kBlockSize-byte block in the target data, regardless
177   // of whether the block is aligned evenly on a block boundary.  The
178   // BlockHash will only store hash entries for the evenly-aligned blocks.
179   //
AddOneIndexHash(int index,uint32_t hash_value)180   void AddOneIndexHash(int index, uint32_t hash_value) {
181     if (index == NextIndexToAdd()) {
182       AddBlock(hash_value);
183     }
184   }
185 
186   // Calls AddBlock() for each kBlockSize-byte block in the range
187   // (last_block_added_ * kBlockSize, end_index), exclusive of the endpoints.
188   // If end_index <= the last index added (last_block_added_ * kBlockSize),
189   // this function does nothing.
190   //
191   // A partial block beginning anywhere up to (end_index - 1) is also added,
192   // unless it extends outside the end of the source data.  Like AddAllBlocks(),
193   // this function computes the hash value for each of the blocks in question
194   // from scratch, so it is not a good option if the hash values have already
195   // been computed for some other purpose.
196   //
197   // Example: assume kBlockSize = 4, last_block_added_ = 1, and there are
198   // 14 bytes of source data.
199   // If AddAllBlocksThroughIndex(9) is invoked, then it will call AddBlock()
200   // only for block number 2 (at index 8).
201   // If, after that, AddAllBlocksThroughIndex(14) is invoked, it will not call
202   // AddBlock() at all, because block 3 (beginning at index 12) would
203   // fall outside the range of source data.
204   //
205   // VCDiffEngine::Encode (in vcdiffengine.cc) uses this function to
206   // add a whole range of data to a target hash when a COPY instruction
207   // is generated.
208   void AddAllBlocksThroughIndex(int end_index);
209 
210   // FindBestMatch takes a position within the unencoded target data
211   // (target_candidate_start) and the hash value of the kBlockSize bytes
212   // beginning at that position (hash_value).  It attempts to find a matching
213   // set of bytes within the source (== dictionary) data, expanding
214   // the match both below and above the target block.  It cannot expand
215   // the match outside the bounds of the source data, or below
216   // target_start within the target data, or past
217   // the end limit of (target_start + target_length).
218   //
219   // target_candidate_start is the start of the candidate block within the
220   // target data for which a match will be sought, while
221   // target_start (which is <= target_candidate_start)
222   // is the start of the target data that has yet to be encoded.
223   //
224   // If a match is found whose size is greater than the size
225   // of best_match, this function populates *best_match with the
226   // size, source_offset, and target_offset of the match found.
227   // best_match->source_offset() will contain the index of the start of the
228   // matching source data, plus starting_offset_
229   // (see description of starting_offset_ for details);
230   // best_match->target_offset() will contain the offset of the match
231   // beginning with target_start = offset 0, such that
232   //     0 <= best_match->target_offset()
233   //              <= (target_candidate_start - target_start);
234   // and best_match->size() will contain the size of the match.
235   // If no such match is found, this function leaves *best_match unmodified.
236   //
237   // On calling FindBestMatch(), best_match must
238   // point to a valid Match object, and cannot be NULL.
239   // The same Match object can be passed
240   // when calling FindBestMatch() on a different BlockHash object
241   // for the same candidate data block, in order to find
242   // the best match possible across both objects.  For example:
243   //
244   //     open_vcdiff::BlockHash::Match best_match;
245   //     uint32_t hash_value =
246   //         RollingHash<BlockHash::kBlockSize>::Hash(target_candidate_start);
247   //     bh1.FindBestMatch(hash_value,
248   //                       target_candidate_start,
249   //                       target_start,
250   //                       target_length,
251   //                       &best_match);
252   //     bh2.FindBestMatch(hash_value,
253   //                       target_candidate_start,
254   //                       target_start,
255   //                       target_length,
256   //                       &best_match);
257   //     if (best_size >= 0) {
258   //       // a match was found; its size, source offset, and target offset
259   //       // can be found in best_match
260   //     }
261   //
262   // hash_value is passed as a separate parameter from target_candidate_start,
263   // (rather than calculated within FindBestMatch) in order to take
264   // advantage of the rolling hash, which quickly calculates the hash value
265   // of the block starting at target_candidate_start based on
266   // the known hash value of the block starting at (target_candidate_start - 1).
267   // See vcdiffengine.cc for more details.
268   //
269   // Example:
270   //    kBlockSize: 4
271   //    target text: "ANDREW LLOYD WEBBER"
272   //                 1^    5^2^         3^
273   //    dictionary: "INSURANCE : LLOYDS OF LONDON"
274   //                           4^
275   //    hashed dictionary blocks:
276   //        "INSU", "RANC", "E : ", "LLOY", "DS O", "F LON"
277   //
278   //    1: target_start (beginning of unencoded data)
279   //    2: target_candidate_start (for the block "LLOY")
280   //    3: target_length (points one byte beyond the last byte of data.)
281   //    4: best_match->source_offset() (after calling FindBestMatch)
282   //    5: best_match->target_offset() (after calling FindBestMatch)
283   //
284   //    Under these conditions, FindBestMatch will find a matching
285   //    hashed dictionary block for "LLOY", and will extend the beginning of
286   //    this match backwards by one byte, and the end of the match forwards
287   //    by one byte, finding that the best match is " LLOYD"
288   //    with best_match->source_offset() = 10
289   //                                  (offset of " LLOYD" in the source string),
290   //         best_match->target_offset() = 6
291   //                                  (offset of " LLOYD" in the target string),
292   //     and best_match->size() = 6.
293   //
294   void FindBestMatch(uint32_t hash_value,
295                      const char* target_candidate_start,
296                      const char* target_start,
297                      size_t target_size,
298                      Match* best_match) const;
299 
300  protected:
301   // FindBestMatch() will not process more than this number
302   // of matching hash entries.
303   //
304   // It is necessary to have a limit on the maximum number of matches
305   // that will be checked in order to avoid the worst-case performance
306   // possible if, for example, all the blocks in the dictionary have
307   // the same hash value.  See the unit test SearchStringFindsTooManyMatches
308   // for an example of such a case.  The encoder uses a loop in
309   // VCDiffEngine::Encode over each target byte, containing a loop in
310   // BlockHash::FindBestMatch over the number of matches (up to a maximum
311   // of the number of source blocks), containing two loops that extend
312   // the match forwards and backwards up to the number of source bytes.
313   // Total complexity in the worst case is
314   //     O([target size] * source_size_ * source_size_)
315   // Placing a limit on the possible number of matches checked changes this to
316   //     O([target size] * source_size_ * kMaxMatchesToCheck)
317   //
318   // In empirical testing on real HTML text, using a block size of 4,
319   // the number of true matches per call to FindBestMatch() did not exceed 78;
320   // with a block size of 32, the number of matches did not exceed 3.
321   //
322   // The expected number of true matches scales super-linearly
323   // with the inverse of kBlockSize, but here a linear scale is used
324   // for block sizes smaller than 32.
325   static const int kMaxMatchesToCheck = (kBlockSize >= 32) ? 32 :
326                                             (32 * (32 / kBlockSize));
327 
328   // Do not skip more than this number of non-matching hash collisions
329   // to find the next matching entry in the hash chain.
330   static const int kMaxProbes = 16;
331 
332   // Internal routine which calculates a hash table size based on kBlockSize and
333   // the dictionary_size.  Will return a power of two if successful, or 0 if an
334   // internal error occurs.  Some calculations (such as GetHashTableIndex())
335   // depend on the table size being a power of two.
336   static size_t CalcTableSize(const size_t dictionary_size);
337 
GetNumberOfBlocks()338   size_t GetNumberOfBlocks() const {
339     return source_size_ / kBlockSize;
340   }
341 
342   // Use the lowest-order bits of the hash value
343   // as the index into the hash table.
GetHashTableIndex(uint32_t hash_value)344   uint32_t GetHashTableIndex(uint32_t hash_value) const {
345     return hash_value & hash_table_mask_;
346   }
347 
348   // The index within source_data_ of the next block
349   // for which AddBlock() should be called.
NextIndexToAdd()350   int NextIndexToAdd() const {
351     return (last_block_added_ + 1) * kBlockSize;
352   }
353 
354   static inline bool TooManyMatches(int* match_counter);
355 
source_data()356   const char* source_data() { return source_data_; }
source_size()357   size_t source_size() { return source_size_; }
358 
359   // Adds an entry to the hash table for one block of source data of length
360   // kBlockSize, starting at source_data_[block_number * kBlockSize],
361   // where block_number is always (last_block_added_ + 1).  That is,
362   // AddBlock() must be called once for each block in source_data_
363   // in increasing order.
364   void AddBlock(uint32_t hash_value);
365 
366   // Calls AddBlock() for each complete kBlockSize-byte block between
367   // source_data_ and (source_data_ + source_size_).  It is equivalent
368   // to calling AddAllBlocksThroughIndex(source_data + source_size).
369   // This function is called when Init(true) is invoked.
370   void AddAllBlocks();
371 
372   // Returns true if the contents of the kBlockSize-byte block
373   // beginning at block1 are identical to the contents of
374   // the block beginning at block2; false otherwise.
375   static bool BlockContentsMatch(const char* block1, const char* block2);
376 
377   // Compares each machine word of the two (possibly unaligned) blocks, rather
378   // than each byte, thus reducing the number of test-and-branch instructions
379   // executed.  Returns a boolean (do the blocks match?) rather than
380   // the signed byte difference returned by memcmp.
381   //
382   // BlockContentsMatch will use either this function or memcmp to do its work,
383   // depending on which is faster for a particular architecture.
384   //
385   // For gcc on x86-based architectures, this function has been shown to run
386   // about twice as fast as the library function memcmp(), and between five and
387   // nine times faster than the assembly instructions (repz and cmpsb) that gcc
388   // uses by default for builtin memcmp.  On other architectures, or using
389   // other compilers, this function has not shown to be faster than memcmp.
390   static bool BlockCompareWords(const char* block1, const char* block2);
391 
392   // Finds the first block number within the hashed data
393   // that represents a match for the given hash value.
394   // Returns -1 if no match was found.
395   //
396   // Init() must have been called and returned true before using
397   // FirstMatchingBlock or NextMatchingBlock.  No check is performed
398   // for this condition; the code will crash if this condition is violated.
399   //
400   // The hash table is initially populated with -1 (not found) values,
401   // so if this function is called before the hash table has been populated
402   // using AddAllBlocks() or AddBlock(), it will simply return -1
403   // for any value of hash_value.
404   int FirstMatchingBlock(uint32_t hash_value, const char* block_ptr) const;
405 
406   // Given a block number returned by FirstMatchingBlock()
407   // or by a previous call to NextMatchingBlock(), returns
408   // the next block number that matches the same hash value.
409   // Returns -1 if no match was found.
410   int NextMatchingBlock(int block_number, const char* block_ptr) const;
411 
412   // Inline version of FirstMatchingBlock.  This saves the cost of a function
413   // call when this routine is called from within the module.  The external
414   // (non-inlined) version is called only by unit tests.
415   inline int FirstMatchingBlockInline(uint32_t hash_value,
416                                       const char* block_ptr) const;
417 
418   // Walk through the hash entry chain, skipping over any false matches
419   // (for which the lowest bits of the fingerprints match,
420   // but the actual block data does not.)  Returns the block number of
421   // the first true match found, or -1 if no true match was found.
422   // If block_number is a matching block, the function will return block_number
423   // without skipping to the next block.
424   int SkipNonMatchingBlocks(int block_number, const char* block_ptr) const;
425 
426   // Returns the number of bytes to the left of source_match_start
427   // that match the corresponding bytes to the left of target_match_start.
428   // Will not examine more than max_bytes bytes, which is to say that
429   // the return value will be in the range [0, max_bytes] inclusive.
430   static int MatchingBytesToLeft(const char* source_match_start,
431                                  const char* target_match_start,
432                                  int max_bytes);
433 
434   // Returns the number of bytes starting at source_match_end
435   // that match the corresponding bytes starting at target_match_end.
436   // Will not examine more than max_bytes bytes, which is to say that
437   // the return value will be in the range [0, max_bytes] inclusive.
438   static int MatchingBytesToRight(const char* source_match_end,
439                                   const char* target_match_end,
440                                   int max_bytes);
441 
442   // The protected functions BlockContentsMatch, FirstMatchingBlock,
443   // NextMatchingBlock, MatchingBytesToLeft, and MatchingBytesToRight
444   // should be made accessible to unit tests.
445   friend class BlockHashTest;
446 
447  private:
448   const char* const  source_data_;
449   const size_t       source_size_;
450 
451   // The size of this array is determined using CalcTableSize().  It has at
452   // least one element for each kBlockSize-byte block in the source data.
453   // GetHashTableIndex() returns an index into this table for a given hash
454   // value.  The value of each element of hash_table_ is the lowest block
455   // number in the source data whose hash value would return the same value from
456   // GetHashTableIndex(), or -1 if there is no matching block.  This value can
457   // then be used as an index into next_block_table_ to retrieve the entire set
458   // of matching block numbers.
459   std::vector<int> hash_table_;
460 
461   // An array containing one element for each source block.  Each element is
462   // either -1 (== not found) or the index of the next block whose hash value
463   // would produce a matching result from GetHashTableIndex().
464   std::vector<int> next_block_table_;
465 
466   // This vector has the same size as next_block_table_.  For every block number
467   // B that is referenced in hash_table_, last_block_table_[B] will contain
468   // the maximum block number that has the same GetHashTableIndex() value
469   // as block B.  This number may be B itself.  For a block number B' that
470   // is not referenced in hash_table_, the value of last_block_table_[B'] is -1.
471   // This table is used only while populating the hash table, not while looking
472   // up hash values in the table.  Keeping track of the last block number in the
473   // chain allows us to construct the block chains as FIFO rather than LIFO
474   // lists, so that the match with the lowest index is returned first.  This
475   // should result in a more compact encoding because the VCDIFF format favors
476   // smaller index values and repeated index values.
477   std::vector<int> last_block_table_;
478 
479   // Performing a bitwise AND with hash_table_mask_ will produce a value ranging
480   // from 0 to the number of elements in hash_table_.
481   uint32_t hash_table_mask_;
482 
483   // The offset of the first byte of source data (the data at source_data_[0]).
484   // For the purpose of computing offsets, the source data and target data
485   // are considered to be concatenated -- not literally in a single memory
486   // buffer, but conceptually as described in the RFC.
487   // The first byte of the previously encoded target data
488   // has an offset that is equal to dictionary_size, i.e., just after
489   // the last byte of source data.
490   // For a hash of source (dictionary) data, starting_offset_ will be zero;
491   // for a hash of previously encoded target data, starting_offset_ will be
492   // equal to the dictionary size.
493   const int starting_offset_;
494 
495   // The last index added by AddBlock().  This determines the block number
496   // for successive calls to AddBlock(), and is also
497   // used to determine the starting block for AddAllBlocksThroughIndex().
498   int last_block_added_;
499 
500   // Making these private avoids implicit copy constructor & assignment operator
501   BlockHash(const BlockHash&);  // NOLINT
502   void operator=(const BlockHash&);
503 };
504 
505 }  // namespace open_vcdiff
506 
507 #endif  // OPEN_VCDIFF_BLOCKHASH_H_
508