mirror of
https://github.com/facebook/zstd.git
synced 2025-08-10 04:43:07 +03:00
changed samples to geometric distribution so that it is huffman compressible
This commit is contained in:
@@ -1344,17 +1344,22 @@ static int generateCorpusWithDict(U32 seed, unsigned numFiles, const char* const
|
|||||||
size_t dictWriteSize = 0;
|
size_t dictWriteSize = 0;
|
||||||
|
|
||||||
/* create random samples */
|
/* create random samples */
|
||||||
unsigned numSamples = RAND(&seed) % 50;
|
unsigned numSamples = 0;
|
||||||
BYTE* samples;
|
BYTE* samples;
|
||||||
unsigned i = 0;
|
unsigned i = 0;
|
||||||
size_t* sampleSizes = malloc(numSamples*sizeof(size_t));
|
size_t* sampleSizes = malloc(numSamples*sizeof(size_t));
|
||||||
{
|
{
|
||||||
size_t* curr = sampleSizes;
|
size_t* curr = sampleSizes;
|
||||||
size_t totalSize = 0;
|
size_t totalSize = 0;
|
||||||
while(i++ < numSamples){
|
while(numSamples < 100){
|
||||||
*curr = RAND(&seed) % dictContentSize;
|
unsigned numReps = ROUND(RAND_exp(&seed, 10));
|
||||||
totalSize += *curr;
|
size_t randSize = RAND(&seed) % dictContentSize;
|
||||||
curr++;
|
while(numReps-- > 0){
|
||||||
|
*curr = randSize;
|
||||||
|
totalSize += *curr;
|
||||||
|
curr++;
|
||||||
|
}
|
||||||
|
numSamples += numReps;
|
||||||
}
|
}
|
||||||
samples = malloc(totalSize);
|
samples = malloc(totalSize);
|
||||||
|
|
||||||
@@ -1374,7 +1379,6 @@ static int generateCorpusWithDict(U32 seed, unsigned numFiles, const char* const
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
{
|
{
|
||||||
/* set dictionary params */
|
/* set dictionary params */
|
||||||
ZDICT_params_t zdictParams;
|
ZDICT_params_t zdictParams;
|
||||||
|
Reference in New Issue
Block a user