diff options
author | Geir Storli <geirst@verizonmedia.com> | 2019-05-08 07:40:36 +0000 |
---|---|---|
committer | Geir Storli <geirst@verizonmedia.com> | 2019-05-08 07:40:36 +0000 |
commit | fd554ae1f32a636009a2bebf5039410ea0acc984 (patch) | |
tree | 5aad24e0b9d913101c5c61660da524003cd95c12 /searchlib/src/tests/postinglistbm | |
parent | 0c379910129e489b8bbd3ecde16690120068a669 (diff) |
Make it possible to specify doc freq for medium and rare words.
Diffstat (limited to 'searchlib/src/tests/postinglistbm')
-rw-r--r-- | searchlib/src/tests/postinglistbm/andstress.cpp | 12 | ||||
-rw-r--r-- | searchlib/src/tests/postinglistbm/andstress.h | 2 | ||||
-rw-r--r-- | searchlib/src/tests/postinglistbm/postinglistbm.cpp | 18 |
3 files changed, 17 insertions, 15 deletions
diff --git a/searchlib/src/tests/postinglistbm/andstress.cpp b/searchlib/src/tests/postinglistbm/andstress.cpp index adca7892464..fcc234ef4e0 100644 --- a/searchlib/src/tests/postinglistbm/andstress.cpp +++ b/searchlib/src/tests/postinglistbm/andstress.cpp @@ -39,7 +39,6 @@ private: search::Rand48 &_rnd; unsigned int _numDocs; - unsigned int _commonDocFreq; std::vector<std::string> _postingTypes; unsigned int _loops; unsigned int _skipCommonPairsRate; @@ -68,8 +67,6 @@ private: public: AndStressMaster(search::Rand48 &rnd, FakeWordSet &wordSet, - unsigned int numDocs, - unsigned int commonDocFreq, const std::vector<std::string> &postingType, unsigned int loops, unsigned int skipCommonPairsRate, @@ -117,8 +114,6 @@ makePosting(FakeWord &fw) AndStressMaster::AndStressMaster(search::Rand48 &rnd, FakeWordSet &wordSet, - unsigned int numDocs, - unsigned int commonDocFreq, const std::vector<std::string> &postingTypes, unsigned int loops, unsigned int skipCommonPairsRate, @@ -126,8 +121,7 @@ AndStressMaster::AndStressMaster(search::Rand48 &rnd, uint32_t stride, bool unpack) : _rnd(rnd), - _numDocs(numDocs), - _commonDocFreq(commonDocFreq), + _numDocs(wordSet.numDocs()), _postingTypes(postingTypes), _loops(loops), _skipCommonPairsRate(skipCommonPairsRate), @@ -399,8 +393,6 @@ AndStress::~AndStress() void AndStress::run(search::Rand48 &rnd, FakeWordSet &wordSet, - unsigned int numDocs, - unsigned int commonDocFreq, const std::vector<std::string> &postingTypes, unsigned int loops, unsigned int skipCommonPairsRate, @@ -410,7 +402,7 @@ AndStress::run(search::Rand48 &rnd, { LOG(debug, "Andstress::run"); AndStressMaster master(rnd, wordSet, - numDocs, commonDocFreq, postingTypes, loops, + postingTypes, loops, skipCommonPairsRate, numTasks, stride, diff --git a/searchlib/src/tests/postinglistbm/andstress.h b/searchlib/src/tests/postinglistbm/andstress.h index fdc99fb42ba..7a7fe2f66f2 100644 --- a/searchlib/src/tests/postinglistbm/andstress.h +++ b/searchlib/src/tests/postinglistbm/andstress.h @@ -21,8 +21,6 @@ public: void run(search::Rand48 &rnd, search::fakedata::FakeWordSet &wordSet, - unsigned int numDocs, - unsigned int commonDocFreq, const std::vector<std::string> &postingTypes, unsigned int loops, unsigned int skipCommonPairsRate, diff --git a/searchlib/src/tests/postinglistbm/postinglistbm.cpp b/searchlib/src/tests/postinglistbm/postinglistbm.cpp index 0a6f99ede11..890ca8fd6eb 100644 --- a/searchlib/src/tests/postinglistbm/postinglistbm.cpp +++ b/searchlib/src/tests/postinglistbm/postinglistbm.cpp @@ -30,6 +30,8 @@ class PostingListBM : public FastOS_Application { private: uint32_t _numDocs; uint32_t _commonDocFreq; + uint32_t _mediumDocFreq; + uint32_t _rareDocFreq; uint32_t _numWordsPerClass; std::vector<std::string> _postingTypes; uint32_t _loops; @@ -54,6 +56,8 @@ usage() "[-C <skipCommonPairsRate>] " "[-T {string, array, weightedSet}] " "[-c <commonDoqFreq>] " + "[-m <mediumDoqFreq>] " + "[-r <rareDoqFreq>] " "[-d <numDocs>] " "[-l <numLoops>] " "[-s <stride>] " @@ -84,6 +88,8 @@ badPostingType(const std::string &postingType) PostingListBM::PostingListBM() : _numDocs(10000000), _commonDocFreq(50000), + _mediumDocFreq(1000), + _rareDocFreq(10), _numWordsPerClass(100), _postingTypes(), _loops(1), @@ -109,7 +115,7 @@ PostingListBM::Main() bool hasElementWeights = false; bool quick = false; - while ((c = GetOpt("C:c:d:l:s:t:uw:T:q", optArg, argi)) != -1) { + while ((c = GetOpt("C:c:m:r:d:l:s:t:uw:T:q", optArg, argi)) != -1) { switch(c) { case 'C': _skipCommonPairsRate = atoi(optArg); @@ -132,6 +138,12 @@ PostingListBM::Main() case 'c': _commonDocFreq = atoi(optArg); break; + case 'm': + _mediumDocFreq = atoi(optArg); + break; + case 'r': + _rareDocFreq = atoi(optArg); + break; case 'd': _numDocs = atoi(optArg); break; @@ -190,11 +202,11 @@ PostingListBM::Main() _postingTypes = getPostingTypes(); } - _wordSet.setupWords(_rnd, _numDocs, _commonDocFreq, _numWordsPerClass); + _wordSet.setupWords(_rnd, _numDocs, _commonDocFreq, _mediumDocFreq, _rareDocFreq, _numWordsPerClass); AndStress andstress; andstress.run(_rnd, _wordSet, - _numDocs, _commonDocFreq, _postingTypes, _loops, + _postingTypes, _loops, _skipCommonPairsRate, numTasks, _stride, |