github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-06-28 16:41:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

fdr_compile: tidy up

Browse Source
This commit is contained in:
Justin Viiret 2016-03-02 10:23:31 +11:00 committed by Matthew Barr
parent 466fc940e5
commit 88e6485e75
1 changed files with 14 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
src/fdr/fdr_compile.cpp
View File

@ -124,10 +124,8 @@ void FDRCompiler::createInitialState(FDR *fdr) {
// Find the minimum length for the literals in this bucket. // Find the minimum length for the literals in this bucket.
const vector<LiteralIndex> &bucket_lits = bucketToLits[b]; const vector<LiteralIndex> &bucket_lits = bucketToLits[b];
u32 min_len = ~0U; u32 min_len = ~0U;
for (vector<LiteralIndex>::const_iterator it = bucket_lits.begin(), for (const LiteralIndex &lit_idx : bucket_lits) {
ite = bucket_lits.end(); min_len = min(min_len, verify_u32(lits[lit_idx].s.length()));
it != ite; ++it) {
min_len = min(min_len, verify_u32(lits[*it].s.length()));
} }
DEBUG_PRINTF("bucket %u has min_len=%u\n", b, min_len); DEBUG_PRINTF("bucket %u has min_len=%u\n", b, min_len);
@ -213,13 +211,11 @@ struct LitOrder {
if (len1 != len2) { if (len1 != len2) {
return len1 < len2; return len1 < len2;
} else { } else {
string::const_reverse_iterator it1, it2; auto p = std::mismatch(i1s.rbegin(), i1s.rend(), i2s.rbegin());
tie(it1, it2) = if (p.first == i1s.rend()) {
std::mismatch(i1s.rbegin(), i1s.rend(), i2s.rbegin());
if (it1 == i1s.rend()) {
return false; return false;
} }
return *it1 < *it2; return *p.first < *p.second;
} }
} }
@ -262,9 +258,8 @@ void FDRCompiler::assignStringsToBuckets() {
stable_sort(vli.begin(), vli.end(), LitOrder(lits)); stable_sort(vli.begin(), vli.end(), LitOrder(lits));
#ifdef DEBUG_ASSIGNMENT #ifdef DEBUG_ASSIGNMENT
for (map<u32, u32>::iterator i = lenCounts.begin(), e = lenCounts.end(); for (const auto &m : lenCounts) {
i != e; ++i) { printf("l<%u>:%u ", m.first, m.second);
printf("l<%d>:%d ", i->first, i->second);
} }
printf("\n"); printf("\n");
#endif #endif
@ -401,8 +396,7 @@ bool getMultiEntriesAtPosition(const FDREngineDescription &eng,
distance = 4; distance = 4;
} }
for (vector<LiteralIndex>::const_iterator i = vl.begin(), e = vl.end(); for (auto i = vl.begin(), e = vl.end(); i != e; ++i) {
i != e; ++i) {
if (e - i > 5) { if (e - i > 5) {
__builtin_prefetch(&lits[*(i + 5)]); __builtin_prefetch(&lits[*(i + 5)]);
} }
@ -456,31 +450,25 @@ void FDRCompiler::setupTab() {
memcpy(tabIndexToMask(i), &defaultMask[0], mask_size); memcpy(tabIndexToMask(i), &defaultMask[0], mask_size);
} }
typedef std::map<u32, ue2::unordered_set<u32> > M2SET;
for (BucketIndex b = 0; b < eng.getNumBuckets(); b++) { for (BucketIndex b = 0; b < eng.getNumBuckets(); b++) {
const vector<LiteralIndex> &vl = bucketToLits[b]; const vector<LiteralIndex> &vl = bucketToLits[b];
SuffixPositionInString pLimit = eng.getBucketWidth(b); SuffixPositionInString pLimit = eng.getBucketWidth(b);
for (SuffixPositionInString pos = 0; pos < pLimit; pos++) { for (SuffixPositionInString pos = 0; pos < pLimit; pos++) {
u32 bit = eng.getSchemeBit(b, pos); u32 bit = eng.getSchemeBit(b, pos);
M2SET m2; map<u32, ue2::unordered_set<u32>> m2;
bool done = getMultiEntriesAtPosition(eng, vl, lits, pos, m2); bool done = getMultiEntriesAtPosition(eng, vl, lits, pos, m2);
if (done) { if (done) {
clearbit(&defaultMask[0], bit); clearbit(&defaultMask[0], bit);
continue; continue;
} }
for (M2SET::const_iterator i = m2.begin(), e = m2.end(); i != e; for (const auto &elem : m2) {
++i) { u32 dc = elem.first;
u32 dc = i->first; const ue2::unordered_set<u32> &mskSet = elem.second;
const ue2::unordered_set<u32> &mskSet = i->second;
u32 v = ~dc; u32 v = ~dc;
do { do {
u32 b2 = v & dc; u32 b2 = v & dc;
for (ue2::unordered_set<u32>::const_iterator for (const u32 &mskVal : mskSet) {
i2 = mskSet.begin(), u32 val = (mskVal & ~dc) | b2;
e2 = mskSet.end();
i2 != e2; ++i2) {
u32 val = (*i2 & ~dc) | b2;
clearbit(tabIndexToMask(val), bit); clearbit(tabIndexToMask(val), bit);
} }
v = (v + (dc & -dc)) | ~dc; v = (v + (dc & -dc)) | ~dc;