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

teddy_compile: style fixes, whitespace

Browse Source
This commit is contained in:
Justin Viiret 2017-06-22 15:21:22 +10:00 committed by Matthew Barr
parent 84a09d35d6
commit cc4a5cc36f
1 changed files with 52 additions and 44 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

68
src/fdr/teddy_compile.cpp
View File

@ -70,6 +70,9 @@ namespace {
//#define TEDDY_DEBUG //#define TEDDY_DEBUG
/** \brief Max number of Teddy masks we use. */
static constexpr size_t MAX_NUM_MASKS = 4;
class TeddyCompiler : noncopyable { class TeddyCompiler : noncopyable {
const TeddyEngineDescription ŋ const TeddyEngineDescription ŋ
const Grey &grey; const Grey &grey;
@ -88,9 +91,6 @@ public:
}; };
class TeddySet { class TeddySet {
/** \brief Max number of Teddy masks we use. */
static constexpr size_t MAX_NUM_MASKS = 4;
/** /**
* \brief Estimate of the max number of literals in a set, used to * \brief Estimate of the max number of literals in a set, used to
* minimise allocations. * minimise allocations.
@ -153,7 +153,8 @@ public:
u8 c_lo = c & 0xf; u8 c_lo = c & 0xf;
nibbleSets[i * 2] = 1 << c_lo; nibbleSets[i * 2] = 1 << c_lo;
if (lit.nocase && ourisalpha(c)) { if (lit.nocase && ourisalpha(c)) {
nibbleSets[i*2+1] = (1 << (c_hi&0xd)) | (1 << (c_hi|0x2)); nibbleSets[i * 2 + 1] =
(1 << (c_hi & 0xd)) | (1 << (c_hi | 0x2));
} else { } else {
nibbleSets[i * 2 + 1] = 1 << c_hi; nibbleSets[i * 2 + 1] = 1 << c_hi;
} }
@ -165,14 +166,6 @@ public:
sort_and_unique(litIds); sort_and_unique(litIds);
} }
void merge(const TeddySet &ts) {
for (u32 i = 0; i < nibbleSets.size(); i++) {
nibbleSets[i] |= ts.nibbleSets[i];
}
litIds.insert(litIds.end(), ts.litIds.begin(), ts.litIds.end());
sort_and_unique(litIds);
}
// return a value p from 0 .. MAXINT64 that gives p/MAXINT64 // return a value p from 0 .. MAXINT64 that gives p/MAXINT64
// likelihood of this TeddySet firing a first-stage accept // likelihood of this TeddySet firing a first-stage accept
// if it was given a bucket of its own and random data were // if it was given a bucket of its own and random data were
@ -206,6 +199,21 @@ public:
} }
return true; return true;
} }
friend TeddySet merge(const TeddySet &a, const TeddySet &b) {
assert(a.nibbleSets.size() == b.nibbleSets.size());
TeddySet m(a);
for (size_t i = 0; i < m.nibbleSets.size(); i++) {
m.nibbleSets[i] |= b.nibbleSets[i];
}
m.litIds.insert(m.litIds.end(), b.litIds.begin(), b.litIds.end());
sort_and_unique(m.litIds);
return m;
}
}; };
bool TeddyCompiler::pack(map<BucketIndex, bool TeddyCompiler::pack(map<BucketIndex,
@ -222,7 +230,8 @@ bool TeddyCompiler::pack(map<BucketIndex,
#ifdef TEDDY_DEBUG #ifdef TEDDY_DEBUG
printf("Size %zu\n", sts.size()); printf("Size %zu\n", sts.size());
for (const TeddySet &ts : sts) { for (const TeddySet &ts : sts) {
printf("\n"); ts.dump(); printf("\n");
ts.dump();
} }
printf("\n===============================================\n"); printf("\n===============================================\n");
#endif #endif
@ -242,9 +251,7 @@ bool TeddyCompiler::pack(map<BucketIndex,
continue; continue;
} }
TeddySet tmpSet(eng.numMasks); TeddySet tmpSet = merge(s1, s2);
tmpSet.merge(s1);
tmpSet.merge(s2);
u64a newScore = tmpSet.heuristic(); u64a newScore = tmpSet.heuristic();
u64a oldScore = s1.heuristic() + s2.heuristic(); u64a oldScore = s1.heuristic() + s2.heuristic();
if (newScore < oldScore) { if (newScore < oldScore) {
@ -272,9 +279,7 @@ bool TeddyCompiler::pack(map<BucketIndex,
} }
// do the merge // do the merge
TeddySet nts(eng.numMasks); TeddySet nts = merge(*m1, *m2);
nts.merge(*m1);
nts.merge(*m2);
#ifdef TEDDY_DEBUG #ifdef TEDDY_DEBUG
printf("Merging\n"); printf("Merging\n");
printf("m1 = \n"); printf("m1 = \n");
@ -305,6 +310,8 @@ bool TeddyCompiler::pack(map<BucketIndex,
} }
bytecode_ptr<FDR> TeddyCompiler::build() { bytecode_ptr<FDR> TeddyCompiler::build() {
assert(eng.numMasks <= MAX_NUM_MASKS);
if (lits.size() > eng.getNumBuckets() * TEDDY_BUCKET_LOAD) { if (lits.size() > eng.getNumBuckets() * TEDDY_BUCKET_LOAD) {
DEBUG_PRINTF("too many literals: %zu\n", lits.size()); DEBUG_PRINTF("too many literals: %zu\n", lits.size());
return nullptr; return nullptr;
@ -383,15 +390,17 @@ bytecode_ptr<FDR> TeddyCompiler::build() {
// fill in masks // fill in masks
for (u32 j = 0; j < eng.numMasks; j++) { for (u32 j = 0; j < eng.numMasks; j++) {
u32 msk_id_lo = j * 2 * maskWidth + (bucket_id / 8); const u32 msk_id_lo = j * 2 * maskWidth + (bucket_id / 8);
u32 msk_id_hi = (j * 2 + 1) * maskWidth + (bucket_id / 8); const u32 msk_id_hi = (j * 2 + 1) * maskWidth + (bucket_id / 8);
const u32 lo_base = msk_id_lo * 16;
const u32 hi_base = msk_id_hi * 16;
// if we don't have a char at this position, fill in i // if we don't have a char at this position, fill in i
// locations in these masks with '1' // locations in these masks with '1'
if (j >= sz) { if (j >= sz) {
for (u32 n = 0; n < 16; n++) { for (u32 n = 0; n < 16; n++) {
baseMsk[msk_id_lo * 16 + n] |= bmsk; baseMsk[lo_base + n] |= bmsk;
baseMsk[msk_id_hi * 16 + n] |= bmsk; baseMsk[hi_base + n] |= bmsk;
} }
} else { } else {
u8 c = l.s[sz - 1 - j]; u8 c = l.s[sz - 1 - j];
@ -410,29 +419,28 @@ bytecode_ptr<FDR> TeddyCompiler::build() {
for (u8 cm = 0; cm < 0x10; cm++) { for (u8 cm = 0; cm < 0x10; cm++) {
if ((cm & m_lo) == (cmp_lo & m_lo)) { if ((cm & m_lo) == (cmp_lo & m_lo)) {
baseMsk[msk_id_lo * 16 + cm] |= bmsk; baseMsk[lo_base + cm] |= bmsk;
} }
if ((cm & m_hi) == (cmp_hi & m_hi)) { if ((cm & m_hi) == (cmp_hi & m_hi)) {
baseMsk[msk_id_hi * 16 + cm] |= bmsk; baseMsk[hi_base + cm] |= bmsk;
} }
} }
} else { } else {
if (l.nocase && ourisalpha(c)) { if (l.nocase && ourisalpha(c)) {
u32 cmHalfClear = (0xdf >> hiShift) & 0xf; u32 cmHalfClear = (0xdf >> hiShift) & 0xf;
u32 cmHalfSet = (0x20 >> hiShift) & 0xf; u32 cmHalfSet = (0x20 >> hiShift) & 0xf;
baseMsk[msk_id_hi * 16 + (n_hi & cmHalfClear)] |= bmsk; baseMsk[hi_base + (n_hi & cmHalfClear)] |= bmsk;
baseMsk[msk_id_hi * 16 + (n_hi | cmHalfSet )] |= bmsk; baseMsk[hi_base + (n_hi | cmHalfSet)] |= bmsk;
} else { } else {
baseMsk[msk_id_hi * 16 + n_hi] |= bmsk; baseMsk[hi_base + n_hi] |= bmsk;
} }
baseMsk[msk_id_lo * 16 + n_lo] |= bmsk; baseMsk[lo_base + n_lo] |= bmsk;
} }
} }
} }
} }
} }
#ifdef TEDDY_DEBUG #ifdef TEDDY_DEBUG
for (u32 i = 0; i < eng.numMasks * 2; i++) { for (u32 i = 0; i < eng.numMasks * 2; i++) {
for (u32 j = 0; j < 16; j++) { for (u32 j = 0; j < 16; j++) {