github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-07-13 05:54:43 +03:00
Code Issues Packages Projects Releases Wiki Activity

dfa: allow smwr to avoid lengthy daddy recalc

Browse Source
This commit is contained in:
Justin Viiret 2017-04-11 12:44:20 +10:00 committed by Matthew Barr
parent 1538d90a9e
commit beac58fcb4
3 changed files with 26 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
src/nfa/mcclellancompile.cpp
View File

@ -802,9 +802,9 @@ flat_set<dstate_id_t> find_daddy_candidates(const dfa_info &info,
#define MAX_SHERMAN_SELF_LOOP 20 #define MAX_SHERMAN_SELF_LOOP 20
static static
void find_better_daddy(dfa_info &info, dstate_id_t curr_id, void find_better_daddy(dfa_info &info, dstate_id_t curr_id, bool using8bit,
bool using8bit, bool any_cyclic_near_anchored_state, bool any_cyclic_near_anchored_state,
const Grey &grey) { bool trust_daddy_states, const Grey &grey) {
if (!grey.allowShermanStates) { if (!grey.allowShermanStates) {
return; return;
} }
@ -839,7 +839,12 @@ void find_better_daddy(dfa_info &info, dstate_id_t curr_id,
dstate_id_t best_daddy = 0; dstate_id_t best_daddy = 0;
dstate &currState = info.states[curr_id]; dstate &currState = info.states[curr_id];
const auto hinted = find_daddy_candidates(info, curr_id); flat_set<dstate_id_t> hinted;
if (trust_daddy_states) {
hinted.insert(currState.daddy);
} else {
hinted = find_daddy_candidates(info, curr_id);
}
for (const dstate_id_t &donor : hinted) { for (const dstate_id_t &donor : hinted) {
assert(donor < curr_id); assert(donor < curr_id);
@ -947,6 +952,7 @@ bool is_cyclic_near(const raw_dfa &raw, dstate_id_t root) {
bytecode_ptr<NFA> mcclellanCompile_i(raw_dfa &raw, accel_dfa_build_strat &strat, bytecode_ptr<NFA> mcclellanCompile_i(raw_dfa &raw, accel_dfa_build_strat &strat,
const CompileContext &cc, const CompileContext &cc,
bool trust_daddy_states,
set<dstate_id_t> *accel_states) { set<dstate_id_t> *accel_states) {
u16 total_daddy = 0; u16 total_daddy = 0;
dfa_info info(strat); dfa_info info(strat);
@ -963,7 +969,7 @@ bytecode_ptr<NFA> mcclellanCompile_i(raw_dfa &raw, accel_dfa_build_strat &strat,
for (u32 i = 0; i < info.size(); i++) { for (u32 i = 0; i < info.size(); i++) {
find_better_daddy(info, i, using8bit, any_cyclic_near_anchored_state, find_better_daddy(info, i, using8bit, any_cyclic_near_anchored_state,
cc.grey); trust_daddy_states, cc.grey);
total_daddy += info.extra[i].daddytaken; total_daddy += info.extra[i].daddytaken;
} }
@ -989,9 +995,10 @@ bytecode_ptr<NFA> mcclellanCompile_i(raw_dfa &raw, accel_dfa_build_strat &strat,
bytecode_ptr<NFA> mcclellanCompile(raw_dfa &raw, const CompileContext &cc, bytecode_ptr<NFA> mcclellanCompile(raw_dfa &raw, const CompileContext &cc,
const ReportManager &rm, const ReportManager &rm,
bool only_accel_init, bool only_accel_init,
bool trust_daddy_states,
set<dstate_id_t> *accel_states) { set<dstate_id_t> *accel_states) {
mcclellan_build_strat mbs(raw, rm, only_accel_init); mcclellan_build_strat mbs(raw, rm, only_accel_init);
return mcclellanCompile_i(raw, mbs, cc, accel_states); return mcclellanCompile_i(raw, mbs, cc, trust_daddy_states, accel_states);
} }
size_t mcclellan_build_strat::accelSize(void) const { size_t mcclellan_build_strat::accelSize(void) const {

3
src/nfa/mcclellancompile.h
View File

@ -71,12 +71,13 @@ private:
bytecode_ptr<NFA> bytecode_ptr<NFA>
mcclellanCompile(raw_dfa &raw, const CompileContext &cc, mcclellanCompile(raw_dfa &raw, const CompileContext &cc,
const ReportManager &rm, bool only_accel_init, const ReportManager &rm, bool only_accel_init,
bool trust_daddy_states = false,
std::set<dstate_id_t> *accel_states = nullptr); std::set<dstate_id_t> *accel_states = nullptr);
/* used internally by mcclellan/haig/gough compile process */ /* used internally by mcclellan/haig/gough compile process */
bytecode_ptr<NFA> bytecode_ptr<NFA>
mcclellanCompile_i(raw_dfa &raw, accel_dfa_build_strat &strat, mcclellanCompile_i(raw_dfa &raw, accel_dfa_build_strat &strat,
const CompileContext &cc, const CompileContext &cc, bool trust_daddy_states = false,
std::set<dstate_id_t> *accel_states = nullptr); std::set<dstate_id_t> *accel_states = nullptr);
/** /**

19
src/smallwrite/smallwrite_build.cpp
View File

@ -692,18 +692,20 @@ bool is_slow(const raw_dfa &rdfa, const set<dstate_id_t> &accel,
static static
bytecode_ptr<NFA> getDfa(raw_dfa &rdfa, const CompileContext &cc, bytecode_ptr<NFA> getDfa(raw_dfa &rdfa, const CompileContext &cc,
const ReportManager &rm, bool has_literals, const ReportManager &rm, bool has_non_literals,
set<dstate_id_t> &accel_states) { set<dstate_id_t> &accel_states) {
// If we determinised literals, then we only need to consider the init // If we determinised only literals, then we only need to consider the init
// states for acceleration. // states for acceleration.
bool only_accel_init = has_literals; bool only_accel_init = !has_non_literals;
bool trust_daddy_states = !has_non_literals;
bytecode_ptr<NFA> dfa = nullptr; bytecode_ptr<NFA> dfa = nullptr;
if (cc.grey.allowSmallWriteSheng) { if (cc.grey.allowSmallWriteSheng) {
dfa = shengCompile(rdfa, cc, rm, only_accel_init, &accel_states); dfa = shengCompile(rdfa, cc, rm, only_accel_init, &accel_states);
} }
if (!dfa) { if (!dfa) {
dfa = mcclellanCompile(rdfa, cc, rm, only_accel_init, &accel_states); dfa = mcclellanCompile(rdfa, cc, rm, only_accel_init,
trust_daddy_states, &accel_states);
} }
return dfa; return dfa;
} }
@ -711,14 +713,14 @@ bytecode_ptr<NFA> getDfa(raw_dfa &rdfa, const CompileContext &cc,
static static
bytecode_ptr<NFA> prepEngine(raw_dfa &rdfa, u32 roseQuality, bytecode_ptr<NFA> prepEngine(raw_dfa &rdfa, u32 roseQuality,
const CompileContext &cc, const ReportManager &rm, const CompileContext &cc, const ReportManager &rm,
bool has_literals, u32 *start_offset, bool has_non_literals, u32 *start_offset,
u32 *small_region) { u32 *small_region) {
*start_offset = remove_leading_dots(rdfa); *start_offset = remove_leading_dots(rdfa);
// Unleash the McClellan! // Unleash the McClellan!
set<dstate_id_t> accel_states; set<dstate_id_t> accel_states;
auto nfa = getDfa(rdfa, cc, rm, has_literals, accel_states); auto nfa = getDfa(rdfa, cc, rm, has_non_literals, accel_states);
if (!nfa) { if (!nfa) {
DEBUG_PRINTF("DFA compile failed for smallwrite NFA\n"); DEBUG_PRINTF("DFA compile failed for smallwrite NFA\n");
return nullptr; return nullptr;
@ -737,7 +739,7 @@ bytecode_ptr<NFA> prepEngine(raw_dfa &rdfa, u32 roseQuality,
return nullptr; return nullptr;
} }
nfa = getDfa(rdfa, cc, rm, has_literals, accel_states); nfa = getDfa(rdfa, cc, rm, has_non_literals, accel_states);
if (!nfa) { if (!nfa) {
DEBUG_PRINTF("DFA compile failed for smallwrite NFA\n"); DEBUG_PRINTF("DFA compile failed for smallwrite NFA\n");
assert(0); /* able to build orig dfa but not the trimmed? */ assert(0); /* able to build orig dfa but not the trimmed? */
@ -768,6 +770,7 @@ unique_ptr<SmallWriteBuild> makeSmallWriteBuilder(size_t num_patterns,
bytecode_ptr<SmallWriteEngine> SmallWriteBuildImpl::build(u32 roseQuality) { bytecode_ptr<SmallWriteEngine> SmallWriteBuildImpl::build(u32 roseQuality) {
const bool has_literals = !is_empty(lit_trie) || !is_empty(lit_trie_nocase); const bool has_literals = !is_empty(lit_trie) || !is_empty(lit_trie_nocase);
const bool has_non_literals = rdfa != nullptr;
if (!rdfa && !has_literals) { if (!rdfa && !has_literals) {
DEBUG_PRINTF("no smallwrite engine\n"); DEBUG_PRINTF("no smallwrite engine\n");
poisoned = true; poisoned = true;
@ -788,7 +791,7 @@ bytecode_ptr<SmallWriteEngine> SmallWriteBuildImpl::build(u32 roseQuality) {
u32 start_offset; u32 start_offset;
u32 small_region; u32 small_region;
auto nfa = prepEngine(*rdfa, roseQuality, cc, rm, has_literals, auto nfa = prepEngine(*rdfa, roseQuality, cc, rm, has_non_literals,
&start_offset, &small_region); &start_offset, &small_region);
if (!nfa) { if (!nfa) {
DEBUG_PRINTF("some smallwrite outfix could not be prepped\n"); DEBUG_PRINTF("some smallwrite outfix could not be prepped\n");