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

accel_dfa_build_strat: make extend() faster

Browse Source
This commit is contained in:
Justin Viiret 2017-08-11 13:29:07 +10:00 committed by Matthew Barr
parent 3f36665e39
commit 3b392d6b70
1 changed files with 22 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

41
src/nfa/accel_dfa_build_strat.cpp
View File

@ -41,6 +41,7 @@
#include "util/verify_types.h" #include "util/verify_types.h"
#include <sstream> #include <sstream>
#include <unordered_map>
#include <unordered_set> #include <unordered_set>
#include <vector> #include <vector>
@ -66,6 +67,17 @@ void dump_paths(const Container &paths) {
DEBUG_PRINTF("%zu paths\n", paths.size()); DEBUG_PRINTF("%zu paths\n", paths.size());
} }
static
vector<CharReach> reverse_alpha_remapping(const raw_dfa &rdfa) {
vector<CharReach> rv(rdfa.alpha_size - 1); /* TOP not required */
for (u32 i = 0; i < N_CHARS; i++) {
rv.at(rdfa.alpha_remap[i]).set(i);
}
return rv;
}
static static
bool is_useful_path(const vector<path> &good, const path &p) { bool is_useful_path(const vector<path> &good, const path &p) {
for (const auto &g : good) { for (const auto &g : good) {
@ -99,9 +111,10 @@ path append(const path &orig, const CharReach &cr, u32 new_dest) {
} }
static static
void extend(const raw_dfa &rdfa, const path &p, void extend(const raw_dfa &rdfa, const vector<CharReach> &rev_map,
map<u32, vector<path>> &all, vector<path> &out) { const path &p, unordered_map<u32, vector<path>> &all,
dstate s = rdfa.states[p.dest]; vector<path> &out) {
const dstate &s = rdfa.states[p.dest];
if (!p.reach.empty() && p.reach.back().none()) { if (!p.reach.empty() && p.reach.back().none()) {
out.push_back(p); out.push_back(p);
@ -126,9 +139,9 @@ void extend(const raw_dfa &rdfa, const path &p,
} }
flat_map<u32, CharReach> dest; flat_map<u32, CharReach> dest;
for (unsigned i = 0; i < N_CHARS; i++) { for (u32 i = 0; i < rev_map.size(); i++) {
u32 succ = s.next[rdfa.alpha_remap[i]]; u32 succ = s.next[i];
dest[succ].set(i); dest[succ] |= rev_map[i];
} }
for (const auto &e : dest) { for (const auto &e : dest) {
@ -149,13 +162,14 @@ void extend(const raw_dfa &rdfa, const path &p,
static static
vector<vector<CharReach>> generate_paths(const raw_dfa &rdfa, vector<vector<CharReach>> generate_paths(const raw_dfa &rdfa,
dstate_id_t base, u32 len) { dstate_id_t base, u32 len) {
const vector<CharReach> rev_map = reverse_alpha_remapping(rdfa);
vector<path> paths{path(base)}; vector<path> paths{path(base)};
map<u32, vector<path>> all; unordered_map<u32, vector<path>> all;
all[base].push_back(path(base)); all[base].push_back(path(base));
for (u32 i = 0; i < len && paths.size() < PATHS_LIMIT; i++) { for (u32 i = 0; i < len && paths.size() < PATHS_LIMIT; i++) {
vector<path> next_gen; vector<path> next_gen;
for (const auto &p : paths) { for (const auto &p : paths) {
extend(rdfa, p, all, next_gen); extend(rdfa, rev_map, p, all, next_gen);
} }
paths = move(next_gen); paths = move(next_gen);
@ -196,17 +210,6 @@ bool better(const AccelScheme &a, const AccelScheme &b) {
return a.cr.count() < b.cr.count(); return a.cr.count() < b.cr.count();
} }
static
vector<CharReach> reverse_alpha_remapping(const raw_dfa &rdfa) {
vector<CharReach> rv(rdfa.alpha_size - 1); /* TOP not required */
for (u32 i = 0; i < N_CHARS; i++) {
rv.at(rdfa.alpha_remap[i]).set(i);
}
return rv;
}
static static
bool double_byte_ok(const AccelScheme &info) { bool double_byte_ok(const AccelScheme &info) {
return !info.double_byte.empty() && return !info.double_byte.empty() &&