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

ng_find_matches: speed up edge lookups

Browse Source 
Improves the performance of step() on graphs with vertices with large
degree.
This commit is contained in:
Justin Viiret 2017-07-12 11:08:45 +10:00 committed by Matthew Barr
parent 815be3fa2b
commit bc232d272f
1 changed files with 25 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
util/ng_find_matches.cpp
View File

@ -752,12 +752,34 @@ bool operator==(const StateSet::State &a, const StateSet::State &b) {
a.som == b.som; a.som == b.som;
} }
/** \brief Cache to speed up edge lookups, rather than hitting the graph. */
struct EdgeCache {
explicit EdgeCache(const NGHolder &g) {
cache.reserve(num_vertices(g));
for (auto e : edges_range(g)) {
cache.emplace(make_pair(source(e, g), target(e, g)), e);
}
}
NFAEdge get(NFAVertex u, NFAVertex v) const {
auto it = cache.find(make_pair(u, v));
if (it != cache.end()) {
return it->second;
}
return NFAEdge();
}
private:
unordered_map<pair<NFAVertex, NFAVertex>, NFAEdge> cache;
};
struct fmstate { struct fmstate {
const size_t num_states; // number of vertices in graph const size_t num_states; // number of vertices in graph
StateSet states; // currently active states StateSet states; // currently active states
StateSet next; // states on after this iteration StateSet next; // states on after this iteration
GraphCache &gc; GraphCache &gc;
vector<NFAVertex> vertices; // mapping from index to vertex vector<NFAVertex> vertices; // mapping from index to vertex
EdgeCache edge_cache;
size_t offset = 0; size_t offset = 0;
unsigned char cur = 0; unsigned char cur = 0;
unsigned char prev = 0; unsigned char prev = 0;
@ -771,7 +793,7 @@ struct fmstate {
states(num_states, edit_distance), states(num_states, edit_distance),
next(num_states, edit_distance), next(num_states, edit_distance),
gc(gc_in), vertices(num_vertices(g), NGHolder::null_vertex()), gc(gc_in), vertices(num_vertices(g), NGHolder::null_vertex()),
utf8(utf8_in), allowStartDs(aSD_in), rm(rm_in) { edge_cache(g), utf8(utf8_in), allowStartDs(aSD_in), rm(rm_in) {
// init states // init states
states.activateState( states.activateState(
StateSet::State {g[g.start].index, 0, 0, StateSet::State {g[g.start].index, 0, 0,
@ -889,7 +911,7 @@ void getAcceptMatches(const NGHolder &g, MatchSet &matches,
eod ? state.gc.vertex_eod_reports_by_level[cur.level][u] eod ? state.gc.vertex_eod_reports_by_level[cur.level][u]
: state.gc.vertex_reports_by_level[cur.level][u]; : state.gc.vertex_reports_by_level[cur.level][u];
NFAEdge e = edge(u, accept_vertex, g); NFAEdge e = state.edge_cache.get(u, accept_vertex);
// we assume edge assertions only exist at level 0 // we assume edge assertions only exist at level 0
if (e && !canReach(g, e, state)) { if (e && !canReach(g, e, state)) {
@ -965,7 +987,7 @@ void step(const NGHolder &g, fmstate &state, StateSet::WorkingData &wd) {
} else { } else {
// we assume edge assertions only exist on level 0 // we assume edge assertions only exist on level 0
const CharReach &cr = g[v].char_reach; const CharReach &cr = g[v].char_reach;
NFAEdge e = edge(u, v, g); NFAEdge e = state.edge_cache.get(u, v);
if (cr.test(state.cur) && if (cr.test(state.cur) &&
(!e || canReach(g, e, state))) { (!e || canReach(g, e, state))) {