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

getMatches: simplify

Browse Source
This commit is contained in:
Justin Viiret 2017-03-08 16:27:41 +11:00 committed by Matthew Barr
parent c81c30b144
commit 749e3e64b9
1 changed files with 45 additions and 43 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

88
util/ng_find_matches.cpp
View File

@ -868,56 +868,58 @@ bool canReach(const NGHolder &g, const NFAEdge &e, struct fmstate &state) {
} }
static static
void getMatches(const NGHolder &g, MatchSet &matches, struct fmstate &state, void getAcceptMatches(const NGHolder &g, MatchSet &matches,
bool allowEodMatches) { struct fmstate &state, NFAVertex accept_vertex) {
flat_set<NFAVertex> accepts {g.accept, g.acceptEod}; assert(accept_vertex == g.accept || accept_vertex == g.acceptEod);
for (auto v : accepts) { const bool eod = accept_vertex == g.acceptEod;
bool eod = v == g.acceptEod; auto active_states = eod ? state.states.getAcceptEodStates(state.gc)
if (eod && !allowEodMatches) { : state.states.getAcceptStates(state.gc);
DEBUG_PRINTF("Number of active states: %zu\n", active_states.size());
for (const auto &cur : active_states) {
auto u = state.vertices[cur.idx];
// we can't accept anything from startDs in between UTF-8 codepoints
if (state.utf8 && u == g.startDs && !isUtf8CodePoint(state.cur)) {
continue; continue;
} }
auto active_states = eod ? state.states.getAcceptEodStates(state.gc) : const auto &reports =
state.states.getAcceptStates(state.gc); eod ? state.gc.vertex_eod_reports_by_level[cur.level][u]
: state.gc.vertex_reports_by_level[cur.level][u];
DEBUG_PRINTF("Number of active states: %zu\n", active_states.size()); NFAEdge e = edge(u, accept_vertex, g);
for (const auto &cur : active_states) { // we assume edge assertions only exist at level 0
auto u = state.vertices[cur.idx]; if (e && !canReach(g, e, state)) {
continue;
// we can't accept anything from startDs in between UTF-8 codepoints
if (state.utf8 && u == g.startDs && !isUtf8CodePoint(state.cur)) {
continue;
}
const auto &reports =
eod ?
state.gc.vertex_eod_reports_by_level[cur.level][u] :
state.gc.vertex_reports_by_level[cur.level][u];
NFAEdge e = edge(u, v, g);
// we assume edge assertions only exist at level 0
if (e && !canReach(g, e, state)) {
continue;
}
DEBUG_PRINTF("%smatch found at %zu\n",
eod ? "eod " : "", state.offset);
assert(!reports.empty());
for (const auto &report_id : reports) {
const Report &ri = state.rm.getReport(report_id);
DEBUG_PRINTF("report %u has offset adjustment %d\n",
report_id, ri.offsetAdjust);
DEBUG_PRINTF("match from (i:%zu,l:%u,t:%u): (%zu,%zu)\n",
cur.idx, cur.level, cur.type, cur.som,
state.offset + ri.offsetAdjust);
matches.emplace(cur.som, state.offset + ri.offsetAdjust);
}
} }
DEBUG_PRINTF("%smatch found at %zu\n", eod ? "eod " : "", state.offset);
assert(!reports.empty());
for (const auto &report_id : reports) {
const Report &ri = state.rm.getReport(report_id);
DEBUG_PRINTF("report %u has offset adjustment %d\n", report_id,
ri.offsetAdjust);
DEBUG_PRINTF("match from (i:%zu,l:%u,t:%u): (%zu,%zu)\n", cur.idx,
cur.level, cur.type, cur.som,
state.offset + ri.offsetAdjust);
matches.emplace(cur.som, state.offset + ri.offsetAdjust);
}
}
}
static
void getMatches(const NGHolder &g, MatchSet &matches, struct fmstate &state,
bool allowEodMatches) {
getAcceptMatches(g, matches, state, g.accept);
if (allowEodMatches) {
getAcceptMatches(g, matches, state, g.acceptEod);
} }
} }