github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-11-17 01:41:51 +03:00
Code Issues Packages Projects Releases Wiki Activity

McClellan refactor

Browse Source 
Taking dfa strat out of McClellan, to be reused by other DFAs
This commit is contained in:
Anatoly Burakov
2016-05-19 14:16:35 +01:00
committed by Matthew Barr
parent a9fddbc400
commit 56bf25b091
9 changed files with 624 additions and 380 deletions
Download Patch File Download Diff File Expand all files Collapse all files

59
src/nfa/mcclellancompile_util.cpp
View File

@@ -336,65 +336,6 @@ size_t hash_dfa(const raw_dfa &rdfa) {
return v;
}
static
bool has_self_loop(dstate_id_t s, const raw_dfa &raw) {
u16 top_remap = raw.alpha_remap[TOP];
for (u32 i = 0; i < raw.states[s].next.size(); i++) {
if (i != top_remap && raw.states[s].next[i] == s) {
return true;
}
}
return false;
}
dstate_id_t get_sds_or_proxy(const raw_dfa &raw) {
if (raw.start_floating != DEAD_STATE) {
DEBUG_PRINTF("has floating start\n");
return raw.start_floating;
}
DEBUG_PRINTF("looking for SDS proxy\n");
dstate_id_t s = raw.start_anchored;
if (has_self_loop(s, raw)) {
return s;
}
u16 top_remap = raw.alpha_remap[TOP];
ue2::unordered_set<dstate_id_t> seen;
while (true) {
seen.insert(s);
DEBUG_PRINTF("basis %hu\n", s);
/* check if we are connected to a state with a self loop */
for (u32 i = 0; i < raw.states[s].next.size(); i++) {
dstate_id_t t = raw.states[s].next[i];
if (i != top_remap && t != DEAD_STATE && has_self_loop(t, raw)) {
return t;
}
}
/* find a neighbour to use as a basis for looking for the sds proxy */
dstate_id_t t = DEAD_STATE;
for (u32 i = 0; i < raw.states[s].next.size(); i++) {
dstate_id_t tt = raw.states[s].next[i];
if (i != top_remap && tt != DEAD_STATE && !contains(seen, tt)) {
t = tt;
break;
}
}
if (t == DEAD_STATE) {
/* we were unable to find a state to use as a SDS proxy */
return DEAD_STATE;
}
s = t;
}
}
static
bool can_die_early(const raw_dfa &raw, dstate_id_t s,
map<dstate_id_t, u32> &visited, u32 age_limit) {