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

smallwrite: use failure map to set dfa daddy state

Browse Source
This commit is contained in:
Justin Viiret 2017-04-11 13:37:56 +10:00 committed by Matthew Barr
parent beac58fcb4
commit 3b3f6b739c
2 changed files with 15 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/nfa/mcclellancompile.cpp
View File

@ -842,6 +842,8 @@ void find_better_daddy(dfa_info &info, dstate_id_t curr_id, bool using8bit,
flat_set<dstate_id_t> hinted; flat_set<dstate_id_t> hinted;
if (trust_daddy_states) { if (trust_daddy_states) {
hinted.insert(currState.daddy); hinted.insert(currState.daddy);
addIfEarlier(hinted, info.raw.start_floating, curr_id);
addIfEarlier(hinted, info.raw.start_anchored, curr_id);
} else { } else {
hinted = find_daddy_candidates(info, curr_id); hinted = find_daddy_candidates(info, curr_id);
} }
@ -896,7 +898,7 @@ void find_better_daddy(dfa_info &info, dstate_id_t curr_id, bool using8bit,
if (self_loop_width > MAX_SHERMAN_SELF_LOOP) { if (self_loop_width > MAX_SHERMAN_SELF_LOOP) {
DEBUG_PRINTF("%hu is banned wide self loop (%u)\n", curr_id, DEBUG_PRINTF("%hu is banned wide self loop (%u)\n", curr_id,
self_loop_width); self_loop_width);
return; return;
} }

17
src/smallwrite/smallwrite_build.cpp
View File

@ -443,11 +443,11 @@ bool isSaneTrie(const LitTrie &trie) {
* edges and reports. * edges and reports.
*/ */
static static
void buildAutomaton(LitTrie &trie) { void buildAutomaton(LitTrie &trie,
map<LitTrieVertex, LitTrieVertex> &failure_map) {
assert(isSaneTrie(trie)); assert(isSaneTrie(trie));
// Find our failure transitions and reports. // Find our failure transitions and reports.
map<LitTrieVertex, LitTrieVertex> failure_map;
vector<LitTrieVertex> ordering; vector<LitTrieVertex> ordering;
ACVisitor ac_vis(trie, failure_map, ordering); ACVisitor ac_vis(trie, failure_map, ordering);
boost::breadth_first_search(trie, trie.root, visitor(ac_vis)); boost::breadth_first_search(trie, trie.root, visitor(ac_vis));
@ -535,7 +535,8 @@ static
unique_ptr<raw_dfa> buildDfa(LitTrie &trie, bool nocase) { unique_ptr<raw_dfa> buildDfa(LitTrie &trie, bool nocase) {
DEBUG_PRINTF("trie has %zu states\n", num_vertices(trie)); DEBUG_PRINTF("trie has %zu states\n", num_vertices(trie));
buildAutomaton(trie); map<LitTrieVertex, LitTrieVertex> failure_map;
buildAutomaton(trie, failure_map);
auto rdfa = make_unique<raw_dfa>(NFA_OUTFIX); auto rdfa = make_unique<raw_dfa>(NFA_OUTFIX);
@ -559,13 +560,19 @@ unique_ptr<raw_dfa> buildDfa(LitTrie &trie, bool nocase) {
DEBUG_PRINTF("state %zu\n", u_state); DEBUG_PRINTF("state %zu\n", u_state);
assert(u_state < rdfa->states.size()); assert(u_state < rdfa->states.size());
auto &ds = rdfa->states[u_state]; auto &ds = rdfa->states[u_state];
ds.daddy = root_state;
ds.reports = trie[u].reports; ds.reports = trie[u].reports;
if (!ds.reports.empty()) { if (!ds.reports.empty()) {
DEBUG_PRINTF("reports: %s\n", as_string_list(ds.reports).c_str()); DEBUG_PRINTF("reports: %s\n", as_string_list(ds.reports).c_str());
} }
// Set daddy state from failure map.
if (u == trie.root) {
ds.daddy = DEAD_STATE;
} else {
assert(contains(failure_map, u));
ds.daddy = trie[failure_map.at(u)].index + 1;
}
// By default, transition back to the root. // By default, transition back to the root.
fill(ds.next.begin(), ds.next.end(), root_state); fill(ds.next.begin(), ds.next.end(), root_state);
// TOP should be a self-loop. // TOP should be a self-loop.