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

look for normal accel schemes using compressed alpha

Browse Source
This commit is contained in:
Alex Coyte 2016-03-04 14:20:53 +11:00 committed by Matthew Barr
parent f53c093baa
commit 6898dc9864
2 changed files with 54 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

64
src/nfa/mcclellancompile_accel.cpp
View File

@ -177,6 +177,7 @@ vector<vector<CharReach> > generate_paths(const raw_dfa &rdfa, dstate_id_t base,
return rv; return rv;
} }
static
escape_info look_for_offset_accel(const raw_dfa &rdfa, dstate_id_t base, escape_info look_for_offset_accel(const raw_dfa &rdfa, dstate_id_t base,
u32 max_allowed_accel_offset) { u32 max_allowed_accel_offset) {
DEBUG_PRINTF("looking for accel for %hu\n", base); DEBUG_PRINTF("looking for accel for %hu\n", base);
@ -191,7 +192,6 @@ escape_info look_for_offset_accel(const raw_dfa &rdfa, dstate_id_t base,
return rv; return rv;
} }
static static
vector<u16> find_nonexit_symbols(const raw_dfa &rdfa, vector<u16> find_nonexit_symbols(const raw_dfa &rdfa,
const CharReach &escape) { const CharReach &escape) {
@ -253,6 +253,17 @@ bool better(const escape_info &a, const escape_info &b) {
return a.outs.count() < b.outs.count(); return a.outs.count() < b.outs.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;
}
map<dstate_id_t, escape_info> populateAccelerationInfo(const raw_dfa &rdfa, map<dstate_id_t, escape_info> populateAccelerationInfo(const raw_dfa &rdfa,
const dfa_build_strat &strat, const dfa_build_strat &strat,
const Grey &grey) { const Grey &grey) {
@ -321,45 +332,58 @@ escape_info find_mcclellan_escape_info(const raw_dfa &rdfa,
u32 max_allowed_accel_offset) { u32 max_allowed_accel_offset) {
escape_info rv; escape_info rv;
const dstate &raw = rdfa.states[this_idx]; const dstate &raw = rdfa.states[this_idx];
const auto &alpha_remap = rdfa.alpha_remap; const vector<CharReach> rev_map = reverse_alpha_remapping(rdfa);
flat_set<pair<u8, u8>> outs2_local; for (u32 i = 0; i < rev_map.size(); i++) {
for (unsigned i = 0; i < N_CHARS; i++) { if (raw.next[i] == this_idx) {
outs2_local.clear(); continue;
}
if (raw.next[alpha_remap[i]] != this_idx) { const CharReach &cr_i = rev_map.at(i);
rv.outs.set(i);
DEBUG_PRINTF("next is %hu\n", raw.next[alpha_remap[i]]); rv.outs |= cr_i;
const dstate &raw_next = rdfa.states[raw.next[alpha_remap[i]]];
DEBUG_PRINTF("next is %hu\n", raw.next[i]);
const dstate &raw_next = rdfa.states[raw.next[i]];
if (!raw_next.reports.empty() && generates_callbacks(rdfa.kind)) { if (!raw_next.reports.empty() && generates_callbacks(rdfa.kind)) {
DEBUG_PRINTF("leads to report\n"); DEBUG_PRINTF("leads to report\n");
rv.outs2_broken = true; /* cannot accelerate over reports */ rv.outs2_broken = true; /* cannot accelerate over reports */
} }
for (unsigned j = 0; !rv.outs2_broken && j < N_CHARS; j++) { if (rv.outs2_broken) {
if (raw_next.next[alpha_remap[j]] == raw.next[alpha_remap[j]]) {
continue; continue;
} }
DEBUG_PRINTF("adding %02x %02x -> %hu to 2 \n", i, j, CharReach cr_all_j;
raw_next.next[alpha_remap[j]]); for (u32 j = 0; j < rev_map.size(); j++) {
outs2_local.emplace((u8)i, (u8)j); if (raw_next.next[j] == raw.next[j]) {
continue;
} }
if (outs2_local.size() > 8) { DEBUG_PRINTF("adding sym %u sym %u -> %hu to 2 \n", i, j,
DEBUG_PRINTF("adding %02x to outs2_single\n", i); raw_next.next[j]);
rv.outs2_single.set(i); cr_all_j |= rev_map.at(j);
} else {
insert(&rv.outs2, outs2_local);
} }
if (cr_i.count() * cr_all_j.count() > 8) {
DEBUG_PRINTF("adding sym %u to outs2_single\n", i);
rv.outs2_single |= cr_i;
} else {
for (auto ii = cr_i.find_first(); ii != CharReach::npos;
ii = cr_i.find_next(ii)) {
for (auto jj = cr_all_j.find_first(); jj != CharReach::npos;
jj = cr_all_j.find_next(jj)) {
rv.outs2.emplace((u8)ii, (u8)jj);
}
}
}
if (rv.outs2.size() > 8) { if (rv.outs2.size() > 8) {
DEBUG_PRINTF("outs2 too big\n"); DEBUG_PRINTF("outs2 too big\n");
rv.outs2_broken = true; rv.outs2_broken = true;
} }
} }
}
DEBUG_PRINTF("this %u, sds proxy %hu\n", this_idx, get_sds_or_proxy(rdfa)); DEBUG_PRINTF("this %u, sds proxy %hu\n", this_idx, get_sds_or_proxy(rdfa));
DEBUG_PRINTF("broken %d\n", rv.outs2_broken); DEBUG_PRINTF("broken %d\n", rv.outs2_broken);

3
src/nfa/mcclellancompile_accel.h
View File

@ -48,9 +48,6 @@ struct Grey;
* than normal states as accelerating sds is important. Matches NFA value */ * than normal states as accelerating sds is important. Matches NFA value */
#define ACCEL_DFA_MAX_FLOATING_STOP_CHAR 192 #define ACCEL_DFA_MAX_FLOATING_STOP_CHAR 192
escape_info look_for_offset_accel(const raw_dfa &rdfa, dstate_id_t base,
u32 max_allowed_accel_offset);
std::map<dstate_id_t, escape_info> populateAccelerationInfo(const raw_dfa &rdfa, std::map<dstate_id_t, escape_info> populateAccelerationInfo(const raw_dfa &rdfa,
const dfa_build_strat &strat, const dfa_build_strat &strat,
const Grey &grey); const Grey &grey);