| 1 | /* |
| 2 | * Copyright (c) 2015-2017, Intel Corporation |
| 3 | * |
| 4 | * Redistribution and use in source and binary forms, with or without |
| 5 | * modification, are permitted provided that the following conditions are met: |
| 6 | * |
| 7 | * * Redistributions of source code must retain the above copyright notice, |
| 8 | * this list of conditions and the following disclaimer. |
| 9 | * * Redistributions in binary form must reproduce the above copyright |
| 10 | * notice, this list of conditions and the following disclaimer in the |
| 11 | * documentation and/or other materials provided with the distribution. |
| 12 | * * Neither the name of Intel Corporation nor the names of its contributors |
| 13 | * may be used to endorse or promote products derived from this software |
| 14 | * without specific prior written permission. |
| 15 | * |
| 16 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
| 17 | * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 18 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 19 | * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE |
| 20 | * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
| 21 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
| 22 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
| 23 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
| 24 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
| 25 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
| 26 | * POSSIBILITY OF SUCH DAMAGE. |
| 27 | */ |
| 28 | |
| 29 | #include "mcclellancompile_util.h" |
| 30 | |
| 31 | #include "rdfa.h" |
| 32 | #include "util/container.h" |
| 33 | #include "util/hash.h" |
| 34 | #include "ue2common.h" |
| 35 | |
| 36 | #include <deque> |
| 37 | #include <map> |
| 38 | |
| 39 | using namespace std; |
| 40 | |
| 41 | namespace ue2 { |
| 42 | |
| 43 | #define INIT_STATE 1 |
| 44 | |
| 45 | static |
| 46 | bool state_has_reports(const raw_dfa &raw, dstate_id_t s) { |
| 47 | const auto &ds = raw.states[s]; |
| 48 | return !ds.reports.empty() || !ds.reports_eod.empty(); |
| 49 | } |
| 50 | |
| 51 | static |
| 52 | u32 count_dots(const raw_dfa &raw) { |
| 53 | assert(raw.start_anchored == INIT_STATE); |
| 54 | |
| 55 | u32 i = INIT_STATE; |
| 56 | for (; i < raw.states.size() && i != raw.start_floating; i++) { |
| 57 | DEBUG_PRINTF("checking %u\n" , i); |
| 58 | assert(raw.states[i].reports.empty()); |
| 59 | assert(raw.states[i].reports_eod.empty()); |
| 60 | |
| 61 | for (symbol_t s = 0; s < raw.getImplAlphaSize(); s++) { |
| 62 | DEBUG_PRINTF("%hu -> %hu\n" , s, raw.states[i].next[s]); |
| 63 | if (raw.states[i].next[s] != i + 1) { |
| 64 | goto validate; |
| 65 | } |
| 66 | } |
| 67 | |
| 68 | if (state_has_reports(raw, raw.states[i].next[0])) { |
| 69 | goto validate; |
| 70 | } |
| 71 | |
| 72 | DEBUG_PRINTF("got dot\n" ); |
| 73 | } |
| 74 | |
| 75 | validate: |
| 76 | u32 dot_count = i - INIT_STATE; |
| 77 | |
| 78 | /* we need to check that no later state has a transition into these leading |
| 79 | * dots */ |
| 80 | for (; i < raw.states.size(); i++) { |
| 81 | for (symbol_t s = 0; s < raw.getImplAlphaSize(); s++) { |
| 82 | DEBUG_PRINTF("%hu -> %hu\n" , s, raw.states[i].next[s]); |
| 83 | dstate_id_t n = raw.states[i].next[s]; |
| 84 | if (n != DEAD_STATE && n <= dot_count) { |
| 85 | return 0; |
| 86 | } |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | return dot_count; |
| 91 | } |
| 92 | |
| 93 | static |
| 94 | void prune_leading_states(raw_dfa &raw, u32 count) { |
| 95 | if (!count) { |
| 96 | return; |
| 97 | } |
| 98 | |
| 99 | for (u32 i = INIT_STATE + count; i < raw.states.size(); i++) { |
| 100 | dstate &curr = raw.states[i - count]; |
| 101 | curr = raw.states[i]; |
| 102 | if (curr.daddy > count) { |
| 103 | curr.daddy -= count; |
| 104 | } else { |
| 105 | curr.daddy = DEAD_STATE; |
| 106 | } |
| 107 | |
| 108 | for (u32 j = 0; j < raw.alpha_size; j++) { |
| 109 | assert(curr.next[j] == DEAD_STATE || curr.next[j] > count); |
| 110 | if (curr.next[j]) { |
| 111 | curr.next[j] -= count; |
| 112 | } |
| 113 | } |
| 114 | } |
| 115 | |
| 116 | raw.states.erase(raw.states.end() - count, raw.states.end()); |
| 117 | } |
| 118 | |
| 119 | u32 remove_leading_dots(raw_dfa &raw) { |
| 120 | u32 count = count_dots(raw); |
| 121 | prune_leading_states(raw, count); |
| 122 | DEBUG_PRINTF("removed %u leading dots\n" , count); |
| 123 | return count; |
| 124 | } |
| 125 | |
| 126 | static never_inline |
| 127 | u32 calc_min_dist_from_bob(raw_dfa &raw, vector<u32> *dist_in) { |
| 128 | vector<u32> &dist = *dist_in; |
| 129 | dist.assign(raw.states.size(), ~0U); |
| 130 | |
| 131 | assert(raw.start_anchored != DEAD_STATE); |
| 132 | |
| 133 | deque<dstate_id_t> to_visit = { raw.start_anchored }; |
| 134 | dist[raw.start_anchored] = 0; |
| 135 | |
| 136 | u32 last_d = 0; |
| 137 | |
| 138 | while (!to_visit.empty()) { |
| 139 | dstate_id_t s = to_visit.front(); |
| 140 | DEBUG_PRINTF("inspecting %u\n" , s); |
| 141 | to_visit.pop_front(); |
| 142 | assert(s != DEAD_STATE); |
| 143 | |
| 144 | u32 d = dist[s]; |
| 145 | assert(d >= last_d); |
| 146 | assert(d != ~0U); |
| 147 | |
| 148 | for (dstate_id_t t : raw.states[s].next) { |
| 149 | if (t == DEAD_STATE) { |
| 150 | continue; |
| 151 | } |
| 152 | if (dist[t] == ~0U) { |
| 153 | to_visit.push_back(t); |
| 154 | dist[t] = d + 1; |
| 155 | } else { |
| 156 | assert(dist[t] <= d + 1); |
| 157 | } |
| 158 | } |
| 159 | |
| 160 | last_d = d; |
| 161 | } |
| 162 | |
| 163 | return last_d; |
| 164 | } |
| 165 | |
| 166 | bool clear_deeper_reports(raw_dfa &raw, u32 max_offset) { |
| 167 | DEBUG_PRINTF("clearing reports on states deeper than %u\n" , max_offset); |
| 168 | vector<u32> bob_dist; |
| 169 | u32 max_min_dist_bob = calc_min_dist_from_bob(raw, &bob_dist); |
| 170 | |
| 171 | if (max_min_dist_bob <= max_offset) { |
| 172 | return false; |
| 173 | } |
| 174 | |
| 175 | bool changed = false; |
| 176 | for (u32 s = DEAD_STATE + 1; s < raw.states.size(); s++) { |
| 177 | if (bob_dist[s] > max_offset && state_has_reports(raw, s)) { |
| 178 | DEBUG_PRINTF("clearing reports on %u (depth %u)\n" , s, bob_dist[s]); |
| 179 | auto &ds = raw.states[s]; |
| 180 | ds.reports.clear(); |
| 181 | ds.reports_eod.clear(); |
| 182 | changed = true; |
| 183 | } |
| 184 | } |
| 185 | |
| 186 | if (!changed) { |
| 187 | return false; |
| 188 | } |
| 189 | |
| 190 | // We may have cleared all reports from the DFA, in which case it should |
| 191 | // become empty. |
| 192 | if (all_of_in(raw.states, [](const dstate &ds) { |
| 193 | return ds.reports.empty() && ds.reports_eod.empty(); |
| 194 | })) { |
| 195 | DEBUG_PRINTF("no reports left at all, dfa is dead\n" ); |
| 196 | raw.start_anchored = DEAD_STATE; |
| 197 | raw.start_floating = DEAD_STATE; |
| 198 | } |
| 199 | |
| 200 | return true; |
| 201 | } |
| 202 | |
| 203 | set<ReportID> all_reports(const raw_dfa &rdfa) { |
| 204 | set<ReportID> all; |
| 205 | for (const auto &ds : rdfa.states) { |
| 206 | insert(&all, ds.reports); |
| 207 | insert(&all, ds.reports_eod); |
| 208 | } |
| 209 | return all; |
| 210 | } |
| 211 | |
| 212 | bool has_eod_accepts(const raw_dfa &rdfa) { |
| 213 | for (const auto &ds : rdfa.states) { |
| 214 | if (!ds.reports_eod.empty()) { |
| 215 | return true; |
| 216 | } |
| 217 | } |
| 218 | return false; |
| 219 | } |
| 220 | |
| 221 | bool has_non_eod_accepts(const raw_dfa &rdfa) { |
| 222 | for (const auto &ds : rdfa.states) { |
| 223 | if (!ds.reports.empty()) { |
| 224 | return true; |
| 225 | } |
| 226 | } |
| 227 | return false; |
| 228 | } |
| 229 | |
| 230 | size_t hash_dfa_no_reports(const raw_dfa &rdfa) { |
| 231 | size_t v = 0; |
| 232 | hash_combine(v, rdfa.alpha_size); |
| 233 | hash_combine(v, rdfa.alpha_remap); |
| 234 | |
| 235 | for (const auto &ds : rdfa.states) { |
| 236 | hash_combine(v, ds.next); |
| 237 | } |
| 238 | |
| 239 | return v; |
| 240 | } |
| 241 | |
| 242 | size_t hash_dfa(const raw_dfa &rdfa) { |
| 243 | size_t v = 0; |
| 244 | hash_combine(v, hash_dfa_no_reports(rdfa)); |
| 245 | hash_combine(v, all_reports(rdfa)); |
| 246 | return v; |
| 247 | } |
| 248 | |
| 249 | static |
| 250 | bool can_die_early(const raw_dfa &raw, dstate_id_t s, |
| 251 | map<dstate_id_t, u32> &visited, u32 age_limit) { |
| 252 | if (contains(visited, s) && visited[s] >= age_limit) { |
| 253 | /* we have already visited (or are in the process of visiting) here with |
| 254 | * a looser limit. */ |
| 255 | return false; |
| 256 | } |
| 257 | visited[s] = age_limit; |
| 258 | |
| 259 | if (s == DEAD_STATE) { |
| 260 | return true; |
| 261 | } |
| 262 | |
| 263 | if (age_limit == 0) { |
| 264 | return false; |
| 265 | } |
| 266 | |
| 267 | for (const auto &next : raw.states[s].next) { |
| 268 | if (can_die_early(raw, next, visited, age_limit - 1)) { |
| 269 | return true; |
| 270 | } |
| 271 | } |
| 272 | |
| 273 | return false; |
| 274 | } |
| 275 | |
| 276 | bool can_die_early(const raw_dfa &raw, u32 age_limit) { |
| 277 | map<dstate_id_t, u32> visited; |
| 278 | return can_die_early(raw, raw.start_anchored, visited, age_limit); |
| 279 | } |
| 280 | |
| 281 | bool is_dead(const raw_dfa &rdfa) { |
| 282 | return rdfa.start_anchored == DEAD_STATE && |
| 283 | rdfa.start_floating == DEAD_STATE; |
| 284 | } |
| 285 | |
| 286 | } // namespace ue2 |
| 287 | |