1 |
|
/****************************************************************************** |
2 |
|
* Top contributors (to current version): |
3 |
|
* Andrew Reynolds, Andres Noetzli, Tianyi Liang |
4 |
|
* |
5 |
|
* This file is part of the cvc5 project. |
6 |
|
* |
7 |
|
* Copyright (c) 2009-2021 by the authors listed in the file AUTHORS |
8 |
|
* in the top-level source directory and their institutional affiliations. |
9 |
|
* All rights reserved. See the file COPYING in the top-level source |
10 |
|
* directory for licensing information. |
11 |
|
* **************************************************************************** |
12 |
|
* |
13 |
|
* Implementation of the theory of strings. |
14 |
|
*/ |
15 |
|
|
16 |
|
#include "theory/strings/sequences_rewriter.h" |
17 |
|
|
18 |
|
#include "expr/attribute.h" |
19 |
|
#include "expr/node_builder.h" |
20 |
|
#include "expr/sequence.h" |
21 |
|
#include "theory/rewriter.h" |
22 |
|
#include "theory/strings/arith_entail.h" |
23 |
|
#include "theory/strings/regexp_entail.h" |
24 |
|
#include "theory/strings/skolem_cache.h" |
25 |
|
#include "theory/strings/strings_rewriter.h" |
26 |
|
#include "theory/strings/theory_strings_utils.h" |
27 |
|
#include "theory/strings/word.h" |
28 |
|
#include "util/rational.h" |
29 |
|
#include "util/regexp.h" |
30 |
|
#include "util/statistics_registry.h" |
31 |
|
#include "util/string.h" |
32 |
|
|
33 |
|
using namespace std; |
34 |
|
using namespace cvc5::kind; |
35 |
|
|
36 |
|
namespace cvc5 { |
37 |
|
namespace theory { |
38 |
|
namespace strings { |
39 |
|
|
40 |
10030 |
SequencesRewriter::SequencesRewriter(HistogramStat<Rewrite>* statistics) |
41 |
10030 |
: d_statistics(statistics), d_stringsEntail(*this) |
42 |
|
{ |
43 |
10030 |
} |
44 |
|
|
45 |
119568 |
Node SequencesRewriter::rewriteEquality(Node node) |
46 |
|
{ |
47 |
119568 |
Assert(node.getKind() == kind::EQUAL); |
48 |
119568 |
if (node[0] == node[1]) |
49 |
|
{ |
50 |
8318 |
Node ret = NodeManager::currentNM()->mkConst(true); |
51 |
4159 |
return returnRewrite(node, ret, Rewrite::EQ_REFL); |
52 |
|
} |
53 |
115409 |
else if (node[0].isConst() && node[1].isConst()) |
54 |
|
{ |
55 |
4362 |
Node ret = NodeManager::currentNM()->mkConst(false); |
56 |
2181 |
return returnRewrite(node, ret, Rewrite::EQ_CONST_FALSE); |
57 |
|
} |
58 |
|
|
59 |
|
// ( ~contains( s, t ) V ~contains( t, s ) ) => ( s == t ---> false ) |
60 |
338998 |
for (unsigned r = 0; r < 2; r++) |
61 |
|
{ |
62 |
|
// must call rewrite contains directly to avoid infinite loop |
63 |
|
// we do a fix point since we may rewrite contains terms to simpler |
64 |
|
// contains terms. |
65 |
452055 |
Node ctn = d_stringsEntail.checkContains(node[r], node[1 - r], false); |
66 |
226285 |
if (!ctn.isNull()) |
67 |
|
{ |
68 |
36326 |
if (!ctn.getConst<bool>()) |
69 |
|
{ |
70 |
515 |
return returnRewrite(node, ctn, Rewrite::EQ_NCTN); |
71 |
|
} |
72 |
|
else |
73 |
|
{ |
74 |
|
// definitely contains but not syntactically equal |
75 |
|
// We may be able to simplify, e.g. |
76 |
|
// str.++( x, "a" ) == "a" ----> x = "" |
77 |
|
} |
78 |
|
} |
79 |
|
} |
80 |
|
|
81 |
|
// ( len( s ) != len( t ) ) => ( s == t ---> false ) |
82 |
|
// This covers cases like str.++( x, x ) == "a" ---> false |
83 |
225426 |
Node len0 = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, node[0]); |
84 |
225426 |
Node len1 = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, node[1]); |
85 |
225426 |
Node len_eq = len0.eqNode(len1); |
86 |
112713 |
len_eq = Rewriter::rewrite(len_eq); |
87 |
112713 |
if (len_eq.isConst() && !len_eq.getConst<bool>()) |
88 |
|
{ |
89 |
5 |
return returnRewrite(node, len_eq, Rewrite::EQ_LEN_DEQ); |
90 |
|
} |
91 |
|
|
92 |
225416 |
std::vector<Node> c[2]; |
93 |
338124 |
for (unsigned i = 0; i < 2; i++) |
94 |
|
{ |
95 |
225416 |
utils::getConcat(node[i], c[i]); |
96 |
|
} |
97 |
|
|
98 |
|
// check if the prefix, suffix mismatches |
99 |
|
// For example, str.++( x, "a", y ) == str.++( x, "bc", z ) ---> false |
100 |
112708 |
unsigned minsize = std::min(c[0].size(), c[1].size()); |
101 |
338051 |
for (unsigned r = 0; r < 2; r++) |
102 |
|
{ |
103 |
238089 |
for (unsigned i = 0; i < minsize; i++) |
104 |
|
{ |
105 |
235856 |
unsigned index1 = r == 0 ? i : (c[0].size() - 1) - i; |
106 |
235856 |
unsigned index2 = r == 0 ? i : (c[1].size() - 1) - i; |
107 |
248556 |
Node s = c[0][index1]; |
108 |
248556 |
Node t = c[1][index2]; |
109 |
235856 |
if (s.isConst() && t.isConst()) |
110 |
|
{ |
111 |
5858 |
size_t lenS = Word::getLength(s); |
112 |
5858 |
size_t lenT = Word::getLength(t); |
113 |
5858 |
size_t lenShort = lenS <= lenT ? lenS : lenT; |
114 |
15600 |
bool isSameFix = r == 1 ? Word::rstrncmp(s, t, lenShort) |
115 |
9742 |
: Word::strncmp(s, t, lenShort); |
116 |
5858 |
if (!isSameFix) |
117 |
|
{ |
118 |
92 |
Node ret = NodeManager::currentNM()->mkConst(false); |
119 |
46 |
return returnRewrite(node, ret, Rewrite::EQ_NFIX); |
120 |
|
} |
121 |
|
} |
122 |
235810 |
if (s != t) |
123 |
|
{ |
124 |
223110 |
break; |
125 |
|
} |
126 |
|
} |
127 |
|
} |
128 |
|
|
129 |
|
// standard ordering |
130 |
112662 |
if (node[0] > node[1]) |
131 |
|
{ |
132 |
49524 |
Node ret = NodeManager::currentNM()->mkNode(kind::EQUAL, node[1], node[0]); |
133 |
24762 |
return returnRewrite(node, ret, Rewrite::EQ_SYM); |
134 |
|
} |
135 |
87900 |
return node; |
136 |
|
} |
137 |
|
|
138 |
113156 |
Node SequencesRewriter::rewriteEqualityExt(Node node) |
139 |
|
{ |
140 |
113156 |
Assert(node.getKind() == EQUAL); |
141 |
113156 |
if (node[0].getType().isInteger()) |
142 |
|
{ |
143 |
20770 |
return rewriteArithEqualityExt(node); |
144 |
|
} |
145 |
92386 |
if (node[0].getType().isStringLike()) |
146 |
|
{ |
147 |
92386 |
return rewriteStrEqualityExt(node); |
148 |
|
} |
149 |
|
return node; |
150 |
|
} |
151 |
|
|
152 |
92386 |
Node SequencesRewriter::rewriteStrEqualityExt(Node node) |
153 |
|
{ |
154 |
92386 |
Assert(node.getKind() == EQUAL && node[0].getType().isStringLike()); |
155 |
184772 |
TypeNode stype = node[0].getType(); |
156 |
|
|
157 |
92386 |
NodeManager* nm = NodeManager::currentNM(); |
158 |
184772 |
std::vector<Node> c[2]; |
159 |
184772 |
Node new_ret; |
160 |
277158 |
for (unsigned i = 0; i < 2; i++) |
161 |
|
{ |
162 |
184772 |
utils::getConcat(node[i], c[i]); |
163 |
|
} |
164 |
|
// ------- equality unification |
165 |
92386 |
bool changed = false; |
166 |
277158 |
for (unsigned i = 0; i < 2; i++) |
167 |
|
{ |
168 |
193944 |
while (!c[0].empty() && !c[1].empty() && c[0].back() == c[1].back()) |
169 |
|
{ |
170 |
4586 |
c[0].pop_back(); |
171 |
4586 |
c[1].pop_back(); |
172 |
4586 |
changed = true; |
173 |
|
} |
174 |
|
// splice constants |
175 |
548175 |
if (!c[0].empty() && !c[1].empty() && c[0].back().isConst() |
176 |
337800 |
&& c[1].back().isConst()) |
177 |
|
{ |
178 |
1194 |
Node cs[2]; |
179 |
|
size_t csl[2]; |
180 |
1791 |
for (unsigned j = 0; j < 2; j++) |
181 |
|
{ |
182 |
1194 |
cs[j] = c[j].back(); |
183 |
1194 |
csl[j] = Word::getLength(cs[j]); |
184 |
|
} |
185 |
597 |
size_t larger = csl[0] > csl[1] ? 0 : 1; |
186 |
597 |
size_t smallerSize = csl[1 - larger]; |
187 |
1194 |
if (cs[1 - larger] |
188 |
1194 |
== (i == 0 ? Word::suffix(cs[larger], smallerSize) |
189 |
|
: Word::prefix(cs[larger], smallerSize))) |
190 |
|
{ |
191 |
577 |
size_t sizeDiff = csl[larger] - smallerSize; |
192 |
577 |
c[larger][c[larger].size() - 1] = |
193 |
1154 |
i == 0 ? Word::prefix(cs[larger], sizeDiff) |
194 |
|
: Word::suffix(cs[larger], sizeDiff); |
195 |
577 |
c[1 - larger].pop_back(); |
196 |
577 |
changed = true; |
197 |
|
} |
198 |
|
} |
199 |
554316 |
for (unsigned j = 0; j < 2; j++) |
200 |
|
{ |
201 |
369544 |
std::reverse(c[j].begin(), c[j].end()); |
202 |
|
} |
203 |
|
} |
204 |
92386 |
if (changed) |
205 |
|
{ |
206 |
|
// e.g. x++y = x++z ---> y = z, "AB" ++ x = "A" ++ y --> "B" ++ x = y |
207 |
5486 |
Node s1 = utils::mkConcat(c[0], stype); |
208 |
5486 |
Node s2 = utils::mkConcat(c[1], stype); |
209 |
4439 |
if (s1 != node[0] || s2 != node[1]) |
210 |
|
{ |
211 |
3392 |
new_ret = s1.eqNode(s2); |
212 |
|
// We generally don't apply the extended equality rewriter if the |
213 |
|
// original node was an equality but we may be able to do additional |
214 |
|
// rewriting here, e.g., |
215 |
|
// x++y = "" --> x = "" and y = "" |
216 |
3392 |
return returnRewrite(node, new_ret, Rewrite::STR_EQ_UNIFY, true); |
217 |
|
} |
218 |
|
} |
219 |
|
|
220 |
|
// ------- homogeneous constants |
221 |
266885 |
for (unsigned i = 0; i < 2; i++) |
222 |
|
{ |
223 |
355841 |
Node cn = StringsEntail::checkHomogeneousString(node[i]); |
224 |
177950 |
if (!cn.isNull() && !Word::isEmpty(cn)) |
225 |
|
{ |
226 |
26278 |
Assert(cn.isConst()); |
227 |
26278 |
Assert(Word::getLength(cn) == 1); |
228 |
|
|
229 |
|
// The operands of the concat on each side of the equality without |
230 |
|
// constant strings |
231 |
78834 |
std::vector<Node> trimmed[2]; |
232 |
|
// Counts the number of `cn`s on each side |
233 |
26278 |
size_t numCns[2] = {0, 0}; |
234 |
78795 |
for (size_t j = 0; j < 2; j++) |
235 |
|
{ |
236 |
|
// Sort the operands of the concats on both sides of the equality |
237 |
|
// (since both sides may only contain one char, the order does not |
238 |
|
// matter) |
239 |
52537 |
std::sort(c[j].begin(), c[j].end()); |
240 |
104990 |
for (const Node& cc : c[j]) |
241 |
|
{ |
242 |
52473 |
if (cc.isConst()) |
243 |
|
{ |
244 |
|
// Count the number of `cn`s in the string constant and make |
245 |
|
// sure that all chars are `cn`s |
246 |
52406 |
std::vector<Node> veccc = Word::getChars(cc); |
247 |
51721 |
for (const Node& cv : veccc) |
248 |
|
{ |
249 |
25528 |
if (cv != cn) |
250 |
|
{ |
251 |
|
// This conflict case should mostly should be taken care of by |
252 |
|
// multiset reasoning in the strings rewriter, but we recognize |
253 |
|
// this conflict just in case. |
254 |
20 |
new_ret = nm->mkConst(false); |
255 |
|
return returnRewrite( |
256 |
20 |
node, new_ret, Rewrite::STR_EQ_CONST_NHOMOG); |
257 |
|
} |
258 |
25508 |
numCns[j]++; |
259 |
|
} |
260 |
|
} |
261 |
|
else |
262 |
|
{ |
263 |
26260 |
trimmed[j].push_back(cc); |
264 |
|
} |
265 |
|
} |
266 |
|
} |
267 |
|
|
268 |
|
// We have to remove the same number of `cn`s from both sides, so the |
269 |
|
// side with less `cn`s determines how many we can remove |
270 |
26258 |
size_t trimmedConst = std::min(numCns[0], numCns[1]); |
271 |
78774 |
for (size_t j = 0; j < 2; j++) |
272 |
|
{ |
273 |
52516 |
size_t diff = numCns[j] - trimmedConst; |
274 |
52516 |
if (diff != 0) |
275 |
|
{ |
276 |
|
// Add a constant string to the side with more `cn`s to restore |
277 |
|
// the difference in number of `cn`s |
278 |
46822 |
std::vector<Node> vec(diff, cn); |
279 |
23411 |
trimmed[j].push_back(Word::mkWordFlatten(vec)); |
280 |
|
} |
281 |
|
} |
282 |
|
|
283 |
52477 |
Node lhs = utils::mkConcat(trimmed[i], stype); |
284 |
52477 |
Node ss = utils::mkConcat(trimmed[1 - i], stype); |
285 |
26258 |
if (lhs != node[i] || ss != node[1 - i]) |
286 |
|
{ |
287 |
|
// e.g. |
288 |
|
// "AA" = y ++ x ---> "AA" = x ++ y if x < y |
289 |
|
// "AAA" = y ++ "A" ++ z ---> "AA" = y ++ z |
290 |
|
// |
291 |
|
// We generally don't apply the extended equality rewriter if the |
292 |
|
// original node was an equality but we may be able to do additional |
293 |
|
// rewriting here. |
294 |
39 |
new_ret = lhs.eqNode(ss); |
295 |
39 |
return returnRewrite(node, new_ret, Rewrite::STR_EQ_HOMOG_CONST, true); |
296 |
|
} |
297 |
|
} |
298 |
|
} |
299 |
|
|
300 |
|
// ------- rewrites for (= "" _) |
301 |
177870 |
Node empty = Word::mkEmptyWord(stype); |
302 |
251697 |
for (size_t i = 0; i < 2; i++) |
303 |
|
{ |
304 |
170426 |
if (node[i] == empty) |
305 |
|
{ |
306 |
110194 |
Node ne = node[1 - i]; |
307 |
58929 |
if (ne.getKind() == STRING_REPLACE) |
308 |
|
{ |
309 |
|
// (= "" (str.replace x y x)) ---> (= x "") |
310 |
2093 |
if (ne[0] == ne[2]) |
311 |
|
{ |
312 |
454 |
Node ret = nm->mkNode(EQUAL, ne[0], empty); |
313 |
227 |
return returnRewrite(node, ret, Rewrite::STR_EMP_REPL_X_Y_X); |
314 |
|
} |
315 |
|
|
316 |
|
// (= "" (str.replace x y "A")) ---> (and (= x "") (not (= y ""))) |
317 |
1866 |
if (StringsEntail::checkNonEmpty(ne[2])) |
318 |
|
{ |
319 |
|
Node ret = |
320 |
|
nm->mkNode(AND, |
321 |
686 |
nm->mkNode(EQUAL, ne[0], empty), |
322 |
1372 |
nm->mkNode(NOT, nm->mkNode(EQUAL, ne[1], empty))); |
323 |
343 |
return returnRewrite(node, ret, Rewrite::STR_EMP_REPL_EMP); |
324 |
|
} |
325 |
|
|
326 |
|
// (= "" (str.replace x "A" "")) ---> (str.prefix x "A") |
327 |
1523 |
if (StringsEntail::checkLengthOne(ne[1], true) && ne[2] == empty) |
328 |
|
{ |
329 |
530 |
Node ret = nm->mkNode(STRING_PREFIX, ne[0], ne[1]); |
330 |
265 |
return returnRewrite(node, ret, Rewrite::STR_EMP_REPL_EMP); |
331 |
|
} |
332 |
|
} |
333 |
56836 |
else if (ne.getKind() == STRING_SUBSTR) |
334 |
|
{ |
335 |
43305 |
Node zero = nm->mkConst(Rational(0)); |
336 |
|
|
337 |
25067 |
if (ArithEntail::check(ne[1], false) && ArithEntail::check(ne[2], true)) |
338 |
|
{ |
339 |
|
// (= "" (str.substr x 0 m)) ---> (= "" x) if m > 0 |
340 |
4215 |
if (ne[1] == zero) |
341 |
|
{ |
342 |
3906 |
Node ret = nm->mkNode(EQUAL, ne[0], empty); |
343 |
1953 |
return returnRewrite(node, ret, Rewrite::STR_EMP_SUBSTR_LEQ_LEN); |
344 |
|
} |
345 |
|
|
346 |
|
// (= "" (str.substr x n m)) ---> (<= (str.len x) n) |
347 |
|
// if n >= 0 and m > 0 |
348 |
4524 |
Node ret = nm->mkNode(LEQ, nm->mkNode(STRING_LENGTH, ne[0]), ne[1]); |
349 |
2262 |
return returnRewrite(node, ret, Rewrite::STR_EMP_SUBSTR_LEQ_LEN); |
350 |
|
} |
351 |
|
|
352 |
|
// (= "" (str.substr "A" 0 z)) ---> (<= z 0) |
353 |
20852 |
if (StringsEntail::checkNonEmpty(ne[0]) && ne[1] == zero) |
354 |
|
{ |
355 |
5228 |
Node ret = nm->mkNode(LEQ, ne[2], zero); |
356 |
2614 |
return returnRewrite(node, ret, Rewrite::STR_EMP_SUBSTR_LEQ_Z); |
357 |
|
} |
358 |
|
} |
359 |
|
} |
360 |
|
} |
361 |
|
|
362 |
|
// ------- rewrites for (= (str.replace _ _ _) _) |
363 |
243277 |
for (size_t i = 0; i < 2; i++) |
364 |
|
{ |
365 |
162495 |
if (node[i].getKind() == STRING_REPLACE) |
366 |
|
{ |
367 |
5609 |
Node repl = node[i]; |
368 |
5609 |
Node x = node[1 - i]; |
369 |
|
|
370 |
|
// (= "A" (str.replace "" x y)) ---> (= "" (str.replace "A" y x)) |
371 |
3049 |
if (StringsEntail::checkNonEmpty(x) && repl[0] == empty) |
372 |
|
{ |
373 |
|
Node ret = nm->mkNode( |
374 |
38 |
EQUAL, empty, nm->mkNode(STRING_REPLACE, x, repl[2], repl[1])); |
375 |
19 |
return returnRewrite(node, ret, Rewrite::STR_EQ_REPL_EMP); |
376 |
|
} |
377 |
|
|
378 |
|
// (= x (str.replace y x y)) ---> (= x y) |
379 |
3030 |
if (repl[0] == repl[2] && x == repl[1]) |
380 |
|
{ |
381 |
4 |
Node ret = nm->mkNode(EQUAL, x, repl[0]); |
382 |
2 |
return returnRewrite(node, ret, Rewrite::STR_EQ_REPL_TO_EQ); |
383 |
|
} |
384 |
|
|
385 |
|
// (= x (str.replace x "A" "B")) ---> (not (str.contains x "A")) |
386 |
3028 |
if (x == repl[0]) |
387 |
|
{ |
388 |
2520 |
Node eq = Rewriter::rewrite(nm->mkNode(EQUAL, repl[1], repl[2])); |
389 |
1492 |
if (eq.isConst() && !eq.getConst<bool>()) |
390 |
|
{ |
391 |
928 |
Node ret = nm->mkNode(NOT, nm->mkNode(STRING_CONTAINS, x, repl[1])); |
392 |
464 |
return returnRewrite(node, ret, Rewrite::STR_EQ_REPL_NOT_CTN); |
393 |
|
} |
394 |
|
} |
395 |
|
|
396 |
|
// (= (str.replace x y z) z) --> (or (= x y) (= x z)) |
397 |
|
// if (str.len y) = (str.len z) |
398 |
2564 |
if (repl[2] == x) |
399 |
|
{ |
400 |
2650 |
Node lenY = nm->mkNode(STRING_LENGTH, repl[1]); |
401 |
2650 |
Node lenZ = nm->mkNode(STRING_LENGTH, repl[2]); |
402 |
1327 |
if (ArithEntail::checkEq(lenY, lenZ)) |
403 |
|
{ |
404 |
|
Node ret = nm->mkNode(OR, |
405 |
8 |
nm->mkNode(EQUAL, repl[0], repl[1]), |
406 |
16 |
nm->mkNode(EQUAL, repl[0], repl[2])); |
407 |
4 |
return returnRewrite(node, ret, Rewrite::STR_EQ_REPL_TO_DIS); |
408 |
|
} |
409 |
|
} |
410 |
|
} |
411 |
|
} |
412 |
|
|
413 |
|
// Try to rewrite (= x y) into a conjunction of equalities based on length |
414 |
|
// entailment. |
415 |
|
// |
416 |
|
// (<= (str.len x) (str.++ y1 ... yn)) AND (= x (str.++ y1 ... yn)) ---> |
417 |
|
// (and (= x (str.++ y1' ... ym')) (= y1'' "") ... (= yk'' "")) |
418 |
|
// |
419 |
|
// where yi' and yi'' correspond to some yj and |
420 |
|
// (<= (str.len x) (str.++ y1' ... ym')) |
421 |
239008 |
for (unsigned i = 0; i < 2; i++) |
422 |
|
{ |
423 |
159941 |
if (node[1 - i].getKind() == STRING_CONCAT) |
424 |
|
{ |
425 |
2159 |
new_ret = StringsEntail::inferEqsFromContains(node[i], node[1 - i]); |
426 |
2159 |
if (!new_ret.isNull()) |
427 |
|
{ |
428 |
1715 |
return returnRewrite(node, new_ret, Rewrite::STR_EQ_CONJ_LEN_ENTAIL); |
429 |
|
} |
430 |
|
} |
431 |
|
} |
432 |
|
|
433 |
79067 |
if (node[0].getKind() == STRING_CONCAT && node[1].getKind() == STRING_CONCAT) |
434 |
|
{ |
435 |
|
// (= (str.++ x_1 ... x_i x_{i + 1} ... x_n) |
436 |
|
// (str.++ y_1 ... y_j y_{j + 1} ... y_m)) ---> |
437 |
|
// (and (= (str.++ x_1 ... x_i) (str.++ y_1 ... y_j)) |
438 |
|
// (= (str.++ x_{i + 1} ... x_n) (str.++ y_{j + 1} ... y_m))) |
439 |
|
// |
440 |
|
// if (str.len (str.++ x_1 ... x_i)) = (str.len (str.++ y_1 ... y_j)) |
441 |
|
// |
442 |
|
// This rewrite performs length-based equality splitting: If we can show |
443 |
|
// that two prefixes have the same length, we can split an equality into |
444 |
|
// two equalities, one over the prefixes and another over the suffixes. |
445 |
28 |
std::vector<Node> v0, v1; |
446 |
18 |
utils::getConcat(node[0], v0); |
447 |
18 |
utils::getConcat(node[1], v1); |
448 |
18 |
size_t startRhs = 0; |
449 |
70 |
for (size_t i = 0, size0 = v0.size(); i <= size0; i++) |
450 |
|
{ |
451 |
112 |
const std::vector<Node> pfxv0(v0.begin(), v0.begin() + i); |
452 |
112 |
Node pfx0 = utils::mkConcat(pfxv0, stype); |
453 |
222 |
for (size_t j = startRhs, size1 = v1.size(); j <= size1; j++) |
454 |
|
{ |
455 |
192 |
if (!(i == 0 && j == 0) && !(i == v0.size() && j == v1.size())) |
456 |
|
{ |
457 |
298 |
std::vector<Node> pfxv1(v1.begin(), v1.begin() + j); |
458 |
298 |
Node pfx1 = utils::mkConcat(pfxv1, stype); |
459 |
298 |
Node lenPfx0 = nm->mkNode(STRING_LENGTH, pfx0); |
460 |
298 |
Node lenPfx1 = nm->mkNode(STRING_LENGTH, pfx1); |
461 |
|
|
462 |
164 |
if (ArithEntail::checkEq(lenPfx0, lenPfx1)) |
463 |
|
{ |
464 |
8 |
std::vector<Node> sfxv0(v0.begin() + i, v0.end()); |
465 |
8 |
std::vector<Node> sfxv1(v1.begin() + j, v1.end()); |
466 |
|
Node ret = nm->mkNode(kind::AND, |
467 |
8 |
pfx0.eqNode(pfx1), |
468 |
8 |
utils::mkConcat(sfxv0, stype) |
469 |
16 |
.eqNode(utils::mkConcat(sfxv1, stype))); |
470 |
4 |
return returnRewrite(node, ret, Rewrite::SPLIT_EQ); |
471 |
|
} |
472 |
160 |
else if (ArithEntail::check(lenPfx1, lenPfx0, true)) |
473 |
|
{ |
474 |
|
// The prefix on the right-hand side is strictly longer than the |
475 |
|
// prefix on the left-hand side, so we try to strip the right-hand |
476 |
|
// prefix by the length of the left-hand prefix |
477 |
|
// |
478 |
|
// Example: |
479 |
|
// (= (str.++ "A" x y) (str.++ x "AB" z)) ---> |
480 |
|
// (and (= (str.++ "A" x) (str.++ x "A")) (= y (str.++ "B" z))) |
481 |
22 |
std::vector<Node> rpfxv1; |
482 |
22 |
if (StringsEntail::stripSymbolicLength( |
483 |
|
pfxv1, rpfxv1, 1, lenPfx0, true)) |
484 |
|
{ |
485 |
|
// The rewrite requires the full left-hand prefix length to be |
486 |
|
// stripped (otherwise we would have to keep parts of the |
487 |
|
// left-hand prefix). |
488 |
|
if (lenPfx0.isConst() && lenPfx0.getConst<Rational>().isZero()) |
489 |
|
{ |
490 |
|
std::vector<Node> sfxv0(v0.begin() + i, v0.end()); |
491 |
|
pfxv1.insert(pfxv1.end(), v1.begin() + j, v1.end()); |
492 |
|
Node ret = |
493 |
|
nm->mkNode(kind::AND, |
494 |
|
pfx0.eqNode(utils::mkConcat(rpfxv1, stype)), |
495 |
|
utils::mkConcat(sfxv0, stype) |
496 |
|
.eqNode(utils::mkConcat(pfxv1, stype))); |
497 |
|
return returnRewrite(node, ret, Rewrite::SPLIT_EQ_STRIP_R); |
498 |
|
} |
499 |
|
} |
500 |
|
|
501 |
|
// If the prefix of the right-hand side is (strictly) longer than |
502 |
|
// the prefix of the left-hand side, we can advance the left-hand |
503 |
|
// side (since the length of the right-hand side is only increasing |
504 |
|
// in the inner loop) |
505 |
22 |
break; |
506 |
|
} |
507 |
138 |
else if (ArithEntail::check(lenPfx0, lenPfx1, true)) |
508 |
|
{ |
509 |
|
// The prefix on the left-hand side is strictly longer than the |
510 |
|
// prefix on the right-hand side, so we try to strip the left-hand |
511 |
|
// prefix by the length of the right-hand prefix |
512 |
|
// |
513 |
|
// Example: |
514 |
|
// (= (str.++ x "AB" z) (str.++ "A" x y)) ---> |
515 |
|
// (and (= (str.++ x "A") (str.++ "A" x)) (= (str.++ "B" z) y)) |
516 |
56 |
std::vector<Node> sfxv0 = pfxv0; |
517 |
56 |
std::vector<Node> rpfxv0; |
518 |
30 |
if (StringsEntail::stripSymbolicLength( |
519 |
|
sfxv0, rpfxv0, 1, lenPfx1, true)) |
520 |
|
{ |
521 |
|
// The rewrite requires the full right-hand prefix length to be |
522 |
|
// stripped (otherwise we would have to keep parts of the |
523 |
|
// right-hand prefix). |
524 |
6 |
if (lenPfx1.isConst() && lenPfx1.getConst<Rational>().isZero()) |
525 |
|
{ |
526 |
4 |
sfxv0.insert(sfxv0.end(), v0.begin() + i, v0.end()); |
527 |
8 |
std::vector<Node> sfxv1(v1.begin() + j, v1.end()); |
528 |
|
Node ret = |
529 |
|
nm->mkNode(kind::AND, |
530 |
8 |
utils::mkConcat(rpfxv0, stype).eqNode(pfx1), |
531 |
8 |
utils::mkConcat(sfxv0, stype) |
532 |
16 |
.eqNode(utils::mkConcat(sfxv1, stype))); |
533 |
4 |
return returnRewrite(node, ret, Rewrite::SPLIT_EQ_STRIP_L); |
534 |
|
} |
535 |
|
} |
536 |
|
|
537 |
|
// If the prefix of the left-hand side is (strictly) longer than |
538 |
|
// the prefix of the right-hand side, then we don't need to check |
539 |
|
// that right-hand prefix for future left-hand prefixes anymore |
540 |
|
// (since they are increasing in length) |
541 |
26 |
startRhs = j + 1; |
542 |
|
} |
543 |
|
} |
544 |
|
} |
545 |
|
} |
546 |
|
} |
547 |
|
|
548 |
79059 |
return node; |
549 |
|
} |
550 |
|
|
551 |
20770 |
Node SequencesRewriter::rewriteArithEqualityExt(Node node) |
552 |
|
{ |
553 |
20770 |
Assert(node.getKind() == EQUAL && node[0].getType().isInteger()); |
554 |
|
|
555 |
|
// cases where we can solve the equality |
556 |
|
|
557 |
|
// notice we cannot rewrite str.to.int(x)=n to x="n" due to leading zeroes. |
558 |
|
|
559 |
20770 |
return node; |
560 |
|
} |
561 |
|
|
562 |
222423 |
Node SequencesRewriter::rewriteLength(Node node) |
563 |
|
{ |
564 |
222423 |
Assert(node.getKind() == STRING_LENGTH); |
565 |
222423 |
NodeManager* nm = NodeManager::currentNM(); |
566 |
222423 |
Kind nk0 = node[0].getKind(); |
567 |
222423 |
if (node[0].isConst()) |
568 |
|
{ |
569 |
13538 |
Node retNode = nm->mkConst(Rational(Word::getLength(node[0]))); |
570 |
6769 |
return returnRewrite(node, retNode, Rewrite::LEN_EVAL); |
571 |
|
} |
572 |
215654 |
else if (nk0 == kind::STRING_CONCAT) |
573 |
|
{ |
574 |
10879 |
Node tmpNode = node[0]; |
575 |
10879 |
if (tmpNode.getKind() == kind::STRING_CONCAT) |
576 |
|
{ |
577 |
21758 |
std::vector<Node> node_vec; |
578 |
42372 |
for (unsigned int i = 0; i < tmpNode.getNumChildren(); ++i) |
579 |
|
{ |
580 |
31493 |
if (tmpNode[i].isConst()) |
581 |
|
{ |
582 |
7640 |
node_vec.push_back( |
583 |
15280 |
nm->mkConst(Rational(Word::getLength(tmpNode[i])))); |
584 |
|
} |
585 |
|
else |
586 |
|
{ |
587 |
23853 |
node_vec.push_back(NodeManager::currentNM()->mkNode( |
588 |
|
kind::STRING_LENGTH, tmpNode[i])); |
589 |
|
} |
590 |
|
} |
591 |
21758 |
Node retNode = NodeManager::currentNM()->mkNode(kind::PLUS, node_vec); |
592 |
10879 |
return returnRewrite(node, retNode, Rewrite::LEN_CONCAT); |
593 |
|
} |
594 |
|
} |
595 |
204775 |
else if (nk0 == STRING_REPLACE || nk0 == STRING_REPLACE_ALL) |
596 |
|
{ |
597 |
1901 |
Node len1 = Rewriter::rewrite(nm->mkNode(STRING_LENGTH, node[0][1])); |
598 |
1901 |
Node len2 = Rewriter::rewrite(nm->mkNode(STRING_LENGTH, node[0][2])); |
599 |
971 |
if (len1 == len2) |
600 |
|
{ |
601 |
|
// len( y ) == len( z ) => len( str.replace( x, y, z ) ) ---> len( x ) |
602 |
82 |
Node retNode = nm->mkNode(STRING_LENGTH, node[0][0]); |
603 |
41 |
return returnRewrite(node, retNode, Rewrite::LEN_REPL_INV); |
604 |
930 |
} |
605 |
|
} |
606 |
203804 |
else if (nk0 == STRING_TOLOWER || nk0 == STRING_TOUPPER || nk0 == STRING_REV |
607 |
203700 |
|| nk0 == STRING_UPDATE) |
608 |
|
{ |
609 |
|
// len( f( x ) ) == len( x ) where f is tolower, toupper, or rev. |
610 |
|
// len( update( x, n, y ) ) = len( x ) |
611 |
240 |
Node retNode = nm->mkNode(STRING_LENGTH, node[0][0]); |
612 |
120 |
return returnRewrite(node, retNode, Rewrite::LEN_CONV_INV); |
613 |
|
} |
614 |
203684 |
else if (nk0 == SEQ_UNIT) |
615 |
|
{ |
616 |
244 |
Node retNode = nm->mkConst(Rational(1)); |
617 |
122 |
return returnRewrite(node, retNode, Rewrite::LEN_SEQ_UNIT); |
618 |
|
} |
619 |
204492 |
return node; |
620 |
|
} |
621 |
|
|
622 |
|
// TODO (#1180) add rewrite |
623 |
|
// str.++( str.substr( x, n1, n2 ), str.substr( x, n1+n2, n3 ) ) ---> |
624 |
|
// str.substr( x, n1, n2+n3 ) |
625 |
57858 |
Node SequencesRewriter::rewriteConcat(Node node) |
626 |
|
{ |
627 |
57858 |
Assert(node.getKind() == kind::STRING_CONCAT); |
628 |
115716 |
Trace("strings-rewrite-debug") |
629 |
57858 |
<< "Strings::rewriteConcat start " << node << std::endl; |
630 |
115716 |
std::vector<Node> node_vec; |
631 |
115716 |
Node preNode = Node::null(); |
632 |
249677 |
for (Node tmpNode : node) |
633 |
|
{ |
634 |
191819 |
if (tmpNode.getKind() == STRING_CONCAT) |
635 |
|
{ |
636 |
1289 |
unsigned j = 0; |
637 |
|
// combine the first term with the previous constant if applicable |
638 |
1289 |
if (!preNode.isNull()) |
639 |
|
{ |
640 |
315 |
if (tmpNode[0].isConst()) |
641 |
|
{ |
642 |
278 |
std::vector<Node> wvec; |
643 |
139 |
wvec.push_back(preNode); |
644 |
139 |
wvec.push_back(tmpNode[0]); |
645 |
139 |
preNode = Word::mkWordFlatten(wvec); |
646 |
139 |
node_vec.push_back(preNode); |
647 |
|
} |
648 |
|
else |
649 |
|
{ |
650 |
176 |
node_vec.push_back(preNode); |
651 |
176 |
node_vec.push_back(tmpNode[0]); |
652 |
|
} |
653 |
315 |
preNode = Node::null(); |
654 |
315 |
++j; |
655 |
|
} |
656 |
|
// insert the middle terms to node_vec |
657 |
1289 |
if (j <= tmpNode.getNumChildren() - 1) |
658 |
|
{ |
659 |
1289 |
node_vec.insert(node_vec.end(), tmpNode.begin() + j, tmpNode.end() - 1); |
660 |
|
} |
661 |
|
// take the last term as the current |
662 |
1289 |
tmpNode = tmpNode[tmpNode.getNumChildren() - 1]; |
663 |
|
} |
664 |
191819 |
if (!tmpNode.isConst()) |
665 |
|
{ |
666 |
91496 |
if (!preNode.isNull()) |
667 |
|
{ |
668 |
17480 |
if (preNode.isConst() && !Word::isEmpty(preNode)) |
669 |
|
{ |
670 |
16731 |
node_vec.push_back(preNode); |
671 |
|
} |
672 |
17480 |
preNode = Node::null(); |
673 |
|
} |
674 |
91496 |
node_vec.push_back(tmpNode); |
675 |
|
} |
676 |
|
else |
677 |
|
{ |
678 |
100323 |
if (preNode.isNull()) |
679 |
|
{ |
680 |
49079 |
preNode = tmpNode; |
681 |
|
} |
682 |
|
else |
683 |
|
{ |
684 |
102488 |
std::vector<Node> vec; |
685 |
51244 |
vec.push_back(preNode); |
686 |
51244 |
vec.push_back(tmpNode); |
687 |
51244 |
preNode = Word::mkWordFlatten(vec); |
688 |
|
} |
689 |
|
} |
690 |
|
} |
691 |
57858 |
if (!preNode.isNull() && (!preNode.isConst() || !Word::isEmpty(preNode))) |
692 |
|
{ |
693 |
30231 |
node_vec.push_back(preNode); |
694 |
|
} |
695 |
|
|
696 |
|
// Sort adjacent operands in str.++ that all result in the same string or the |
697 |
|
// empty string. |
698 |
|
// |
699 |
|
// E.g.: (str.++ ... (str.replace "A" x "") "A" (str.substr "A" 0 z) ...) --> |
700 |
|
// (str.++ ... [sort those 3 arguments] ... ) |
701 |
57858 |
size_t lastIdx = 0; |
702 |
115716 |
Node lastX; |
703 |
198932 |
for (size_t i = 0, nsize = node_vec.size(); i < nsize; i++) |
704 |
|
{ |
705 |
282148 |
Node s = StringsEntail::getStringOrEmpty(node_vec[i]); |
706 |
141074 |
bool nextX = false; |
707 |
141074 |
if (s != lastX) |
708 |
|
{ |
709 |
137666 |
nextX = true; |
710 |
|
} |
711 |
|
|
712 |
141074 |
if (nextX) |
713 |
|
{ |
714 |
137666 |
std::sort(node_vec.begin() + lastIdx, node_vec.begin() + i); |
715 |
137666 |
lastX = s; |
716 |
137666 |
lastIdx = i; |
717 |
|
} |
718 |
|
} |
719 |
57858 |
std::sort(node_vec.begin() + lastIdx, node_vec.end()); |
720 |
|
|
721 |
115716 |
TypeNode tn = node.getType(); |
722 |
115716 |
Node retNode = utils::mkConcat(node_vec, tn); |
723 |
115716 |
Trace("strings-rewrite-debug") |
724 |
57858 |
<< "Strings::rewriteConcat end " << retNode << std::endl; |
725 |
57858 |
if (retNode != node) |
726 |
|
{ |
727 |
19814 |
return returnRewrite(node, retNode, Rewrite::CONCAT_NORM); |
728 |
|
} |
729 |
38044 |
return node; |
730 |
|
} |
731 |
|
|
732 |
2864 |
Node SequencesRewriter::rewriteConcatRegExp(TNode node) |
733 |
|
{ |
734 |
2864 |
Assert(node.getKind() == kind::REGEXP_CONCAT); |
735 |
2864 |
NodeManager* nm = NodeManager::currentNM(); |
736 |
5728 |
Trace("strings-rewrite-debug") |
737 |
2864 |
<< "Strings::rewriteConcatRegExp flatten " << node << std::endl; |
738 |
5728 |
Node retNode = node; |
739 |
5728 |
std::vector<Node> vec; |
740 |
2864 |
bool changed = false; |
741 |
5728 |
Node emptyRe; |
742 |
|
|
743 |
|
// get the string type that are members of this regular expression |
744 |
5728 |
TypeNode rtype = node.getType(); |
745 |
5728 |
TypeNode stype; |
746 |
2864 |
if (rtype.isRegExp()) |
747 |
|
{ |
748 |
|
// standard regular expressions are for strings |
749 |
2864 |
stype = nm->stringType(); |
750 |
|
} |
751 |
|
else |
752 |
|
{ |
753 |
|
Unimplemented(); |
754 |
|
} |
755 |
|
|
756 |
10948 |
for (const Node& c : node) |
757 |
|
{ |
758 |
8120 |
if (c.getKind() == REGEXP_CONCAT) |
759 |
|
{ |
760 |
218 |
changed = true; |
761 |
698 |
for (const Node& cc : c) |
762 |
|
{ |
763 |
480 |
vec.push_back(cc); |
764 |
|
} |
765 |
|
} |
766 |
19174 |
else if (c.getKind() == STRING_TO_REGEXP && c[0].isConst() |
767 |
19094 |
&& Word::isEmpty(c[0])) |
768 |
|
{ |
769 |
128 |
changed = true; |
770 |
128 |
emptyRe = c; |
771 |
|
} |
772 |
7774 |
else if (c.getKind() == REGEXP_EMPTY) |
773 |
|
{ |
774 |
|
// re.++( ..., empty, ... ) ---> empty |
775 |
72 |
Node ret = nm->mkNode(REGEXP_EMPTY); |
776 |
36 |
return returnRewrite(node, ret, Rewrite::RE_CONCAT_EMPTY); |
777 |
|
} |
778 |
|
else |
779 |
|
{ |
780 |
7738 |
vec.push_back(c); |
781 |
|
} |
782 |
|
} |
783 |
2828 |
if (changed) |
784 |
|
{ |
785 |
|
// flatten |
786 |
|
// this handles nested re.++ and elimination or str.to.re(""), e.g.: |
787 |
|
// re.++( re.++( R1, R2 ), str.to.re(""), R3 ) ---> re.++( R1, R2, R3 ) |
788 |
328 |
if (vec.empty()) |
789 |
|
{ |
790 |
2 |
Assert(!emptyRe.isNull()); |
791 |
2 |
retNode = emptyRe; |
792 |
|
} |
793 |
|
else |
794 |
|
{ |
795 |
326 |
retNode = vec.size() == 1 ? vec[0] : nm->mkNode(REGEXP_CONCAT, vec); |
796 |
|
} |
797 |
328 |
return returnRewrite(node, retNode, Rewrite::RE_CONCAT_FLATTEN); |
798 |
|
} |
799 |
5000 |
Trace("strings-rewrite-debug") |
800 |
2500 |
<< "Strings::rewriteConcatRegExp start " << node << std::endl; |
801 |
5000 |
std::vector<Node> cvec; |
802 |
|
// the current accumulation of constant strings |
803 |
5000 |
std::vector<Node> preReStr; |
804 |
|
// whether the last component was (_)* |
805 |
2500 |
bool lastAllStar = false; |
806 |
5000 |
String emptyStr = String(""); |
807 |
|
// this loop checks to see if components can be combined or dropped |
808 |
12324 |
for (unsigned i = 0, size = vec.size(); i <= size; i++) |
809 |
|
{ |
810 |
19648 |
Node curr; |
811 |
9824 |
if (i < size) |
812 |
|
{ |
813 |
7324 |
curr = vec[i]; |
814 |
7324 |
Assert(curr.getKind() != REGEXP_CONCAT); |
815 |
|
} |
816 |
|
// update preReStr |
817 |
9824 |
if (!curr.isNull() && curr.getKind() == STRING_TO_REGEXP) |
818 |
|
{ |
819 |
3008 |
lastAllStar = false; |
820 |
3008 |
preReStr.push_back(curr[0]); |
821 |
3008 |
curr = Node::null(); |
822 |
|
} |
823 |
6816 |
else if (!preReStr.empty()) |
824 |
|
{ |
825 |
2552 |
Assert(!lastAllStar); |
826 |
|
// this groups consecutive strings a++b ---> ab |
827 |
5104 |
Node acc = nm->mkNode(STRING_TO_REGEXP, utils::mkConcat(preReStr, stype)); |
828 |
2552 |
cvec.push_back(acc); |
829 |
2552 |
preReStr.clear(); |
830 |
|
} |
831 |
4264 |
else if (!curr.isNull() && lastAllStar) |
832 |
|
{ |
833 |
|
// if empty, drop it |
834 |
|
// e.g. this ensures we rewrite (_)* ++ (a)* ---> (_)* |
835 |
324 |
if (RegExpEntail::isConstRegExp(curr) |
836 |
324 |
&& RegExpEntail::testConstStringInRegExp(emptyStr, 0, curr)) |
837 |
|
{ |
838 |
44 |
curr = Node::null(); |
839 |
|
} |
840 |
|
} |
841 |
9824 |
if (!curr.isNull()) |
842 |
|
{ |
843 |
4272 |
lastAllStar = false; |
844 |
4272 |
if (curr.getKind() == REGEXP_STAR) |
845 |
|
{ |
846 |
|
// we can group stars (a)* ++ (a)* ---> (a)* |
847 |
2252 |
if (!cvec.empty() && cvec.back() == curr) |
848 |
|
{ |
849 |
8 |
curr = Node::null(); |
850 |
|
} |
851 |
2244 |
else if (curr[0].getKind() == REGEXP_SIGMA) |
852 |
|
{ |
853 |
902 |
Assert(!lastAllStar); |
854 |
902 |
lastAllStar = true; |
855 |
|
// go back and remove empty ones from back of cvec |
856 |
|
// e.g. this ensures we rewrite (a)* ++ (_)* ---> (_)* |
857 |
2572 |
while (!cvec.empty() && RegExpEntail::isConstRegExp(cvec.back()) |
858 |
3452 |
&& RegExpEntail::testConstStringInRegExp( |
859 |
722 |
emptyStr, 0, cvec.back())) |
860 |
|
{ |
861 |
8 |
cvec.pop_back(); |
862 |
|
} |
863 |
|
} |
864 |
|
} |
865 |
|
} |
866 |
9824 |
if (!curr.isNull()) |
867 |
|
{ |
868 |
4264 |
cvec.push_back(curr); |
869 |
|
} |
870 |
|
} |
871 |
2500 |
Assert(!cvec.empty()); |
872 |
2500 |
retNode = utils::mkConcat(cvec, rtype); |
873 |
2500 |
if (retNode != node) |
874 |
|
{ |
875 |
|
// handles all cases where consecutive re constants are combined or |
876 |
|
// dropped as described in the loop above. |
877 |
198 |
return returnRewrite(node, retNode, Rewrite::RE_CONCAT); |
878 |
|
} |
879 |
|
|
880 |
|
// flipping adjacent star arguments |
881 |
2302 |
changed = false; |
882 |
6356 |
for (size_t i = 0, size = cvec.size() - 1; i < size; i++) |
883 |
|
{ |
884 |
4054 |
if (cvec[i].getKind() == REGEXP_STAR && cvec[i][0] == cvec[i + 1]) |
885 |
|
{ |
886 |
|
// by convention, flip the order (a*)++a ---> a++(a*) |
887 |
64 |
std::swap(cvec[i], cvec[i + 1]); |
888 |
64 |
changed = true; |
889 |
|
} |
890 |
|
} |
891 |
2302 |
if (changed) |
892 |
|
{ |
893 |
48 |
retNode = utils::mkConcat(cvec, rtype); |
894 |
48 |
return returnRewrite(node, retNode, Rewrite::RE_CONCAT_OPT); |
895 |
|
} |
896 |
2254 |
return node; |
897 |
|
} |
898 |
|
|
899 |
1170 |
Node SequencesRewriter::rewriteStarRegExp(TNode node) |
900 |
|
{ |
901 |
1170 |
Assert(node.getKind() == REGEXP_STAR); |
902 |
1170 |
NodeManager* nm = NodeManager::currentNM(); |
903 |
2340 |
Node retNode = node; |
904 |
1170 |
if (node[0].getKind() == REGEXP_STAR) |
905 |
|
{ |
906 |
|
// ((R)*)* ---> R* |
907 |
4 |
return returnRewrite(node, node[0], Rewrite::RE_STAR_NESTED_STAR); |
908 |
|
} |
909 |
3962 |
else if (node[0].getKind() == STRING_TO_REGEXP && node[0][0].isConst() |
910 |
3902 |
&& Word::isEmpty(node[0][0])) |
911 |
|
{ |
912 |
|
// ("")* ---> "" |
913 |
18 |
return returnRewrite(node, node[0], Rewrite::RE_STAR_EMPTY_STRING); |
914 |
|
} |
915 |
1148 |
else if (node[0].getKind() == REGEXP_EMPTY) |
916 |
|
{ |
917 |
|
// (empty)* ---> "" |
918 |
8 |
retNode = nm->mkNode(STRING_TO_REGEXP, nm->mkConst(String(""))); |
919 |
8 |
return returnRewrite(node, retNode, Rewrite::RE_STAR_EMPTY); |
920 |
|
} |
921 |
1140 |
else if (node[0].getKind() == REGEXP_UNION) |
922 |
|
{ |
923 |
|
// simplification of unions under star |
924 |
256 |
if (RegExpEntail::hasEpsilonNode(node[0])) |
925 |
|
{ |
926 |
38 |
bool changed = false; |
927 |
38 |
std::vector<Node> node_vec; |
928 |
120 |
for (const Node& nc : node[0]) |
929 |
|
{ |
930 |
240 |
if (nc.getKind() == STRING_TO_REGEXP && nc[0].isConst() |
931 |
236 |
&& Word::isEmpty(nc[0])) |
932 |
|
{ |
933 |
|
// can be removed |
934 |
38 |
changed = true; |
935 |
|
} |
936 |
|
else |
937 |
|
{ |
938 |
44 |
node_vec.push_back(nc); |
939 |
|
} |
940 |
|
} |
941 |
38 |
if (changed) |
942 |
|
{ |
943 |
38 |
retNode = node_vec.size() == 1 ? node_vec[0] |
944 |
|
: nm->mkNode(REGEXP_UNION, node_vec); |
945 |
38 |
retNode = nm->mkNode(REGEXP_STAR, retNode); |
946 |
|
// simplification of union beneath star based on loop above |
947 |
|
// for example, ( "" | "a" )* ---> ("a")* |
948 |
38 |
return returnRewrite(node, retNode, Rewrite::RE_STAR_UNION); |
949 |
|
} |
950 |
|
} |
951 |
|
} |
952 |
1102 |
return node; |
953 |
|
} |
954 |
|
|
955 |
1080 |
Node SequencesRewriter::rewriteAndOrRegExp(TNode node) |
956 |
|
{ |
957 |
1080 |
Kind nk = node.getKind(); |
958 |
1080 |
Assert(nk == REGEXP_UNION || nk == REGEXP_INTER); |
959 |
2160 |
Trace("strings-rewrite-debug") |
960 |
1080 |
<< "Strings::rewriteAndOrRegExp start " << node << std::endl; |
961 |
2160 |
std::vector<Node> node_vec; |
962 |
2160 |
std::vector<Node> polRegExp[2]; |
963 |
3463 |
for (const Node& ni : node) |
964 |
|
{ |
965 |
2385 |
if (ni.getKind() == nk) |
966 |
|
{ |
967 |
42 |
for (const Node& nic : ni) |
968 |
|
{ |
969 |
28 |
if (std::find(node_vec.begin(), node_vec.end(), nic) == node_vec.end()) |
970 |
|
{ |
971 |
26 |
node_vec.push_back(nic); |
972 |
|
} |
973 |
|
} |
974 |
|
} |
975 |
2371 |
else if (ni.getKind() == REGEXP_EMPTY) |
976 |
|
{ |
977 |
65 |
if (nk == REGEXP_INTER) |
978 |
|
{ |
979 |
2 |
return returnRewrite(node, ni, Rewrite::RE_AND_EMPTY); |
980 |
|
} |
981 |
|
// otherwise, can ignore |
982 |
|
} |
983 |
2306 |
else if (ni.getKind() == REGEXP_STAR && ni[0].getKind() == REGEXP_SIGMA) |
984 |
|
{ |
985 |
8 |
if (nk == REGEXP_UNION) |
986 |
|
{ |
987 |
|
return returnRewrite(node, ni, Rewrite::RE_OR_ALL); |
988 |
|
} |
989 |
|
// otherwise, can ignore |
990 |
|
} |
991 |
2298 |
else if (std::find(node_vec.begin(), node_vec.end(), ni) == node_vec.end()) |
992 |
|
{ |
993 |
2259 |
node_vec.push_back(ni); |
994 |
2259 |
uint32_t pindex = ni.getKind() == REGEXP_COMPLEMENT ? 1 : 0; |
995 |
4518 |
Node nia = pindex == 1 ? ni[0] : ni; |
996 |
2259 |
polRegExp[pindex].push_back(nia); |
997 |
|
} |
998 |
|
} |
999 |
1078 |
NodeManager* nm = NodeManager::currentNM(); |
1000 |
|
// use inclusion tests |
1001 |
1274 |
for (const Node& negMem : polRegExp[1]) |
1002 |
|
{ |
1003 |
346 |
for (const Node& posMem : polRegExp[0]) |
1004 |
|
{ |
1005 |
284 |
Node m1 = nk == REGEXP_INTER ? negMem : posMem; |
1006 |
284 |
Node m2 = nk == REGEXP_INTER ? posMem : negMem; |
1007 |
|
// inclusion test for conflicting case m1 contains m2 |
1008 |
|
// (re.inter (re.comp R1) R2) --> re.none where R1 includes R2 |
1009 |
|
// (re.union R1 (re.comp R2)) --> (re.* re.allchar) where R1 includes R2 |
1010 |
150 |
if (RegExpEntail::regExpIncludes(m1, m2)) |
1011 |
|
{ |
1012 |
32 |
Node retNode; |
1013 |
16 |
if (nk == REGEXP_INTER) |
1014 |
|
{ |
1015 |
16 |
retNode = nm->mkNode(kind::REGEXP_EMPTY); |
1016 |
|
} |
1017 |
|
else |
1018 |
|
{ |
1019 |
|
retNode = nm->mkNode(REGEXP_STAR, nm->mkNode(REGEXP_SIGMA)); |
1020 |
|
} |
1021 |
16 |
return returnRewrite(node, retNode, Rewrite::RE_ANDOR_INC_CONFLICT); |
1022 |
|
} |
1023 |
|
} |
1024 |
|
} |
1025 |
2124 |
Node retNode; |
1026 |
1062 |
if (node_vec.empty()) |
1027 |
|
{ |
1028 |
|
if (nk == REGEXP_INTER) |
1029 |
|
{ |
1030 |
|
retNode = nm->mkNode(REGEXP_STAR, nm->mkNode(REGEXP_SIGMA)); |
1031 |
|
} |
1032 |
|
else |
1033 |
|
{ |
1034 |
|
retNode = nm->mkNode(kind::REGEXP_EMPTY); |
1035 |
|
} |
1036 |
|
} |
1037 |
|
else |
1038 |
|
{ |
1039 |
1062 |
retNode = node_vec.size() == 1 ? node_vec[0] : nm->mkNode(nk, node_vec); |
1040 |
|
} |
1041 |
1062 |
if (retNode != node) |
1042 |
|
{ |
1043 |
|
// flattening and removing children, based on loop above |
1044 |
104 |
return returnRewrite(node, retNode, Rewrite::RE_ANDOR_FLATTEN); |
1045 |
|
} |
1046 |
958 |
return node; |
1047 |
|
} |
1048 |
|
|
1049 |
24 |
Node SequencesRewriter::rewriteLoopRegExp(TNode node) |
1050 |
|
{ |
1051 |
24 |
Assert(node.getKind() == REGEXP_LOOP); |
1052 |
48 |
Node retNode = node; |
1053 |
48 |
Node r = node[0]; |
1054 |
24 |
if (r.getKind() == REGEXP_STAR) |
1055 |
|
{ |
1056 |
|
return returnRewrite(node, r, Rewrite::RE_LOOP_STAR); |
1057 |
|
} |
1058 |
24 |
NodeManager* nm = NodeManager::currentNM(); |
1059 |
48 |
cvc5::Rational rMaxInt(String::maxSize()); |
1060 |
24 |
uint32_t l = utils::getLoopMinOccurrences(node); |
1061 |
48 |
std::vector<Node> vec_nodes; |
1062 |
110 |
for (unsigned i = 0; i < l; i++) |
1063 |
|
{ |
1064 |
86 |
vec_nodes.push_back(r); |
1065 |
|
} |
1066 |
|
Node n = |
1067 |
24 |
vec_nodes.size() == 0 |
1068 |
24 |
? nm->mkNode(STRING_TO_REGEXP, nm->mkConst(String(""))) |
1069 |
48 |
: vec_nodes.size() == 1 ? r : nm->mkNode(REGEXP_CONCAT, vec_nodes); |
1070 |
24 |
uint32_t u = utils::getLoopMaxOccurrences(node); |
1071 |
24 |
if (u < l) |
1072 |
|
{ |
1073 |
8 |
std::vector<Node> nvec; |
1074 |
4 |
retNode = nm->mkNode(REGEXP_EMPTY, nvec); |
1075 |
|
} |
1076 |
20 |
else if (u == l) |
1077 |
|
{ |
1078 |
10 |
retNode = n; |
1079 |
|
} |
1080 |
|
else |
1081 |
|
{ |
1082 |
20 |
std::vector<Node> vec2; |
1083 |
10 |
vec2.push_back(n); |
1084 |
20 |
TypeNode rtype = nm->regExpType(); |
1085 |
52 |
for (uint32_t j = l; j < u; j++) |
1086 |
|
{ |
1087 |
42 |
vec_nodes.push_back(r); |
1088 |
42 |
n = utils::mkConcat(vec_nodes, rtype); |
1089 |
42 |
vec2.push_back(n); |
1090 |
|
} |
1091 |
10 |
retNode = nm->mkNode(REGEXP_UNION, vec2); |
1092 |
|
} |
1093 |
48 |
Trace("strings-lp") << "Strings::lp " << node << " => " << retNode |
1094 |
24 |
<< std::endl; |
1095 |
24 |
if (retNode != node) |
1096 |
|
{ |
1097 |
24 |
return returnRewrite(node, retNode, Rewrite::RE_LOOP); |
1098 |
|
} |
1099 |
|
return node; |
1100 |
|
} |
1101 |
|
|
1102 |
4 |
Node SequencesRewriter::rewriteRepeatRegExp(TNode node) |
1103 |
|
{ |
1104 |
4 |
Assert(node.getKind() == REGEXP_REPEAT); |
1105 |
4 |
NodeManager* nm = NodeManager::currentNM(); |
1106 |
|
// ((_ re.^ n) R) --> ((_ re.loop n n) R) |
1107 |
4 |
unsigned r = utils::getRepeatAmount(node); |
1108 |
8 |
Node lop = nm->mkConst(RegExpLoop(r, r)); |
1109 |
8 |
Node retNode = nm->mkNode(REGEXP_LOOP, lop, node[0]); |
1110 |
8 |
return returnRewrite(node, retNode, Rewrite::RE_REPEAT_ELIM); |
1111 |
|
} |
1112 |
|
|
1113 |
31 |
Node SequencesRewriter::rewriteOptionRegExp(TNode node) |
1114 |
|
{ |
1115 |
31 |
Assert(node.getKind() == REGEXP_OPT); |
1116 |
31 |
NodeManager* nm = NodeManager::currentNM(); |
1117 |
|
Node retNode = |
1118 |
|
nm->mkNode(REGEXP_UNION, |
1119 |
62 |
nm->mkNode(STRING_TO_REGEXP, nm->mkConst(String(""))), |
1120 |
124 |
node[0]); |
1121 |
62 |
return returnRewrite(node, retNode, Rewrite::RE_OPT_ELIM); |
1122 |
|
} |
1123 |
|
|
1124 |
45 |
Node SequencesRewriter::rewritePlusRegExp(TNode node) |
1125 |
|
{ |
1126 |
45 |
Assert(node.getKind() == REGEXP_PLUS); |
1127 |
45 |
NodeManager* nm = NodeManager::currentNM(); |
1128 |
|
Node retNode = |
1129 |
90 |
nm->mkNode(REGEXP_CONCAT, node[0], nm->mkNode(REGEXP_STAR, node[0])); |
1130 |
90 |
return returnRewrite(node, retNode, Rewrite::RE_PLUS_ELIM); |
1131 |
|
} |
1132 |
|
|
1133 |
55 |
Node SequencesRewriter::rewriteDifferenceRegExp(TNode node) |
1134 |
|
{ |
1135 |
55 |
Assert(node.getKind() == REGEXP_DIFF); |
1136 |
55 |
NodeManager* nm = NodeManager::currentNM(); |
1137 |
|
Node retNode = |
1138 |
110 |
nm->mkNode(REGEXP_INTER, node[0], nm->mkNode(REGEXP_COMPLEMENT, node[1])); |
1139 |
110 |
return returnRewrite(node, retNode, Rewrite::RE_DIFF_ELIM); |
1140 |
|
} |
1141 |
|
|
1142 |
376 |
Node SequencesRewriter::rewriteRangeRegExp(TNode node) |
1143 |
|
{ |
1144 |
376 |
Assert(node.getKind() == REGEXP_RANGE); |
1145 |
|
unsigned ch[2]; |
1146 |
1128 |
for (size_t i = 0; i < 2; ++i) |
1147 |
|
{ |
1148 |
752 |
if (!node[i].isConst() || node[i].getConst<String>().size() != 1) |
1149 |
|
{ |
1150 |
|
// not applied to characters, it is not handled |
1151 |
|
return node; |
1152 |
|
} |
1153 |
752 |
ch[i] = node[i].getConst<String>().front(); |
1154 |
|
} |
1155 |
|
|
1156 |
376 |
NodeManager* nm = NodeManager::currentNM(); |
1157 |
376 |
if (node[0] == node[1]) |
1158 |
|
{ |
1159 |
|
Node retNode = nm->mkNode(STRING_TO_REGEXP, node[0]); |
1160 |
|
// re.range( "A", "A" ) ---> str.to_re( "A" ) |
1161 |
|
return returnRewrite(node, retNode, Rewrite::RE_RANGE_SINGLE); |
1162 |
|
} |
1163 |
|
|
1164 |
376 |
if (ch[0] > ch[1]) |
1165 |
|
{ |
1166 |
|
// re.range( "B", "A" ) ---> re.none |
1167 |
4 |
Node retNode = nm->mkNode(REGEXP_EMPTY); |
1168 |
2 |
return returnRewrite(node, retNode, Rewrite::RE_RANGE_EMPTY); |
1169 |
|
} |
1170 |
374 |
return node; |
1171 |
|
} |
1172 |
|
|
1173 |
11572 |
Node SequencesRewriter::rewriteMembership(TNode node) |
1174 |
|
{ |
1175 |
11572 |
NodeManager* nm = NodeManager::currentNM(); |
1176 |
23144 |
Node x = node[0]; |
1177 |
23144 |
Node r = node[1]; |
1178 |
|
|
1179 |
23144 |
TypeNode stype = x.getType(); |
1180 |
23144 |
TypeNode rtype = r.getType(); |
1181 |
|
|
1182 |
11572 |
if(r.getKind() == kind::REGEXP_EMPTY) |
1183 |
|
{ |
1184 |
132 |
Node retNode = NodeManager::currentNM()->mkConst(false); |
1185 |
66 |
return returnRewrite(node, retNode, Rewrite::RE_IN_EMPTY); |
1186 |
|
} |
1187 |
11506 |
else if (x.isConst() && RegExpEntail::isConstRegExp(r)) |
1188 |
|
{ |
1189 |
|
// test whether x in node[1] |
1190 |
3882 |
cvc5::String s = x.getConst<String>(); |
1191 |
1941 |
bool test = RegExpEntail::testConstStringInRegExp(s, 0, r); |
1192 |
3882 |
Node retNode = NodeManager::currentNM()->mkConst(test); |
1193 |
1941 |
return returnRewrite(node, retNode, Rewrite::RE_IN_EVAL); |
1194 |
|
} |
1195 |
9565 |
else if (r.getKind() == kind::REGEXP_SIGMA) |
1196 |
|
{ |
1197 |
124 |
Node one = nm->mkConst(Rational(1)); |
1198 |
124 |
Node retNode = one.eqNode(nm->mkNode(STRING_LENGTH, x)); |
1199 |
62 |
return returnRewrite(node, retNode, Rewrite::RE_IN_SIGMA); |
1200 |
|
} |
1201 |
9503 |
else if (r.getKind() == kind::REGEXP_STAR) |
1202 |
|
{ |
1203 |
3145 |
if (x.isConst()) |
1204 |
|
{ |
1205 |
40 |
size_t xlen = Word::getLength(x); |
1206 |
40 |
if (xlen == 0) |
1207 |
|
{ |
1208 |
36 |
Node retNode = nm->mkConst(true); |
1209 |
|
// e.g. (str.in.re "" (re.* (str.to.re x))) ----> true |
1210 |
18 |
return returnRewrite(node, retNode, Rewrite::RE_EMPTY_IN_STR_STAR); |
1211 |
|
} |
1212 |
22 |
else if (xlen == 1) |
1213 |
|
{ |
1214 |
22 |
if (r[0].getKind() == STRING_TO_REGEXP) |
1215 |
|
{ |
1216 |
4 |
Node retNode = r[0][0].eqNode(x); |
1217 |
|
// e.g. (str.in.re "A" (re.* (str.to.re x))) ----> "A" = x |
1218 |
2 |
return returnRewrite(node, retNode, Rewrite::RE_CHAR_IN_STR_STAR); |
1219 |
|
} |
1220 |
|
} |
1221 |
|
} |
1222 |
3105 |
else if (x.getKind() == STRING_CONCAT) |
1223 |
|
{ |
1224 |
|
// (str.in.re (str.++ x1 ... xn) (re.* R)) --> |
1225 |
|
// (str.in.re x1 (re.* R)) AND ... AND (str.in.re xn (re.* R)) |
1226 |
|
// if the length of all strings in R is one. |
1227 |
1193 |
Node flr = RegExpEntail::getFixedLengthForRegexp(r[0]); |
1228 |
692 |
if (!flr.isNull()) |
1229 |
|
{ |
1230 |
261 |
Node one = nm->mkConst(Rational(1)); |
1231 |
226 |
if (flr == one) |
1232 |
|
{ |
1233 |
382 |
NodeBuilder nb(AND); |
1234 |
671 |
for (const Node& xc : x) |
1235 |
|
{ |
1236 |
480 |
nb << nm->mkNode(STRING_IN_REGEXP, xc, r); |
1237 |
|
} |
1238 |
|
return returnRewrite( |
1239 |
191 |
node, nb.constructNode(), Rewrite::RE_IN_DIST_CHAR_STAR); |
1240 |
|
} |
1241 |
|
} |
1242 |
|
} |
1243 |
2934 |
if (r[0].getKind() == kind::REGEXP_SIGMA) |
1244 |
|
{ |
1245 |
202 |
Node retNode = NodeManager::currentNM()->mkConst(true); |
1246 |
101 |
return returnRewrite(node, retNode, Rewrite::RE_IN_SIGMA_STAR); |
1247 |
|
} |
1248 |
|
} |
1249 |
6358 |
else if (r.getKind() == kind::REGEXP_CONCAT) |
1250 |
|
{ |
1251 |
4445 |
bool allSigma = true; |
1252 |
4445 |
bool allSigmaStrict = true; |
1253 |
4445 |
unsigned allSigmaMinSize = 0; |
1254 |
8716 |
Node constStr; |
1255 |
4445 |
size_t constIdx = 0; |
1256 |
4445 |
size_t nchildren = r.getNumChildren(); |
1257 |
10136 |
for (size_t i = 0; i < nchildren; i++) |
1258 |
|
{ |
1259 |
15112 |
Node rc = r[i]; |
1260 |
9421 |
Assert(rc.getKind() != kind::REGEXP_EMPTY); |
1261 |
9421 |
if (rc.getKind() == kind::REGEXP_SIGMA) |
1262 |
|
{ |
1263 |
982 |
allSigmaMinSize++; |
1264 |
|
} |
1265 |
8439 |
else if (rc.getKind() == REGEXP_STAR && rc[0].getKind() == REGEXP_SIGMA) |
1266 |
|
{ |
1267 |
2093 |
allSigmaStrict = false; |
1268 |
|
} |
1269 |
6346 |
else if (rc.getKind() == STRING_TO_REGEXP) |
1270 |
|
{ |
1271 |
3584 |
if (constStr.isNull()) |
1272 |
|
{ |
1273 |
2616 |
constStr = rc[0]; |
1274 |
2616 |
constIdx = i; |
1275 |
|
} |
1276 |
|
else |
1277 |
|
{ |
1278 |
968 |
allSigma = false; |
1279 |
968 |
break; |
1280 |
|
} |
1281 |
|
} |
1282 |
|
else |
1283 |
|
{ |
1284 |
2762 |
allSigma = false; |
1285 |
2762 |
break; |
1286 |
|
} |
1287 |
|
} |
1288 |
4445 |
if (allSigma) |
1289 |
|
{ |
1290 |
715 |
if (constStr.isNull()) |
1291 |
|
{ |
1292 |
|
// x in re.++(_*, _, _) ---> str.len(x) >= 2 |
1293 |
40 |
Node num = nm->mkConst(Rational(allSigmaMinSize)); |
1294 |
40 |
Node lenx = nm->mkNode(STRING_LENGTH, x); |
1295 |
40 |
Node retNode = nm->mkNode(allSigmaStrict ? EQUAL : GEQ, lenx, num); |
1296 |
20 |
return returnRewrite(node, retNode, Rewrite::RE_CONCAT_PURE_ALLCHAR); |
1297 |
|
} |
1298 |
695 |
else if (allSigmaMinSize == 0 && nchildren >= 3 && constIdx != 0 |
1299 |
154 |
&& constIdx != nchildren - 1) |
1300 |
|
{ |
1301 |
|
// x in re.++(_*, "abc", _*) ---> str.contains(x, "abc") |
1302 |
308 |
Node retNode = nm->mkNode(STRING_CONTAINS, x, constStr); |
1303 |
154 |
return returnRewrite(node, retNode, Rewrite::RE_CONCAT_TO_CONTAINS); |
1304 |
|
} |
1305 |
|
} |
1306 |
|
} |
1307 |
3826 |
else if (r.getKind() == kind::REGEXP_INTER |
1308 |
1913 |
|| r.getKind() == kind::REGEXP_UNION) |
1309 |
|
{ |
1310 |
924 |
std::vector<Node> mvec; |
1311 |
1462 |
for (unsigned i = 0; i < r.getNumChildren(); i++) |
1312 |
|
{ |
1313 |
1000 |
mvec.push_back( |
1314 |
2000 |
NodeManager::currentNM()->mkNode(kind::STRING_IN_REGEXP, x, r[i])); |
1315 |
|
} |
1316 |
|
Node retNode = NodeManager::currentNM()->mkNode( |
1317 |
924 |
r.getKind() == kind::REGEXP_INTER ? kind::AND : kind::OR, mvec); |
1318 |
462 |
return returnRewrite(node, retNode, Rewrite::RE_IN_ANDOR); |
1319 |
|
} |
1320 |
1451 |
else if (r.getKind() == kind::STRING_TO_REGEXP) |
1321 |
|
{ |
1322 |
1724 |
Node retNode = x.eqNode(r[0]); |
1323 |
862 |
return returnRewrite(node, retNode, Rewrite::RE_IN_CSTRING); |
1324 |
|
} |
1325 |
589 |
else if (r.getKind() == REGEXP_RANGE) |
1326 |
|
{ |
1327 |
|
// x in re.range( char_i, char_j ) ---> i <= str.code(x) <= j |
1328 |
|
// we do not do this if the arguments are not constant |
1329 |
354 |
if (RegExpEntail::isConstRegExp(r)) |
1330 |
|
{ |
1331 |
708 |
Node xcode = nm->mkNode(STRING_TO_CODE, x); |
1332 |
|
Node retNode = |
1333 |
|
nm->mkNode(AND, |
1334 |
708 |
nm->mkNode(LEQ, nm->mkNode(STRING_TO_CODE, r[0]), xcode), |
1335 |
1416 |
nm->mkNode(LEQ, xcode, nm->mkNode(STRING_TO_CODE, r[1]))); |
1336 |
354 |
return returnRewrite(node, retNode, Rewrite::RE_IN_RANGE); |
1337 |
|
} |
1338 |
|
} |
1339 |
235 |
else if (r.getKind() == REGEXP_COMPLEMENT) |
1340 |
|
{ |
1341 |
410 |
Node retNode = nm->mkNode(STRING_IN_REGEXP, x, r[0]).negate(); |
1342 |
205 |
return returnRewrite(node, retNode, Rewrite::RE_IN_COMPLEMENT); |
1343 |
|
} |
1344 |
|
|
1345 |
|
// do simple consumes |
1346 |
14268 |
Node retNode = node; |
1347 |
7134 |
if (r.getKind() == kind::REGEXP_STAR) |
1348 |
|
{ |
1349 |
8322 |
for (unsigned dir = 0; dir <= 1; dir++) |
1350 |
|
{ |
1351 |
11143 |
std::vector<Node> mchildren; |
1352 |
5654 |
utils::getConcat(x, mchildren); |
1353 |
5654 |
bool success = true; |
1354 |
17208 |
while (success) |
1355 |
|
{ |
1356 |
5800 |
success = false; |
1357 |
11577 |
std::vector<Node> children; |
1358 |
5800 |
utils::getConcat(r[0], children); |
1359 |
11577 |
Node scn = RegExpEntail::simpleRegexpConsume(mchildren, children, dir); |
1360 |
5800 |
if (!scn.isNull()) |
1361 |
|
{ |
1362 |
46 |
Trace("regexp-ext-rewrite") |
1363 |
23 |
<< "Regexp star : const conflict : " << node << std::endl; |
1364 |
23 |
return returnRewrite(node, scn, Rewrite::RE_CONSUME_S_CCONF); |
1365 |
|
} |
1366 |
5777 |
else if (children.empty()) |
1367 |
|
{ |
1368 |
|
// fully consumed one copy of the STAR |
1369 |
154 |
if (mchildren.empty()) |
1370 |
|
{ |
1371 |
|
Trace("regexp-ext-rewrite") |
1372 |
|
<< "Regexp star : full consume : " << node << std::endl; |
1373 |
|
Node ret = NodeManager::currentNM()->mkConst(true); |
1374 |
|
return returnRewrite(node, ret, Rewrite::RE_CONSUME_S_FULL); |
1375 |
|
} |
1376 |
|
else |
1377 |
|
{ |
1378 |
308 |
Node prev = retNode; |
1379 |
462 |
retNode = nm->mkNode( |
1380 |
308 |
STRING_IN_REGEXP, utils::mkConcat(mchildren, stype), r); |
1381 |
|
// Iterate again if the node changed. It may not have changed if |
1382 |
|
// nothing was consumed from mchildren (e.g. if the body of the |
1383 |
|
// re.* accepts the empty string. |
1384 |
154 |
success = (retNode != prev); |
1385 |
|
} |
1386 |
|
} |
1387 |
|
} |
1388 |
5631 |
if (retNode != node) |
1389 |
|
{ |
1390 |
284 |
Trace("regexp-ext-rewrite") << "Regexp star : rewrite " << node |
1391 |
142 |
<< " -> " << retNode << std::endl; |
1392 |
142 |
return returnRewrite(node, retNode, Rewrite::RE_CONSUME_S); |
1393 |
|
} |
1394 |
|
} |
1395 |
|
} |
1396 |
|
else |
1397 |
|
{ |
1398 |
7899 |
std::vector<Node> children; |
1399 |
4301 |
utils::getConcat(r, children); |
1400 |
7899 |
std::vector<Node> mchildren; |
1401 |
4301 |
utils::getConcat(x, mchildren); |
1402 |
4301 |
unsigned prevSize = children.size() + mchildren.size(); |
1403 |
7899 |
Node scn = RegExpEntail::simpleRegexpConsume(mchildren, children); |
1404 |
4301 |
if (!scn.isNull()) |
1405 |
|
{ |
1406 |
306 |
Trace("regexp-ext-rewrite") |
1407 |
153 |
<< "Regexp : const conflict : " << node << std::endl; |
1408 |
153 |
return returnRewrite(node, scn, Rewrite::RE_CONSUME_CCONF); |
1409 |
|
} |
1410 |
4148 |
else if ((children.size() + mchildren.size()) != prevSize) |
1411 |
|
{ |
1412 |
|
// Given a membership (str.++ x1 ... xn) in (re.++ r1 ... rm), |
1413 |
|
// above, we strip components to construct an equivalent membership: |
1414 |
|
// (str.++ xi .. xj) in (re.++ rk ... rl). |
1415 |
1100 |
Node xn = utils::mkConcat(mchildren, stype); |
1416 |
1100 |
Node emptyStr = Word::mkEmptyWord(stype); |
1417 |
550 |
if (children.empty()) |
1418 |
|
{ |
1419 |
|
// If we stripped all components on the right, then the left is |
1420 |
|
// equal to the empty string. |
1421 |
|
// e.g. (str.++ "a" x) in (re.++ (str.to.re "a")) ---> (= x "") |
1422 |
66 |
retNode = xn.eqNode(emptyStr); |
1423 |
|
} |
1424 |
|
else |
1425 |
|
{ |
1426 |
|
// otherwise, construct the updated regular expression |
1427 |
484 |
retNode = |
1428 |
968 |
nm->mkNode(STRING_IN_REGEXP, xn, utils::mkConcat(children, rtype)); |
1429 |
|
} |
1430 |
1100 |
Trace("regexp-ext-rewrite") |
1431 |
550 |
<< "Regexp : rewrite : " << node << " -> " << retNode << std::endl; |
1432 |
550 |
return returnRewrite(node, retNode, Rewrite::RE_SIMPLE_CONSUME); |
1433 |
|
} |
1434 |
|
} |
1435 |
6266 |
return node; |
1436 |
|
} |
1437 |
|
|
1438 |
530658 |
RewriteResponse SequencesRewriter::postRewrite(TNode node) |
1439 |
|
{ |
1440 |
1061316 |
Trace("sequences-postrewrite") |
1441 |
530658 |
<< "Strings::SequencesRewriter::postRewrite start " << node << std::endl; |
1442 |
1061316 |
Node retNode = node; |
1443 |
530658 |
Kind nk = node.getKind(); |
1444 |
530658 |
if (nk == kind::STRING_CONCAT) |
1445 |
|
{ |
1446 |
57858 |
retNode = rewriteConcat(node); |
1447 |
|
} |
1448 |
472800 |
else if (nk == kind::EQUAL) |
1449 |
|
{ |
1450 |
119568 |
retNode = rewriteEquality(node); |
1451 |
|
} |
1452 |
353232 |
else if (nk == kind::STRING_LENGTH) |
1453 |
|
{ |
1454 |
222423 |
retNode = rewriteLength(node); |
1455 |
|
} |
1456 |
130809 |
else if (nk == kind::STRING_CHARAT) |
1457 |
|
{ |
1458 |
107 |
retNode = rewriteCharAt(node); |
1459 |
|
} |
1460 |
130702 |
else if (nk == kind::STRING_SUBSTR) |
1461 |
|
{ |
1462 |
45866 |
retNode = rewriteSubstr(node); |
1463 |
|
} |
1464 |
84836 |
else if (nk == kind::STRING_UPDATE) |
1465 |
|
{ |
1466 |
180 |
retNode = rewriteUpdate(node); |
1467 |
|
} |
1468 |
84656 |
else if (nk == kind::STRING_CONTAINS) |
1469 |
|
{ |
1470 |
26985 |
retNode = rewriteContains(node); |
1471 |
|
} |
1472 |
57671 |
else if (nk == kind::STRING_INDEXOF) |
1473 |
|
{ |
1474 |
5962 |
retNode = rewriteIndexof(node); |
1475 |
|
} |
1476 |
51709 |
else if (nk == kind::STRING_INDEXOF_RE) |
1477 |
|
{ |
1478 |
937 |
retNode = rewriteIndexofRe(node); |
1479 |
|
} |
1480 |
50772 |
else if (nk == kind::STRING_REPLACE) |
1481 |
|
{ |
1482 |
3356 |
retNode = rewriteReplace(node); |
1483 |
|
} |
1484 |
47416 |
else if (nk == kind::STRING_REPLACE_ALL) |
1485 |
|
{ |
1486 |
198 |
retNode = rewriteReplaceAll(node); |
1487 |
|
} |
1488 |
47218 |
else if (nk == kind::STRING_REPLACE_RE) |
1489 |
|
{ |
1490 |
403 |
retNode = rewriteReplaceRe(node); |
1491 |
|
} |
1492 |
46815 |
else if (nk == kind::STRING_REPLACE_RE_ALL) |
1493 |
|
{ |
1494 |
261 |
retNode = rewriteReplaceReAll(node); |
1495 |
|
} |
1496 |
46554 |
else if (nk == STRING_REV) |
1497 |
|
{ |
1498 |
243 |
retNode = rewriteStrReverse(node); |
1499 |
|
} |
1500 |
46311 |
else if (nk == kind::STRING_PREFIX || nk == kind::STRING_SUFFIX) |
1501 |
|
{ |
1502 |
153 |
retNode = rewritePrefixSuffix(node); |
1503 |
|
} |
1504 |
46158 |
else if (nk == kind::STRING_IN_REGEXP) |
1505 |
|
{ |
1506 |
11572 |
retNode = rewriteMembership(node); |
1507 |
|
} |
1508 |
34586 |
else if (nk == REGEXP_CONCAT) |
1509 |
|
{ |
1510 |
2864 |
retNode = rewriteConcatRegExp(node); |
1511 |
|
} |
1512 |
31722 |
else if (nk == REGEXP_UNION || nk == REGEXP_INTER) |
1513 |
|
{ |
1514 |
1080 |
retNode = rewriteAndOrRegExp(node); |
1515 |
|
} |
1516 |
30642 |
else if (nk == REGEXP_DIFF) |
1517 |
|
{ |
1518 |
55 |
retNode = rewriteDifferenceRegExp(node); |
1519 |
|
} |
1520 |
30587 |
else if (nk == REGEXP_STAR) |
1521 |
|
{ |
1522 |
1170 |
retNode = rewriteStarRegExp(node); |
1523 |
|
} |
1524 |
29417 |
else if (nk == REGEXP_PLUS) |
1525 |
|
{ |
1526 |
45 |
retNode = rewritePlusRegExp(node); |
1527 |
|
} |
1528 |
29372 |
else if (nk == REGEXP_OPT) |
1529 |
|
{ |
1530 |
31 |
retNode = rewriteOptionRegExp(node); |
1531 |
|
} |
1532 |
29341 |
else if (nk == REGEXP_RANGE) |
1533 |
|
{ |
1534 |
376 |
retNode = rewriteRangeRegExp(node); |
1535 |
|
} |
1536 |
28965 |
else if (nk == REGEXP_LOOP) |
1537 |
|
{ |
1538 |
24 |
retNode = rewriteLoopRegExp(node); |
1539 |
|
} |
1540 |
28941 |
else if (nk == REGEXP_REPEAT) |
1541 |
|
{ |
1542 |
4 |
retNode = rewriteRepeatRegExp(node); |
1543 |
|
} |
1544 |
28937 |
else if (nk == SEQ_UNIT) |
1545 |
|
{ |
1546 |
403 |
retNode = rewriteSeqUnit(node); |
1547 |
|
} |
1548 |
28534 |
else if (nk == SEQ_NTH || nk == SEQ_NTH_TOTAL) |
1549 |
|
{ |
1550 |
388 |
retNode = rewriteSeqNth(node); |
1551 |
|
} |
1552 |
|
|
1553 |
1061316 |
Trace("sequences-postrewrite") |
1554 |
530658 |
<< "Strings::SequencesRewriter::postRewrite returning " << retNode |
1555 |
530658 |
<< std::endl; |
1556 |
530658 |
if (node != retNode) |
1557 |
|
{ |
1558 |
195184 |
Trace("strings-rewrite-debug") << "Strings::SequencesRewriter::postRewrite " |
1559 |
97592 |
<< node << " to " << retNode << std::endl; |
1560 |
97592 |
return RewriteResponse(REWRITE_AGAIN_FULL, retNode); |
1561 |
|
} |
1562 |
433066 |
Trace("strings-rewrite-nf") << "No rewrites for : " << node << std::endl; |
1563 |
433066 |
return RewriteResponse(REWRITE_DONE, retNode); |
1564 |
|
} |
1565 |
|
|
1566 |
327788 |
RewriteResponse SequencesRewriter::preRewrite(TNode node) |
1567 |
|
{ |
1568 |
327788 |
return RewriteResponse(REWRITE_DONE, node); |
1569 |
|
} |
1570 |
|
|
1571 |
687 |
TrustNode SequencesRewriter::expandDefinition(Node node) |
1572 |
|
{ |
1573 |
1374 |
Trace("strings-exp-def") << "SequencesRewriter::expandDefinition : " << node |
1574 |
687 |
<< std::endl; |
1575 |
|
|
1576 |
687 |
if (node.getKind() == kind::SEQ_NTH) |
1577 |
|
{ |
1578 |
2 |
NodeManager* nm = NodeManager::currentNM(); |
1579 |
4 |
Node s = node[0]; |
1580 |
4 |
Node n = node[1]; |
1581 |
|
// seq.nth(s, n) --> ite(0 <= n < len(s), seq.nth_total(s,n), Uf(s, n)) |
1582 |
|
Node cond = nm->mkNode(AND, |
1583 |
4 |
nm->mkNode(LEQ, nm->mkConst(Rational(0)), n), |
1584 |
8 |
nm->mkNode(LT, n, nm->mkNode(STRING_LENGTH, s))); |
1585 |
4 |
Node ss = nm->mkNode(SEQ_NTH_TOTAL, s, n); |
1586 |
4 |
Node uf = SkolemCache::mkSkolemSeqNth(s.getType(), "Uf"); |
1587 |
4 |
Node u = nm->mkNode(APPLY_UF, uf, s, n); |
1588 |
4 |
Node ret = nm->mkNode(ITE, cond, ss, u); |
1589 |
2 |
Trace("strings-exp-def") << "...return " << ret << std::endl; |
1590 |
2 |
return TrustNode::mkTrustRewrite(node, ret, nullptr); |
1591 |
|
} |
1592 |
685 |
return TrustNode::null(); |
1593 |
|
} |
1594 |
|
|
1595 |
388 |
Node SequencesRewriter::rewriteSeqNth(Node node) |
1596 |
|
{ |
1597 |
388 |
Assert(node.getKind() == SEQ_NTH || node.getKind() == SEQ_NTH_TOTAL); |
1598 |
776 |
Node s = node[0]; |
1599 |
776 |
Node i = node[1]; |
1600 |
388 |
if (s.isConst() && i.isConst()) |
1601 |
|
{ |
1602 |
28 |
size_t len = Word::getLength(s); |
1603 |
28 |
if (i.getConst<Rational>().sgn() != -1) |
1604 |
|
{ |
1605 |
28 |
size_t pos = i.getConst<Rational>().getNumerator().toUnsignedInt(); |
1606 |
28 |
if (pos < len) |
1607 |
|
{ |
1608 |
20 |
std::vector<Node> elements = s.getConst<Sequence>().getVec(); |
1609 |
10 |
const Node& ret = elements[pos]; |
1610 |
10 |
return returnRewrite(node, ret, Rewrite::SEQ_NTH_EVAL); |
1611 |
|
} |
1612 |
|
} |
1613 |
18 |
if (node.getKind() == SEQ_NTH_TOTAL) |
1614 |
|
{ |
1615 |
|
// return arbitrary term |
1616 |
4 |
Node ret = s.getType().getSequenceElementType().mkGroundValue(); |
1617 |
2 |
return returnRewrite(node, ret, Rewrite::SEQ_NTH_TOTAL_OOB); |
1618 |
|
} |
1619 |
|
else |
1620 |
|
{ |
1621 |
16 |
return node; |
1622 |
|
} |
1623 |
|
} |
1624 |
|
else |
1625 |
|
{ |
1626 |
360 |
return node; |
1627 |
|
} |
1628 |
|
} |
1629 |
|
|
1630 |
107 |
Node SequencesRewriter::rewriteCharAt(Node node) |
1631 |
|
{ |
1632 |
107 |
Assert(node.getKind() == STRING_CHARAT); |
1633 |
107 |
NodeManager* nm = NodeManager::currentNM(); |
1634 |
214 |
Node one = nm->mkConst(Rational(1)); |
1635 |
214 |
Node retNode = nm->mkNode(STRING_SUBSTR, node[0], node[1], one); |
1636 |
214 |
return returnRewrite(node, retNode, Rewrite::CHARAT_ELIM); |
1637 |
|
} |
1638 |
|
|
1639 |
45878 |
Node SequencesRewriter::rewriteSubstr(Node node) |
1640 |
|
{ |
1641 |
45878 |
Assert(node.getKind() == kind::STRING_SUBSTR); |
1642 |
|
|
1643 |
45878 |
NodeManager* nm = NodeManager::currentNM(); |
1644 |
45878 |
if (node[0].isConst()) |
1645 |
|
{ |
1646 |
10951 |
if (Word::isEmpty(node[0])) |
1647 |
|
{ |
1648 |
1888 |
Node ret = node[0]; |
1649 |
944 |
return returnRewrite(node, ret, Rewrite::SS_EMPTYSTR); |
1650 |
|
} |
1651 |
|
// rewriting for constant arguments |
1652 |
10007 |
if (node[1].isConst() && node[2].isConst()) |
1653 |
|
{ |
1654 |
10016 |
Node s = node[0]; |
1655 |
10016 |
cvc5::Rational rMaxInt(String::maxSize()); |
1656 |
|
uint32_t start; |
1657 |
5008 |
if (node[1].getConst<Rational>() > rMaxInt) |
1658 |
|
{ |
1659 |
|
// start beyond the maximum size of strings |
1660 |
|
// thus, it must be beyond the end point of this string |
1661 |
8 |
Node ret = Word::mkEmptyWord(node.getType()); |
1662 |
4 |
return returnRewrite(node, ret, Rewrite::SS_CONST_START_MAX_OOB); |
1663 |
|
} |
1664 |
5004 |
else if (node[1].getConst<Rational>().sgn() < 0) |
1665 |
|
{ |
1666 |
|
// start before the beginning of the string |
1667 |
1018 |
Node ret = Word::mkEmptyWord(node.getType()); |
1668 |
509 |
return returnRewrite(node, ret, Rewrite::SS_CONST_START_NEG); |
1669 |
|
} |
1670 |
|
else |
1671 |
|
{ |
1672 |
4495 |
start = node[1].getConst<Rational>().getNumerator().toUnsignedInt(); |
1673 |
4495 |
if (start >= Word::getLength(node[0])) |
1674 |
|
{ |
1675 |
|
// start beyond the end of the string |
1676 |
766 |
Node ret = Word::mkEmptyWord(node.getType()); |
1677 |
383 |
return returnRewrite(node, ret, Rewrite::SS_CONST_START_OOB); |
1678 |
|
} |
1679 |
|
} |
1680 |
4112 |
if (node[2].getConst<Rational>() > rMaxInt) |
1681 |
|
{ |
1682 |
|
// take up to the end of the string |
1683 |
1 |
size_t lenS = Word::getLength(s); |
1684 |
2 |
Node ret = Word::suffix(s, lenS - start); |
1685 |
1 |
return returnRewrite(node, ret, Rewrite::SS_CONST_LEN_MAX_OOB); |
1686 |
|
} |
1687 |
4111 |
else if (node[2].getConst<Rational>().sgn() <= 0) |
1688 |
|
{ |
1689 |
738 |
Node ret = Word::mkEmptyWord(node.getType()); |
1690 |
369 |
return returnRewrite(node, ret, Rewrite::SS_CONST_LEN_NON_POS); |
1691 |
|
} |
1692 |
|
else |
1693 |
|
{ |
1694 |
|
uint32_t len = |
1695 |
3742 |
node[2].getConst<Rational>().getNumerator().toUnsignedInt(); |
1696 |
3742 |
if (start + len > Word::getLength(node[0])) |
1697 |
|
{ |
1698 |
|
// take up to the end of the string |
1699 |
325 |
size_t lenS = Word::getLength(s); |
1700 |
650 |
Node ret = Word::suffix(s, lenS - start); |
1701 |
325 |
return returnRewrite(node, ret, Rewrite::SS_CONST_END_OOB); |
1702 |
|
} |
1703 |
|
else |
1704 |
|
{ |
1705 |
|
// compute the substr using the constant string |
1706 |
6834 |
Node ret = Word::substr(s, start, len); |
1707 |
3417 |
return returnRewrite(node, ret, Rewrite::SS_CONST_SS); |
1708 |
|
} |
1709 |
|
} |
1710 |
|
} |
1711 |
|
} |
1712 |
79852 |
Node zero = nm->mkConst(cvc5::Rational(0)); |
1713 |
|
|
1714 |
|
// if entailed non-positive length or negative start point |
1715 |
39926 |
if (ArithEntail::check(zero, node[1], true)) |
1716 |
|
{ |
1717 |
82 |
Node ret = Word::mkEmptyWord(node.getType()); |
1718 |
41 |
return returnRewrite(node, ret, Rewrite::SS_START_NEG); |
1719 |
|
} |
1720 |
39885 |
else if (ArithEntail::check(zero, node[2])) |
1721 |
|
{ |
1722 |
1570 |
Node ret = Word::mkEmptyWord(node.getType()); |
1723 |
785 |
return returnRewrite(node, ret, Rewrite::SS_LEN_NON_POS); |
1724 |
|
} |
1725 |
|
|
1726 |
39100 |
if (node[0].getKind() == STRING_SUBSTR) |
1727 |
|
{ |
1728 |
|
// (str.substr (str.substr x a b) c d) ---> "" if c >= b |
1729 |
|
// |
1730 |
|
// Note that this rewrite can be generalized to: |
1731 |
|
// |
1732 |
|
// (str.substr x a b) ---> "" if a >= (str.len x) |
1733 |
|
// |
1734 |
|
// This can be done when we generalize our entailment methods to |
1735 |
|
// accept an optional context. Then we could conjecture that |
1736 |
|
// (str.substr x a b) rewrites to "" and do a case analysis: |
1737 |
|
// |
1738 |
|
// - a < 0 or b < 0 (the result is trivially empty in these cases) |
1739 |
|
// - a >= (str.len x) assuming that { a >= 0, b >= 0 } |
1740 |
|
// |
1741 |
|
// For example, for (str.substr (str.substr x a a) a a), we could |
1742 |
|
// then deduce that under those assumptions, "a" is an |
1743 |
|
// over-approximation of the length of (str.substr x a a), which |
1744 |
|
// then allows us to reason that the result of the whole term must |
1745 |
|
// be empty. |
1746 |
3767 |
if (ArithEntail::check(node[1], node[0][2])) |
1747 |
|
{ |
1748 |
68 |
Node ret = Word::mkEmptyWord(node.getType()); |
1749 |
34 |
return returnRewrite(node, ret, Rewrite::SS_START_GEQ_LEN); |
1750 |
|
} |
1751 |
|
} |
1752 |
35333 |
else if (node[0].getKind() == STRING_REPLACE) |
1753 |
|
{ |
1754 |
|
// (str.substr (str.replace x y z) 0 n) |
1755 |
|
// ---> (str.replace (str.substr x 0 n) y z) |
1756 |
|
// if (str.len y) = 1 and (str.len z) = 1 |
1757 |
354 |
if (node[1] == zero) |
1758 |
|
{ |
1759 |
492 |
if (StringsEntail::checkLengthOne(node[0][1], true) |
1760 |
492 |
&& StringsEntail::checkLengthOne(node[0][2], true)) |
1761 |
|
{ |
1762 |
|
Node ret = nm->mkNode( |
1763 |
|
kind::STRING_REPLACE, |
1764 |
8 |
nm->mkNode(kind::STRING_SUBSTR, node[0][0], node[1], node[2]), |
1765 |
|
node[0][1], |
1766 |
16 |
node[0][2]); |
1767 |
4 |
return returnRewrite(node, ret, Rewrite::SUBSTR_REPL_SWAP); |
1768 |
|
} |
1769 |
|
} |
1770 |
|
} |
1771 |
|
|
1772 |
78124 |
std::vector<Node> n1; |
1773 |
39062 |
utils::getConcat(node[0], n1); |
1774 |
78124 |
TypeNode stype = node.getType(); |
1775 |
|
|
1776 |
|
// definite inclusion |
1777 |
39062 |
if (node[1] == zero) |
1778 |
|
{ |
1779 |
27178 |
Node curr = node[2]; |
1780 |
27178 |
std::vector<Node> childrenr; |
1781 |
13956 |
if (StringsEntail::stripSymbolicLength(n1, childrenr, 1, curr)) |
1782 |
|
{ |
1783 |
734 |
if (curr != zero && !n1.empty()) |
1784 |
|
{ |
1785 |
579 |
childrenr.push_back(nm->mkNode( |
1786 |
386 |
kind::STRING_SUBSTR, utils::mkConcat(n1, stype), node[1], curr)); |
1787 |
|
} |
1788 |
1468 |
Node ret = utils::mkConcat(childrenr, stype); |
1789 |
734 |
return returnRewrite(node, ret, Rewrite::SS_LEN_INCLUDE); |
1790 |
|
} |
1791 |
|
} |
1792 |
|
|
1793 |
|
// symbolic length analysis |
1794 |
111727 |
for (unsigned r = 0; r < 2; r++) |
1795 |
|
{ |
1796 |
|
// the amount of characters we can strip |
1797 |
148561 |
Node curr; |
1798 |
75162 |
if (r == 0) |
1799 |
|
{ |
1800 |
38328 |
if (node[1] != zero) |
1801 |
|
{ |
1802 |
|
// strip up to start point off the start of the string |
1803 |
25106 |
curr = node[1]; |
1804 |
|
} |
1805 |
|
} |
1806 |
36834 |
else if (r == 1) |
1807 |
|
{ |
1808 |
|
Node tot_len = |
1809 |
73603 |
Rewriter::rewrite(nm->mkNode(kind::STRING_LENGTH, node[0])); |
1810 |
73603 |
Node end_pt = Rewriter::rewrite(nm->mkNode(kind::PLUS, node[1], node[2])); |
1811 |
36834 |
if (node[2] != tot_len) |
1812 |
|
{ |
1813 |
35108 |
if (ArithEntail::check(node[2], tot_len)) |
1814 |
|
{ |
1815 |
|
// end point beyond end point of string, map to tot_len |
1816 |
80 |
Node ret = nm->mkNode(kind::STRING_SUBSTR, node[0], node[1], tot_len); |
1817 |
40 |
return returnRewrite(node, ret, Rewrite::SS_END_PT_NORM); |
1818 |
|
} |
1819 |
|
else |
1820 |
|
{ |
1821 |
|
// strip up to ( str.len(node[0]) - end_pt ) off the end of the string |
1822 |
35068 |
curr = Rewriter::rewrite(nm->mkNode(kind::MINUS, tot_len, end_pt)); |
1823 |
|
} |
1824 |
|
} |
1825 |
|
|
1826 |
|
// (str.substr s x y) --> "" if x < len(s) |= 0 >= y |
1827 |
|
Node n1_lt_tot_len = |
1828 |
73563 |
Rewriter::rewrite(nm->mkNode(kind::LT, node[1], tot_len)); |
1829 |
36794 |
if (ArithEntail::checkWithAssumption(n1_lt_tot_len, zero, node[2], false)) |
1830 |
|
{ |
1831 |
36 |
Node ret = Word::mkEmptyWord(node.getType()); |
1832 |
18 |
return returnRewrite(node, ret, Rewrite::SS_START_ENTAILS_ZERO_LEN); |
1833 |
|
} |
1834 |
|
|
1835 |
|
// (str.substr s x y) --> "" if 0 < y |= x >= str.len(s) |
1836 |
|
Node non_zero_len = |
1837 |
73545 |
Rewriter::rewrite(nm->mkNode(kind::LT, zero, node[2])); |
1838 |
36776 |
if (ArithEntail::checkWithAssumption( |
1839 |
|
non_zero_len, node[1], tot_len, false)) |
1840 |
|
{ |
1841 |
4 |
Node ret = Word::mkEmptyWord(node.getType()); |
1842 |
2 |
return returnRewrite(node, ret, Rewrite::SS_NON_ZERO_LEN_ENTAILS_OOB); |
1843 |
|
} |
1844 |
|
|
1845 |
|
// (str.substr s x y) --> "" if x >= 0 |= 0 >= str.len(s) |
1846 |
|
Node geq_zero_start = |
1847 |
73543 |
Rewriter::rewrite(nm->mkNode(kind::GEQ, node[1], zero)); |
1848 |
36774 |
if (ArithEntail::checkWithAssumption( |
1849 |
|
geq_zero_start, zero, tot_len, false)) |
1850 |
|
{ |
1851 |
4 |
Node ret = Word::mkEmptyWord(node.getType()); |
1852 |
|
return returnRewrite( |
1853 |
2 |
node, ret, Rewrite::SS_GEQ_ZERO_START_ENTAILS_EMP_S); |
1854 |
|
} |
1855 |
|
|
1856 |
|
// (str.substr s x x) ---> "" if (str.len s) <= 1 |
1857 |
36772 |
if (node[1] == node[2] && StringsEntail::checkLengthOne(node[0])) |
1858 |
|
{ |
1859 |
6 |
Node ret = Word::mkEmptyWord(node.getType()); |
1860 |
3 |
return returnRewrite(node, ret, Rewrite::SS_LEN_ONE_Z_Z); |
1861 |
|
} |
1862 |
|
} |
1863 |
75097 |
if (!curr.isNull()) |
1864 |
|
{ |
1865 |
|
// strip off components while quantity is entailed positive |
1866 |
60149 |
int dir = r == 0 ? 1 : -1; |
1867 |
118600 |
std::vector<Node> childrenr; |
1868 |
60149 |
if (StringsEntail::stripSymbolicLength(n1, childrenr, dir, curr)) |
1869 |
|
{ |
1870 |
1698 |
if (r == 0) |
1871 |
|
{ |
1872 |
|
Node ret = nm->mkNode( |
1873 |
2988 |
kind::STRING_SUBSTR, utils::mkConcat(n1, stype), curr, node[2]); |
1874 |
1494 |
return returnRewrite(node, ret, Rewrite::SS_STRIP_START_PT); |
1875 |
|
} |
1876 |
|
else |
1877 |
|
{ |
1878 |
|
Node ret = nm->mkNode(kind::STRING_SUBSTR, |
1879 |
408 |
utils::mkConcat(n1, stype), |
1880 |
|
node[1], |
1881 |
816 |
node[2]); |
1882 |
204 |
return returnRewrite(node, ret, Rewrite::SS_STRIP_END_PT); |
1883 |
|
} |
1884 |
|
} |
1885 |
|
} |
1886 |
|
} |
1887 |
|
// combine substr |
1888 |
36565 |
if (node[0].getKind() == kind::STRING_SUBSTR) |
1889 |
|
{ |
1890 |
6567 |
Node start_inner = node[0][1]; |
1891 |
6567 |
Node start_outer = node[1]; |
1892 |
3641 |
if (ArithEntail::check(start_outer) && ArithEntail::check(start_inner)) |
1893 |
|
{ |
1894 |
|
// both are positive |
1895 |
|
// thus, start point is definitely start_inner+start_outer. |
1896 |
|
// We can rewrite if it for certain what the length is |
1897 |
|
|
1898 |
|
// the length of a string from the inner substr subtracts the start point |
1899 |
|
// of the outer substr |
1900 |
|
Node len_from_inner = |
1901 |
4003 |
Rewriter::rewrite(nm->mkNode(kind::MINUS, node[0][2], start_outer)); |
1902 |
4003 |
Node len_from_outer = node[2]; |
1903 |
4003 |
Node new_len; |
1904 |
|
// take quantity that is for sure smaller than the other |
1905 |
2359 |
if (len_from_inner == len_from_outer) |
1906 |
|
{ |
1907 |
9 |
new_len = len_from_inner; |
1908 |
|
} |
1909 |
2350 |
else if (ArithEntail::check(len_from_inner, len_from_outer)) |
1910 |
|
{ |
1911 |
499 |
new_len = len_from_outer; |
1912 |
|
} |
1913 |
1851 |
else if (ArithEntail::check(len_from_outer, len_from_inner)) |
1914 |
|
{ |
1915 |
207 |
new_len = len_from_inner; |
1916 |
|
} |
1917 |
2359 |
if (!new_len.isNull()) |
1918 |
|
{ |
1919 |
1430 |
Node new_start = nm->mkNode(kind::PLUS, start_inner, start_outer); |
1920 |
|
Node ret = |
1921 |
1430 |
nm->mkNode(kind::STRING_SUBSTR, node[0][0], new_start, new_len); |
1922 |
715 |
return returnRewrite(node, ret, Rewrite::SS_COMBINE); |
1923 |
|
} |
1924 |
|
} |
1925 |
|
} |
1926 |
35850 |
return node; |
1927 |
|
} |
1928 |
|
|
1929 |
180 |
Node SequencesRewriter::rewriteUpdate(Node node) |
1930 |
|
{ |
1931 |
180 |
Assert(node.getKind() == kind::STRING_UPDATE); |
1932 |
360 |
Node s = node[0]; |
1933 |
180 |
if (s.isConst()) |
1934 |
|
{ |
1935 |
168 |
if (Word::isEmpty(s)) |
1936 |
|
{ |
1937 |
|
return returnRewrite(node, s, Rewrite::UPD_EMPTYSTR); |
1938 |
|
} |
1939 |
|
// rewriting for constant arguments |
1940 |
168 |
if (node[1].isConst()) |
1941 |
|
{ |
1942 |
144 |
cvc5::Rational rMaxInt(String::maxSize()); |
1943 |
124 |
if (node[1].getConst<Rational>() > rMaxInt) |
1944 |
|
{ |
1945 |
|
// start beyond the maximum size of strings |
1946 |
|
// thus, it must be beyond the end point of this string |
1947 |
|
return returnRewrite(node, s, Rewrite::UPD_CONST_INDEX_MAX_OOB); |
1948 |
|
} |
1949 |
124 |
else if (node[1].getConst<Rational>().sgn() < 0) |
1950 |
|
{ |
1951 |
|
// start before the beginning of the string |
1952 |
|
return returnRewrite(node, s, Rewrite::UPD_CONST_INDEX_NEG); |
1953 |
|
} |
1954 |
|
uint32_t start = |
1955 |
124 |
node[1].getConst<Rational>().getNumerator().toUnsignedInt(); |
1956 |
124 |
size_t len = Word::getLength(s); |
1957 |
124 |
if (start >= len) |
1958 |
|
{ |
1959 |
|
// start beyond the end of the string |
1960 |
|
return returnRewrite(node, s, Rewrite::UPD_CONST_INDEX_OOB); |
1961 |
|
} |
1962 |
124 |
if (node[2].isConst()) |
1963 |
|
{ |
1964 |
208 |
Node ret = Word::update(s, start, node[2]); |
1965 |
104 |
return returnRewrite(node, ret, Rewrite::UPD_EVAL); |
1966 |
|
} |
1967 |
|
} |
1968 |
|
} |
1969 |
|
|
1970 |
76 |
return node; |
1971 |
|
} |
1972 |
|
|
1973 |
253525 |
Node SequencesRewriter::rewriteContains(Node node) |
1974 |
|
{ |
1975 |
253525 |
Assert(node.getKind() == kind::STRING_CONTAINS); |
1976 |
253525 |
NodeManager* nm = NodeManager::currentNM(); |
1977 |
|
|
1978 |
253525 |
if (node[0] == node[1]) |
1979 |
|
{ |
1980 |
1326 |
Node ret = NodeManager::currentNM()->mkConst(true); |
1981 |
663 |
return returnRewrite(node, ret, Rewrite::CTN_EQ); |
1982 |
|
} |
1983 |
252862 |
if (node[0].isConst()) |
1984 |
|
{ |
1985 |
71437 |
if (node[1].isConst()) |
1986 |
|
{ |
1987 |
9870 |
Node ret = nm->mkConst(Word::find(node[0], node[1]) != std::string::npos); |
1988 |
4935 |
return returnRewrite(node, ret, Rewrite::CTN_CONST); |
1989 |
|
} |
1990 |
|
else |
1991 |
|
{ |
1992 |
128852 |
Node t = node[1]; |
1993 |
66502 |
if (Word::isEmpty(node[0])) |
1994 |
|
{ |
1995 |
|
Node len1 = |
1996 |
65101 |
NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, node[1]); |
1997 |
32665 |
if (ArithEntail::check(len1, true)) |
1998 |
|
{ |
1999 |
|
// we handle the false case here since the rewrite for equality |
2000 |
|
// uses this function, hence we want to conclude false if possible. |
2001 |
|
// len(x)>0 => contains( "", x ) ---> false |
2002 |
458 |
Node ret = NodeManager::currentNM()->mkConst(false); |
2003 |
229 |
return returnRewrite(node, ret, Rewrite::CTN_LHS_EMPTYSTR); |
2004 |
|
} |
2005 |
|
} |
2006 |
33837 |
else if (StringsEntail::checkLengthOne(t)) |
2007 |
|
{ |
2008 |
7590 |
std::vector<Node> vec = Word::getChars(node[0]); |
2009 |
7590 |
Node emp = Word::mkEmptyWord(t.getType()); |
2010 |
7590 |
NodeBuilder nb(OR); |
2011 |
3795 |
nb << emp.eqNode(t); |
2012 |
8804 |
for (const Node& c : vec) |
2013 |
|
{ |
2014 |
5009 |
Assert(c.getType() == t.getType()); |
2015 |
5009 |
nb << c.eqNode(t); |
2016 |
|
} |
2017 |
|
|
2018 |
|
// str.contains("ABCabc", t) ---> |
2019 |
|
// t = "" v t = "A" v t = "B" v t = "C" v t = "a" v t = "b" v t = "c" |
2020 |
|
// if len(t) <= 1 |
2021 |
7590 |
Node ret = nb; |
2022 |
3795 |
return returnRewrite(node, ret, Rewrite::CTN_SPLIT); |
2023 |
|
} |
2024 |
30042 |
else if (node[1].getKind() == kind::STRING_CONCAT) |
2025 |
|
{ |
2026 |
|
int firstc, lastc; |
2027 |
2823 |
if (!StringsEntail::canConstantContainConcat( |
2028 |
|
node[0], node[1], firstc, lastc)) |
2029 |
|
{ |
2030 |
256 |
Node ret = NodeManager::currentNM()->mkConst(false); |
2031 |
128 |
return returnRewrite(node, ret, Rewrite::CTN_NCONST_CTN_CONCAT); |
2032 |
|
} |
2033 |
|
} |
2034 |
|
} |
2035 |
|
} |
2036 |
243775 |
if (node[1].isConst()) |
2037 |
|
{ |
2038 |
63136 |
size_t len = Word::getLength(node[1]); |
2039 |
63136 |
if (len == 0) |
2040 |
|
{ |
2041 |
|
// contains( x, "" ) ---> true |
2042 |
64776 |
Node ret = NodeManager::currentNM()->mkConst(true); |
2043 |
32388 |
return returnRewrite(node, ret, Rewrite::CTN_RHS_EMPTYSTR); |
2044 |
|
} |
2045 |
30748 |
else if (len == 1) |
2046 |
|
{ |
2047 |
|
// The following rewrites are specific to a single character second |
2048 |
|
// argument of contains, where we can reason that this character is |
2049 |
|
// not split over multiple components in the first argument. |
2050 |
16315 |
if (node[0].getKind() == STRING_CONCAT) |
2051 |
|
{ |
2052 |
2410 |
std::vector<Node> nc1; |
2053 |
1205 |
utils::getConcat(node[0], nc1); |
2054 |
2410 |
NodeBuilder nb(OR); |
2055 |
4381 |
for (const Node& ncc : nc1) |
2056 |
|
{ |
2057 |
3176 |
nb << nm->mkNode(STRING_CONTAINS, ncc, node[1]); |
2058 |
|
} |
2059 |
2410 |
Node ret = nb.constructNode(); |
2060 |
|
// str.contains( x ++ y, "A" ) ---> |
2061 |
|
// str.contains( x, "A" ) OR str.contains( y, "A" ) |
2062 |
1205 |
return returnRewrite(node, ret, Rewrite::CTN_CONCAT_CHAR); |
2063 |
|
} |
2064 |
15110 |
else if (node[0].getKind() == STRING_REPLACE) |
2065 |
|
{ |
2066 |
1043 |
Node rplDomain = d_stringsEntail.checkContains(node[0][1], node[1]); |
2067 |
587 |
if (!rplDomain.isNull() && !rplDomain.getConst<bool>()) |
2068 |
|
{ |
2069 |
262 |
Node d1 = nm->mkNode(STRING_CONTAINS, node[0][0], node[1]); |
2070 |
|
Node d2 = |
2071 |
|
nm->mkNode(AND, |
2072 |
262 |
nm->mkNode(STRING_CONTAINS, node[0][0], node[0][1]), |
2073 |
524 |
nm->mkNode(STRING_CONTAINS, node[0][2], node[1])); |
2074 |
262 |
Node ret = nm->mkNode(OR, d1, d2); |
2075 |
|
// If str.contains( y, "A" ) ---> false, then: |
2076 |
|
// str.contains( str.replace( x, y, z ), "A" ) ---> |
2077 |
|
// str.contains( x, "A" ) OR |
2078 |
|
// ( str.contains( x, y ) AND str.contains( z, "A" ) ) |
2079 |
131 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_CHAR); |
2080 |
|
} |
2081 |
|
} |
2082 |
|
} |
2083 |
|
} |
2084 |
420102 |
std::vector<Node> nc1; |
2085 |
210051 |
utils::getConcat(node[0], nc1); |
2086 |
420102 |
std::vector<Node> nc2; |
2087 |
210051 |
utils::getConcat(node[1], nc2); |
2088 |
|
|
2089 |
|
// component-wise containment |
2090 |
420102 |
std::vector<Node> nc1rb; |
2091 |
420102 |
std::vector<Node> nc1re; |
2092 |
210051 |
if (d_stringsEntail.componentContains(nc1, nc2, nc1rb, nc1re) != -1) |
2093 |
|
{ |
2094 |
7814 |
Node ret = NodeManager::currentNM()->mkConst(true); |
2095 |
3907 |
return returnRewrite(node, ret, Rewrite::CTN_COMPONENT); |
2096 |
|
} |
2097 |
412288 |
TypeNode stype = node[0].getType(); |
2098 |
|
|
2099 |
|
// strip endpoints |
2100 |
412288 |
std::vector<Node> nb; |
2101 |
412288 |
std::vector<Node> ne; |
2102 |
206144 |
if (StringsEntail::stripConstantEndpoints(nc1, nc2, nb, ne)) |
2103 |
|
{ |
2104 |
|
Node ret = NodeManager::currentNM()->mkNode( |
2105 |
610 |
kind::STRING_CONTAINS, utils::mkConcat(nc1, stype), node[1]); |
2106 |
305 |
return returnRewrite(node, ret, Rewrite::CTN_STRIP_ENDPT); |
2107 |
|
} |
2108 |
|
|
2109 |
522368 |
for (const Node& n : nc2) |
2110 |
|
{ |
2111 |
316702 |
if (n.getKind() == kind::STRING_REPLACE) |
2112 |
|
{ |
2113 |
|
// (str.contains x (str.replace y z w)) --> false |
2114 |
|
// if (str.contains x y) = false and (str.contains x w) = false |
2115 |
|
// |
2116 |
|
// Reasoning: (str.contains x y) checks that x does not contain y if the |
2117 |
|
// replacement does not change y. (str.contains x w) checks that if the |
2118 |
|
// replacement changes anything in y, the w makes it impossible for it to |
2119 |
|
// occur in x. |
2120 |
7843 |
Node ctnConst = d_stringsEntail.checkContains(node[0], n[0]); |
2121 |
4008 |
if (!ctnConst.isNull() && !ctnConst.getConst<bool>()) |
2122 |
|
{ |
2123 |
484 |
Node ctnConst2 = d_stringsEntail.checkContains(node[0], n[2]); |
2124 |
246 |
if (!ctnConst2.isNull() && !ctnConst2.getConst<bool>()) |
2125 |
|
{ |
2126 |
16 |
Node res = nm->mkConst(false); |
2127 |
8 |
return returnRewrite(node, res, Rewrite::CTN_RPL_NON_CTN); |
2128 |
|
} |
2129 |
|
} |
2130 |
|
|
2131 |
|
// (str.contains x (str.++ w (str.replace x y x) z)) ---> |
2132 |
|
// (and (= w "") (= x (str.replace x y x)) (= z "")) |
2133 |
|
// |
2134 |
|
// TODO: Remove with under-/over-approximation |
2135 |
4000 |
if (node[0] == n[0] && node[0] == n[2]) |
2136 |
|
{ |
2137 |
330 |
Node ret; |
2138 |
165 |
if (nc2.size() > 1) |
2139 |
|
{ |
2140 |
46 |
Node emp = Word::mkEmptyWord(stype); |
2141 |
46 |
NodeBuilder nb2(kind::AND); |
2142 |
69 |
for (const Node& n2 : nc2) |
2143 |
|
{ |
2144 |
46 |
if (n2 == n) |
2145 |
|
{ |
2146 |
23 |
nb2 << nm->mkNode(kind::EQUAL, node[0], node[1]); |
2147 |
|
} |
2148 |
|
else |
2149 |
|
{ |
2150 |
23 |
nb2 << nm->mkNode(kind::EQUAL, emp, n2); |
2151 |
|
} |
2152 |
|
} |
2153 |
23 |
ret = nb2.constructNode(); |
2154 |
|
} |
2155 |
|
else |
2156 |
|
{ |
2157 |
142 |
ret = nm->mkNode(kind::EQUAL, node[0], node[1]); |
2158 |
|
} |
2159 |
165 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_SELF); |
2160 |
|
} |
2161 |
|
} |
2162 |
|
} |
2163 |
|
|
2164 |
|
// length entailment |
2165 |
411332 |
Node len_n1 = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, node[0]); |
2166 |
411332 |
Node len_n2 = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, node[1]); |
2167 |
205666 |
if (ArithEntail::check(len_n2, len_n1, true)) |
2168 |
|
{ |
2169 |
|
// len( n2 ) > len( n1 ) => contains( n1, n2 ) ---> false |
2170 |
320 |
Node ret = NodeManager::currentNM()->mkConst(false); |
2171 |
160 |
return returnRewrite(node, ret, Rewrite::CTN_LEN_INEQ); |
2172 |
|
} |
2173 |
|
|
2174 |
|
// multi-set reasoning |
2175 |
|
// For example, contains( str.++( x, "b" ), str.++( "a", x ) ) ---> false |
2176 |
|
// since the number of a's in the second argument is greater than the number |
2177 |
|
// of a's in the first argument |
2178 |
205506 |
if (StringsEntail::checkMultisetSubset(node[0], node[1])) |
2179 |
|
{ |
2180 |
96 |
Node ret = nm->mkConst(false); |
2181 |
48 |
return returnRewrite(node, ret, Rewrite::CTN_MSET_NSS); |
2182 |
|
} |
2183 |
|
|
2184 |
205458 |
if (ArithEntail::check(len_n2, len_n1, false)) |
2185 |
|
{ |
2186 |
|
// len( n2 ) >= len( n1 ) => contains( n1, n2 ) ---> n1 = n2 |
2187 |
82460 |
Node ret = node[0].eqNode(node[1]); |
2188 |
41230 |
return returnRewrite(node, ret, Rewrite::CTN_LEN_INEQ_NSTRICT); |
2189 |
|
} |
2190 |
|
|
2191 |
|
// splitting |
2192 |
164228 |
if (node[0].getKind() == kind::STRING_CONCAT) |
2193 |
|
{ |
2194 |
55444 |
if (node[1].isConst()) |
2195 |
|
{ |
2196 |
4424 |
Node t = node[1]; |
2197 |
|
// Below, we are looking for a constant component of node[0] |
2198 |
|
// has no overlap with node[1], which means we can split. |
2199 |
|
// Notice that if the first or last components had no |
2200 |
|
// overlap, these would have been removed by strip |
2201 |
|
// constant endpoints above. |
2202 |
|
// Hence, we consider only the inner children. |
2203 |
4364 |
for (unsigned i = 1; i < (node[0].getNumChildren() - 1); i++) |
2204 |
|
{ |
2205 |
|
// constant contains |
2206 |
2189 |
if (node[0][i].isConst()) |
2207 |
|
{ |
2208 |
|
// if no overlap, we can split into disjunction |
2209 |
271 |
if (Word::noOverlapWith(node[0][i], node[1])) |
2210 |
|
{ |
2211 |
148 |
std::vector<Node> nc0; |
2212 |
74 |
utils::getConcat(node[0], nc0); |
2213 |
148 |
std::vector<Node> spl[2]; |
2214 |
74 |
spl[0].insert(spl[0].end(), nc0.begin(), nc0.begin() + i); |
2215 |
74 |
Assert(i < nc0.size() - 1); |
2216 |
74 |
spl[1].insert(spl[1].end(), nc0.begin() + i + 1, nc0.end()); |
2217 |
|
Node ret = NodeManager::currentNM()->mkNode( |
2218 |
|
kind::OR, |
2219 |
296 |
NodeManager::currentNM()->mkNode(kind::STRING_CONTAINS, |
2220 |
148 |
utils::mkConcat(spl[0], stype), |
2221 |
|
node[1]), |
2222 |
296 |
NodeManager::currentNM()->mkNode(kind::STRING_CONTAINS, |
2223 |
148 |
utils::mkConcat(spl[1], stype), |
2224 |
370 |
node[1])); |
2225 |
74 |
return returnRewrite(node, ret, Rewrite::CTN_SPLIT); |
2226 |
|
} |
2227 |
|
} |
2228 |
|
} |
2229 |
|
} |
2230 |
|
} |
2231 |
108784 |
else if (node[0].getKind() == kind::STRING_SUBSTR) |
2232 |
|
{ |
2233 |
|
// (str.contains (str.substr x n (str.len y)) y) ---> |
2234 |
|
// (= (str.substr x n (str.len y)) y) |
2235 |
|
// |
2236 |
|
// TODO: Remove with under-/over-approximation |
2237 |
13732 |
if (node[0][2] == nm->mkNode(kind::STRING_LENGTH, node[1])) |
2238 |
|
{ |
2239 |
56 |
Node ret = nm->mkNode(kind::EQUAL, node[0], node[1]); |
2240 |
28 |
return returnRewrite(node, ret, Rewrite::CTN_SUBSTR); |
2241 |
|
} |
2242 |
|
} |
2243 |
95052 |
else if (node[0].getKind() == kind::STRING_REPLACE) |
2244 |
|
{ |
2245 |
1892 |
if (node[1].isConst() && node[0][1].isConst() && node[0][2].isConst()) |
2246 |
|
{ |
2247 |
596 |
if (Word::noOverlapWith(node[1], node[0][1]) |
2248 |
596 |
&& Word::noOverlapWith(node[1], node[0][2])) |
2249 |
|
{ |
2250 |
|
// (str.contains (str.replace x c1 c2) c3) ---> (str.contains x c3) |
2251 |
|
// if there is no overlap between c1 and c3 and none between c2 and c3 |
2252 |
4 |
Node ret = nm->mkNode(STRING_CONTAINS, node[0][0], node[1]); |
2253 |
2 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_CNSTS_TO_CTN); |
2254 |
|
} |
2255 |
|
} |
2256 |
|
|
2257 |
1890 |
if (node[0][0] == node[0][2]) |
2258 |
|
{ |
2259 |
|
// (str.contains (str.replace x y x) y) ---> (str.contains x y) |
2260 |
326 |
if (node[0][1] == node[1]) |
2261 |
|
{ |
2262 |
24 |
Node ret = nm->mkNode(kind::STRING_CONTAINS, node[0][0], node[1]); |
2263 |
12 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_TO_CTN); |
2264 |
|
} |
2265 |
|
|
2266 |
|
// (str.contains (str.replace x y x) z) ---> (str.contains x z) |
2267 |
|
// if (str.len z) <= 1 |
2268 |
314 |
if (StringsEntail::checkLengthOne(node[1])) |
2269 |
|
{ |
2270 |
24 |
Node ret = nm->mkNode(kind::STRING_CONTAINS, node[0][0], node[1]); |
2271 |
12 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_LEN_ONE_TO_CTN); |
2272 |
|
} |
2273 |
|
} |
2274 |
|
|
2275 |
|
// (str.contains (str.replace x y z) z) ---> |
2276 |
|
// (or (str.contains x y) (str.contains x z)) |
2277 |
1866 |
if (node[0][2] == node[1]) |
2278 |
|
{ |
2279 |
|
Node ret = |
2280 |
|
nm->mkNode(OR, |
2281 |
144 |
nm->mkNode(STRING_CONTAINS, node[0][0], node[0][1]), |
2282 |
288 |
nm->mkNode(STRING_CONTAINS, node[0][0], node[0][2])); |
2283 |
72 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_TO_CTN_DISJ); |
2284 |
|
} |
2285 |
|
|
2286 |
|
// (str.contains (str.replace x y z) w) ---> |
2287 |
|
// (str.contains (str.replace x y "") w) |
2288 |
|
// if (str.contains z w) ---> false and (str.len w) = 1 |
2289 |
1794 |
if (StringsEntail::checkLengthOne(node[1])) |
2290 |
|
{ |
2291 |
480 |
Node ctn = d_stringsEntail.checkContains(node[0][2], node[1]); |
2292 |
295 |
if (!ctn.isNull() && !ctn.getConst<bool>()) |
2293 |
|
{ |
2294 |
220 |
Node empty = Word::mkEmptyWord(stype); |
2295 |
|
Node ret = nm->mkNode( |
2296 |
|
kind::STRING_CONTAINS, |
2297 |
220 |
nm->mkNode(kind::STRING_REPLACE, node[0][0], node[0][1], empty), |
2298 |
440 |
node[1]); |
2299 |
110 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_SIMP_REPL); |
2300 |
|
} |
2301 |
|
} |
2302 |
|
} |
2303 |
|
|
2304 |
163918 |
if (node[1].getKind() == kind::STRING_REPLACE) |
2305 |
|
{ |
2306 |
|
// (str.contains x (str.replace y x y)) ---> |
2307 |
|
// (str.contains x y) |
2308 |
1961 |
if (node[0] == node[1][1] && node[1][0] == node[1][2]) |
2309 |
|
{ |
2310 |
12 |
Node ret = nm->mkNode(kind::STRING_CONTAINS, node[0], node[1][0]); |
2311 |
6 |
return returnRewrite(node, ret, Rewrite::CTN_REPL); |
2312 |
|
} |
2313 |
|
|
2314 |
|
// (str.contains x (str.replace "" x y)) ---> |
2315 |
|
// (= "" (str.replace "" x y)) |
2316 |
|
// |
2317 |
|
// Note: Length-based reasoning is not sufficient to get this rewrite. We |
2318 |
|
// can neither show that str.len(str.replace("", x, y)) - str.len(x) >= 0 |
2319 |
|
// nor str.len(x) - str.len(str.replace("", x, y)) >= 0 |
2320 |
3908 |
Node emp = Word::mkEmptyWord(stype); |
2321 |
1955 |
if (node[0] == node[1][1] && node[1][0] == emp) |
2322 |
|
{ |
2323 |
4 |
Node ret = nm->mkNode(kind::EQUAL, emp, node[1]); |
2324 |
2 |
return returnRewrite(node, ret, Rewrite::CTN_REPL_EMPTY); |
2325 |
|
} |
2326 |
|
} |
2327 |
|
|
2328 |
163910 |
return node; |
2329 |
|
} |
2330 |
|
|
2331 |
5962 |
Node SequencesRewriter::rewriteIndexof(Node node) |
2332 |
|
{ |
2333 |
5962 |
Assert(node.getKind() == kind::STRING_INDEXOF); |
2334 |
5962 |
NodeManager* nm = NodeManager::currentNM(); |
2335 |
|
|
2336 |
5962 |
if (node[2].isConst() && node[2].getConst<Rational>().sgn() < 0) |
2337 |
|
{ |
2338 |
|
// z<0 implies str.indexof( x, y, z ) --> -1 |
2339 |
354 |
Node negone = nm->mkConst(Rational(-1)); |
2340 |
177 |
return returnRewrite(node, negone, Rewrite::IDOF_NEG); |
2341 |
|
} |
2342 |
|
|
2343 |
|
// the string type |
2344 |
11570 |
TypeNode stype = node[0].getType(); |
2345 |
|
|
2346 |
|
// evaluation and simple cases |
2347 |
11570 |
std::vector<Node> children0; |
2348 |
5785 |
utils::getConcat(node[0], children0); |
2349 |
5785 |
if (children0[0].isConst() && node[1].isConst() && node[2].isConst()) |
2350 |
|
{ |
2351 |
1034 |
cvc5::Rational rMaxInt(cvc5::String::maxSize()); |
2352 |
891 |
if (node[2].getConst<Rational>() > rMaxInt) |
2353 |
|
{ |
2354 |
|
// We know that, due to limitations on the size of string constants |
2355 |
|
// in our implementation, that accessing a position greater than |
2356 |
|
// rMaxInt is guaranteed to be out of bounds. |
2357 |
|
Node negone = nm->mkConst(Rational(-1)); |
2358 |
|
return returnRewrite(node, negone, Rewrite::IDOF_MAX); |
2359 |
|
} |
2360 |
891 |
Assert(node[2].getConst<Rational>().sgn() >= 0); |
2361 |
1034 |
Node s = children0[0]; |
2362 |
1034 |
Node t = node[1]; |
2363 |
|
uint32_t start = |
2364 |
891 |
node[2].getConst<Rational>().getNumerator().toUnsignedInt(); |
2365 |
891 |
std::size_t ret = Word::find(s, t, start); |
2366 |
891 |
if (ret != std::string::npos) |
2367 |
|
{ |
2368 |
656 |
Node retv = nm->mkConst(Rational(static_cast<unsigned>(ret))); |
2369 |
328 |
return returnRewrite(node, retv, Rewrite::IDOF_FIND); |
2370 |
|
} |
2371 |
563 |
else if (children0.size() == 1) |
2372 |
|
{ |
2373 |
840 |
Node negone = nm->mkConst(Rational(-1)); |
2374 |
420 |
return returnRewrite(node, negone, Rewrite::IDOF_NFIND); |
2375 |
|
} |
2376 |
|
} |
2377 |
|
|
2378 |
5037 |
if (node[0] == node[1]) |
2379 |
|
{ |
2380 |
102 |
if (node[2].isConst()) |
2381 |
|
{ |
2382 |
18 |
if (node[2].getConst<Rational>().sgn() == 0) |
2383 |
|
{ |
2384 |
|
// indexof( x, x, 0 ) --> 0 |
2385 |
36 |
Node zero = nm->mkConst(Rational(0)); |
2386 |
18 |
return returnRewrite(node, zero, Rewrite::IDOF_EQ_CST_START); |
2387 |
|
} |
2388 |
|
} |
2389 |
84 |
if (ArithEntail::check(node[2], true)) |
2390 |
|
{ |
2391 |
|
// y>0 implies indexof( x, x, y ) --> -1 |
2392 |
|
Node negone = nm->mkConst(Rational(-1)); |
2393 |
|
return returnRewrite(node, negone, Rewrite::IDOF_EQ_NSTART); |
2394 |
|
} |
2395 |
140 |
Node emp = Word::mkEmptyWord(stype); |
2396 |
84 |
if (node[0] != emp) |
2397 |
|
{ |
2398 |
|
// indexof( x, x, z ) ---> indexof( "", "", z ) |
2399 |
56 |
Node ret = nm->mkNode(STRING_INDEXOF, emp, emp, node[2]); |
2400 |
28 |
return returnRewrite(node, ret, Rewrite::IDOF_EQ_NORM); |
2401 |
|
} |
2402 |
|
} |
2403 |
|
|
2404 |
9982 |
Node len0 = nm->mkNode(STRING_LENGTH, node[0]); |
2405 |
9982 |
Node len1 = nm->mkNode(STRING_LENGTH, node[1]); |
2406 |
9982 |
Node len0m2 = nm->mkNode(MINUS, len0, node[2]); |
2407 |
|
|
2408 |
4991 |
if (node[1].isConst()) |
2409 |
|
{ |
2410 |
4441 |
if (Word::isEmpty(node[1])) |
2411 |
|
{ |
2412 |
82 |
if (ArithEntail::check(len0, node[2]) && ArithEntail::check(node[2])) |
2413 |
|
{ |
2414 |
|
// len(x)>=z ^ z >=0 implies indexof( x, "", z ) ---> z |
2415 |
10 |
return returnRewrite(node, node[2], Rewrite::IDOF_EMP_IDOF); |
2416 |
|
} |
2417 |
|
} |
2418 |
|
} |
2419 |
|
|
2420 |
4981 |
if (ArithEntail::check(len1, len0m2, true)) |
2421 |
|
{ |
2422 |
|
// len(x)-z < len(y) implies indexof( x, y, z ) ----> -1 |
2423 |
16 |
Node negone = nm->mkConst(Rational(-1)); |
2424 |
8 |
return returnRewrite(node, negone, Rewrite::IDOF_LEN); |
2425 |
|
} |
2426 |
|
|
2427 |
9946 |
Node fstr = node[0]; |
2428 |
4973 |
if (!node[2].isConst() || node[2].getConst<Rational>().sgn() != 0) |
2429 |
|
{ |
2430 |
1072 |
fstr = nm->mkNode(kind::STRING_SUBSTR, node[0], node[2], len0); |
2431 |
1072 |
fstr = Rewriter::rewrite(fstr); |
2432 |
|
} |
2433 |
|
|
2434 |
9946 |
Node cmp_conr = d_stringsEntail.checkContains(fstr, node[1]); |
2435 |
9946 |
Trace("strings-rewrite-debug") << "For " << node << ", check contains(" |
2436 |
4973 |
<< fstr << ", " << node[1] << ")" << std::endl; |
2437 |
4973 |
Trace("strings-rewrite-debug") << "...got " << cmp_conr << std::endl; |
2438 |
9946 |
std::vector<Node> children1; |
2439 |
4973 |
utils::getConcat(node[1], children1); |
2440 |
4973 |
if (!cmp_conr.isNull()) |
2441 |
|
{ |
2442 |
282 |
if (cmp_conr.getConst<bool>()) |
2443 |
|
{ |
2444 |
277 |
if (node[2].isConst() && node[2].getConst<Rational>().sgn() == 0) |
2445 |
|
{ |
2446 |
|
// past the first position in node[0] that contains node[1], we can drop |
2447 |
335 |
std::vector<Node> nb; |
2448 |
335 |
std::vector<Node> ne; |
2449 |
199 |
int cc = d_stringsEntail.componentContains( |
2450 |
199 |
children0, children1, nb, ne, true, 1); |
2451 |
199 |
if (cc != -1 && !ne.empty()) |
2452 |
|
{ |
2453 |
|
// For example: |
2454 |
|
// str.indexof(str.++(x,y,z),y,0) ---> str.indexof(str.++(x,y),y,0) |
2455 |
102 |
Node nn = utils::mkConcat(children0, stype); |
2456 |
102 |
Node ret = nm->mkNode(kind::STRING_INDEXOF, nn, node[1], node[2]); |
2457 |
51 |
return returnRewrite(node, ret, Rewrite::IDOF_DEF_CTN); |
2458 |
|
} |
2459 |
|
|
2460 |
|
// Strip components from the beginning that are guaranteed not to match |
2461 |
148 |
if (StringsEntail::stripConstantEndpoints( |
2462 |
|
children0, children1, nb, ne, 1)) |
2463 |
|
{ |
2464 |
|
// str.indexof(str.++("AB", x, "C"), "C", 0) ---> |
2465 |
|
// 2 + str.indexof(str.++(x, "C"), "C", 0) |
2466 |
|
Node ret = nm->mkNode( |
2467 |
|
kind::PLUS, |
2468 |
24 |
nm->mkNode(kind::STRING_LENGTH, utils::mkConcat(nb, stype)), |
2469 |
48 |
nm->mkNode(kind::STRING_INDEXOF, |
2470 |
24 |
utils::mkConcat(children0, stype), |
2471 |
|
node[1], |
2472 |
48 |
node[2])); |
2473 |
12 |
return returnRewrite(node, ret, Rewrite::IDOF_STRIP_CNST_ENDPTS); |
2474 |
|
} |
2475 |
|
} |
2476 |
|
// To show that the first argument definitely contains the second, the |
2477 |
|
// index must be a valid index in the first argument. This ensures that |
2478 |
|
// (str.indexof t "" n) is not rewritten to something other than -1 when n |
2479 |
|
// is beyond the length of t. This is not required for the above rewrites, |
2480 |
|
// which only apply when n=0. |
2481 |
214 |
if (ArithEntail::check(node[2]) && ArithEntail::check(len0, node[2])) |
2482 |
|
{ |
2483 |
|
// strip symbolic length |
2484 |
278 |
Node new_len = node[2]; |
2485 |
278 |
std::vector<Node> nr; |
2486 |
142 |
if (StringsEntail::stripSymbolicLength(children0, nr, 1, new_len)) |
2487 |
|
{ |
2488 |
|
// For example: |
2489 |
|
// z>=0 and z>str.len( x1 ) and str.contains( x2, y )-->true |
2490 |
|
// implies |
2491 |
|
// str.indexof( str.++( x1, x2 ), y, z ) ---> |
2492 |
|
// str.len( x1 ) + str.indexof( x2, y, z-str.len(x1) ) |
2493 |
12 |
Node nn = utils::mkConcat(children0, stype); |
2494 |
|
Node ret = |
2495 |
|
nm->mkNode(PLUS, |
2496 |
12 |
nm->mkNode(MINUS, node[2], new_len), |
2497 |
24 |
nm->mkNode(STRING_INDEXOF, nn, node[1], new_len)); |
2498 |
6 |
return returnRewrite(node, ret, Rewrite::IDOF_STRIP_SYM_LEN); |
2499 |
|
} |
2500 |
|
} |
2501 |
|
} |
2502 |
|
else |
2503 |
|
{ |
2504 |
|
// str.contains( x, y ) --> false implies str.indexof(x,y,z) --> -1 |
2505 |
10 |
Node negone = nm->mkConst(Rational(-1)); |
2506 |
5 |
return returnRewrite(node, negone, Rewrite::IDOF_NCTN); |
2507 |
|
} |
2508 |
|
} |
2509 |
|
else |
2510 |
|
{ |
2511 |
9349 |
Node new_len = node[2]; |
2512 |
9349 |
std::vector<Node> nr; |
2513 |
4691 |
if (StringsEntail::stripSymbolicLength(children0, nr, 1, new_len)) |
2514 |
|
{ |
2515 |
|
// Normalize the string before the start index. |
2516 |
|
// |
2517 |
|
// For example: |
2518 |
|
// str.indexof(str.++("ABCD", x), y, 3) ---> |
2519 |
|
// str.indexof(str.++("AAAD", x), y, 3) |
2520 |
121 |
Node nodeNr = utils::mkConcat(nr, stype); |
2521 |
121 |
Node normNr = lengthPreserveRewrite(nodeNr); |
2522 |
77 |
if (normNr != nodeNr) |
2523 |
|
{ |
2524 |
66 |
std::vector<Node> normNrChildren; |
2525 |
33 |
utils::getConcat(normNr, normNrChildren); |
2526 |
66 |
std::vector<Node> children(normNrChildren); |
2527 |
33 |
children.insert(children.end(), children0.begin(), children0.end()); |
2528 |
66 |
Node nn = utils::mkConcat(children, stype); |
2529 |
66 |
Node res = nm->mkNode(kind::STRING_INDEXOF, nn, node[1], node[2]); |
2530 |
33 |
return returnRewrite(node, res, Rewrite::IDOF_NORM_PREFIX); |
2531 |
|
} |
2532 |
|
} |
2533 |
|
} |
2534 |
|
|
2535 |
4866 |
if (node[2].isConst() && node[2].getConst<Rational>().sgn() == 0) |
2536 |
|
{ |
2537 |
7668 |
std::vector<Node> cb; |
2538 |
7668 |
std::vector<Node> ce; |
2539 |
3838 |
if (StringsEntail::stripConstantEndpoints(children0, children1, cb, ce, -1)) |
2540 |
|
{ |
2541 |
16 |
Node ret = utils::mkConcat(children0, stype); |
2542 |
8 |
ret = nm->mkNode(STRING_INDEXOF, ret, node[1], node[2]); |
2543 |
|
// For example: |
2544 |
|
// str.indexof( str.++( x, "A" ), "B", 0 ) ---> str.indexof( x, "B", 0 ) |
2545 |
8 |
return returnRewrite(node, ret, Rewrite::RPL_PULL_ENDPT); |
2546 |
|
} |
2547 |
|
} |
2548 |
|
|
2549 |
4858 |
return node; |
2550 |
|
} |
2551 |
|
|
2552 |
937 |
Node SequencesRewriter::rewriteIndexofRe(Node node) |
2553 |
|
{ |
2554 |
937 |
Assert(node.getKind() == STRING_INDEXOF_RE); |
2555 |
937 |
NodeManager* nm = NodeManager::currentNM(); |
2556 |
1874 |
Node s = node[0]; |
2557 |
1874 |
Node r = node[1]; |
2558 |
1874 |
Node n = node[2]; |
2559 |
1874 |
Node zero = nm->mkConst(Rational(0)); |
2560 |
1874 |
Node slen = nm->mkNode(STRING_LENGTH, s); |
2561 |
|
|
2562 |
937 |
if (ArithEntail::check(zero, n, true) || ArithEntail::check(n, slen, true)) |
2563 |
|
{ |
2564 |
68 |
Node ret = nm->mkConst(Rational(-1)); |
2565 |
34 |
return returnRewrite(node, ret, Rewrite::INDEXOF_RE_INVALID_INDEX); |
2566 |
|
} |
2567 |
|
|
2568 |
903 |
if (RegExpEntail::isConstRegExp(r)) |
2569 |
|
{ |
2570 |
635 |
if (s.isConst() && n.isConst()) |
2571 |
|
{ |
2572 |
310 |
Rational nrat = n.getConst<Rational>(); |
2573 |
310 |
cvc5::Rational rMaxInt(cvc5::String::maxSize()); |
2574 |
155 |
if (nrat > rMaxInt) |
2575 |
|
{ |
2576 |
|
// We know that, due to limitations on the size of string constants |
2577 |
|
// in our implementation, that accessing a position greater than |
2578 |
|
// rMaxInt is guaranteed to be out of bounds. |
2579 |
|
Node negone = nm->mkConst(Rational(-1)); |
2580 |
|
return returnRewrite(node, negone, Rewrite::INDEXOF_RE_MAX_INDEX); |
2581 |
|
} |
2582 |
|
|
2583 |
155 |
uint32_t start = nrat.getNumerator().toUnsignedInt(); |
2584 |
310 |
Node rem = nm->mkConst(s.getConst<String>().substr(start)); |
2585 |
155 |
std::pair<size_t, size_t> match = firstMatch(rem, r); |
2586 |
|
Node ret = nm->mkConst( |
2587 |
405 |
Rational(match.first == string::npos |
2588 |
|
? -1 |
2589 |
405 |
: static_cast<int64_t>(start + match.first))); |
2590 |
155 |
return returnRewrite(node, ret, Rewrite::INDEXOF_RE_EVAL); |
2591 |
|
} |
2592 |
|
|
2593 |
480 |
if (ArithEntail::check(n, zero) && ArithEntail::check(slen, n)) |
2594 |
|
{ |
2595 |
676 |
String emptyStr(""); |
2596 |
342 |
if (RegExpEntail::testConstStringInRegExp(emptyStr, 0, r)) |
2597 |
|
{ |
2598 |
8 |
return returnRewrite(node, n, Rewrite::INDEXOF_RE_EMP_RE); |
2599 |
|
} |
2600 |
|
} |
2601 |
|
} |
2602 |
740 |
return node; |
2603 |
|
} |
2604 |
|
|
2605 |
3356 |
Node SequencesRewriter::rewriteReplace(Node node) |
2606 |
|
{ |
2607 |
3356 |
Assert(node.getKind() == kind::STRING_REPLACE); |
2608 |
3356 |
NodeManager* nm = NodeManager::currentNM(); |
2609 |
|
|
2610 |
3356 |
if (node[1].isConst() && Word::isEmpty(node[1])) |
2611 |
|
{ |
2612 |
846 |
Node ret = nm->mkNode(STRING_CONCAT, node[2], node[0]); |
2613 |
423 |
return returnRewrite(node, ret, Rewrite::RPL_RPL_EMPTY); |
2614 |
|
} |
2615 |
|
// the string type |
2616 |
5866 |
TypeNode stype = node.getType(); |
2617 |
|
|
2618 |
5866 |
std::vector<Node> children0; |
2619 |
2933 |
utils::getConcat(node[0], children0); |
2620 |
|
|
2621 |
2933 |
if (node[1].isConst() && children0[0].isConst()) |
2622 |
|
{ |
2623 |
885 |
Node s = children0[0]; |
2624 |
885 |
Node t = node[1]; |
2625 |
876 |
std::size_t p = Word::find(s, t); |
2626 |
876 |
if (p == std::string::npos) |
2627 |
|
{ |
2628 |
556 |
if (children0.size() == 1) |
2629 |
|
{ |
2630 |
547 |
return returnRewrite(node, node[0], Rewrite::RPL_CONST_NFIND); |
2631 |
|
} |
2632 |
|
} |
2633 |
|
else |
2634 |
|
{ |
2635 |
640 |
Node s1 = Word::substr(s, 0, p); |
2636 |
640 |
Node s3 = Word::substr(s, p + Word::getLength(t)); |
2637 |
640 |
std::vector<Node> children; |
2638 |
320 |
if (!Word::isEmpty(s1)) |
2639 |
|
{ |
2640 |
51 |
children.push_back(s1); |
2641 |
|
} |
2642 |
320 |
children.push_back(node[2]); |
2643 |
320 |
if (!Word::isEmpty(s3)) |
2644 |
|
{ |
2645 |
90 |
children.push_back(s3); |
2646 |
|
} |
2647 |
320 |
children.insert(children.end(), children0.begin() + 1, children0.end()); |
2648 |
640 |
Node ret = utils::mkConcat(children, stype); |
2649 |
320 |
return returnRewrite(node, ret, Rewrite::RPL_CONST_FIND); |
2650 |
|
} |
2651 |
|
} |
2652 |
|
|
2653 |
|
// rewrites that apply to both replace and replaceall |
2654 |
4132 |
Node rri = rewriteReplaceInternal(node); |
2655 |
2066 |
if (!rri.isNull()) |
2656 |
|
{ |
2657 |
|
// printing of the rewrite managed by the call above |
2658 |
135 |
return rri; |
2659 |
|
} |
2660 |
|
|
2661 |
1931 |
if (node[0] == node[2]) |
2662 |
|
{ |
2663 |
|
// ( len( y )>=len(x) ) => str.replace( x, y, x ) ---> x |
2664 |
617 |
Node l0 = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, node[0]); |
2665 |
617 |
Node l1 = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, node[1]); |
2666 |
354 |
if (ArithEntail::check(l1, l0)) |
2667 |
|
{ |
2668 |
79 |
return returnRewrite(node, node[0], Rewrite::RPL_RPL_LEN_ID); |
2669 |
|
} |
2670 |
|
|
2671 |
|
// (str.replace x y x) ---> (str.replace x (str.++ y1 ... yn) x) |
2672 |
|
// if 1 >= (str.len x) and (= y "") ---> (= y1 "") ... (= yn "") |
2673 |
275 |
if (StringsEntail::checkLengthOne(node[0])) |
2674 |
|
{ |
2675 |
176 |
Node empty = Word::mkEmptyWord(stype); |
2676 |
|
Node rn1 = Rewriter::rewrite( |
2677 |
176 |
rewriteEqualityExt(nm->mkNode(EQUAL, node[1], empty))); |
2678 |
94 |
if (rn1 != node[1]) |
2679 |
|
{ |
2680 |
176 |
std::vector<Node> emptyNodes; |
2681 |
|
bool allEmptyEqs; |
2682 |
94 |
std::tie(allEmptyEqs, emptyNodes) = utils::collectEmptyEqs(rn1); |
2683 |
|
|
2684 |
94 |
if (allEmptyEqs) |
2685 |
|
{ |
2686 |
108 |
Node nn1 = utils::mkConcat(emptyNodes, stype); |
2687 |
60 |
if (node[1] != nn1) |
2688 |
|
{ |
2689 |
24 |
Node ret = nm->mkNode(STRING_REPLACE, node[0], nn1, node[2]); |
2690 |
12 |
return returnRewrite(node, ret, Rewrite::RPL_X_Y_X_SIMP); |
2691 |
|
} |
2692 |
|
} |
2693 |
|
} |
2694 |
|
} |
2695 |
|
} |
2696 |
|
|
2697 |
3680 |
std::vector<Node> children1; |
2698 |
1840 |
utils::getConcat(node[1], children1); |
2699 |
|
|
2700 |
|
// check if contains definitely does (or does not) hold |
2701 |
3680 |
Node cmp_con = nm->mkNode(kind::STRING_CONTAINS, node[0], node[1]); |
2702 |
3680 |
Node cmp_conr = Rewriter::rewrite(cmp_con); |
2703 |
1840 |
if (cmp_conr.isConst()) |
2704 |
|
{ |
2705 |
211 |
if (cmp_conr.getConst<bool>()) |
2706 |
|
{ |
2707 |
|
// component-wise containment |
2708 |
367 |
std::vector<Node> cb; |
2709 |
367 |
std::vector<Node> ce; |
2710 |
205 |
int cc = d_stringsEntail.componentContains( |
2711 |
205 |
children0, children1, cb, ce, true, 1); |
2712 |
205 |
if (cc != -1) |
2713 |
|
{ |
2714 |
103 |
if (cc == 0 && children0[0] == children1[0]) |
2715 |
|
{ |
2716 |
|
// definitely a prefix, can do the replace |
2717 |
|
// for example, |
2718 |
|
// str.replace( str.++( x, "ab" ), str.++( x, "a" ), y ) ---> |
2719 |
|
// str.++( y, "b" ) |
2720 |
56 |
std::vector<Node> cres; |
2721 |
28 |
cres.push_back(node[2]); |
2722 |
28 |
cres.insert(cres.end(), ce.begin(), ce.end()); |
2723 |
56 |
Node ret = utils::mkConcat(cres, stype); |
2724 |
28 |
return returnRewrite(node, ret, Rewrite::RPL_CCTN_RPL); |
2725 |
|
} |
2726 |
75 |
else if (!ce.empty()) |
2727 |
|
{ |
2728 |
|
// we can pull remainder past first definite containment |
2729 |
|
// for example, |
2730 |
|
// str.replace( str.++( x, "ab" ), "a", y ) ---> |
2731 |
|
// str.++( str.replace( str.++( x, "a" ), "a", y ), "b" ) |
2732 |
|
// this is independent of whether the second argument may be empty |
2733 |
30 |
std::vector<Node> scc; |
2734 |
45 |
scc.push_back(NodeManager::currentNM()->mkNode( |
2735 |
|
kind::STRING_REPLACE, |
2736 |
30 |
utils::mkConcat(children0, stype), |
2737 |
|
node[1], |
2738 |
|
node[2])); |
2739 |
15 |
scc.insert(scc.end(), ce.begin(), ce.end()); |
2740 |
30 |
Node ret = utils::mkConcat(scc, stype); |
2741 |
15 |
return returnRewrite(node, ret, Rewrite::RPL_CCTN); |
2742 |
|
} |
2743 |
|
} |
2744 |
|
} |
2745 |
|
else |
2746 |
|
{ |
2747 |
|
// ~contains( t, s ) => ( replace( t, s, r ) ----> t ) |
2748 |
6 |
return returnRewrite(node, node[0], Rewrite::RPL_NCTN); |
2749 |
|
} |
2750 |
|
} |
2751 |
1629 |
else if (cmp_conr.getKind() == kind::EQUAL || cmp_conr.getKind() == kind::AND) |
2752 |
|
{ |
2753 |
|
// Rewriting the str.contains may return equalities of the form (= x ""). |
2754 |
|
// In that case, we can substitute the variables appearing in those |
2755 |
|
// equalities with the empty string in the third argument of the |
2756 |
|
// str.replace. For example: |
2757 |
|
// |
2758 |
|
// (str.replace x (str.++ x y) y) --> (str.replace x (str.++ x y) "") |
2759 |
|
// |
2760 |
|
// This can be done because str.replace changes x iff (str.++ x y) is in x |
2761 |
|
// but that means that y must be empty in that case. Thus, we can |
2762 |
|
// substitute y with "" in the third argument. Note that the third argument |
2763 |
|
// does not matter when the str.replace does not apply. |
2764 |
|
// |
2765 |
618 |
Node empty = Word::mkEmptyWord(stype); |
2766 |
|
|
2767 |
618 |
std::vector<Node> emptyNodes; |
2768 |
|
bool allEmptyEqs; |
2769 |
318 |
std::tie(allEmptyEqs, emptyNodes) = utils::collectEmptyEqs(cmp_conr); |
2770 |
|
|
2771 |
318 |
if (emptyNodes.size() > 0) |
2772 |
|
{ |
2773 |
|
// Perform the substitutions |
2774 |
346 |
std::vector<TNode> substs(emptyNodes.size(), TNode(empty)); |
2775 |
|
Node nn2 = node[2].substitute( |
2776 |
346 |
emptyNodes.begin(), emptyNodes.end(), substs.begin(), substs.end()); |
2777 |
|
|
2778 |
|
// If the contains rewrites to a conjunction of empty-string equalities |
2779 |
|
// and we are doing the replacement in an empty string, we can rewrite |
2780 |
|
// the string-to-replace with a concatenation of all the terms that must |
2781 |
|
// be empty: |
2782 |
|
// |
2783 |
|
// (str.replace "" y z) ---> (str.replace "" (str.++ y1 ... yn) z) |
2784 |
|
// if (str.contains "" y) ---> (and (= y1 "") ... (= yn "")) |
2785 |
182 |
if (node[0] == empty && allEmptyEqs) |
2786 |
|
{ |
2787 |
192 |
std::vector<Node> emptyNodesList(emptyNodes.begin(), emptyNodes.end()); |
2788 |
192 |
Node nn1 = utils::mkConcat(emptyNodesList, stype); |
2789 |
100 |
if (nn1 != node[1] || nn2 != node[2]) |
2790 |
|
{ |
2791 |
16 |
Node res = nm->mkNode(kind::STRING_REPLACE, node[0], nn1, nn2); |
2792 |
8 |
return returnRewrite(node, res, Rewrite::RPL_EMP_CNTS_SUBSTS); |
2793 |
|
} |
2794 |
|
} |
2795 |
|
|
2796 |
174 |
if (nn2 != node[2]) |
2797 |
|
{ |
2798 |
20 |
Node res = nm->mkNode(kind::STRING_REPLACE, node[0], node[1], nn2); |
2799 |
10 |
return returnRewrite(node, res, Rewrite::RPL_CNTS_SUBSTS); |
2800 |
|
} |
2801 |
|
} |
2802 |
|
} |
2803 |
|
|
2804 |
1773 |
if (cmp_conr != cmp_con) |
2805 |
|
{ |
2806 |
589 |
if (StringsEntail::checkNonEmpty(node[1])) |
2807 |
|
{ |
2808 |
|
// pull endpoints that can be stripped |
2809 |
|
// for example, |
2810 |
|
// str.replace( str.++( "b", x, "b" ), "a", y ) ---> |
2811 |
|
// str.++( "b", str.replace( x, "a", y ), "b" ) |
2812 |
657 |
std::vector<Node> cb; |
2813 |
657 |
std::vector<Node> ce; |
2814 |
337 |
if (StringsEntail::stripConstantEndpoints(children0, children1, cb, ce)) |
2815 |
|
{ |
2816 |
34 |
std::vector<Node> cc; |
2817 |
17 |
cc.insert(cc.end(), cb.begin(), cb.end()); |
2818 |
17 |
cc.push_back( |
2819 |
68 |
NodeManager::currentNM()->mkNode(kind::STRING_REPLACE, |
2820 |
34 |
utils::mkConcat(children0, stype), |
2821 |
|
node[1], |
2822 |
|
node[2])); |
2823 |
17 |
cc.insert(cc.end(), ce.begin(), ce.end()); |
2824 |
34 |
Node ret = utils::mkConcat(cc, stype); |
2825 |
17 |
return returnRewrite(node, ret, Rewrite::RPL_PULL_ENDPT); |
2826 |
|
} |
2827 |
|
} |
2828 |
|
} |
2829 |
|
|
2830 |
1756 |
children1.clear(); |
2831 |
1756 |
utils::getConcat(node[1], children1); |
2832 |
3512 |
Node lastChild1 = children1[children1.size() - 1]; |
2833 |
1756 |
if (lastChild1.getKind() == kind::STRING_SUBSTR) |
2834 |
|
{ |
2835 |
|
// (str.replace x (str.++ t (str.substr y i j)) z) ---> |
2836 |
|
// (str.replace x (str.++ t |
2837 |
|
// (str.substr y i (+ (str.len x) 1 (- (str.len t))))) z) |
2838 |
|
// if j > len(x) |
2839 |
|
// |
2840 |
|
// Reasoning: If the string to be replaced is longer than x, then it does |
2841 |
|
// not matter how much longer it is, the result is always x. Thus, it is |
2842 |
|
// fine to only look at the prefix of length len(x) + 1 - len(t). |
2843 |
|
|
2844 |
112 |
children1.pop_back(); |
2845 |
|
// Length of the non-substr components in the second argument |
2846 |
|
Node partLen1 = |
2847 |
216 |
nm->mkNode(kind::STRING_LENGTH, utils::mkConcat(children1, stype)); |
2848 |
216 |
Node maxLen1 = nm->mkNode(kind::PLUS, partLen1, lastChild1[2]); |
2849 |
|
|
2850 |
216 |
Node zero = nm->mkConst(Rational(0)); |
2851 |
216 |
Node one = nm->mkConst(Rational(1)); |
2852 |
216 |
Node len0 = nm->mkNode(kind::STRING_LENGTH, node[0]); |
2853 |
216 |
Node len0_1 = nm->mkNode(kind::PLUS, len0, one); |
2854 |
|
// Check len(t) + j > len(x) + 1 |
2855 |
112 |
if (ArithEntail::check(maxLen1, len0_1, true)) |
2856 |
|
{ |
2857 |
8 |
children1.push_back(nm->mkNode( |
2858 |
|
kind::STRING_SUBSTR, |
2859 |
|
lastChild1[0], |
2860 |
|
lastChild1[1], |
2861 |
16 |
nm->mkNode( |
2862 |
16 |
kind::PLUS, len0, one, nm->mkNode(kind::UMINUS, partLen1)))); |
2863 |
|
Node res = nm->mkNode(kind::STRING_REPLACE, |
2864 |
|
node[0], |
2865 |
16 |
utils::mkConcat(children1, stype), |
2866 |
32 |
node[2]); |
2867 |
8 |
return returnRewrite(node, res, Rewrite::REPL_SUBST_IDX); |
2868 |
|
} |
2869 |
|
} |
2870 |
|
|
2871 |
1748 |
if (node[0].getKind() == STRING_REPLACE) |
2872 |
|
{ |
2873 |
388 |
Node x = node[0]; |
2874 |
388 |
Node y = node[1]; |
2875 |
388 |
Node z = node[2]; |
2876 |
200 |
if (x[0] == x[2] && x[0] == y) |
2877 |
|
{ |
2878 |
|
// (str.replace (str.replace y w y) y z) --> |
2879 |
|
// (str.replace (str.replace y w z) y z) |
2880 |
|
// if (str.len w) >= (str.len z) and w != z |
2881 |
|
// |
2882 |
|
// Reasoning: There are two cases: (1) w does not appear in y and (2) w |
2883 |
|
// does appear in y. |
2884 |
|
// |
2885 |
|
// Case (1): In this case, the reasoning is trivial. The |
2886 |
|
// inner replace does not do anything, so we can just replace its third |
2887 |
|
// argument with any string. |
2888 |
|
// |
2889 |
|
// Case (2): After the inner replace, we are guaranteed to have a string |
2890 |
|
// that contains y at the index of w in the original string y. The outer |
2891 |
|
// replace then replaces that y with z, so we can short-circuit that |
2892 |
|
// replace by directly replacing w with z in the inner replace. We can |
2893 |
|
// only do that if the result of the new inner replace does not contain |
2894 |
|
// y, otherwise we end up doing two replaces that are different from the |
2895 |
|
// original expression. We enforce that by requiring that the length of w |
2896 |
|
// has to be greater or equal to the length of z and that w and z have to |
2897 |
|
// be different. This makes sure that an inner replace changes a string |
2898 |
|
// to a string that is shorter than y, making it impossible for the outer |
2899 |
|
// replace to match. |
2900 |
136 |
Node w = x[1]; |
2901 |
|
|
2902 |
|
// (str.len w) >= (str.len z) |
2903 |
136 |
Node wlen = nm->mkNode(kind::STRING_LENGTH, w); |
2904 |
136 |
Node zlen = nm->mkNode(kind::STRING_LENGTH, z); |
2905 |
74 |
if (ArithEntail::check(wlen, zlen)) |
2906 |
|
{ |
2907 |
|
// w != z |
2908 |
72 |
Node wEqZ = Rewriter::rewrite(nm->mkNode(kind::EQUAL, w, z)); |
2909 |
42 |
if (wEqZ.isConst() && !wEqZ.getConst<bool>()) |
2910 |
|
{ |
2911 |
|
Node ret = nm->mkNode(kind::STRING_REPLACE, |
2912 |
24 |
nm->mkNode(kind::STRING_REPLACE, y, w, z), |
2913 |
|
y, |
2914 |
48 |
z); |
2915 |
12 |
return returnRewrite(node, ret, Rewrite::REPL_REPL_SHORT_CIRCUIT); |
2916 |
|
} |
2917 |
|
} |
2918 |
|
} |
2919 |
|
} |
2920 |
|
|
2921 |
1736 |
if (node[1].getKind() == STRING_REPLACE) |
2922 |
|
{ |
2923 |
148 |
if (node[1][0] == node[0]) |
2924 |
|
{ |
2925 |
46 |
if (node[1][0] == node[1][2] && node[1][0] == node[2]) |
2926 |
|
{ |
2927 |
|
// str.replace( x, str.replace( x, y, x ), x ) ---> x |
2928 |
|
return returnRewrite(node, node[0], Rewrite::REPL_REPL2_INV_ID); |
2929 |
|
} |
2930 |
46 |
bool dualReplIteSuccess = false; |
2931 |
92 |
Node cmp_con2 = d_stringsEntail.checkContains(node[1][0], node[1][2]); |
2932 |
46 |
if (!cmp_con2.isNull() && !cmp_con2.getConst<bool>()) |
2933 |
|
{ |
2934 |
|
// str.contains( x, z ) ---> false |
2935 |
|
// implies |
2936 |
|
// str.replace( x, str.replace( x, y, z ), w ) ---> |
2937 |
|
// ite( str.contains( x, y ), x, w ) |
2938 |
|
dualReplIteSuccess = true; |
2939 |
|
} |
2940 |
|
else |
2941 |
|
{ |
2942 |
|
// str.contains( y, z ) ---> false and str.contains( z, y ) ---> false |
2943 |
|
// implies |
2944 |
|
// str.replace( x, str.replace( x, y, z ), w ) ---> |
2945 |
|
// ite( str.contains( x, y ), x, w ) |
2946 |
46 |
cmp_con2 = d_stringsEntail.checkContains(node[1][1], node[1][2]); |
2947 |
46 |
if (!cmp_con2.isNull() && !cmp_con2.getConst<bool>()) |
2948 |
|
{ |
2949 |
8 |
cmp_con2 = d_stringsEntail.checkContains(node[1][2], node[1][1]); |
2950 |
8 |
if (!cmp_con2.isNull() && !cmp_con2.getConst<bool>()) |
2951 |
|
{ |
2952 |
|
dualReplIteSuccess = true; |
2953 |
|
} |
2954 |
|
} |
2955 |
|
} |
2956 |
46 |
if (dualReplIteSuccess) |
2957 |
|
{ |
2958 |
|
Node res = nm->mkNode(ITE, |
2959 |
|
nm->mkNode(STRING_CONTAINS, node[0], node[1][1]), |
2960 |
|
node[0], |
2961 |
|
node[2]); |
2962 |
|
return returnRewrite(node, res, Rewrite::REPL_DUAL_REPL_ITE); |
2963 |
|
} |
2964 |
|
} |
2965 |
|
|
2966 |
148 |
bool invSuccess = false; |
2967 |
148 |
if (node[1][1] == node[0]) |
2968 |
|
{ |
2969 |
74 |
if (node[1][0] == node[1][2]) |
2970 |
|
{ |
2971 |
|
// str.replace(x, str.replace(y, x, y), w) ---> str.replace(x, y, w) |
2972 |
|
invSuccess = true; |
2973 |
|
} |
2974 |
74 |
else if (node[1][1] == node[2] || node[1][0] == node[2]) |
2975 |
|
{ |
2976 |
|
// str.contains(y, z) ----> false and ( y == w or x == w ) implies |
2977 |
|
// implies |
2978 |
|
// str.replace(x, str.replace(y, x, z), w) ---> str.replace(x, y, w) |
2979 |
44 |
Node cmp_con2 = d_stringsEntail.checkContains(node[1][0], node[1][2]); |
2980 |
22 |
invSuccess = !cmp_con2.isNull() && !cmp_con2.getConst<bool>(); |
2981 |
|
} |
2982 |
|
} |
2983 |
|
else |
2984 |
|
{ |
2985 |
|
// str.contains(x, z) ----> false and str.contains(x, w) ----> false |
2986 |
|
// implies |
2987 |
|
// str.replace(x, str.replace(y, z, w), u) ---> str.replace(x, y, u) |
2988 |
148 |
Node cmp_con2 = d_stringsEntail.checkContains(node[0], node[1][1]); |
2989 |
74 |
if (!cmp_con2.isNull() && !cmp_con2.getConst<bool>()) |
2990 |
|
{ |
2991 |
16 |
cmp_con2 = d_stringsEntail.checkContains(node[0], node[1][2]); |
2992 |
16 |
invSuccess = !cmp_con2.isNull() && !cmp_con2.getConst<bool>(); |
2993 |
|
} |
2994 |
|
} |
2995 |
148 |
if (invSuccess) |
2996 |
|
{ |
2997 |
|
Node res = nm->mkNode(kind::STRING_REPLACE, node[0], node[1][0], node[2]); |
2998 |
|
return returnRewrite(node, res, Rewrite::REPL_REPL2_INV); |
2999 |
|
} |
3000 |
|
} |
3001 |
1736 |
if (node[2].getKind() == STRING_REPLACE) |
3002 |
|
{ |
3003 |
56 |
if (node[2][1] == node[0]) |
3004 |
|
{ |
3005 |
|
// str.contains( z, w ) ----> false implies |
3006 |
|
// str.replace( x, w, str.replace( z, x, y ) ) ---> str.replace( x, w, z ) |
3007 |
24 |
Node cmp_con2 = d_stringsEntail.checkContains(node[2][0], node[1]); |
3008 |
12 |
if (!cmp_con2.isNull() && !cmp_con2.getConst<bool>()) |
3009 |
|
{ |
3010 |
|
Node res = |
3011 |
|
nm->mkNode(kind::STRING_REPLACE, node[0], node[1], node[2][0]); |
3012 |
|
return returnRewrite(node, res, Rewrite::REPL_REPL3_INV); |
3013 |
|
} |
3014 |
|
} |
3015 |
56 |
if (node[2][0] == node[1]) |
3016 |
|
{ |
3017 |
4 |
bool success = false; |
3018 |
4 |
if (node[2][0] == node[2][2] && node[2][1] == node[0]) |
3019 |
|
{ |
3020 |
|
// str.replace( x, y, str.replace( y, x, y ) ) ---> x |
3021 |
|
success = true; |
3022 |
|
} |
3023 |
|
else |
3024 |
|
{ |
3025 |
|
// str.contains( x, z ) ----> false implies |
3026 |
|
// str.replace( x, y, str.replace( y, z, w ) ) ---> x |
3027 |
4 |
cmp_con = d_stringsEntail.checkContains(node[0], node[2][1]); |
3028 |
4 |
success = !cmp_con.isNull() && !cmp_con.getConst<bool>(); |
3029 |
|
} |
3030 |
4 |
if (success) |
3031 |
|
{ |
3032 |
|
return returnRewrite(node, node[0], Rewrite::REPL_REPL3_INV_ID); |
3033 |
|
} |
3034 |
|
} |
3035 |
|
} |
3036 |
|
// miniscope based on components that do not contribute to contains |
3037 |
|
// for example, |
3038 |
|
// str.replace( x ++ y ++ x ++ y, "A", z ) --> |
3039 |
|
// str.replace( x ++ y, "A", z ) ++ x ++ y |
3040 |
|
// since if "A" occurs in x ++ y ++ x ++ y, then it must occur in x ++ y. |
3041 |
1736 |
if (StringsEntail::checkLengthOne(node[1])) |
3042 |
|
{ |
3043 |
1352 |
Node lastLhs; |
3044 |
680 |
unsigned lastCheckIndex = 0; |
3045 |
692 |
for (unsigned i = 1, iend = children0.size(); i < iend; i++) |
3046 |
|
{ |
3047 |
124 |
unsigned checkIndex = children0.size() - i; |
3048 |
136 |
std::vector<Node> checkLhs; |
3049 |
124 |
checkLhs.insert( |
3050 |
248 |
checkLhs.end(), children0.begin(), children0.begin() + checkIndex); |
3051 |
136 |
Node lhs = utils::mkConcat(checkLhs, stype); |
3052 |
136 |
Node rhs = children0[checkIndex]; |
3053 |
136 |
Node ctn = d_stringsEntail.checkContains(lhs, rhs); |
3054 |
124 |
if (!ctn.isNull() && ctn.getConst<bool>()) |
3055 |
|
{ |
3056 |
12 |
lastLhs = lhs; |
3057 |
12 |
lastCheckIndex = checkIndex; |
3058 |
|
} |
3059 |
|
else |
3060 |
|
{ |
3061 |
112 |
break; |
3062 |
|
} |
3063 |
|
} |
3064 |
680 |
if (!lastLhs.isNull()) |
3065 |
|
{ |
3066 |
16 |
std::vector<Node> remc(children0.begin() + lastCheckIndex, |
3067 |
32 |
children0.end()); |
3068 |
16 |
Node rem = utils::mkConcat(remc, stype); |
3069 |
|
Node ret = |
3070 |
|
nm->mkNode(STRING_CONCAT, |
3071 |
16 |
nm->mkNode(STRING_REPLACE, lastLhs, node[1], node[2]), |
3072 |
32 |
rem); |
3073 |
|
// for example: |
3074 |
|
// str.replace( x ++ x, "A", y ) ---> str.replace( x, "A", y ) ++ x |
3075 |
|
// Since we know that the first occurrence of "A" cannot be in the |
3076 |
|
// second occurrence of x. Notice this is specific to single characters |
3077 |
|
// due to complications with finds that span multiple components for |
3078 |
|
// non-characters. |
3079 |
8 |
return returnRewrite(node, ret, Rewrite::REPL_CHAR_NCONTRIB_FIND); |
3080 |
|
} |
3081 |
|
} |
3082 |
|
|
3083 |
|
// TODO (#1180) incorporate these? |
3084 |
|
// contains( t, s ) => |
3085 |
|
// replace( replace( x, t, s ), s, r ) ----> replace( x, t, r ) |
3086 |
|
// contains( t, s ) => |
3087 |
|
// contains( replace( t, s, r ), r ) ----> true |
3088 |
|
|
3089 |
1728 |
return node; |
3090 |
|
} |
3091 |
|
|
3092 |
198 |
Node SequencesRewriter::rewriteReplaceAll(Node node) |
3093 |
|
{ |
3094 |
198 |
Assert(node.getKind() == STRING_REPLACE_ALL); |
3095 |
|
|
3096 |
396 |
TypeNode stype = node.getType(); |
3097 |
|
|
3098 |
198 |
if (node[0].isConst() && node[1].isConst()) |
3099 |
|
{ |
3100 |
80 |
std::vector<Node> children; |
3101 |
80 |
Node s = node[0]; |
3102 |
80 |
Node t = node[1]; |
3103 |
40 |
if (Word::isEmpty(s) || Word::isEmpty(t)) |
3104 |
|
{ |
3105 |
10 |
return returnRewrite(node, node[0], Rewrite::REPLALL_EMPTY_FIND); |
3106 |
|
} |
3107 |
30 |
std::size_t sizeS = Word::getLength(s); |
3108 |
30 |
std::size_t sizeT = Word::getLength(t); |
3109 |
30 |
std::size_t index = 0; |
3110 |
30 |
std::size_t curr = 0; |
3111 |
65 |
do |
3112 |
|
{ |
3113 |
95 |
curr = Word::find(s, t, index); |
3114 |
95 |
if (curr != std::string::npos) |
3115 |
|
{ |
3116 |
65 |
if (curr > index) |
3117 |
|
{ |
3118 |
32 |
children.push_back(Word::substr(s, index, curr - index)); |
3119 |
|
} |
3120 |
65 |
children.push_back(node[2]); |
3121 |
65 |
index = curr + sizeT; |
3122 |
|
} |
3123 |
|
else |
3124 |
|
{ |
3125 |
30 |
children.push_back(Word::substr(s, index, sizeS - index)); |
3126 |
|
} |
3127 |
95 |
} while (curr != std::string::npos && curr < sizeS); |
3128 |
|
// constant evaluation |
3129 |
60 |
Node res = utils::mkConcat(children, stype); |
3130 |
30 |
return returnRewrite(node, res, Rewrite::REPLALL_CONST); |
3131 |
|
} |
3132 |
|
|
3133 |
|
// rewrites that apply to both replace and replaceall |
3134 |
316 |
Node rri = rewriteReplaceInternal(node); |
3135 |
158 |
if (!rri.isNull()) |
3136 |
|
{ |
3137 |
|
// printing of the rewrite managed by the call above |
3138 |
4 |
return rri; |
3139 |
|
} |
3140 |
|
|
3141 |
154 |
return node; |
3142 |
|
} |
3143 |
|
|
3144 |
2224 |
Node SequencesRewriter::rewriteReplaceInternal(Node node) |
3145 |
|
{ |
3146 |
2224 |
Kind nk = node.getKind(); |
3147 |
2224 |
Assert(nk == STRING_REPLACE || nk == STRING_REPLACE_ALL); |
3148 |
|
|
3149 |
2224 |
if (node[1] == node[2]) |
3150 |
|
{ |
3151 |
60 |
return returnRewrite(node, node[0], Rewrite::RPL_ID); |
3152 |
|
} |
3153 |
|
|
3154 |
2164 |
if (node[0] == node[1]) |
3155 |
|
{ |
3156 |
|
// only holds for replaceall if non-empty |
3157 |
107 |
if (nk == STRING_REPLACE || StringsEntail::checkNonEmpty(node[1])) |
3158 |
|
{ |
3159 |
79 |
return returnRewrite(node, node[2], Rewrite::RPL_REPLACE); |
3160 |
|
} |
3161 |
|
} |
3162 |
|
|
3163 |
2085 |
return Node::null(); |
3164 |
|
} |
3165 |
|
|
3166 |
403 |
Node SequencesRewriter::rewriteReplaceRe(Node node) |
3167 |
|
{ |
3168 |
403 |
Assert(node.getKind() == STRING_REPLACE_RE); |
3169 |
403 |
NodeManager* nm = NodeManager::currentNM(); |
3170 |
806 |
Node x = node[0]; |
3171 |
806 |
Node y = node[1]; |
3172 |
806 |
Node z = node[2]; |
3173 |
|
|
3174 |
403 |
if (RegExpEntail::isConstRegExp(y)) |
3175 |
|
{ |
3176 |
271 |
if (x.isConst()) |
3177 |
|
{ |
3178 |
|
// str.replace_re("ZABCZ", re.++("A", _*, "C"), y) ---> "Z" ++ y ++ "Z" |
3179 |
75 |
std::pair<size_t, size_t> match = firstMatch(x, y); |
3180 |
75 |
if (match.first != string::npos) |
3181 |
|
{ |
3182 |
80 |
String s = x.getConst<String>(); |
3183 |
|
Node ret = nm->mkNode(STRING_CONCAT, |
3184 |
80 |
nm->mkConst(s.substr(0, match.first)), |
3185 |
|
z, |
3186 |
160 |
nm->mkConst(s.substr(match.second))); |
3187 |
40 |
return returnRewrite(node, ret, Rewrite::REPLACE_RE_EVAL); |
3188 |
|
} |
3189 |
|
else |
3190 |
|
{ |
3191 |
35 |
return returnRewrite(node, x, Rewrite::REPLACE_RE_EVAL); |
3192 |
|
} |
3193 |
|
} |
3194 |
|
// str.replace_re( x, y, z ) ---> z ++ x if "" in y ---> true |
3195 |
388 |
String emptyStr(""); |
3196 |
196 |
if (RegExpEntail::testConstStringInRegExp(emptyStr, 0, y)) |
3197 |
|
{ |
3198 |
8 |
Node ret = nm->mkNode(STRING_CONCAT, z, x); |
3199 |
4 |
return returnRewrite(node, ret, Rewrite::REPLACE_RE_EMP_RE); |
3200 |
|
} |
3201 |
|
} |
3202 |
324 |
return node; |
3203 |
|
} |
3204 |
|
|
3205 |
261 |
Node SequencesRewriter::rewriteReplaceReAll(Node node) |
3206 |
|
{ |
3207 |
261 |
Assert(node.getKind() == STRING_REPLACE_RE_ALL); |
3208 |
261 |
NodeManager* nm = NodeManager::currentNM(); |
3209 |
522 |
Node x = node[0]; |
3210 |
522 |
Node y = node[1]; |
3211 |
522 |
Node z = node[2]; |
3212 |
|
|
3213 |
261 |
if (RegExpEntail::isConstRegExp(y)) |
3214 |
|
{ |
3215 |
207 |
if (x.isConst()) |
3216 |
|
{ |
3217 |
|
// str.replace_re_all("ZABCZAB", re.++("A", _*, "C"), y) ---> |
3218 |
|
// "Z" ++ y ++ "Z" ++ y |
3219 |
138 |
TypeNode t = x.getType(); |
3220 |
138 |
Node emp = Word::mkEmptyWord(t); |
3221 |
|
Node yp = Rewriter::rewrite( |
3222 |
138 |
nm->mkNode(REGEXP_DIFF, y, nm->mkNode(STRING_TO_REGEXP, emp))); |
3223 |
138 |
std::vector<Node> res; |
3224 |
138 |
String rem = x.getConst<String>(); |
3225 |
69 |
std::pair<size_t, size_t> match(0, 0); |
3226 |
183 |
while (rem.size() != 0) |
3227 |
|
{ |
3228 |
67 |
match = firstMatch(nm->mkConst(rem), yp); |
3229 |
67 |
if (match.first == string::npos) |
3230 |
|
{ |
3231 |
10 |
break; |
3232 |
|
} |
3233 |
57 |
res.push_back(nm->mkConst(rem.substr(0, match.first))); |
3234 |
57 |
res.push_back(z); |
3235 |
57 |
rem = rem.substr(match.second); |
3236 |
|
} |
3237 |
69 |
res.push_back(nm->mkConst(rem)); |
3238 |
138 |
Node ret = utils::mkConcat(res, t); |
3239 |
69 |
return returnRewrite(node, ret, Rewrite::REPLACE_RE_ALL_EVAL); |
3240 |
|
} |
3241 |
|
} |
3242 |
|
|
3243 |
192 |
return node; |
3244 |
|
} |
3245 |
|
|
3246 |
297 |
std::pair<size_t, size_t> SequencesRewriter::firstMatch(Node n, Node r) |
3247 |
|
{ |
3248 |
297 |
Assert(n.isConst() && n.getType().isStringLike()); |
3249 |
297 |
Assert(r.getType().isRegExp()); |
3250 |
297 |
NodeManager* nm = NodeManager::currentNM(); |
3251 |
|
|
3252 |
594 |
Node sigmaStar = nm->mkNode(REGEXP_STAR, nm->mkNode(REGEXP_SIGMA)); |
3253 |
594 |
Node re = nm->mkNode(REGEXP_CONCAT, r, sigmaStar); |
3254 |
594 |
String s = n.getConst<String>(); |
3255 |
|
|
3256 |
297 |
if (s.size() == 0) |
3257 |
|
{ |
3258 |
63 |
if (RegExpEntail::testConstStringInRegExp(s, 0, r)) |
3259 |
|
{ |
3260 |
6 |
return std::make_pair(0, 0); |
3261 |
|
} |
3262 |
|
else |
3263 |
|
{ |
3264 |
57 |
return std::make_pair(string::npos, string::npos); |
3265 |
|
} |
3266 |
|
} |
3267 |
|
|
3268 |
341 |
for (size_t i = 0, size = s.size(); i < size; i++) |
3269 |
|
{ |
3270 |
293 |
if (RegExpEntail::testConstStringInRegExp(s, i, re)) |
3271 |
|
{ |
3272 |
635 |
for (size_t j = i; j <= size; j++) |
3273 |
|
{ |
3274 |
1084 |
String substr = s.substr(i, j - i); |
3275 |
635 |
if (RegExpEntail::testConstStringInRegExp(substr, 0, r)) |
3276 |
|
{ |
3277 |
186 |
return std::make_pair(i, j); |
3278 |
|
} |
3279 |
|
} |
3280 |
|
} |
3281 |
|
} |
3282 |
|
|
3283 |
48 |
return std::make_pair(string::npos, string::npos); |
3284 |
|
} |
3285 |
|
|
3286 |
243 |
Node SequencesRewriter::rewriteStrReverse(Node node) |
3287 |
|
{ |
3288 |
243 |
Assert(node.getKind() == STRING_REV); |
3289 |
243 |
NodeManager* nm = NodeManager::currentNM(); |
3290 |
486 |
Node x = node[0]; |
3291 |
243 |
if (x.isConst()) |
3292 |
|
{ |
3293 |
|
// reverse the characters in the constant |
3294 |
130 |
Node retNode = Word::reverse(x); |
3295 |
65 |
return returnRewrite(node, retNode, Rewrite::STR_CONV_CONST); |
3296 |
|
} |
3297 |
178 |
else if (x.getKind() == STRING_CONCAT) |
3298 |
|
{ |
3299 |
78 |
std::vector<Node> children; |
3300 |
119 |
for (const Node& nc : x) |
3301 |
|
{ |
3302 |
80 |
children.push_back(nm->mkNode(STRING_REV, nc)); |
3303 |
|
} |
3304 |
39 |
std::reverse(children.begin(), children.end()); |
3305 |
|
// rev( x1 ++ x2 ) --> rev( x2 ) ++ rev( x1 ) |
3306 |
78 |
Node retNode = nm->mkNode(STRING_CONCAT, children); |
3307 |
39 |
return returnRewrite(node, retNode, Rewrite::STR_REV_MINSCOPE_CONCAT); |
3308 |
|
} |
3309 |
139 |
else if (x.getKind() == STRING_REV) |
3310 |
|
{ |
3311 |
|
// rev( rev( x ) ) --> x |
3312 |
10 |
Node retNode = x[0]; |
3313 |
5 |
return returnRewrite(node, retNode, Rewrite::STR_REV_IDEM); |
3314 |
|
} |
3315 |
134 |
return node; |
3316 |
|
} |
3317 |
|
|
3318 |
153 |
Node SequencesRewriter::rewritePrefixSuffix(Node n) |
3319 |
|
{ |
3320 |
153 |
Assert(n.getKind() == kind::STRING_PREFIX |
3321 |
|
|| n.getKind() == kind::STRING_SUFFIX); |
3322 |
153 |
bool isPrefix = n.getKind() == kind::STRING_PREFIX; |
3323 |
153 |
if (n[0] == n[1]) |
3324 |
|
{ |
3325 |
16 |
Node ret = NodeManager::currentNM()->mkConst(true); |
3326 |
8 |
return returnRewrite(n, ret, Rewrite::SUF_PREFIX_EQ); |
3327 |
|
} |
3328 |
145 |
if (n[0].isConst()) |
3329 |
|
{ |
3330 |
58 |
if (Word::isEmpty(n[0])) |
3331 |
|
{ |
3332 |
8 |
Node ret = NodeManager::currentNM()->mkConst(true); |
3333 |
4 |
return returnRewrite(n, ret, Rewrite::SUF_PREFIX_EMPTY_CONST); |
3334 |
|
} |
3335 |
|
} |
3336 |
141 |
if (n[1].isConst()) |
3337 |
|
{ |
3338 |
47 |
Node s = n[1]; |
3339 |
47 |
size_t lenS = Word::getLength(s); |
3340 |
47 |
if (n[0].isConst()) |
3341 |
|
{ |
3342 |
|
Node ret = NodeManager::currentNM()->mkConst(false); |
3343 |
|
Node t = n[0]; |
3344 |
|
size_t lenT = Word::getLength(t); |
3345 |
|
if (lenS >= lenT) |
3346 |
|
{ |
3347 |
|
if ((isPrefix && t == Word::prefix(s, lenT)) |
3348 |
|
|| (!isPrefix && t == Word::suffix(s, lenT))) |
3349 |
|
{ |
3350 |
|
ret = NodeManager::currentNM()->mkConst(true); |
3351 |
|
} |
3352 |
|
} |
3353 |
|
return returnRewrite(n, ret, Rewrite::SUF_PREFIX_CONST); |
3354 |
|
} |
3355 |
47 |
else if (lenS == 0) |
3356 |
|
{ |
3357 |
8 |
Node ret = n[0].eqNode(n[1]); |
3358 |
4 |
return returnRewrite(n, ret, Rewrite::SUF_PREFIX_EMPTY); |
3359 |
|
} |
3360 |
43 |
else if (lenS == 1) |
3361 |
|
{ |
3362 |
|
// (str.prefix x "A") and (str.suffix x "A") are equivalent to |
3363 |
|
// (str.contains "A" x ) |
3364 |
|
Node ret = |
3365 |
86 |
NodeManager::currentNM()->mkNode(kind::STRING_CONTAINS, n[1], n[0]); |
3366 |
43 |
return returnRewrite(n, ret, Rewrite::SUF_PREFIX_CTN); |
3367 |
|
} |
3368 |
|
} |
3369 |
188 |
Node lens = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, n[0]); |
3370 |
188 |
Node lent = NodeManager::currentNM()->mkNode(kind::STRING_LENGTH, n[1]); |
3371 |
188 |
Node val; |
3372 |
94 |
if (isPrefix) |
3373 |
|
{ |
3374 |
67 |
val = NodeManager::currentNM()->mkConst(::cvc5::Rational(0)); |
3375 |
|
} |
3376 |
|
else |
3377 |
|
{ |
3378 |
27 |
val = NodeManager::currentNM()->mkNode(kind::MINUS, lent, lens); |
3379 |
|
} |
3380 |
|
|
3381 |
|
// Check if we can turn the prefix/suffix into equalities by showing that the |
3382 |
|
// prefix/suffix is at least as long as the string |
3383 |
188 |
Node eqs = StringsEntail::inferEqsFromContains(n[1], n[0]); |
3384 |
94 |
if (!eqs.isNull()) |
3385 |
|
{ |
3386 |
8 |
return returnRewrite(n, eqs, Rewrite::SUF_PREFIX_TO_EQS); |
3387 |
|
} |
3388 |
|
|
3389 |
|
// general reduction to equality + substr |
3390 |
|
Node retNode = n[0].eqNode( |
3391 |
172 |
NodeManager::currentNM()->mkNode(kind::STRING_SUBSTR, n[1], val, lens)); |
3392 |
|
|
3393 |
86 |
return returnRewrite(n, retNode, Rewrite::SUF_PREFIX_ELIM); |
3394 |
|
} |
3395 |
|
|
3396 |
81 |
Node SequencesRewriter::lengthPreserveRewrite(Node n) |
3397 |
|
{ |
3398 |
81 |
NodeManager* nm = NodeManager::currentNM(); |
3399 |
162 |
Node len = Rewriter::rewrite(nm->mkNode(kind::STRING_LENGTH, n)); |
3400 |
162 |
Node res = canonicalStrForSymbolicLength(len, n.getType()); |
3401 |
162 |
return res.isNull() ? n : res; |
3402 |
|
} |
3403 |
|
|
3404 |
93 |
Node SequencesRewriter::canonicalStrForSymbolicLength(Node len, TypeNode stype) |
3405 |
|
{ |
3406 |
93 |
NodeManager* nm = NodeManager::currentNM(); |
3407 |
|
|
3408 |
186 |
Node res; |
3409 |
93 |
if (len.getKind() == CONST_RATIONAL) |
3410 |
|
{ |
3411 |
|
// c -> "A" repeated c times |
3412 |
162 |
Rational ratLen = len.getConst<Rational>(); |
3413 |
81 |
Assert(ratLen.getDenominator() == 1); |
3414 |
162 |
Integer intLen = ratLen.getNumerator(); |
3415 |
81 |
uint32_t u = intLen.getUnsignedInt(); |
3416 |
81 |
if (stype.isString()) // string-only |
3417 |
|
{ |
3418 |
81 |
res = nm->mkConst(String(std::string(u, 'A'))); |
3419 |
|
} |
3420 |
|
// we could do this for sequences, but we need to be careful: some |
3421 |
|
// sorts do not permit values that the solver can handle (e.g. uninterpreted |
3422 |
|
// sorts and arrays). |
3423 |
|
} |
3424 |
12 |
else if (len.getKind() == PLUS) |
3425 |
|
{ |
3426 |
|
// x + y -> norm(x) + norm(y) |
3427 |
8 |
NodeBuilder concatBuilder(STRING_CONCAT); |
3428 |
12 |
for (const auto& n : len) |
3429 |
|
{ |
3430 |
16 |
Node sn = canonicalStrForSymbolicLength(n, stype); |
3431 |
8 |
if (sn.isNull()) |
3432 |
|
{ |
3433 |
|
return Node::null(); |
3434 |
|
} |
3435 |
16 |
std::vector<Node> snChildren; |
3436 |
8 |
utils::getConcat(sn, snChildren); |
3437 |
8 |
concatBuilder.append(snChildren); |
3438 |
|
} |
3439 |
4 |
res = concatBuilder.constructNode(); |
3440 |
|
} |
3441 |
20 |
else if (len.getKind() == MULT && len.getNumChildren() == 2 |
3442 |
20 |
&& len[0].isConst()) |
3443 |
|
{ |
3444 |
|
// c * x -> norm(x) repeated c times |
3445 |
8 |
Rational ratReps = len[0].getConst<Rational>(); |
3446 |
4 |
Assert(ratReps.getDenominator() == 1); |
3447 |
8 |
Integer intReps = ratReps.getNumerator(); |
3448 |
|
|
3449 |
8 |
Node nRep = canonicalStrForSymbolicLength(len[1], stype); |
3450 |
4 |
if (nRep.isNull()) |
3451 |
|
{ |
3452 |
|
return Node::null(); |
3453 |
|
} |
3454 |
8 |
std::vector<Node> nRepChildren; |
3455 |
4 |
utils::getConcat(nRep, nRepChildren); |
3456 |
8 |
NodeBuilder concatBuilder(STRING_CONCAT); |
3457 |
12 |
for (size_t i = 0, reps = intReps.getUnsignedInt(); i < reps; i++) |
3458 |
|
{ |
3459 |
8 |
concatBuilder.append(nRepChildren); |
3460 |
|
} |
3461 |
4 |
res = concatBuilder.constructNode(); |
3462 |
|
} |
3463 |
4 |
else if (len.getKind() == STRING_LENGTH) |
3464 |
|
{ |
3465 |
|
// len(x) -> x |
3466 |
4 |
res = len[0]; |
3467 |
|
} |
3468 |
93 |
return res; |
3469 |
|
} |
3470 |
|
|
3471 |
403 |
Node SequencesRewriter::rewriteSeqUnit(Node node) |
3472 |
|
{ |
3473 |
403 |
NodeManager* nm = NodeManager::currentNM(); |
3474 |
403 |
if (node[0].isConst()) |
3475 |
|
{ |
3476 |
274 |
std::vector<Node> seq; |
3477 |
137 |
seq.push_back(node[0]); |
3478 |
274 |
TypeNode stype = node[0].getType(); |
3479 |
274 |
Node ret = nm->mkConst(Sequence(stype, seq)); |
3480 |
137 |
return returnRewrite(node, ret, Rewrite::SEQ_UNIT_EVAL); |
3481 |
|
} |
3482 |
266 |
return node; |
3483 |
|
} |
3484 |
|
|
3485 |
194481 |
Node SequencesRewriter::returnRewrite(Node node, |
3486 |
|
Node ret, |
3487 |
|
Rewrite r, |
3488 |
|
bool rewriteEqAgain) |
3489 |
|
{ |
3490 |
388962 |
Trace("strings-rewrite") << "Rewrite " << node << " to " << ret << " by " << r |
3491 |
194481 |
<< "." << std::endl; |
3492 |
|
|
3493 |
194481 |
NodeManager* nm = NodeManager::currentNM(); |
3494 |
|
|
3495 |
194481 |
if (d_statistics != nullptr) |
3496 |
|
{ |
3497 |
194415 |
(*d_statistics) << r; |
3498 |
|
} |
3499 |
|
|
3500 |
|
// standard post-processing |
3501 |
|
// We rewrite (string) equalities immediately here. This allows us to forego |
3502 |
|
// the standard invariant on equality rewrites (that s=t must rewrite to one |
3503 |
|
// of { s=t, t=s, true, false } ). |
3504 |
194481 |
Kind retk = ret.getKind(); |
3505 |
194481 |
if (retk == OR || retk == AND) |
3506 |
|
{ |
3507 |
15690 |
std::vector<Node> children; |
3508 |
7845 |
bool childChanged = false; |
3509 |
26462 |
for (const Node& cret : ret) |
3510 |
|
{ |
3511 |
37234 |
Node creter = cret; |
3512 |
18617 |
if (cret.getKind() == EQUAL) |
3513 |
|
{ |
3514 |
12306 |
creter = rewriteEqualityExt(cret); |
3515 |
|
} |
3516 |
6311 |
else if (cret.getKind() == NOT && cret[0].getKind() == EQUAL) |
3517 |
|
{ |
3518 |
360 |
creter = nm->mkNode(NOT, rewriteEqualityExt(cret[0])); |
3519 |
|
} |
3520 |
18617 |
childChanged = childChanged || cret != creter; |
3521 |
18617 |
children.push_back(creter); |
3522 |
|
} |
3523 |
7845 |
if (childChanged) |
3524 |
|
{ |
3525 |
2464 |
ret = nm->mkNode(retk, children); |
3526 |
7845 |
} |
3527 |
|
} |
3528 |
186636 |
else if (retk == NOT && ret[0].getKind() == EQUAL) |
3529 |
|
{ |
3530 |
|
ret = nm->mkNode(NOT, rewriteEqualityExt(ret[0])); |
3531 |
|
} |
3532 |
186636 |
else if (retk == EQUAL && (rewriteEqAgain || node.getKind() != EQUAL)) |
3533 |
|
{ |
3534 |
91866 |
Trace("strings-rewrite") |
3535 |
45933 |
<< "Apply extended equality rewrite on " << ret << std::endl; |
3536 |
45933 |
ret = rewriteEqualityExt(ret); |
3537 |
|
} |
3538 |
194481 |
return ret; |
3539 |
|
} |
3540 |
|
|
3541 |
|
} // namespace strings |
3542 |
|
} // namespace theory |
3543 |
29286 |
} // namespace cvc5 |