/usr/lib/gcc/x86_64-linux-gnu/5/include/d/std/regex/internal/kickstart.d is in libphobos-5-dev 5.5.0-12ubuntu1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 | /*
Kickstart is a coarse-grained "filter" engine that finds likely matches
to be verified by full-blown matcher.
*/
module std.regex.internal.kickstart;
package(std.regex):
import std.regex.internal.ir;
import std.algorithm, std.range, std.utf;
//utility for shiftOr, returns a minimum number of bytes to test in a Char
uint effectiveSize(Char)()
{
static if(is(Char == char))
return 1;
else static if(is(Char == wchar))
return 2;
else static if(is(Char == dchar))
return 3;
else
static assert(0);
}
/*
Kickstart engine using ShiftOr algorithm,
a bit parallel technique for inexact string searching.
*/
struct ShiftOr(Char)
{
private:
uint[] table;
uint fChar;
uint n_length;
enum charSize = effectiveSize!Char();
//maximum number of chars in CodepointSet to process
enum uint charsetThreshold = 32_000;
static struct ShiftThread
{
uint[] tab;
uint mask;
uint idx;
uint pc, counter, hops;
this(uint newPc, uint newCounter, uint[] table)
{
pc = newPc;
counter = newCounter;
mask = 1;
idx = 0;
hops = 0;
tab = table;
}
void setMask(uint idx, uint mask)
{
tab[idx] |= mask;
}
void setInvMask(uint idx, uint mask)
{
tab[idx] &= ~mask;
}
void set(alias setBits = setInvMask)(dchar ch)
{
static if(charSize == 3)
{
uint val = ch, tmask = mask;
setBits(val&0xFF, tmask);
tmask <<= 1;
val >>= 8;
setBits(val&0xFF, tmask);
tmask <<= 1;
val >>= 8;
assert(val <= 0x10);
setBits(val, tmask);
tmask <<= 1;
}
else
{
Char[dchar.sizeof/Char.sizeof] buf;
uint tmask = mask;
size_t total = encode(buf, ch);
for(size_t i = 0; i < total; i++, tmask<<=1)
{
static if(charSize == 1)
setBits(buf[i], tmask);
else static if(charSize == 2)
{
setBits(buf[i]&0xFF, tmask);
tmask <<= 1;
setBits(buf[i]>>8, tmask);
}
}
}
}
void add(dchar ch){ return set!setInvMask(ch); }
void advance(uint s)
{
mask <<= s;
idx += s;
}
@property bool full(){ return !mask; }
}
static ShiftThread fork(ShiftThread t, uint newPc, uint newCounter)
{
ShiftThread nt = t;
nt.pc = newPc;
nt.counter = newCounter;
return nt;
}
@trusted static ShiftThread fetch(ref ShiftThread[] worklist)
{
auto t = worklist[$-1];
worklist.length -= 1;
if(!__ctfe)
cast(void)worklist.assumeSafeAppend();
return t;
}
static uint charLen(uint ch)
{
assert(ch <= 0x10FFFF);
return codeLength!Char(cast(dchar)ch)*charSize;
}
public:
@trusted this(ref Regex!Char re, uint[] memory)
{
import std.conv;
assert(memory.length == 256);
fChar = uint.max;
L_FindChar:
for(size_t i = 0;;)
{
switch(re.ir[i].code)
{
case IR.Char:
fChar = re.ir[i].data;
static if(charSize != 3)
{
Char[dchar.sizeof/Char.sizeof] buf;
encode(buf, fChar);
fChar = buf[0];
}
fChar = fChar & 0xFF;
break L_FindChar;
case IR.GroupStart, IR.GroupEnd:
i += IRL!(IR.GroupStart);
break;
case IR.Bol, IR.Wordboundary, IR.Notwordboundary:
i += IRL!(IR.Bol);
break;
default:
break L_FindChar;
}
}
table = memory;
table[] = uint.max;
ShiftThread[] trs;
ShiftThread t = ShiftThread(0, 0, table);
//locate first fixed char if any
n_length = 32;
for(;;)
{
L_Eval_Thread:
for(;;)
{
switch(re.ir[t.pc].code)
{
case IR.Char:
uint s = charLen(re.ir[t.pc].data);
if(t.idx+s > n_length)
goto L_StopThread;
t.add(re.ir[t.pc].data);
t.advance(s);
t.pc += IRL!(IR.Char);
break;
case IR.OrChar://assumes IRL!(OrChar) == 1
uint len = re.ir[t.pc].sequence;
uint end = t.pc + len;
uint[Bytecode.maxSequence] s;
uint numS;
for(uint i = 0; i < len; i++)
{
auto x = charLen(re.ir[t.pc+i].data);
if(countUntil(s[0..numS], x) < 0)
s[numS++] = x;
}
for(uint i = t.pc; i < end; i++)
{
t.add(re.ir[i].data);
}
for(uint i = 0; i < numS; i++)
{
auto tx = fork(t, t.pc + len, t.counter);
if(tx.idx + s[i] <= n_length)
{
tx.advance(s[i]);
trs ~= tx;
}
}
if(!trs.empty)
t = fetch(trs);
else
goto L_StopThread;
break;
case IR.CodepointSet:
case IR.Trie:
auto set = re.charsets[re.ir[t.pc].data];
uint[4] s;
uint numS;
static if(charSize == 3)
{
s[0] = charSize;
numS = 1;
}
else
{
static if(charSize == 1)
static immutable codeBounds = [0x0, 0x7F, 0x80, 0x7FF, 0x800, 0xFFFF, 0x10000, 0x10FFFF];
else //== 2
static immutable codeBounds = [0x0, 0xFFFF, 0x10000, 0x10FFFF];
uint[] arr = new uint[set.byInterval.length * 2];
size_t ofs = 0;
foreach(ival; set.byInterval)
{
arr[ofs++] = ival.a;
arr[ofs++] = ival.b;
}
auto srange = assumeSorted!"a <= b"(arr);
for(uint i = 0; i < codeBounds.length/2; i++)
{
auto start = srange.lowerBound(codeBounds[2*i]).length;
auto end = srange.lowerBound(codeBounds[2*i+1]).length;
if(end > start || (end == start && (end & 1)))
s[numS++] = (i+1)*charSize;
}
}
if(numS == 0 || t.idx + s[numS-1] > n_length)
goto L_StopThread;
auto chars = set.length;
if(chars > charsetThreshold)
goto L_StopThread;
foreach(ch; set.byCodepoint)
{
//avoid surrogate pairs
if(0xD800 <= ch && ch <= 0xDFFF)
continue;
t.add(ch);
}
for(uint i = 0; i < numS; i++)
{
auto tx = fork(t, t.pc + IRL!(IR.CodepointSet), t.counter);
tx.advance(s[i]);
trs ~= tx;
}
if(!trs.empty)
t = fetch(trs);
else
goto L_StopThread;
break;
case IR.Any:
goto L_StopThread;
case IR.GotoEndOr:
t.pc += IRL!(IR.GotoEndOr)+re.ir[t.pc].data;
assert(re.ir[t.pc].code == IR.OrEnd);
goto case;
case IR.OrEnd:
t.pc += IRL!(IR.OrEnd);
break;
case IR.OrStart:
t.pc += IRL!(IR.OrStart);
goto case;
case IR.Option:
uint next = t.pc + re.ir[t.pc].data + IRL!(IR.Option);
//queue next Option
if(re.ir[next].code == IR.Option)
{
trs ~= fork(t, next, t.counter);
}
t.pc += IRL!(IR.Option);
break;
case IR.RepeatStart:case IR.RepeatQStart:
t.pc += IRL!(IR.RepeatStart)+re.ir[t.pc].data;
goto case IR.RepeatEnd;
case IR.RepeatEnd:
case IR.RepeatQEnd:
uint len = re.ir[t.pc].data;
uint step = re.ir[t.pc+2].raw;
uint min = re.ir[t.pc+3].raw;
if(t.counter < min)
{
t.counter += step;
t.pc -= len;
break;
}
uint max = re.ir[t.pc+4].raw;
if(t.counter < max)
{
trs ~= fork(t, t.pc - len, t.counter + step);
t.counter = t.counter%step;
t.pc += IRL!(IR.RepeatEnd);
}
else
{
t.counter = t.counter%step;
t.pc += IRL!(IR.RepeatEnd);
}
break;
case IR.InfiniteStart, IR.InfiniteQStart:
t.pc += re.ir[t.pc].data + IRL!(IR.InfiniteStart);
goto case IR.InfiniteEnd; //both Q and non-Q
case IR.InfiniteEnd:
case IR.InfiniteQEnd:
uint len = re.ir[t.pc].data;
uint pc1, pc2; //branches to take in priority order
if(++t.hops == 32)
goto L_StopThread;
pc1 = t.pc + IRL!(IR.InfiniteEnd);
pc2 = t.pc - len;
trs ~= fork(t, pc2, t.counter);
t.pc = pc1;
break;
case IR.GroupStart, IR.GroupEnd:
t.pc += IRL!(IR.GroupStart);
break;
case IR.Bol, IR.Wordboundary, IR.Notwordboundary:
t.pc += IRL!(IR.Bol);
break;
case IR.LookaheadStart, IR.NeglookaheadStart, IR.LookbehindStart, IR.NeglookbehindStart:
t.pc += IRL!(IR.LookaheadStart) + IRL!(IR.LookaheadEnd) + re.ir[t.pc].data;
break;
default:
L_StopThread:
assert(re.ir[t.pc].code >= 0x80, text(re.ir[t.pc].code));
debug (fred_search) writeln("ShiftOr stumbled on ",re.ir[t.pc].mnemonic);
n_length = min(t.idx, n_length);
break L_Eval_Thread;
}
}
if(trs.empty)
break;
t = fetch(trs);
}
debug(std_regex_search)
{
writeln("Min length: ", n_length);
}
}
@property bool empty() const { return n_length == 0; }
@property uint length() const{ return n_length/charSize; }
// lookup compatible bit pattern in haystack, return starting index
// has a useful trait: if supplied with valid UTF indexes,
// returns only valid UTF indexes
// (that given the haystack in question is valid UTF string)
@trusted size_t search(const(Char)[] haystack, size_t idx)
{//@BUG: apparently assumes little endian machines
import std.conv, core.stdc.string;
assert(!empty);
auto p = cast(const(ubyte)*)(haystack.ptr+idx);
uint state = uint.max;
uint limit = 1u<<(n_length - 1u);
debug(std_regex_search) writefln("Limit: %32b",limit);
if(fChar != uint.max)
{
const(ubyte)* end = cast(ubyte*)(haystack.ptr + haystack.length);
const orginalAlign = cast(size_t)p & (Char.sizeof-1);
while(p != end)
{
if(!~state)
{//speed up seeking first matching place
for(;;)
{
assert(p <= end, text(p," vs ", end));
p = cast(ubyte*)memchr(p, fChar, end - p);
if(!p)
return haystack.length;
if((cast(size_t)p & (Char.sizeof-1)) == orginalAlign)
break;
if(++p == end)
return haystack.length;
}
state = ~1u;
assert((cast(size_t)p & (Char.sizeof-1)) == orginalAlign);
static if(charSize == 3)
{
state = (state<<1) | table[p[1]];
state = (state<<1) | table[p[2]];
p += 4;
}
else
p++;
//first char is tested, see if that's all
if(!(state & limit))
return (p-cast(ubyte*)haystack.ptr)/Char.sizeof
-length;
}
else
{//have some bits/states for possible matches,
//use the usual shift-or cycle
static if(charSize == 3)
{
state = (state<<1) | table[p[0]];
state = (state<<1) | table[p[1]];
state = (state<<1) | table[p[2]];
p += 4;
}
else
{
state = (state<<1) | table[p[0]];
p++;
}
if(!(state & limit))
return (p-cast(ubyte*)haystack.ptr)/Char.sizeof
-length;
}
debug(std_regex_search) writefln("State: %32b", state);
}
}
else
{
//normal path, partially unrolled for char/wchar
static if(charSize == 3)
{
const(ubyte)* end = cast(ubyte*)(haystack.ptr + haystack.length);
while(p != end)
{
state = (state<<1) | table[p[0]];
state = (state<<1) | table[p[1]];
state = (state<<1) | table[p[2]];
p += 4;
if(!(state & limit))//division rounds down for dchar
return (p-cast(ubyte*)haystack.ptr)/Char.sizeof
-length;
}
}
else
{
auto len = cast(ubyte*)(haystack.ptr + haystack.length) - p;
size_t i = 0;
if(len & 1)
{
state = (state<<1) | table[p[i++]];
if(!(state & limit))
return idx+i/Char.sizeof-length;
}
while(i < len)
{
state = (state<<1) | table[p[i++]];
if(!(state & limit))
return idx+i/Char.sizeof
-length;
state = (state<<1) | table[p[i++]];
if(!(state & limit))
return idx+i/Char.sizeof
-length;
debug(std_regex_search) writefln("State: %32b", state);
}
}
}
return haystack.length;
}
@system debug static void dump(uint[] table)
{//@@@BUG@@@ writef(ln) is @system
import std.stdio;
for(size_t i = 0; i < table.length; i += 4)
{
writefln("%32b %32b %32b %32b",table[i], table[i+1], table[i+2], table[i+3]);
}
}
}
unittest
{
import std.conv, std.regex;
@trusted void test_fixed(alias Kick)()
{
foreach(i, v; TypeTuple!(char, wchar, dchar))
{
alias Char = v;
alias String = immutable(v)[];
auto r = regex(to!String(`abc$`));
auto kick = Kick!Char(r, new uint[256]);
assert(kick.length == 3, text(Kick.stringof," ",v.stringof, " == ", kick.length));
auto r2 = regex(to!String(`(abc){2}a+`));
kick = Kick!Char(r2, new uint[256]);
assert(kick.length == 7, text(Kick.stringof,v.stringof," == ", kick.length));
auto r3 = regex(to!String(`\b(a{2}b{3}){2,4}`));
kick = Kick!Char(r3, new uint[256]);
assert(kick.length == 10, text(Kick.stringof,v.stringof," == ", kick.length));
auto r4 = regex(to!String(`\ba{2}c\bxyz`));
kick = Kick!Char(r4, new uint[256]);
assert(kick.length == 6, text(Kick.stringof,v.stringof, " == ", kick.length));
auto r5 = regex(to!String(`\ba{2}c\b`));
kick = Kick!Char(r5, new uint[256]);
size_t x = kick.search("aabaacaa", 0);
assert(x == 3, text(Kick.stringof,v.stringof," == ", kick.length));
x = kick.search("aabaacaa", x+1);
assert(x == 8, text(Kick.stringof,v.stringof," == ", kick.length));
}
}
@trusted void test_flex(alias Kick)()
{
foreach(i, v;TypeTuple!(char, wchar, dchar))
{
alias Char = v;
alias String = immutable(v)[];
auto r = regex(to!String(`abc[a-z]`));
auto kick = Kick!Char(r, new uint[256]);
auto x = kick.search(to!String("abbabca"), 0);
assert(x == 3, text("real x is ", x, " ",v.stringof));
auto r2 = regex(to!String(`(ax|bd|cdy)`));
String s2 = to!String("abdcdyabax");
kick = Kick!Char(r2, new uint[256]);
x = kick.search(s2, 0);
assert(x == 1, text("real x is ", x));
x = kick.search(s2, x+1);
assert(x == 3, text("real x is ", x));
x = kick.search(s2, x+1);
assert(x == 8, text("real x is ", x));
auto rdot = regex(to!String(`...`));
kick = Kick!Char(rdot, new uint[256]);
assert(kick.length == 0);
auto rN = regex(to!String(`a(b+|c+)x`));
kick = Kick!Char(rN, new uint[256]);
assert(kick.length == 3);
assert(kick.search("ababx",0) == 2);
assert(kick.search("abaacba",0) == 3);//expected inexact
}
}
test_fixed!(ShiftOr)();
test_flex!(ShiftOr)();
}
alias Kickstart = ShiftOr;
|