1 | /* |
2 | ** String scanning. |
3 | ** Copyright (C) 2005-2014 Mike Pall. See Copyright Notice in luajit.h |
4 | */ |
5 | |
6 | #include <math.h> |
7 | |
8 | #define lj_strscan_c |
9 | #define LUA_CORE |
10 | |
11 | #include "lj_obj.h" |
12 | #include "lj_char.h" |
13 | #include "lj_strscan.h" |
14 | |
15 | /* -- Scanning numbers ---------------------------------------------------- */ |
16 | |
17 | /* |
18 | ** Rationale for the builtin string to number conversion library: |
19 | ** |
20 | ** It removes a dependency on libc's strtod(), which is a true portability |
21 | ** nightmare. Mainly due to the plethora of supported OS and toolchain |
22 | ** combinations. Sadly, the various implementations |
23 | ** a) are often buggy, incomplete (no hex floats) and/or imprecise, |
24 | ** b) sometimes crash or hang on certain inputs, |
25 | ** c) return non-standard NaNs that need to be filtered out, and |
26 | ** d) fail if the locale-specific decimal separator is not a dot, |
27 | ** which can only be fixed with atrocious workarounds. |
28 | ** |
29 | ** Also, most of the strtod() implementations are hopelessly bloated, |
30 | ** which is not just an I-cache hog, but a problem for static linkage |
31 | ** on embedded systems, too. |
32 | ** |
33 | ** OTOH the builtin conversion function is very compact. Even though it |
34 | ** does a lot more, like parsing long longs, octal or imaginary numbers |
35 | ** and returning the result in different formats: |
36 | ** a) It needs less than 3 KB (!) of machine code (on x64 with -Os), |
37 | ** b) it doesn't perform any dynamic allocation and, |
38 | ** c) it needs only around 600 bytes of stack space. |
39 | ** |
40 | ** The builtin function is faster than strtod() for typical inputs, e.g. |
41 | ** "123", "1.5" or "1e6". Arguably, it's slower for very large exponents, |
42 | ** which are not very common (this could be fixed, if needed). |
43 | ** |
44 | ** And most importantly, the builtin function is equally precise on all |
45 | ** platforms. It correctly converts and rounds any input to a double. |
46 | ** If this is not the case, please send a bug report -- but PLEASE verify |
47 | ** that the implementation you're comparing to is not the culprit! |
48 | ** |
49 | ** The implementation quickly pre-scans the entire string first and |
50 | ** handles simple integers on-the-fly. Otherwise, it dispatches to the |
51 | ** base-specific parser. Hex and octal is straightforward. |
52 | ** |
53 | ** Decimal to binary conversion uses a fixed-length circular buffer in |
54 | ** base 100. Some simple cases are handled directly. For other cases, the |
55 | ** number in the buffer is up-scaled or down-scaled until the integer part |
56 | ** is in the proper range. Then the integer part is rounded and converted |
57 | ** to a double which is finally rescaled to the result. Denormals need |
58 | ** special treatment to prevent incorrect 'double rounding'. |
59 | */ |
60 | |
61 | /* Definitions for circular decimal digit buffer (base 100 = 2 digits/byte). */ |
62 | #define STRSCAN_DIG 1024 |
63 | #define STRSCAN_MAXDIG 800 /* 772 + extra are sufficient. */ |
64 | #define STRSCAN_DDIG (STRSCAN_DIG/2) |
65 | #define STRSCAN_DMASK (STRSCAN_DDIG-1) |
66 | |
67 | /* Helpers for circular buffer. */ |
68 | #define DNEXT(a) (((a)+1) & STRSCAN_DMASK) |
69 | #define DPREV(a) (((a)-1) & STRSCAN_DMASK) |
70 | #define DLEN(lo, hi) ((int32_t)(((lo)-(hi)) & STRSCAN_DMASK)) |
71 | |
72 | #define casecmp(c, k) (((c) | 0x20) == k) |
73 | |
74 | /* Final conversion to double. */ |
75 | static void strscan_double(uint64_t x, TValue *o, int32_t ex2, int32_t neg) |
76 | { |
77 | double n; |
78 | |
79 | /* Avoid double rounding for denormals. */ |
80 | if (LJ_UNLIKELY(ex2 <= -1075 && x != 0)) { |
81 | /* NYI: all of this generates way too much code on 32 bit CPUs. */ |
82 | #if defined(__GNUC__) && LJ_64 |
83 | int32_t b = (int32_t)(__builtin_clzll(x)^63); |
84 | #else |
85 | int32_t b = (x>>32) ? 32+(int32_t)lj_fls((uint32_t)(x>>32)) : |
86 | (int32_t)lj_fls((uint32_t)x); |
87 | #endif |
88 | if ((int32_t)b + ex2 <= -1023 && (int32_t)b + ex2 >= -1075) { |
89 | uint64_t rb = (uint64_t)1 << (-1075-ex2); |
90 | if ((x & rb) && ((x & (rb+rb+rb-1)))) x += rb+rb; |
91 | x = (x & ~(rb+rb-1)); |
92 | } |
93 | } |
94 | |
95 | /* Convert to double using a signed int64_t conversion, then rescale. */ |
96 | lua_assert((int64_t)x >= 0); |
97 | n = (double)(int64_t)x; |
98 | if (neg) n = -n; |
99 | if (ex2) n = ldexp(n, ex2); |
100 | o->n = n; |
101 | } |
102 | |
103 | /* Parse hexadecimal number. */ |
104 | static StrScanFmt strscan_hex(const uint8_t *p, TValue *o, |
105 | StrScanFmt fmt, uint32_t opt, |
106 | int32_t ex2, int32_t neg, uint32_t dig) |
107 | { |
108 | uint64_t x = 0; |
109 | uint32_t i; |
110 | |
111 | /* Scan hex digits. */ |
112 | for (i = dig > 16 ? 16 : dig ; i; i--, p++) { |
113 | uint32_t d = (*p != '.' ? *p : *++p); if (d > '9') d += 9; |
114 | x = (x << 4) + (d & 15); |
115 | } |
116 | |
117 | /* Summarize rounding-effect of excess digits. */ |
118 | for (i = 16; i < dig; i++, p++) |
119 | x |= ((*p != '.' ? *p : *++p) != '0'), ex2 += 4; |
120 | |
121 | /* Format-specific handling. */ |
122 | switch (fmt) { |
123 | case STRSCAN_INT: |
124 | if (!(opt & STRSCAN_OPT_TONUM) && x < 0x80000000u+neg) { |
125 | o->i = neg ? -(int32_t)x : (int32_t)x; |
126 | return STRSCAN_INT; /* Fast path for 32 bit integers. */ |
127 | } |
128 | if (!(opt & STRSCAN_OPT_C)) { fmt = STRSCAN_NUM; break; } |
129 | /* fallthrough */ |
130 | case STRSCAN_U32: |
131 | if (dig > 8) return STRSCAN_ERROR; |
132 | o->i = neg ? -(int32_t)x : (int32_t)x; |
133 | return STRSCAN_U32; |
134 | case STRSCAN_I64: |
135 | case STRSCAN_U64: |
136 | if (dig > 16) return STRSCAN_ERROR; |
137 | o->u64 = neg ? (uint64_t)-(int64_t)x : x; |
138 | return fmt; |
139 | default: |
140 | break; |
141 | } |
142 | |
143 | /* Reduce range then convert to double. */ |
144 | if ((x & U64x(c0000000,0000000))) { x = (x >> 2) | (x & 3); ex2 += 2; } |
145 | strscan_double(x, o, ex2, neg); |
146 | return fmt; |
147 | } |
148 | |
149 | /* Parse octal number. */ |
150 | static StrScanFmt strscan_oct(const uint8_t *p, TValue *o, |
151 | StrScanFmt fmt, int32_t neg, uint32_t dig) |
152 | { |
153 | uint64_t x = 0; |
154 | |
155 | /* Scan octal digits. */ |
156 | if (dig > 22 || (dig == 22 && *p > '1')) return STRSCAN_ERROR; |
157 | while (dig-- > 0) { |
158 | if (!(*p >= '0' && *p <= '7')) return STRSCAN_ERROR; |
159 | x = (x << 3) + (*p++ & 7); |
160 | } |
161 | |
162 | /* Format-specific handling. */ |
163 | switch (fmt) { |
164 | case STRSCAN_INT: |
165 | if (x >= 0x80000000u+neg) fmt = STRSCAN_U32; |
166 | /* fallthrough */ |
167 | case STRSCAN_U32: |
168 | if ((x >> 32)) return STRSCAN_ERROR; |
169 | o->i = neg ? -(int32_t)x : (int32_t)x; |
170 | break; |
171 | default: |
172 | case STRSCAN_I64: |
173 | case STRSCAN_U64: |
174 | o->u64 = neg ? (uint64_t)-(int64_t)x : x; |
175 | break; |
176 | } |
177 | return fmt; |
178 | } |
179 | |
180 | /* Parse decimal number. */ |
181 | static StrScanFmt strscan_dec(const uint8_t *p, TValue *o, |
182 | StrScanFmt fmt, uint32_t opt, |
183 | int32_t ex10, int32_t neg, uint32_t dig) |
184 | { |
185 | uint8_t xi[STRSCAN_DDIG], *xip = xi; |
186 | |
187 | if (dig) { |
188 | uint32_t i = dig; |
189 | if (i > STRSCAN_MAXDIG) { |
190 | ex10 += (int32_t)(i - STRSCAN_MAXDIG); |
191 | i = STRSCAN_MAXDIG; |
192 | } |
193 | /* Scan unaligned leading digit. */ |
194 | if (((ex10^i) & 1)) |
195 | *xip++ = ((*p != '.' ? *p : *++p) & 15), i--, p++; |
196 | /* Scan aligned double-digits. */ |
197 | for ( ; i > 1; i -= 2) { |
198 | uint32_t d = 10 * ((*p != '.' ? *p : *++p) & 15); p++; |
199 | *xip++ = d + ((*p != '.' ? *p : *++p) & 15); p++; |
200 | } |
201 | /* Scan and realign trailing digit. */ |
202 | if (i) *xip++ = 10 * ((*p != '.' ? *p : *++p) & 15), ex10--, p++; |
203 | |
204 | /* Summarize rounding-effect of excess digits. */ |
205 | if (dig > STRSCAN_MAXDIG) { |
206 | do { |
207 | if ((*p != '.' ? *p : *++p) != '0') { xip[-1] |= 1; break; } |
208 | p++; |
209 | } while (--dig > STRSCAN_MAXDIG); |
210 | dig = STRSCAN_MAXDIG; |
211 | } else { /* Simplify exponent. */ |
212 | while (ex10 > 0 && dig <= 18) *xip++ = 0, ex10 -= 2, dig += 2; |
213 | } |
214 | } else { /* Only got zeros. */ |
215 | ex10 = 0; |
216 | xi[0] = 0; |
217 | } |
218 | |
219 | /* Fast path for numbers in integer format (but handles e.g. 1e6, too). */ |
220 | if (dig <= 20 && ex10 == 0) { |
221 | uint8_t *xis; |
222 | uint64_t x = xi[0]; |
223 | double n; |
224 | for (xis = xi+1; xis < xip; xis++) x = x * 100 + *xis; |
225 | if (!(dig == 20 && (xi[0] > 18 || (int64_t)x >= 0))) { /* No overflow? */ |
226 | /* Format-specific handling. */ |
227 | switch (fmt) { |
228 | case STRSCAN_INT: |
229 | if (!(opt & STRSCAN_OPT_TONUM) && x < 0x80000000u+neg) { |
230 | o->i = neg ? -(int32_t)x : (int32_t)x; |
231 | return STRSCAN_INT; /* Fast path for 32 bit integers. */ |
232 | } |
233 | if (!(opt & STRSCAN_OPT_C)) { fmt = STRSCAN_NUM; goto plainnumber; } |
234 | /* fallthrough */ |
235 | case STRSCAN_U32: |
236 | if ((x >> 32) != 0) return STRSCAN_ERROR; |
237 | o->i = neg ? -(int32_t)x : (int32_t)x; |
238 | return STRSCAN_U32; |
239 | case STRSCAN_I64: |
240 | case STRSCAN_U64: |
241 | o->u64 = neg ? (uint64_t)-(int64_t)x : x; |
242 | return fmt; |
243 | default: |
244 | plainnumber: /* Fast path for plain numbers < 2^63. */ |
245 | if ((int64_t)x < 0) break; |
246 | n = (double)(int64_t)x; |
247 | if (neg) n = -n; |
248 | o->n = n; |
249 | return fmt; |
250 | } |
251 | } |
252 | } |
253 | |
254 | /* Slow non-integer path. */ |
255 | if (fmt == STRSCAN_INT) { |
256 | if ((opt & STRSCAN_OPT_C)) return STRSCAN_ERROR; |
257 | fmt = STRSCAN_NUM; |
258 | } else if (fmt > STRSCAN_INT) { |
259 | return STRSCAN_ERROR; |
260 | } |
261 | { |
262 | uint32_t hi = 0, lo = (uint32_t)(xip-xi); |
263 | int32_t ex2 = 0, idig = (int32_t)lo + (ex10 >> 1); |
264 | |
265 | lua_assert(lo > 0 && (ex10 & 1) == 0); |
266 | |
267 | /* Handle simple overflow/underflow. */ |
268 | if (idig > 310/2) { if (neg) setminfV(o); else setpinfV(o); return fmt; } |
269 | else if (idig < -326/2) { o->n = neg ? -0.0 : 0.0; return fmt; } |
270 | |
271 | /* Scale up until we have at least 17 or 18 integer part digits. */ |
272 | while (idig < 9 && idig < DLEN(lo, hi)) { |
273 | uint32_t i, cy = 0; |
274 | ex2 -= 6; |
275 | for (i = DPREV(lo); ; i = DPREV(i)) { |
276 | uint32_t d = (xi[i] << 6) + cy; |
277 | cy = (((d >> 2) * 5243) >> 17); d = d - cy * 100; /* Div/mod 100. */ |
278 | xi[i] = (uint8_t)d; |
279 | if (i == hi) break; |
280 | if (d == 0 && i == DPREV(lo)) lo = i; |
281 | } |
282 | if (cy) { |
283 | hi = DPREV(hi); |
284 | if (xi[DPREV(lo)] == 0) lo = DPREV(lo); |
285 | else if (hi == lo) { lo = DPREV(lo); xi[DPREV(lo)] |= xi[lo]; } |
286 | xi[hi] = (uint8_t)cy; idig++; |
287 | } |
288 | } |
289 | |
290 | /* Scale down until no more than 17 or 18 integer part digits remain. */ |
291 | while (idig > 9) { |
292 | uint32_t i, cy = 0; |
293 | ex2 += 6; |
294 | for (i = hi; i != lo; i = DNEXT(i)) { |
295 | cy += xi[i]; |
296 | xi[i] = (cy >> 6); |
297 | cy = 100 * (cy & 0x3f); |
298 | if (xi[i] == 0 && i == hi) hi = DNEXT(hi), idig--; |
299 | } |
300 | while (cy) { |
301 | if (hi == lo) { xi[DPREV(lo)] |= 1; break; } |
302 | xi[lo] = (cy >> 6); lo = DNEXT(lo); |
303 | cy = 100 * (cy & 0x3f); |
304 | } |
305 | } |
306 | |
307 | /* Collect integer part digits and convert to rescaled double. */ |
308 | { |
309 | uint64_t x = xi[hi]; |
310 | uint32_t i; |
311 | for (i = DNEXT(hi); --idig > 0 && i != lo; i = DNEXT(i)) |
312 | x = x * 100 + xi[i]; |
313 | if (i == lo) { |
314 | while (--idig >= 0) x = x * 100; |
315 | } else { /* Gather round bit from remaining digits. */ |
316 | x <<= 1; ex2--; |
317 | do { |
318 | if (xi[i]) { x |= 1; break; } |
319 | i = DNEXT(i); |
320 | } while (i != lo); |
321 | } |
322 | strscan_double(x, o, ex2, neg); |
323 | } |
324 | } |
325 | return fmt; |
326 | } |
327 | |
328 | /* Scan string containing a number. Returns format. Returns value in o. */ |
329 | StrScanFmt lj_strscan_scan(const uint8_t *p, TValue *o, uint32_t opt) |
330 | { |
331 | int32_t neg = 0; |
332 | |
333 | /* Remove leading space, parse sign and non-numbers. */ |
334 | if (LJ_UNLIKELY(!lj_char_isdigit(*p))) { |
335 | while (lj_char_isspace(*p)) p++; |
336 | if (*p == '+' || *p == '-') neg = (*p++ == '-'); |
337 | if (LJ_UNLIKELY(*p >= 'A')) { /* Parse "inf", "infinity" or "nan". */ |
338 | TValue tmp; |
339 | setnanV(&tmp); |
340 | if (casecmp(p[0],'i') && casecmp(p[1],'n') && casecmp(p[2],'f')) { |
341 | if (neg) setminfV(&tmp); else setpinfV(&tmp); |
342 | p += 3; |
343 | if (casecmp(p[0],'i') && casecmp(p[1],'n') && casecmp(p[2],'i') && |
344 | casecmp(p[3],'t') && casecmp(p[4],'y')) p += 5; |
345 | } else if (casecmp(p[0],'n') && casecmp(p[1],'a') && casecmp(p[2],'n')) { |
346 | p += 3; |
347 | } |
348 | while (lj_char_isspace(*p)) p++; |
349 | if (*p) return STRSCAN_ERROR; |
350 | o->u64 = tmp.u64; |
351 | return STRSCAN_NUM; |
352 | } |
353 | } |
354 | |
355 | /* Parse regular number. */ |
356 | { |
357 | StrScanFmt fmt = STRSCAN_INT; |
358 | int cmask = LJ_CHAR_DIGIT; |
359 | int base = (opt & STRSCAN_OPT_C) && *p == '0' ? 0 : 10; |
360 | const uint8_t *sp, *dp = NULL; |
361 | uint32_t dig = 0, hasdig = 0, x = 0; |
362 | int32_t ex = 0; |
363 | |
364 | /* Determine base and skip leading zeros. */ |
365 | if (LJ_UNLIKELY(*p <= '0')) { |
366 | if (*p == '0' && casecmp(p[1], 'x')) |
367 | base = 16, cmask = LJ_CHAR_XDIGIT, p += 2; |
368 | for ( ; ; p++) { |
369 | if (*p == '0') { |
370 | hasdig = 1; |
371 | } else if (*p == '.') { |
372 | if (dp) return STRSCAN_ERROR; |
373 | dp = p; |
374 | } else { |
375 | break; |
376 | } |
377 | } |
378 | } |
379 | |
380 | /* Preliminary digit and decimal point scan. */ |
381 | for (sp = p; ; p++) { |
382 | if (LJ_LIKELY(lj_char_isa(*p, cmask))) { |
383 | x = x * 10 + (*p & 15); /* For fast path below. */ |
384 | dig++; |
385 | } else if (*p == '.') { |
386 | if (dp) return STRSCAN_ERROR; |
387 | dp = p; |
388 | } else { |
389 | break; |
390 | } |
391 | } |
392 | if (!(hasdig | dig)) return STRSCAN_ERROR; |
393 | |
394 | /* Handle decimal point. */ |
395 | if (dp) { |
396 | fmt = STRSCAN_NUM; |
397 | if (dig) { |
398 | ex = (int32_t)(dp-(p-1)); dp = p-1; |
399 | while (ex < 0 && *dp-- == '0') ex++, dig--; /* Skip trailing zeros. */ |
400 | if (base == 16) ex *= 4; |
401 | } |
402 | } |
403 | |
404 | /* Parse exponent. */ |
405 | if (casecmp(*p, (uint32_t)(base == 16 ? 'p' : 'e'))) { |
406 | uint32_t xx; |
407 | int negx = 0; |
408 | fmt = STRSCAN_NUM; p++; |
409 | if (*p == '+' || *p == '-') negx = (*p++ == '-'); |
410 | if (!lj_char_isdigit(*p)) return STRSCAN_ERROR; |
411 | xx = (*p++ & 15); |
412 | while (lj_char_isdigit(*p)) { |
413 | if (xx < 65536) xx = xx * 10 + (*p & 15); |
414 | p++; |
415 | } |
416 | ex += negx ? -(int32_t)xx : (int32_t)xx; |
417 | } |
418 | |
419 | /* Parse suffix. */ |
420 | if (*p) { |
421 | /* I (IMAG), U (U32), LL (I64), ULL/LLU (U64), L (long), UL/LU (ulong). */ |
422 | /* NYI: f (float). Not needed until cp_number() handles non-integers. */ |
423 | if (casecmp(*p, 'i')) { |
424 | if (!(opt & STRSCAN_OPT_IMAG)) return STRSCAN_ERROR; |
425 | p++; fmt = STRSCAN_IMAG; |
426 | } else if (fmt == STRSCAN_INT) { |
427 | if (casecmp(*p, 'u')) p++, fmt = STRSCAN_U32; |
428 | if (casecmp(*p, 'l')) { |
429 | p++; |
430 | if (casecmp(*p, 'l')) p++, fmt += STRSCAN_I64 - STRSCAN_INT; |
431 | else if (!(opt & STRSCAN_OPT_C)) return STRSCAN_ERROR; |
432 | else if (sizeof(long) == 8) fmt += STRSCAN_I64 - STRSCAN_INT; |
433 | } |
434 | if (casecmp(*p, 'u') && (fmt == STRSCAN_INT || fmt == STRSCAN_I64)) |
435 | p++, fmt += STRSCAN_U32 - STRSCAN_INT; |
436 | if ((fmt == STRSCAN_U32 && !(opt & STRSCAN_OPT_C)) || |
437 | (fmt >= STRSCAN_I64 && !(opt & STRSCAN_OPT_LL))) |
438 | return STRSCAN_ERROR; |
439 | } |
440 | while (lj_char_isspace(*p)) p++; |
441 | if (*p) return STRSCAN_ERROR; |
442 | } |
443 | |
444 | /* Fast path for decimal 32 bit integers. */ |
445 | if (fmt == STRSCAN_INT && base == 10 && |
446 | (dig < 10 || (dig == 10 && *sp <= '2' && x < 0x80000000u+neg))) { |
447 | int32_t y = neg ? -(int32_t)x : (int32_t)x; |
448 | if ((opt & STRSCAN_OPT_TONUM)) { |
449 | o->n = (double)y; |
450 | return STRSCAN_NUM; |
451 | } else { |
452 | o->i = y; |
453 | return STRSCAN_INT; |
454 | } |
455 | } |
456 | |
457 | /* Dispatch to base-specific parser. */ |
458 | if (base == 0 && !(fmt == STRSCAN_NUM || fmt == STRSCAN_IMAG)) |
459 | return strscan_oct(sp, o, fmt, neg, dig); |
460 | if (base == 16) |
461 | fmt = strscan_hex(sp, o, fmt, opt, ex, neg, dig); |
462 | else |
463 | fmt = strscan_dec(sp, o, fmt, opt, ex, neg, dig); |
464 | |
465 | /* Try to convert number to integer, if requested. */ |
466 | if (fmt == STRSCAN_NUM && (opt & STRSCAN_OPT_TOINT)) { |
467 | double n = o->n; |
468 | int32_t i = lj_num2int(n); |
469 | if (n == (lua_Number)i) { o->i = i; return STRSCAN_INT; } |
470 | } |
471 | return fmt; |
472 | } |
473 | } |
474 | |
475 | int LJ_FASTCALL lj_strscan_num(GCstr *str, TValue *o) |
476 | { |
477 | StrScanFmt fmt = lj_strscan_scan((const uint8_t *)strdata(str), o, |
478 | STRSCAN_OPT_TONUM); |
479 | lua_assert(fmt == STRSCAN_ERROR || fmt == STRSCAN_NUM); |
480 | return (fmt != STRSCAN_ERROR); |
481 | } |
482 | |
483 | #if LJ_DUALNUM |
484 | int LJ_FASTCALL lj_strscan_number(GCstr *str, TValue *o) |
485 | { |
486 | StrScanFmt fmt = lj_strscan_scan((const uint8_t *)strdata(str), o, |
487 | STRSCAN_OPT_TOINT); |
488 | lua_assert(fmt == STRSCAN_ERROR || fmt == STRSCAN_NUM || fmt == STRSCAN_INT); |
489 | if (fmt == STRSCAN_INT) setitype(o, LJ_TISNUM); |
490 | return (fmt != STRSCAN_ERROR); |
491 | } |
492 | #endif |
493 | |
494 | #undef DNEXT |
495 | #undef DPREV |
496 | #undef DLEN |
497 | |
498 | |