… | |
… | |
27 | #define F_SPACE_AFTER 0x00000040UL |
27 | #define F_SPACE_AFTER 0x00000040UL |
28 | #define F_ALLOW_NONREF 0x00000100UL |
28 | #define F_ALLOW_NONREF 0x00000100UL |
29 | #define F_SHRINK 0x00000200UL |
29 | #define F_SHRINK 0x00000200UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
|
|
32 | #define F_RELAXED 0x00001000UL |
|
|
33 | |
32 | #define F_MAXDEPTH 0xf8000000UL |
34 | #define F_MAXDEPTH 0xf8000000UL |
33 | #define S_MAXDEPTH 27 |
35 | #define S_MAXDEPTH 27 |
34 | #define F_MAXSIZE 0x01f00000UL |
36 | #define F_MAXSIZE 0x01f00000UL |
35 | #define S_MAXSIZE 20 |
37 | #define S_MAXSIZE 20 |
36 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
38 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
… | |
… | |
48 | |
50 | |
49 | #define SB do { |
51 | #define SB do { |
50 | #define SE } while (0) |
52 | #define SE } while (0) |
51 | |
53 | |
52 | #if __GNUC__ >= 3 |
54 | #if __GNUC__ >= 3 |
53 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
55 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
54 | # define inline inline |
56 | # define INLINE static inline |
55 | #else |
57 | #else |
56 | # define expect(expr,value) (expr) |
58 | # define expect(expr,value) (expr) |
57 | # define inline static |
59 | # define INLINE static |
58 | #endif |
60 | #endif |
59 | |
61 | |
60 | #define expect_false(expr) expect ((expr) != 0, 0) |
62 | #define expect_false(expr) expect ((expr) != 0, 0) |
61 | #define expect_true(expr) expect ((expr) != 0, 1) |
63 | #define expect_true(expr) expect ((expr) != 0, 1) |
62 | |
64 | |
… | |
… | |
78 | } JSON; |
80 | } JSON; |
79 | |
81 | |
80 | ///////////////////////////////////////////////////////////////////////////// |
82 | ///////////////////////////////////////////////////////////////////////////// |
81 | // utility functions |
83 | // utility functions |
82 | |
84 | |
83 | inline void |
85 | INLINE void |
84 | shrink (SV *sv) |
86 | shrink (SV *sv) |
85 | { |
87 | { |
86 | sv_utf8_downgrade (sv, 1); |
88 | sv_utf8_downgrade (sv, 1); |
87 | if (SvLEN (sv) > SvCUR (sv) + 1) |
89 | if (SvLEN (sv) > SvCUR (sv) + 1) |
88 | { |
90 | { |
… | |
… | |
97 | // decode an utf-8 character and return it, or (UV)-1 in |
99 | // decode an utf-8 character and return it, or (UV)-1 in |
98 | // case of an error. |
100 | // case of an error. |
99 | // we special-case "safe" characters from U+80 .. U+7FF, |
101 | // we special-case "safe" characters from U+80 .. U+7FF, |
100 | // but use the very good perl function to parse anything else. |
102 | // but use the very good perl function to parse anything else. |
101 | // note that we never call this function for a ascii codepoints |
103 | // note that we never call this function for a ascii codepoints |
102 | inline UV |
104 | INLINE UV |
103 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
105 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
104 | { |
106 | { |
105 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
107 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
106 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
108 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
107 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
109 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
… | |
… | |
126 | char *end; // SvEND (sv) |
128 | char *end; // SvEND (sv) |
127 | SV *sv; // result scalar |
129 | SV *sv; // result scalar |
128 | JSON json; |
130 | JSON json; |
129 | U32 indent; // indentation level |
131 | U32 indent; // indentation level |
130 | U32 maxdepth; // max. indentation/recursion level |
132 | U32 maxdepth; // max. indentation/recursion level |
|
|
133 | UV limit; // escape character values >= this value when encoding |
131 | } enc_t; |
134 | } enc_t; |
132 | |
135 | |
133 | inline void |
136 | INLINE void |
134 | need (enc_t *enc, STRLEN len) |
137 | need (enc_t *enc, STRLEN len) |
135 | { |
138 | { |
136 | if (expect_false (enc->cur + len >= enc->end)) |
139 | if (expect_false (enc->cur + len >= enc->end)) |
137 | { |
140 | { |
138 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
141 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
… | |
… | |
140 | enc->cur = SvPVX (enc->sv) + cur; |
143 | enc->cur = SvPVX (enc->sv) + cur; |
141 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
144 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
142 | } |
145 | } |
143 | } |
146 | } |
144 | |
147 | |
145 | inline void |
148 | INLINE void |
146 | encode_ch (enc_t *enc, char ch) |
149 | encode_ch (enc_t *enc, char ch) |
147 | { |
150 | { |
148 | need (enc, 1); |
151 | need (enc, 1); |
149 | *enc->cur++ = ch; |
152 | *enc->cur++ = ch; |
150 | } |
153 | } |
… | |
… | |
204 | { |
207 | { |
205 | uch = ch; |
208 | uch = ch; |
206 | clen = 1; |
209 | clen = 1; |
207 | } |
210 | } |
208 | |
211 | |
209 | if (uch > 0x10FFFFUL) |
212 | if (uch < 0x20 || uch >= enc->limit) |
210 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
211 | |
|
|
212 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
|
|
213 | { |
213 | { |
214 | if (uch > 0xFFFFUL) |
214 | if (uch > 0xFFFFUL) |
215 | { |
215 | { |
|
|
216 | if (uch > 0x10FFFFUL) |
|
|
217 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
218 | |
216 | need (enc, len += 11); |
219 | need (enc, len += 11); |
217 | sprintf (enc->cur, "\\u%04x\\u%04x", |
220 | sprintf (enc->cur, "\\u%04x\\u%04x", |
218 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
221 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
219 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
222 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
220 | enc->cur += 12; |
223 | enc->cur += 12; |
… | |
… | |
259 | |
262 | |
260 | --len; |
263 | --len; |
261 | } |
264 | } |
262 | } |
265 | } |
263 | |
266 | |
264 | inline void |
267 | INLINE void |
265 | encode_indent (enc_t *enc) |
268 | encode_indent (enc_t *enc) |
266 | { |
269 | { |
267 | if (enc->json.flags & F_INDENT) |
270 | if (enc->json.flags & F_INDENT) |
268 | { |
271 | { |
269 | int spaces = enc->indent * INDENT_STEP; |
272 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
272 | memset (enc->cur, ' ', spaces); |
275 | memset (enc->cur, ' ', spaces); |
273 | enc->cur += spaces; |
276 | enc->cur += spaces; |
274 | } |
277 | } |
275 | } |
278 | } |
276 | |
279 | |
277 | inline void |
280 | INLINE void |
278 | encode_space (enc_t *enc) |
281 | encode_space (enc_t *enc) |
279 | { |
282 | { |
280 | need (enc, 1); |
283 | need (enc, 1); |
281 | encode_ch (enc, ' '); |
284 | encode_ch (enc, ' '); |
282 | } |
285 | } |
283 | |
286 | |
284 | inline void |
287 | INLINE void |
285 | encode_nl (enc_t *enc) |
288 | encode_nl (enc_t *enc) |
286 | { |
289 | { |
287 | if (enc->json.flags & F_INDENT) |
290 | if (enc->json.flags & F_INDENT) |
288 | { |
291 | { |
289 | need (enc, 1); |
292 | need (enc, 1); |
290 | encode_ch (enc, '\n'); |
293 | encode_ch (enc, '\n'); |
291 | } |
294 | } |
292 | } |
295 | } |
293 | |
296 | |
294 | inline void |
297 | INLINE void |
295 | encode_comma (enc_t *enc) |
298 | encode_comma (enc_t *enc) |
296 | { |
299 | { |
297 | encode_ch (enc, ','); |
300 | encode_ch (enc, ','); |
298 | |
301 | |
299 | if (enc->json.flags & F_INDENT) |
302 | if (enc->json.flags & F_INDENT) |
… | |
… | |
310 | int i, len = av_len (av); |
313 | int i, len = av_len (av); |
311 | |
314 | |
312 | if (enc->indent >= enc->maxdepth) |
315 | if (enc->indent >= enc->maxdepth) |
313 | croak ("data structure too deep (hit recursion limit)"); |
316 | croak ("data structure too deep (hit recursion limit)"); |
314 | |
317 | |
315 | encode_ch (enc, '['); encode_nl (enc); |
318 | encode_ch (enc, '['); |
316 | ++enc->indent; |
319 | |
|
|
320 | if (len >= 0) |
|
|
321 | { |
|
|
322 | encode_nl (enc); ++enc->indent; |
317 | |
323 | |
318 | for (i = 0; i <= len; ++i) |
324 | for (i = 0; i <= len; ++i) |
319 | { |
325 | { |
320 | SV **svp = av_fetch (av, i, 0); |
326 | SV **svp = av_fetch (av, i, 0); |
321 | |
327 | |
322 | encode_indent (enc); |
328 | encode_indent (enc); |
323 | |
329 | |
324 | if (svp) |
330 | if (svp) |
325 | encode_sv (enc, *svp); |
331 | encode_sv (enc, *svp); |
326 | else |
332 | else |
327 | encode_str (enc, "null", 4, 0); |
333 | encode_str (enc, "null", 4, 0); |
328 | |
334 | |
329 | if (i < len) |
335 | if (i < len) |
330 | encode_comma (enc); |
336 | encode_comma (enc); |
331 | } |
337 | } |
332 | |
338 | |
|
|
339 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
340 | } |
|
|
341 | |
333 | encode_nl (enc); |
342 | encode_ch (enc, ']'); |
334 | |
|
|
335 | --enc->indent; |
|
|
336 | encode_indent (enc); encode_ch (enc, ']'); |
|
|
337 | } |
343 | } |
338 | |
344 | |
339 | static void |
345 | static void |
340 | encode_he (enc_t *enc, HE *he) |
346 | encode_hk (enc_t *enc, HE *he) |
341 | { |
347 | { |
342 | encode_ch (enc, '"'); |
348 | encode_ch (enc, '"'); |
343 | |
349 | |
344 | if (HeKLEN (he) == HEf_SVKEY) |
350 | if (HeKLEN (he) == HEf_SVKEY) |
345 | { |
351 | { |
… | |
… | |
358 | encode_ch (enc, '"'); |
364 | encode_ch (enc, '"'); |
359 | |
365 | |
360 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
366 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
361 | encode_ch (enc, ':'); |
367 | encode_ch (enc, ':'); |
362 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
368 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
363 | encode_sv (enc, HeVAL (he)); |
|
|
364 | } |
369 | } |
365 | |
370 | |
366 | // compare hash entries, used when all keys are bytestrings |
371 | // compare hash entries, used when all keys are bytestrings |
367 | static int |
372 | static int |
368 | he_cmp_fast (const void *a_, const void *b_) |
373 | he_cmp_fast (const void *a_, const void *b_) |
… | |
… | |
373 | HE *b = *(HE **)b_; |
378 | HE *b = *(HE **)b_; |
374 | |
379 | |
375 | STRLEN la = HeKLEN (a); |
380 | STRLEN la = HeKLEN (a); |
376 | STRLEN lb = HeKLEN (b); |
381 | STRLEN lb = HeKLEN (b); |
377 | |
382 | |
378 | if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
383 | if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
379 | cmp = la - lb; |
384 | cmp = lb - la; |
380 | |
385 | |
381 | return cmp; |
386 | return cmp; |
382 | } |
387 | } |
383 | |
388 | |
384 | // compare hash entries, used when some keys are sv's or utf-x |
389 | // compare hash entries, used when some keys are sv's or utf-x |
385 | static int |
390 | static int |
386 | he_cmp_slow (const void *a, const void *b) |
391 | he_cmp_slow (const void *a, const void *b) |
387 | { |
392 | { |
388 | return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
393 | return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
389 | } |
394 | } |
390 | |
395 | |
391 | static void |
396 | static void |
392 | encode_hv (enc_t *enc, HV *hv) |
397 | encode_hv (enc_t *enc, HV *hv) |
393 | { |
398 | { |
|
|
399 | HE *he; |
394 | int count, i; |
400 | int count; |
395 | |
401 | |
396 | if (enc->indent >= enc->maxdepth) |
402 | if (enc->indent >= enc->maxdepth) |
397 | croak ("data structure too deep (hit recursion limit)"); |
403 | croak ("data structure too deep (hit recursion limit)"); |
398 | |
404 | |
399 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
405 | encode_ch (enc, '{'); |
400 | |
406 | |
401 | if ((count = hv_iterinit (hv))) |
|
|
402 | { |
|
|
403 | // for canonical output we have to sort by keys first |
407 | // for canonical output we have to sort by keys first |
404 | // actually, this is mostly due to the stupid so-called |
408 | // actually, this is mostly due to the stupid so-called |
405 | // security workaround added somewhere in 5.8.x. |
409 | // security workaround added somewhere in 5.8.x. |
406 | // that randomises hash orderings |
410 | // that randomises hash orderings |
407 | if (enc->json.flags & F_CANONICAL) |
411 | if (enc->json.flags & F_CANONICAL) |
|
|
412 | { |
|
|
413 | int count = hv_iterinit (hv); |
|
|
414 | |
|
|
415 | if (SvMAGICAL (hv)) |
408 | { |
416 | { |
|
|
417 | // need to count by iterating. could improve by dynamically building the vector below |
|
|
418 | // but I don't care for the speed of this special case. |
|
|
419 | // note also that we will run into undefined behaviour when the two iterations |
|
|
420 | // do not result in the same count, something I might care for in some later release. |
|
|
421 | |
|
|
422 | count = 0; |
|
|
423 | while (hv_iternext (hv)) |
|
|
424 | ++count; |
|
|
425 | |
|
|
426 | hv_iterinit (hv); |
|
|
427 | } |
|
|
428 | |
|
|
429 | if (count) |
|
|
430 | { |
409 | int fast = 1; |
431 | int i, fast = 1; |
410 | HE *he; |
|
|
411 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
432 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
412 | HE **hes = _alloca (count * sizeof (HE)); |
433 | HE **hes = _alloca (count * sizeof (HE)); |
413 | #else |
434 | #else |
414 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
435 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
415 | #endif |
436 | #endif |
… | |
… | |
442 | |
463 | |
443 | FREETMPS; |
464 | FREETMPS; |
444 | LEAVE; |
465 | LEAVE; |
445 | } |
466 | } |
446 | |
467 | |
447 | for (i = 0; i < count; ++i) |
468 | encode_nl (enc); ++enc->indent; |
|
|
469 | |
|
|
470 | while (count--) |
448 | { |
471 | { |
449 | encode_indent (enc); |
472 | encode_indent (enc); |
|
|
473 | he = hes [count]; |
450 | encode_he (enc, hes [i]); |
474 | encode_hk (enc, he); |
|
|
475 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
451 | |
476 | |
452 | if (i < count - 1) |
477 | if (count) |
453 | encode_comma (enc); |
478 | encode_comma (enc); |
454 | } |
479 | } |
455 | |
480 | |
456 | encode_nl (enc); |
481 | encode_nl (enc); --enc->indent; encode_indent (enc); |
457 | } |
482 | } |
|
|
483 | } |
458 | else |
484 | else |
|
|
485 | { |
|
|
486 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
|
|
487 | if ((he = hv_iternext (hv))) |
459 | { |
488 | { |
460 | HE *he = hv_iternext (hv); |
489 | encode_nl (enc); ++enc->indent; |
461 | |
490 | |
462 | for (;;) |
491 | for (;;) |
463 | { |
492 | { |
464 | encode_indent (enc); |
493 | encode_indent (enc); |
465 | encode_he (enc, he); |
494 | encode_hk (enc, he); |
|
|
495 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
466 | |
496 | |
467 | if (!(he = hv_iternext (hv))) |
497 | if (!(he = hv_iternext (hv))) |
468 | break; |
498 | break; |
469 | |
499 | |
470 | encode_comma (enc); |
500 | encode_comma (enc); |
471 | } |
501 | } |
472 | |
502 | |
473 | encode_nl (enc); |
503 | encode_nl (enc); --enc->indent; encode_indent (enc); |
474 | } |
504 | } |
475 | } |
505 | } |
476 | |
506 | |
477 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
507 | encode_ch (enc, '}'); |
478 | } |
508 | } |
479 | |
509 | |
480 | // encode objects, arrays and special \0=false and \1=true values. |
510 | // encode objects, arrays and special \0=false and \1=true values. |
481 | static void |
511 | static void |
482 | encode_rv (enc_t *enc, SV *sv) |
512 | encode_rv (enc_t *enc, SV *sv) |
… | |
… | |
651 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
681 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
652 | enc.cur = SvPVX (enc.sv); |
682 | enc.cur = SvPVX (enc.sv); |
653 | enc.end = SvEND (enc.sv); |
683 | enc.end = SvEND (enc.sv); |
654 | enc.indent = 0; |
684 | enc.indent = 0; |
655 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
685 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
|
|
686 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
|
|
687 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
|
|
688 | : 0x10FFFFUL; |
656 | |
689 | |
657 | SvPOK_only (enc.sv); |
690 | SvPOK_only (enc.sv); |
658 | encode_sv (&enc, scalar); |
691 | encode_sv (&enc, scalar); |
659 | |
692 | |
660 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
693 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
… | |
… | |
681 | JSON json; |
714 | JSON json; |
682 | U32 depth; // recursion depth |
715 | U32 depth; // recursion depth |
683 | U32 maxdepth; // recursion depth limit |
716 | U32 maxdepth; // recursion depth limit |
684 | } dec_t; |
717 | } dec_t; |
685 | |
718 | |
686 | inline void |
719 | INLINE void |
|
|
720 | decode_comment (dec_t *dec) |
|
|
721 | { |
|
|
722 | // only '#'-style comments allowed a.t.m. |
|
|
723 | |
|
|
724 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
725 | ++dec->cur; |
|
|
726 | } |
|
|
727 | |
|
|
728 | INLINE void |
687 | decode_ws (dec_t *dec) |
729 | decode_ws (dec_t *dec) |
688 | { |
730 | { |
689 | for (;;) |
731 | for (;;) |
690 | { |
732 | { |
691 | char ch = *dec->cur; |
733 | char ch = *dec->cur; |
692 | |
734 | |
693 | if (ch > 0x20 |
735 | if (ch > 0x20) |
|
|
736 | { |
|
|
737 | if (expect_false (ch == '#')) |
|
|
738 | { |
|
|
739 | if (dec->json.flags & F_RELAXED) |
|
|
740 | decode_comment (dec); |
|
|
741 | else |
|
|
742 | break; |
|
|
743 | } |
|
|
744 | else |
|
|
745 | break; |
|
|
746 | } |
694 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
747 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
695 | break; |
748 | break; // parse error, but let higher level handle it, gives better error messages |
696 | |
749 | |
697 | ++dec->cur; |
750 | ++dec->cur; |
698 | } |
751 | } |
699 | } |
752 | } |
700 | |
753 | |
… | |
… | |
951 | |
1004 | |
952 | if (!is_nv) |
1005 | if (!is_nv) |
953 | { |
1006 | { |
954 | int len = dec->cur - start; |
1007 | int len = dec->cur - start; |
955 | |
1008 | |
956 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
1009 | // special case the rather common 1..5-digit-int case |
957 | if (*start == '-') |
1010 | if (*start == '-') |
958 | switch (len) |
1011 | switch (len) |
959 | { |
1012 | { |
960 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1013 | case 2: return newSViv (-( start [1] - '0' * 1)); |
961 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1014 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
962 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1015 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
963 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1016 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
|
|
1017 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
964 | } |
1018 | } |
965 | else |
1019 | else |
966 | switch (len) |
1020 | switch (len) |
967 | { |
1021 | { |
968 | case 1: return newSViv ( start [0] - '0' * 1); |
1022 | case 1: return newSViv ( start [0] - '0' * 1); |
969 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1023 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
970 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1024 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
971 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1025 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1026 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
972 | } |
1027 | } |
973 | |
1028 | |
974 | { |
1029 | { |
975 | UV uv; |
1030 | UV uv; |
976 | int numtype = grok_number (start, len, &uv); |
1031 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1037 | |
1092 | |
1038 | if (*dec->cur != ',') |
1093 | if (*dec->cur != ',') |
1039 | ERR (", or ] expected while parsing array"); |
1094 | ERR (", or ] expected while parsing array"); |
1040 | |
1095 | |
1041 | ++dec->cur; |
1096 | ++dec->cur; |
|
|
1097 | |
|
|
1098 | decode_ws (dec); |
|
|
1099 | |
|
|
1100 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1101 | { |
|
|
1102 | ++dec->cur; |
|
|
1103 | break; |
|
|
1104 | } |
1042 | } |
1105 | } |
1043 | |
1106 | |
1044 | DEC_DEC_DEPTH; |
1107 | DEC_DEC_DEPTH; |
1045 | return newRV_noinc ((SV *)av); |
1108 | return newRV_noinc ((SV *)av); |
1046 | |
1109 | |
… | |
… | |
1062 | if (*dec->cur == '}') |
1125 | if (*dec->cur == '}') |
1063 | ++dec->cur; |
1126 | ++dec->cur; |
1064 | else |
1127 | else |
1065 | for (;;) |
1128 | for (;;) |
1066 | { |
1129 | { |
1067 | decode_ws (dec); EXPECT_CH ('"'); |
1130 | EXPECT_CH ('"'); |
1068 | |
1131 | |
1069 | // heuristic: assume that |
1132 | // heuristic: assume that |
1070 | // a) decode_str + hv_store_ent are abysmally slow. |
1133 | // a) decode_str + hv_store_ent are abysmally slow. |
1071 | // b) most hash keys are short, simple ascii text. |
1134 | // b) most hash keys are short, simple ascii text. |
1072 | // => try to "fast-match" such strings to avoid |
1135 | // => try to "fast-match" such strings to avoid |
… | |
… | |
1086 | if (!key) |
1149 | if (!key) |
1087 | goto fail; |
1150 | goto fail; |
1088 | |
1151 | |
1089 | decode_ws (dec); EXPECT_CH (':'); |
1152 | decode_ws (dec); EXPECT_CH (':'); |
1090 | |
1153 | |
|
|
1154 | decode_ws (dec); |
1091 | value = decode_sv (dec); |
1155 | value = decode_sv (dec); |
1092 | if (!value) |
1156 | if (!value) |
1093 | { |
1157 | { |
1094 | SvREFCNT_dec (key); |
1158 | SvREFCNT_dec (key); |
1095 | goto fail; |
1159 | goto fail; |
… | |
… | |
1107 | int len = p - key; |
1171 | int len = p - key; |
1108 | dec->cur = p + 1; |
1172 | dec->cur = p + 1; |
1109 | |
1173 | |
1110 | decode_ws (dec); EXPECT_CH (':'); |
1174 | decode_ws (dec); EXPECT_CH (':'); |
1111 | |
1175 | |
|
|
1176 | decode_ws (dec); |
1112 | value = decode_sv (dec); |
1177 | value = decode_sv (dec); |
1113 | if (!value) |
1178 | if (!value) |
1114 | goto fail; |
1179 | goto fail; |
1115 | |
1180 | |
1116 | hv_store (hv, key, len, value, 0); |
1181 | hv_store (hv, key, len, value, 0); |
… | |
… | |
1132 | |
1197 | |
1133 | if (*dec->cur != ',') |
1198 | if (*dec->cur != ',') |
1134 | ERR (", or } expected while parsing object/hash"); |
1199 | ERR (", or } expected while parsing object/hash"); |
1135 | |
1200 | |
1136 | ++dec->cur; |
1201 | ++dec->cur; |
|
|
1202 | |
|
|
1203 | decode_ws (dec); |
|
|
1204 | |
|
|
1205 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1206 | { |
|
|
1207 | ++dec->cur; |
|
|
1208 | break; |
|
|
1209 | } |
1137 | } |
1210 | } |
1138 | |
1211 | |
1139 | DEC_DEC_DEPTH; |
1212 | DEC_DEC_DEPTH; |
1140 | sv = newRV_noinc ((SV *)hv); |
1213 | sv = newRV_noinc ((SV *)hv); |
1141 | |
1214 | |
… | |
… | |
1206 | } |
1279 | } |
1207 | |
1280 | |
1208 | static SV * |
1281 | static SV * |
1209 | decode_sv (dec_t *dec) |
1282 | decode_sv (dec_t *dec) |
1210 | { |
1283 | { |
1211 | decode_ws (dec); |
|
|
1212 | |
|
|
1213 | // the beauty of JSON: you need exactly one character lookahead |
1284 | // the beauty of JSON: you need exactly one character lookahead |
1214 | // to parse anything. |
1285 | // to parse anything. |
1215 | switch (*dec->cur) |
1286 | switch (*dec->cur) |
1216 | { |
1287 | { |
1217 | case '"': ++dec->cur; return decode_str (dec); |
1288 | case '"': ++dec->cur; return decode_str (dec); |
… | |
… | |
1225 | |
1296 | |
1226 | case 't': |
1297 | case 't': |
1227 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1298 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1228 | { |
1299 | { |
1229 | dec->cur += 4; |
1300 | dec->cur += 4; |
|
|
1301 | #if JSON_SLOW |
|
|
1302 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
|
|
1303 | #endif |
1230 | return SvREFCNT_inc (json_true); |
1304 | return SvREFCNT_inc (json_true); |
1231 | } |
1305 | } |
1232 | else |
1306 | else |
1233 | ERR ("'true' expected"); |
1307 | ERR ("'true' expected"); |
1234 | |
1308 | |
… | |
… | |
1236 | |
1310 | |
1237 | case 'f': |
1311 | case 'f': |
1238 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1312 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1239 | { |
1313 | { |
1240 | dec->cur += 5; |
1314 | dec->cur += 5; |
|
|
1315 | #if JSON_SLOW |
|
|
1316 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
|
|
1317 | #endif |
1241 | return SvREFCNT_inc (json_false); |
1318 | return SvREFCNT_inc (json_false); |
1242 | } |
1319 | } |
1243 | else |
1320 | else |
1244 | ERR ("'false' expected"); |
1321 | ERR ("'false' expected"); |
1245 | |
1322 | |
… | |
… | |
1295 | |
1372 | |
1296 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1373 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1297 | dec.json.flags |= F_HOOK; |
1374 | dec.json.flags |= F_HOOK; |
1298 | |
1375 | |
1299 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1376 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1377 | |
|
|
1378 | decode_ws (&dec); |
1300 | sv = decode_sv (&dec); |
1379 | sv = decode_sv (&dec); |
1301 | |
1380 | |
1302 | if (!(offset_return || !sv)) |
1381 | if (!(offset_return || !sv)) |
1303 | { |
1382 | { |
1304 | // check for trailing garbage |
1383 | // check for trailing garbage |
… | |
… | |
1384 | { |
1463 | { |
1385 | SV *pv = NEWSV (0, sizeof (JSON)); |
1464 | SV *pv = NEWSV (0, sizeof (JSON)); |
1386 | SvPOK_only (pv); |
1465 | SvPOK_only (pv); |
1387 | Zero (SvPVX (pv), 1, JSON); |
1466 | Zero (SvPVX (pv), 1, JSON); |
1388 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1467 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1389 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); |
1468 | XPUSHs (sv_2mortal (sv_bless ( |
|
|
1469 | newRV_noinc (pv), |
|
|
1470 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
|
|
1471 | ))); |
1390 | } |
1472 | } |
1391 | |
1473 | |
1392 | void ascii (JSON *self, int enable = 1) |
1474 | void ascii (JSON *self, int enable = 1) |
1393 | ALIAS: |
1475 | ALIAS: |
1394 | ascii = F_ASCII |
1476 | ascii = F_ASCII |
… | |
… | |
1401 | pretty = F_PRETTY |
1483 | pretty = F_PRETTY |
1402 | allow_nonref = F_ALLOW_NONREF |
1484 | allow_nonref = F_ALLOW_NONREF |
1403 | shrink = F_SHRINK |
1485 | shrink = F_SHRINK |
1404 | allow_blessed = F_ALLOW_BLESSED |
1486 | allow_blessed = F_ALLOW_BLESSED |
1405 | convert_blessed = F_CONV_BLESSED |
1487 | convert_blessed = F_CONV_BLESSED |
|
|
1488 | relaxed = F_RELAXED |
1406 | PPCODE: |
1489 | PPCODE: |
1407 | { |
1490 | { |
1408 | if (enable) |
1491 | if (enable) |
1409 | self->flags |= ix; |
1492 | self->flags |= ix; |
1410 | else |
1493 | else |
1411 | self->flags &= ~ix; |
1494 | self->flags &= ~ix; |
1412 | |
1495 | |
1413 | XPUSHs (ST (0)); |
1496 | XPUSHs (ST (0)); |
1414 | } |
1497 | } |
1415 | |
1498 | |
|
|
1499 | void get_ascii (JSON *self) |
|
|
1500 | ALIAS: |
|
|
1501 | get_ascii = F_ASCII |
|
|
1502 | get_latin1 = F_LATIN1 |
|
|
1503 | get_utf8 = F_UTF8 |
|
|
1504 | get_indent = F_INDENT |
|
|
1505 | get_canonical = F_CANONICAL |
|
|
1506 | get_space_before = F_SPACE_BEFORE |
|
|
1507 | get_space_after = F_SPACE_AFTER |
|
|
1508 | get_allow_nonref = F_ALLOW_NONREF |
|
|
1509 | get_shrink = F_SHRINK |
|
|
1510 | get_allow_blessed = F_ALLOW_BLESSED |
|
|
1511 | get_convert_blessed = F_CONV_BLESSED |
|
|
1512 | get_relaxed = F_RELAXED |
|
|
1513 | PPCODE: |
|
|
1514 | XPUSHs (boolSV (self->flags & ix)); |
|
|
1515 | |
1416 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1516 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1417 | PPCODE: |
1517 | PPCODE: |
1418 | { |
1518 | { |
1419 | UV log2 = 0; |
1519 | UV log2 = 0; |
1420 | |
1520 | |
… | |
… | |
1426 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1526 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1427 | |
1527 | |
1428 | XPUSHs (ST (0)); |
1528 | XPUSHs (ST (0)); |
1429 | } |
1529 | } |
1430 | |
1530 | |
|
|
1531 | U32 get_max_depth (JSON *self) |
|
|
1532 | CODE: |
|
|
1533 | RETVAL = DEC_DEPTH (self->flags); |
|
|
1534 | OUTPUT: |
|
|
1535 | RETVAL |
|
|
1536 | |
1431 | void max_size (JSON *self, UV max_size = 0) |
1537 | void max_size (JSON *self, UV max_size = 0) |
1432 | PPCODE: |
1538 | PPCODE: |
1433 | { |
1539 | { |
1434 | UV log2 = 0; |
1540 | UV log2 = 0; |
1435 | |
1541 | |
… | |
… | |
1441 | |
1547 | |
1442 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
1548 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
1443 | |
1549 | |
1444 | XPUSHs (ST (0)); |
1550 | XPUSHs (ST (0)); |
1445 | } |
1551 | } |
|
|
1552 | |
|
|
1553 | int get_max_size (JSON *self) |
|
|
1554 | CODE: |
|
|
1555 | RETVAL = DEC_SIZE (self->flags); |
|
|
1556 | OUTPUT: |
|
|
1557 | RETVAL |
1446 | |
1558 | |
1447 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1559 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1448 | PPCODE: |
1560 | PPCODE: |
1449 | { |
1561 | { |
1450 | SvREFCNT_dec (self->cb_object); |
1562 | SvREFCNT_dec (self->cb_object); |
… | |
… | |
1497 | SvREFCNT_dec (self->cb_sk_object); |
1609 | SvREFCNT_dec (self->cb_sk_object); |
1498 | SvREFCNT_dec (self->cb_object); |
1610 | SvREFCNT_dec (self->cb_object); |
1499 | |
1611 | |
1500 | PROTOTYPES: ENABLE |
1612 | PROTOTYPES: ENABLE |
1501 | |
1613 | |
1502 | void to_json (SV *scalar) |
1614 | void encode_json (SV *scalar) |
1503 | PPCODE: |
1615 | PPCODE: |
1504 | { |
1616 | { |
1505 | JSON json = { F_DEFAULT | F_UTF8 }; |
1617 | JSON json = { F_DEFAULT | F_UTF8 }; |
1506 | XPUSHs (encode_json (scalar, &json)); |
1618 | XPUSHs (encode_json (scalar, &json)); |
1507 | } |
1619 | } |
1508 | |
1620 | |
1509 | void from_json (SV *jsonstr) |
1621 | void decode_json (SV *jsonstr) |
1510 | PPCODE: |
1622 | PPCODE: |
1511 | { |
1623 | { |
1512 | JSON json = { F_DEFAULT | F_UTF8 }; |
1624 | JSON json = { F_DEFAULT | F_UTF8 }; |
1513 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1625 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1514 | } |
1626 | } |