vndb/fu
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

jsonparse: A bunch of performance improvements

Browse source 
Turrns out JSON::XS had some pretty good ideas that I could borrow.
This commit is contained in:
Yorhel 2025-01-31 16:35:47 +01:00
parent ca8d1b72be
commit ebe84167e7
5 changed files with 144 additions and 111 deletions
Download patch file Download diff file Expand all files Collapse all files

68
FU/Benchmarks.pod
View file

@ -55,7 +55,7 @@ API object from L<JSON::XS> documentation.
Cpanel::JSON::XS 112845/s 102903/s 101843/s Cpanel::JSON::XS 112845/s 102903/s 101843/s
JSON::SIMD 125127/s 116569/s 117446/s JSON::SIMD 125127/s 116569/s 117446/s
JSON::XS 128792/s 118616/s 117257/s JSON::XS 128792/s 118616/s 117257/s
FU::Util 126324/s 106597/s 62633/s FU::Util 136151/s 115564/s 120184/s
Object (small) Object (small)
@ -64,7 +64,7 @@ Object (small)
Cpanel::JSON::XS 42755/s 27849/s 18920/s Cpanel::JSON::XS 42755/s 27849/s 18920/s
JSON::SIMD 48487/s 30266/s 22527/s JSON::SIMD 48487/s 30266/s 22527/s
JSON::XS 48980/s 30558/s 24065/s JSON::XS 48980/s 30558/s 24065/s
FU::Util 43172/s 24882/s 6818/s FU::Util 49346/s 26316/s 20768/s
Object (large) Object (large)
@ -73,7 +73,7 @@ Object (large)
Cpanel::JSON::XS 29889/s 11629/s 16264/s Cpanel::JSON::XS 29889/s 11629/s 16264/s
JSON::SIMD 33950/s 12050/s 22649/s JSON::SIMD 33950/s 12050/s 22649/s
JSON::XS 34534/s 12399/s 23418/s JSON::XS 34534/s 12399/s 23418/s
FU::Util 37298/s 12001/s 5413/s FU::Util 40482/s 13735/s 18161/s
Object (large, mixed unicode) Object (large, mixed unicode)
@ -82,7 +82,7 @@ Object (large, mixed unicode)
Cpanel::JSON::XS 20563/s 1402/s 7590/s Cpanel::JSON::XS 20563/s 1402/s 7590/s
JSON::SIMD 24996/s 1401/s 15959/s JSON::SIMD 24996/s 1401/s 15959/s
JSON::XS 26915/s 1454/s 8848/s JSON::XS 26915/s 1454/s 8848/s
FU::Util 24280/s 10461/s 5998/s FU::Util 24706/s 11403/s 9379/s
Small integers Small integers
@ -91,7 +91,7 @@ Small integers
Cpanel::JSON::XS 7235/s 6008/s Cpanel::JSON::XS 7235/s 6008/s
JSON::SIMD 8058/s 4335/s JSON::SIMD 8058/s 4335/s
JSON::XS 8036/s 5998/s JSON::XS 8036/s 5998/s
FU::Util 8344/s 5656/s FU::Util 8701/s 5902/s
Large integers Large integers
@ -100,7 +100,7 @@ Large integers
Cpanel::JSON::XS 31337/s 48514/s Cpanel::JSON::XS 31337/s 48514/s
JSON::SIMD 36876/s 52446/s JSON::SIMD 36876/s 52446/s
JSON::XS 36665/s 48689/s JSON::XS 36665/s 48689/s
FU::Util 115559/s 62931/s FU::Util 122700/s 62048/s
ASCII strings ASCII strings
@ -109,7 +109,7 @@ ASCII strings
Cpanel::JSON::XS 117604/s 43652/s Cpanel::JSON::XS 117604/s 43652/s
JSON::SIMD 136214/s 50865/s JSON::SIMD 136214/s 50865/s
JSON::XS 136486/s 40599/s JSON::XS 136486/s 40599/s
FU::Util 168713/s 32742/s FU::Util 192974/s 43327/s
Unicode strings Unicode strings
@ -118,7 +118,7 @@ Unicode strings
Cpanel::JSON::XS 92871/s 66032/s Cpanel::JSON::XS 92871/s 66032/s
JSON::SIMD 106664/s 100555/s JSON::SIMD 106664/s 100555/s
JSON::XS 104716/s 60538/s JSON::XS 104716/s 60538/s
FU::Util 108032/s 36331/s FU::Util 124756/s 49825/s
String escaping (few) String escaping (few)
@ -127,7 +127,7 @@ String escaping (few)
Cpanel::JSON::XS 133222/s 104174/s Cpanel::JSON::XS 133222/s 104174/s
JSON::SIMD 155205/s 148939/s JSON::SIMD 155205/s 148939/s
JSON::XS 123154/s 92594/s JSON::XS 123154/s 92594/s
FU::Util 206172/s 76491/s FU::Util 202792/s 85352/s
String escaping (many) String escaping (many)
@ -136,7 +136,7 @@ String escaping (many)
Cpanel::JSON::XS 139049/s 99260/s Cpanel::JSON::XS 139049/s 99260/s
JSON::SIMD 152317/s 113037/s JSON::SIMD 152317/s 113037/s
JSON::XS 153421/s 107918/s JSON::XS 153421/s 107918/s
FU::Util 132774/s 64880/s FU::Util 132523/s 78199/s
@ -144,122 +144,122 @@ String escaping (many)
# Cached data used by bench.PL. Same as the formatted tables above but easier to parse. # Cached data used by bench.PL. Same as the formatted tables above but easier to parse.
json/api Canonical Cpanel::JSON::XS 102903 json/api Canonical Cpanel::JSON::XS 102903
json/api Canonical FU::Util 106597 json/api Canonical FU::Util 115564
json/api Canonical JSON::PP 4973 json/api Canonical JSON::PP 4973
json/api Canonical JSON::SIMD 116569 json/api Canonical JSON::SIMD 116569
json/api Canonical JSON::XS 118616 json/api Canonical JSON::XS 118616
json/api Decode Cpanel::JSON::XS 101843 json/api Decode Cpanel::JSON::XS 101843
json/api Decode FU::Util 62633 json/api Decode FU::Util 120184
json/api Decode JSON::PP 1290 json/api Decode JSON::PP 1290
json/api Decode JSON::SIMD 117446 json/api Decode JSON::SIMD 117446
json/api Decode JSON::XS 117257 json/api Decode JSON::XS 117257
json/api Encode Cpanel::JSON::XS 112845 json/api Encode Cpanel::JSON::XS 112845
json/api Encode FU::Util 126324 json/api Encode FU::Util 136151
json/api Encode JSON::PP 5397 json/api Encode JSON::PP 5397
json/api Encode JSON::SIMD 125127 json/api Encode JSON::SIMD 125127
json/api Encode JSON::XS 128792 json/api Encode JSON::XS 128792
json/intl Decode Cpanel::JSON::XS 48514 json/intl Decode Cpanel::JSON::XS 48514
json/intl Decode FU::Util 62931 json/intl Decode FU::Util 62048
json/intl Decode JSON::PP 330 json/intl Decode JSON::PP 330
json/intl Decode JSON::SIMD 52446 json/intl Decode JSON::SIMD 52446
json/intl Decode JSON::XS 48689 json/intl Decode JSON::XS 48689
json/intl Encode Cpanel::JSON::XS 31337 json/intl Encode Cpanel::JSON::XS 31337
json/intl Encode FU::Util 115559 json/intl Encode FU::Util 122700
json/intl Encode JSON::PP 2174 json/intl Encode JSON::PP 2174
json/intl Encode JSON::SIMD 36876 json/intl Encode JSON::SIMD 36876
json/intl Encode JSON::XS 36665 json/intl Encode JSON::XS 36665
json/ints Decode Cpanel::JSON::XS 6008 json/ints Decode Cpanel::JSON::XS 6008
json/ints Decode FU::Util 5656 json/ints Decode FU::Util 5902
json/ints Decode JSON::PP 29 json/ints Decode JSON::PP 29
json/ints Decode JSON::SIMD 4335 json/ints Decode JSON::SIMD 4335
json/ints Decode JSON::XS 5998 json/ints Decode JSON::XS 5998
json/ints Encode Cpanel::JSON::XS 7235 json/ints Encode Cpanel::JSON::XS 7235
json/ints Encode FU::Util 8344 json/ints Encode FU::Util 8701
json/ints Encode JSON::PP 113 json/ints Encode JSON::PP 113
json/ints Encode JSON::SIMD 8058 json/ints Encode JSON::SIMD 8058
json/ints Encode JSON::XS 8036 json/ints Encode JSON::XS 8036
json/objl Canonical Cpanel::JSON::XS 11629 json/objl Canonical Cpanel::JSON::XS 11629
json/objl Canonical FU::Util 12001 json/objl Canonical FU::Util 13735
json/objl Canonical JSON::PP 744 json/objl Canonical JSON::PP 744
json/objl Canonical JSON::SIMD 12050 json/objl Canonical JSON::SIMD 12050
json/objl Canonical JSON::XS 12399 json/objl Canonical JSON::XS 12399
json/objl Decode Cpanel::JSON::XS 16264 json/objl Decode Cpanel::JSON::XS 16264
json/objl Decode FU::Util 5413 json/objl Decode FU::Util 18161
json/objl Decode JSON::PP 105 json/objl Decode JSON::PP 105
json/objl Decode JSON::SIMD 22649 json/objl Decode JSON::SIMD 22649
json/objl Decode JSON::XS 23418 json/objl Decode JSON::XS 23418
json/objl Encode Cpanel::JSON::XS 29889 json/objl Encode Cpanel::JSON::XS 29889
json/objl Encode FU::Util 37298 json/objl Encode FU::Util 40482
json/objl Encode JSON::PP 916 json/objl Encode JSON::PP 916
json/objl Encode JSON::SIMD 33950 json/objl Encode JSON::SIMD 33950
json/objl Encode JSON::XS 34534 json/objl Encode JSON::XS 34534
json/objs Canonical Cpanel::JSON::XS 27849 json/objs Canonical Cpanel::JSON::XS 27849
json/objs Canonical FU::Util 24882 json/objs Canonical FU::Util 26316
json/objs Canonical JSON::PP 827 json/objs Canonical JSON::PP 827
json/objs Canonical JSON::SIMD 30266 json/objs Canonical JSON::SIMD 30266
json/objs Canonical JSON::XS 30558 json/objs Canonical JSON::XS 30558
json/objs Decode Cpanel::JSON::XS 18920 json/objs Decode Cpanel::JSON::XS 18920
json/objs Decode FU::Util 6818 json/objs Decode FU::Util 20768
json/objs Decode JSON::PP 206 json/objs Decode JSON::PP 206
json/objs Decode JSON::SIMD 22527 json/objs Decode JSON::SIMD 22527
json/objs Decode JSON::XS 24065 json/objs Decode JSON::XS 24065
json/objs Encode Cpanel::JSON::XS 42755 json/objs Encode Cpanel::JSON::XS 42755
json/objs Encode FU::Util 43172 json/objs Encode FU::Util 49346
json/objs Encode JSON::PP 905 json/objs Encode JSON::PP 905
json/objs Encode JSON::SIMD 48487 json/objs Encode JSON::SIMD 48487
json/objs Encode JSON::XS 48980 json/objs Encode JSON::XS 48980
json/obju Canonical Cpanel::JSON::XS 1402 json/obju Canonical Cpanel::JSON::XS 1402
json/obju Canonical FU::Util 10461 json/obju Canonical FU::Util 11403
json/obju Canonical JSON::PP 699 json/obju Canonical JSON::PP 699
json/obju Canonical JSON::SIMD 1401 json/obju Canonical JSON::SIMD 1401
json/obju Canonical JSON::XS 1454 json/obju Canonical JSON::XS 1454
json/obju Decode Cpanel::JSON::XS 7590 json/obju Decode Cpanel::JSON::XS 7590
json/obju Decode FU::Util 5998 json/obju Decode FU::Util 9379
json/obju Decode JSON::PP 86 json/obju Decode JSON::PP 86
json/obju Decode JSON::SIMD 15959 json/obju Decode JSON::SIMD 15959
json/obju Decode JSON::XS 8848 json/obju Decode JSON::XS 8848
json/obju Encode Cpanel::JSON::XS 20563 json/obju Encode Cpanel::JSON::XS 20563
json/obju Encode FU::Util 24280 json/obju Encode FU::Util 24706
json/obju Encode JSON::PP 845 json/obju Encode JSON::PP 845
json/obju Encode JSON::SIMD 24996 json/obju Encode JSON::SIMD 24996
json/obju Encode JSON::XS 26915 json/obju Encode JSON::XS 26915
json/strel Decode Cpanel::JSON::XS 99260 json/strel Decode Cpanel::JSON::XS 99260
json/strel Decode FU::Util 64880 json/strel Decode FU::Util 78199
json/strel Decode JSON::PP 360 json/strel Decode JSON::PP 360
json/strel Decode JSON::SIMD 113037 json/strel Decode JSON::SIMD 113037
json/strel Decode JSON::XS 107918 json/strel Decode JSON::XS 107918
json/strel Encode Cpanel::JSON::XS 139049 json/strel Encode Cpanel::JSON::XS 139049
json/strel Encode FU::Util 132774 json/strel Encode FU::Util 132523
json/strel Encode JSON::PP 2230 json/strel Encode JSON::PP 2230
json/strel Encode JSON::SIMD 152317 json/strel Encode JSON::SIMD 152317
json/strel Encode JSON::XS 153421 json/strel Encode JSON::XS 153421
json/stres Decode Cpanel::JSON::XS 104174 json/stres Decode Cpanel::JSON::XS 104174
json/stres Decode FU::Util 76491 json/stres Decode FU::Util 85352
json/stres Decode JSON::PP 350 json/stres Decode JSON::PP 350
json/stres Decode JSON::SIMD 148939 json/stres Decode JSON::SIMD 148939
json/stres Decode JSON::XS 92594 json/stres Decode JSON::XS 92594
json/stres Encode Cpanel::JSON::XS 133222 json/stres Encode Cpanel::JSON::XS 133222
json/stres Encode FU::Util 206172 json/stres Encode FU::Util 202792
json/stres Encode JSON::PP 4258 json/stres Encode JSON::PP 4258
json/stres Encode JSON::SIMD 155205 json/stres Encode JSON::SIMD 155205
json/stres Encode JSON::XS 123154 json/stres Encode JSON::XS 123154
json/strs Decode Cpanel::JSON::XS 43652 json/strs Decode Cpanel::JSON::XS 43652
json/strs Decode FU::Util 32742 json/strs Decode FU::Util 43327
json/strs Decode JSON::PP 341 json/strs Decode JSON::PP 341
json/strs Decode JSON::SIMD 50865 json/strs Decode JSON::SIMD 50865
json/strs Decode JSON::XS 40599 json/strs Decode JSON::XS 40599
json/strs Encode Cpanel::JSON::XS 117604 json/strs Encode Cpanel::JSON::XS 117604
json/strs Encode FU::Util 168713 json/strs Encode FU::Util 192974
json/strs Encode JSON::PP 2922 json/strs Encode JSON::PP 2922
json/strs Encode JSON::SIMD 136214 json/strs Encode JSON::SIMD 136214
json/strs Encode JSON::XS 136486 json/strs Encode JSON::XS 136486
json/stru Decode Cpanel::JSON::XS 66032 json/stru Decode Cpanel::JSON::XS 66032
json/stru Decode FU::Util 36331 json/stru Decode FU::Util 49825
json/stru Decode JSON::PP 254 json/stru Decode JSON::PP 254
json/stru Decode JSON::SIMD 100555 json/stru Decode JSON::SIMD 100555
json/stru Decode JSON::XS 60538 json/stru Decode JSON::XS 60538
json/stru Encode Cpanel::JSON::XS 92871 json/stru Encode Cpanel::JSON::XS 92871
json/stru Encode FU::Util 108032 json/stru Encode FU::Util 124756
json/stru Encode JSON::PP 5144 json/stru Encode JSON::PP 5144
json/stru Encode JSON::SIMD 106664 json/stru Encode JSON::SIMD 106664
json/stru Encode JSON::XS 104716 json/stru Encode JSON::XS 104716

51
c/common.c
View file

@ -2,29 +2,47 @@
typedef struct { typedef struct {
SV *sv; SV *sv;
SV *mortal;
char *cur; char *cur;
char *end; char *end;
size_t maxlen; size_t maxlen;
int setutf8;
char sbuf[4096];
} fustr; } fustr;
/* sv must be a new SV with a preallocated buffer */ static void fustr_init_(pTHX_ fustr *s, SV *mortal, size_t maxlen) {
static void fustr_init_(pTHX_ fustr *s, SV *sv, size_t maxlen) { s->sv = NULL;
s->sv = sv; s->cur = s->sbuf;
SvPOK_only(s->sv); s->end = s->sbuf + (maxlen > sizeof s->sbuf ? sizeof s->sbuf : maxlen);
s->cur = SvPVX(s->sv);
s->end = SvEND(s->sv);
s->maxlen = maxlen; s->maxlen = maxlen;
s->mortal = mortal;
s->setutf8 = 0;
} }
#define fustr_start(s) (((s)->sv ? SvPVX((s)->sv) : (s)->sbuf))
static void fustr_grow(pTHX_ fustr *s, size_t add) { static void fustr_grow(pTHX_ fustr *s, size_t add) {
size_t off = s->cur - SvPVX(s->sv); size_t off = s->cur - (s->sv ? SvPVX(s->sv) : s->sbuf);
size_t newlen = 64; size_t newlen = sizeof s->sbuf;
char *buf;
add += off; add += off;
if (add > s->maxlen) croak("maximum string length exceeded"); if (add > s->maxlen) croak("maximum string length exceeded");
/* Increment to next power of two; SvGROW's default strategy is slow */ /* Increment to next power of two; SvGROW's default strategy is slow */
while (newlen < add) newlen <<= 1; while (newlen < add) newlen <<= 1;
if (newlen > s->maxlen) newlen = s->maxlen; if (newlen > s->maxlen) newlen = s->maxlen;
char *buf = SvGROW(s->sv, newlen); if (s->sv) {
buf = SvGROW(s->sv, newlen);
} else {
if (s->mortal) {
s->sv = s->mortal;
sv_setpv_bufsize(s->sv, off, newlen);
} else {
s->sv = newSV(newlen);
}
SvPOK_only(s->sv);
buf = SvPVX(s->sv);
memcpy(buf, s->sbuf, off);
}
s->cur = buf + off; s->cur = buf + off;
s->end = buf + (SvLEN(s->sv) > s->maxlen ? s->maxlen : SvLEN(s->sv)); s->end = buf + (SvLEN(s->sv) > s->maxlen ? s->maxlen : SvLEN(s->sv));
} }
@ -39,6 +57,11 @@ static inline void fustr_write_(pTHX_ fustr *s, const char *str, size_t n) {
s->cur += n; s->cur += n;
} }
static inline void fustr_write_ch_(pTHX_ fustr *s, char x) {
fustr_reserve_(aTHX_ s, 1);
*(s->cur++) = x;
}
/* Adds n uninitialized bytes to the string and returns a buffer to write the data to */ /* Adds n uninitialized bytes to the string and returns a buffer to write the data to */
static inline char *fustr_write_buf_(pTHX_ fustr *s, size_t n) { static inline char *fustr_write_buf_(pTHX_ fustr *s, size_t n) {
fustr_reserve_(aTHX_ s, n); fustr_reserve_(aTHX_ s, n);
@ -50,13 +73,19 @@ static inline char *fustr_write_buf_(pTHX_ fustr *s, size_t n) {
static SV *fustr_done_(pTHX_ fustr *s) { static SV *fustr_done_(pTHX_ fustr *s) {
fustr_reserve_(aTHX_ s, 1); fustr_reserve_(aTHX_ s, 1);
*s->cur = 0; *s->cur = 0;
SvCUR_set(s->sv, s->cur - SvPVX(s->sv)); if (s->sv) {
// TODO: SvPV_shrink_to_cur? SvCUR_set(s->sv, s->cur - SvPVX(s->sv));
// TODO: SvPV_shrink_to_cur?
} else {
s->sv = newSVpvn_flags(s->sbuf, s->cur - s->sbuf, s->mortal ? SVs_TEMP : 0);
}
if (s->setutf8) SvUTF8_on(s->sv);
return s->sv; return s->sv;
} }
#define fustr_init(a,b,c) fustr_init_(aTHX_ a,b,c) #define fustr_init(a,b,c) fustr_init_(aTHX_ a,b,c)
#define fustr_reserve(a,b) fustr_reserve_(aTHX_ a,b) #define fustr_reserve(a,b) fustr_reserve_(aTHX_ a,b)
#define fustr_write(a,b,c) fustr_write_(aTHX_ a,b,c) #define fustr_write(a,b,c) fustr_write_(aTHX_ a,b,c)
#define fustr_write_ch(a,b) fustr_write_ch_(aTHX_ a,b)
#define fustr_write_buf(a,b) fustr_write_buf_(aTHX_ a,b) #define fustr_write_buf(a,b) fustr_write_buf_(aTHX_ a,b)
#define fustr_done(a) fustr_done_(aTHX_ a) #define fustr_done(a) fustr_done_(aTHX_ a)

24
c/jsonfmt.c
View file

@ -30,7 +30,7 @@ static void fujson_fmt_str(pTHX_ fujson_fmt_ctx *ctx, const char *stri, size_t l
croak("invalid codepoint encountered in string, cannot format to JSON"); croak("invalid codepoint encountered in string, cannot format to JSON");
} }
fustr_write(&ctx->out, "\"", 1); fustr_write_ch(&ctx->out, '\"');
fustr_reserve(&ctx->out, len); fustr_reserve(&ctx->out, len);
while (off < len) { while (off < len) {
@ -78,7 +78,7 @@ static void fujson_fmt_str(pTHX_ fujson_fmt_ctx *ctx, const char *stri, size_t l
} }
} }
fustr_write(&ctx->out, "\"", 1); fustr_write_ch(&ctx->out, '\"');
} }
static const char fujson_digits[] = static const char fujson_digits[] =
@ -109,7 +109,7 @@ static void fujson_fmt_int(pTHX_ fujson_fmt_ctx *ctx, SV *val) {
} }
if (uv == 0) { if (uv == 0) {
fustr_write(&ctx->out, "0", 1); fustr_write_ch(&ctx->out, '0');
return; return;
} }
@ -125,10 +125,10 @@ static void fujson_fmt_int(pTHX_ fujson_fmt_ctx *ctx, SV *val) {
static void fujson_fmt_av(pTHX_ fujson_fmt_ctx *ctx, AV *av) { static void fujson_fmt_av(pTHX_ fujson_fmt_ctx *ctx, AV *av) {
int i, len = av_count(av); int i, len = av_count(av);
fustr_write(&ctx->out, "[", 1); fustr_write_ch(&ctx->out, '[');
ctx->pretty++; ctx->pretty++;
for (i=0; i<len; i++) { for (i=0; i<len; i++) {
if (i) fustr_write(&ctx->out, ",", 1); if (i) fustr_write_ch(&ctx->out, ',');
fujson_fmt_indent(aTHX_ ctx); fujson_fmt_indent(aTHX_ ctx);
SV **sv = av_fetch(av, i, 0); SV **sv = av_fetch(av, i, 0);
if (sv) fujson_fmt(aTHX_ ctx, *sv); /* sv will have magic if av is tied, but fujson_fmt() handles that. */ if (sv) fujson_fmt(aTHX_ ctx, *sv); /* sv will have magic if av is tied, but fujson_fmt() handles that. */
@ -136,7 +136,7 @@ static void fujson_fmt_av(pTHX_ fujson_fmt_ctx *ctx, AV *av) {
} }
ctx->pretty--; ctx->pretty--;
if (i) fujson_fmt_indent(aTHX_ ctx); if (i) fujson_fmt_indent(aTHX_ ctx);
fustr_write(&ctx->out, "]", 1); fustr_write_ch(&ctx->out, ']');
} }
static int fujson_fmt_hvcmp(const void *pa, const void *pb) { static int fujson_fmt_hvcmp(const void *pa, const void *pb) {
@ -159,12 +159,12 @@ static int fujson_fmt_hvcmp(const void *pa, const void *pb) {
static void fujson_fmt_hvkv(pTHX_ fujson_fmt_ctx *ctx, HV *hv, HE *he, char **hestr) { static void fujson_fmt_hvkv(pTHX_ fujson_fmt_ctx *ctx, HV *hv, HE *he, char **hestr) {
STRLEN helen; STRLEN helen;
if (*hestr) fustr_write(&ctx->out, ",", 1); if (*hestr) fustr_write_ch(&ctx->out, ',');
fujson_fmt_indent(aTHX_ ctx); fujson_fmt_indent(aTHX_ ctx);
*hestr = HePV(he, helen); *hestr = HePV(he, helen);
fujson_fmt_str(aTHX_ ctx, *hestr, helen, HeUTF8(he)); fujson_fmt_str(aTHX_ ctx, *hestr, helen, HeUTF8(he));
if (ctx->pretty > 0) fustr_write(&ctx->out, " : ", 3); if (ctx->pretty > 0) fustr_write(&ctx->out, " : ", 3);
else fustr_write(&ctx->out, ":", 1); else fustr_write_ch(&ctx->out, ':');
fujson_fmt(aTHX_ ctx, UNLIKELY(SvMAGICAL(hv)) ? hv_iterval(hv, he) : HeVAL(he)); fujson_fmt(aTHX_ ctx, UNLIKELY(SvMAGICAL(hv)) ? hv_iterval(hv, he) : HeVAL(he));
} }
@ -173,7 +173,7 @@ static void fujson_fmt_hv(pTHX_ fujson_fmt_ctx *ctx, HV *hv) {
char *hestr = NULL; char *hestr = NULL;
int numkeys = hv_iterinit(hv); int numkeys = hv_iterinit(hv);
fustr_write(&ctx->out, "{", 1); fustr_write_ch(&ctx->out, '{');
ctx->pretty++; ctx->pretty++;
/* Canonical order on tied hashes is not supported. Cpanel::JSON::XS has /* Canonical order on tied hashes is not supported. Cpanel::JSON::XS has
@ -204,7 +204,7 @@ static void fujson_fmt_hv(pTHX_ fujson_fmt_ctx *ctx, HV *hv) {
} }
ctx->pretty--; ctx->pretty--;
if (hestr) fujson_fmt_indent(aTHX_ ctx); if (hestr) fujson_fmt_indent(aTHX_ ctx);
fustr_write(&ctx->out, "}", 1); fustr_write_ch(&ctx->out, '}');
} }
static void fujson_fmt_obj(pTHX_ fujson_fmt_ctx *ctx, SV *rv, SV *obj) { static void fujson_fmt_obj(pTHX_ fujson_fmt_ctx *ctx, SV *rv, SV *obj) {
@ -305,9 +305,9 @@ static SV *fujson_fmt_xs(pTHX_ I32 ax, I32 argc, SV *val) {
if (ctx.out.maxlen == 0) ctx.out.maxlen = 1<<30; if (ctx.out.maxlen == 0) ctx.out.maxlen = 1<<30;
if (ctx.depth == 0) ctx.depth = 512; if (ctx.depth == 0) ctx.depth = 512;
fustr_init(&ctx.out, sv_2mortal(newSV(128)), ctx.out.maxlen); fustr_init(&ctx.out, sv_newmortal(), ctx.out.maxlen);
fujson_fmt(aTHX_ &ctx, val); fujson_fmt(aTHX_ &ctx, val);
if (ctx.pretty >= 0) fustr_write(&ctx.out, "\n", 1); if (ctx.pretty >= 0) fustr_write_ch(&ctx.out, '\n');
r = fustr_done(&ctx.out); r = fustr_done(&ctx.out);
if (!encutf8) SvUTF8_on(r); if (!encutf8) SvUTF8_on(r);
return r; return r;

107
c/jsonparse.c
View file

@ -14,19 +14,19 @@ static void fujson_parse_ws(pTHX_ fujson_parse_ctx *ctx) {
} }
} }
static int fujson_parse_string_escape(pTHX_ fujson_parse_ctx *ctx, fustr *r) { static inline int fujson_parse_string_escape(fujson_parse_ctx *ctx, fustr *r) {
unsigned int n, s; unsigned int n, s;
ctx->buf++; /* '\\' */ ctx->buf++; /* '\\' */
if (ctx->buf == ctx->end) return 1; if (UNLIKELY(ctx->buf == ctx->end)) return 1;
switch (*(ctx->buf++)) { switch (*(ctx->buf++)) {
case '"': fustr_write(r, "\"", 1); break; case '"': *(r->cur++) = '\"'; break;
case '\\': fustr_write(r, "\\", 1); break; case '\\':*(r->cur++) = '\\'; break;
case '/': fustr_write(r, "/", 1); break; /* We don't escape this one */ case '/': *(r->cur++) = '/'; break; /* We don't escape this one */
case 'b': fustr_write(r, "\x08", 1); break; case 'b': *(r->cur++) = 0x08; break;
case 't': fustr_write(r, "\x09", 1); break; case 't': *(r->cur++) = 0x09; break;
case 'n': fustr_write(r, "\x0a", 1); break; case 'n': *(r->cur++) = 0x0a; break;
case 'f': fustr_write(r, "\x0c", 1); break; case 'f': *(r->cur++) = 0x0c; break;
case 'r': fustr_write(r, "\x0d", 1); break; case 'r': *(r->cur++) = 0x0d; break;
case 'u': case 'u':
/* (awful code adapted from ncdu) */ /* (awful code adapted from ncdu) */
#define INV (1<<16) #define INV (1<<16)
@ -44,8 +44,8 @@ static int fujson_parse_string_escape(pTHX_ fujson_parse_ctx *ctx, fustr *r) {
n = 0x10000 + (((n & 0x03ff) << 10) | (s & 0x03ff)); n = 0x10000 + (((n & 0x03ff) << 10) | (s & 0x03ff));
ctx->buf += 6; ctx->buf += 6;
} }
fustr_reserve(r, 4);
r->cur = (char *)uvchr_to_utf8((U8 *)r->cur, n); r->cur = (char *)uvchr_to_utf8((U8 *)r->cur, n);
if (n >= 0x80) r->setutf8 = 1;
break; break;
#undef INV #undef INV
#undef hn #undef hn
@ -56,45 +56,41 @@ static int fujson_parse_string_escape(pTHX_ fujson_parse_ctx *ctx, fustr *r) {
return 0; return 0;
} }
static SV *fujson_parse_string(pTHX_ fujson_parse_ctx *ctx) { static int fujson_parse_string_buf(pTHX_ fujson_parse_ctx *ctx, fustr *r) {
fustr r; size_t len;
size_t len, maxlen; unsigned char x;
unsigned char x = 0;
fustr_init(&r, newSV(32), SIZE_MAX);
ctx->buf++; /* '"' */ ctx->buf++; /* '"' */
while (true) { while (true) {
/* Fast path: ASCII, no unescaping needed */ fustr_reserve(aTHX_ r, 4);
len = 0; if (UNLIKELY(ctx->buf == ctx->end)) return 1;
maxlen = ctx->end - ctx->buf; x = *ctx->buf;
while (len < maxlen) { if (UNLIKELY(x == '"')) {
x = ctx->buf[len];
/* While we always escape 0x7f when formatting, JSON does permit it unescaped */
if (x <= 0x1f || x == '"' || x == '\\' || x >= 0x80) break;
len++;
}
if (len == maxlen) goto err;
fustr_write(&r, (const char *)ctx->buf, len);
ctx->buf += len;
/* Slow path */
if (x == '"') {
ctx->buf++; ctx->buf++;
SvUTF8_on(fustr_done(&r)); return 0;
return r.sv; } else if (UNLIKELY(x == '\\')) {
} else if (x == '\\') { if (fujson_parse_string_escape(ctx, r)) return 1;
if (fujson_parse_string_escape(aTHX_ ctx, &r)) goto err;
} else if (x >= 0x80) { } else if (x >= 0x80) {
len = isC9_STRICT_UTF8_CHAR(ctx->buf, ctx->end); if (UNLIKELY((len = isC9_STRICT_UTF8_CHAR(ctx->buf, ctx->end)) == 0)) return 1;
if (len == 0) goto err; memcpy(r->cur, ctx->buf, len);
fustr_write(&r, (const char *)ctx->buf, len); r->cur += len;
ctx->buf += len; ctx->buf += len;
} else { r->setutf8 = 1;
goto err; } else if (x >= 0x20) {
} *(r->cur++) = x;
ctx->buf++;
} else return 1;
}
}
static SV *fujson_parse_string(pTHX_ fujson_parse_ctx *ctx) {
fustr r;
fustr_init(&r, NULL, SIZE_MAX);
if (fujson_parse_string_buf(aTHX_ ctx, &r)) {
if (r.sv) SvREFCNT_dec(r.sv);
return NULL;
} else {
return fustr_done(&r);
} }
err:
SvREFCNT_dec(r.sv);
return NULL;
} }
/* Validate JSON grammar of a number, increments ctx->buf to the end of the /* Validate JSON grammar of a number, increments ctx->buf to the end of the
@ -180,8 +176,12 @@ err:
static SV *fujson_parse_obj(pTHX_ fujson_parse_ctx *ctx) { static SV *fujson_parse_obj(pTHX_ fujson_parse_ctx *ctx) {
HV *hv = newHV(); HV *hv = newHV();
SV *key = NULL;
SV *val; SV *val;
char *keystart;
UV keyhash;
fustr key;
fustr_init(&key, NULL, SIZE_MAX);
ctx->buf++; /* '{' */ ctx->buf++; /* '{' */
fujson_parse_ws(aTHX_ ctx); fujson_parse_ws(aTHX_ ctx);
if (ctx->buf == ctx->end) goto err; if (ctx->buf == ctx->end) goto err;
@ -189,9 +189,11 @@ static SV *fujson_parse_obj(pTHX_ fujson_parse_ctx *ctx) {
while (true) { while (true) {
/* key */ /* key */
if (*ctx->buf != '"') goto err; if (*ctx->buf != '"') goto err;
if (!(key = fujson_parse_string(aTHX_ ctx))) goto err; if (fujson_parse_string_buf(aTHX_ ctx, &key)) goto err;
/* TODO: Use precomputed hash */ keystart = fustr_start(&key);
if (hv_exists_ent(hv, key, 0)) goto err; if (key.setutf8) keyhash = 0;
else PERL_HASH(keyhash, keystart, key.cur - keystart);
if (hv_common(hv, NULL, keystart, key.cur - keystart, key.setutf8, HV_FETCH_ISEXISTS, NULL, keyhash)) goto err;
/* ':' */ /* ':' */
fujson_parse_ws(aTHX_ ctx); fujson_parse_ws(aTHX_ ctx);
@ -201,9 +203,9 @@ static SV *fujson_parse_obj(pTHX_ fujson_parse_ctx *ctx) {
/* value */ /* value */
if (!(val = fujson_parse(aTHX_ ctx))) goto err; if (!(val = fujson_parse(aTHX_ ctx))) goto err;
hv_store_ent(hv, key, val, 0); hv_common(hv, NULL, keystart, key.cur - keystart, key.setutf8, HV_FETCH_ISSTORE|HV_FETCH_JUST_SV, val, keyhash);
SvREFCNT_dec(key); /* TODO: can reuse buffer */ key.cur = keystart;
key = NULL; key.setutf8 = 0;
fujson_parse_ws(aTHX_ ctx); fujson_parse_ws(aTHX_ ctx);
if (ctx->buf == ctx->end) goto err; if (ctx->buf == ctx->end) goto err;
@ -213,10 +215,11 @@ static SV *fujson_parse_obj(pTHX_ fujson_parse_ctx *ctx) {
fujson_parse_ws(aTHX_ ctx); fujson_parse_ws(aTHX_ ctx);
} }
done: done:
if (key.sv) SvREFCNT_dec(key.sv);
ctx->buf++; /* '}' */ ctx->buf++; /* '}' */
return newRV_noinc((SV *)hv); return newRV_noinc((SV *)hv);
err: err:
if (key) SvREFCNT_dec(key); if (key.sv) SvREFCNT_dec(key.sv);
SvREFCNT_dec((SV *)hv); SvREFCNT_dec((SV *)hv);
return NULL; return NULL;
} }

5
t/json_parse.t
View file

@ -49,6 +49,7 @@ my @error = (
'{"":1,}', '{"":1,}',
'{,}', '{,}',
'{"":1,"":2}', '{"":1,"":2}',
'{"ë":1,"ë":1}',
); );
for my $s (@error) { for my $s (@error) {
ok !eval { json_parse($s); 1 }; ok !eval { json_parse($s); 1 };
@ -70,10 +71,10 @@ sub str($in, $exp) {
utf8::encode(my $str = $in); utf8::encode(my $str = $in);
my $out = json_parse($in); my $out = json_parse($in);
is $out, $exp, $str; is $out, $exp, $str;
ok utf8::is_utf8($out); ok utf8::is_utf8($out) || $out =~ /^[\x00-\x7f]*$/;
$out = json_parse($str, utf8 => 1); $out = json_parse($str, utf8 => 1);
is $out, $exp, $str; is $out, $exp, $str;
ok utf8::is_utf8($out); ok utf8::is_utf8($out) || $out =~ /^[\x00-\x7f]*$/;
} }
str '""', ''; str '""', '';
str '"hello, world"', 'hello, world'; str '"hello, world"', 'hello, world';