2129 lines
56 KiB
Perl
2129 lines
56 KiB
Perl
package GT::JSON::PP;
|
|
|
|
# GT::JSON-2.0
|
|
|
|
use 5.005;
|
|
use strict;
|
|
use base qw(Exporter);
|
|
use overload;
|
|
|
|
use Carp ();
|
|
use B ();
|
|
#use Devel::Peek;
|
|
|
|
$GT::JSON::PP::VERSION = '2.22010';
|
|
|
|
@GT::JSON::PP::EXPORT = qw(encode_json decode_json from_json to_json);
|
|
|
|
# instead of hash-access, i tried index-access for speed.
|
|
# but this method is not faster than what i expected. so it will be changed.
|
|
|
|
use constant P_ASCII => 0;
|
|
use constant P_LATIN1 => 1;
|
|
use constant P_UTF8 => 2;
|
|
use constant P_INDENT => 3;
|
|
use constant P_CANONICAL => 4;
|
|
use constant P_SPACE_BEFORE => 5;
|
|
use constant P_SPACE_AFTER => 6;
|
|
use constant P_ALLOW_NONREF => 7;
|
|
use constant P_SHRINK => 8;
|
|
use constant P_ALLOW_BLESSED => 9;
|
|
use constant P_CONVERT_BLESSED => 10;
|
|
use constant P_RELAXED => 11;
|
|
|
|
use constant P_LOOSE => 12;
|
|
use constant P_ALLOW_BIGNUM => 13;
|
|
use constant P_ALLOW_BAREKEY => 14;
|
|
use constant P_ALLOW_SINGLEQUOTE => 15;
|
|
use constant P_ESCAPE_SLASH => 16;
|
|
use constant P_AS_NONBLESSED => 17;
|
|
|
|
use constant P_ALLOW_UNKNOWN => 18;
|
|
|
|
BEGIN {
|
|
my @xs_compati_bit_properties = qw(
|
|
latin1 ascii utf8 indent canonical space_before space_after allow_nonref shrink
|
|
allow_blessed convert_blessed relaxed allow_unknown
|
|
);
|
|
my @pp_bit_properties = qw(
|
|
allow_singlequote allow_bignum loose
|
|
allow_barekey escape_slash as_nonblessed
|
|
);
|
|
|
|
# Perl version check, Unicode handling is enable?
|
|
# Helper module sets @GT::JSON::PP::_properties.
|
|
|
|
my $helper = $] >= 5.008 ? 'GT::JSON::PP58'
|
|
: $] >= 5.006 ? 'GT::JSON::PP56'
|
|
: 'GT::JSON::PP5005'
|
|
;
|
|
|
|
eval qq| require $helper |;
|
|
if ($@) { Carp::croak $@; }
|
|
|
|
for my $name (@xs_compati_bit_properties, @pp_bit_properties) {
|
|
my $flag_name = 'P_' . uc($name);
|
|
|
|
eval qq/
|
|
sub $name {
|
|
my \$enable = defined \$_[1] ? \$_[1] : 1;
|
|
|
|
if (\$enable) {
|
|
\$_[0]->{PROPS}->[$flag_name] = 1;
|
|
}
|
|
else {
|
|
\$_[0]->{PROPS}->[$flag_name] = 0;
|
|
}
|
|
|
|
\$_[0];
|
|
}
|
|
|
|
sub get_$name {
|
|
\$_[0]->{PROPS}->[$flag_name] ? 1 : '';
|
|
}
|
|
/;
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
# Functions
|
|
|
|
my %encode_allow_method
|
|
= map {($_ => 1)} qw/utf8 pretty allow_nonref latin1 self_encode escape_slash
|
|
allow_blessed convert_blessed indent indent_length allow_bignum
|
|
as_nonblessed
|
|
/;
|
|
my %decode_allow_method
|
|
= map {($_ => 1)} qw/utf8 allow_nonref loose allow_singlequote allow_bignum
|
|
allow_barekey max_size relaxed/;
|
|
|
|
|
|
my $JSON; # cache
|
|
|
|
sub encode_json ($) { # encode
|
|
($JSON ||= __PACKAGE__->new->utf8)->encode(@_);
|
|
}
|
|
|
|
|
|
sub decode_json { # decode
|
|
($JSON ||= __PACKAGE__->new->utf8)->decode(@_);
|
|
}
|
|
|
|
# Obsoleted
|
|
|
|
sub to_json($) {
|
|
Carp::croak ("GT::JSON::PP::to_json has been renamed to encode_json.");
|
|
}
|
|
|
|
|
|
sub from_json($) {
|
|
Carp::croak ("GT::JSON::PP::from_json has been renamed to decode_json.");
|
|
}
|
|
|
|
|
|
# Methods
|
|
|
|
sub new {
|
|
my $class = shift;
|
|
my $self = {
|
|
max_depth => 512,
|
|
max_size => 0,
|
|
indent => 0,
|
|
FLAGS => 0,
|
|
fallback => sub { encode_error('Invalid value. JSON can only reference.') },
|
|
indent_length => 3,
|
|
};
|
|
|
|
bless $self, $class;
|
|
}
|
|
|
|
|
|
sub encode {
|
|
return $_[0]->PP_encode_json($_[1]);
|
|
}
|
|
|
|
|
|
sub decode {
|
|
return $_[0]->PP_decode_json($_[1], 0x00000000);
|
|
}
|
|
|
|
|
|
sub decode_prefix {
|
|
return $_[0]->PP_decode_json($_[1], 0x00000001);
|
|
}
|
|
|
|
|
|
# accessor
|
|
|
|
|
|
# pretty printing
|
|
|
|
sub pretty {
|
|
my ($self, $v) = @_;
|
|
my $enable = defined $v ? $v : 1;
|
|
|
|
if ($enable) { # indent_length(3) for JSON::XS compatibility
|
|
$self->indent(1)->indent_length(3)->space_before(1)->space_after(1);
|
|
}
|
|
else {
|
|
$self->indent(0)->space_before(0)->space_after(0);
|
|
}
|
|
|
|
$self;
|
|
}
|
|
|
|
# etc
|
|
|
|
sub max_depth {
|
|
my $max = defined $_[1] ? $_[1] : 0x80000000;
|
|
$_[0]->{max_depth} = $max;
|
|
$_[0];
|
|
}
|
|
|
|
|
|
sub get_max_depth { $_[0]->{max_depth}; }
|
|
|
|
|
|
sub max_size {
|
|
my $max = defined $_[1] ? $_[1] : 0;
|
|
$_[0]->{max_size} = $max;
|
|
$_[0];
|
|
}
|
|
|
|
|
|
sub get_max_size { $_[0]->{max_size}; }
|
|
|
|
|
|
sub filter_json_object {
|
|
$_[0]->{cb_object} = defined $_[1] ? $_[1] : 0;
|
|
$_[0]->{F_HOOK} = ($_[0]->{cb_object} or $_[0]->{cb_sk_object}) ? 1 : 0;
|
|
$_[0];
|
|
}
|
|
|
|
sub filter_json_single_key_object {
|
|
if (@_ > 1) {
|
|
$_[0]->{cb_sk_object}->{$_[1]} = $_[2];
|
|
}
|
|
$_[0]->{F_HOOK} = ($_[0]->{cb_object} or $_[0]->{cb_sk_object}) ? 1 : 0;
|
|
$_[0];
|
|
}
|
|
|
|
sub indent_length {
|
|
if (!defined $_[1] or $_[1] > 15 or $_[1] < 0) {
|
|
Carp::carp "The acceptable range of indent_length() is 0 to 15.";
|
|
}
|
|
else {
|
|
$_[0]->{indent_length} = $_[1];
|
|
}
|
|
$_[0];
|
|
}
|
|
|
|
sub get_indent_length {
|
|
$_[0]->{indent_length};
|
|
}
|
|
|
|
sub sort_by {
|
|
$_[0]->{sort_by} = defined $_[1] ? $_[1] : 1;
|
|
$_[0];
|
|
}
|
|
|
|
sub allow_bigint {
|
|
Carp::carp("allow_bigint() is obsoleted. use allow_bignum() insted.");
|
|
}
|
|
|
|
###############################
|
|
|
|
###
|
|
### Perl => JSON
|
|
###
|
|
|
|
|
|
{ # Convert
|
|
|
|
my $max_depth;
|
|
my $indent;
|
|
my $ascii;
|
|
my $latin1;
|
|
my $utf8;
|
|
my $space_before;
|
|
my $space_after;
|
|
my $canonical;
|
|
my $allow_blessed;
|
|
my $convert_blessed;
|
|
|
|
my $indent_length;
|
|
my $escape_slash;
|
|
my $bignum;
|
|
my $as_nonblessed;
|
|
|
|
my $depth;
|
|
my $indent_count;
|
|
my $keysort;
|
|
|
|
|
|
sub PP_encode_json {
|
|
my $self = shift;
|
|
my $obj = shift;
|
|
|
|
$indent_count = 0;
|
|
$depth = 0;
|
|
|
|
my $idx = $self->{PROPS};
|
|
|
|
($ascii, $latin1, $utf8, $indent, $canonical, $space_before, $space_after, $allow_blessed,
|
|
$convert_blessed, $escape_slash, $bignum, $as_nonblessed)
|
|
= @{$idx}[P_ASCII .. P_SPACE_AFTER, P_ALLOW_BLESSED, P_CONVERT_BLESSED,
|
|
P_ESCAPE_SLASH, P_ALLOW_BIGNUM, P_AS_NONBLESSED];
|
|
|
|
($max_depth, $indent_length) = @{$self}{qw/max_depth indent_length/};
|
|
|
|
$keysort = $canonical ? sub { $a cmp $b } : undef;
|
|
|
|
if ($self->{sort_by}) {
|
|
$keysort = ref($self->{sort_by}) eq 'CODE' ? $self->{sort_by}
|
|
: $self->{sort_by} =~ /\D+/ ? $self->{sort_by}
|
|
: sub { $a cmp $b };
|
|
}
|
|
|
|
encode_error("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)")
|
|
if(!ref $obj and !$idx->[ P_ALLOW_NONREF ]);
|
|
|
|
my $str = $self->object_to_json($obj);
|
|
|
|
unless ($ascii or $latin1 or $utf8) {
|
|
utf8::upgrade($str);
|
|
}
|
|
|
|
if ($idx->[ P_SHRINK ]) {
|
|
utf8::downgrade($str, 1);
|
|
}
|
|
|
|
return $str;
|
|
}
|
|
|
|
|
|
sub object_to_json {
|
|
my ($self, $obj) = @_;
|
|
my $type = ref($obj);
|
|
|
|
if($type eq 'HASH'){
|
|
return $self->hash_to_json($obj);
|
|
}
|
|
elsif($type eq 'ARRAY'){
|
|
return $self->array_to_json($obj);
|
|
}
|
|
elsif ($type) { # blessed object?
|
|
if (blessed($obj)) {
|
|
|
|
return $self->value_to_json($obj) if ( $obj->isa('GT::JSON::PP::Boolean') );
|
|
|
|
if ( $convert_blessed and $obj->can('TO_JSON') ) {
|
|
my $result = $obj->TO_JSON();
|
|
if ( defined $result and $obj eq $result ) {
|
|
encode_error( sprintf(
|
|
"%s::TO_JSON method returned same object as was passed instead of a new one",
|
|
ref $obj
|
|
) );
|
|
}
|
|
return $self->object_to_json( $result );
|
|
}
|
|
|
|
return "$obj" if ( $bignum and _is_bignum($obj) );
|
|
return $self->blessed_to_json($obj) if ($allow_blessed and $as_nonblessed); # will be removed.
|
|
|
|
encode_error( sprintf("encountered object '%s', but neither allow_blessed "
|
|
. "nor convert_blessed settings are enabled", $obj)
|
|
) unless ($allow_blessed);
|
|
|
|
return 'null';
|
|
}
|
|
else {
|
|
return $self->value_to_json($obj);
|
|
}
|
|
}
|
|
else{
|
|
return $self->value_to_json($obj);
|
|
}
|
|
}
|
|
|
|
|
|
sub hash_to_json {
|
|
my ($self, $obj) = @_;
|
|
my ($k,$v);
|
|
my %res;
|
|
|
|
encode_error("json text or perl structure exceeds maximum nesting level (max_depth set too low?)")
|
|
if (++$depth > $max_depth);
|
|
|
|
my ($pre, $post) = $indent ? $self->_up_indent() : ('', '');
|
|
my $del = ($space_before ? ' ' : '') . ':' . ($space_after ? ' ' : '');
|
|
|
|
if ( my $tie_class = tied %$obj ) {
|
|
if ( $tie_class->can('TIEHASH') ) {
|
|
$tie_class =~ s/=.+$//;
|
|
tie %res, $tie_class;
|
|
}
|
|
}
|
|
|
|
# In the old Perl verions, tied hashes in bool context didn't work.
|
|
# So, we can't use such a way (%res ? a : b)
|
|
my $has;
|
|
|
|
for my $k (keys %$obj) {
|
|
my $v = $obj->{$k};
|
|
$res{$k} = $self->object_to_json($v) || $self->value_to_json($v);
|
|
$has = 1 unless ( $has );
|
|
}
|
|
|
|
--$depth;
|
|
$self->_down_indent() if ($indent);
|
|
|
|
return '{' . ( $has ? $pre : '' ) # indent
|
|
. ( $has ? join(",$pre", map { utf8::decode($_) if ($] < 5.008); # key for Perl 5.6
|
|
string_to_json($self, $_) . $del . $res{$_} # key : value
|
|
} _sort( $self, \%res )
|
|
) . $post # indent
|
|
: ''
|
|
)
|
|
. '}';
|
|
}
|
|
|
|
|
|
sub array_to_json {
|
|
my ($self, $obj) = @_;
|
|
my @res;
|
|
|
|
encode_error("json text or perl structure exceeds maximum nesting level (max_depth set too low?)")
|
|
if (++$depth > $max_depth);
|
|
|
|
my ($pre, $post) = $indent ? $self->_up_indent() : ('', '');
|
|
|
|
if (my $tie_class = tied @$obj) {
|
|
if ( $tie_class->can('TIEARRAY') ) {
|
|
$tie_class =~ s/=.+$//;
|
|
tie @res, $tie_class;
|
|
}
|
|
}
|
|
|
|
for my $v (@$obj){
|
|
push @res, $self->object_to_json($v) || $self->value_to_json($v);
|
|
}
|
|
|
|
--$depth;
|
|
$self->_down_indent() if ($indent);
|
|
|
|
return '[' . ( @res ? $pre : '' ) . ( @res ? join( ",$pre", @res ) . $post : '' ) . ']';
|
|
}
|
|
|
|
|
|
sub value_to_json {
|
|
my ($self, $value) = @_;
|
|
|
|
return 'null' if(!defined $value);
|
|
|
|
my $b_obj = B::svref_2object(\$value); # for round trip problem
|
|
my $flags = $b_obj->FLAGS;
|
|
|
|
return $value # as is
|
|
if ( ( $flags & B::SVf_IOK or $flags & B::SVp_IOK
|
|
or $flags & B::SVf_NOK or $flags & B::SVp_NOK
|
|
) and !($flags & B::SVf_POK )
|
|
); # SvTYPE is IV or NV?
|
|
|
|
my $type = ref($value);
|
|
|
|
if(!$type){
|
|
return string_to_json($self, $value);
|
|
}
|
|
elsif( blessed($value) and $value->isa('GT::JSON::PP::Boolean') ){
|
|
return $$value == 1 ? 'true' : 'false';
|
|
}
|
|
elsif ($type) {
|
|
if ((overload::StrVal($value) =~ /=(\w+)/)[0]) {
|
|
return $self->value_to_json("$value");
|
|
}
|
|
|
|
if ($type eq 'SCALAR' and defined $$value) {
|
|
return $$value eq '1' ? 'true'
|
|
: $$value eq '0' ? 'false'
|
|
: $self->{PROPS}->[ P_ALLOW_UNKNOWN ] ? 'null'
|
|
: encode_error("cannot encode reference to scalar");
|
|
}
|
|
|
|
if ( $self->{PROPS}->[ P_ALLOW_UNKNOWN ] ) {
|
|
return 'null';
|
|
}
|
|
else {
|
|
if ( $type eq 'SCALAR' or $type eq 'REF' ) {
|
|
encode_error("cannot encode reference to scalar");
|
|
}
|
|
else {
|
|
encode_error("encountered $value, but JSON can only represent references to arrays or hashes");
|
|
}
|
|
}
|
|
|
|
}
|
|
else {
|
|
return $self->{fallback}->($value)
|
|
if ($self->{fallback} and ref($self->{fallback}) eq 'CODE');
|
|
return 'null';
|
|
}
|
|
|
|
}
|
|
|
|
|
|
my %esc = (
|
|
"\n" => '\n',
|
|
"\r" => '\r',
|
|
"\t" => '\t',
|
|
"\f" => '\f',
|
|
"\b" => '\b',
|
|
"\"" => '\"',
|
|
"\\" => '\\\\',
|
|
"\'" => '\\\'',
|
|
);
|
|
|
|
|
|
sub string_to_json {
|
|
my ($self, $arg) = @_;
|
|
|
|
$arg =~ s/([\x22\x5c\n\r\t\f\b])/$esc{$1}/eg;
|
|
$arg =~ s/\//\\\//g if ($escape_slash);
|
|
$arg =~ s/([\x00-\x08\x0b\x0e-\x1f])/'\\u00' . unpack('H2', $1)/eg;
|
|
|
|
if ($ascii) {
|
|
$arg = JSON_PP_encode_ascii($arg);
|
|
}
|
|
|
|
if ($latin1) {
|
|
$arg = JSON_PP_encode_latin1($arg);
|
|
}
|
|
|
|
if ($utf8) {
|
|
utf8::encode($arg);
|
|
}
|
|
|
|
return '"' . $arg . '"';
|
|
}
|
|
|
|
|
|
sub blessed_to_json {
|
|
my $b_obj = B::svref_2object($_[1]);
|
|
if ($b_obj->isa('B::HV')) {
|
|
return $_[0]->hash_to_json($_[1]);
|
|
}
|
|
elsif ($b_obj->isa('B::AV')) {
|
|
return $_[0]->array_to_json($_[1]);
|
|
}
|
|
else {
|
|
return 'null';
|
|
}
|
|
}
|
|
|
|
|
|
sub encode_error {
|
|
my $error = shift;
|
|
Carp::croak "$error";
|
|
}
|
|
|
|
|
|
sub _sort {
|
|
my ($self, $res) = @_;
|
|
defined $keysort ? (sort $keysort (keys %$res)) : keys %$res;
|
|
}
|
|
|
|
|
|
sub _up_indent {
|
|
my $self = shift;
|
|
my $space = ' ' x $indent_length;
|
|
|
|
my ($pre,$post) = ('','');
|
|
|
|
$post = "\n" . $space x $indent_count;
|
|
|
|
$indent_count++;
|
|
|
|
$pre = "\n" . $space x $indent_count;
|
|
|
|
return ($pre,$post);
|
|
}
|
|
|
|
|
|
sub _down_indent { $indent_count--; }
|
|
|
|
|
|
sub PP_encode_box {
|
|
{
|
|
depth => $depth,
|
|
indent_count => $indent_count,
|
|
};
|
|
}
|
|
|
|
} # Convert
|
|
|
|
|
|
sub _encode_ascii {
|
|
join('',
|
|
map {
|
|
$_ <= 127 ?
|
|
chr($_) :
|
|
$_ <= 65535 ?
|
|
sprintf('\u%04x', $_) : sprintf('\u%x\u%x', _encode_surrogates($_));
|
|
} unpack('U*', $_[0])
|
|
);
|
|
}
|
|
|
|
|
|
sub _encode_latin1 {
|
|
join('',
|
|
map {
|
|
$_ <= 255 ?
|
|
chr($_) :
|
|
$_ <= 65535 ?
|
|
sprintf('\u%04x', $_) : sprintf('\u%x\u%x', _encode_surrogates($_));
|
|
} unpack('U*', $_[0])
|
|
);
|
|
}
|
|
|
|
|
|
sub _encode_surrogates { # from perlunicode
|
|
my $uni = $_[0] - 0x10000;
|
|
return ($uni / 0x400 + 0xD800, $uni % 0x400 + 0xDC00);
|
|
}
|
|
|
|
|
|
sub _is_bignum {
|
|
$_[0]->isa('Math::BigInt') or $_[0]->isa('Math::BigFloat');
|
|
}
|
|
|
|
|
|
|
|
#
|
|
# JSON => Perl
|
|
#
|
|
|
|
my $max_intsize;
|
|
|
|
BEGIN {
|
|
my $checkint = 1111;
|
|
for my $d (5..30) {
|
|
$checkint .= 1;
|
|
my $int = eval qq| $checkint |;
|
|
if ($int =~ /[eE]/) {
|
|
$max_intsize = $d - 1;
|
|
last;
|
|
}
|
|
}
|
|
}
|
|
|
|
{ # PARSE
|
|
|
|
my %escapes = ( # by Jeremy Muhlich <jmuhlich [at] bitflood.org>
|
|
b => "\x8",
|
|
t => "\x9",
|
|
n => "\xA",
|
|
f => "\xC",
|
|
r => "\xD",
|
|
'\\' => '\\',
|
|
'"' => '"',
|
|
'/' => '/',
|
|
);
|
|
|
|
my $text; # json data
|
|
my $at; # offset
|
|
my $ch; # 1chracter
|
|
my $len; # text length (changed according to UTF8 or NON UTF8)
|
|
# INTERNAL
|
|
my $depth; # nest counter
|
|
my $encoding; # json text encoding
|
|
my $is_valid_utf8; # temp variable
|
|
my $utf8_len; # utf8 byte length
|
|
# FLAGS
|
|
my $utf8; # must be utf8
|
|
my $max_depth; # max nest nubmer of objects and arrays
|
|
my $max_size;
|
|
my $relaxed;
|
|
my $cb_object;
|
|
my $cb_sk_object;
|
|
|
|
my $F_HOOK;
|
|
|
|
my $allow_bigint; # using Math::BigInt
|
|
my $singlequote; # loosely quoting
|
|
my $loose; #
|
|
my $allow_barekey; # bareKey
|
|
|
|
# $opt flag
|
|
# 0x00000001 .... decode_prefix
|
|
|
|
sub PP_decode_json {
|
|
my ($self, $opt); # $opt is an effective flag during this decode_json.
|
|
|
|
($self, $text, $opt) = @_;
|
|
|
|
($at, $ch, $depth) = (0, '', 0);
|
|
|
|
if (!defined $text or ref $text) {
|
|
decode_error("malformed text data.");
|
|
}
|
|
|
|
my $idx = $self->{PROPS};
|
|
|
|
($utf8, $relaxed, $loose, $allow_bigint, $allow_barekey, $singlequote)
|
|
= @{$idx}[P_UTF8, P_RELAXED, P_LOOSE .. P_ALLOW_SINGLEQUOTE];
|
|
|
|
if ( $utf8 ) {
|
|
utf8::downgrade( $text, 1 ) or Carp::croak("Wide character in subroutine entry");
|
|
}
|
|
else {
|
|
utf8::upgrade( $text );
|
|
}
|
|
|
|
$len = length $text;
|
|
|
|
($max_depth, $max_size, $cb_object, $cb_sk_object, $F_HOOK)
|
|
= @{$self}{qw/max_depth max_size cb_object cb_sk_object F_HOOK/};
|
|
|
|
if ($max_size > 1) {
|
|
use bytes;
|
|
my $bytes = length $text;
|
|
decode_error(
|
|
sprintf("attempted decode of JSON text of %s bytes size, but max_size is set to %s"
|
|
, $bytes, $max_size), 1
|
|
) if ($bytes > $max_size);
|
|
}
|
|
|
|
# Currently no effect
|
|
# should use regexp
|
|
my @octets = unpack('C4', $text);
|
|
$encoding = ( $octets[0] and $octets[1]) ? 'UTF-8'
|
|
: (!$octets[0] and $octets[1]) ? 'UTF-16BE'
|
|
: (!$octets[0] and !$octets[1]) ? 'UTF-32BE'
|
|
: ( $octets[2] ) ? 'UTF-16LE'
|
|
: (!$octets[2] ) ? 'UTF-32LE'
|
|
: 'unknown';
|
|
|
|
my $result = value();
|
|
|
|
if (!$idx->[ P_ALLOW_NONREF ] and !ref $result) {
|
|
decode_error(
|
|
'JSON text must be an object or array (but found number, string, true, false or null,'
|
|
. ' use allow_nonref to allow this)', 1);
|
|
}
|
|
|
|
if ($len >= $at) {
|
|
my $consumed = $at - 1;
|
|
white();
|
|
if ($ch) {
|
|
decode_error("garbage after JSON object") unless ($opt & 0x00000001);
|
|
return ($result, $consumed);
|
|
}
|
|
}
|
|
|
|
$result;
|
|
}
|
|
|
|
|
|
sub next_chr {
|
|
return $ch = undef if($at >= $len);
|
|
$ch = substr($text, $at++, 1);
|
|
}
|
|
|
|
|
|
sub value {
|
|
white();
|
|
return if(!defined $ch);
|
|
return object() if($ch eq '{');
|
|
return array() if($ch eq '[');
|
|
return string() if($ch eq '"' or ($singlequote and $ch eq "'"));
|
|
return number() if($ch =~ /[0-9]/ or $ch eq '-');
|
|
return word();
|
|
}
|
|
|
|
sub string {
|
|
my ($i, $s, $t, $u);
|
|
my $utf16;
|
|
my $is_utf8;
|
|
|
|
($is_valid_utf8, $utf8_len) = ('', 0);
|
|
|
|
$s = ''; # basically UTF8 flag on
|
|
|
|
if($ch eq '"' or ($singlequote and $ch eq "'")){
|
|
my $boundChar = $ch if ($singlequote);
|
|
|
|
OUTER: while( defined(next_chr()) ){
|
|
|
|
if((!$singlequote and $ch eq '"') or ($singlequote and $ch eq $boundChar)){
|
|
next_chr();
|
|
|
|
if ($utf16) {
|
|
decode_error("missing low surrogate character in surrogate pair");
|
|
}
|
|
|
|
utf8::decode($s) if($is_utf8);
|
|
|
|
return $s;
|
|
}
|
|
elsif($ch eq '\\'){
|
|
next_chr();
|
|
if(exists $escapes{$ch}){
|
|
$s .= $escapes{$ch};
|
|
}
|
|
elsif($ch eq 'u'){ # UNICODE handling
|
|
my $u = '';
|
|
|
|
for(1..4){
|
|
$ch = next_chr();
|
|
last OUTER if($ch !~ /[0-9a-fA-F]/);
|
|
$u .= $ch;
|
|
}
|
|
|
|
# U+D800 - U+DBFF
|
|
if ($u =~ /^[dD][89abAB][0-9a-fA-F]{2}/) { # UTF-16 high surrogate?
|
|
$utf16 = $u;
|
|
}
|
|
# U+DC00 - U+DFFF
|
|
elsif ($u =~ /^[dD][c-fC-F][0-9a-fA-F]{2}/) { # UTF-16 low surrogate?
|
|
unless (defined $utf16) {
|
|
decode_error("missing high surrogate character in surrogate pair");
|
|
}
|
|
$is_utf8 = 1;
|
|
$s .= JSON_PP_decode_surrogates($utf16, $u) || next;
|
|
$utf16 = undef;
|
|
}
|
|
else {
|
|
if (defined $utf16) {
|
|
decode_error("surrogate pair expected");
|
|
}
|
|
|
|
if ( ( my $hex = hex( $u ) ) > 127 ) {
|
|
$is_utf8 = 1;
|
|
$s .= JSON_PP_decode_unicode($u) || next;
|
|
}
|
|
else {
|
|
$s .= chr $hex;
|
|
}
|
|
}
|
|
|
|
}
|
|
else{
|
|
unless ($loose) {
|
|
decode_error('illegal backslash escape sequence in string');
|
|
}
|
|
$s .= $ch;
|
|
}
|
|
}
|
|
else{
|
|
|
|
if ( ord $ch > 127 ) {
|
|
if ( $utf8 ) {
|
|
unless( $ch = is_valid_utf8($ch) ) {
|
|
$at -= 1;
|
|
decode_error("malformed UTF-8 character in JSON string");
|
|
}
|
|
else {
|
|
$at += $utf8_len - 1;
|
|
}
|
|
}
|
|
else {
|
|
utf8::encode( $ch );
|
|
}
|
|
|
|
$is_utf8 = 1;
|
|
}
|
|
|
|
if (!$loose) {
|
|
if ($ch =~ /[\x00-\x1f\x22\x5c]/) { # '/' ok
|
|
$at--;
|
|
decode_error('invalid character encountered while parsing JSON string');
|
|
}
|
|
}
|
|
|
|
$s .= $ch;
|
|
}
|
|
}
|
|
}
|
|
|
|
decode_error("unexpected end of string while parsing JSON string");
|
|
}
|
|
|
|
|
|
sub white {
|
|
while( defined $ch ){
|
|
if($ch le ' '){
|
|
next_chr();
|
|
}
|
|
elsif($ch eq '/'){
|
|
next_chr();
|
|
if(defined $ch and $ch eq '/'){
|
|
1 while(defined(next_chr()) and $ch ne "\n" and $ch ne "\r");
|
|
}
|
|
elsif(defined $ch and $ch eq '*'){
|
|
next_chr();
|
|
while(1){
|
|
if(defined $ch){
|
|
if($ch eq '*'){
|
|
if(defined(next_chr()) and $ch eq '/'){
|
|
next_chr();
|
|
last;
|
|
}
|
|
}
|
|
else{
|
|
next_chr();
|
|
}
|
|
}
|
|
else{
|
|
decode_error("Unterminated comment");
|
|
}
|
|
}
|
|
next;
|
|
}
|
|
else{
|
|
$at--;
|
|
decode_error("malformed JSON string, neither array, object, number, string or atom");
|
|
}
|
|
}
|
|
else{
|
|
if ($relaxed and $ch eq '#') { # correctly?
|
|
pos($text) = $at;
|
|
$text =~ /\G([^\n]*(?:\r\n|\r|\n))/g;
|
|
$at = pos($text);
|
|
next_chr;
|
|
next;
|
|
}
|
|
|
|
last;
|
|
}
|
|
}
|
|
}
|
|
|
|
|
|
sub array {
|
|
my $a = [];
|
|
|
|
decode_error('json text or perl structure exceeds maximum nesting level (max_depth set too low?)')
|
|
if (++$depth > $max_depth);
|
|
|
|
next_chr();
|
|
white();
|
|
|
|
if(defined $ch and $ch eq ']'){
|
|
--$depth;
|
|
next_chr();
|
|
return $a;
|
|
}
|
|
else {
|
|
while(defined($ch)){
|
|
push @$a, value();
|
|
|
|
white();
|
|
|
|
if (!defined $ch) {
|
|
last;
|
|
}
|
|
|
|
if($ch eq ']'){
|
|
--$depth;
|
|
next_chr();
|
|
return $a;
|
|
}
|
|
|
|
if($ch ne ','){
|
|
last;
|
|
}
|
|
|
|
next_chr();
|
|
white();
|
|
|
|
if ($relaxed and $ch eq ']') {
|
|
--$depth;
|
|
next_chr();
|
|
return $a;
|
|
}
|
|
|
|
}
|
|
}
|
|
|
|
decode_error(", or ] expected while parsing array");
|
|
}
|
|
|
|
|
|
sub object {
|
|
my $o = {};
|
|
my $k;
|
|
|
|
decode_error('json text or perl structure exceeds maximum nesting level (max_depth set too low?)')
|
|
if (++$depth > $max_depth);
|
|
next_chr();
|
|
white();
|
|
|
|
if(defined $ch and $ch eq '}'){
|
|
--$depth;
|
|
next_chr();
|
|
if ($F_HOOK) {
|
|
return _json_object_hook($o);
|
|
}
|
|
return $o;
|
|
}
|
|
else {
|
|
while (defined $ch) {
|
|
$k = ($allow_barekey and $ch ne '"' and $ch ne "'") ? bareKey() : string();
|
|
white();
|
|
|
|
if(!defined $ch or $ch ne ':'){
|
|
$at--;
|
|
decode_error("':' expected");
|
|
}
|
|
|
|
next_chr();
|
|
$o->{$k} = value();
|
|
white();
|
|
|
|
last if (!defined $ch);
|
|
|
|
if($ch eq '}'){
|
|
--$depth;
|
|
next_chr();
|
|
if ($F_HOOK) {
|
|
return _json_object_hook($o);
|
|
}
|
|
return $o;
|
|
}
|
|
|
|
if($ch ne ','){
|
|
last;
|
|
}
|
|
|
|
next_chr();
|
|
white();
|
|
|
|
if ($relaxed and $ch eq '}') {
|
|
--$depth;
|
|
next_chr();
|
|
if ($F_HOOK) {
|
|
return _json_object_hook($o);
|
|
}
|
|
return $o;
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
$at--;
|
|
decode_error(", or } expected while parsing object/hash");
|
|
}
|
|
|
|
|
|
sub bareKey { # doesn't strictly follow Standard ECMA-262 3rd Edition
|
|
my $key;
|
|
while($ch =~ /[^\x00-\x23\x25-\x2F\x3A-\x40\x5B-\x5E\x60\x7B-\x7F]/){
|
|
$key .= $ch;
|
|
next_chr();
|
|
}
|
|
return $key;
|
|
}
|
|
|
|
|
|
sub word {
|
|
my $word = substr($text,$at-1,4);
|
|
|
|
if($word eq 'true'){
|
|
$at += 3;
|
|
next_chr;
|
|
return $GT::JSON::PP::true;
|
|
}
|
|
elsif($word eq 'null'){
|
|
$at += 3;
|
|
next_chr;
|
|
return undef;
|
|
}
|
|
elsif($word eq 'fals'){
|
|
$at += 3;
|
|
if(substr($text,$at,1) eq 'e'){
|
|
$at++;
|
|
next_chr;
|
|
return $GT::JSON::PP::false;
|
|
}
|
|
}
|
|
|
|
$at--; # for decode_error report
|
|
|
|
decode_error("'null' expected") if ($word =~ /^n/);
|
|
decode_error("'true' expected") if ($word =~ /^t/);
|
|
decode_error("'false' expected") if ($word =~ /^f/);
|
|
decode_error("malformed JSON string, neither array, object, number, string or atom");
|
|
}
|
|
|
|
|
|
sub number {
|
|
my $n = '';
|
|
my $v;
|
|
|
|
# According to RFC4627, hex or oct digts are invalid.
|
|
if($ch eq '0'){
|
|
my $peek = substr($text,$at,1);
|
|
my $hex = $peek =~ /[xX]/; # 0 or 1
|
|
|
|
if($hex){
|
|
decode_error("malformed number (leading zero must not be followed by another digit)");
|
|
($n) = ( substr($text, $at+1) =~ /^([0-9a-fA-F]+)/);
|
|
}
|
|
else{ # oct
|
|
($n) = ( substr($text, $at) =~ /^([0-7]+)/);
|
|
if (defined $n and length $n > 1) {
|
|
decode_error("malformed number (leading zero must not be followed by another digit)");
|
|
}
|
|
}
|
|
|
|
if(defined $n and length($n)){
|
|
if (!$hex and length($n) == 1) {
|
|
decode_error("malformed number (leading zero must not be followed by another digit)");
|
|
}
|
|
$at += length($n) + $hex;
|
|
next_chr;
|
|
return $hex ? hex($n) : oct($n);
|
|
}
|
|
}
|
|
|
|
if($ch eq '-'){
|
|
$n = '-';
|
|
next_chr;
|
|
if (!defined $ch or $ch !~ /\d/) {
|
|
decode_error("malformed number (no digits after initial minus)");
|
|
}
|
|
}
|
|
|
|
while(defined $ch and $ch =~ /\d/){
|
|
$n .= $ch;
|
|
next_chr;
|
|
}
|
|
|
|
if(defined $ch and $ch eq '.'){
|
|
$n .= '.';
|
|
|
|
next_chr;
|
|
if (!defined $ch or $ch !~ /\d/) {
|
|
decode_error("malformed number (no digits after decimal point)");
|
|
}
|
|
else {
|
|
$n .= $ch;
|
|
}
|
|
|
|
while(defined(next_chr) and $ch =~ /\d/){
|
|
$n .= $ch;
|
|
}
|
|
}
|
|
|
|
if(defined $ch and ($ch eq 'e' or $ch eq 'E')){
|
|
$n .= $ch;
|
|
next_chr;
|
|
|
|
if(defined($ch) and ($ch eq '+' or $ch eq '-')){
|
|
$n .= $ch;
|
|
next_chr;
|
|
if (!defined $ch or $ch =~ /\D/) {
|
|
decode_error("malformed number (no digits after exp sign)");
|
|
}
|
|
$n .= $ch;
|
|
}
|
|
elsif(defined($ch) and $ch =~ /\d/){
|
|
$n .= $ch;
|
|
}
|
|
else {
|
|
decode_error("malformed number (no digits after exp sign)");
|
|
}
|
|
|
|
while(defined(next_chr) and $ch =~ /\d/){
|
|
$n .= $ch;
|
|
}
|
|
|
|
}
|
|
|
|
$v .= $n;
|
|
|
|
if ($v !~ /[.eE]/ and length $v > $max_intsize) {
|
|
if ($allow_bigint) { # from Adam Sussman
|
|
require Math::BigInt;
|
|
return Math::BigInt->new($v);
|
|
}
|
|
else {
|
|
return "$v";
|
|
}
|
|
}
|
|
elsif ($allow_bigint) {
|
|
require Math::BigFloat;
|
|
return Math::BigFloat->new($v);
|
|
}
|
|
|
|
return 0+$v;
|
|
}
|
|
|
|
|
|
sub is_valid_utf8 {
|
|
|
|
$utf8_len = $_[0] =~ /[\x00-\x7F]/ ? 1
|
|
: $_[0] =~ /[\xC2-\xDF]/ ? 2
|
|
: $_[0] =~ /[\xE0-\xEF]/ ? 3
|
|
: $_[0] =~ /[\xF0-\xF4]/ ? 4
|
|
: 0
|
|
;
|
|
|
|
return unless $utf8_len;
|
|
|
|
my $is_valid_utf8 = substr($text, $at - 1, $utf8_len);
|
|
|
|
return ( $is_valid_utf8 =~ /^(?:
|
|
[\x00-\x7F]
|
|
|[\xC2-\xDF][\x80-\xBF]
|
|
|[\xE0][\xA0-\xBF][\x80-\xBF]
|
|
|[\xE1-\xEC][\x80-\xBF][\x80-\xBF]
|
|
|[\xED][\x80-\x9F][\x80-\xBF]
|
|
|[\xEE-\xEF][\x80-\xBF][\x80-\xBF]
|
|
|[\xF0][\x90-\xBF][\x80-\xBF][\x80-\xBF]
|
|
|[\xF1-\xF3][\x80-\xBF][\x80-\xBF][\x80-\xBF]
|
|
|[\xF4][\x80-\x8F][\x80-\xBF][\x80-\xBF]
|
|
)$/x ) ? $is_valid_utf8 : '';
|
|
}
|
|
|
|
|
|
sub decode_error {
|
|
my $error = shift;
|
|
my $no_rep = shift;
|
|
my $str = defined $text ? substr($text, $at) : '';
|
|
my $mess = '';
|
|
my $type = $] >= 5.008 ? 'U*'
|
|
: $] < 5.006 ? 'C*'
|
|
: utf8::is_utf8( $str ) ? 'U*' # 5.6
|
|
: 'C*'
|
|
;
|
|
|
|
for my $c ( unpack( $type, $str ) ) { # emulate pv_uni_display() ?
|
|
$mess .= $c == 0x07 ? '\a'
|
|
: $c == 0x09 ? '\t'
|
|
: $c == 0x0a ? '\n'
|
|
: $c == 0x0d ? '\r'
|
|
: $c == 0x0c ? '\f'
|
|
: $c < 0x20 ? sprintf('\x{%x}', $c)
|
|
: $c < 0x80 ? chr($c)
|
|
: sprintf('\x{%x}', $c)
|
|
;
|
|
if ( length $mess >= 20 ) {
|
|
$mess .= '...';
|
|
last;
|
|
}
|
|
}
|
|
|
|
unless ( length $mess ) {
|
|
$mess = '(end of string)';
|
|
}
|
|
|
|
Carp::croak (
|
|
$no_rep ? "$error" : "$error, at character offset $at [\"$mess\"]"
|
|
);
|
|
}
|
|
|
|
|
|
sub _json_object_hook {
|
|
my $o = $_[0];
|
|
my @ks = keys %{$o};
|
|
|
|
if ( $cb_sk_object and @ks == 1 and exists $cb_sk_object->{ $ks[0] } and ref $cb_sk_object->{ $ks[0] } ) {
|
|
my @val = $cb_sk_object->{ $ks[0] }->( $o->{$ks[0]} );
|
|
if (@val == 1) {
|
|
return $val[0];
|
|
}
|
|
}
|
|
|
|
my @val = $cb_object->($o) if ($cb_object);
|
|
if (@val == 0 or @val > 1) {
|
|
return $o;
|
|
}
|
|
else {
|
|
return $val[0];
|
|
}
|
|
}
|
|
|
|
|
|
sub PP_decode_box {
|
|
{
|
|
text => $text,
|
|
at => $at,
|
|
ch => $ch,
|
|
len => $len,
|
|
depth => $depth,
|
|
encoding => $encoding,
|
|
is_valid_utf8 => $is_valid_utf8,
|
|
};
|
|
}
|
|
|
|
} # PARSE
|
|
|
|
|
|
sub _decode_surrogates { # from perlunicode
|
|
my $uni = 0x10000 + (hex($_[0]) - 0xD800) * 0x400 + (hex($_[1]) - 0xDC00);
|
|
my $un = pack('U*', $uni);
|
|
utf8::encode( $un );
|
|
return $un;
|
|
}
|
|
|
|
|
|
sub _decode_unicode {
|
|
my $un = pack('U', hex shift);
|
|
utf8::encode( $un );
|
|
return $un;
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
###############################
|
|
# Utilities
|
|
#
|
|
|
|
BEGIN {
|
|
eval 'require Scalar::Util';
|
|
unless($@){
|
|
*GT::JSON::PP::blessed = \&Scalar::Util::blessed;
|
|
}
|
|
else{ # This code is from Sclar::Util.
|
|
# warn $@;
|
|
eval 'sub UNIVERSAL::a_sub_not_likely_to_be_here { ref($_[0]) }';
|
|
*GT::JSON::PP::blessed = sub {
|
|
local($@, $SIG{__DIE__}, $SIG{__WARN__});
|
|
ref($_[0]) ? eval { $_[0]->a_sub_not_likely_to_be_here } : undef;
|
|
};
|
|
}
|
|
}
|
|
|
|
|
|
# shamely copied and modified from JSON::XS code.
|
|
|
|
$GT::JSON::PP::true = do { bless \(my $dummy = 1), "GT::JSON::PP::Boolean" };
|
|
$GT::JSON::PP::false = do { bless \(my $dummy = 0), "GT::JSON::PP::Boolean" };
|
|
|
|
sub is_bool { defined $_[0] and UNIVERSAL::isa($_[0], "GT::JSON::PP::Boolean"); }
|
|
|
|
sub true { $GT::JSON::PP::true }
|
|
sub false { $GT::JSON::PP::false }
|
|
sub null { undef; }
|
|
|
|
###############################
|
|
|
|
package GT::JSON::PP::Boolean;
|
|
|
|
|
|
use overload (
|
|
"0+" => sub { ${$_[0]} },
|
|
"++" => sub { $_[0] = ${$_[0]} + 1 },
|
|
"--" => sub { $_[0] = ${$_[0]} - 1 },
|
|
fallback => 1,
|
|
);
|
|
|
|
|
|
###############################
|
|
|
|
package GT::JSON::PP::IncrParser;
|
|
|
|
use strict;
|
|
|
|
use constant INCR_M_WS => 0; # initial whitespace skipping
|
|
use constant INCR_M_STR => 1; # inside string
|
|
use constant INCR_M_BS => 2; # inside backslash
|
|
use constant INCR_M_JSON => 3; # outside anything, count nesting
|
|
|
|
$GT::JSON::PP::IncrParser::VERSION = '1.01';
|
|
|
|
my $unpack_format = $] < 5.006 ? 'C*' : 'U*';
|
|
|
|
sub new {
|
|
my ( $class ) = @_;
|
|
|
|
bless {
|
|
incr_nest => 0,
|
|
incr_text => undef,
|
|
incr_parsing => 0,
|
|
incr_p => 0,
|
|
|
|
}, $class;
|
|
}
|
|
|
|
|
|
sub incr_parse {
|
|
my ( $self, $coder, $text ) = @_;
|
|
|
|
$self->{incr_text} = '' unless ( defined $self->{incr_text} );
|
|
|
|
if ( defined $text ) {
|
|
if ( utf8::is_utf8( $text ) and !utf8::is_utf8( $self->{incr_text} ) ) {
|
|
utf8::upgrade( $self->{incr_text} ) ;
|
|
utf8::decode( $self->{incr_text} ) ;
|
|
}
|
|
$self->{incr_text} .= $text;
|
|
}
|
|
|
|
|
|
my $max_size = $coder->get_max_size;
|
|
|
|
if ( defined wantarray ) {
|
|
|
|
$self->{incr_mode} = INCR_M_WS;
|
|
|
|
if ( wantarray ) {
|
|
my @ret;
|
|
|
|
$self->{incr_parsing} = 1;
|
|
|
|
do {
|
|
push @ret, $self->_incr_parse( $coder, $self->{incr_text} );
|
|
|
|
unless ( !$self->{incr_nest} and $self->{incr_mode} == INCR_M_JSON ) {
|
|
$self->{incr_mode} = INCR_M_WS;
|
|
}
|
|
|
|
} until ( !$self->{incr_text} );
|
|
|
|
$self->{incr_parsing} = 0;
|
|
|
|
return @ret;
|
|
}
|
|
else { # in scalar context
|
|
$self->{incr_parsing} = 1;
|
|
my $obj = $self->_incr_parse( $coder, $self->{incr_text} );
|
|
$self->{incr_parsing} = 0;
|
|
return $obj;
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
sub _incr_parse {
|
|
my ( $self, $coder, $text, $skip ) = @_;
|
|
my $p = $self->{incr_p};
|
|
my $restore = $p;
|
|
|
|
my @obj;
|
|
my $len = length $text;
|
|
|
|
if ( $self->{incr_mode} == INCR_M_WS ) {
|
|
while ( $len > $p ) {
|
|
my $s = substr( $text, $p, 1 );
|
|
$p++ and next if ( 0x20 >= unpack($unpack_format, $s) );
|
|
$self->{incr_mode} = INCR_M_JSON;
|
|
last;
|
|
}
|
|
}
|
|
|
|
while ( $len > $p ) {
|
|
my $s = substr( $text, $p++, 1 );
|
|
|
|
if ( $s eq '"' ) {
|
|
if ( $self->{incr_mode} != INCR_M_STR ) {
|
|
$self->{incr_mode} = INCR_M_STR;
|
|
}
|
|
else {
|
|
$self->{incr_mode} = INCR_M_JSON;
|
|
unless ( $self->{incr_nest} ) {
|
|
last;
|
|
}
|
|
}
|
|
}
|
|
|
|
if ( $self->{incr_mode} == INCR_M_JSON ) {
|
|
|
|
if ( $s eq '[' or $s eq '{' ) {
|
|
if ( ++$self->{incr_nest} > $coder->get_max_depth ) {
|
|
Carp::croak('json text or perl structure exceeds maximum nesting level (max_depth set too low?)');
|
|
}
|
|
}
|
|
elsif ( $s eq ']' or $s eq '}' ) {
|
|
last if ( --$self->{incr_nest} <= 0 );
|
|
}
|
|
}
|
|
|
|
}
|
|
|
|
$self->{incr_p} = $p;
|
|
|
|
return if ( $self->{incr_mode} == INCR_M_JSON and $self->{incr_nest} > 0 );
|
|
|
|
return unless ( length substr( $self->{incr_text}, 0, $p ) );
|
|
|
|
local $Carp::CarpLevel = 2;
|
|
|
|
$self->{incr_p} = $restore;
|
|
$self->{incr_c} = $p;
|
|
|
|
my ( $obj, $tail ) = $coder->decode_prefix( substr( $self->{incr_text}, 0, $p ) );
|
|
|
|
$self->{incr_text} = substr( $self->{incr_text}, $p );
|
|
$self->{incr_p} = 0;
|
|
|
|
return $obj;
|
|
}
|
|
|
|
|
|
sub incr_text {
|
|
if ( $_[0]->{incr_parsing} ) {
|
|
Carp::croak("incr_text can not be called when the incremental parser already started parsing");
|
|
}
|
|
$_[0]->{incr_text};
|
|
}
|
|
|
|
|
|
sub incr_skip {
|
|
my $self = shift;
|
|
$self->{incr_text} = substr( $self->{incr_text}, $self->{incr_c} );
|
|
$self->{incr_p} = 0;
|
|
}
|
|
|
|
|
|
sub incr_reset {
|
|
my $self = shift;
|
|
$self->{incr_text} = undef;
|
|
$self->{incr_p} = 0;
|
|
$self->{incr_mode} = 0;
|
|
$self->{incr_nest} = 0;
|
|
$self->{incr_parsing} = 0;
|
|
}
|
|
|
|
###############################
|
|
|
|
|
|
1;
|
|
__END__
|
|
=pod
|
|
|
|
=head1 NAME
|
|
|
|
GT::JSON::PP - JSON::XS compatible pure-Perl module.
|
|
|
|
=head1 SYNOPSIS
|
|
|
|
use GT::JSON::PP;
|
|
|
|
# exported functions, they croak on error
|
|
# and expect/generate UTF-8
|
|
|
|
$utf8_encoded_json_text = encode_json $perl_hash_or_arrayref;
|
|
$perl_hash_or_arrayref = decode_json $utf8_encoded_json_text;
|
|
|
|
# OO-interface
|
|
|
|
$coder = GT::JSON::PP->new->ascii->pretty->allow_nonref;
|
|
$pretty_printed_unencoded = $coder->encode ($perl_scalar);
|
|
$perl_scalar = $coder->decode ($unicode_json_text);
|
|
|
|
# Note that GT::JSON version 2.0 and above will automatically use
|
|
# JSON::XS or GT::JSON::PP, so you should be able to just:
|
|
|
|
use GT::JSON;
|
|
|
|
=head1 DESCRIPTION
|
|
|
|
This module is L<JSON::XS> compatible pure Perl module.
|
|
(Perl 5.8 or later is recommended)
|
|
|
|
JSON::XS is the fastest and most proper JSON module on CPAN.
|
|
It is written by Marc Lehmann in C, so must be compiled and
|
|
installed in the used environment.
|
|
|
|
GT::JSON::PP is a pure-Perl module and has compatibility to JSON::XS.
|
|
|
|
|
|
=head2 FEATURES
|
|
|
|
=over
|
|
|
|
=item * correct unicode handling
|
|
|
|
This module knows how to handle Unicode (depending on Perl version).
|
|
|
|
See to L<JSON::XS/A FEW NOTES ON UNICODE AND PERL> and L<UNICODE HANDLING ON PERLS>.
|
|
|
|
|
|
=item * round-trip integrity
|
|
|
|
When you serialise a perl data structure using only datatypes supported by JSON,
|
|
the deserialised data structure is identical on the Perl level.
|
|
(e.g. the string "2.0" doesn't suddenly become "2" just because it looks like a number).
|
|
|
|
=item * strict checking of JSON correctness
|
|
|
|
There is no guessing, no generating of illegal JSON texts by default,
|
|
and only JSON is accepted as input by default (the latter is a security feature).
|
|
But when some options are set, loose chcking features are available.
|
|
|
|
=back
|
|
|
|
=head1 FUNCTIONS
|
|
|
|
Basically, check to L<GT::JSON> or L<JSON::XS>.
|
|
|
|
=head2 encode_json
|
|
|
|
$json_text = encode_json $perl_scalar
|
|
|
|
=head2 decode_json
|
|
|
|
$perl_scalar = decode_json $json_text
|
|
|
|
=head2 GT::JSON::PP::true
|
|
|
|
Returns JSON true value which is blessed object.
|
|
It C<isa> GT::JSON::PP::Boolean object.
|
|
|
|
=head2 GT::JSON::PP::false
|
|
|
|
Returns JSON false value which is blessed object.
|
|
It C<isa> GT::JSON::PP::Boolean object.
|
|
|
|
=head2 GT::JSON::PP::null
|
|
|
|
Returns C<undef>.
|
|
|
|
=head1 METHODS
|
|
|
|
Basically, check to L<GT::JSON> or L<JSON::XS>.
|
|
|
|
=head2 new
|
|
|
|
$json = new GT::JSON::PP
|
|
|
|
Rturns a new GT::JSON::PP object that can be used to de/encode JSON
|
|
strings.
|
|
|
|
=head2 ascii
|
|
|
|
$json = $json->ascii([$enable])
|
|
|
|
$enabled = $json->get_ascii
|
|
|
|
If $enable is true (or missing), then the encode method will not generate characters outside
|
|
the code range 0..127. Any Unicode characters outside that range will be escaped using either
|
|
a single \uXXXX or a double \uHHHH\uLLLLL escape sequence, as per RFC4627.
|
|
(See to L<JSON::XS/OBJECT-ORIENTED INTERFACE>).
|
|
|
|
In Perl 5.005, there is no character having high value (more than 255).
|
|
See to L<UNICODE HANDLING ON PERLS>.
|
|
|
|
If $enable is false, then the encode method will not escape Unicode characters unless
|
|
required by the JSON syntax or other flags. This results in a faster and more compact format.
|
|
|
|
GT::JSON::PP->new->ascii(1)->encode([chr 0x10401])
|
|
=> ["\ud801\udc01"]
|
|
|
|
=head2 latin1
|
|
|
|
$json = $json->latin1([$enable])
|
|
|
|
$enabled = $json->get_latin1
|
|
|
|
If $enable is true (or missing), then the encode method will encode the resulting JSON
|
|
text as latin1 (or iso-8859-1), escaping any characters outside the code range 0..255.
|
|
|
|
If $enable is false, then the encode method will not escape Unicode characters
|
|
unless required by the JSON syntax or other flags.
|
|
|
|
JSON::XS->new->latin1->encode (["\x{89}\x{abc}"]
|
|
=> ["\x{89}\\u0abc"] # (perl syntax, U+abc escaped, U+89 not)
|
|
|
|
See to L<UNICODE HANDLING ON PERLS>.
|
|
|
|
=head2 utf8
|
|
|
|
$json = $json->utf8([$enable])
|
|
|
|
$enabled = $json->get_utf8
|
|
|
|
If $enable is true (or missing), then the encode method will encode the JSON result
|
|
into UTF-8, as required by many protocols, while the decode method expects to be handled
|
|
an UTF-8-encoded string. Please note that UTF-8-encoded strings do not contain any
|
|
characters outside the range 0..255, they are thus useful for bytewise/binary I/O.
|
|
|
|
(In Perl 5.005, any character outside the range 0..255 does not exist.
|
|
See to L<UNICODE HANDLING ON PERLS>.)
|
|
|
|
In future versions, enabling this option might enable autodetection of the UTF-16 and UTF-32
|
|
encoding families, as described in RFC4627.
|
|
|
|
If $enable is false, then the encode method will return the JSON string as a (non-encoded)
|
|
Unicode string, while decode expects thus a Unicode string. Any decoding or encoding
|
|
(e.g. to UTF-8 or UTF-16) needs to be done yourself, e.g. using the Encode module.
|
|
|
|
Example, output UTF-16BE-encoded JSON:
|
|
|
|
use Encode;
|
|
$jsontext = encode "UTF-16BE", JSON::XS->new->encode ($object);
|
|
|
|
Example, decode UTF-32LE-encoded JSON:
|
|
|
|
use Encode;
|
|
$object = JSON::XS->new->decode (decode "UTF-32LE", $jsontext);
|
|
|
|
|
|
=head2 pretty
|
|
|
|
$json = $json->pretty([$enable])
|
|
|
|
This enables (or disables) all of the C<indent>, C<space_before> and
|
|
C<space_after> flags in one call to generate the most readable
|
|
(or most compact) form possible.
|
|
|
|
=head2 indent
|
|
|
|
$json = $json->indent([$enable])
|
|
|
|
$enabled = $json->get_indent
|
|
|
|
The default indent space lenght is three.
|
|
You can use C<indent_length> to change the length.
|
|
|
|
=head2 space_before
|
|
|
|
$json = $json->space_before([$enable])
|
|
|
|
$enabled = $json->get_space_before
|
|
|
|
=head2 space_after
|
|
|
|
$json = $json->space_after([$enable])
|
|
|
|
$enabled = $json->get_space_after
|
|
|
|
=head2 relaxed
|
|
|
|
$json = $json->relaxed([$enable])
|
|
|
|
$enabled = $json->get_relaxed
|
|
|
|
=head2 canonical
|
|
|
|
$json = $json->canonical([$enable])
|
|
|
|
$enabled = $json->get_canonical
|
|
|
|
If you want your own sorting routine, you can give a code referece
|
|
or a subroutine name to C<sort_by>. See to C<GT::JSON::PP OWN METHODS>.
|
|
|
|
=head2 allow_nonref
|
|
|
|
$json = $json->allow_nonref([$enable])
|
|
|
|
$enabled = $json->get_allow_nonref
|
|
|
|
=head2 allow_unknown
|
|
|
|
$json = $json->allow_unknown ([$enable])
|
|
|
|
$enabled = $json->get_allow_unknown
|
|
|
|
=head2 allow_blessed
|
|
|
|
$json = $json->allow_blessed([$enable])
|
|
|
|
$enabled = $json->get_allow_blessed
|
|
|
|
=head2 convert_blessed
|
|
|
|
$json = $json->convert_blessed([$enable])
|
|
|
|
$enabled = $json->get_convert_blessed
|
|
|
|
=head2 filter_json_object
|
|
|
|
$json = $json->filter_json_object([$coderef])
|
|
|
|
=head2 filter_json_single_key_object
|
|
|
|
$json = $json->filter_json_single_key_object($key [=> $coderef])
|
|
|
|
=head2 shrink
|
|
|
|
$json = $json->shrink([$enable])
|
|
|
|
$enabled = $json->get_shrink
|
|
|
|
In JSON::XS, this flag resizes strings generated by either
|
|
C<encode> or C<decode> to their minimum size possible.
|
|
It will also try to downgrade any strings to octet-form if possible.
|
|
|
|
In GT::JSON::PP, it is noop about resizing strings but tries
|
|
C<utf8::downgrade> to the returned string by C<encode>.
|
|
See to L<utf8>.
|
|
|
|
See to L<JSON::XS/OBJECT-ORIENTED INTERFACE>
|
|
|
|
=head2 max_depth
|
|
|
|
$json = $json->max_depth([$maximum_nesting_depth])
|
|
|
|
$max_depth = $json->get_max_depth
|
|
|
|
Sets the maximum nesting level (default C<512>) accepted while encoding
|
|
or decoding. If a higher nesting level is detected in JSON text or a Perl
|
|
data structure, then the encoder and decoder will stop and croak at that
|
|
point.
|
|
|
|
Nesting level is defined by number of hash- or arrayrefs that the encoder
|
|
needs to traverse to reach a given point or the number of C<{> or C<[>
|
|
characters without their matching closing parenthesis crossed to reach a
|
|
given character in a string.
|
|
|
|
If no argument is given, the highest possible setting will be used, which
|
|
is rarely useful.
|
|
|
|
See L<JSON::XS/SSECURITY CONSIDERATIONS> for more info on why this is useful.
|
|
|
|
When a large value (100 or more) was set and it de/encodes a deep nested object/text,
|
|
it may raise a warning 'Deep recursion on subroutin' at the perl runtime phase.
|
|
|
|
=head2 max_size
|
|
|
|
$json = $json->max_size([$maximum_string_size])
|
|
|
|
$max_size = $json->get_max_size
|
|
|
|
Set the maximum length a JSON text may have (in bytes) where decoding is
|
|
being attempted. The default is C<0>, meaning no limit. When C<decode>
|
|
is called on a string that is longer then this many bytes, it will not
|
|
attempt to decode the string but throw an exception. This setting has no
|
|
effect on C<encode> (yet).
|
|
|
|
If no argument is given, the limit check will be deactivated (same as when
|
|
C<0> is specified).
|
|
|
|
See L<JSON::XS/SSECURITY CONSIDERATIONS> for more info on why this is useful.
|
|
|
|
=head2 encode
|
|
|
|
$json_text = $json->encode($perl_scalar)
|
|
|
|
=head2 decode
|
|
|
|
$perl_scalar = $json->decode($json_text)
|
|
|
|
=head2 decode_prefix
|
|
|
|
($perl_scalar, $characters) = $json->decode_prefix($json_text)
|
|
|
|
|
|
|
|
=head1 INCREMENTAL PARSING
|
|
|
|
In JSON::XS 2.2, incremental parsing feature of JSON
|
|
texts was experimentally implemented.
|
|
Please check to L<JSON::XS/INCREMENTAL PARSING>.
|
|
|
|
=over 4
|
|
|
|
=item [void, scalar or list context] = $json->incr_parse ([$string])
|
|
|
|
This is the central parsing function. It can both append new text and
|
|
extract objects from the stream accumulated so far (both of these
|
|
functions are optional).
|
|
|
|
If C<$string> is given, then this string is appended to the already
|
|
existing JSON fragment stored in the C<$json> object.
|
|
|
|
After that, if the function is called in void context, it will simply
|
|
return without doing anything further. This can be used to add more text
|
|
in as many chunks as you want.
|
|
|
|
If the method is called in scalar context, then it will try to extract
|
|
exactly I<one> JSON object. If that is successful, it will return this
|
|
object, otherwise it will return C<undef>. If there is a parse error,
|
|
this method will croak just as C<decode> would do (one can then use
|
|
C<incr_skip> to skip the errornous part). This is the most common way of
|
|
using the method.
|
|
|
|
And finally, in list context, it will try to extract as many objects
|
|
from the stream as it can find and return them, or the empty list
|
|
otherwise. For this to work, there must be no separators between the JSON
|
|
objects or arrays, instead they must be concatenated back-to-back. If
|
|
an error occurs, an exception will be raised as in the scalar context
|
|
case. Note that in this case, any previously-parsed JSON texts will be
|
|
lost.
|
|
|
|
=item $lvalue_string = $json->incr_text
|
|
|
|
This method returns the currently stored JSON fragment as an lvalue, that
|
|
is, you can manipulate it. This I<only> works when a preceding call to
|
|
C<incr_parse> in I<scalar context> successfully returned an object. Under
|
|
all other circumstances you must not call this function (I mean it.
|
|
although in simple tests it might actually work, it I<will> fail under
|
|
real world conditions). As a special exception, you can also call this
|
|
method before having parsed anything.
|
|
|
|
This function is useful in two cases: a) finding the trailing text after a
|
|
JSON object or b) parsing multiple JSON objects separated by non-JSON text
|
|
(such as commas).
|
|
|
|
In Perl 5.005, C<lvalue> attribute is not available.
|
|
You must write codes like the below:
|
|
|
|
$string = $json->incr_text;
|
|
$string =~ s/\s*,\s*//;
|
|
$json->incr_text( $string );
|
|
|
|
=item $json->incr_skip
|
|
|
|
This will reset the state of the incremental parser and will remove the
|
|
parsed text from the input buffer. This is useful after C<incr_parse>
|
|
died, in which case the input buffer and incremental parser state is left
|
|
unchanged, to skip the text parsed so far and to reset the parse state.
|
|
|
|
=back
|
|
|
|
|
|
|
|
=head1 GT::JSON::PP OWN METHODS
|
|
|
|
=head2 allow_singlequote
|
|
|
|
$json = $json->allow_singlequote([$enable])
|
|
|
|
If C<$enable> is true (or missing), then C<decode> will accept
|
|
JSON strings quoted by single quotations that are invalid JSON
|
|
format.
|
|
|
|
$json->allow_singlequote->decode({"foo":'bar'});
|
|
$json->allow_singlequote->decode({'foo':"bar"});
|
|
$json->allow_singlequote->decode({'foo':'bar'});
|
|
|
|
As same as the C<relaxed> option, this option may be used to parse
|
|
application-specific files written by humans.
|
|
|
|
|
|
=head2 allow_barekey
|
|
|
|
$json = $json->allow_barekey([$enable])
|
|
|
|
If C<$enable> is true (or missing), then C<decode> will accept
|
|
bare keys of JSON object that are invalid JSON format.
|
|
|
|
As same as the C<relaxed> option, this option may be used to parse
|
|
application-specific files written by humans.
|
|
|
|
$json->allow_barekey->decode('{foo:"bar"}');
|
|
|
|
=head2 allow_bignum
|
|
|
|
$json = $json->allow_bignum([$enable])
|
|
|
|
If C<$enable> is true (or missing), then C<decode> will convert
|
|
the big integer Perl cannot handle as integer into a L<Math::BigInt>
|
|
object and convert a floating number (any) into a L<Math::BigFloat>.
|
|
|
|
On the contary, C<encode> converts C<Math::BigInt> objects and C<Math::BigFloat>
|
|
objects into JSON numbers with C<allow_blessed> enable.
|
|
|
|
$json->allow_nonref->allow_blessed->allow_bignum;
|
|
$bigfloat = $json->decode('2.000000000000000000000000001');
|
|
print $json->encode($bigfloat);
|
|
# => 2.000000000000000000000000001
|
|
|
|
See to L<JSON::XS/MAPPING> aboout the normal conversion of JSON number.
|
|
|
|
=head2 loose
|
|
|
|
$json = $json->loose([$enable])
|
|
|
|
The unescaped [\x00-\x1f\x22\x2f\x5c] strings are invalid in JSON strings
|
|
and the module doesn't allow to C<decode> to these (except for \x2f).
|
|
If C<$enable> is true (or missing), then C<decode> will accept these
|
|
unescaped strings.
|
|
|
|
$json->loose->decode(qq|["abc
|
|
def"]|);
|
|
|
|
See L<JSON::XS/SSECURITY CONSIDERATIONS>.
|
|
|
|
=head2 escape_slash
|
|
|
|
$json = $json->escape_slash([$enable])
|
|
|
|
According to JSON Grammar, I<slash> (U+002F) is escaped. But default
|
|
GT::JSON::PP (as same as JSON::XS) encodes strings without escaping slash.
|
|
|
|
If C<$enable> is true (or missing), then C<encode> will escape slashes.
|
|
|
|
=head2 (OBSOLETED)as_nonblessed
|
|
|
|
$json = $json->as_nonblessed
|
|
|
|
(OBSOLETED) If C<$enable> is true (or missing), then C<encode> will convert
|
|
a blessed hash reference or a blessed array reference (contains
|
|
other blessed references) into JSON members and arrays.
|
|
|
|
This feature is effective only when C<allow_blessed> is enable.
|
|
|
|
=head2 indent_length
|
|
|
|
$json = $json->indent_length($length)
|
|
|
|
JSON::XS indent space length is 3 and cannot be changed.
|
|
GT::JSON::PP set the indent space length with the given $length.
|
|
The default is 3. The acceptable range is 0 to 15.
|
|
|
|
=head2 sort_by
|
|
|
|
$json = $json->sort_by($function_name)
|
|
$json = $json->sort_by($subroutine_ref)
|
|
|
|
If $function_name or $subroutine_ref are set, its sort routine are used
|
|
in encoding JSON objects.
|
|
|
|
$js = $pc->sort_by(sub { $GT::JSON::PP::a cmp $GT::JSON::PP::b })->encode($obj);
|
|
# is($js, q|{"a":1,"b":2,"c":3,"d":4,"e":5,"f":6,"g":7,"h":8,"i":9}|);
|
|
|
|
$js = $pc->sort_by('own_sort')->encode($obj);
|
|
# is($js, q|{"a":1,"b":2,"c":3,"d":4,"e":5,"f":6,"g":7,"h":8,"i":9}|);
|
|
|
|
sub GT::JSON::PP::own_sort { $GT::JSON::PP::a cmp $GT::JSON::PP::b }
|
|
|
|
As the sorting routine runs in the GT::JSON::PP scope, the given
|
|
subroutine name and the special variables C<$a>, C<$b> will begin
|
|
'GT::JSON::PP::'.
|
|
|
|
If $integer is set, then the effect is same as C<canonical> on.
|
|
|
|
=head1 INTERNAL
|
|
|
|
For developers.
|
|
|
|
=over
|
|
|
|
=item PP_encode_box
|
|
|
|
Returns
|
|
|
|
{
|
|
depth => $depth,
|
|
indent_count => $indent_count,
|
|
}
|
|
|
|
|
|
=item PP_decode_box
|
|
|
|
Returns
|
|
|
|
{
|
|
text => $text,
|
|
at => $at,
|
|
ch => $ch,
|
|
len => $len,
|
|
depth => $depth,
|
|
encoding => $encoding,
|
|
is_valid_utf8 => $is_valid_utf8,
|
|
};
|
|
|
|
=back
|
|
|
|
=head1 MAPPING
|
|
|
|
See to L<JSON::XS/MAPPING>.
|
|
|
|
|
|
=head1 UNICODE HANDLING ON PERLS
|
|
|
|
If you do not know about Unicode on Perl well,
|
|
please check L<JSON::XS/A FEW NOTES ON UNICODE AND PERL>.
|
|
|
|
=head2 Perl 5.8 and later
|
|
|
|
Perl can handle Unicode and the GT::JSON::PP de/encode methods also work properly.
|
|
|
|
$json->allow_nonref->encode(chr hex 3042);
|
|
$json->allow_nonref->encode(chr hex 12345);
|
|
|
|
Reuturns C<"\u3042"> and C<"\ud808\udf45"> respectively.
|
|
|
|
$json->allow_nonref->decode('"\u3042"');
|
|
$json->allow_nonref->decode('"\ud808\udf45"');
|
|
|
|
Returns UTF-8 encoded strings with UTF8 flag, regarded as C<U+3042> and C<U+12345>.
|
|
|
|
Note that the versions from Perl 5.8.0 to 5.8.2, Perl built-in C<join> was broken,
|
|
so GT::JSON::PP wraps the C<join> with a subroutine. Thus GT::JSON::PP works slow in the versions.
|
|
|
|
|
|
=head2 Perl 5.6
|
|
|
|
Perl can handle Unicode and the GT::JSON::PP de/encode methods also work.
|
|
|
|
=head2 Perl 5.005
|
|
|
|
Perl 5.005 is a byte sementics world -- all strings are sequences of bytes.
|
|
That means the unicode handling is not available.
|
|
|
|
In encoding,
|
|
|
|
$json->allow_nonref->encode(chr hex 3042); # hex 3042 is 12354.
|
|
$json->allow_nonref->encode(chr hex 12345); # hex 12345 is 74565.
|
|
|
|
Returns C<B> and C<E>, as C<chr> takes a value more than 255, it treats
|
|
as C<$value % 256>, so the above codes are equivalent to :
|
|
|
|
$json->allow_nonref->encode(chr 66);
|
|
$json->allow_nonref->encode(chr 69);
|
|
|
|
In decoding,
|
|
|
|
$json->decode('"\u00e3\u0081\u0082"');
|
|
|
|
The returned is a byte sequence C<0xE3 0x81 0x82> for UTF-8 encoded
|
|
japanese character (C<HIRAGANA LETTER A>).
|
|
And if it is represented in Unicode code point, C<U+3042>.
|
|
|
|
Next,
|
|
|
|
$json->decode('"\u3042"');
|
|
|
|
We ordinary expect the returned value is a Unicode character C<U+3042>.
|
|
But here is 5.005 world. This is C<0xE3 0x81 0x82>.
|
|
|
|
$json->decode('"\ud808\udf45"');
|
|
|
|
This is not a character C<U+12345> but bytes - C<0xf0 0x92 0x8d 0x85>.
|
|
|
|
|
|
=head1 TODO
|
|
|
|
=over
|
|
|
|
=item speed
|
|
|
|
=item memory saving
|
|
|
|
=back
|
|
|
|
|
|
=head1 SEE ALSO
|
|
|
|
Most of the document are copied and modified from JSON::XS doc.
|
|
|
|
L<JSON::XS>
|
|
|
|
RFC4627 (L<http://www.ietf.org/rfc/rfc4627.txt>)
|
|
|
|
=head1 AUTHOR
|
|
|
|
Makamaka Hannyaharamitu, E<lt>makamaka[at]cpan.orgE<gt>
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE
|
|
|
|
Copyright 2008 by Makamaka Hannyaharamitu
|
|
|
|
This library is free software; you can redistribute it and/or modify
|
|
it under the same terms as Perl itself.
|
|
|
|
=cut
|