—use
strict;
use
warnings;
package
JSON::Schema::Modern::Result;
# vim: set ts=8 sts=2 sw=2 tw=100 et :
# ABSTRACT: Contains the result of a JSON Schema evaluation
our
$VERSION
=
'0.607'
;
use
5.020;
use
Moo;
use
strictures 2;
no
autovivification
warn
=>
qw(fetch store exists delete)
;
use
if
"$]"
>= 5.022,
experimental
=>
're_strict'
;
no
if
"$]"
>= 5.031009,
feature
=>
'indirect'
;
no
if
"$]"
>= 5.033001,
feature
=>
'multidimensional'
;
no
if
"$]"
>= 5.033006,
feature
=>
'bareword_filehandles'
;
use
MooX::TypeTiny;
use
JSON::PP ();
use
Safe::Isa;
use
namespace::clean;
use
overload
'bool'
=>
sub
{
croak
'boolean overload is deprecated and could be removed anytime after 2026-02-01'
;
$_
[0]->valid;
},
'&'
=> \
&combine
,
'""'
=>
sub
{
$_
[0]->stringify },
fallback
=> 1;
has
valid
=> (
is
=>
'ro'
,
isa
=> Bool,
coerce
=> 1,
required
=> 1,
);
sub
result {
goto
\
&valid
}
# backcompat only
has
exception
=> (
is
=>
'ro'
,
isa
=> Bool,
lazy
=> 1,
default
=>
sub
(
$self
) { any {
$_
->exception or
$_
->error =~ /^EXCEPTION: / }
$self
->errors },
);
# turn hashrefs in _errors or _annotations into blessed objects
has
$_
.
's'
=> (
is
=>
'bare'
,
reader
=>
'__'
.
$_
.
's'
,
isa
=> ArrayRef[InstanceOf[
'JSON::Schema::Modern::'
.
ucfirst
]],
lazy
=> 1,
default
=>
do
{
my
$type
=
$_
;
sub
(
$self
) {
return
[]
if
not ((
$self
->{
'_'
.
$type
.
's'
}//[])->@*);
# E() and A() in ::Utilities returns an unblessed hashref, which is used to create a real object
# by its BUILDARGS sub
return
[
map
+((
'JSON::Schema::Modern::'
.
ucfirst
(
$type
))->new(
$_
)),
$self
->{
'_'
.
$type
.
's'
}->@* ];
};
},
)
foreach
qw(error annotation)
;
sub
errors {
$_
[0]->__errors->@* }
sub
error_count {
scalar
((
$_
[0]->{errors}//[])->@*) ||
scalar
((
$_
[0]->{_errors}//[])->@*) }
sub
annotations {
$_
[0]->__annotations->@* }
sub
annotation_count {
scalar
((
$_
[0]->{annotations}//[])->@*) ||
scalar
((
$_
[0]->{_annotations}//[])->@*) }
has
recommended_response
=> (
is
=>
'rw'
,
isa
=> Maybe[Tuple[PositiveInt, Str]],
lazy
=> 1,
default
=>
sub
(
$self
) {
return
if
not
$self
->errors;
for
my
$error
(
$self
->errors) {
my
$pe
=
$error
->recommended_response;
return
$pe
if
$pe
;
}
return
[ 500,
'Internal Server Error'
]
if
$self
->exception;
return
[ 400, (
$self
->errors)[0]->stringify ];
},
);
# strict_basic can only be used with draft2019-09.
has
output_format
=> (
is
=>
'rw'
,
isa
=> Enum(OUTPUT_FORMATS),
default
=>
'basic'
,
);
has
formatted_annotations
=> (
is
=>
'ro'
,
isa
=> Bool,
default
=> 1,
);
around
BUILDARGS
=>
sub
(
$orig
,
$class
,
@args
) {
my
$args
=
$class
->
$orig
(
@args
);
# set unblessed hashrefs aside, and defer creation of blessed objects until needed
$args
->{_errors} =
delete
$args
->{errors}
if
exists
$args
->{errors} and any { !blessed(
$_
) }
$args
->{errors}->@*;
$args
->{_annotations} =
delete
$args
->{annotations}
if
exists
$args
->{annotations} and any { !blessed(
$_
) }
$args
->{annotations}->@*;
return
$args
;
};
sub
BUILD (
$self
,
$args
) {
warn
'result is false but there are no errors'
if
not
$self
->valid and not
$self
->error_count;
$self
->{_errors} =
$args
->{_errors}
if
exists
$args
->{_errors};
$self
->{_annotations} =
$args
->{_annotations}
if
exists
$args
->{_annotations};
}
sub
format
(
$self
,
$style
,
$formatted_annotations
=
undef
) {
$formatted_annotations
//=
$self
->formatted_annotations;
if
(
$style
eq
'flag'
) {
return
+{
valid
=>
$self
->valid ? true : false };
}
elsif
(
$style
eq
'basic'
) {
return
+{
valid
=>
$self
->valid ? true : false,
$self
->valid
? (
$formatted_annotations
&&
$self
->annotation_count ? (
annotations
=> [
map
$_
->TO_JSON,
$self
->annotations ]) : ())
: (
errors
=> [
map
$_
->TO_JSON,
$self
->errors ]),
};
}
# note: strict_basic will NOT be supported after draft 2019-09!
elsif
(
$style
eq
'strict_basic'
) {
return
+{
valid
=> (
$self
->valid ? true : false),
$self
->valid
? (
$formatted_annotations
&&
$self
->annotation_count ? (
annotations
=> [
map
_map_uris(
$_
->TO_JSON),
$self
->annotations ]) : ())
: (
errors
=> [
map
_map_uris(
$_
->TO_JSON),
$self
->errors ]),
};
}
elsif
(
$style
eq
'terse'
) {
my
(
%instance_locations
,
%keyword_locations
);
my
@errors
=
grep
{
my
(
$keyword
,
$error
) = (
$_
->keyword,
$_
->error);
my
$keep
= 0+!!(
not
$keyword
or (
not
grep
$keyword
eq
$_
,
qw(allOf anyOf if then else dependentSchemas contains propertyNames)
and (
$keyword
ne
'oneOf'
or
$error
ne
'no subschemas are valid'
)
and (
$keyword
ne
'prefixItems'
or
$error
eq
'item not permitted'
)
and (
$keyword
ne
'items'
or
$error
eq
'item not permitted'
or
$error
eq
'additional item not permitted'
)
and (
$keyword
ne
'additionalItems'
or
$error
eq
'additional item not permitted'
)
and (not
grep
$keyword
eq
$_
,
qw(properties patternProperties)
or
$error
eq
'property not permitted'
)
and (
$keyword
ne
'additionalProperties'
or
$error
eq
'additional property not permitted'
))
and (
$keyword
ne
'dependentRequired'
or
$error
ne
'not all dependencies are satisfied'
)
);
++
$instance_locations
{
$_
->instance_location}
if
$keep
;
++
$keyword_locations
{
$_
->keyword_location}
if
$keep
;
$keep
;
}
$self
->errors;
die
'uh oh, have no errors left to report'
if
not
$self
->valid and not
@errors
;
return
+{
valid
=>
$self
->valid ? true : false,
$self
->valid
? (
$formatted_annotations
&&
$self
->annotation_count ? (
annotations
=> [
map
$_
->TO_JSON,
$self
->annotations ]) : ())
: (
errors
=> [
map
$_
->TO_JSON,
@errors
]),
};
}
elsif
(
$style
eq
'data_only'
) {
return
'valid'
if
not
$self
->error_count;
# Note: this output is going to be confusing when coming from a schema with a 'oneOf', 'not',
# etc. Perhaps generating the strings with indentation levels, as derived from a nested format,
# might be more readable.
return
join
(
"\n"
, uniq(
map
$_
->stringify,
$self
->errors));
}
# TODO: support detailed, verbose ?
die
'unsupported output format'
;
}
sub
count {
$_
[0]->valid ?
$_
[0]->annotation_count :
$_
[0]->error_count }
sub
combine (
$self
,
$other
,
$swap
) {
die
'wrong type for & operation'
if
not
$other
->
$_isa
(__PACKAGE__);
return
$self
if
refaddr(
$other
) == refaddr(
$self
);
return
ref
(
$self
)->new(
valid
=>
$self
->valid &&
$other
->valid,
annotations
=> [
$self
->annotations,
$other
->annotations,
],
errors
=> [
$self
->errors,
$other
->errors,
],
output_format
=>
$self
->output_format,
formatted_annotations
=>
$self
->formatted_annotations ||
$other
->formatted_annotations,
);
}
sub
stringify (
$self
) {
return
$self
->
format
(
'data_only'
);
}
sub
TO_JSON (
$self
) {
die
'cannot produce JSON output for data_only format'
if
$self
->output_format eq
'data_only'
;
$self
->
format
(
$self
->output_format);
}
sub
dump
(
$self
) {
my
$encoder
= JSON::Schema::Modern::_JSON_BACKEND()->new
->utf8(0)
->convert_blessed(1)
->canonical(1)
->indent(1)
->space_after(1);
$encoder
->indent_length(2)
if
$encoder
->can(
'indent_length'
);
$encoder
->encode(
$self
);
}
# turns the JSON pointers in instance_location, keyword_location into a URI fragments,
# for strict draft-201909 adherence
sub
_map_uris (
$data
) {
return
+{
%$data
,
map
+(
$_
=> Mojo::URL->new->fragment(
$data
->{
$_
})->to_string),
qw(instanceLocation keywordLocation)
,
};
}
1;
__END__
=pod
=encoding UTF-8
=head1 NAME
JSON::Schema::Modern::Result - Contains the result of a JSON Schema evaluation
=head1 VERSION
version 0.607
=head1 SYNOPSIS
use JSON::Schema::Modern;
my $js = JSON::Schema::Modern->new;
my $result = $js->evaluate($data, $schema);
my @errors = $result->errors;
my $result_data_encoded = encode_json($result); # calls TO_JSON
# use in numeric and boolean context
say sprintf('got %d %ss', $result, ($result ? 'annotation' : 'error'));
# use in string context
say 'full results: ', $result;
# combine two results into one:
my $overall_result = $result1 & $result2;
=head1 DESCRIPTION
This object holds the complete results of evaluating a data payload against a JSON Schema using
L<JSON::Schema::Modern>.
=head1 OVERLOADS
The object contains a string overload, which evaluates to a potentially multi-line string
summarizing the errors within (if any); this is intended to be used as a user-oriented error message
that references data locations, but not schema locations.
=for stopwords iff
The object also contains a I<bitwise AND> overload (C<&>), for combining two results into one (the
result is valid iff both inputs are valid; annotations and errors from the second argument are
appended to those of the first in a new Result object).
=head1 ATTRIBUTES
=head2 valid
A boolean. Indicates whether validation was successful or failed.
=head2 errors
Returns an array of L<JSON::Schema::Modern::Error> objects.
=head2 annotations
Returns an array of L<JSON::Schema::Modern::Annotation> objects.
=head2 output_format
=for stopwords subschemas
One of: C<flag>, C<basic>, C<strict_basic>, C<terse>, C<data_only>. Defaults to C<basic>.
=over 4
=item *
C<flag> returns just the result of the evaluation: either C<{"valid": true}> or C<{"valid": false}>.
=item *
C<basic> adds the list of C<errors> or C<annotations> to the boolean evaluation result. C<instance_location> and C<keyword_location> are always included, as JSON pointers, describing the path to the evaluation location; C<absolute_keyword_location> is added (as a resolved URI) whenever it is known and different from C<keyword_location>.
=item *
C<strict_basic> is like C<basic> but follows the draft-2019-09 specification precisely, including replicating an error fixed in the next draft, in that C<instance_location> and C<keyword_location> values are provided as fragment-only URI references rather than JSON pointers.
=item *
C<terse> is not described in any specification; it is like C<basic>, but omits some redundant errors (for example the one for the C<allOf> keyword that is added when any of the subschemas under C<allOf> failed evaluation).
=item *
C<data_only> returns a string, not a data structure: it contains a list of errors identified only by their C<instance_location> and error message (or C<keyword_location>, when the error occurred while loading the schema itself). This format is suitable for generating errors when the schema is not published, or for describing errors with the schema itself. This is not an official specification format and may change slightly over time, as it is tested in production environments.
=back
=head2 formatted_annotations
A boolean flag indicating whether L</format> should include annotations in the output. Defaults to true.
=head2 exception
Indicates that evaluation stopped due to a severe error.
=head2 recommended_response
=for stopwords OpenAPI
A tuple, consisting of C<[ integer, string ]>, indicating the recommended HTTP response code and
string to use for this result (if validating an HTTP request). This could exist for things like a
failed authentication check in OpenAPI validation, in which case it would contain
C<[ 401, 'Unauthorized' ]>.
Only populated when there are errors; when not explicitly set by an evaluator, defaults to
C<< [ 500, 'Internal Server Error' ] >> if any errors indicate an exception, and
C<< [ 400, <first error string> ] >> otherwise. The exact error string is hidden in the case of 500
errors because you should not leak internal issues with your application, but you may also wish to
obfuscate normal validation errors, in which case you should check for C<400> and change the string
to C<'Bad Request'>.
=head1 METHODS
=for Pod::Coverage BUILD BUILDARGS OUTPUT_FORMATS result stringify annotation_count error_count
true false
=head2 format
Returns a data structure suitable for serialization; requires one argument specifying the output
format to use, which corresponds to the formats documented in
L<https://json-schema.org/draft/2019-09/json-schema-core.html#rfc.section.10.4>. The only supported
formats at this time are C<flag>, C<basic>, C<strict_basic>, and C<terse>.
=head2 TO_JSON
Calls L</format> with the style configured in L</output_format>.
=head2 count
Returns the number of annotations when the result is true, or the number of errors when the result
is false.
=head2 combine
When provided with another result object, returns a new object with the combination of all results.
See C<&> at L</OVERLOADS>.
=head2 dump
Returns a JSON string representing the result object, using the requested L</format>, according to
the L<draft2019-09 specification|https://json-schema.org/draft/2019-09/json-schema-core.html#rfc.section.10>
and the L<draft2020-12 specification|https://json-schema.org/draft/2020-12/json-schema-core#section-12>.
=head1 SERIALIZATION
Results (and their contained errors and annotations) can be serialized in a number of ways.
Results have defined L</output_format>s, which can be generated as nested unblessed hashes/arrays
and are suitable for serializing using a JSON encoder for use in another application. A JSON string of
the result can be obtained directly using L</dump>.
If it is preferable to omit direct references to the schema (for example in an application where the
schema is not published), but still convey some semantic information about the nature of the errors,
stringify the object directly. This also means that result objects can be thrown as exceptions, or
embedded in error messages.
If you are embedding the full result inside another data structure, perhaps to be serialized to JSON
(or another format) later on, use L</TO_JSON> or L</format>.
=head1 SUPPORT
Bugs may be submitted through L<https://github.com/karenetheridge/JSON-Schema-Modern/issues>.
I am also usually active on irc, as 'ether' at C<irc.perl.org> and C<irc.libera.chat>.
=for stopwords OpenAPI
You can also find me on the L<JSON Schema Slack server|https://json-schema.slack.com> and L<OpenAPI Slack
server|https://open-api.slack.com>, which are also great resources for finding help.
=head1 AUTHOR
Karen Etheridge <ether@cpan.org>
=head1 COPYRIGHT AND LICENCE
This software is copyright (c) 2020 by Karen Etheridge.
This is free software; you can redistribute it and/or modify it under
the same terms as the Perl 5 programming language system itself.
=cut