# jesse [![Build Status][2]][1]
jesse (JSON Schema Erlang) is an implementation of a JSON Schema validator
for Erlang, though it can work just as well as a CLI tool.
jesse implements the following specifications:
* [Draft 03](http://tools.ietf.org/html/draft-zyp-json-schema-03)
* [Draft 04](http://tools.ietf.org/html/draft-zyp-json-schema-04)
* [Draft 06](https://datatracker.ietf.org/doc/html/draft-wright-json-schema-00)
Install from git or https://hex.pm/packages/jesse .
For convenience, [available as a Docker image too](https://hub.docker.com/r/ysoftwareab/jesse): `docker run --rm -it -v $PWD:$PWD -w $PWD ysoftwareab/jesse ...`.
And [whalebrew package](https://github.com/whalebrew/whalebrew): `brew install whalebrew && sudo whalebrew install ysoftwareab/jesse; jesse ...`
## Erlang API Docs
Automatically generated docs are available https://dev.erldocs.com/github.com/for-get/jesse/ .
Please keep in mind that the public API is the `jesse.erl` module alone.
## Command-Line Interface
You can either build the `bin/jesse` executable yourself (just type `make`),
or you can use a Docker image a call it with `docker run ysoftwareab/jesse`.
You can fire up `jesse` from the CLI, with
```bash
bin/jesse [path_to_json_schema] path_to_json_schema -- path_to_json_instance [path_to_json_instance]
```
You can also output the result in JSON format, with `--json`, and beautify it e.g. with python
```bash
bin/jesse [path_to_json_schema] path_to_json_schema --json -- path_to_json_instance [path_to_json_instance] | python -m json.tool
```
You can pass multiple JSON schemas which should be loaded into jesse in-memory storage, but JSON instances will be validated against the last JSON schema passed.
## Erlang Interface
There are two ways of using jesse:
* to use jesse internal in-memory storage to keep all your schema definitions
In this case jesse will look up a schema definition in its own storage,
and then validate given a JSON instance.
* it is also possible to provide jesse with schema definitions when jesse is called.
## Examples
NOTE: jesse doesn't have any parsing functionality. It currently works with four
formats: mochijson2, jiffy, jsx and Erlang 17+ maps, so JSON needs to be
parsed in advance, or you can specify a callback which jesse will use to
parse JSON.
In examples below and in jesse test suite jiffy parser is used.
* Use jesse's internal in-memory storage:
(parse JSON in advance)
```erlang
1> Schema = jiffy:decode(<<"{\"items\": {\"type\": \"integer\"}}">>).
{[{<<"items">>,{[{<<"type">>,<<"integer">>}]}}]}
2> jesse:add_schema("some_key", Schema).
ok
3> Json1 = jiffy:decode(<<"[1, 2, 3]">>).
[1,2,3]
4> jesse:validate("some_key", Json1).
{ok,[1,2,3]}
5> Json2 = jiffy:decode(<<"[1, \"x\"]">>).
[1,<<"x">>]
6> jesse:validate("some_key", Json2).
{error,[{data_invalid,{[{<<"type">>,<<"integer">>}]},
wrong_type,<<"x">>,
[1]}]}
```
The `[1]` in the error is the path in the original value to `<<"x">>` where the
validation failed. See *Validation errors* below for the full error format.
(using a callback)
```erlang
1> jesse:add_schema("some_key",
1> <<"{\"uniqueItems\": true}">>,
1> [{parser_fun, fun jiffy:decode/1}]).
ok
2> jesse:validate("some_key",
2> <<"[1, 2]">>,
2> [{parser_fun, fun jiffy:decode/1}]).
{ok,[1, 2]}
3> jesse:validate("some_key",
3> <<"[{\"foo\": \"bar\"}, {\"foo\": \"bar\"}] ">>,
3> [{parser_fun, fun jiffy:decode/1}]).
{error,[{data_invalid,{[{<<"uniqueItems">>,true}]},
{not_unique,{[{<<"foo">>,<<"bar">>}]}},
[{[{<<"foo">>,<<"bar">>}]},{[{<<"foo">>,<<"bar">>}]}],
[]}]}
```
* Call jesse with schema definition in place (do not use internal storage)
(parse JSON in advance)
```erlang
1> Schema = jiffy:decode(<<"{\"pattern\": \"^a*$\"}">>).
{[{<<"pattern">>,<<"^a*$">>}]}
2> Json1 = jiffy:decode(<<"\"aaa\"">>).
<<"aaa">>
3> jesse:validate_with_schema(Schema, Json1).
{ok,<<"aaa">>}
4> Json2 = jiffy:decode(<<"\"abc\"">>).
<<"abc">>
5> jesse:validate_with_schema(Schema, Json2).
{error,[{data_invalid,{[{<<"pattern">>,<<"^a*$">>}]},
no_match,
<<"abc">>,[]}]}
```
(using a callback)
```erlang
1> Schema = <<"{\"patternProperties\": {\"f.*o\": {\"type\": \"integer\"}}}">>.
<<"{\"patternProperties\": {\"f.*o\": {\"type\": \"integer\"}}}">>
2> jesse:validate_with_schema(Schema,
2> <<"{\"foo\": 1, \"foooooo\" : 2}">>,
2> [{parser_fun, fun jiffy:decode/1}]).
{ok,{[{<<"foo">>,1},{<<"foooooo">>,2}]}}
3> jesse:validate_with_schema(Schema,
3> <<"{\"foo\": \"bar\", \"fooooo\": 2}">>,
3> [{parser_fun, fun jiffy:decode/1}]).
{error,[{data_invalid,{[{<<"type">>,<<"integer">>}]},
wrong_type,<<"bar">>,
[<<"foo">>]}]}
```
* Since 0.4.0 it's possible to instruct jesse to collect errors, and not stop
immediately when it finds an error in the given JSON instance:
```erlang
1> Schema = <<"{\"properties\": {\"a\": {\"type\": \"integer\"}, \"b\": {\"type\": \"string\"}, \"c\": {\"type\": \"boolean\"}}}">>.
<<"{\"properties\": {\"a\": {\"type\": \"integer\"}, \"b\": {\"type\": \"string\"}, \"c\": {\"type\": \"boolean\"}}}">>
2> jesse:validate_with_schema(Schema,
2> <<"{\"a\": 1, \"b\": \"b\", \"c\": true}">>,
2> [{parser_fun, fun jiffy:decode/1}]).
{ok,{[{<<"a">>,1},{<<"b">>,<<"b">>},{<<"c">>,true}]}}
```
now let's change the value of the field "b" to an integer
```erlang
3> jesse:validate_with_schema(Schema,
3> <<"{\"a\": 1, \"b\": 2, \"c\": true}">>,
3> [{parser_fun, fun jiffy:decode/1}]).
{error,[{data_invalid,{[{<<"type">>,<<"string">>}]},
wrong_type,2,
[<<"b">>]}]}
```
works as expected, but let's change the value of the field "c" as well
```erlang
4> jesse:validate_with_schema(Schema,
4> <<"{\"a\": 1, \"b\": 2, \"c\": 3}">>,
4> [{parser_fun, fun jiffy:decode/1}]).
{error,[{data_invalid,{[{<<"type">>,<<"string">>}]},
wrong_type,2,
[<<"b">>]}]}
```
still works as expected, jesse stops validating as soon as finds an error.
Let's use the `allowed_errors` option, and set it to 1
```erlang
5> jesse:validate_with_schema(Schema,
5> <<"{\"a\": 1, \"b\": 2, \"c\": 3}">>,
5> [{parser_fun, fun jiffy:decode/1},
5> {allowed_errors, 1}]).
{error,[{data_invalid,{[{<<"type">>,<<"boolean">>}]},
wrong_type,3,
[<<"c">>]},
{data_invalid,{[{<<"type">>,<<"string">>}]},
wrong_type,2,
[<<"b">>]}]}
```
now we got a list of two errors.
Let's now change the value of the field "a" to a boolean
```erlang
6> jesse:validate_with_schema(Schema,
6> <<"{\"a\": true, \"b\": 2, \"c\": 3}">>,
6> [{parser_fun, fun jiffy:decode/1},
6> {allowed_errors, 1}]).
{error,[{data_invalid,{[{<<"type">>,<<"string">>}]},
wrong_type,2,
[<<"b">>]},
{data_invalid,{[{<<"type">>,<<"integer">>}]},
wrong_type,true,
[<<"a">>]}]}
```
we still got only two errors.
Let's try using 'infinity' as the argument for the `allowed_errors` option
```erlang
7> jesse:validate_with_schema(Schema,
7> <<"{\"a\": true, \"b\": 2, \"c\": 3}">>,
7> [{parser_fun, fun jiffy:decode/1},
7> {allowed_errors, infinity}]).
{error,[{data_invalid,{[{<<"type">>,<<"boolean">>}]},
wrong_type,3,
[<<"c">>]},
{data_invalid,{[{<<"type">>,<<"string">>}]},
wrong_type,2,
[<<"b">>]},
{data_invalid,{[{<<"type">>,<<"integer">>}]},
wrong_type,true,
[<<"a">>]}]}
```
Maps example
```erlang
8> jesse:validate_with_schema(Schema,
8> <<"{\"a\": 1, \"b\": 2, \"c\": true}">>,
8> [{parser_fun, fun(Bin) -> jiffy:decode(Bin, [return_maps]) end}]).
{error,[{data_invalid,#{<<"type">> => <<"string">>},
wrong_type,2,
[<<"b">>]}]}
9> jesse:validate_with_schema(Schema,
9> <<"{\"a\": 1, \"b\": \"val\", \"c\": true}">>,
9> [{parser_fun, fun(Bin) -> jiffy:decode(Bin, [return_maps]) end}]).
{ok, #{<<"a">> => 1, <<"b">> => <<"val">>, <<"c">> => true}}
```
## JSON Schema versions
jesse currently supports JSON Schema draft3, draft4 and draft6. To decide which
validator to use jesse tries to read `$schema` property from the given schema,
and checks if it's a supported one, otherwise it will return an error.
If `$schema` property isn't provided in the given schema, jesse will use the
default validator (currently the validator for draft3).
To specify which validator to use by default (if there's no `schema` property in
the given schema), one should use 'default_schema_ver' option when call
`jesse:validate/3` or `jesse:validate_with_schema/3`, the value should be
a binary consisting a schema path,
i.e. <<"http://json-schema.org/draft-03/schema#">>.
## Validation errors
The validation functions `jesse:validate/2` and `jesse:validate_with_schema/2,3`
return `{ok, Value}` on success and `{error, ListOfErrors}` on failure. An error
is either `data_invalid` or `schema_invalid`.
A `data_invalid` error is a tuple on the form `{data_invalid, Schema, ErrorType,
Value, Path}` where
* Schema is the part of the schema where validation failed
* ErrorType is the type of error, usually an atom such as `wrong_type`,
`not_in_range` or `no_match`
* Value is The part of the value where failed validation against Schema
* Path is a path to where validation failed within the original value. The path
is a list of property names and zero-based array indices referencing the
properties and array items within a JSON document; e.g. in the JSON document
`{"foo": [42, 43, 44]}`, the path `[<<"foo">>, 0]` refers to the value 42. An
empty list refers to the whole JSON document.
A `schema_invalid` error is a tuple on the form `{schema_invalid, Schema,
ErrorType}` where
* Schema is the part of the schema which is invalid
* ErrorType is an atom such as `missing_id_field` or a tuple such as
`{wrong_type_dependency, Dependency}`.
## Caveats
* pattern and patternProperty attributes:
jesse uses standard erlang module `re` for regexp matching, therefore there
could be some incompatible regular expressions in schemas you define.
From the [erlang docs](https://www.erlang.org/doc/man/re.html#description):
> The matching algorithms of the library are based on the PCRE library, but
> not all of the PCRE library is interfaced and some parts of the library go
> beyond what PCRE offers.
Most common cases should work fine. Note that jesse provides an application
environment setting, `re_options` (default: `[unicode, ucp]`), for customizing
its use of the `re` module. `ucp` provides the widest compatibility for
matching unicode code points beyond ISO Latin-1 in character classes like
`\w`, `\s`, and `\d`, but hurts performance. If maximum compatibility is not
required and performance is a significant concern, set `re_options` to
`[unicode]` instead. See notes on the `ucp` option at
[`re:compile/2`](https://www.erlang.org/doc/man/re.html#compile-2) for more
details.
* internal references (id attribute) are NOT supported
http://json-schema.org/latest/json-schema-core.html#rfc.section.8.2.1
## Contributing
If you see something missing or incorrect, a pull request is most welcome!
## License
[Apache 2.0](LICENSE)
## Stargazers over time
[](https://starchart.cc/for-GET/jesse)
[1]: https://github.com/for-GET/jesse/actions?query=workflow%3ACI+branch%3Amaster
[2]: https://github.com/for-GET/jesse/workflows/CI/badge.svg?branch=master
