Hi, is trimming leading zeros from numeric string a feature or a bug?

- 0907123456

valid mobile number gets parsed as 907123456 integer. I know I can use quotes to parse it as string. I was just used to not having to use quotes in neon syntax. If this is a feature it could help someone else if it was mentioned in docs

I'd like include a regexp version of
Strict comparison using === between 'OBJECT'|PhpParser\Node\Expr and 'ARRAY_A' will always evaluate to false. in @phpstan 's config.

This is invalid.

message: '#^Strict comparison using === between 'OBJECT'\|PhpParser\\Node\\Expr and 'ARRAY_A' will always evaluate to false\.$#'

Please advise.

Several times we had to replace array by another array instead of merging them:

For example:

[default]
    test: [a,b,c,d]
[test < default]
    test: [b,f,m,l]

And we need to give array('b', 'f', 'm', 'l') instead of array('a', 'b', 'c', 'd','f', 'm', 'l')

It could be useful to have possibility to "force replace" for example: using prefix before the array name.

[default]
    test: [a,b,c,d]
[test < default]
    !test: [b,f,m,l]

Hey all,

I couldn't find this in the forums, and googling for vim neon mostly gives me colorschemes: is there any known vim plugin to get syntax highlighting to work in .neon files?

Prologue

I'd like to propose a new NEON features. I named them anchors, references, extend. It basically came from YAML (https://en.wikipedia.org/wiki/YAML#Syntax, https://blog.daemonl.com/2016/02/yaml.html).

When you are using NEON in nette-like applications, you are probably using it for service definitions. I consider current syntax as complete and don't miss anything.

However, when you are using NEON as standalone tool for some kind of definition schema, I found myself to miss some features like extending and referencing. I have 1000+ rows in one NEON file and DRY principle comes very handy in here.

Anchors & References

YAML

foo: &anchor
 K1: "One"
 K2: "Two"

bar: *anchor

Result

{
    "foo": {
        "K1": "One",
        "K2": "Two"
    },
    "bar": {
        "K1": "One",
        "K2": "Two"
    }
}

NEON

# A
foo: &anchor
bar: *anchor

# B
foo: &(anchor)
bar: $(anchor)

# C
foo: =$(anchor)
bar: $(anchor)

& and * has no special meaning in NEON, am I right?

Extending

YAML

foo: &anchor
  K1: "One"
  K2: "Two"

bar:
  <<: *anchor
  K2: "I Changed"
  K3: "Three"

Result

{
    "foo": {
        "K1": "One",
        "K2": "Two"
    },
    "bar": {
        "K1": "One",
        "K2": "I Changed",
        "K3": "Three"
    }
}

NEON

# A
foo: # depends on anchor 
bar: <<$anchor

# B
foo: # depends on anchor
bar: ...$anchor

It's an idea of adding extra features to NEON. Maybe someone think the same way. Thanks for a feedback.

List of resources

• https://ne-on.org/
• https://en.wikipedia.org/wiki/YAML#Syntax
• https://blog.daemonl.com/2016/02/yaml.html
• https://docs.gitlab.com/ee/ci/yaml
• http://blogs.perl.org/users/tinita/2019/05/reusing-data-with-yaml-anchors-aliases-and-merge-keys.html

This

foo:
    bar

is parsed as

array
    foo => bar

is it ok? I cannot find any test...

When I did a typo in a file like "q" on the last line:

Fatal error: Uncaught exception 'Nette\Neon\Exception' with message 'Unexpected 'end' on line 71, column 2.' in /Neon/Decoder.php:329

• bug report? no
• feature request? yes

Description

I don't think there is super big chance for this feature, but let's try it... :) As you know in PHP it's possible to use group declarations for namespaces. I found it very useful. And now I am missing this syntax in NEON format. Instead of this:

services:
  - Adeira\Connector\Devices\Application\Service\WeatherStation\AddWeatherStationRecordService
  - Adeira\Connector\Devices\Application\Service\WeatherStation\ViewAllWeatherStationsService
  - Adeira\Connector\Devices\Application\Service\WeatherStation\ViewSingleWeatherStationService

It would be awesome to write just this:

services:
  - Adeira\Connector\Devices\Application\Service\WeatherStation\{
      AddWeatherStationRecordService,
      ViewAllWeatherStationsService,
      ViewSingleWeatherStationService
    }

Also it opens new ways of configuring multiple services at once:

services:
  - class: Adeira\Connector\Devices\Application\Service\WeatherStation\{
      AddWeatherStationRecordService,
      ViewAllWeatherStationsService,
      ViewSingleWeatherStationService
    }
    arguments:
      transactionalSession: @doctrineSession
      ownerService: @userIdOwnerService

But I would be ok with just simple group format without possibilities to configure these services. Unfortunately it means that NEON syntax will be more complicated, but it should be without BC break (works only with [a-z]\\{ ...). What do you think? Or maybe it's just stupid idea... :)

This is maby more related to the Nette\DI package, but it's NEON syntax...

Version: 2.4.3

Bug Description

parse ' - - ' string in yaml error.

Expected Behavior

run the follow code:

use Nette\Neon\Neon;

$a = '
$query:
  - - same
    - id:     1
  - - case
    - isTop:  1
      fail:   1
$seek:
  - start/503102
  - field:  id
    take:   20
';
$result = Neon::decode(trim($a));

Let's have following messagess.neon file:

a:
    b: Some text
    c: Some other text
    d: someKey:
        e: Some next message
f: Hello world

We want to parse this file by Decoder, the decoded values encode by Encoder and save the file back, because we want fix "Neon coding standards" in this file. Like remove unneeded ",' etc...

The common problem is, that the Encoder properly fix the Neon coding standards, but generates also unneeded newlines and tabs.

Let's have following code:

$file = file_get_contents('messages.neon');
$decoder = new Decoder;
$decoded = $decoder->decode($file);
$encoder = new Encoder;
$encoded = $encoder->encode($decoded, Encoder::BLOCK);
file_put_contents('messages.neon', $encoded);

And the result of messages.neon?

a:
    b: Some text
    c: Some other text
    d: someKey:
        e: Some next message
        #two tabs
    #one tab
f: Hello world

Reported year ago – nette/nette#1057. However today I found out that only part of the problem was actually solved. {"abc":123} is still not parsed properly.

I've written a fix but it breaks a test which relies on literal starting with colon.

DateTime representation in generated Neon isn't properly parsed as a DateTime-compatible string and is therefore decoded as a string instead of a DateTime instance.

This breaks stuff, as $data !== decode(encode($data)).

PHP 7.0.3 CLI from MacPorts.

$data = [
    'myDate' => new DateTime('2016-06-03T19:00:00+02:00'),
];

$neon = Nette\Neon\Neon::encode($data, Nette\Neon\Neon::BLOCK);
$decoded = Nette\Neon\Neon::decode($neon);

// Fatal error: Call to a member function format() on a string
$decoded['myDate']->format('Y-m-d H:i:s O');

The problem is that the encoder formats a DateTime using the Y-m-d H:i:s O format string; the O format character doesn't add a colon between the timezone offset's hours and minutes (unlike the P format character). The regular expression in the decoder won't match the string without the colon in the timezone offset (the specific part of the regexp in question being Z|[-+]\d\d?(?::\d\d)?).

Fixing this is as simple as putting a single question mark in the decoder regexp after the offending colon, making it Z|[-+]\d\d?(?::?\d\d)?. I'd fix this and send a PR, but I won't get around to doing it today..

• bug report? yes
• feature request? no
• version: *

Description

There are few examples on website, but otherwise there is nothing else.

To fully understand the syntax of the file a grammar definitions are required. Take a look at http://json.org/ for a good example.

Steps To Reproduce

1. Try to read the documentation.
2. Bug: There is no documentation.

I'm not sure there is anything we should do about it.

http://seriot.ch/parsing_json.html

Version: 3.1.0

encoder::encode() Might return false, because e.g. json-encode can return false and therefore this might violate the defined return type of string (php fatal error).

I guess in this case I run into this problem phpstan/phpstan#2956

Any suggestions/ideas on how to handle this problem?

Version: 3.1.0

Bug Description

Cannot escape single quote inside single quoted string

Steps To Reproduce

'I''m'

Expected Behavior

Some way (ideally compatible with YAML) to escape single quotes inside single quoted strings

Possible Solution

Use double single quote (like YAML)

name: d:g

is fine, but

name: d=g # name= d=g

throws Error: Unexpected '=' on line 1, column 8.

I think that it is better to return large integers as string instead of casting them to float.

What do you think? Any tips, how to efficiently decide whether string contains integer larger than MAX_INT?

• bug report? probably

I assumed that the following three should be equal, yet they are not.

- App\Model\AnalyzersRegistry(analyzers: {phpstan: @App\Model\PhpStanAnalyzer})
- App\Model\AnalyzersRegistry(
	analyzers: {
		phpstan: @App\Model\PhpStanAnalyzer
	}
)
- App\Model\AnalyzersRegistry(
	analyzers:
		phpstan: @App\Model\PhpStanAnalyzer
)

The last example is parsed differently.

Hi There,

On a line with a string value, how would someone escape a '(' or ')'?

MyString: This (unfortunately) won't work.

There appears to be no way to do this?

Version: 3.2.1

Bug Description

In case of NEON:

urls:
    - https://cdnjs.cloudflare.com/ajax/libs/babel-polyfill/6.23.0/polyfill.min.js
    - https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.2/MathJax.js?config=TeX-AMS_HTML
    - https://baraja.cz/css/styles.css

Is it parsed to:

Because the second URL contains = character.

Expected Behavior & Possible Solution

I think in the context of the URL (starting with #^(https?|ftp)://#) NEON should ignore character = in the URL and keep it as a string.

Thanks!

Version: 3.1.1

it is ok to use keys like 2020-02 without quotes, but 2020-02-02 is a syntax error.

foo:
    2020-02: bar # ok
    2020-02-02: baz # fails

i would expect the behavior to be the same regardless of number of dashes, but maybe the first case working is this some side effect of parsing numbers in sci notation?

anyway, it would be really nice if this works : ]

I just discovered Neon and love the way it works. Especially, I love the notion of entity.

Now, I have a question about the supported syntax inside an Entity.

I've noticed that this is OK:

entity0: Column(name=Homer, phones=[ home: 555-6528, work: 555-7334 ])

This is ok too:

entity: Column(
    name: Homer
    phones: [ home: 555-6528, work: 555-7334 ]
)

But this yields what I would deem to be an unexpected result:

entity2: Column(
    name: Homer
    phones:
        home: 555-6528
        work: 555-7334
)

In the attributes array, "home" and "work" are at the same level as "name" and "phones". I would have expected those keys to be in the "phones" key.

Also, this piece of code fails to parse:

entity2: Column(
    name: Homer
    phones:
     - 555-6528
     - 555-7334
)

Is this considered a bug? If not, is there a specification of the NEON format somewhere explaining what is ok and what is not?
Could this be considered as a possible enhancement?

Version: 3.3.3

Steps To Reproduce

Neon::encode([INF], Neon::BLOCK);

Expected Behavior

Better exception or ???

Possible Solution

Problem is here

Prologue

This RFC aims to keep original NEON format and prevent data losing. It's in correlation to #51.

Picture you have really big NEON file holding big definition schema with comments. It's very hand to have this kind of definition schema, because people keep their comments there and after all this NEON file is converted to JSON.

At this time we handle it like this:

• There is table in DB holding NEON column and JSON column.
• Someone update NEON and both columns get updated.

So far so good.

Unless you need to apply some automatic migrations, in current situation you can't. We'd loose user comments and they are really needed.

Current API

foo.neon

# List of users
users: [1,2,3,4]
# Revision v0.0.1

test.php

$content = file_get_contents('./foo.neon');

$data = Nette\Neon\Neon::decode($content);
$data->users[] = 5;

file_put_contents('./foo.neon', Nette\Neon\Neon::encode($data));

foo.neon

users: [1,2,3,4,5]

I totally understand how this API works and it works great for encoding/decoding, but it does not prevent original content.

Proposed API

AST

I am not sure how this API should look like. Maybe it would be needed to create some kind of AST parser and understand comments properly.

AST parsers is maybe too heavy and someone would bring up simpler solution.

Context Merging

Method Neon::encode does not have any context of original file, passing original file instead of string could be the way. I am not sure.

$data = Nette\Neon\Neon::merge('./foo.neon', ['users' => [5]);
file_put_contents('./foo.neon', $data);

It's an idea of adding extra feature to NEON. Maybe someone think the same way. Thanks for a feedback.

There is no check whether next token is array-bullet, so now it allows following syntax:

foo:
bar

which is parsed as

array
    foo => bar

But more complex syntax like this one:

foo:
bar
lorem:
ipsum

is not supported. ("Unexpected 'lorem'....")

So there are two possible ways how to solve this:

1. allow only array (to be more specific, only block array with array bullet) [possible BC break], see this commit
2. allow everything and fix that more complex syntax

Which one do you prefer?

To increase compatibility with JSON ecosystem

Currently seems to be unsupported HHVM:
Fatal error: Arrays are not allowed in class constants in /vendor/nette/neon/src/Neon/Decoder.php on line 20

Will it be improved to support HHVM?

When I write chmod octal value, leading zero is stripped and an integer returned:

Example code:
fileMode: 0777 -> returns int(777)

In PHP (and eg. Symfony Container as well) is it intepreted as an octal value and converted to integer:
var_dump(0777); // int(511)

I hope this is not intended behaviour, may we solve it?