The validator thinks -0 is a string. not a number

[coderextreme]

See attached CC.txt JSON document and Schema.txt JSON schema

CC.txt
Schema.txt

[erosb]

@coderextreme thanks for reporting it. I'm afraid I can't fix it, because this problem is caused by a bug of the underlying org.json API (more specifically the JSONTokener#nextValue() and/or the JSONObject#stringToValue()` methods).

I've tried to find a workaround for it, but it seems the parser notoriously parses the -0 literal as a string. Trying to convert the String instance as a Number would not make sense, because if I'd do it that way, then a schema with {"type" : "number"} would also accept the "-0" JSON string literal as a number (which is against the JSON Schema spec). That would mean fixing a bug by creating a new bug, obviously doesn't make sense.

edit: minimal snippet about my findings:

JSONObject input = new JSONObject("{\"prop\" : \"-0\"}");
System.out.println(input.get("prop") + " " + input.get("prop").getClass());
input = new JSONObject("{\"prop\" : -0}");
System.out.println(input.get("prop") + " " + input.get("prop").getClass());
input = new JSONObject("{\"prop\" : -0.0}");
System.out.println(input.get("prop") + " " + input.get("prop").getClass());

It prints:

-0 class java.lang.String
-0 class java.lang.String
-0.0 class java.lang.Double

So both -0 and "-0" are parsed into a String instance (while -0.0 is properly parsed as Double), therefore the validator cannot decide after parsing if the original literal was a string or a number.

[coderextreme]

Can we use something besides JSONTokener?
On Jan 27, 2016 8:10 AM, "Bence Eros" notifications@github.com wrote:

@coderextreme https://github.com/coderextreme thanks for reporting it.
I'm afraid I can't fix it, because this problem is caused by a bug of the
underlying org.json API (more specifically the JSONTokener#nextValue()
and/or the JSONObject#stringToValue()` methods).

I've tried to find a workaround for it, but it seems the parser
notoriously parses the -0 literal as a string. Trying to convert the
String instance as a Number would not make sense, because if I'd do it
that way, then a schema with {"type" : "number"} would also accept the
"-0" JSON string literal as a number (which is against the JSON Schema
spec). That would mean fixing a bug by creating a new bug, obviously
doesn't make sense.

—
Reply to this email directly or view it on GitHub
#25 (comment)
.

[erosb]

Well I'm not going to reimplement the tokener from scratch. I'll look after how much time does it take to fix the org.json implementation.

[coderextreme]

I was thinking of doing a JSON.parse() and working with that result, but
I'm not familiar with parsers in Java, just JavaScript.
On Jan 27, 2016 8:26 AM, "Bence Eros" notifications@github.com wrote:

Well I'm not going to reimplement the tokener from scratch. I'll look
after how much time does it take to fix the org.json implementation.

—
Reply to this email directly or view it on GitHub
#25 (comment)
.

[coderextreme]

Or maybe we could switch to gson if it's better.
On Jan 27, 2016 8:29 AM, "John Carlson" yottzumm@gmail.com wrote:

I was thinking of doing a JSON.parse() and working with that result, but
I'm not familiar with parsers in Java, just JavaScript.
On Jan 27, 2016 8:26 AM, "Bence Eros" notifications@github.com wrote:

Well I'm not going to reimplement the tokener from scratch. I'll look
after how much time does it take to fix the org.json implementation.

—
Reply to this email directly or view it on GitHub
#25 (comment)
.

[coderextreme]

I filed an issue here against org.json: stleary/JSON-java#187

[erosb]

As the most simple temporary solution I recommend using -0.0 in your JSON instead of -0

[coderextreme]

I don’t control the JSON but I can make a recommendation.

John

On Jan 27, 2016, at 8:51 AM, Bence Eros notifications@github.com wrote:

As the most simple temporary solution I recommend using -0.0 in your JSON instead of -0

—
Reply to this email directly or view it on GitHub #25 (comment).

[stleary]

A pull request containing the fix can be found here: stleary/JSON-java#188. Feel free to try it out and let us know whether it addresses the issue.

[coderextreme]

Thanks. Our JSON passes through now (at least that area). We had over 6300 instances of -0 so this was pretty critical.

John

On Jan 27, 2016, at 10:42 PM, Sean Leary notifications@github.com wrote:

A pull request containing the fix can be found here: stleary/JSON-java#188 stleary/JSON-java#188. Feel free to try it out and let us know whether it addresses the issue.

—
Reply to this email directly or view it on GitHub #25 (comment).

[erosb]

@stleary how are the artifacts deployed to maven central? From my side, fixing this issue will only mean upgrading (in my pom.xml) to the version containing the bugfix.