Ignore trailing '#' in meta-schema URLs

[GREsau]

A single trailing # should be present for draft-07 and earlier, but should not be present for 2019-09 or 2020-12 (e.g. see the $schema value in draft2020-12/schema.json). But for simplicity and robustness, we can just ignore it in all cases.

With this change, the output of this example is now false:

let schema = JSONSchema::compile(&json!({
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "unevaluatedProperties": false
}))
.unwrap();

let value = json!({
    "blah": 1
});

println!("{}", schema.is_valid(&value));

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.05%. Comparing base (f001fe6) to head (79c8de5).
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #497      +/-   ##
==========================================
+ Coverage   89.87%   90.05%   +0.18%     
==========================================
  Files          58       58              
  Lines        9945     9866      -79     
==========================================
- Hits         8938     8885      -53     
+ Misses       1007      981      -26     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[codspeed-hq]

CodSpeed Performance Report

Merging #497 will degrade performances by 49.53%

_{Comparing GREsau:meta_schema_trailing_hash (79c8de5) with master (c449e7a)}

Summary

⚡ 17 improvements
❌ 15 regressions
✅ 288 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	master	GREsau:meta_schema_trailing_hash	Change
⚡	additional_properties_and_properties bartruefoo1[jsonschema/is_valid/invalid]	1.4 µs	1.2 µs	+12.63%
⚡	additional_properties_false [jsonschema/is_valid/valid]	427.8 ns	360.6 ns	+18.64%
⚡	additional_properties_false [jsonschema/validate/valid]	2.6 µs	2.3 µs	+13.45%
⚡	additional_properties_false foo1[jsonschema/is_valid/invalid]	645.6 ns	391.7 ns	+64.82%
⚡	additional_properties_false_and_properties bar2foo1[jsonschema/is_valid/invalid]	1,050 ns	923.9 ns	+13.65%
⚡	additional_properties_single foo1[jsonschema/is_valid/invalid]	955.3 ns	793.6 ns	+20.37%
⚡	additional_properties_single foobar[jsonschema/is_valid/valid]	960.8 ns	823.9 ns	+16.62%
❌	any_of 1[jsonschema/is_valid/valid]	611.1 ns	698.6 ns	-12.52%
❌	any_of_multiple_types foo[jsonschema/is_valid/valid]	736.9 ns	882.8 ns	-16.52%
❌	any_of_multiple_types null[jsonschema/is_valid/invalid]	737.8 ns	883.6 ns	-16.5%
❌	boolean_false 1[jsonschema/is_valid/invalid]	29.7 ns	58.9 ns	-49.53%
❌	const 1[jsonschema/is_valid/valid]	392.2 ns	479.7 ns	-18.24%
❌	const foo[jsonschema/is_valid/invalid]	240.3 ns	327.8 ns	-26.69%
❌	format_date 06/19/1963[jsonschema/is_valid/invalid]	985.3 ns	1,135.8 ns	-13.26%
⚡	format_email foo[jsonschema/is_valid/invalid]	585.6 ns	527.2 ns	+11.06%
⚡	format_ipv4 127.0.0.999[jsonschema/is_valid/invalid]	1.7 µs	1.5 µs	+16.28%
⚡	format_ipv4 20010db885a3000000008a2e03707334[jsonschema/is_valid/invalid]	1.8 µs	1.6 µs	+12.49%
⚡	format_ipv6 20010db885a3000000008a2e03707334[jsonschema/is_valid/valid]	1.8 µs	1.6 µs	+16.97%
❌	format_time 083006.283185Z[jsonschema/validate/valid]	5 µs	5.8 µs	-12.7%
❌	format_uri_template http//example.com/dictionary/term1/term[jsonschema/validate/valid]	5.3 µs	6.1 µs	-13.9%
...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

[Stranger6667]

Thanks!

Should the example be evaluated as true? Given the example from learnjsonschema.com:

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "unevaluatedProperties": false
}

All object instances fail against the false schema

Am I correct that this change exposes #496 more clearly?

[GREsau]

Should the example be evaluated as true?

Whoops, sorry that was a typo. I meant it currently outputs true, but with this PR, it now outputs false!

[GREsau]

Am I correct that this change exposes #496 more clearly?

Kinda, in that you no longer need to specify .with_draft(Draft::Draft202012), as it will now infer that it's a 2020-12 schema from $schema

[Stranger6667]

Whoops, sorry that was a typo. I meant it currently outputs true, but with this PR, it now outputs false!

Aha, got it! So, the other part of my comment is irrelevant then :) I haven't checked it myself yet, but it would not fix #496, right? as it seems to me that #496 is caused by some issue in UnevaluatedPropertiesValidator::is_valid

[GREsau]

I haven't checked it myself yet, but it would not fix #496, right? as it seems to me that #496 is caused by some issue in UnevaluatedPropertiesValidator::is_valid

Correct, they're entirely separate issues

[Stranger6667]

Good. I am going through the failures in bowtie.report and will push some fixes this week. For today, I plan to fix the issue of $ref ignoring $schema inside the resolved schema.