feat(validator): use jsonschema Draft2020

[Breee]

Uses current jsonschema as draft: https://json-schema.org/draft/2020-12

I also hat to add c.AssertFormat() to fix the tests

This fixes my issue #345

tested with:

$ make 
$ go build -o bin/kubeconform cmd/kubeconform/main.go
$ go test -v ./...
?       github.com/yannh/kubeconform/cmd/kubeconform    [no test files]
?       github.com/yannh/kubeconform/examples   [no test files]
?       github.com/yannh/kubeconform/pkg/cache  [no test files]
=== RUN   TestSkipKindMaps
--- PASS: TestSkipKindMaps (0.00s)
=== RUN   TestFromFlags
--- PASS: TestFromFlags (0.00s)
PASS
ok      github.com/yannh/kubeconform/pkg/config 0.002s
=== RUN   TestHTTPURLLoader_Load
=== RUN   TestHTTPURLLoader_Load/successful_load
=== RUN   TestHTTPURLLoader_Load/not_found_error
=== RUN   TestHTTPURLLoader_Load/server_error
=== RUN   TestHTTPURLLoader_Load/cache_hit
=== RUN   TestHTTPURLLoader_Load/Partial_response_from_server
--- PASS: TestHTTPURLLoader_Load (0.01s)
    --- PASS: TestHTTPURLLoader_Load/successful_load (0.00s)
    --- PASS: TestHTTPURLLoader_Load/not_found_error (0.00s)
    --- PASS: TestHTTPURLLoader_Load/server_error (0.00s)
    --- PASS: TestHTTPURLLoader_Load/cache_hit (0.00s)
    --- PASS: TestHTTPURLLoader_Load/Partial_response_from_server (0.00s)
=== RUN   TestHTTPURLLoader_Load_Retries
=== RUN   TestHTTPURLLoader_Load_Retries/retries_on_503
=== RUN   TestHTTPURLLoader_Load_Retries/fails_when_hitting_max_retries
=== RUN   TestHTTPURLLoader_Load_Retries/retry_on_connection_reset
=== RUN   TestHTTPURLLoader_Load_Retries/retry_on_connection_reset#01
--- PASS: TestHTTPURLLoader_Load_Retries (8.01s)
    --- PASS: TestHTTPURLLoader_Load_Retries/retries_on_503 (1.00s)
    --- PASS: TestHTTPURLLoader_Load_Retries/fails_when_hitting_max_retries (3.00s)
    --- PASS: TestHTTPURLLoader_Load_Retries/retry_on_connection_reset (1.00s)
    --- PASS: TestHTTPURLLoader_Load_Retries/retry_on_connection_reset#01 (3.00s)
PASS
ok      github.com/yannh/kubeconform/pkg/loader 8.027s
=== RUN   TestJSONWrite
--- PASS: TestJSONWrite (0.00s)
=== RUN   TestJunitWrite
--- PASS: TestJunitWrite (0.00s)
=== RUN   TestPrettyTextWrite
--- PASS: TestPrettyTextWrite (0.00s)
=== RUN   TestTapWrite
--- PASS: TestTapWrite (0.00s)
=== RUN   TestTextWrite
--- PASS: TestTextWrite (0.00s)
PASS
ok      github.com/yannh/kubeconform/pkg/output 0.012s
=== RUN   TestSchemaPath
--- PASS: TestSchemaPath (0.00s)
PASS
ok      github.com/yannh/kubeconform/pkg/registry       0.010s
=== RUN   TestIsYamlFile
--- PASS: TestIsYamlFile (0.00s)
=== RUN   TestIsJSONFile
--- PASS: TestIsJSONFile (0.00s)
=== RUN   TestFindResourcesInReader
--- PASS: TestFindResourcesInReader (0.00s)
=== RUN   TestSignatureFromBytes
--- PASS: TestSignatureFromBytes (0.00s)
=== RUN   TestSignatureFromMap
--- PASS: TestSignatureFromMap (0.00s)
=== RUN   TestResources
--- PASS: TestResources (0.00s)
=== RUN   TestFromStream
--- PASS: TestFromStream (0.01s)
PASS
ok      github.com/yannh/kubeconform/pkg/resource       0.010s
=== RUN   TestValidate
--- PASS: TestValidate (0.00s)
=== RUN   TestValidationErrors
--- PASS: TestValidationErrors (0.00s)
=== RUN   TestValidateFile
--- PASS: TestValidateFile (0.00s)
PASS
ok      github.com/yannh/kubeconform/pkg/validator      0.010s

$ ./bin/kubeconform -schema-location default -schema-location 'https://raw.githubusercontent.com/datreeio/CRDs-catalog/refs/heads/main/{{.Group}}/{{.ResourceKind}}_{{.ResourceAPIVersion}}.json' test.yaml 

$ git clone https://github.com/gravitee-io/gravitee-kubernetes-operator /tmp/gravitee-kubernetes-operator

script:

set -euo pipefail

# Configuration
KUBECONFORM_BIN="$(dirname "$0")/bin/kubeconform"
EXAMPLES_DIR="${EXAMPLES_DIR:-/tmp/gravitee-kubernetes-operator}"
SCHEMA_LOCATIONS=(
    "default"
    "https://raw.githubusercontent.com/datreeio/CRDs-catalog/refs/heads/main/{{.Group}}/{{.ResourceKind}}_{{.ResourceAPIVersion}}.json"
)

# Check if kubeconform binary exists
if [ ! -f "$KUBECONFORM_BIN" ]; then
    echo "Error: kubeconform binary not found at $KUBECONFORM_BIN"
    echo "Please run: go build -o bin/kubeconform cmd/kubeconform/main.go"
    exit 1
fi

# Check if examples directory exists
if [ ! -d "$EXAMPLES_DIR" ]; then
    echo "Error: Examples directory not found at $EXAMPLES_DIR"
    exit 1
fi

# Build schema location arguments
SCHEMA_ARGS=()
for loc in "${SCHEMA_LOCATIONS[@]}"; do
    SCHEMA_ARGS+=("-schema-location" "$loc")
done

# Find all YAML files
find "$EXAMPLES_DIR" -type f \( -name "*.yaml" -o -name "*.yml" \) | while read -r file; do
    echo "Validating: $file"
    "$KUBECONFORM_BIN" "${SCHEMA_ARGS[@]}" "$file"
done

[Breee]

@yannh would you be so kind to review this?

[yannh]

@Breee hi! Thanks for the contribution! Could you provide an explanation what this would change/ why this is needed? This would change the default behavior, is the new spec fully backwards compatible or is there a risk of breakage?

[Breee]

So in my case i want to run kubeconform with this CRD:

https://github.com/datreeio/CRDs-catalog/blob/main/gravitee.io/apidefinition_v1alpha1.json

here is some minimal example:

package main

import (
	"encoding/json"
	"fmt"
	"io"
	"net/http"

	jsonschema "github.com/santhosh-tekuri/jsonschema/v6"
)

func main() {
	schemaURL := "https://raw.githubusercontent.com/datreeio/CRDs-catalog/refs/heads/main/gravitee.io/apidefinition_v1alpha1.json"

	resp, err := http.Get(schemaURL)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	schemaBytes, err := io.ReadAll(resp.Body)
	if err != nil {
		panic(err)
	}

	var schemaData interface{}
	json.Unmarshal(schemaBytes, &schemaData)

	// OLD CODE (master branch)
	fmt.Println("OLD CODE (Draft4):")
	c1 := jsonschema.NewCompiler()
	c1.DefaultDraft(jsonschema.Draft4)
	c1.AddResource(schemaURL, schemaData)
	_, err = c1.Compile(schemaURL)
	if err != nil {
		fmt.Printf("FAILED: %v\n\n", err)
	} else {
		fmt.Println("SUCCESS\n")
	}

	// NEW CODE (feature/jsonschema-draft2020 branch)
	fmt.Println("NEW CODE (Draft2020 + AssertFormat):")
	c2 := jsonschema.NewCompiler()
	c2.DefaultDraft(jsonschema.Draft2020)
	c2.AssertFormat()
	c2.AddResource(schemaURL, schemaData)
	_, err = c2.Compile(schemaURL)
	if err != nil {
		fmt.Printf("FAILED: %v\n", err)
	} else {
		fmt.Println("SUCCESS")
	}
}

yields:

OLD CODE (Draft4):
FAILED: "https://raw.githubusercontent.com/datreeio/CRDs-catalog/refs/heads/main/gravitee.io/apidefinition_v1alpha1.json#" is not valid against metaschema: jsonschema validation failed with 'http://json-schema.org/draft-04/schema#'
- at '/properties/spec/properties/additionalProperties': got boolean, want object

NEW CODE (Draft2020 + AssertFormat):
SUCCESS

so its about how the additionalProperties are handled in the latest version of the json spec. I can dig deeper and probably just add a cli flag?