weatherreporter/internal/generatedtext/json_test.go

package generatedtext

import (
	"bytes"
	"fmt"
	"strings"
	"testing"

	"gitea.maximumdirect.net/eric/weatherreporter/internal/promptassets"
	"github.com/santhosh-tekuri/jsonschema/v6"
)

func TestGeneratedTextValidatorsRequireExactObjectFields(t *testing.T) {
	validators := []struct {
		name     string
		valid    string
		validate func([]byte) error
	}{
		{name: "daily", valid: `{"summary":"Summary","forecast_discussion":["Discussion"],"precipitation_timing":""}`, validate: validateDailyJSON},
		{name: "today", valid: `{"summary":"Summary","forecast_discussion":["Discussion"],"precipitation_timing":""}`, validate: validateTodayJSON},
		{name: "tomorrow", valid: `{"summary":"Summary","forecast_discussion":["Discussion"],"precipitation_timing":""}`, validate: validateTomorrowJSON},
		{name: "hourly", valid: `{"summary":"Summary","forecast_discussion":"Discussion","precipitation_timing":""}`, validate: validateHourlyJSON},
	}
	for _, validator := range validators {
		t.Run(validator.name, func(t *testing.T) {
			for _, test := range []struct {
				name  string
				input string
				valid bool
			}{
				{name: "canonical", input: validator.valid, valid: true},
				{name: "missing", input: strings.Replace(validator.valid, `"summary":"Summary",`, "", 1)},
				{name: "additional", input: strings.Replace(validator.valid, "}", `,"extra":"value"}`, 1)},
				{name: "case variant summary", input: strings.Replace(validator.valid, `"summary"`, `"Summary"`, 1)},
				{name: "case variant discussion", input: strings.Replace(validator.valid, `"forecast_discussion"`, `"Forecast_Discussion"`, 1)},
				{name: "duplicate", input: strings.Replace(validator.valid, `"summary":"Summary",`, `"summary":"Summary","summary":"Other",`, 1)},
			} {
				t.Run(test.name, func(t *testing.T) {
					err := validator.validate([]byte(test.input))
					if (err == nil) != test.valid {
						t.Fatalf("validate(%s) error = %v, want valid = %t", test.input, err, test.valid)
					}
				})
			}
		})
	}
}

func TestTypedValidatorsMatchEmbeddedSchemaObjectShape(t *testing.T) {
	validators := []struct {
		name     string
		valid    string
		validate func([]byte) error
	}{
		{name: "daily", valid: `{"summary":"Summary","forecast_discussion":["Discussion"],"precipitation_timing":""}`, validate: validateDailyJSON},
		{name: "today", valid: `{"summary":"Summary","forecast_discussion":["Discussion"],"precipitation_timing":""}`, validate: validateTodayJSON},
		{name: "tomorrow", valid: `{"summary":"Summary","forecast_discussion":["Discussion"],"precipitation_timing":""}`, validate: validateTomorrowJSON},
		{name: "hourly", valid: `{"summary":"Summary","forecast_discussion":"Discussion","precipitation_timing":""}`, validate: validateHourlyJSON},
	}
	for _, validator := range validators {
		t.Run(validator.name, func(t *testing.T) {
			schema := generatedTextSchema(t, validator.name)
			for _, test := range []struct {
				name  string
				input string
			}{
				{name: "canonical", input: validator.valid},
				{name: "missing", input: strings.Replace(validator.valid, `"summary":"Summary",`, "", 1)},
				{name: "additional", input: strings.Replace(validator.valid, "}", `,"extra":"value"}`, 1)},
				{name: "case variant", input: strings.Replace(validator.valid, `"summary"`, `"Summary"`, 1)},
				{name: "null", input: strings.Replace(validator.valid, `"precipitation_timing":""`, `"precipitation_timing":null`, 1)},
				{name: "wrong type", input: strings.Replace(validator.valid, `"summary":"Summary"`, `"summary":false`, 1)},
			} {
				t.Run(test.name, func(t *testing.T) {
					instance, err := jsonschema.UnmarshalJSON(strings.NewReader(test.input))
					if err != nil {
						t.Fatalf("parse schema instance: %v", err)
					}
					schemaValid := schema.Validate(instance) == nil
					typedValid := validator.validate([]byte(test.input)) == nil
					if schemaValid != typedValid {
						t.Fatalf("schema valid = %t, typed valid = %t for %s", schemaValid, typedValid, test.input)
					}
				})
			}
		})
	}
}

func TestTypedValidatorsMatchEmbeddedSchemaContentLimits(t *testing.T) {
	validators := []struct {
		name     string
		valid    string
		invalid  []string
		validate func([]byte) error
	}{
		{
			name:     "daily",
			valid:    fmt.Sprintf(`{"summary":%q,"forecast_discussion":[%q],"precipitation_timing":""}`, strings.Repeat("s", 4_000), strings.Repeat("d", 4_000)),
			invalid:  []string{fmt.Sprintf(`{"summary":%q,"forecast_discussion":["discussion"],"precipitation_timing":""}`, strings.Repeat("s", 4_001)), fmt.Sprintf(`{"summary":"summary","forecast_discussion":[%s],"precipitation_timing":""}`, strings.TrimSuffix(strings.Repeat(`"discussion",`, 13), ",")), `{"summary":"summary","forecast_discussion":[null],"precipitation_timing":""}`},
			validate: validateDailyJSON,
		},
		{
			name:     "today",
			valid:    fmt.Sprintf(`{"summary":%q,"forecast_discussion":[%q],"precipitation_timing":""}`, strings.Repeat("s", 4_000), strings.Repeat("d", 4_000)),
			invalid:  []string{fmt.Sprintf(`{"summary":"summary","forecast_discussion":[%q],"precipitation_timing":""}`, strings.Repeat("d", 4_001))},
			validate: validateTodayJSON,
		},
		{
			name:     "tomorrow",
			valid:    fmt.Sprintf(`{"summary":%q,"forecast_discussion":[%q],"precipitation_timing":""}`, strings.Repeat("s", 4_000), strings.Repeat("d", 4_000)),
			invalid:  []string{fmt.Sprintf(`{"summary":"summary","forecast_discussion":["discussion"],"precipitation_timing":%q}`, strings.Repeat("t", 4_001))},
			validate: validateTomorrowJSON,
		},
		{
			name:     "hourly",
			valid:    fmt.Sprintf(`{"summary":%q,"forecast_discussion":%q,"precipitation_timing":""}`, strings.Repeat("s", 4_000), strings.Repeat("d", 12_000)),
			invalid:  []string{fmt.Sprintf(`{"summary":"summary","forecast_discussion":%q,"precipitation_timing":""}`, strings.Repeat("d", 12_001))},
			validate: validateHourlyJSON,
		},
	}
	for _, validator := range validators {
		t.Run(validator.name, func(t *testing.T) {
			schema := generatedTextSchema(t, validator.name)
			for _, input := range append([]string{validator.valid}, validator.invalid...) {
				instance, err := jsonschema.UnmarshalJSON(strings.NewReader(input))
				if err != nil {
					t.Fatalf("parse schema instance: %v", err)
				}
				schemaValid := schema.Validate(instance) == nil
				typedValid := validator.validate([]byte(input)) == nil
				if schemaValid != typedValid {
					t.Fatalf("schema valid = %t, typed valid = %t", schemaValid, typedValid)
				}
			}
		})
	}
}

func generatedTextSchema(t *testing.T, id string) *jsonschema.Schema {
	t.Helper()
	data, err := promptassets.Schema(id)
	if err != nil {
		t.Fatalf("Schema(%q) error = %v", id, err)
	}
	document, err := jsonschema.UnmarshalJSON(bytes.NewReader(data))
	if err != nil {
		t.Fatalf("parse schema %q: %v", id, err)
	}
	compiler := jsonschema.NewCompiler()
	location := fmt.Sprintf("https://weatherreporter.example/schemas/%s.json", id)
	if err := compiler.AddResource(location, document); err != nil {
		t.Fatalf("add schema %q: %v", id, err)
	}
	schema, err := compiler.Compile(location)
	if err != nil {
		t.Fatalf("compile schema %q: %v", id, err)
	}
	return schema
}

func validateDailyJSON(data []byte) error {
	_, err := ValidateDaily(data)
	return err
}

func validateTodayJSON(data []byte) error {
	_, err := ValidateToday(data)
	return err
}

func validateTomorrowJSON(data []byte) error {
	_, err := ValidateTomorrow(data)
	return err
}

func validateHourlyJSON(data []byte) error {
	_, err := ValidateHourly(data)
	return err
}