feat: support URL input for /v1/validate

README.md

@@ -26,6 +26,13 @@ curl -X QUERY "http://localhost:3000/v1/validate" \
 --data-binary $'publiccodeYmlVersion: "0.5"\ndevelopmentStatus: stable\n [...]\n'
 ```
 
+or by URL query parameter:
+
+```console
+curl -G -X QUERY "http://localhost:3000/v1/validate" \
+--data-urlencode "url=https://example.com/publiccode.yml"
+```
+
 ### Example response (valid publiccode.yml)
 
 ```json

internal/handlers/validate.go

@@ -3,6 +3,12 @@
 import (
 	"bytes"
 	"errors"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"strings"
+	"time"
 
 	"github.com/gofiber/fiber/v3"
 	publiccodeParser "github.com/italia/publiccode-parser-go/v5"
@@ -12,6 +18,7 @@
 type PubliccodeymlValidatorHandler struct {
 	parser               *publiccodeParser.Parser
 	parserExternalChecks *publiccodeParser.Parser
+	httpClient           *http.Client
 }
 
 func NewPubliccodeymlValidatorHandler() *PubliccodeymlValidatorHandler {
@@ -25,10 +32,14 @@
 		panic("can't create a publiccode.yml parser: " + err.Error())
 	}
 
-	return &PubliccodeymlValidatorHandler{parser: parser, parserExternalChecks: parserExternalChecks}
+	return &PubliccodeymlValidatorHandler{
+		parser:               parser,
+		parserExternalChecks: parserExternalChecks,
+		httpClient:           &http.Client{Timeout: 10 * time.Second},
+	}
 }
 
 func (vh *PubliccodeymlValidatorHandler) Query(ctx fiber.Ctx) error {
 	var normalized *string
 
 	valid := true
@@ -38,13 +49,24 @@
 		parser = vh.parserExternalChecks
 	}
 
-	if len(ctx.Body()) == 0 {
-		return common.Error(fiber.StatusBadRequest, "empty body", "need a body to validate")
+	input := ctx.Body()
+	if len(input) == 0 {
+		rawURL := strings.TrimSpace(fiber.Query[string](ctx, "url", ""))
+		if rawURL == "" {
+			return common.Error(fiber.StatusBadRequest, "empty body", "need a body to validate")
+		}
+
+		content, err := vh.fetchURL(rawURL)
+		if err != nil {
+			return err
+		}
+
+		input = content
 	}
 
 	results := make(publiccodeParser.ValidationResults, 0)
 
-	reader := bytes.NewReader(ctx.Body())
+	reader := bytes.NewReader(input)
 
 	parsed, err := parser.ParseStream(reader)
 	if err != nil {
@@ -72,3 +94,44 @@
 	//nolint:wrapcheck
 	return ctx.JSON(fiber.Map{"valid": valid, "results": results, "normalized": normalized})
 }
+
+func (vh *PubliccodeymlValidatorHandler) fetchURL(rawURL string) ([]byte, error) {
+	parsedURL, err := url.ParseRequestURI(rawURL)
+	if err != nil || parsedURL.Host == "" {
+		return nil, common.Error(fiber.StatusBadRequest, "invalid url", "query parameter 'url' must be a valid http(s) URL")
+	}
+
+	if parsedURL.Scheme != "http" && parsedURL.Scheme != "https" {
+		return nil, common.Error(fiber.StatusBadRequest, "invalid url", "query parameter 'url' must use http or https")
+	}
+
+	req, err := http.NewRequest(http.MethodGet, parsedURL.String(), nil)
+	if err != nil {
+		return nil, common.Error(fiber.StatusBadRequest, "invalid url", "query parameter 'url' is invalid")
+	}
+
+	resp, err := vh.httpClient.Do(req)
+	if err != nil {
+		return nil, common.Error(fiber.StatusBadRequest, "url fetch failed", fmt.Sprintf("failed to fetch URL: %v", err))
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode < http.StatusOK || resp.StatusCode >= http.StatusMultipleChoices {
+		return nil, common.Error(
+			fiber.StatusBadRequest,
+			"url fetch failed",
+			fmt.Sprintf("failed to fetch URL: HTTP %d", resp.StatusCode),
+		)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return nil, common.Error(fiber.StatusBadRequest, "url fetch failed", "failed to read response body")
+	}
+
+	if len(body) == 0 {
+		return nil, common.Error(fiber.StatusBadRequest, "empty body", "the URL response is empty")
+	}
+
+	return body, nil
+}

main_test.go

@@ -5,6 +5,7 @@ import (
 	"io"
 	"log"
 	"net/http"
+	"net/http/httptest"
 	"net/url"
 	"os"
 	"strings"
@@ -129,6 +130,23 @@ func TestApi(t *testing.T) {
 func TestValidateEndpoint(t *testing.T) {
 	validYml := loadTestdata(t, "valid.publiccode.yml")
 	invalidYml := loadTestdata(t, "invalid.publiccode.yml")
+	sourceServer := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		switch r.URL.Path {
+		case "/valid.publiccode.yml":
+			_, _ = w.Write([]byte(validYml))
+		case "/invalid.publiccode.yml":
+			_, _ = w.Write([]byte(invalidYml))
+		case "/empty.publiccode.yml":
+			// Return 200 with an empty response body
+		default:
+			w.WriteHeader(http.StatusNotFound)
+		}
+	}))
+	defer sourceServer.Close()
+
+	validYmlURL := url.QueryEscape(sourceServer.URL + "/valid.publiccode.yml")
+	emptyYmlURL := url.QueryEscape(sourceServer.URL + "/empty.publiccode.yml")
+	notFoundYmlURL := url.QueryEscape(sourceServer.URL + "/missing.publiccode.yml")
 
 	tests := []TestCase{
 		{
@@ -167,6 +185,40 @@ func TestValidateEndpoint(t *testing.T) {
 				assert.Nil(t, response["normalized"])
 			},
 		},
+		{
+			description:         "validate: valid file from URL query parameter",
+			query:               "QUERY /v1/validate?url=" + validYmlURL,
+			expectedCode:        200,
+			expectedContentType: "application/json; charset=utf-8",
+			validateFunc: func(t *testing.T, response map[string]any) {
+				assert.Equal(t, true, response["valid"])
+				results, ok := response["results"].([]any)
+				assert.True(t, ok)
+				assert.Len(t, results, 0)
+				assert.NotNil(t, response["normalized"])
+			},
+		},
+		{
+			description:         "validate: invalid URL query parameter",
+			query:               "QUERY /v1/validate?url=not-a-url",
+			expectedCode:        400,
+			expectedBody:        `{"title":"invalid url","detail":"query parameter 'url' must be a valid http(s) URL","status":400}`,
+			expectedContentType: "application/problem+json",
+		},
+		{
+			description:         "validate: URL query parameter returns empty body",
+			query:               "QUERY /v1/validate?url=" + emptyYmlURL,
+			expectedCode:        400,
+			expectedBody:        `{"title":"empty body","detail":"the URL response is empty","status":400}`,
+			expectedContentType: "application/problem+json",
+		},
+		{
+			description:         "validate: URL query parameter returns non-2xx",
+			query:               "QUERY /v1/validate?url=" + notFoundYmlURL,
+			expectedCode:        400,
+			expectedBody:        `{"title":"url fetch failed","detail":"failed to fetch URL: HTTP 404","status":400}`,
+			expectedContentType: "application/problem+json",
+		},
 	}
 
 	runTestCases(t, tests)