[#188] Improve content-type detector

Signed-off-by: Aleksey Kravchenko <al.kravchenko@yadro.com>
2024-12-24 18:42:02 +03:00 · 2024-12-24 18:42:02 +03:00 · f0c999d9a2
commit f0c999d9a2
parent 1db62f9d95
3 changed files with 87 additions and 20 deletions
--- a/internal/handler/reader_test.go
+++ b/internal/handler/reader_test.go
@ -10,39 +10,80 @@ import (
 	"github.com/stretchr/testify/require"
 )

+const (
+	txtContentType        = "text/plain; charset=utf-8"
+	cssContentType        = "text/css; charset=utf-8"
+	htmlContentType       = "text/html; charset=utf-8"
+	javascriptContentType = "text/javascript; charset=utf-8"
+
+	htmlBody = "<!DOCTYPE html><html ><head><meta charset=\"utf-8\"><title>Test Html</title>"
+)
+
 func TestDetector(t *testing.T) {
-	txtContentType := "text/plain; charset=utf-8"
 	sb := strings.Builder{}
 	for i := 0; i < 10; i++ {
 		sb.WriteString("Some txt content. Content-Type must be detected properly by detector.")
 	}

 	for _, tc := range []struct {
-		Name        string
-		ContentType string
-		Expected    string
+		Name                string
+		ExpectedContentType string
+		Content             string
+		FileName            string
 	}{
 		{
-			Name:        "less than 512b",
-			ContentType: txtContentType,
-			Expected:    sb.String()[:256],
+			Name:                "less than 512b",
+			ExpectedContentType: txtContentType,
+			Content:             sb.String()[:256],
+			FileName:            "test.txt",
 		},
 		{
-			Name:        "more than 512b",
-			ContentType: txtContentType,
-			Expected:    sb.String(),
+			Name:                "more than 512b",
+			ExpectedContentType: txtContentType,
+			Content:             sb.String(),
+			FileName:            "test.txt",
+		},
+		{
+			Name:                "css content type",
+			ExpectedContentType: cssContentType,
+			Content:             sb.String(),
+			FileName:            "test.css",
+		},
+		{
+			Name:                "javascript content type",
+			ExpectedContentType: javascriptContentType,
+			Content:             sb.String(),
+			FileName:            "test.js",
+		},
+		{
+			Name:                "html content type by file content",
+			ExpectedContentType: htmlContentType,
+			Content:             htmlBody,
+			FileName:            "test.detect-by-content",
+		},
+		{
+			Name:                "html content type by file extension",
+			ExpectedContentType: htmlContentType,
+			Content:             sb.String(),
+			FileName:            "test.html",
+		},
+		{
+			Name:                "empty file extension",
+			ExpectedContentType: txtContentType,
+			Content:             sb.String(),
+			FileName:            "test",
 		},
 	} {
 		t.Run(tc.Name, func(t *testing.T) {
-			contentType, data, err := readContentType(uint64(len(tc.Expected)),
+			contentType, data, err := readContentType(uint64(len(tc.Content)),
 				func(uint64) (io.Reader, error) {
-					return strings.NewReader(tc.Expected), nil
-				},
+					return strings.NewReader(tc.Content), nil
+				}, tc.FileName,
 			)

 			require.NoError(t, err)
-			require.Equal(t, tc.ContentType, contentType)
-			require.True(t, strings.HasPrefix(tc.Expected, string(data)))
+			require.Equal(t, tc.ExpectedContentType, contentType)
+			require.True(t, strings.HasPrefix(tc.Content, string(data)))
 		})
 	}
 }