chore: add File normalizer and utilize in runner

file.go

@@ -5,13 +5,35 @@ import (
 	"encoding/json"
 	"errors"
 	"io"
+	"mime"
+	"path/filepath"
 	"strings"
 	"sync"
 	"sync/atomic"
+	"time"
 )
 
+var initMime = func() func() {
+	var once sync.Once
+	return func() {
+		once.Do(func() {
+			_ = mime.AddExtensionType(".xml", "application/xml")
+			_ = mime.AddExtensionType(".jsonld", "application/ld+json")
+			_ = mime.AddExtensionType(".jsonnet", "application/jsonnet")
+			_ = mime.AddExtensionType(".yaml", "text/yaml")
+			_ = mime.AddExtensionType(".yml", "text/yaml")
+		})
+	}
+}()
+
+const contentTypeOctetStream = "application/octet-stream"
+
+var nowFn = time.Now
+
 // File represents a response that is a response body. The runner is in charge
-// of getting the contents to the destination. The metadata will be received.
+// of getting the contents to the destination. The metadata will be received. One
+// note, we call NormalizeFile on the File's in the Runner's that execute the handler.
+// Testing through the Run function will illustrate this.
 type File struct {
 	ContentType string        `json:"content_type"`
 	Encoding    string        `json:"encoding"`
@@ -25,6 +47,20 @@ func (f File) MarshalJSON() ([]byte, error) {
 	return json.Marshal(alias(f))
 }
 
+// NormalizeFile normalizes a file so that all fields are set with sane defaults.
+func NormalizeFile(f File) File {
+	if f.ContentType == "" {
+		f.ContentType = normalizeContentType(f.Filename)
+	}
+	if f.Encoding == "" {
+		f.Encoding = normalizeEncoding(f.Filename)
+	}
+	if f.Filename == "" {
+		f.Filename = normalizeFilename(f.ContentType, f.Encoding, nowFn())
+	}
+	return f
+}
+
 // CompressGzip compresses a files contents with gzip compression.
 func CompressGzip(file File) File {
 	switch {
@@ -99,3 +135,68 @@ func (c *compressorGzip) Close() error {
 
 	return errors.Join(errs...)
 }
+
+func normalizeContentType(filename string) string {
+	initMime()
+	parts := strings.Split(filepath.Base(filename), ".")
+	if len(parts) < 2 {
+		return contentTypeOctetStream
+	}
+	for _, ext := range parts[1:] {
+		if ct := mime.TypeByExtension("." + ext); ct != "" {
+			return ct
+		}
+	}
+	return contentTypeOctetStream
+}
+
+func normalizeEncoding(filename string) string {
+	parts := strings.SplitN(filepath.Base(filename), ".", 2)
+	if len(parts) == 1 {
+		return ""
+	}
+
+	mapping := map[string]string{
+		"br":  "brotli",
+		"gz":  "gzip",
+		"zst": "zstd",
+	}
+	var out []string
+	for _, part := range strings.Split(parts[1], ".") {
+		if encoding, ok := mapping[part]; ok {
+			out = append(out, encoding)
+		}
+	}
+	return strings.Join(out, ", ")
+}
+
+var compressionToExt = map[string]string{
+	"brotli": "br",
+	"gzip":   "gz",
+	"zstd":   "zst",
+}
+
+func normalizeFilename(contentType, encoding string, now time.Time) string {
+	filename := "upload_" + now.Format(time.RFC3339)
+	if encoding != "" {
+		var converted []string
+		for _, enc := range strings.Split(strings.ReplaceAll(encoding, " ", ""), ",") {
+			if ext := compressionToExt[enc]; ext != "" {
+				converted = append(converted, ext)
+			}
+		}
+		if len(converted) > 0 {
+			encoding = "." + strings.Join(converted, ".")
+		}
+	}
+
+	var ctExt string
+	if contentType != contentTypeOctetStream {
+		initMime()
+		exts, _ := mime.ExtensionsByType(contentType)
+		if len(exts) > 0 {
+			ctExt = exts[0]
+		}
+	}
+	return filename + ctExt + encoding
+}

file_internal_test.go

@@ -0,0 +1,272 @@
+package fdk
+
+import (
+	"testing"
+	"time"
+)
+
+func TestNormalizeFile(t *testing.T) {
+	var start = time.Time{}.UTC().Add(time.Hour)
+
+	tests := []struct {
+		name  string
+		nowFn func() time.Time
+		in    File
+		want  File
+	}{
+		{
+			name: "file with all values set should remain the same",
+			in: File{
+				ContentType: "application/json",
+				Encoding:    "gzip",
+				Filename:    "test.json",
+			},
+			want: File{
+				ContentType: "application/json",
+				Encoding:    "gzip",
+				Filename:    "test.json",
+			},
+		},
+		{
+			name: "file missing encoding should remain the same",
+			in: File{
+				ContentType: "application/json",
+				Filename:    "test.json",
+			},
+			want: File{
+				ContentType: "application/json",
+				Filename:    "test.json",
+			},
+		},
+		{
+			name: "file missing content type for filename with json extension should add content type to application/ld+json",
+			in: File{
+				Filename: "test.jsonld",
+			},
+			want: File{
+				ContentType: "application/ld+json",
+				Filename:    "test.jsonld",
+			},
+		},
+		{
+			name: "file missing content type for filename with jsonld extension should add content type to application/json",
+			in: File{
+				Filename: "test.json",
+			},
+			want: File{
+				ContentType: "application/json",
+				Filename:    "test.json",
+			},
+		},
+		{
+			name: "file missing content type for filename with jsonld extension should add content type to application/json",
+			in: File{
+				Filename: "test.js",
+			},
+			want: File{
+				ContentType: "text/javascript; charset=utf-8",
+				Filename:    "test.js",
+			},
+		},
+		{
+			name: "file missing content type for filename with xml extension should add content type to application/xml",
+			in: File{
+				Filename: "test.xml",
+			},
+			want: File{
+				ContentType: "application/xml",
+				Filename:    "test.xml",
+			},
+		},
+		{
+			name: "file missing content type for filename with tar extension should add content type to application/x-tar",
+			in: File{
+				Filename: "test.tar",
+			},
+			want: File{
+				ContentType: "application/x-tar",
+				Filename:    "test.tar",
+			},
+		},
+		{
+			name: "file missing content type for filename with zip extension should add content type to application/zip",
+			in: File{
+				Filename: "test.zip",
+			},
+			want: File{
+				ContentType: "application/zip",
+				Filename:    "test.zip",
+			},
+		},
+		{
+			name: "file missing content type for filename with html extension should add content type to text/html",
+			in: File{
+				Filename: "test.html",
+			},
+			want: File{
+				ContentType: "text/html; charset=utf-8",
+				Filename:    "test.html",
+			},
+		},
+		{
+			name: "file missing content type for filename with yaml extension should add content type to text/yaml",
+			in: File{
+				Filename: "test.yaml",
+			},
+			want: File{
+				ContentType: "text/yaml; charset=utf-8",
+				Filename:    "test.yaml",
+			},
+		},
+		{
+			name: "file missing content type for filename with yml extension should add content type to text/yaml",
+			in: File{
+				Filename: "test.yml",
+			},
+			want: File{
+				ContentType: "text/yaml; charset=utf-8",
+				Filename:    "test.yml",
+			},
+		},
+		{
+			name: "file missing content type for filename with txt extension should add content type to text/plain",
+			in: File{
+				Filename: "test.txt",
+			},
+			want: File{
+				ContentType: "text/plain; charset=utf-8",
+				Filename:    "test.txt",
+			},
+		},
+		{
+			name: "file missing content encoding for gzipped json filename should add content type and encoding",
+			in: File{
+				Filename: "test.json.gz",
+			},
+			want: File{
+				ContentType: "application/json",
+				Encoding:    "gzip",
+				Filename:    "test.json.gz",
+			},
+		},
+		{
+			name: "file missing content encoding for gzipped yaml filename should add content type and encoding",
+			in: File{
+				Filename: "test.yaml.gz",
+			},
+			want: File{
+				ContentType: "text/yaml; charset=utf-8",
+				Encoding:    "gzip",
+				Filename:    "test.yaml.gz",
+			},
+		},
+		{
+			name: "file missing content type and filename should set filename to timestamp",
+			nowFn: func() time.Time {
+				return start
+			},
+			in: File{},
+			want: File{
+				ContentType: "application/octet-stream",
+				Filename:    "upload_" + start.Format(time.RFC3339),
+			},
+		},
+		{
+			name: "file missing filename with content type json set filename to timestamp.json file",
+			nowFn: func() time.Time {
+				return start
+			},
+			in: File{
+				ContentType: "application/json",
+			},
+			want: File{
+				ContentType: "application/json",
+				Filename:    "upload_" + start.Format(time.RFC3339) + ".json",
+			},
+		},
+		{
+			name: "file missing filename with content type json and gz encoding set filename to timestamp.json.gz file",
+			nowFn: func() time.Time {
+				return start
+			},
+			in: File{
+				ContentType: "application/json",
+				Encoding:    "gzip",
+			},
+			want: File{
+				ContentType: "application/json",
+				Encoding:    "gzip",
+				Filename:    "upload_" + start.Format(time.RFC3339) + ".json.gz",
+			},
+		},
+		{
+			name: "file missing filename with content type json and gz zstd encodings set filename to timestamp.json.gz file",
+			nowFn: func() time.Time {
+				return start
+			},
+			in: File{
+				ContentType: "application/json",
+				Encoding:    "zstd, gzip",
+			},
+			want: File{
+				ContentType: "application/json",
+				Encoding:    "zstd, gzip",
+				Filename:    "upload_" + start.Format(time.RFC3339) + ".json.zst.gz",
+			},
+		},
+		{
+			name: "file missing filename with content type plain/javascript and gz zstd encodings set filename to timestamp.json.gz file",
+			nowFn: func() time.Time {
+				return start
+			},
+			in: File{
+				ContentType: "text/javascript; charset=utf-8",
+				Encoding:    "zstd, gzip",
+			},
+			want: File{
+				ContentType: "text/javascript; charset=utf-8",
+				Encoding:    "zstd, gzip",
+				Filename:    "upload_" + start.Format(time.RFC3339) + ".js.zst.gz",
+			},
+		},
+		{
+			name: "file missing filename with content type xml and gz zstd encodings set filename to timestamp.json.gz file",
+			nowFn: func() time.Time {
+				return start
+			},
+			in: File{
+				ContentType: "application/xml",
+				Encoding:    "zstd, gzip",
+			},
+			want: File{
+				ContentType: "application/xml",
+				Encoding:    "zstd, gzip",
+				Filename:    "upload_" + start.Format(time.RFC3339) + ".xml.zst.gz",
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if tt.nowFn != nil {
+				old := nowFn
+				nowFn = tt.nowFn
+				t.Cleanup(func() { nowFn = old })
+			}
+
+			got := NormalizeFile(tt.in)
+			EqualVals(t, tt.want.Filename, got.Filename)
+			EqualVals(t, tt.want.ContentType, got.ContentType)
+			EqualVals(t, tt.want.Encoding, got.Encoding)
+		})
+	}
+}
+
+func EqualVals[T comparable](t testing.TB, want, got T) bool {
+	t.Helper()
+
+	match := want == got
+	if !match {
+		t.Errorf("values not equal:\n\t\twant:\t%#v\n\t\tgot:\t%#v", want, got)
+	}
+	return match
+}