media: Also consider extension in FromContent

As used in `resources.GetRemote`.

This will now reject image files with text and text files with images.

hugolib/resource_chain_test.go

@@ -22,16 +22,15 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"os"
-
-	"github.com/gohugoio/hugo/config"
-
-	"github.com/gohugoio/hugo/resources/resource_transformers/tocss/dartsass"
-
 	"path/filepath"
 	"strings"
 	"testing"
 	"time"
 
+	"github.com/gohugoio/hugo/config"
+
+	"github.com/gohugoio/hugo/resources/resource_transformers/tocss/dartsass"
+
 	jww "github.com/spf13/jwalterweatherman"
 
 	"github.com/gohugoio/hugo/common/herrors"
@@ -57,7 +56,6 @@ func TestSCSSWithIncludePaths(t *testing.T) {
 		{"libsass", func() bool { return scss.Supports() }},
 		{"dartsass", func() bool { return dartsass.Supports() }},
 	} {
-
 		c.Run(test.name, func(c *qt.C) {
 			if !test.supports() {
 				c.Skip(fmt.Sprintf("Skip %s", test.name))
@@ -107,9 +105,7 @@ T1: {{ $r.Content }}
 
 			b.AssertFileContent(filepath.Join(workDir, "public/index.html"), `T1: moo{color:#fff}`)
 		})
-
 	}
-
 }
 
 func TestSCSSWithRegularCSSImport(t *testing.T) {
@@ -122,7 +118,6 @@ func TestSCSSWithRegularCSSImport(t *testing.T) {
 		{"libsass", func() bool { return scss.Supports() }},
 		{"dartsass", func() bool { return dartsass.Supports() }},
 	} {
-
 		c.Run(test.name, func(c *qt.C) {
 			if !test.supports() {
 				c.Skip(fmt.Sprintf("Skip %s", test.name))
@@ -202,11 +197,9 @@ moo {
 }
 
 /* foo */`)
-
 			}
 		})
 	}
-
 }
 
 func TestSCSSWithThemeOverrides(t *testing.T) {
@@ -219,7 +212,6 @@ func TestSCSSWithThemeOverrides(t *testing.T) {
 		{"libsass", func() bool { return scss.Supports() }},
 		{"dartsass", func() bool { return dartsass.Supports() }},
 	} {
-
 		c.Run(test.name, func(c *qt.C) {
 			if !test.supports() {
 				c.Skip(fmt.Sprintf("Skip %s", test.name))
@@ -319,7 +311,6 @@ T1: {{ $r.Content }}
 			)
 		})
 	}
-
 }
 
 // https://github.com/gohugoio/hugo/issues/6274
@@ -333,7 +324,6 @@ func TestSCSSWithIncludePathsSass(t *testing.T) {
 		{"libsass", func() bool { return scss.Supports() }},
 		{"dartsass", func() bool { return dartsass.Supports() }},
 	} {
-
 		c.Run(test.name, func(c *qt.C) {
 			if !test.supports() {
 				c.Skip(fmt.Sprintf("Skip %s", test.name))
@@ -620,6 +610,7 @@ func TestResourceChains(t *testing.T) {
 			return
 
 		case "/authenticated/":
+			w.Header().Set("Content-Type", "text/plain")
 			if r.Header.Get("Authorization") == "Bearer abcd" {
 				w.Write([]byte(`Welcome`))
 				return
@@ -628,6 +619,7 @@ func TestResourceChains(t *testing.T) {
 			return
 
 		case "/post":
+			w.Header().Set("Content-Type", "text/plain")
 			if r.Method == http.MethodPost {
 				body, err := ioutil.ReadAll(r.Body)
 				if err != nil {

media/mediaType.go

@@ -28,6 +28,8 @@ import (
 	"github.com/mitchellh/mapstructure"
 )
 
+var zero Type
+
 const (
 	defaultDelimiter = "."
 )
@@ -64,16 +66,14 @@ type SuffixInfo struct {
 // FromContent resolve the Type primarily using http.DetectContentType.
 // If http.DetectContentType resolves to application/octet-stream, a zero Type is returned.
 // If http.DetectContentType  resolves to text/plain or application/xml, we try to get more specific using types and ext.
-func FromContent(types Types, ext string, content []byte) Type {
-	ext = strings.TrimPrefix(ext, ".")
+func FromContent(types Types, extensionHints []string, content []byte) Type {
 	t := strings.Split(http.DetectContentType(content), ";")[0]
-	var m Type
 	if t == "application/octet-stream" {
-		return m
+		return zero
 	}
 
 	var found bool
-	m, found = types.GetByType(t)
+	m, found := types.GetByType(t)
 	if !found {
 		if t == "text/xml" {
 			// This is how it's configured in Hugo by default.
@@ -81,19 +81,36 @@ func FromContent(types Types, ext string, content []byte) Type {
 		}
 	}
 
-	if !found || ext == "" {
+	if !found {
+		return zero
+	}
+
+	var mm Type
+
+	for _, extension := range extensionHints {
+		extension = strings.TrimPrefix(extension, ".")
+		mm, _, found = types.GetFirstBySuffix(extension)
+		if found {
+			break
+		}
+	}
+
+	if found {
+		if m == mm {
 			return m
 		}
 
-	if m.Type() == "text/plain" || m.Type() == "application/xml" {
+		if m.IsText() && mm.IsText() {
 			// http.DetectContentType isn't brilliant when it comes to common text formats, so we need to do better.
 			// For now we say that if it's detected to be a text format and the extension/content type in header reports
 			// it to be a text format, then we use that.
-		mm, _, found := types.GetFirstBySuffix(ext)
-		if found && mm.IsText() {
 			return mm
 		}
+
+		// E.g. an image with a *.js extension.
+		return zero
 	}
+
 	return m
 }
 

media/mediaType_test.go

@@ -15,7 +15,6 @@ package media
 
 import (
 	"encoding/json"
-	"fmt"
 	"io/ioutil"
 	"path/filepath"
 	"sort"
@@ -194,15 +193,39 @@ func TestFromContent(t *testing.T) {
 			content, err := ioutil.ReadFile(filename)
 			c.Assert(err, qt.IsNil)
 			ext := strings.TrimPrefix(paths.Ext(filename), ".")
-			fmt.Println("=>", ext)
+			var exts []string
+			if ext == "jpg" {
+				exts = append(exts, "foo", "bar", "jpg")
+			} else {
+				exts = []string{ext}
+			}
 			expected, _, found := mtypes.GetFirstBySuffix(ext)
 			c.Assert(found, qt.IsTrue)
-			got := FromContent(mtypes, ext, content)
+			got := FromContent(mtypes, exts, content)
 			c.Assert(got, qt.Equals, expected)
 		})
 	}
 }
 
+func TestFromContentFakes(t *testing.T) {
+	c := qt.New(t)
+
+	files, err := filepath.Glob("./testdata/fake.*")
+	c.Assert(err, qt.IsNil)
+	mtypes := DefaultTypes
+
+	for _, filename := range files {
+		name := filepath.Base(filename)
+		c.Run(name, func(c *qt.C) {
+			content, err := ioutil.ReadFile(filename)
+			c.Assert(err, qt.IsNil)
+			ext := strings.TrimPrefix(paths.Ext(filename), ".")
+			got := FromContent(mtypes, []string{ext}, content)
+			c.Assert(got, qt.Equals, zero)
+		})
+	}
+}
+
 func TestDecodeTypes(t *testing.T) {
 	c := qt.New(t)
 

media/testdata/fake.png

@@ -0,0 +1,3 @@
+function foo() {
+    return "foo";
+}

resources/resource_factories/create/remote.go

@@ -110,21 +110,30 @@ func (c *Client) FromRemote(uri string, optionsm map[string]interface{}) (resour
 		}
 	}
 
-	var extensionHint string
+	var extensionHints []string
 
-	if arr, _ := mime.ExtensionsByType(res.Header.Get("Content-Type")); len(arr) == 1 {
-		extensionHint = arr[0]
+	contentType := res.Header.Get("Content-Type")
+
+	// mime.ExtensionsByType gives a long list of extensions for text/plain,
+	// just use ".txt".
+	if strings.HasPrefix(contentType, "text/plain") {
+		extensionHints = []string{".txt"}
+	} else {
+		exts, _ := mime.ExtensionsByType(contentType)
+		if exts != nil {
+			extensionHints = exts
+		}
 	}
 
-	// Look for a file extention
-	if extensionHint == "" {
+	// Look for a file extention. If it's .txt, look for a more specific.
+	if extensionHints == nil || extensionHints[0] == ".txt" {
 		if ext := path.Ext(filename); ext != "" {
-			extensionHint = ext
+			extensionHints = []string{ext}
 		}
 	}
 
 	// Now resolve the media type primarily using the content.
-	mediaType := media.FromContent(c.rs.MediaTypes, extensionHint, body)
+	mediaType := media.FromContent(c.rs.MediaTypes, extensionHints, body)
 	if mediaType.IsZero() {
 		return nil, errors.Errorf("failed to resolve media type for remote resource %q", uri)
 	}
@@ -140,7 +149,6 @@ func (c *Client) FromRemote(uri string, optionsm map[string]interface{}) (resour
 			},
 			RelTargetFilename: filepath.Clean(resourceID),
 		})
-
 }
 
 func (c *Client) validateFromRemoteArgs(uri string, options fromRemoteOptions) error {
@@ -213,7 +221,7 @@ func (o fromRemoteOptions) BodyReader() io.Reader {
 }
 
 func decodeRemoteOptions(optionsm map[string]interface{}) (fromRemoteOptions, error) {
-	var options = fromRemoteOptions{
+	options := fromRemoteOptions{
 		Method: "GET",
 	}
 
@@ -224,5 +232,4 @@ func decodeRemoteOptions(optionsm map[string]interface{}) (fromRemoteOptions, er
 	options.Method = strings.ToUpper(options.Method)
 
 	return options, nil
-
 }