shithub: hugo

Info  •  Files  •  Log  •  Branches

source: Normalize file name to NFC

Fixes #2203

--- a/source/filesystem.go

+++ b/source/filesystem.go

@@ -18,9 +18,11 @@

 	"os"

 	"path/filepath"

 	"regexp"

+	"runtime"

 	"strings"

 	"github.com/spf13/hugo/hugofs"

+	"golang.org/x/text/unicode/norm"

 	"github.com/spf13/viper"

@@ -65,6 +67,11 @@

 // add populates a file in the Filesystem.files

 func (f *Filesystem) add(name string, reader io.Reader) (err error) {

 	var file *File

+

+	if runtime.GOOS == "darwin" {

+		// When a file system is HFS+, its filepath is in NFD form.

+		name = norm.NFC.String(name)

+	}

 	file, err = NewFileFromAbs(f.Base, name, reader)

--- a/source/filesystem_test.go

+++ b/source/filesystem_test.go

@@ -16,6 +16,8 @@

 import (

 	"bytes"

 	"path/filepath"

+	"runtime"

+	"strings"

 	"testing"

 )

@@ -81,4 +83,29 @@

 			}

 		}

 	}

+}

+

+func TestUnicodeNorm(t *testing.T) {

+	if runtime.GOOS != "darwin" {

+		// Normalization code is only for Mac OS, since it is not necessary for other OSes.

+		return

+	}

+

+	paths := []struct {

+		NFC string

+		NFD string

+	}{

+		{NFC: "å", NFD: "\x61\xcc\x8a"},

+		{NFC: "é", NFD: "\x65\xcc\x81"},

+	}

+

+	for _, path := range paths {

+		src := new(Filesystem)

+		_ = src.add(path.NFD, strings.NewReader(""))

+		f := src.Files()[0]

+		if f.BaseFileName() != path.NFC {

+			t.Fatalf("file name in NFD form should be normalized (%s)", path.NFC)

+		}

+	}

+

 }