fix: special characters issues with filenames (#736)

pkg/modules/api/context.go

@@ -12,14 +12,11 @@ import (
 	"path/filepath"
 	"strings"
 	"time"
-	"unicode"
 
 	"github.com/google/uuid"
 	"github.com/labstack/echo/v4"
 	"github.com/mholt/archiver/v3"
 	"go.uber.org/zap"
-	"golang.org/x/text/runes"
-	"golang.org/x/text/transform"
 	"golang.org/x/text/unicode/norm"
 
 	"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
@@ -124,15 +121,6 @@ func newContext(echoCtx echo.Context, logger *zap.Logger, fs *gotenberg.FileSyst
 	ctx.files = make(map[string]string)
 
 	copyToDisk := func(fh *multipart.FileHeader) error {
-		// Avoid directory traversal and normalize filename.
-		// See https://github.com/gotenberg/gotenberg/issues/104.
-		t := transform.Chain(norm.NFD, runes.Remove(runes.In(unicode.Mn)), norm.NFC)
-
-		filename, _, err := transform.String(t, filepath.Base(fh.Filename))
-		if err != nil {
-			return fmt.Errorf("transform filename: %w", err)
-		}
-
 		in, err := fh.Open()
 		if err != nil {
 			return fmt.Errorf("open multipart file: %w", err)
@@ -145,6 +133,10 @@ func newContext(echoCtx echo.Context, logger *zap.Logger, fs *gotenberg.FileSyst
 			}
 		}()
 
+		// Avoid directory traversal and make sure filename characters are
+		// normalized.
+		// See: https://github.com/gotenberg/gotenberg/issues/662.
+		filename := norm.NFC.String(filepath.Base(fh.Filename))
 		path := fmt.Sprintf("%s/%s", ctx.dirPath, filename)
 
 		out, err := os.Create(path)

pkg/modules/libreoffice/api/libreoffice.go

@@ -4,12 +4,15 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"io"
 	"net"
 	"os"
+	"path/filepath"
 	"sync"
 	"sync/atomic"
 	"time"
 
+	"github.com/google/uuid"
 	"go.uber.org/zap"
 
 	"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
@@ -290,6 +293,11 @@ func (p *libreOfficeProcess) pdf(ctx context.Context, logger *zap.Logger, inputP
 		)
 	}
 
+	inputPath, err := nonBasicLatinCharactersGuard(logger, inputPath)
+	if err != nil {
+		return fmt.Errorf("non-basic latin characters guard: %w", err)
+	}
+
 	args = append(args, "--output", outputPath, inputPath)
 
 	cmd, err := gotenberg.CommandContext(ctx, logger, p.arguments.unoBinPath, args...)
@@ -321,6 +329,65 @@ func (p *libreOfficeProcess) pdf(ctx context.Context, logger *zap.Logger, inputP
 	return fmt.Errorf("convert to PDF: %w", err)
 }
 
+// LibreOffice cannot convert a file with a name containing non-basic Latin
+// characters.
+// See:
+// https://github.com/gotenberg/gotenberg/issues/104
+// https://github.com/gotenberg/gotenberg/issues/730
+func nonBasicLatinCharactersGuard(logger *zap.Logger, inputPath string) (string, error) {
+	hasNonBasicLatinChars := func(str string) bool {
+		for _, r := range str {
+			// Check if the character is outside basic Latin.
+			if r != '.' && (r < ' ' || r > '~') {
+				return true
+			}
+		}
+		return false
+	}
+
+	filename := filepath.Base(inputPath)
+	if !hasNonBasicLatinChars(filename) {
+		logger.Debug("no non-basic latin characters in filename, skip copy")
+		return inputPath, nil
+	}
+
+	logger.Warn("non-basic latin characters in filename, copy to a file with a valid filename")
+	basePath := filepath.Dir(inputPath)
+	ext := filepath.Ext(inputPath)
+	newInputPath := filepath.Join(basePath, fmt.Sprintf("%s%s", uuid.NewString(), ext))
+
+	in, err := os.Open(inputPath)
+	if err != nil {
+		return "", fmt.Errorf("open file: %w", err)
+	}
+
+	defer func() {
+		err := in.Close()
+		if err != nil {
+			logger.Error(fmt.Sprintf("close file: %s", err))
+		}
+	}()
+
+	out, err := os.Create(newInputPath)
+	if err != nil {
+		return "", fmt.Errorf("create new file: %w", err)
+	}
+
+	defer func() {
+		err := out.Close()
+		if err != nil {
+			logger.Error(fmt.Sprintf("close new file: %s", err))
+		}
+	}()
+
+	_, err = io.Copy(out, in)
+	if err != nil {
+		return "", fmt.Errorf("copy file to new file: %w", err)
+	}
+
+	return newInputPath, nil
+}
+
 // Interface guards.
 var (
 	_ gotenberg.Process = (*libreOfficeProcess)(nil)

pkg/modules/libreoffice/api/libreoffice_test.go

@@ -545,7 +545,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
 			err := tc.libreOffice.pdf(
 				ctx,
 				logger,
-				fmt.Sprintf("file://%s/document.txt", tc.fs.WorkingDirPath()),
+				fmt.Sprintf("%s/document.txt", tc.fs.WorkingDirPath()),
 				fmt.Sprintf("%s/%s.pdf", tc.fs.WorkingDirPath(), uuid.NewString()),
 				tc.options,
 			)
@@ -564,3 +564,87 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
 		})
 	}
 }
+
+func TestNonBasicLatinCharactersGuard(t *testing.T) {
+	for _, tc := range []struct {
+		scenario            string
+		fs                  *gotenberg.FileSystem
+		filename            string
+		expectSameInputPath bool
+		expectError         bool
+	}{
+		{
+			scenario: "basic latin characters",
+			fs: func() *gotenberg.FileSystem {
+				fs := gotenberg.NewFileSystem()
+
+				err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
+				if err != nil {
+					t.Fatalf(fmt.Sprintf("expected no error but got: %v", err))
+				}
+
+				err = os.WriteFile(fmt.Sprintf("%s/document.txt", fs.WorkingDirPath()), []byte("Basic latin characters"), 0o755)
+				if err != nil {
+					t.Fatalf("expected no error but got: %v", err)
+				}
+
+				return fs
+			}(),
+			filename:            "document.txt",
+			expectSameInputPath: true,
+			expectError:         false,
+		},
+		{
+			scenario: "non-basic latin characters",
+			fs: func() *gotenberg.FileSystem {
+				fs := gotenberg.NewFileSystem()
+
+				err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
+				if err != nil {
+					t.Fatalf(fmt.Sprintf("expected no error but got: %v", err))
+				}
+
+				err = os.WriteFile(fmt.Sprintf("%s/éèßàùä.txt", fs.WorkingDirPath()), []byte("Non-basic latin characters"), 0o755)
+				if err != nil {
+					t.Fatalf("expected no error but got: %v", err)
+				}
+
+				return fs
+			}(),
+			filename:            "éèßàùä.txt",
+			expectSameInputPath: false,
+			expectError:         false,
+		},
+	} {
+		t.Run(tc.scenario, func(t *testing.T) {
+			defer func() {
+				err := os.RemoveAll(tc.fs.WorkingDirPath())
+				if err != nil {
+					t.Fatalf("expected no error while cleaning up, but got: %v", err)
+				}
+			}()
+
+			inputPath := fmt.Sprintf("%s/%s", tc.fs.WorkingDirPath(), tc.filename)
+			newInputPath, err := nonBasicLatinCharactersGuard(
+				zap.NewNop(),
+				inputPath,
+			)
+
+			if !tc.expectError && err != nil {
+				t.Fatalf("expected no error but got: %v", err)
+			}
+
+			if tc.expectError && err == nil {
+				t.Fatal("expected error but got none")
+			}
+
+			if tc.expectSameInputPath && newInputPath != inputPath {
+				t.Fatalf("expected same input path, but got '%s'", newInputPath)
+			}
+
+			if !tc.expectSameInputPath && newInputPath == inputPath {
+				t.Fatalf("expected different input path, but got same '%s'", newInputPath)
+			}
+		})
+	}
+}