Fix #755

pdfcpu · Dec 13, 2023 · e33b502 · e33b502
1 parent e3358c4
commit e33b502
Show file tree

Hide file tree

Showing 8 changed files with 80 additions and 29 deletions.
diff --git a/pkg/pdfcpu/crypto.go b/pkg/pdfcpu/crypto.go
@@ -1087,7 +1087,7 @@ func decryptBytes(b []byte, objNr, genNr int, encKey []byte, needAES bool, r int
 // decryptString decrypts s using RC4 or AES.
 func decryptString(s string, objNr, genNr int, key []byte, needAES bool, r int) ([]byte, error) {
 
-	bb, err := types.Unescape(s, true)
+	bb, err := types.Unescape(s)
 	if err != nil {
 		return nil, err
 	}

diff --git a/pkg/pdfcpu/model/dereference.go b/pkg/pdfcpu/model/dereference.go
@@ -377,7 +377,7 @@ func (xRefTable *XRefTable) DereferenceStringEntryBytes(d types.Dict, key string
 
 	switch o := o.(type) {
 	case types.StringLiteral:
-		bb, err := types.Unescape(o.Value(), false)
+		bb, err := types.Unescape(o.Value())
 		if err != nil {
 			return nil, err
 		}

diff --git a/pkg/pdfcpu/model/xreftable.go b/pkg/pdfcpu/model/xreftable.go
@@ -1563,7 +1563,7 @@ func (xRefTable *XRefTable) IDFirstElement() (id []byte, err error) {
 		return nil, errors.New("pdfcpu: ID must contain hex literals or string literals")
 	}
 
-	bb, err := types.Unescape(sl.Value(), false)
+	bb, err := types.Unescape(sl.Value())
 	if err != nil {
 		return nil, err
 	}

diff --git a/pkg/pdfcpu/types/dict.go b/pkg/pdfcpu/types/dict.go
@@ -561,7 +561,7 @@ func (d Dict) StringEntryBytes(key string) ([]byte, error) {
 
 	s := d.StringLiteralEntry(key)
 	if s != nil {
-		bb, err := Unescape(s.Value(), false)
+		bb, err := Unescape(s.Value())
 		if err != nil {
 			return nil, err
 		}

diff --git a/pkg/pdfcpu/types/string.go b/pkg/pdfcpu/types/string.go
@@ -107,10 +107,7 @@ func regularChar(c byte, esc bool) bool {
 }
 
 // Unescape resolves all escape sequences of s.
-func Unescape(s string, enc bool) ([]byte, error) {
-
-	// TODO remove "enc" parameter.
-
+func Unescape(s string) ([]byte, error) {
 	var esc bool
 	var longEol bool
 	var octalCode string
@@ -129,6 +126,21 @@ func Unescape(s string, enc bool) ([]byte, error) {
 			}
 		}
 
+		if len(octalCode) > 0 {
+			if strings.ContainsRune("01234567", rune(c)) {
+				octalCode = octalCode + string(c)
+				if len(octalCode) == 3 {
+					b.WriteByte(ByteForOctalString(octalCode))
+					octalCode = ""
+					esc = false
+				}
+				continue
+			}
+			b.WriteByte(ByteForOctalString(octalCode))
+			octalCode = ""
+			esc = false
+		}
+
 		if regularChar(c, esc) {
 			b.WriteByte(c)
 			continue
@@ -149,19 +161,6 @@ func Unescape(s string, enc bool) ([]byte, error) {
 
 		// escaped = true && any other than \
 
-		if len(octalCode) > 0 {
-			if !strings.ContainsRune("01234567", rune(c)) {
-				return nil, errors.Errorf("Unescape: illegal octal sequence detected %X", octalCode)
-			}
-			octalCode = octalCode + string(c)
-			if len(octalCode) == 3 {
-				b.WriteByte(ByteForOctalString(octalCode))
-				octalCode = ""
-				esc = false
-			}
-			continue
-		}
-
 		// Ignore \eol line breaks.
 		if c == 0x0A {
 			esc = false
@@ -189,6 +188,10 @@ func Unescape(s string, enc bool) ([]byte, error) {
 		esc = false
 	}
 
+	if len(octalCode) > 0 {
+		b.WriteByte(ByteForOctalString(octalCode))
+	}
+
 	return b.Bytes(), nil
 }
 

diff --git a/pkg/pdfcpu/types/string_test.go b/pkg/pdfcpu/types/string_test.go
@@ -17,6 +17,7 @@ limitations under the License.
 package types
 
 import (
+	"bytes"
 	"testing"
 )
 
@@ -52,9 +53,56 @@ func TestByteForOctalString(t *testing.T) {
 	}
 	for _, test := range tests {
 		t.Run(test.input, func(t *testing.T) {
-			actual := ByteForOctalString(test.input)
-			if actual != test.expected {
-				t.Errorf("got %x; want %x", test.expected, actual)
+			got := ByteForOctalString(test.input)
+			if got != test.expected {
+				t.Errorf("got %x; want %x", got, test.expected)
+			}
+		})
+	}
+}
+
+func TestUnescapeStringWithOctal(t *testing.T) {
+	tests := []struct {
+		input    string
+		expected []byte
+	}{
+		{
+			"\\5",
+			[]byte{0x05},
+		},
+		{
+			"\\5a",
+			[]byte{0x05, 'a'},
+		},
+		{
+			"\\5\\5",
+			[]byte{0x05, 0x05},
+		},
+		{
+			"\\53",
+			[]byte{'+'},
+		},
+		{
+			"\\53a",
+			[]byte{'+', 'a'},
+		},
+		{
+			"\\053",
+			[]byte{'+'},
+		},
+		{
+			"\\0053",
+			[]byte{0x05, '3'},
+		},
+	}
+	for _, test := range tests {
+		t.Run(test.input, func(t *testing.T) {
+			got, err := Unescape(test.input)
+			if err != nil {
+				t.Fail()
+			}
+			if !bytes.Equal(got, test.expected) {
+				t.Errorf("got %x; want %x", got, test.expected)
 			}
 		})
 	}
@@ -116,12 +164,12 @@ func TestDecodeName(t *testing.T) {
 	}
 	for _, test := range tests {
 		t.Run(test.input, func(t *testing.T) {
-			actual, err := DecodeName(test.input)
+			got, err := DecodeName(test.input)
 			if err != nil {
 				t.Fail()
 			}
-			if actual != test.expected {
-				t.Errorf("got %x; want %x", test.expected, actual)
+			if got != test.expected {
+				t.Errorf("got %x; want %x", got, test.expected)
 			}
 		})
 	}

diff --git a/pkg/pdfcpu/types/utf16.go b/pkg/pdfcpu/types/utf16.go
@@ -127,7 +127,7 @@ func EscapeUTF16String(s string) (*string, error) {
 
 // StringLiteralToString returns the best possible string rep for a string literal.
 func StringLiteralToString(sl StringLiteral) (string, error) {
-	bb, err := Unescape(sl.Value(), false)
+	bb, err := Unescape(sl.Value())
 	if err != nil {
 		return "", err
 	}

diff --git a/pkg/pdfcpu/writeImage.go b/pkg/pdfcpu/writeImage.go
@@ -127,7 +127,7 @@ func colorLookupTable(xRefTable *model.XRefTable, o types.Object) ([]byte, error
 	switch o := o.(type) {
 
 	case types.StringLiteral:
-		return types.Unescape(o.Value(), false)
+		return types.Unescape(o.Value())
 
 	case types.HexLiteral:
 		return o.Bytes()