[feaLib]Escape nameid strings when writing feature files

https://github.com/fonttools/fonttools/issues/780

Lib/fontTools/feaLib/ast.py

@@ -1,6 +1,8 @@
 from __future__ import print_function, division, absolute_import
 from __future__ import unicode_literals
+from fontTools.misc.py23 import *
 from fontTools.feaLib.error import FeatureLibError
+from fontTools.misc.encodingTools import getEncoding
 from collections import OrderedDict
 import itertools
 
@@ -1059,10 +1061,27 @@ class NameRecord(Statement):
             self.platEncID, self.langID, self.string)
 
     def asFea(self, indent=""):
-        plat = simplify_name_attributes(self.platformID, self.platEncID, self.langID)
+        def escape(c, escape_pattern):
+            # Also escape U+0022 QUOTATION MARK and U+005C REVERSE SOLIDUS
+            if c >= 0x20 and c <= 0x7E and c not in (0x22, 0x5C):
+                return unichr(c)
+            else:
+                return escape_pattern % c
+        encoding = getEncoding(self.platformID, self.platEncID, self.langID)
+        if encoding is None:
+            raise FeatureLibError("Unsupported encoding", self.location)
+        s = tobytes(self.string, encoding=encoding)
+        if encoding == "utf_16_be":
+            escaped_string = "".join([
+                escape(byteord(s[i]) * 256 + byteord(s[i + 1]), r"\%04x")
+                for i in range(0, len(s), 2)])
+        else:
+            escaped_string = "".join([escape(byteord(b), r"\%02x") for b in s])
+        plat = simplify_name_attributes(
+            self.platformID, self.platEncID, self.langID)
         if plat != "":
             plat += " "
-        return "nameid {} {}\"{}\";".format(self.nameID, plat, self.string)
+        return "nameid {} {}\"{}\";".format(self.nameID, plat, escaped_string)
 
 
 class FeatureNameStatement(NameRecord):

Tests/feaLib/data/spec9e.fea

@@ -1,6 +1,4 @@
 table name {
-#test-fea2fea: nameid 9 "Joachim Müller-Lancé";
     nameid 9 "Joachim M\00fcller-Lanc\00e9";    # Windows (Unicode)
-#test-fea2fea: nameid 9 1 "Joachim Müller-Lancé";
     nameid 9 1 "Joachim M\9fller-Lanc\8e";      # Macintosh (Mac Roman)
 } name;

Tests/feaLib/lexer_test.py

@@ -87,7 +87,15 @@ class LexerTest(unittest.TestCase):
                          [(Lexer.STRING, "foo"), (Lexer.STRING, "bar")])
         self.assertEqual(lex('"foo \nbar\r baz \r\nqux\n\n "'),
                          [(Lexer.STRING, "foo bar baz qux ")])
-        self.assertRaises(FeatureLibError, lambda: lex('"foo\n bar'))
+        # The lexer should preserve escape sequences because they have
+        # different interpretations depending on context. For better
+        # or for worse, that is how the OpenType Feature File Syntax
+        # has been specified; see section 9.e (name table) for examples.
+        self.assertEqual(lex(r'"M\00fcller-Lanc\00e9"'),  # 'nameid 9'
+                         [(Lexer.STRING, r"M\00fcller-Lanc\00e9")])
+        self.assertEqual(lex(r'"M\9fller-Lanc\8e"'),  # 'nameid 9 1'
+                         [(Lexer.STRING, r"M\9fller-Lanc\8e")])
+        self.assertRaises(FeatureLibError, lex, '"foo\n bar')
 
     def test_bad_character(self):
         self.assertRaises(FeatureLibError, lambda: lex("123 \u0001"))

Tests/feaLib/parser_test.py

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 from __future__ import print_function, division, absolute_import
 from __future__ import unicode_literals
 from fontTools.feaLib.error import FeatureLibError
@@ -901,6 +902,69 @@ class ParserTest(unittest.TestCase):
         self.assertEqual(mc.glyphSet(), ("acute", "grave"))
         self.assertEqual((mc.anchor.x, mc.anchor.y), (350, 3))
 
+    def test_nameid_windows_utf16(self):
+        doc = self.parse(
+            r'table name { nameid 9 "M\00fcller-Lanc\00e9"; } name;')
+        name = doc.statements[0].statements[0]
+        self.assertIsInstance(name, ast.NameRecord)
+        self.assertEquals(name.nameID, 9)
+        self.assertEquals(name.platformID, 3)
+        self.assertEquals(name.platEncID, 1)
+        self.assertEquals(name.langID, 0x0409)
+        self.assertEquals(name.string, "Müller-Lancé")
+        self.assertEquals(name.asFea(), r'nameid 9 "M\00fcller-Lanc\00e9";')
+
+    def test_nameid_windows_utf16_backslash(self):
+        doc = self.parse(r'table name { nameid 9 "Back\005cslash"; } name;')
+        name = doc.statements[0].statements[0]
+        self.assertEquals(name.string, r"Back\slash")
+        self.assertEquals(name.asFea(), r'nameid 9 "Back\005cslash";')
+
+    def test_nameid_windows_utf16_quotation_mark(self):
+        doc = self.parse(
+            r'table name { nameid 9 "Quotation \0022Mark\0022"; } name;')
+        name = doc.statements[0].statements[0]
+        self.assertEquals(name.string, 'Quotation "Mark"')
+        self.assertEquals(name.asFea(), r'nameid 9 "Quotation \0022Mark\0022";')
+
+    def test_nameid_windows_utf16_surroates(self):
+        pass
+        # TODO: https://github.com/fonttools/fonttools/issues/842
+        # doc = self.parse(r'table name { nameid 9 "Carrot \D83E\DD55"; } name;')
+        # name = doc.statements[0].statements[0]
+        # self.assertEquals(name.string, r"Carrot 🥕")
+        # self.assertEquals(name.asFea(), r'nameid 9 "Carrot \d83e\dd55";')
+
+    def test_nameid_mac_roman(self):
+        doc = self.parse(
+            r'table name { nameid 9 1 "Joachim M\9fller-Lanc\8e"; } name;')
+        name = doc.statements[0].statements[0]
+        self.assertIsInstance(name, ast.NameRecord)
+        self.assertEquals(name.nameID, 9)
+        self.assertEquals(name.platformID, 1)
+        self.assertEquals(name.platEncID, 0)
+        self.assertEquals(name.langID, 0)
+        self.assertEquals(name.string, "Joachim Müller-Lancé")
+        self.assertEquals(name.asFea(),
+                          r'nameid 9 1 "Joachim M\9fller-Lanc\8e";')
+
+    def test_nameid_mac_croatian(self):
+        doc = self.parse(
+            r'table name { nameid 9 1 0 18 "Jovica Veljovi\e6"; } name;')
+        name = doc.statements[0].statements[0]
+        self.assertEquals(name.nameID, 9)
+        self.assertEquals(name.platformID, 1)
+        self.assertEquals(name.platEncID, 0)
+        self.assertEquals(name.langID, 18)
+        # TODO: https://github.com/fonttools/fonttools/issues/842
+        # self.assertEquals(name.string, "Jovica Veljović")
+        # self.assertEquals(name.asFea(), r'nameid 9 1 0 18 "Jovica Veljovi\e6";')
+
+    def test_nameid_unsupported_platform(self):
+        self.assertRaisesRegex(
+            FeatureLibError, "Expected platform id 1 or 3",
+            self.parse, 'table name { nameid 9 666 "Foo"; } name;')
+
     def test_rsub_format_a(self):
         doc = self.parse("feature test {rsub a [b B] c' d [e E] by C;} test;")
         rsub = doc.statements[0].statements[0]