[unicode] Consolidate surrogate range and max code point checks

osh/word_compile.py

@@ -136,17 +136,23 @@ def EvalSingleQuoted(id_, tokens):
             # EvalCStringToken() redoes some of this work, but right now it's
             # shared with dynamic echo -e / printf, which don't have tokens.
 
-            if t.id == Id.Char_Unicode8:  # check for invalid \U00110000
+            i = -1
+            if t.id in (Id.Char_Unicode4, Id.Char_Unicode8):
                 s = lexer.TokenSliceLeft(t, 2)
                 i = int(s, 16)
-                if i > 0x10ffff:
-                    p_die("Code point can't be greater than U+10ffff", t)
 
-            elif t.id == Id.Char_UBraced:  # check for invalid \u{110000}
+            elif t.id == Id.Char_UBraced:
                 s = lexer.TokenSlice(t, 3, -1)
                 i = int(s, 16)
+
+            if i != -1:
+                # check for invalid \U00110000 or \u{110000}
                 if i > 0x10ffff:
                     p_die("Code point can't be greater than U+10ffff", t)
+                if 0xD800 <= i and i < 0xE000:
+                    p_die(
+                        r"%s escape is illegal because it's in the surrogate range"
+                        % lexer.TokenVal(t), t)
 
             strs.append(EvalCStringToken(t.id, lexer.TokenVal(t)))
 

osh/word_parse.py

@@ -736,14 +736,6 @@ def ReadSingleQuoted(self, lex_mode, left_token, out_tokens, is_ysh_expr):
                 p_die(
                     r"%s escapes not allowed in u'' strings" %
                     lexer.TokenVal(tok), tok)
-            # \u{dc00} isn't valid
-            if tok.id == Id.Char_UBraced:
-                h = lexer.TokenSlice(tok, 3, -1)  # \u{123456}
-                i = int(h, 16)
-                if 0xD800 <= i and i < 0xE000:
-                    p_die(
-                        r"%s escape is illegal because it's in the surrogate range"
-                        % lexer.TokenVal(tok), tok)
 
         out_tokens.extend(tokens)
         return self.cur_token

spec/unicode.test.sh

@@ -107,6 +107,42 @@ py-repr $'\U00110000'
 '\xf4\x90\x80\x80'
 ## END
 
+#### $'' checks for surrogate range at parse time
+
+case $SH in mksh) exit ;; esac
+
+$SH << 'EOF'
+x=$'\udc00'
+EOF
+if test $? -ne 0; then
+  echo pass
+else
+  echo fail
+fi
+
+$SH << 'EOF'
+x=$'\U0000dc00'
+EOF
+if test $? -ne 0; then
+  echo pass
+else
+  echo fail
+fi
+
+
+## STDOUT:
+pass
+pass
+## END
+
+## BUG bash STDOUT:
+fail
+fail
+## END
+
+## BUG mksh STDOUT:
+## END
+
 
 #### printf / echo -e check that 0x110000 is too big at runtime
 case $SH in mksh) exit ;; esac