diff options
Diffstat (limited to 'generic/tclParse.c')
| -rw-r--r-- | generic/tclParse.c | 23 |
1 files changed, 18 insertions, 5 deletions
diff --git a/generic/tclParse.c b/generic/tclParse.c index befe208..b5fa16d 100644 --- a/generic/tclParse.c +++ b/generic/tclParse.c @@ -817,7 +817,7 @@ TclParseBackslash( count = 2; switch (*p) { /* - * Note: in the conversions below, use absolute values (e.g., 0xa) + * Note: in the conversions below, use absolute values (e.g., 0xA) * rather than symbolic values (e.g. \n) that get converted by the * compiler. It's possible that compilers on some platforms will do * the symbolic conversions differently, which could result in @@ -831,19 +831,19 @@ TclParseBackslash( result = 0x8; break; case 'f': - result = 0xc; + result = 0xC; break; case 'n': - result = 0xa; + result = 0xA; break; case 'r': - result = 0xd; + result = 0xD; break; case 't': result = 0x9; break; case 'v': - result = 0xb; + result = 0xB; break; case 'x': count += TclParseHex(p+1, (numBytes > 3) ? 2 : numBytes-2, &result); @@ -867,6 +867,16 @@ TclParseBackslash( * No hexdigits -> This is just "u". */ result = 'u'; + } else if (((result & 0xDC00) == 0xD800) && (count == 6) + && (p[5] == '\\') && (p[6] == 'u') && (numBytes >= 10)) { + /* If high surrogate is immediately followed by a low surrogate + * escape, combine them into one character. */ + int low; + int count2 = TclParseHex(p+7, 4, &low); + if ((count2 == 4) && ((low & 0xDC00) == 0xDC00)) { + result = ((result & 0x3FF)<<10 | (low & 0x3FF)) + 0x10000; + count += count2 + 2; + } } break; case 'U': @@ -876,6 +886,9 @@ TclParseBackslash( * No hexdigits -> This is just "U". */ result = 'U'; + } else if ((result | 0x7FF) == 0xDFFF) { + /* Upper or lower surrogate, not allowed in this syntax. */ + result = 0xFFFD; } break; case '\n': |