diff options
| -rw-r--r-- | generic/tclStringObj.c | 44 | ||||
| -rw-r--r-- | tests/string.test | 8 |
2 files changed, 45 insertions, 7 deletions
diff --git a/generic/tclStringObj.c b/generic/tclStringObj.c index 33b2139..bdc9c99 100644 --- a/generic/tclStringObj.c +++ b/generic/tclStringObj.c @@ -2899,6 +2899,9 @@ TclStringReverse( { String *stringPtr; Tcl_UniChar ch = 0; +#if TCL_UTF_MAX <= 4 + int needFlip = 0; +#endif if (TclIsPureByteArray(objPtr)) { int numBytes; @@ -2917,10 +2920,9 @@ TclStringReverse( if (stringPtr->hasUnicode) { Tcl_UniChar *from = Tcl_GetUnicode(objPtr); Tcl_UniChar *src = from + stringPtr->numChars; + Tcl_UniChar *to; if (Tcl_IsShared(objPtr)) { - Tcl_UniChar *to; - /* * Create a non-empty, pure unicode value, so we can coax * Tcl_SetObjLength into growing the unicode rep buffer. @@ -2930,19 +2932,54 @@ TclStringReverse( Tcl_SetObjLength(objPtr, stringPtr->numChars); to = Tcl_GetUnicode(objPtr); while (--src >= from) { +#if TCL_UTF_MAX <= 4 + ch = *src; + if ((ch & 0xF800) == 0xD800) { + needFlip = 1; + } + *to++ = ch; +#else *to++ = *src; +#endif } } else { /* * Reversing in place. */ +#if TCL_UTF_MAX <= 4 + to = src; +#endif while (--src > from) { ch = *src; +#if TCL_UTF_MAX <= 4 + if ((ch & 0xF800) == 0xD800) { + needFlip = 1; + } +#endif *src = *from; *from++ = ch; } } +#if TCL_UTF_MAX <= 4 + if (needFlip) { + /* + * Flip back surrogate pairs. + */ + + from = to - stringPtr->numChars; + while (--to >= from) { + ch = *to; + if ((ch & 0xFC00) == 0xD800) { + if ((to-1 >= from) && ((to[-1] & 0xFC00) == 0xDC00)) { + to[0] = to[-1]; + to[-1] = ch; + --to; + } + } + } + } +#endif } if (objPtr->bytes) { @@ -2968,6 +3005,7 @@ TclStringReverse( int charCount = 0; int bytesLeft = numBytes; + int chw; while (bytesLeft) { /* @@ -2976,7 +3014,7 @@ TclStringReverse( * skip calling Tcl_UtfCharComplete() here. */ - int bytesInChar = TclUtfToUniChar(from, &ch); + int bytesInChar = TclUtfToUCS4(from, &chw); ReverseBytes((unsigned char *)to, (unsigned char *)from, bytesInChar); diff --git a/tests/string.test b/tests/string.test index 3ac3060..66eb6ad 100644 --- a/tests/string.test +++ b/tests/string.test @@ -1828,19 +1828,19 @@ test string-24.15 {string reverse command - pure bytearray} { binary scan [tcl::string::reverse [binary format H* 010203]] H* x set x } 030201 -test string-24.16 {string reverse command - surrogates} knownBug { +test string-24.16 {string reverse command - surrogates} { string reverse \u0444bulb\uD83D\uDE02 } \uD83D\uDE02blub\u0444 -test string-24.17 {string reverse command - surrogates} knownBug { +test string-24.17 {string reverse command - surrogates} { string reverse \uD83D\uDE02hello\uD83D\uDE02 } \uD83D\uDE02olleh\uD83D\uDE02 -test string-24.18 {string reverse command - surrogates} knownBug { +test string-24.18 {string reverse command - surrogates} { set s \u0444bulb\uD83D\uDE02 # shim shimmery ... string index $s 0 string reverse $s } \uD83D\uDE02blub\u0444 -test string-24.19 {string reverse command - surrogates} knownBug { +test string-24.19 {string reverse command - surrogates} { set s \uD83D\uDE02hello\uD83D\uDE02 # shim shimmery ... string index $s 0 |