From 1889ded1144a4dbd44d0c6f03e72a01d70115a51 Mon Sep 17 00:00:00 2001 From: "jan.nijtmans" Date: Sat, 11 Mar 2023 22:00:29 +0000 Subject: Proposed fix for [db7a085bd9]: encoding convertfrom -strict utf-16 accepts partial surrogates. TODO: testcases, and implement for 8.7 too --- generic/tclCmdAH.c | 2 +- generic/tclEncoding.c | 30 ++++++++++++++++++++++++++---- 2 files changed, 27 insertions(+), 5 deletions(-) diff --git a/generic/tclCmdAH.c b/generic/tclCmdAH.c index 4df1216..ac504d0 100644 --- a/generic/tclCmdAH.c +++ b/generic/tclCmdAH.c @@ -514,7 +514,7 @@ EncodingConvertfromObjCmd( char buf[TCL_INTEGER_SPACE]; sprintf(buf, "%" TCL_Z_MODIFIER "u", result); Tcl_SetObjResult(interp, Tcl_ObjPrintf("unexpected byte sequence starting at index %" - TCL_Z_MODIFIER "u: '\\x%X'", result, UCHAR(bytesPtr[result]))); + TCL_Z_MODIFIER "u: '\\x%02X'", result, UCHAR(bytesPtr[result]))); Tcl_SetErrorCode(interp, "TCL", "ENCODING", "ILLEGALSEQUENCE", buf, NULL); Tcl_DStringFree(&ds); diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c index fc3397a..4f334bb 100644 --- a/generic/tclEncoding.c +++ b/generic/tclEncoding.c @@ -2603,6 +2603,7 @@ Utf32ToUtfProc( dst += Tcl_UniCharToUtf(-1, dst); } #endif + if ((unsigned)ch > 0x10FFFF) { ch = 0xFFFD; if (STOPONERROR) { @@ -2639,6 +2640,7 @@ Utf32ToUtfProc( dst += Tcl_UniCharToUtf(-1, dst); } #endif + if ((flags & TCL_ENCODING_END) && (result == TCL_CONVERT_MULTIBYTE)) { /* We have a single byte left-over at the end */ if (dst > dstEnd) { @@ -2846,6 +2848,13 @@ Utf16ToUtfProc( ch = (src[0] & 0xFF) << 8 | (src[1] & 0xFF); } if (((prev & ~0x3FF) == 0xD800) && ((ch & ~0x3FF) != 0xDC00)) { + if (((flags & TCL_ENCODING_STRICT) == TCL_ENCODING_STRICT)) { + result = TCL_CONVERT_UNKNOWN; + src -= 2; /* Go back to before the high surrogate */ + dst--; /* Also undo writing a single byte too much */ + numChars--; + break; + } /* Bug [10c2c17c32]. If Hi surrogate not followed by Lo surrogate, finish 3-byte UTF-8 */ dst += Tcl_UniCharToUtf(-1, dst); } @@ -2855,17 +2864,30 @@ Utf16ToUtfProc( * unsigned short-size data. */ - if (ch && ch < 0x80) { + if ((unsigned)ch - 1 < 0x7F) { *dst++ = (ch & 0xFF); - } else { + } else if (((prev & ~0x3FF) == 0xD800) || ((ch & ~0x3FF) == 0xD800)) { dst += Tcl_UniCharToUtf(ch | TCL_COMBINE, dst); + } else if (((ch & ~0x3FF) == 0xDC00) && ((flags & TCL_ENCODING_STRICT) == TCL_ENCODING_STRICT)) { + /* Lo surrogate not preceded by Hi surrogate */ + result = TCL_CONVERT_UNKNOWN; + break; + } else { + dst += Tcl_UniCharToUtf(ch, dst); } src += sizeof(unsigned short); } if ((ch & ~0x3FF) == 0xD800) { - /* Bug [10c2c17c32]. If Hi surrogate, finish 3-byte UTF-8 */ - dst += Tcl_UniCharToUtf(-1, dst); + if ((flags & TCL_ENCODING_STRICT) == TCL_ENCODING_STRICT) { + result = TCL_CONVERT_UNKNOWN; + src -= 2; + dst--; + numChars--; + } else { + /* Bug [10c2c17c32]. If Hi surrogate, finish 3-byte UTF-8 */ + dst += Tcl_UniCharToUtf(-1, dst); + } } if ((flags & TCL_ENCODING_END) && (result == TCL_CONVERT_MULTIBYTE)) { /* We have a single byte left-over at the end */ -- cgit v0.12