diff options
Diffstat (limited to 'generic')
| -rw-r--r-- | generic/tclCmdMZ.c | 46 | ||||
| -rw-r--r-- | generic/tclCompCmdsSZ.c | 7 | ||||
| -rw-r--r-- | generic/tclInt.h | 8 | ||||
| -rw-r--r-- | generic/tclUtil.c | 8 |
4 files changed, 57 insertions, 12 deletions
diff --git a/generic/tclCmdMZ.c b/generic/tclCmdMZ.c index d477216..9b136b4 100644 --- a/generic/tclCmdMZ.c +++ b/generic/tclCmdMZ.c @@ -18,7 +18,6 @@ #include "tclInt.h" #include "tclRegexp.h" -#include "tclStringTrim.h" static inline Tcl_Obj * During(Tcl_Interp *interp, int resultCode, Tcl_Obj *oldOptions, Tcl_Obj *errorInfo); @@ -32,6 +31,39 @@ static int TryPostHandler(ClientData data[], Tcl_Interp *interp, int result); static int UniCharIsAscii(int character); static int UniCharIsHexDigit(int character); + +/* + * Default set of characters to trim in [string trim] and friends. This is a + * UTF-8 literal string containing all Unicode space characters [TIP #413] + */ + +const char tclDefaultTrimSet[] = + "\x09\x0a\x0b\x0c\x0d " /* ASCII */ + "\xc0\x80" /* nul (U+0000) */ + "\xc2\x85" /* next line (U+0085) */ + "\xc2\xa0" /* non-breaking space (U+00a0) */ + "\xe1\x9a\x80" /* ogham space mark (U+1680) */ + "\xe1\xa0\x8e" /* mongolian vowel separator (U+180e) */ + "\xe2\x80\x80" /* en quad (U+2000) */ + "\xe2\x80\x81" /* em quad (U+2001) */ + "\xe2\x80\x82" /* en space (U+2002) */ + "\xe2\x80\x83" /* em space (U+2003) */ + "\xe2\x80\x84" /* three-per-em space (U+2004) */ + "\xe2\x80\x85" /* four-per-em space (U+2005) */ + "\xe2\x80\x86" /* six-per-em space (U+2006) */ + "\xe2\x80\x87" /* figure space (U+2007) */ + "\xe2\x80\x88" /* punctuation space (U+2008) */ + "\xe2\x80\x89" /* thin space (U+2009) */ + "\xe2\x80\x8a" /* hair space (U+200a) */ + "\xe2\x80\x8b" /* zero width space (U+200b) */ + "\xe2\x80\xa8" /* line separator (U+2028) */ + "\xe2\x80\xa9" /* paragraph separator (U+2029) */ + "\xe2\x80\xaf" /* narrow no-break space (U+202f) */ + "\xe2\x81\x9f" /* medium mathematical space (U+205f) */ + "\xe2\x81\xa0" /* word joiner (U+2060) */ + "\xe3\x80\x80" /* ideographic space (U+3000) */ + "\xef\xbb\xbf" /* zero width no-break space (U+feff) */ +; /* *---------------------------------------------------------------------- @@ -3158,8 +3190,8 @@ StringTrimCmd( if (objc == 3) { string2 = TclGetStringFromObj(objv[2], &length2); } else if (objc == 2) { - string2 = DEFAULT_TRIM_SET; - length2 = strlen(DEFAULT_TRIM_SET); + string2 = tclDefaultTrimSet; + length2 = strlen(tclDefaultTrimSet); } else { Tcl_WrongNumArgs(interp, 1, objv, "string ?chars?"); return TCL_ERROR; @@ -3206,8 +3238,8 @@ StringTrimLCmd( if (objc == 3) { string2 = TclGetStringFromObj(objv[2], &length2); } else if (objc == 2) { - string2 = DEFAULT_TRIM_SET; - length2 = strlen(DEFAULT_TRIM_SET); + string2 = tclDefaultTrimSet; + length2 = strlen(tclDefaultTrimSet); } else { Tcl_WrongNumArgs(interp, 1, objv, "string ?chars?"); return TCL_ERROR; @@ -3252,8 +3284,8 @@ StringTrimRCmd( if (objc == 3) { string2 = TclGetStringFromObj(objv[2], &length2); } else if (objc == 2) { - string2 = DEFAULT_TRIM_SET; - length2 = strlen(DEFAULT_TRIM_SET); + string2 = tclDefaultTrimSet; + length2 = strlen(tclDefaultTrimSet); } else { Tcl_WrongNumArgs(interp, 1, objv, "string ?chars?"); return TCL_ERROR; diff --git a/generic/tclCompCmdsSZ.c b/generic/tclCompCmdsSZ.c index 0f2790f..9f97e8c 100644 --- a/generic/tclCompCmdsSZ.c +++ b/generic/tclCompCmdsSZ.c @@ -17,7 +17,6 @@ #include "tclInt.h" #include "tclCompile.h" -#include "tclStringTrim.h" /* * Prototypes for procedures defined later in this file: @@ -810,7 +809,7 @@ TclCompileStringTrimLCmd( tokenPtr = TokenAfter(tokenPtr); CompileWord(envPtr, tokenPtr, interp, 2); } else { - PushLiteral(envPtr, DEFAULT_TRIM_SET, strlen(DEFAULT_TRIM_SET)); + PushLiteral(envPtr, tclDefaultTrimSet, strlen(tclDefaultTrimSet)); } OP( STR_TRIM_LEFT); return TCL_OK; @@ -838,7 +837,7 @@ TclCompileStringTrimRCmd( tokenPtr = TokenAfter(tokenPtr); CompileWord(envPtr, tokenPtr, interp, 2); } else { - PushLiteral(envPtr, DEFAULT_TRIM_SET, strlen(DEFAULT_TRIM_SET)); + PushLiteral(envPtr, tclDefaultTrimSet, strlen(tclDefaultTrimSet)); } OP( STR_TRIM_RIGHT); return TCL_OK; @@ -866,7 +865,7 @@ TclCompileStringTrimCmd( tokenPtr = TokenAfter(tokenPtr); CompileWord(envPtr, tokenPtr, interp, 2); } else { - PushLiteral(envPtr, DEFAULT_TRIM_SET, strlen(DEFAULT_TRIM_SET)); + PushLiteral(envPtr, tclDefaultTrimSet, strlen(tclDefaultTrimSet)); } OP( STR_TRIM); return TCL_OK; diff --git a/generic/tclInt.h b/generic/tclInt.h index 3aaa30b..daeeb81 100644 --- a/generic/tclInt.h +++ b/generic/tclInt.h @@ -2649,6 +2649,14 @@ MODULE_SCOPE Tcl_ScaleTimeProc *tclScaleTimeProcPtr; MODULE_SCOPE ClientData tclTimeClientData; /* + * Default set of characters to trim in [string trim] and friends. This is a + * UTF-8 literal string containing all Unicode space characters [TIP #413] + */ + +MODULE_SCOPE const char tclDefaultTrimSet[]; + + +/* * Variables denoting the Tcl object types defined in the core. */ diff --git a/generic/tclUtil.c b/generic/tclUtil.c index 2d00adf..65d01f7 100644 --- a/generic/tclUtil.c +++ b/generic/tclUtil.c @@ -14,7 +14,6 @@ #include "tclInt.h" #include "tclParse.h" -#include "tclStringTrim.h" #include <math.h> /* @@ -1768,6 +1767,13 @@ TclTrimLeft( *---------------------------------------------------------------------- */ +/* + * The whitespace trimming set used when [concat]enating. This is a subset of + * tclDefaultTrimSet, and deliberately so. + */ + +#define CONCAT_TRIM_SET " \f\v\r\t\n" + /* The whitespace characters trimmed during [concat] operations */ #define CONCAT_WS_SIZE (int) (sizeof(CONCAT_TRIM_SET "") - 1) |