diff options
-rw-r--r-- | ChangeLog | 8 | ||||
-rw-r--r-- | generic/tclCompCmds.c | 141 |
2 files changed, 44 insertions, 105 deletions
@@ -20,9 +20,17 @@ * tests/parse.test: Tests for expanded literals quoting detection. + * generic/tclCompCmds.c: New TclFindElement() is also a better + fit for the [switch] compiler. + * generic/tclInt.h: Replace TclCountSpaceRuns() with * generic/tclListObj.c: TclMaxListLength() which is the function we * generic/tclUtil.c: actually want. + * generic/tclCompCmds.c: + + * generic/tclCompCmds.c: Rewrite of parts of the switch compiler to + better use the powers of TclFindElement() and do less parsing on + its own. 2011-04-28 Don Porter <dgp@users.sourceforge.net> diff --git a/generic/tclCompCmds.c b/generic/tclCompCmds.c index ddd2242..f2d1bfb 100644 --- a/generic/tclCompCmds.c +++ b/generic/tclCompCmds.c @@ -3907,7 +3907,6 @@ TclCompileSwitchCmd( int savedStackDepth = envPtr->currStackDepth; int noCase; /* Has the -nocase flag been given? */ int foundMode = 0; /* Have we seen a mode flag yet? */ - int isListedArms = 0; int i, valueIndex; DefineLineInformation; /* TIP #280 */ int* clNext = envPtr->clNext; @@ -4047,89 +4046,40 @@ TclCompileSwitchCmd( */ if (numWords == 1) { - Tcl_DString bodyList; - const char **argv = NULL, *tokenStartPtr, *p; + CONST char *bytes; + int maxLen, numBytes; int bline; /* TIP #280: line of the pattern/action list, * and start of list for when tracking the * location. This list comes immediately after * the value we switch on. */ - int isTokenBraced; - - /* - * Test that we've got a suitable body list as a simple (i.e. braced) - * word, and that the elements of the body are simple words too. This - * is really rather nasty indeed. - */ if (tokenPtr->type != TCL_TOKEN_SIMPLE_WORD) { return TCL_ERROR; } + bytes = tokenPtr[1].start; + numBytes = tokenPtr[1].size; - Tcl_DStringInit(&bodyList); - Tcl_DStringAppend(&bodyList, tokenPtr[1].start, tokenPtr[1].size); - if (Tcl_SplitList(NULL, Tcl_DStringValue(&bodyList), &numWords, - &argv) != TCL_OK) { - Tcl_DStringFree(&bodyList); - return TCL_ERROR; - } - Tcl_DStringFree(&bodyList); - - /* - * Now we know what the switch arms are, we've got to see whether we - * can synthesize tokens for the arms. First check whether we've got a - * valid number of arms since we can do that now. - */ - - if (numWords == 0 || numWords % 2) { - ckfree((char *) argv); + /* Allocate enough space to work in. */ + maxLen = TclMaxListLength(bytes, numBytes, NULL); + if (maxLen < 2) { return TCL_ERROR; } - - isListedArms = 1; - bodyTokenArray = (Tcl_Token *) ckalloc(sizeof(Tcl_Token) * numWords); - bodyToken = (Tcl_Token **) ckalloc(sizeof(Tcl_Token *) * numWords); - bodyLines = (int *) ckalloc(sizeof(int) * numWords); - bodyNext = (int **) ckalloc(sizeof(int*) * numWords); - - /* - * Locate the start of the arms within the overall word. - */ + bodyTokenArray = (Tcl_Token *) ckalloc(sizeof(Tcl_Token) * maxLen); + bodyToken = (Tcl_Token **) ckalloc(sizeof(Tcl_Token *) * maxLen); + bodyLines = (int *) ckalloc(sizeof(int) * maxLen); + bodyNext = (int **) ckalloc(sizeof(int*) * maxLen); bline = mapPtr->loc[eclIndex].line[valueIndex+1]; - p = tokenStartPtr = tokenPtr[1].start; - while (isspace(UCHAR(*tokenStartPtr))) { - tokenStartPtr++; - } - if (*tokenStartPtr == '{') { - tokenStartPtr++; - isTokenBraced = 1; - } else { - isTokenBraced = 0; - } + numWords = 0; - /* - * TIP #280: Count lines within the literal list. - */ - - for (i=0 ; i<numWords ; i++) { - bodyTokenArray[i].type = TCL_TOKEN_TEXT; - bodyTokenArray[i].start = tokenStartPtr; - bodyTokenArray[i].size = strlen(argv[i]); - bodyTokenArray[i].numComponents = 0; - bodyToken[i] = bodyTokenArray+i; - tokenStartPtr += bodyTokenArray[i].size; - - /* - * Test to see if we have guessed the end of the word correctly; - * if not, we can't feed the real string to the sub-compilation - * engine, and we're then stuck and so have to punt out to doing - * everything at runtime. - */ + while (numBytes > 0) { + CONST char *prevBytes = bytes; + int literal; - if ((isTokenBraced && *(tokenStartPtr++) != '}') || - (tokenStartPtr < tokenPtr[1].start+tokenPtr[1].size - && !isspace(UCHAR(*tokenStartPtr)))) { - ckfree((char *) argv); + if (TCL_OK != TclFindElement(NULL, bytes, numBytes, + &(bodyTokenArray[numWords].start), &bytes, + &(bodyTokenArray[numWords].size), &literal) || !literal) { + abort: ckfree((char *) bodyToken); ckfree((char *) bodyTokenArray); ckfree((char *) bodyLines); @@ -4137,48 +4087,30 @@ TclCompileSwitchCmd( return TCL_ERROR; } + bodyTokenArray[numWords].type = TCL_TOKEN_TEXT; + bodyTokenArray[numWords].numComponents = 0; + bodyToken[numWords] = bodyTokenArray + numWords; + /* * TIP #280: Now determine the line the list element starts on * (there is no need to do it earlier, due to the possibility of * aborting, see above). */ - TclAdvanceLines(&bline, p, bodyTokenArray[i].start); + TclAdvanceLines(&bline, prevBytes, bodyTokenArray[numWords].start); TclAdvanceContinuations (&bline, &clNext, - bodyTokenArray[i].start - envPtr->source); - bodyLines[i] = bline; - bodyNext[i] = clNext; - p = bodyTokenArray[i].start; - - while (isspace(UCHAR(*tokenStartPtr))) { - tokenStartPtr++; - if (tokenStartPtr >= tokenPtr[1].start+tokenPtr[1].size) { - break; - } - } - if (*tokenStartPtr == '{') { - tokenStartPtr++; - isTokenBraced = 1; - } else { - isTokenBraced = 0; - } + bodyTokenArray[numWords].start - envPtr->source); + bodyLines[numWords] = bline; + bodyNext[numWords] = clNext; + TclAdvanceLines(&bline, bodyTokenArray[numWords].start, bytes); + TclAdvanceContinuations (&bline, &clNext, bytes - envPtr->source); + + numBytes -= (bytes - prevBytes); + numWords++; } - ckfree((char *) argv); - - /* - * Check that we've parsed everything we thought we were going to - * parse. If not, something odd is going on (I believe it is possible - * to defeat the code above) and we should bail out. - */ - - if (tokenStartPtr != tokenPtr[1].start+tokenPtr[1].size) { - ckfree((char *) bodyToken); - ckfree((char *) bodyTokenArray); - ckfree((char *) bodyLines); - ckfree((char *) bodyNext); - return TCL_ERROR; + if (numWords % 2) { + goto abort; } - } else if (numWords % 2 || numWords == 0) { /* * Odd number of words (>1) available, or no words at all available. @@ -4205,8 +4137,7 @@ TclCompileSwitchCmd( * traces, etc. */ - if (tokenPtr->type != TCL_TOKEN_SIMPLE_WORD || - tokenPtr->numComponents != 1) { + if (tokenPtr->type != TCL_TOKEN_SIMPLE_WORD) { ckfree((char *) bodyToken); ckfree((char *) bodyLines); ckfree((char *) bodyNext); @@ -4255,7 +4186,7 @@ TclCompileSwitchCmd( * but it handles the most common case well enough. */ - if (isListedArms && mode == Switch_Exact && !noCase) { + if (mode == Switch_Exact) { JumptableInfo *jtPtr; int infoIndex, isNew, *finalFixups, numRealBodies = 0, jumpLocation; int mustGenerate, jumpToDefault; |