From 2f2d04eac2f4d40cf16733c6803429519a649be4 Mon Sep 17 00:00:00 2001 From: dgp Date: Tue, 6 Sep 2011 17:48:10 +0000 Subject: 3401704 New patch to enable nancy(), influence(), 99bottles(). --- generic/tclCompExpr.c | 81 ++++++++++++++++++++++++--------------------------- 1 file changed, 38 insertions(+), 43 deletions(-) diff --git a/generic/tclCompExpr.c b/generic/tclCompExpr.c index d5300db..077d595 100644 --- a/generic/tclCompExpr.c +++ b/generic/tclCompExpr.c @@ -651,12 +651,6 @@ ParseExpr( Tcl_Obj *literal; /* Filled by the ParseLexeme() call when * a literal is parsed that has a Tcl_Obj * rep worth preserving. */ - const char *lastStart = start - scanned; - /* Compute where the lexeme parsed the - * previous pass through the loop began. - * This is helpful for detecting invalid - * octals and providing more complete error - * messages. */ /* * Each pass through this loop adds up to one more OpNode. Allocate @@ -742,29 +736,30 @@ ParseExpr( " or \"%.*s%s(...)\" or ...", (scanned < limit) ? scanned : limit - 3, start, (scanned < limit) ? "" : "..."); - if (NotOperator(lastParsed)) { - if ((lastStart[0] == '0') - && ((lastStart[1] == 'o') - || (lastStart[1] == 'O')) - && (lastStart[2] >= '0') - && (lastStart[2] <= '9')) { - const char *end = lastStart + 2; - Tcl_Obj* copy; - while (isdigit(UCHAR(*end))) { - end++; - } - copy = Tcl_NewStringObj(lastStart, - end - lastStart); - if (TclCheckBadOctal(NULL, - Tcl_GetString(copy))) { + if (start[0] == '0') { + const char *stop; + TclParseNumber(NULL, NULL, NULL, start, scanned, + &stop, TCL_PARSE_NO_WHITESPACE); + + if (isdigit(UCHAR(*stop)) || (stop == start + 1)) { + parsePtr->errorType = TCL_PARSE_BAD_NUMBER; + + switch (start[1]) { + case 'b': Tcl_AppendToObj(post, - "(invalid octal number?)", -1); + " (invalid binary number?)", -1); + break; + case 'o': + Tcl_AppendToObj(post, + " (invalid octal number?)", -1); + default: + if (isdigit(UCHAR(start[1]))) { + Tcl_AppendToObj(post, + " (invalid octal number?)", -1); + } + break; } - Tcl_DecrRefCount(copy); } - scanned = 0; - insertMark = 1; - parsePtr->errorType = TCL_PARSE_BAD_NUMBER; } goto error; } @@ -809,18 +804,8 @@ ParseExpr( if (NotOperator(lastParsed)) { msg = Tcl_ObjPrintf("missing operator at %s", mark); - if (lastStart[0] == '0') { - Tcl_Obj *copy = Tcl_NewStringObj(lastStart, - start + scanned - lastStart); - if (TclCheckBadOctal(NULL, Tcl_GetString(copy))) { - TclNewLiteralStringObj(post, - "looks like invalid octal number"); - } - Tcl_DecrRefCount(copy); - } scanned = 0; insertMark = 1; - parsePtr->errorType = TCL_PARSE_BAD_NUMBER; /* Free any literal to avoid a memleak. */ if ((lexeme == NUMBER) || (lexeme == BOOLEAN)) { @@ -1934,14 +1919,24 @@ ParseLexeme( literal = Tcl_NewObj(); if (TclParseNumber(NULL, literal, NULL, start, numBytes, &end, TCL_PARSE_NO_WHITESPACE) == TCL_OK) { - TclInitStringRep(literal, start, end-start); - *lexemePtr = NUMBER; - if (literalPtr) { - *literalPtr = literal; + if (!isalnum(UCHAR(*end)) && UCHAR(*end) != '_') { + + number: + TclInitStringRep(literal, start, end-start); + *lexemePtr = NUMBER; + if (literalPtr) { + *literalPtr = literal; + } else { + Tcl_DecrRefCount(literal); + } + return (end-start); } else { - Tcl_DecrRefCount(literal); + unsigned char lexeme; + ParseLexeme(end, numBytes-(end-start), &lexeme, NULL); + if ((NODE_TYPE & lexeme) == BINARY) { + goto number; + } } - return (end-start); } if (Tcl_UtfCharComplete(start, numBytes)) { @@ -1952,7 +1947,7 @@ ParseLexeme( utfBytes[numBytes] = '\0'; scanned = Tcl_UtfToUniChar(utfBytes, &ch); } - if (!isalpha(UCHAR(ch))) { + if (!isalnum(UCHAR(ch))) { *lexemePtr = INVALID; Tcl_DecrRefCount(literal); return scanned; -- cgit v0.12 From ce6c10424834a81beb5c714d76a1bd6670d320b5 Mon Sep 17 00:00:00 2001 From: dgp Date: Tue, 6 Sep 2011 20:37:58 +0000 Subject: Don't extend numbers with non-alphanumeric characters into barewords. --- generic/tclCompExpr.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/generic/tclCompExpr.c b/generic/tclCompExpr.c index 077d595..3fdcc22 100644 --- a/generic/tclCompExpr.c +++ b/generic/tclCompExpr.c @@ -1932,6 +1932,12 @@ ParseLexeme( return (end-start); } else { unsigned char lexeme; + const char *p = start; + while (p < end) { + if (!isalnum(UCHAR(*p++))) { + goto number; + } + } ParseLexeme(end, numBytes-(end-start), &lexeme, NULL); if ((NODE_TYPE & lexeme) == BINARY) { goto number; -- cgit v0.12 From 27306c24b9e0af8fc80742dc287acbe68b072119 Mon Sep 17 00:00:00 2001 From: dgp Date: Wed, 7 Sep 2011 14:30:47 +0000 Subject: Tidiness, comments, and tests. --- generic/tclCompExpr.c | 33 +++++++++++++++++++++++----- tests/parseExpr.test | 61 +++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 89 insertions(+), 5 deletions(-) diff --git a/generic/tclCompExpr.c b/generic/tclCompExpr.c index 3fdcc22..36e1c9d 100644 --- a/generic/tclCompExpr.c +++ b/generic/tclCompExpr.c @@ -1919,7 +1919,8 @@ ParseLexeme( literal = Tcl_NewObj(); if (TclParseNumber(NULL, literal, NULL, start, numBytes, &end, TCL_PARSE_NO_WHITESPACE) == TCL_OK) { - if (!isalnum(UCHAR(*end)) && UCHAR(*end) != '_') { + if (end < start + numBytes && !isalnum(UCHAR(*end)) + && UCHAR(*end) != '_') { number: TclInitStringRep(literal, start, end-start); @@ -1932,16 +1933,38 @@ ParseLexeme( return (end-start); } else { unsigned char lexeme; - const char *p = start; - while (p < end) { - if (!isalnum(UCHAR(*p++))) { - goto number; + + /* + * We have a number followed directly by bareword characters + * (alpha, digit, underscore). Is this a number followed by + * bareword syntax error? Or should we join into one bareword? + * Example: Inf + luence + () becomes a valid function call. + * [Bug 3401704] + */ + if (literal->typePtr == &tclDoubleType) { + const char *p = start; + while (p < end) { + if (!isalnum(UCHAR(*p++))) { + /* + * The number has non-bareword characters, so we + * must treat it as a number. + */ + goto number; + } } } ParseLexeme(end, numBytes-(end-start), &lexeme, NULL); if ((NODE_TYPE & lexeme) == BINARY) { + /* + * The bareword characters following the number take the + * form of an operator (eq, ne, in, ni, ...) so we treat + * as number + operator. + */ goto number; } + /* + * Otherwise, fall through and parse the whole as a bareword. + */ } } diff --git a/tests/parseExpr.test b/tests/parseExpr.test index 7673dbe..1620a62 100644 --- a/tests/parseExpr.test +++ b/tests/parseExpr.test @@ -997,6 +997,67 @@ test parseExpr-21.63 {error message} -body { } -returnCodes error -result "missing close-brace in expression \"...12345678901234567890*\[\{abcdefghijklmnopqrstuv...\"" +test parseExpr-18.1 {LogSyntaxError procedure, error in expr longer than 60 chars} -constraints testexprparser -body { + testexprparser {(+0123456)*(+0123456)*(+0123456)*(+0123456)*(+0123456)*(+0123456)*(+0123456)/} -1 +} -returnCodes error -match glob -result * + +test parseExpr-22.1 {Bug 3401704} -constraints testexprparser -body { + testexprparser 2a() 1 +} -result {- {} 0 subexpr 2 1 text 2 0 {}} +test parseExpr-22.2 {Bug 3401704} -constraints testexprparser -body { + testexprparser nana() 3 +} -result {- {} 0 subexpr nan 1 text nan 0 {}} +test parseExpr-22.3 {Bug 3401704} -constraints testexprparser -body { + testexprparser 2a() -1 +} -result {- {} 0 subexpr 2a() 1 operator 2a 0 {}} +test parseExpr-22.4 {Bug 3401704} -constraints testexprparser -body { + testexprparser nana() -1 +} -result {- {} 0 subexpr nana() 1 operator nana 0 {}} +test parseExpr-22.5 {Bug 3401704} -constraints testexprparser -body { + testexprparser nan9() -1 +} -result {- {} 0 subexpr nan9() 1 operator nan9 0 {}} +test parseExpr-22.6 {Bug 3401704} -constraints testexprparser -body { + testexprparser 2_() -1 +} -result {- {} 0 subexpr 2_() 1 operator 2_ 0 {}} +test parseExpr-22.7 {Bug 3401704} -constraints testexprparser -body { + testexprparser nan_() -1 +} -result {- {} 0 subexpr nan_() 1 operator nan_ 0 {}} +test parseExpr-22.8 {Bug 3401704} -constraints testexprparser -body { + testexprparser nan!() -1 +} -returnCodes error -match glob -result * +test parseExpr-22.9 {Bug 3401704} -constraints testexprparser -body { + testexprparser 1e3_() -1 +} -result {- {} 0 subexpr 1e3_() 1 operator 1e3_ 0 {}} +test parseExpr-22.10 {Bug 3401704} -constraints testexprparser -body { + testexprparser 1.3_() -1 +} -returnCodes error -match glob -result * +test parseExpr-22.11 {Bug 3401704} -constraints testexprparser -body { + testexprparser 1e-3_() -1 +} -returnCodes error -match glob -result * +test parseExpr-22.12 {Bug 3401704} -constraints testexprparser -body { + testexprparser naneq() -1 +} -returnCodes error -match glob -result * +test parseExpr-22.13 {Bug 3401704} -constraints testexprparser -body { + testexprparser naner() -1 +} -result {- {} 0 subexpr naner() 1 operator naner 0 {}} + +test parseExpr-22.14 {Bug 3401704} -constraints testexprparser -body { + testexprparser 08 -1 +} -returnCodes error -match glob -result {*invalid octal number*} +test parseExpr-22.15 {Bug 3401704} -constraints testexprparser -body { + testexprparser 0o8 -1 +} -returnCodes error -match glob -result {*invalid octal number*} +test parseExpr-22.16 {Bug 3401704} -constraints testexprparser -body { + testexprparser 0o08 -1 +} -returnCodes error -match glob -result {*invalid octal number*} +test parseExpr-22.17 {Bug 3401704} -constraints testexprparser -body { + testexprparser 0b2 -1 +} -returnCodes error -match glob -result {*invalid binary number*} +test parseExpr-22.18 {Bug 3401704} -constraints testexprparser -body { + testexprparser 0b02 -1 +} -returnCodes error -match glob -result {*invalid binary number*} + + # cleanup ::tcltest::cleanupTests return -- cgit v0.12